Beyond Chronic Rejection: Tissue Remodelling in Obliterative Bronchiolitis after Lung Transplantation Masaaki Sato, MD

Transcription

1 Beyond Chronic Rejection: Tissue Remodelling in Obliterative Bronchiolitis after Lung Transplantation by Masaaki Sato, MD A thesis submitted in conformity with the requirements for the degree of doctor of philosophy Institute of Medical Science University of Toronto Copyright by Masaaki Sato 2009

2 Beyond Chronic Rejection: Tissue Remodelling in Obliterative Bronchiolitis after Lung Transplantation Masaaki Sato Doctor of Philosophy, 2009 Institute of Medical Science, University of Toronto Abstract The long-term success of lung transplantation has been challenged by chronic graft dysfunction, which is manifested as obliterative bronchiolitis (OB). We demonstrated that allograft airway fibrosis is a dynamic process of tissue remodelling, in which cellular and matrix components dynamically change before or after complete obliteration of the airway lumen. This dynamic process was associated with changes in expression and activity of matrix metalloproteinases (MMPs). The early inflammatory phase was associated with MMP-dependent migration of blood-borne fibrocytes, which highly express MMP-9 and MMP-12. Established fibrosis was associated with MMP-2 and MMP-14 expressed by myofibroblasts in both human OB lesions and their animal models. In established allograft airway fibrosis, general MMP inhibition resulted in apoptosis of myofibroblasts in vivo and in vitro, while low-doses of MMP-inhibitor treatment induced upregulation of MMP-2, increased collagenolytic activity, and significantly decreased myofibroblasts and collagen. The dynamic process of tissue remodelling in established allograft airway fibrosis was supported by underlying continuous alloimmune responses, in particular, direct T-cell-myofibroblast contact. ii

3 Modulation of tissue remodelling using a low-dose MMP inhibitor in combination with cyclosporine induced partial regression of fibrosis after its establishment. We further demonstrated the mechanism of alloimmune responses unique to the lung. Human and animal studies demonstrated that bronchioles develop de novo lymphoid tissue characterized by formation of high endothelial venules and homing of effector memory T-cells. A following study demonstrated the important role of local immunological memory maintained by the intrapulmonary lymphoid tissue in exerting effector function in allograft rejection. Collectively, the present studies support the hypothesis that tissue remodelling is an important mechanism of allograft airway fibrosis. Regulation of tissue remodelling and underlying tissue injury is important not only to arrest aberrant remodelling of allograft airways but likely to reverse aberrant remodelling and to regenerate normal tissue architecture in airways after lung transplantation. iii

4 Co-authorship This thesis contains results from previously published or submitted manuscripts and manuscripts to be submitted that are co-authored by Mingyao Liu, Masaki Anraku, Takenori Ogura, Geoffrey D Cruz, Edward Kim, Pascal Duchesneau, Shin Hirayama, Humberto Lara-Guera, Dirk Wagnetz, Benjamin A Alman, Li Zhang, David Hwang, Thomas K Waddell, and Shaf Keshavjee. All of the experimental data presented in this thesis was performed by Masaaki Sato except: mixed lymphocyte reaction assay conducted by Edward Kim (Chapter 4); orthotopic rat lung transplantation conducted by Shin Hirayama (Chapter 5 and 9); a part of semi-quantitative and quantitative analysis for histological and immunohistochemical slides were assisted by Takenori Ogura, Masaki Anraku, and Dirk Wagnetz (Chapters 5 9); a part of flow cytometric analysis was assisted by Pascal Duchesneau (Chapter 8); Geoffrey D Cruz contributed in establishing the technique to measure enzymatic activity of matrix metalloproteinases (Chapter 6). Preliminary results in Chapter 10 were produced by Humberto Lara- Guera (spectrotyping was conducted); Dirk Wagnetz and Shin Hirayama (mouse intrapulmonary tracheal transplantation and cyclosporine treatment). Chapters 1 10 of this thesis were written by Masaaki Sato. iv

5 Acknowledgments I would first and foremost like to express my deep gratitude to my supervisor, Prof. Shaf Keshavjee. I truly appreciate his patience during my struggle sometimes the project appeared to spread out as if going nowhere. However, his patience allowed me to develop and culture ideas of my own and finally integrate them into a more comprehensive theory of tissue remodelling that bridges matrix biology to immunology. At the same time, I learned a lot about his leadership as a surgeon scientist during the last five years. His indefinite interest in many fields of biology, technology, and broad clinical medicine has always amazed and stimulated me. Such capacity makes him my role model as an academic surgeon. I was fortunate to be able to work in the Latner Thoracic Surgery Research Laboratories. I am deeply indebted to Prof. Mingyao Liu, my co-supervisor. I could never develop my research career without Mingyao. His education was particularly important in many proposals for grants and fellowships as well as in my Ph.D. transfer exam. His guidance provided key insight and knowledge that will undoubtedly help me learn how to survive as a researcher. Important advice was also provided by Dr. Tom Waddell that had significant impact on my research plan. Tom is another important role model to me as an academic surgeon. His profound insight into both clinical and basic science is impressive and obtaining such high levels of wisdom is one of my own future goals. The pathological support from Dr. David Hwang was indispensable to the development of my research all through my Ph.D studies. I am extremely grateful for his warm support and his professional viewpoint. I would like to thank my committee members, Dr. Benjamin Alman and Dr. Rama Khokha. The clue leading to the core concept of tissue remodelling in this dissertation was first provided by Dr. Alman. Advice regarding MMP studies from Dr. Rama Khokha were always important to my studies as well the biological effect of the group of enzymes was so complex that I definitely needed help from an expert. I would also like to thank Dr. Li Zhang, even though she was not one of my committee members. She has given me precious advice from her immunological expertise. Since studies on posttransplant graft dysfunction can never be free of transplant immunology, her support has been and will always be an indispensable part of our research in this field. Collectively, this invaluable support represents the unique environment of the University of Toronto, in which world-class scientists work as our neighbours. I would also like to thank all the members of the Latner Thoracic Surgery Research Laboratories. My Ph.D was a long journey during which I met many students, fellows and technicians of various nationalities and cultural backgrounds. Looking back, it is amazing to see how many people I have v

6 worked with. Among these people, I particularly thank Dr. Shin Hirayama for his time spent on orthotopic lung transplantation; his surgical technique in combination with mine realized unique studies that were not possible otherwise. I also thank Dr. Helen Chan for her support in reviewing and editing this dissertation. I would like to thank Prof. Hiromi Wada, the previous professor of the department of Thoracic Surgery, Kyoto University in Japan, where I underwent my surgical residency training. He first introduced me to Prof. Keshavjee in 1999 when he was visiting Japan when I was a first-year surgical resident. We had a conversation over a glass of champagne at the poolside for an hour and the day changed my life completely. With this regard, Prof. Wada is the very person who made everything in this dissertation happen in the first place. During my Ph.D. training, Prof. Wada continuously provided me with remote but the warmest support from Japan. I cannot fully thank Prof. Wada for the opportunity he has given to me. I thank the following foundations who funded my program: Rotary International Scholarship Program ( ); the industrial partner fellowship program of The Canadian Institutes of Health Research and Wyeth Pharmaceuticals ( ); and The Canadian Cystic Fibrosis Foundation ( ). The Canadian Cystic Fibrosis Foundation has continuous financial support to the project regarding obliterative bronchiolitis after lung transplantation during my Ph.D. training. Last but definitely not least, I thank my wife, Yukiko, her family, and my parents for their patience during my Ph.D. studies. After completing my surgical residency in 2003, it might have been questionable to them as to why I chose to pursue this Ph.D. despite so many social and economical disadvantages overseas. Without their continuous and patient support, I could never complete the research training for five years. vi

7 Dedication This work is dedicated to my wife, Yukiko; and my son, Yoshinori (Eric), born on December 24, vii

8 Table of Contents Abstract Co-authorship Acknowledgments Dedication Table of Contents List of Figures List of Tables List of Abbreviations ii iv v vii viii xiii xv xvi Chapter 1 Introduction, rationale, Hypothesis Obliterative bronchiolitis in clinical lung transplantation Introduction Contributing factors to OB/BOS Immune-mediated tissue injury Alloimmune-independent mechanisms of tissue injury Fibrosis Current preventive and therapeutic strategies for OB/BOS Immunosuppression Treatment options other than immunosuppression Failure of tissue remodelling in response to immune-mediated and alloimmune-independent injury Aberrant tissue remodelling exacerbates ongoing tissue injury Aberrant remodelling of the epithelium Aberrant remodelling in the vasculature Aberrant remodelling in the stroma Remodelled lymphoid system The link between primary graft dysfunction and chronic graft dysfunction (BOS) Conclusion and future directions Animal Studies on OB: Current Approaches and Limitations Experimental models of obliterative bronchiolitis Orthotopic lung transplant models 24 a. Rat orthotopic lung transplant models 24 b. Large animal orthotopic lung transplant models Heterotopic airway transplant models 25 a. Subcutaneous and intra-omentum tracheal transplant models of rodents 25 b. Novel intrapulmonary tracheal transplant model Orthotopic and semi-orthotopic tracheal transplant models Selection of an animal model in the investigation of OB after lung transplantation Investigation of mechanisms and therapeutic strategies using animal models of OB Allorecognition 34 a. Direct allorecognition 34 b. Indirect recognition of MHC-derived peptides 35 c. Recognition of minor antigens 35 d. B-cell mediated allorecognition 37 e. Recognition of self-antigen Costimulatory signals 38 a. The CD28-B7 pathway 38 b. The CD40-CD40L pathway Suppression of alloimmune responses 39 a. Conventional immunosuppression 39 b. Tolerance induction Inflammatory mediators in alloimmune responses 41 viii

9 a. Chemokines 41 b. Proinflammatory cytokines 42 c. Complements 44 d. Arachidonic acid derivatives 44 e. Nitric oxide (NO) 44 f. Anti-inflammatory mediators Alloantigen-independent pathways 47 a. Chemical injury 47 b. Brain-death related lung injury 47 c. Viral infection Anti-inflammatory strategies Epithelial injury 49 a. Loss of epithelium 50 b. Epithelial-mesenchymal crosstalk Epithelial protection and regeneration Fibroproliferation 52 a. Profibrotic growth factors 52 b. Th2 cytokines 54 c. Angiogenic factors promoting fibrosis 54 d. The angiotensin system 55 e. The MMP-TIMP systems Anti-fibroproliferative strategies The concept of tissue remodelling integrating repair, regeneration and aberrant remodelling: the rationale Tissue remodelling integrating repair, regeneration and aberrant remodelling Tissue remodelling: not equivalent to fibrosis but a dynamic process in tissue Transition from injury to remodelling Regeneration vs. aberrant remodelling Reversibility of aberrant remodelling and required conditions Potential mechanisms of tissue remodelling that promote or resist fibrosis regression: the rationale of the study Models to be used to investigate tissue remodelling and immune responses after lung transplantation Hypothesis and Study Aims Hypothesis Specific study aims 67 Chapter 2 Materials and Methods Human tissue samples of BOS lungs and normal control lungs Animal Models Animals Tracheal transplant models of OB and combination with orthotopic lung transplantation Epithelial denudation of a tracheal graft In vivo drug Treatment Cell culture Fibrocyte isolation and culture MMP inhibitor treatment for fibrocytes and wound healing assay Fibroblast culture and induction of the myofibroblast phenotype SC080 treatment for myofibroblasts Myofibroblast culture and co-culture with T lymphocytes PBMC labelling and injection into allograft recipient animals Histological, immunohistochemical, and immunofluorescence assessments 73 ix

10 Histology Immunohistochemistry and Immunofluorescence for tissue Quantification of lymphoid tissue in the lung Immunofluorescence with terminal deoxynucleotidyl transferase-mediated dutp nick end labelling (TUNEL) Morphometric myofibroblast viability/differentiation analysis Flow Cytometry Analysis Quantitative Reverse Transcriptase Polymerase Chain Reaction (RT-PCR) Analysis Assessment of proteinase activities Assessment of gelatinolytic MMPs Collagenolytic activity assay T-cell Proliferation Assay and Mixed Lymphocyte Reaction (MLR) Assay ELISA for TGF-β Statistics 81 Chapter 3 Allograft airway fibrosis in the pulmonary milieu: A disorder of tissue remodelling Abstract Introduction Study Design Results Establishment of fibrosis in allografts by day Dynamic tissue remodelling in allografts after complete fibrotic obliteration in the airway Dynamic changes of MMPs in allografts MMP inhibition reduces allograft airway fibrosis but does not block T cell activation MMP inhibition reduces the progression of developing fibrosis Established fibrosis is modulated by MMP inhibition Discussion Tissue remodelling is a continuous process after fibrosis establishment Tissue remodelling is manipulatable MMP inhibition is effective even after the initiation of the fibrotic process Limitations of the study and future directions to modify tissue remodelling in OB Conclusion 104 Chapter 4 Metalloproteinase-dependent migration of myofibroblast progenitors of extrapulmonary origin contributes to post-transplant allograft airway fibrosis in the lung Abstract Introduction Results Myofibroblasts of extra-pulmonary origin in allograft airway fibrosis in the lung Cultured circulating fibrocytes can differentiate into myofibroblasts MMP expression and MMP-dependent migration of fibrocytes in vitro MMP inhibition reduces obliterative fibrosis even after the initiation of fibrotic process while cyclosporine does not 119 x

11 MMP inhibitors reduce myofibroblasts without significantly changing the fibroblastmyofibroblast ratio, proliferation, and cell death Discussion Fibroproliferative tissue remodelling is an important therapeutic target of OB to overcome the limitations of current immunosuppression Concurrent intrapulmonary tracheal and orthotopic lung transplantation enables the direct evaluation of myofibroblast progenitor migration of extra-pulmonary origin into the lung Fibrocytes depend on MMPs for their migration MMP inhibition could have effects other than inhibition of fibrocyte migration Conclusion 130 Chapter 5 The mechanisms of myofibroblast-mediated metalloproteinase-dependent tissue remodelling after lung transplantation Abstract Introduction Results MMP-2 and MMP-14 are localized to myofibroblasts in BOS lungs Localization of MMP-2, MMP-14, and matrix degradation to myofibroblasts in experimental OB MMP inhibitor treatment for animals MMP inhibitor treatment for animals MMP inhibitor treatment modulates cellular and matrix components of established fibrosis Dose-dependent decrease in procollagen gene expression by SC080 treatment Low-dose SC080 upregulates gene expression of MMP-2 and increases collagenolytic activity MMP-inhibitor treatment decreases myofibroblasts in fibrous tissue Discussion MMP-mediated tissue remodelling of allograft airway fibrosis is potentially bidirectional MMP-2 and MMP-14 expressed by myofibroblasts are the key MMPs involved in tissue remodelling of OB Upregulation of collagenolytic MMPs is important to fibrosis regression Treatment of underlying tissue injury may be necessary to realize fibrosis regression Induction of myofibroblast apoptosis is another therapeutic niche for MMP modulation Conclusion 152 Chapter 6 Myofibroblast-T-cell interaction supporting persistent post-transplant allograft airway fibrosis Abstract Introduction Results Alloimmune-related animal OB lesions accompany continuous T-cell infiltration, but isograft OB lesions do not Alloimmune-related animal OB lesions are associated with increased myofibroblasts and collagen deposition compared with alloimmune-free animal OB lesions Cytokines and chemokines did not show significant difference between isograft OB and allograft OB lesions Direct contact with T cells facilitate persistence of the myofibroblast phenotype in vitro 163 xi

12 Modification of tissue remodelling in combination with immunosuppression induces regression of established obliterative fibrosis Immunosuppression reduces the number of myofibroblasts and changes their morphology SC080 treatment affects T-cell apoptosis while cyclosporine reduces collagen deposition The therapeutic effect of combined treatment is reduced in the long term Discussion Treatment of underlying injury is essential to reverse fibrosis Modification of tissue remodelling is necessary to reverse allograft airway fibrosis Direct myofibroblast-t-cell contact is one of the pathways that keeps activating myofibroblasts in pathological fibrosis Regeneration of the epithelium is missing in the present approach to fibrosis Conclusion 175 Chapter 7 The role of intrapulmonary de novo lymphoid tissue in obliterative bronchiolitis after lung transplantation Abstract Introduction Results Effector memory T cells in small airways of BOS lungs Development of high endothelial venules in small airways after lung transplantation Alloantigen-dependent ELT formation after rat intrapulmonary tracheal transplantation Early lymphocyte aggregates are insufficient for allograft airway rejection Memory lymphocytes homing to the lung exert effector function Discussion A novel type of de novo lymphoid tissue in the lung The role of de novo lymphoid tissue in obliterative bronchiolitis after lung transplantation The role of de novo lymphoid tissue in the transition from acute to chronic rejection Questions remaining in de novo lymphoid tissue in the lung after transplantation Conclusion 199 Chapter 8 Discussion and Future Directions Discussion The concept of tissue remodelling in OB after lung transplantation Tissue remodelling, a novel therapeutic niche in lung transplantation Ongoing immune responses, T-cell-myofibroblast crosstalk, and reversibility of fibrosis De novo lymphoid tissue formation in the lung after transplantation Future directions Hypothesis and specific aims of the future study To investigate the role of lymphoid tissue in the lung after transplantation 208 a. To investigate human intrapulmonary lymphoid tissue in stable lung transplant recipients and in those who developed OB/BOS. 208 b. To investigate dendritic cells involved in de novo lymphoid tissue in the lung after transplantation 209 xii

13 c. To determine the role of secondary and ectopic lymphoid tissue in allograft airway rejection using transgenic mice lacking secondary lymphoid tissue To investigate the long-term immunoregulatory effect of IL-10, including its effect on lymphoid tissue formation after transplantation 213 a. Adenovirus-mediated IL-10 gene therapy in an intrapulmonary tracheal transplant model 213 b. Lentivirus-mediated long-term IL-10 gene transfer To investigate strategies to modulate tissue remodelling in allograft airway fibrosis To investigate the interaction between the immune system and tissue remodelling in allograft airway fibrosis 221 a. Molecular mechanisms of cell-cell contact between myofibroblasts and T cells 221 b. Identification of T-cell phenotypes are important to myofibroblast persistence 222 c. Phenotypical changes of myofibroblasts and T cells through cell-cell interaction Conclusion 223 List of Figures Figure 1-1. Kaplan-Meier survival of adult lung transplantation by era 5 Figure 1-2. Freedom from OB/BOS for adult lung transplant recipients 6 Figure 1-3. The continuous injury-remodelling cycle in OB/BOS 15 Figure 1-4. Epithelial and vascular injury that exacerbate aberrant tissue remodelling 18 Figure 1-5. Crosstalk between stromal cells and the immune system perpetuates the cycle of repetitive injury and stromal remodelling 20 Figure 1-6. The number of publications using heterotopic tracheal transplant models 27 Figure 1-7. A rat intrapulmonary tracheal transplant model 29 Figure 1-8. Remaining airway epithelium in human obliterative bronchiolitis and its animal model 31 Figure 1-9. Theories explaining the relationships between fibrosis and MMP-TIMP balance 57 Figure Normal tissue repair and regeneration vs. aberrant tissue remodelling after injury 63 Figure 2-1. A technique of microdissection 78 Figure 3-1. Schematics of the study design using an MMP inhibitor 86 Figure 3-2. Fibrotic obliteration of allograft airways in the lung 88 Figure 3-3. Dynamic changes of collagen, fibroblasts, and myofibroblasts in allografts 90 Figure 3-4. Dynamic changes in activity and expression of matrix metalloproteinase (MMP) in allografts 92 Figure 3-5. SC080 treatment from day 0 to day 28 prevents allograft airway fibrosis 94 Figure 3-6. SC080 treatment from day 0 to day 28 inhibits lumenal leukocyte infiltration, but not T cell activation in allografts 96 Figure 3-7. SC080 treatment from day 14 to day 28 reduced airway obliteration 98 Figure 3-8. SC080 treatment from day 21 to day 35 modulated airway fibrosis 100 Figure 4-1. Myofibroblasts in allograft airway fibrosis are of extra-pulmonary origin 112 Figure 4-2. Cultured blood-borne fibrocytes and their gene expression 114 Figure 4-3. PKH26-labelled PBMCs differentiate into myofibroblasts in allograft airway fibrosis 116 Figure 4-4. MMP expression and MMP-dependent migration of cultured fibrocytes 118 Figure 4-5. The effect of cyclosporine, SC080, and MMI270 on obliterative allograft airway fibrosis 121 xiii

14 Figure 4-6. Fibroblasts, myofibroblasts, and collagen in allograft fibrosis after treatment 124 Figure 4-7. Double immunofluorescence of Ki67 + or TUNEL and vimentin or α-sma 125 Figure 5-1. Localization of MMP-2, MMP-9, and MMP-14 in human OB after lung transplantation 136 Figure 5-2. Localization of MMPs to myofibroblasts and peri-cellular matrix degradation 138 Figure 5-3. SC080 treatment for established allograft airway fibrosis 140 Figure 5-4. Low-dose SC080 increases MMP-2 gene expression and collagen degradation 142 Figure 5-5. The effect of SC080 treatment on myofibroblasts in vivo 144 Figure 5-6. Accumulation of macrophages and T cells in fibrous tissue after high-dose SC080 treatment 145 Figure 5-7. Induction of the myofibroblast phenotype in primary cultured pulmonary fibroblasts 147 Figure 5-8. SC080 induces apoptosis in myofibroblasts in vitro 148 Figure 6-1. The effect of epithelial denudation on graft obliteration and T-cell infiltration 158 Figure 6-2. The effect of alloimmune responses on myofibroblasts and collagen 160 Figure 6-3. Cytokines and chemokines are not different between isograft fibrosis and allograft fibrosis 162 Figure 6-4. Direct co-culture of myofibroblasts with T cells facilitates the persistence of the myofibroblast phenotype in vitro 164 Figure 6-5. Combination of SC080 and cyclosporine induces fibrosis regression in vivo 166 Figure 6-6. The effect of CsA, SC080, and their combination on myofibroblasts 168 Figure 6-7. The effect of SC080, CsA, and their combination on T cells and collagen deposition 170 Figure 6-8. Allograft airways at day 56 after 4 weeks of treatment of CsA, SC080, and their combination Figure 7-1. Effector memory T cells in airways after lung transplantation 182 Figure 7-2. Development of high endothelial venules (HEVs) in small airways after lung transplantation 184 Figure 7-3. HEVs are associated with lymphocytic bronchiolitis and active OB 185 Figure 7-4. Development of effector lymphoid tissue in airways after intrapulmonary tracheal transplantation 188 Figure 7-5. Lymphocyte aggregates are immature at day 7 after intrapulmonary allograft tracheal transplantation 190 Figure 7-6. Lymphocyte aggregates become mature by day 28 after intrapulmonary allograft tracheal transplantation 192 Figure 7-7. Stable homing of T cells derived from the initial tracheal transplant recipient and memory T cells in the lung 193 Figure 7-8. The effector function of lymphocytes in the lung 195 Figure 8-1. Overview of the mechanisms of tissue remodelling 203 Figure 8-2. Microdissection and specrotyping of rat intrapulmonary lymphoid tissue 212 Figure 8-3. A preliminary result of initial immunosuppression after rat intrapulmonary tracheal transplantation 215 Figure 8-4 Preliminary results of modified intrapulmonary tracheal transplant model and the effect of a novel MMP inhibitor 220 xiv

15 List of Tables Table 1-1. Advantages, disadvantages, and examples of model studies 33 Table 2-1. Antibodies used in immunohistochemical and immunofluorescence assessments 76 Table 2-2. PCR primers used in the study 79 Table 4-1. MMP inhibition does not change the ratio of myofibroblast differentiation, fibroblast/myofibroblast proliferation or apoptosis 126 Table 7-1. Demographics of lung transplant recipients who were diagnosed with OB/BOS 181 xv

16 List of Abbreviations α-sma Alpha Smooth Muscle Actin ACE Angiotensin converting enzyme ALG/ATG Anti-lymphocyte/anti-thymocyte globulins ANOVA Analysis of variance APC Antigen-presenting cell ARDS Acute respiratory distress syndrome AUC Area under curve BALT Bronchus associated lymphoid tissue BN Brown-Norway BOS Bronchiolitis Obliterans Syndrome CMV Cytomegalovirus CO Carbon monoxide CsA Cyclosporine A CTLA4 Cytotoxic T lymphocyte-associated protein 4 DMSO Dimethyl sulfoxide DNA Deoxyribonucleic acid ECM Excessive extracellular matrix EDTA ethylene diamine tetraacetic acid ELISA Enzyme linked immunosorbent assay ELT Effector lymphoid tissue EMT Epithelial-mesenchymal transition enos Endothelial NO synthase ET-1 Endothelin-1 FAK Focal adhesion kinase FBS Fetal bovine serum FEF25-75 Forced mid-expiratory flow rate FEV1 Forced expiratory volume in one second FGF Fibroblast growth factor GF Growth factor GVHD Graft-versus-host disease H&E Haematoxylin and eiosin HD High dose HEV High endothelial venule HLA Human leukocyte antigen xvi

17 HMG-CoA 3-hydroxy-3-methyl-glutaryl-CoA HO-1 Heme oxygenase-1 hrtgf-β Human recombinant transforming Growth Factor Beta ibalt Inducible bronchus associated lymphoid tissue ICAM-1 Intercellular adhesion molecule-1 IFN-γ Interferon Gamma Ig Immunoglobulin IL Interleukin IL-2Rα Interleukin-2 receptor alpha inos Inducible NO synthase LD Low dose LFA-1 lymphocyte function-associated antigen-1 LTα Lymphotoxin alpha LYVE-1 Lymphatic vessel endothelial receptor 1 MadCAM-1 Mucosal addressin cell-adhesion molecule 1 MCP-1 Monocyte chemotactic protein-1 mhag Minor histocompatibility antigen MHC Major histocompatibility complex MMF Mycophenolate mofetil MMP Matrix Metalloproteinase NF-κB Nuclear factor-kappa B OAD Obliterative airway disease OB Obliterative Bronchiolitis PARS Poly (ADP)-ribose synthetase PBMC Peripheral blood mononuclear cell PBS Phosphate Buffered Saline PDGF Platelet-derived growth factor PI3K Phosphatidylinositol 3-kinase PNAd Post-nodal addressin PSR Picrosirius Red RANTES Regulated upon activation, normal T cell expressed and secreted RNA Ribonucleic acid RT-PCR Reverse Transcriptase Polymerase Chain Reaction SDS-PAGE sodium dodecyl sulfate polyacrylamide gel electrophoresis SEM Standard Error of the Mean xvii

18 TGF-β TIMP TLO TNF-α TUNEL VEGF Transforming Growth Factor Beta Tissue Inhibitor of Matrix Metalloproteinase Tertiary lymphoid organ Tumor Necrosis Factor Alpha Terminal deoxynucleotidyl transferase-mediated dutp nick end labelling Vascular endothelial growth factor xviii

19 Dissemination of Thesis Content Publications Sato M, Liu M, Anraku M, Ogura T, D Cruz G, Alman BA, Waddell TK, Kim E, Zhang L, Keshavjee S. Allograft airway fibrosis in the pulmonary milieu: a disorder of tissue remodeling. Am Journal of T Transplantation. 2008;8: The content of this manuscript is used with permission from Blackwell Publishing Sato M, Keshavjee S. Bronchiolitis Obliterans Syndrome: Alloimmune-Dependent and Independent Injury with Aberrant Tissue Remodeling. Seminers in Thoracic and Cardiovascular Surgery (in press). The content of this manuscript is used with permission from the Elsevier Limited. xix

20 Chapter 1 Introduction, rationale, Hypothesis A part of this chapter was published in Seminars in Thoracic and Cardiovascular Surgery. Sato M, Keshavjee S. Bronchiolitis obliterans syndrome: alloimmune-dependent and -independent injury with aberrant tissue remodeling. 2008;20(2):

21 1.1. Obliterative bronchiolitis in clinical lung transplantation Long-term success in lung transplantation continues to be challenged by chronic graft dysfunction, which is manifested as bronchiolitis obliterans syndrome (BOS). The mechanisms of BOS involve both immune-mediated pathways (rejection, auto-immune-like mechanisms) and alloimmune-independent pathways (infection, aspiration, ischemia, primary graft failure), which lead to fibroproliferative responses. BOS correlates histologically with obliterative bronchiolitis in terminal bronchioles and evidence of aberrant remodelling in the airway epithelium, vasculature, stroma, and lymphoid system. A potentially important mechanism that supports the progressive and therapy-resistant nature of BOS is a continuous cycle of ongoing injury and aberrant remodelling. Namely, anatomical and functional abnormalities induce and exacerbate immune-mediated and alloimmune-independent pathways through various mechanisms (e.g. epithelial remodelling decreases mucociliary clearance that exacerbates aspiration-related injury). From this viewpoint, we will review current therapeutic strategies and revisit the role of transplant surgeons in attenuating the initial transplant-related injuries to prevent lung grafts from entering the remodelling-injury cycle Introduction Over the last two decades, lung transplantation has become the mainstay of treatment for terminal pulmonary disorders. With the improvement of early post-transplant management, chronic graft dysfunction has become the major challenge (Error! Reference source not found.). Chronic graft dysfunction of transplanted lungs is manifested as obliterative bronchiolitis (OB) and its clinical correlate, bronchiolitis obliterans syndrome (BOS). OB is a histological diagnosis referring to inflammatory and fibroproliferative obliteration of bronchioles usually seen in transbronchial biopsies (1), while BOS is the associated clinical diagnosis referring to a decline in pulmonary function (forced expiratory volume in one second, FEV1 and/or forced mid-expiratory flow rates, FEF ) (2). OB/BOS affects about 50% of lung transplant recipients five years after lung transplantation and is the major cause of morbidity and mortality in long-term survivors of lung transplantation (3) (Figure 1-2). 2

22 OB/BOS is considered to be a multifactorial process, involving alloimmune-dependent and independent factors, and relatively novel autoimmune-like pathways. In this review, molecular mechanisms of these injurious processes and current preventive and therapeutic strategies for OB/BOS will be updated. Secondly, we shed light on aberrant tissue remodelling, which has been considered to be an end result of tissue injury, but may be an important exacerbating factor in ongoing inflammatory and immune responses. Thirdly, based on this relatively novel viewpoint, we will revisit the potentially important role of transplant surgeons in attenuating the initial transplant-related injuries to prevent the lung grafts from entering the injury-remodelling cycle Contributing factors to OB/BOS Immune-mediated tissue injury The important contributions of alloimmune-mediated mechanisms have been supported by the analyses of risk factors for OB/BOS including episodes of acute rejection (4), lymphocytic bronchitis/bronchiolitis (5), mismatch in human leukocyte antigen (HLA)-A, -B(6), and -DR (7) loci, and existence of pre-transplant anti-human HLA antibodies (8, 9). CD4 + T cells in lung transplant recipients who developed BOS have been demonstrated to be highly sensitized to mismatched class I (10, 11) and class II (11, 12) MHC compared with those from BOSfree patients. Indirect allorecognition is a form of antigen recognition, in which recipient T cells recognize a processed peptide derived from donor mismatched MHC molecules presented by recipient antigen-presenting cells (APCs). Increasing evidence suggests more important roles of indirect recognition in chronic rejection in contrast to the major role of direct recognition mediated by donorderived passenger APCs in acute rejection (13). Humoral immune mechanisms may also be important to the development of BOS. Development of anti-mhc antibody increases the risk of lymphocytic bronchiolitis (9) and BOS (9, 14). Anti-MHC class I antibody is associated with the early development of BOS (15, 16) and decreased patient 3

23 survival (17). Anti-class II MHC antibody also develops, but rarely and the significance remains unclear (17). Interestingly, anti-mhc antibodies do not simply activate complement-mediated tissue injury, but can also induce cellular activation. Anti-MHC antibodies appear to activate airway epithelial cells to induce fibroproliferative growth factors, such as platelet-derived growth factor (PDGF), which may promote proliferation of fibroblasts (18). In addition to the allorecognition of mismatched MHC molecules, recognition of non-mhc antigens potentially contributes to the development of OB/BOS. Degradation and release of fragmented type V collagen has been demonstrated to activate autoimmune-like immune responses in animal models of OB (19). In clinical lung transplantation, BOS has been associated with an increase in type V collagenspecific interferon-γ-producing Th1 cells and a decrease in type V collagen-specific IL-10-producing T cells (20). A recent study has demonstrated the involvement of Th17-dependent cellular immunity against type V collagen (21). Th17 cells are considered to be a subset of CD4 + T cells, distinct from conventional Th1 and Th2 cells that play important roles in autoimmune diseases in part through their production of IL-17 (22). Antibodies against non-mhc antigens may also be important in the development of BOS. Non-MHC antibodies directed to airway epithelium have been demonstrated to develop in lung transplant recipients in association with BOS (23). A recent study has identified anti-kα1 tubulin antibody that is directed to airway epithelial cells (24). Interestingly, this antibody may also stimulate epithelial cells and augment fibroproliferative responses through the production of pro-fibrotic growth factors (23, 24). 4

24 Survival (%) (N=4,392) (N=6,726) /2004 (N=9,419) : 1/2-life = 3.9 Years; Conditional 1/2-life = 7.0 Years : 1/2-life = 4.5 Years; Conditional 1/2-life = 7.1 Years /2004: 1/2-life = 5.3 Years; Conditional 1/2-life = 6.4 Years Years Figure 1-1. Kaplan-Meier survival of adult lung transplantation by era. This data was cited from the Registry data of the International Society for Heart and Lung Transplantation (3). Note the improved overall survival in recent lung transplantation is mostly attributed to improvement early after transplantation, while the survival curves by era are parallel after 1 year, suggesting that chronic graft dysfunction may be the limiting factor in long-term survival of lung transplant recipients. 5

25 % Freedom from OB Years Figure 1-2. Freedom from OB/BOS for adult lung transplant recipients. This data was cited from the Registry data of the International Society for Heart and Lung Transplantation (3). Follow-ups: April 1994-June 2005, conditional on survival to 14 days (n = 10,173). 6

26 Alloimmune-independent mechanisms of tissue injury Alloimmune-independent factors such as cytomegalovirus (CMV) pneumonitis (25), gastroesophageal reflux disease (26), and primary graft dysfunction (27), also increase the risk of OB/BOS. These factors may elicit innate immune responses (and adaptive immune responses to cytomegalovirus) that induce collateral damage to airways. Importantly, recent studies suggest that innate immunity plays an important role in the acceleration of alloimmune responses after organ transplantation. Theoretically, any alloantigen-independent insults including infection, chemical irritants (e.g. aspiration), and ischemia-reperfusion injury at the time of lung transplantation can lead to the release of danger signals such as viral double-stranded RNA, bacterial CpGDNA, or endogenous heatshock proteins (28). These danger signals can activate professional antigen-presenting cells (e.g. dendritic cells) by signalling through pathogen recognition receptors, like Toll-like receptors, leading to effective antigen presentation to alloreactive T cells (28). In fact, polymorphisms in Toll-like receptor 4 (recipient heterozygosity for either Asp299Gly or Thr399Ile) have been shown to be significantly associated with a reduced frequency of acute rejection and a trend towards a decrease in severe BOS (29). Innate immunity appears to be an important accelerator to alloimmune responses. Involvement of innate immunity in BOS has also been supported by the strong correlation of BOS with airway neutrophilia and high levels of IL-8, which is a potent chemoattractant to neutrophils (30, 31). Neutrophilia can be caused by infection and various other alloantigen-independent tissue injuries. Bacterial colonization (32) and bile acid aspiration (33) have been associated with airway neutrophilia, elevated IL-8, and a decline in pulmonary function. Conversely, neutrophilia can be induced by adaptive immune responses. For example, lymphocyte-derived IL-17 induces the release of IL-8 in airway epithelial and endothelial cells (34). Either way, neutrophils can cause alloimmune-independent tissue injury through their production of reactive oxygen species, myeloperoxydases, and matrix metalloproteinases. Airway neutrophilia has been associated with overproduction of MMP-8 and 7

27 MMP-9 in BOS patients (35). Thus, blocking neutrophil-dependent pathways may attenuate tissue injury in OB/BOS Fibrosis Tissue injury involving an alloantigen-dependent and -independent processes induces a fibrotic reaction. Although activation of fibroblasts and accumulation of extracellular matrix is an important process in normal wound repair, the reparative process appears to be excessive in transplanted lungs. Upon activation, fibroblasts differentiate into myofibroblasts, which express alpha smooth muscle actin (α-sma) and exhibit an intermediate phenotype between fibroblasts and smooth muscle cells. Myofibroblasts are commonly observed in fibroproliferative disorders, including OB after lung transplantation (36), and contribute to excessive production of extracellular matrix (37). Inflammatory and immune responses are the primary contributors to excessive fibroblast activation. In general, macrophages that infiltrate inflammatory tissue are an important source of various profibrotic growth factors such as transforming growth factor (TGF)-β and PDGF; and this also appears to be true in OB lesions after lung transplantation (38, 39). TGF-β has been recognized to be an important differentiation and survival factor for myofibroblasts (37). IL-13, a Th2 cytokine, has recently been demonstrated to contribute to fibroproliferation in OB/BOS (40, 41). In allergic asthma, IL-13 is considered to be one of the most detrimental tissue remodelling factors (42). Moreover, accumulating evidence suggests that IL-13 exerts its potent profibrotic effect by directly stimulating extracellular matrix production, in addition to increasing expression and activation of TGF-β (42). The interaction between fibroblasts and epithelial cells is another important component of fibroproliferation in airways. In asthma studies, normal airway epithelial cells have been demonstrated to have potent inhibitory effects on subepithelial fibroblasts, in part through the production of prostaglandin E2, which inhibits TGF-β-dependent fibroblast activation (43). The inhibitory role of the epithelium has been confirmed in animal models of OB, in which enzymatic denudation of the epithelium resulted in obliterative airway fibrosis even in syngenic grafts (44). After clinical lung 8

28 transplantation, normal epithelial regeneration may be hindered by multiple injurious factors as well as calcineurin-inhibitor immunosuppressants like cyclosporine and tacrolimus (45). Damage to the epithelium may reduce the natural inhibitory effect of the epithelium on fibroblasts. Conversely, as mentioned above, epithelial cells can be activated by anti-mhc and non-mhc antibodies to produce profibrotic growth factors that promote fibroproliferation (18, 24). Besides resident airway fibroblasts, other cells may contribute to the myofibroblast population in OB lesions. A portion (15-30%) of myofibroblasts in established OB lesions have been identified to be of recipient origin using sex-mismatched transplant samples and gene polymorphisms (36). This finding suggests a contribution of circulating fibrocytes a group of blood-borne mononuclear cells expressing collagen I that are capable of differentiating into fibroblast-like cells in peripheral tissue (46). Circulating fibrocytes are recruited in part through a chemokine-dependent mechanism (e.g. CXCL12 (47)) in various inflammatory and fibrous conditions such as pulmonary fibrosis (47). Another potential source of myofibroblasts in OB lesions is epithelial-mesenchymal transition (EMT), in which epithelial cells transdifferentiate into cells of mesenchymal phenotype like fibroblasts and myofibroblasts. In the kidney, EMT is a well established concept (48) and the most important signalling pathway leading to the induction of EMT is considered to be through TGF-β and downstream signalling through Smad 2, 3, and 4. The signalling cascades of bone-morphogenic protein 7 may have the opposite effect (48). In lung transplantation, evidence of EMT remains to be confirmed (49). Blocking myofibroblast differentiation may provide an opportunity to therapeutically target the fibroproliferation process in OB/BOS Current preventive and therapeutic strategies for OB/BOS Immunosuppression Augmentation and modification of maintenance immunosuppression has been the mainstay of BOS management. Since the 1990s, several new immunosuppressive agents have been introduced to lung transplantation. Although these new drugs appear to provide some benefit to lung transplant recipients 9

29 by delaying the onset or attenuating the progression of BOS, they do not prevent BOS nor cure the disease. Lung transplant recipients usually receive a triple-drug maintenance regimen that includes a calcineurin inhibitor, an antimetabolite, and a corticosteroid. The calcineurin inhibitors, cyclosporine and tacrolimus (FK506) interfere with signal transduction from the cell surface receptors to the nucleus in T lymphocytes, preventing transcription of IL-2 genes involved in T cell activation. In a prospective comparison of cyclosporine and tacrolimus in combination with azathioprine and prednisolone, a significantly lower incidence of BOS was reported in the tacrolimus group (38% vs. 22%, P = 0.025) (50). A retrospective multi-center study on 134 patients with post-lung transplant BOS reported that conversion from cyclosporine to tacrolimus for BOS allowed for short-term stabilization of lung function in most patients (51). Although a control arm was lacking in this study, many transplant centers have changed maintenance immunosuppression from cyclosporine to tacrolimus for patients with recurrent acute rejection or for those who have developed BOS (52). Azathioprine and Mycophenolate mofetil (MMF) are antimetabolic agents that interfere with nucleotide metabolism inhibiting not only T-cell, but also B-cell proliferation. MMF has been increasingly used over azathioprine (3). In a non-randomised study in 22 lung transplant recipients, MMF was reported to have a lower prevalence of BOS at one year compared with azathioprine (53). However, in a recent prospective multi-center randomized trial comparing MMF and azathioprine in combination with induction therapy, cyclosporine and corticosteroids, no differences were seen in the incidence of acute rejection or BOS at one and two years after lung transplantation (54). Currently, there is no solid evidence to support the use of MMF over azathioprine to reduce the incidence of BOS. Sirolimus and everolimus are macrocyclic lactones that inhibit growth factor-stimulated proliferation of lymphocytes and mesenchymal cells (55, 56). In a prospective comparison of everolimus and azathioprine in combination with cyclosporine and corticosteroids, everolimus slowed the decline in pulmonary function and decreased the incidence of acute rejection at 12 months (57). A large, multi- 10

30 center, open-label Phase III study of everolimus versus MMF in combination with cyclosporine and corticosteroids is currently under way. The clinical effectiveness of sirolimus in stabilizing pulmonary function in patients with BOS has been reported in small case series (58, 59). It has been reported that conversion from calcineurin inhibitor-based immunosuppression to MMF and sirolimus after the diagnosis of BOS attenuates progression of the disease in 10 patients (60). Currently, a large, multicenter, open-label Phase IV study of sirolimus versus azathioprine in combination with tacrolimus and corticosteroids is underway. Corticosteroids are commonly included in triple-drug regimens of immunosuppression after lung transplantation. A short-course of high-dose intravenous methylprednisolone is the typical empirical treatment when BOS is diagnosed (i.e. pulmonary function decline without evidence of other causes), although evidence to support this approach is limited (52). Inhaled cyclosporine has been reported to stabilize pulmonary function of BOS patients in a small case series (61). Interestingly, a recent randomised, double-blinded trial of inhaled cyclosporine for BOSfree patients demonstrated that inhaled cyclosporine did not improve the rate of acute rejection, but improved survival and extended BOS-free survival (62). Induction therapies using polyclonal agents (anti-lymphocyte/anti-thymocyte globulins, ALG/ATG), monoclonal anti-cd25 antibodies, or campath (monoclonal anti-cd52 antibody) are applied in fifty percent of transplant recipients at the time of lung transplantation (3). However, their ability to reduce the development of BOS remains to be clarified. Other treatments that have been reported to arrest or reverse some loss of pulmonary function include methotrexate (63), cyclophosphamide (64), photopheresis (65) and total lymphoid irradiation (66). Because these are small, retrospective studies, further investigation is clearly required. In summary, because the alloimmune response is a central contributing mechanism to chronic graft dysfunction after lung transplantation, the development of better immunosuppressive strategies is no 11

31 doubt mandatory in order to improve the clinical outcome of lung transplantation Treatment options other than immunosuppression Azithromycin is a macrolide antibiotic that has been used as an immunomodulatory or antiinflammatory agent in diffuse pan bronchiolitis (a chronic inflammatory airway disease seen almost exclusively in Japan). Several small clinical studies have reported that low-dose maintenance azithromycin delays the progress of OB/BOS or even improves pulmonary function in a portion of patients (67-69). Azithromycin has been demonstrated to reduce airway neutrophilia and bronchoalveolar lavage IL-8 levels in BOS patients (70). Azithromycin may inhibit IL-17-dependent IL-8 release from airway smooth muscle cells (71). The anti-inflammatory and immunomodulatory effects of azithromycin are likely exerted at lower doses than that required to exert its antibiotic effect (72). Azithromycin may play a major role as an adjuvant treatment for conventional immunosuppression, especially by modulating innate immune responses. Anti-reflux surgery is another modality performed to reduce the risk of OB/BOS after lung transplantation in some patients. Early fundoplication has been reported to decrease the incidence of OB/BOS (100% BOS-free survival at 3 years in 14 patients) in a retrospective study compared to patients who did not have fundoplication (60% BOS-free survival) (73). Fundoplication may also reverse pulmonary function in a portion of BOS patients (74). Gastroesophageal reflux is a common problem among lung transplant candidates (75) and the transplant operation can exacerbate reflux even though many patients remain asymptomatic (76). Since the prevalence of gastroesophageal reflux after lung transplantation has been reported to be as high as 70% during 24 hour ph monitoring (26), aggressive intervention in appropriate patients may improve the outcome of lung transplantation. Statins (3-hydroxy-3-methyl-glutaryl-CoA (HMG-CoA) reductase inhibitors) are agents commonly used to treat hypercholesterolemia. Statins have been demonstrated to reduce the risk of BOS after lung transplantation (77). Although the statins were prescribed to lung transplant recipients for hyperlipidemia, they may have been beneficial with respect to immunomodulatory effects (78). 12

32 Pravastatin has been demonstrated to down-regulate MHC class II molecules in a rat lung transplant model in association with prolonged allograft survival (79). Statins have also been reported to induce fibroblast apoptosis (80) and inhibit TNF-α-dependent myofibroblast invasion in vitro (81). Thus, statins are likely to have complex effects on inflammatory and immune responses as well as fibroproliferation. Further investigation and clinical trials are necessary to confirm the beneficial effect of statins in lung transplant recipients Failure of tissue remodelling in response to immune-mediated and alloimmuneindependent injury Aberrant tissue remodelling exacerbates ongoing tissue injury Unfortunately, once the diagnosis of BOS is made, the course of the disease is generally progressively downhill (82). There are many potential explanations regarding the progressive and therapy-resistant nature of BOS. For example, current immunosuppression may be less effective in blocking indirect antigen recognition of T cell- or B-cell-mediated immune responses compared to direct antigen recognition of T cells (83). Cross activation of T cells with viral antigens before transplantation may induce memory-type alloreactive T cells so that T cells in lung transplant recipient patients may not be as naïve as those in rodent models (84). Recent strategies like azithromycin and anti-reflux surgery are effective, but unlikely to be sufficient to arrest the progression of the disease completely. Besides these current limitations, we may have under-appreciated aberrant tissue remodelling per se as an exacerbating factor in ongoing tissue injury. Established fibrosis is generally considered to be the end result of injurious processes; however, from histological observations, it is evident that airway fibrosis does not affect the whole transplanted lung simultaneously; rather, it heterogeneously affects and progresses over time. Our recent animal experimental work suggests that airway fibrosis is not a static, but rather a dynamic process in which tissue destruction and remodelling is continuous (85). Moreover, in transplanted lungs, the aberrant tissue remodelling is not limited to obliterative airway fibrosis, but also exists in the airway epithelium (squamous metaplasia and ulceration), vasculature 13

33 (vascular sclerosis and abnormal angiogenesis), peripheral lung tissue (interstitial fibrosis in alveoli) (86), and possibly in the lymphoid tissue (de novo formation of ectopic lymphoid structures) (87). Of important note is that once aberrant tissue remodelling results in abnormal tissue architecture, the anatomical and functional abnormalities may exacerbate ongoing tissue injury through alloantigendependent and alloantigen-independent pathways, forming a positive feedback loop of injury and aberrant remodelling (Figure 1-). We propose that the continuous cycle of injury and remodelling is a potentially important mechanism underlying OB/BOS and will review the plausible molecular mechanisms. 14

34 Alloimmunedependent factors Adaptive immunity Alloimmuneindependent factors Innate immunity Tissue Injury Functional abnormality Defective innate defence against microorganisms Augmented adaptive immune responses Physical defects (e.g. microcirculation, mucociliary clearance) Aberrant tissue remodeling Normal tissue repair and regeneration Epithelial remodeling Vascular sclerosis Remodeling in the stroma Lymphoid neogenesis Figure 1-3. The continuous injury-remodelling cycle in OB/BOS. The continuous cycle of ongoing injury and aberrant tissue remodelling perpetuates the progressive nature of the disease. Following initial insults and failure of normal tissue repair, typical OB lesions develop along with other types of aberrant tissue remodelling in the lung. The associated structural and functional abnormalities in the lung further aggravate tissue injury through multiple mechanisms. 15

35 Aberrant remodelling of the epithelium The epithelium in BOS patients often exhibits attenuated squamous metaplasia and ulceration (86), which are relatively defective with respect to the innate defence mechanisms normally found in an intact epithelial lining (88). Because the epithelial lining provides an important physiological barrier against infectious organisms, loss of barrier function by disturbance of intercellular tight junctions results in a leaky epithelial lining with an increased risk of infection (88). Remodelled epithelium may lose the ability to effectively produce protective proteins like the Clara cell secretory protein, which is decreased in BOS patients (89). Surfactant proteins (SP-A and SP-D) and phospholipids have also been shown to be reduced in association with bile acid aspiration and development of BOS (90). Reduced epithelial cell ciliary motility leads to decreased airway mucociliary clearance and contributes to increased inflammation, mucus stasis, and plugging that further increases the risk of infection or aspiration-related injury. Azithromycin may exert its beneficial effect in part by promoting ciliary motility (91). In addition to alloantigen-dependent or -independent insults to the epithelium, impaired CXCL12- dependent recruitment of circulating epithelial progenitor cells may be a mechanism of persistent squamous metaplasia (92). If repetitive epithelial repair and regeneration cycles deplete the pool of local and/or circulating epithelial progenitors, supplying epithelial progenitors might be a potential novel therapeutic approach to this problem (93). The mechanisms of epithelial remodelling causing tissue injury are shown in Figure 1-A Aberrant remodelling in the vasculature Vascular remodelling in transplanted lungs includes vascular sclerosis and abnormal angiogenesis. Vascular or arteriosclerosis is commonly observed in transplanted solid organs. Evidence suggests that coronary vasculopathy after cardiac transplantation is multifactorial, involving transplant-related insults (e.g. ischemia-reperfusion injury), alloimmune responses to graft endothelium, and immunosuppression-related conditions including hyperlipidemia, hyperglycemia, and hypertension 16

36 (94). Parallel mechanisms are conceivable in pulmonary vasculopathy after lung transplantation, although detailed clinical analysis is yet to be conducted. Microvascular vasculopathy has been proposed to occur before the onset of OB (95). It is possible that chronic ischemia in airways impairs epithelial regeneration and renders them more vulnerable to fibroproliferative changes. Statins might reduce the incidence of BOS (77) in part by a vascular protective effect. Conversely, abnormal angiogenesis has been suggested to support fibroproliferation in airways (95). Angiogenesis in allograft airways has been demonstrated to be partially dependent on CXCR2 and vascular endothelial growth factor pathways (96, 97). Importantly, endothelial regeneration and angiogenesis in the arterial wall are major events in the pathogenesis of transplant vasculopathy both circulating and bone-marrow-derived endothelial progenitor cells have been demonstrated to contribute to this as well (98). Thus, prevention or treatment of chronic airway ischemia may not be solved solely by facilitating endothelial regeneration (e.g. supplying endothelial progenitor cells). Fibroproliferation and wound healing factors would also be important and necessary considerations. The mechanisms of vascular remodelling exacerbating tissue injury are shown in Figure 1-4B. 17

37 A B Figure 1-4. Epithelial and vascular injury that exacerbate aberrant tissue remodelling. (A) The cycle of epithelial injury and aberrant remodelling. Normal airway epithelial cells have innate defence mechanisms including mucociliary clearance, production of protective proteins, and a mechanical barrier in cell-cell tight junctions. Following injury and abnormal repair, the epithelium shows squamous metaplasia or ulceration and loses its defence mechanisms, allowing for mucus stasis and invasion of microorganisms which lead to secondary injury. The points of potential beneficial effects of anti-reflux surgery in aspiration injury and azithromycin in antibiotic and anti-inflammatory mechanisms are indicated. (B) The cycle of vascular injury and aberrant remodelling. Microcirculation around airways may be impaired directly by ischemia at the time of transplantation or chronic vascular sclerosis in the microvasculature (vasculopathy). Chronic airway ischemia may impair epithelial recovery after injury and contribute to aberrant airway remodelling. Allograft vasculopathy is attributed to many factors including alloimmune responses and immunosuppression-related hypercholesterolemia. Statins attenuate vasculopathy and may indirectly delay the development of BOS. Once aberrant airway remodelling is initiated by myofibroblasts, abnormal angiogenesis supports the process. 18

38 Aberrant remodelling in the stroma In the lung interstitial tissue, parenchymal injury and inflammation activates stromal fibroblasts, but the fibroblasts themselves may be an important accelerator of adaptive immune responses. Stimulated fibroblasts in inflammation can produce pro-inflammatory cytokines (e.g. IL-1β) (99) and also directly activate T cells through cell-cell interactions mediated by co-stimulatory molecules (e.g. CD40, CD40L) (100). Furthermore, circulating fibrocytes that migrate into the stroma during inflammation have been demonstrated to play roles as potent antigen-presenting and lymphocyte-activating cells by expressing MHC class II molecules and costimulatory molecules (e.g. CD80 and CD86) (101). In this regard, activated fibroblasts and fibrocytes behave in a similar manner to immune responsive cells such as macrophages and dendritic cells. Furthermore, activated fibroblasts may exhibit a more invasive phenotype by expressing matrix metalloproteinases (MMPs) and directly contribute to tissue destruction (81). We have recently demonstrated that MMPs, in particular MMP-2 and MMP-14 expressed by myofibroblasts, contribute to chronic tissue remodelling in allograft airway fibrosis in an animal model of OB (85). Moreover, the matrix remodelling may induce or exacerbate autoimmune mechanisms of OB/BOS through degradation and release of type V collagen (19). Thus, strategies to regulate fibroblasts/myofibroblasts may be important not only to prevent fibrosis, but also to attenuate ongoing immune responses in transplanted lungs. The mechanism of aberrant stromal remodelling is shown in Figure

39 Figure 1-5. Crosstalk between stromal cells and the immune system perpetuates the cycle of repetitive injury and stromal remodelling. After an initial injury, stromal cells represented by fibroblasts signal immune competent cells (e.g. macrophages and lymphocytes) through direct cellcell contact and/or soluble chemokines, cytokines, and growth factors. Circulating fibrocytes also contribute to the fibroblast population and interact with immune cells. Stimulated fibroblasts differentiate into myofibroblasts and mediate stromal tissue remodelling. Proteolytic processes in stromal remodelling may result in autoantigen exposure of type V collagen, which leads to augmentation of adaptive immune responses in a manner similar to autoimmune diseases. ICAM-1, intercellular adhesion molecule-1; LFA-1, lymphocyte function-associated antigen-1. ; TNF, tumor necrosis factor; IL-1, interleukin-1; TGF-β, transforming growth factor-beta. 20

40 Remodelled lymphoid system Lymphoid neogenesis or remodelling of the lymphoid network is another potential mechanism that exacerbates ongoing adaptive immune responses after organ transplantation. Lymphoid neogenesis is defined as ectopic or de novo organization of lymph node-like structures in peripheral tissues containing T cells, B cells, dendritic cells, and high endothelial venules that allow lymphocytes to move in or out of the lymph node-like structures from the circulatory system (102). Lymphoid neogenesis has been observed in various chronic inflammatory diseases including infection and autoimmune diseases and it is believed to contribute to maintaining local immune responses against persistent antigens (102). Lymphoid neogenesis has also been observed in transplantation of the heart (103), kidney (104), and liver (104). Potentially important effector mechanisms of ectopic lymphoid tissue include priming of naïve lymphocytes (102), development of antibody-mediated immune responses (104), and the development and homing of memory lymphocytes in peripheral tissues (105). None of these, however, have been clearly demonstrated in human organ transplantation (87). Moreover, it is also possible that ectopic lymphoid tissues harbour regulatory T cells and contribute to immunological tolerance (106). Clearly, lymphoid neogenesis is a relatively novel and potentially important area in transplant immunology. Thus far, the role of lymphoid neogenesis after lung transplantation remains uncertain, partly due to difficulties in distinguishing neogenesis from preexisting bronchus-associated lymphoid tissue (BALT) or lymphocytic infiltrates. Interestingly, in cardiac transplantation, Quilty lesions, which are now considered to be a form of lymphoid neogenesis, have long been confused with infiltration of lymphocytes associated with acute rejection until recent demonstration of CD21 + follicular dendritic cells (107). Thus, careful exploration will be required in elucidating the role of lymphoid neogenesis in the lung after transplantation and its contribution to chronic graft dysfunction The link between primary graft dysfunction and chronic graft dysfunction (BOS) Various pharmacologic approaches have been undertaken to reduce the incidence of BOS. On the other hand, the potential role of the transplant surgeon in the prevention of OB/BOS may have been 21

41 underappreciated. Given the continuous cycle of injury and subsequent aberrant tissue remodelling, transplant surgeons might well be the first-line defence to set the stage to modify, attenuate, or prevent the initial transplant-related injuries that occur at the time of lung transplantation. Such transplant-related injuries include donor lung injury due to brain death, aspiration, trauma, ventilationinduced injury, infection, cold ischemia, and reperfusion injury (108). These injuries contribute to primary graft dysfunction, which is clinically manifested by lung edema and hypoxemia, and is biologically associated with an increase in inflammatory cytokines and activation of innate immune responses (108). Considering the underlying mechanisms of tissue injury at the time of transplantation, it should not be surprising that primary graft dysfunction might be the earliest initiation into the cycle of recurrent injury and aberrant remodelling leading to OB/BOS. Indeed, primary graft dysfunction has been demonstrated to be a potential risk factor for OB/BOS in a relatively small group of patients (109). A larger retrospective cohort study has further demonstrated the association of primary graft dysfunction with an increased risk of BOS, independent of acute rejection, lymphocytic bronchitis and communityacquired respiratory viral infections (27). Importantly, the risk of BOS has been demonstrated to be directly related to the severity of primary graft dysfunction (27). These data emphasize the importance of early graft optimization in reducing the risk of BOS in the long term. It is true that ideal donor and recipient selection is not always possible and that overly conservative donor-lung selection due to fear of primary graft dysfunction leads to a decrease in the number of transplants, which translates into increased mortality on the waiting list. We and others have investigated strategies to expand the donor pool as well as to improve donor lungs and attenuate ischemia reperfusion injury using gene therapy (110). More recently, we have combined gene therapeutic repair with ex-vivo lung perfusion maintenance (111) in a strategy that will allow us to repair and immunologically optimize injured donor lungs before transplantation. These novel strategies might not only improve the short-term outcome of lung transplantation, but may also lead to long-term 22

42 benefits by allowing for transplantation of pre-prepared organs of better quality decreasing the magnitude of transplant-related donor lung injury and preventing early entry into the vicious cycle of injury and aberrant remodelling Conclusion and future directions Chronic lung allograft dysfunction (OB/BOS) is not simply graft rejection, but a complex multifactorial process involving immune-mediated and alloimmune-dependent tissue injuries and aberrant reparative responses or remodelling. Furthermore, the aberrant remodelling might not simply be a consequence of excessive tissue injury, but an important exacerbating factor in ongoing tissue injury, perpetuating the relentless progression of the disease. In addition to current therapeutic strategies targeting alloimmune-mediated and alloimmune-independent processes, future strategies should target the attenuation or interruption of the continuous cycle of injury and aberrant remodelling. Chronic graft dysfunction is the result of a multitude of injuries experienced by the transplanted lung. The injury process starts in the donor lung and is further exacerbated with sequential injury related to the transplant operation and subsequent innate and acquired host responses. Clearly, in order to improve the overall long term outcome of lung transplantation, each component of this multifactorial process will need to be addressed. 23

43 1.2. Animal Studies on OB: Current Approaches and Limitations Experimental models of obliterative bronchiolitis Clinical investigation of OB is limited by the availability of biological material. To study the mechanisms of OB, several animal models have been developed. However, animal studies have also been challenged by technical difficulties, reproducibility, relevance to clinical OB, and different immunological background and sensitivity to immunosuppression among species. Since none of these models perfectly represent clinical OB after lung transplantation, it is important for researchers to select an appropriate animal model that fits the purpose of their studies Orthotopic lung transplant models a. Rat orthotopic lung transplant models Rat orthotopic left lung transplant models have been extensively used to investigate ischemiareperfusion injury and acute rejection; however, creation of OB-like lesions in these models may not be reliable. Without immunosuppression, rat lung allografts are rejected and become necrotic within several days, whereas short-term immunosuppression (e.g. cyclosporine) enables long-term acceptance of allografts. Despite these difficulties, limited use of cyclosporine at the beginning of the postoperative course enabled OB-like late airway changes including granulation and submucosal fibrosis at day 100 in a total major histocompatibility (MHC) mismatched combination (Brown-Norway to Lewis) (112). OB-like lesions were also reproduced using the moderately histo-incompatible strains of Fisher 344 donors (RT lvl ) and Wistar Kyoto rat recipients (RT l ) without immunosuppression. In this model, acute rejection is not as severe as to result in whole lung necrosis and, after 49 days, the grafts show lesions such as peribronchiolar fibrosis, interstitial fibrosis, vasculitis and intimal hyperplasia (113). Following studies using the same strain combination have reported the development of OB-like lesions in small airways (19, 114). Despite these findings, reproducibility of this strain combination remains controversial (115). b. Large animal orthotopic lung transplant models 24

44 Although large animal models are costly and require special facilities for long survival experiments, several authors have described the development of OB-like lesions in large animal orthotopic lung transplant models. Using genetically defined miniature swine for orthotopic lung transplantation, development of OB-like lesions was observed after 3 months of immunosuppression and its tapering over a month (116). Reproducible OB-like lesions have also been reported using MHC-inbred miniature swine, using MHC-matched minor-antigen mismatched combinations (117) and MHC class I mismatched combinations (118). In these models, a 12-day course of postoperative cyclosporine resulted in OB-like lesions in the long term (>100 days after transplantation). Although these large animal models are expensive and not easily available, the potential advantages over rat orthotopic lung transplant models include continuous monitoring of individual animals by repeated biopsy, bronchoalveolar lavage, and computer tomography and better reflection of human immune system, especially constitutive expression of MHC II antigens in endothelial and bronchial epithelial cells (119, 120). On the other hand, like rat lung transplant models, the immunological background that necessitates immunosuppression appears to be largely different from that of human being. c. Mouse orthotopic lung transplant model Recently, a mouse orthotopic lung transplant model has been established (121). This model appears to be highly demanding technically because of the small size of the bronchus and vasculature. The obvious advantage of this model lies with the use of transgenic mice, which will enable us to dissect biological pathways that are important to lung transplantation. On the other hand, development of OB has not been demonstrated in this model. As an animal model of OB, this model may have limitations similar to those of rat orthotopic lung transplant models Heterotopic airway transplant models a. Subcutaneous and intra-omentum tracheal transplant models of rodents Heterotopic tracheal transplant models are represented by mouse or rat subcutaneous tracheal 25

45 transplantation and intra-omentum tracheal transplantation. Most importantly, in these models, isograft trachea recovers its normal tissue architecture while allografts develop obliterative fibrosis histologically similar to human OB usually within 3-4 weeks if immunosuppression is not used. Heterotopic tracheal transplant models have been extensively used as simple animal models of OB for the ease of the surgical procedure, reliable reproducibility, and low cost. Historically, tracheal isografts implanted in the omentum had been recognized to undergo revascularization and epithelial regeneration without vascular anastomosis (122, 123). Hertz et al. applied the characteristics of heterotopic tracheal transplantation to mouse subcutaneous tissue, and established a novel tracheal transplant model of OB (124). Huang et al. described rat models of heterotopic tracheal transplantation in various locations including subcutaneous tissue, the omentum, and subrenal capsule with similar end results of allograft airway fibrosis (125). Boehler et al. described the histological changes in detail in both iso- and allografts over time in subcutaneous and intraomentum rat models; the results of both were basically the same (126). There is no doubt that heterotopic tracheal transplant models have played a significant role in dissecting the mechanisms of OB. Despite that, these models have recently been criticized for their limitations which include: the use of large airways as transplant grafts instead of small airways; different donor-recipient interface (i.e. the milieu surrounding the graft is that of recipient instead of the donor lung tissue in orthotopic lung transplantation); non-vascularised graft transplantation; and lack of the contact with air as usual airways have. As a result, the number of publications using this model has declined over the last few years (Figure 1-6). This decline may also reflect an increase of availability of other animal models in this field. In other words, researchers today have more options for studying OB than before, and will encounter different types of challenges in choosing appropriate animal models for a given scientific hypothesis. We will discuss this important issue in Section of this chapter. 26

46 Figure 1-6. The number of publications using heterotopic tracheal transplant models. The number of publications was examined using PubMed* to search for literature including the terms trachea and transplantation or obliterative bronchiolitis or bronchiolitis obliterans or obliterative airway disease. The use of a heterotopic tracheal transplant model was examined from the abstract of each resulting publication. * 27

47 b. Novel intrapulmonary tracheal transplant model We have recently developed a new model of rat intrapulmonary tracheal transplantation and have demonstrated the connection of the grafts with the pulmonary circulation (127). Technically, this animal model is similar to other heterotopic tracheal transplant models in that a piece of a trachea is used as a graft. Following thoracotomy, the tracheal graft is implanted in the parenchyma of the left lung of a recipient rat, using a specially devised apparatus for graft implantation and a surgical clip for haemorrhage and air-leak control (Figure 1-7). The technical details of this animal model are also available as video clips attached to this thesis. This model succeeds the advantages of conventional heterotopic tracheal transplantation, namely, reliable reproducibility, alloimmune-dependent formation of obliterative fibrosis, development of obliterative fibrosis in the short-term and relatively easy surgical technique. The intrapulmonary model is theoretically an improved model in that transplanted tracheae are subjected to pulmonary milieu which may have a significant relevance to the accommodation of implanted lung tissue (128). Moreover, the model can provide a potential opportunity to examine immune responses unique to the lung (as discussed in Chapters 7 in detail). 28

48 Figure 1-7. A rat intrapulmonary tracheal transplant model. (a) Two donor rat tracheae after retrieval. One of them is half inserted into the 10GA angiocath, which is used later for implantation of the graft in the recipient lung parenchyma. (b) Implantation of a rat tracheal graft in the lung using 10GA angiocath. The trachea is ejected into the lung using a tip-blunted 10GA needle. (c) A higher magnification of (b). The interrupted line indicates the estimated location where the tracheal graft will be implanted. (d) After implantation of the trachea. The arrow indicates where a surgical clip is placed on the hole of the lung made to implant the tracheal graft. 29

49 c. Subcutaneous small airway transplant models of pigs Subcutaneous small airway transplant models of pigs have been used. Massilta et al. implanted pieces of a lobar bronchus subcutaneously into domestic pigs (129) whereas Ikonen et al. implanted lung fragments (1cm 3 ) with airways, and 1-2 mm diameter bronchi alone subcutaneously in pigs (130). These models were reported to develop obliterative fibrosis similar to those of rodent heterotopic tracheal transplant models. The kinetics of immune cells and epithelium with or without immunosuppression also appear to be comparable (131, 132). Advantages of these models are (i) multiple grafts implantable into one recipient, and (ii) an MHC II expression pattern on airway epithelial cells similar to humans. Unlike inbred rodents, however, the heterogeneous genetic background of randomly bred domestic pigs might be a disadvantage Orthotopic and semi-orthotopic tracheal transplant models In the allograft heterotopic tracheal transplant model without immunosuppression, the epithelium is usually denuded by rejection; however, in human OB lesions, the epithelium may or may not be completely lost (Figure 1-8A-a, b). To study the role of the epithelium or air-epithelium interface in OB, several animal models are proposed that use direct anastomosis between graft and host tracheae (Figure 1-8A-c, d). Ikonen documented a rat tracheal transplant model using a tracheal graft interposing the recipient trachea (133). Genden proposed a mouse interposing tracheal transplant model (134). Hyun Sung et al. proposed an end-to-side anastomosis model (135) and Minamoto et al. documented a double lumen tracheal transplant model in mice (136). In these models, even allografts show maintained lumenal patency, but pathologic changes such as fibroproliferation in submucosal area also coexist (Figure 1-8B). In these models, recipient-derived epithelial cells are believed to migrate and cover the interior surface of the allograft lumen (137). The rejection-free epithelium allows for unique opportunities to investigate the relationship between epithelium and fibrosis. On the other hand, because the airway epithelium in actual human lung allografts is composed predominantly of donor-derived cells likely with some degree of chimerism (138), the total replacement of the epithelium with recipient-derived cells is an artificial circumstance limiting these models. 30

50 A B Graft Graft Orthotopic Side-to-end anastomosis Double lumen Figure 1-8. Remaining airway epithelium in human obliterative bronchiolitis and its animal model. (A)(a) Total obliteration of airways in human OB after lung transplantation. (b) Subepithelial fibrosis (*) with remaining epithelium (arrows) from the same patient as (a). (c, d) A rat allograft trachea observed 21 days after side-to-end tracheal anastomosis (conducted following a publication) (135). Allograft lumen remains open with epithelial lining (arrow). Subepithelial fibrosis (*) is also observed. (B) Orthotopic and semi-orthotopic tracheal transplant models of OB. These models using tracheal anastomosis simulate the interface with the air and coexisting epithelium and subepithelial fibrosis in allograft airways. Arrows indicate migration of recipient-derived epithelial cells into the graft. 31

51 Selection of an animal model in the investigation of OB after lung transplantation A major challenge for researchers in the field of OB is selection of animal models. As discussed in this chapter, there are several options available and the fields to be investigated range from immunology to regenerative medicine. Heterotopic tracheal transplant models are often criticized mainly because they are not lung transplantation. However, when we discuss animal models, it is highly important to revisit the philosophy behind the models and to reflect over the hypothesis or scientific questions to be tested or answered using the models in question. In general, scientific models are an abstraction used in the creation of a predictive formula (Wikipedia. typically in physics but sometimes in medical science. Such a formula is seen in epidemiological studies in medicine, for example, in the formula of the Cox proportional hazard model. If not a mathematical formula, all the models, like cell culture for example, do not reflect the environment/circumstances of the real world we live in. In this regard, there are no perfect models. In other words, perfect models would not be models any more because they would reflect the same complexities of the real world. Models exist to help us to study and dissect the complex reality and to represent a certain aspect of the real world for scientists to focus upon. As such, all the OB models presented could be valid models so long as the hypothesis is appropriate for the model. Advantages and disadvantages of each animal model, and good examples for which the model are used appropriately are presented in Table

52 Model Orthotopic lung transplant Heterotopic tracheal transplant Orthotopic tracheal transplant Anatomically similar to clinical lung transplant Repetitive biopsy (large animal) Reproducibility Productivity Advantage Formation of fibrosis Epithelial-air interface Disadvantage Controversial reproducibility Preclinical trial (large animal) Technical challenge (rodents) Anatomical difference (vascular supply, large airway, lack of the interface with air) Anatomical difference (large airway, vascular supply) Usually recipient-derived epithelium Use of model (example) Examine responses in small airway epithelium of the donor High through-put studies Tissue remodelling, angiogenesis, epithelial regeneration (syngenic) The role of the epithelium Table 1-1. Advantages, disadvantages, and examples of model studies 33

53 Investigation of mechanisms and therapeutic strategies using animal models of OB Allorecognition Animal and human studies suggest the pivotal role of the alloimmune response in the initiation of obliterative airway disease (OAD). The best described is heterotopic tracheal transplant models of rodents. The immune-mediated mechanism of OAD is well-described as only allografts, not isografts, show prominent lymphocytic infiltration followed by OAD development (126). The alloimmunemediated mechanisms of OAD were further clarified by experiments using severe combined immunodeficiency mice lacking B and T lymphocytes these did not develop OAD after receiving total MHC-mismatched grafts (139, 140). Recent elegant studies using rodent heterotopic tracheal transplant models are aiding in the elucidation of detailed alloimmune-mediated mechanisms of OAD. It appears that allorecognition in OAD involves both direct and indirect recognition of donor MHC molecules in a redundant manner. Minor antigens and self antigens may also be involved. a. Direct allorecognition Direct allorecognition is the recognition of donor-derived intact MHC molecules by recipient lymphocytes as a consequence of the structural similarity between allogenic and self MHC molecules. A high frequency (1 10 %) of T cells are reported to directly recognize foreign MHC molecules expressed on donor cells (141). The allo-mhc class I antigens are recognized by recipient CD8 + T cells, whereas the allo-mhc class II antigens are recognized by recipient CD4 + T cells. The priming of naïve alloreactive T cells results from their interaction with donor-derived antigen-presenting cells (APCs) that can provide costimulatory signals (13). Szeto et al. demonstrated the necessity of donortype APCs in the development of OAD by reconstituting donor tracheal grafts with recipient-type APCs preoperatively by bone marrow transplant this resulted in significantly reduced OAD after transplantation (142). Richards et al. demonstrated the critical role of direct recognition of donorderived class I MHC molecules by recipient CD8 + T cells by adoptively transfusing T cell receptor transgenic CD8 + T cells that specifically recognize donor class I MHC molecule (143). 34

54 b. Indirect recognition of MHC-derived peptides Indirect allorecognition is a classic form of antigen recognition, in which recipient T cells can recognize a processed peptide derived from donor mismatched MHC molecules displayed by recipient APCs in the context of recipient MHC molecules. Increasing evidence suggests that the indirect recognition pathway plays more important roles in chronic rejection than the direct recognition pathway, which is considered to play a major role in acute rejection (13, 141). Using human leukocyte antigen (HLA)-A2 transgenic mice as donors in murine-mhc matched combination, Smith et al. demonstrated that indirect recognition of processed MHC peptides induces OAD and the presence of anti-hla-a2 antibodies (144). Furthermore, from (HLA)-A2 transgenic mice donors, CD4 -/- recipients and, to a lesser degree, CD8 -/- recipients showed delayed OAD development when compared with wild-type recipients, suggesting that both CD4 + and CD8 + T cells contribute to indirect allorecognition following rejection (145). Antigen processing and presentation on MHC II molecules by APCs are blocked intracellularly in H2- DMα -/- mice. MHC II -/- and H2-DMα -/- recipients receiving a wild-type trachea in total MHCmismatched transplantation showed significantly delayed OAD development when compared with wild type or MHC I -/- recipients, suggesting the importance of indirect allorecognition through MHC II- CD4 + dependent pathway (146). In the same report, the authors concluded that MHC I is an important alloantigen that is recognized through the indirect allorecognition pathway based on the following observations: (i) MHC II -/- allografts placed into wild-type recipients did not show diminished rejection. (ii) OAD was reduced in MHC II -/- and H2-DMα -/- recipients. Kelly et al. also implied the relative importance of MHC I as an alloantigen, observing significantly delayed OAD in single class IIdiscrepant but not MHC I-discrepant combination compared with total MHC mismatched transplantation (140). c. Recognition of minor antigens The minor histocompatibility antigens (mhags) are donor-derived foreign peptides to recipients 35

55 other than MHC molecules. The mechanism of mhag recognition is primarily the same as indirect recognition processed peptides are presented to recipient T cells by recipient APCs. mhags were originally recognized by their ability to induce skin graft rejection in MHC-matched mice.(147) The clinical importance of human mhags has been established in the development of graft-versus-host disease (GVHD) in recipients of MHC-matched bone marrow transplant (148). mhags can be presented by recipient MHC class I or II molecules. Grafts from MHC I/II -/- transgenic mice were eventually rejected in a mouse heterotopic tracheal transplant model, indicating that mhags sufficiently cause OAD (146). Richards et al. demonstrated that a single donor mhag (ova albumin) recognized by recipient CD4 + T cells causes OAD equally well to fully MHC-mismatched mouse heterotopic tracheal transplantation using mhag-specific T-cell-receptor-transgenic CD4 + T cells adoptive transferred into recipient mice (149). Although this experiment might have overemphasized the role of mhag by adoptive transfer of transgenic T cells, Higuchi et al. demonstrated delayed but significant development of OAD by day 90 in a single mhag-mismatched combination using a mismatch in the polymorphic H13 allel, the product of which is known to be presented in the context of H2D b MHC I. Furthermore, pre-sensitization of recipient mice by this antigenic peptide accelerated OAD development whereas pre-tolerization abrogated the CD8 + cytotoxic-t-cell-mediated response (150). Qu et al. transplanted tracheae of FVB/N mice human epithelial glycoprotein-2 is expressed specifically on epithelial cells subcutaneously into non-transgenic syngenic mice and showed that mhag-directed epithelial injury causes progressive abnormality of the epithelium and mild OAD (151). Richards et al. suggested that Y-chromosome-derived HY antigen can induce OAD in a mouse subcutaneous model (149). Collectively, minor antigen mismatch is capable of inducing OAD in animal models and the significance is detectable if MHC is matched. Notably, minor antigen mismatch can be a sensitive tool for detecting a contributing factor to graft rejection by avoiding too aggressive MHC-mismatched rejection. For example, the significance of Toll-like receptor signalling on graft survival was able to be detected in HY-mismatched skin graft 36

56 transplantation (152), but not in MHC-full mismatched settings (153). (See Chapter 8 for discussion of the use of HY-mismatched transplantation to detect the influence of the milieu in which a tracheal graft is placed). d. B-cell mediated allorecognition Detection of the anti-hla antibody and its relation to BOS development has been reported (9, 14, 18). Surprisingly, however, reports of antibody-mediated immune responses in animal models are limited. When HLA-A*0201-transgenic tracheal allografts were transplanted into Rag1 -/- mice treated with the anti-hla class I mab, OAD developed by day 90, suggesting that the anti-hla antibody-mediated immune response can induce OAD (154). That animal studies have not focused on antibody-mediated allograft rejection or failed to detect its significance (145) may be explained by the significant contribution of T-cell mediated immune responses without immunosuppression. This is one of the areas where further animal investigation is necessary to understand the mechanisms of OB/BOS after lung transplantation. e. Recognition of self-antigen Although self-antigen is not alloantigen by definition, its recognition by the immune system is an emerging mechanism in allograft rejection after organ transplantation. It has been indicated that myosin and heat shock proteins can be targets of the immune response during cardiac and skin graft rejection, respectively (155, 156). Yasufuku et al. demonstrated that induced oral tolerance to type V collagen can prevent OB-like lesions in the orthotopic rat lung transplant model from Fisher 344 to Wistar Kyoto (19). Type V collagen is a highly conserved protein located in peribronchiolar and perivascular connective tissue in the lung. This and other studies suggest that lung allograft rejection involves both allo- and autoimmune responses, and graft destruction during rejection may expose allograft-infiltrating T cells to potentially antigenic epitopes in type V collagen ( ). Another possibility is that the danger signal (160) provoked by transplant-induced injury breaks down the anergy state of lymphocytes against the potentially antigenic self-antigen. Taking recent evidence in 37

57 human studies into consideration (20, 21, 24), self-antigen is likely to be an important mechanism contributing to OB/BOS after lung transplantation Costimulatory signals a. The CD28-B7 pathway T-cell activation requires costimulatory signals like CD28-B7 interactions. In the absence of costimulation, T cells that encounter antigens either die by apoptosis or enter a state of unresponsiveness called anergy. Cytotoxic T lymphocyte-associated protein (CTLA)4 is a receptor expressed by T cells that binds to B7 molecules expressed by APCs resulting in negative regulation of immune responses. CD28 and CTLA-4 share common ligands but have opposite effects. The role of B7/CD28 pathway has been examined using CTLA4-Ig, which is a fusion protein that inhibits T cell activation by blocking B7-CD28 interaction. Injection of CTLA4-Ig in a mouse heterotopic tracheal transplant model moderately inhibited epithelial loss and fibrosis development in allografts by day 35 (161). Adenovirus-mediated CTLA-Ig gene transfer prevented OAD in a heterotopic rat allograft tracheae (162). A single injection of CTLA-Ig was reported to delay OAD significantly in a rat heterotopic tracheal transplantation and this effect of CTLA4-Ig was demonstrated to be mainly due to the blockade of CD28/B7-2 via a mutant form of CTLA4-Ig (163). b. The CD40-CD40L pathway Activated CD4 + T cells express CD40L (CD154), the ligand of CD40 expressed by macrophages and B cells. CD40-CD40L interaction plays multiple important roles in effector function of adaptive immune responses including stimulation of macrophages to produce IL-12 and enhancement their phagocytotic capacity, and stimulation of B cells in antibody class switch and their differentiation into plasma cells. Rumbley et al. demonstrated the relative importance of CD40-mediated signalling in a mouse heterotopic tracheal transplant model using CD28 -/- and CD40L -/- mice (164). CD40L -/- recipients showed complete preservation of airway epithelium and no evidence of OAD in allografts by 5 weeks whereas CD28 -/- recipients showed slight protection by 2 weeks but the lumen was completely 38

58 obliterated by 5 weeks. They also suggested that the significant impact of CD40 is independent of B cells by using B-cell -/- mice. The effect of CD40-CD40L interaction was also examined using anti-cd40l monoclonal antibody (MR-1) in a mouse heterotopic tracheal transplant model. Although OAD was not observed by day 28, moderate cellular infiltration, epithelial metaplasia, and a minimal narrowing of the tracheal lumen emerged by day 90 and OAD developed by day 120 (165). Chalermskulrat et al. later reported that the combination of donor-specific transfusion of splenocytes and a short course of anti-cd40l antibody markedly prolonged the lumen patency and survival (>100 days) in a mouse heterotopic model (166). It is noteworthy that the blocking of costimulatory signals has been intended to test tolerance induction and that early administration of antibodies or viral vectors resulted in measurable plasma concentration of target proteins only for the first several weeks. Although early costimulatory pathway blockade alone did not induce tolerance in these animal models (161, 165), the theoretical benefit of silencing innate-immune-mediated activation of alloimmune responses remains attractive. Further investigation will be necessary in a more relevant clinical setting such as in combination with immunosuppression Suppression of alloimmune responses a. Conventional immunosuppression The effect of immunosuppressive agents has been tested extensively in animal models of OB. The tested agents, dose regimen and results in rodent heterotopic models are well summarized in a review by Hele et al. (167). In short, 1) immunosuppressive agents including cyclosporine, rapamycin and leflunomide are generally effective as a single treatment to prevent OAD in rodent models if treatment is started on transplantation whereas deoxyspergualin or mycophenolate mofetil (MMF) alone may be less effective (168, 169); 2) the preventive effect of cyclosporine on OAD is dose-dependent (168, 170); 3) delayed start of treatment (e.g. day 7 or 14 after transplantation) is less effective; and 4) the effect of immunosuppression tends to be lost in the long term (e.g. >50 days) (169, 171). 39

59 In a pig subcutaneous model, the combination of cyclosporine, azathioprine and methylprednisolone delayed OAD development but total obliteration was observed around day 30, whereas the combination of cyclosporine and MP with a rapamycin-derived macrolide, everolimus, prevented OAD development until day 90 (131, 132). Alho et al. from the same group reported that the superior effect of the immunosuppression regimen including everolimus was related to reduced epithelial cell apoptosis (172). Studies using primary cultured lung fibroblasts obtained from lung transplant recipients suggested that everolimus and MMF are the most potent anti-fibroproliferative drugs that are effective at safe concentrations in vivo (55). From a clinical viewpoint, the effectiveness of delayed treatment (i.e. treatment for developing or established OAD) and long-term effectiveness are of special importance. The only exception of effective delayed treatment reported is that of everolimus used on a mouse subcutaneous model from day 14 to 28 (173), which might be explained by the blocking effect of everolimus on a calciumindependent signalling pathway in lymphocytes and inhibitory effects on fibroblast proliferation (174, 175). The limited observation period of animal models might be giving the impression that immunosuppressants work in animals but not in humans. This impression, however, is not correct. It is likely that even in animal models preventive effects of immunosuppression are gradually overcome in the long term by multiple mechanisms that compensate or circumvent alloimmune responses (169, 171). On the other hand, it would not be true either that immunosuppression is ineffective in human OB/BOS; it is reasonable that immunosuppression delays the development of OB in humans, too. However, it does not perfectly suppress the development of OB in the long term, resulting in as high as 50% of prevalence after 5 years (3), even though years of observation are unrealistic in animal studies. b. Tolerance induction Tolerance induction would be one of the ultimate solutions for rejection-related graft injury. Extensive efforts have been made in this area in organ transplantation and some encouraging results have been reported. However, these results are often not reproducible in different MHC matching or different 40

60 models. Significant steps are still required to realize true tolerance in clinical lung transplantation and to prevent OB completely. As discussed earlier, costimulatory blockade alone does not induce tolerance. These costimulatory blockades may prevent OAD more efficiently in combination with other strategies. Two studies have demonstrated the synergistic effects of CTLA4-Ig combined with a new immunosuppressive agent, FTY720, which induces T-cell apoptosis and sequestration of circulating mature lymphocytes (176, 177). Mixed haematopoietic chimerism is a state in which bone marrow haematopoietic stem cells from two genetically different animals coexist. Mixed haematopoietic chimerism reconstituted in lethally irradiated rats 4 weeks before transplantation have been reported to induce donor-specific tolerance in orthotopic lung transplantation (178) and heterotopic rat tracheal transplantation (179). The maximum observation periods were 317 and 150 days, respectively, with no evidence of rejection or GVHD. Tolerance induction is an ultimate goal of transplant immunology. Conversely, Suematsu et al. showed that pretreatment-induced chimerism (injection of donor splenocytes and bone marrow cells under immunosuppression) does not preserve allograft trachea in a murine model (180). Since mixed chimerism has been demonstrated to induce the most robust tolerance status, this line of investigation is highly important to realize transplant tolerance (181) Inflammatory mediators in alloimmune responses Soluble mediators, such as cytokines and chemokines, play important roles in innate and adaptive immune responses as well as in bridging from innate to adaptive immunity and amplifying alloimmune responses. a. Chemokines Our laboratory has demonstrated upregulation of Th1-cytokines in rat heterotopic allografts (182), where a beta-chemokine, monocyte chemotactic protein (MCP)-1 was also upregulated at day 3 and 41

61 RANTES reached its peak by day 21. In contrast, MIP-2 upregulation was observed in the early posttransplant period and was not restricted to allografts alone, which is considered to reflect transplantrelated alloantigen-independent stimuli such as ischemia that are present in both allografts and isografts. We further demonstrated that neutralizing antibody against RANTES infused to allografts decreased the number of CD4 + infiltrating cells and preserved lumenal patency in a rat heterotopic tracheal transplant model whereas RANTES infusion in isografts did not induce OAD, even though CD4 + cell migration was increased (183). Farivar et al. also demonstrated the importance of MCP-1 and RANTES by d14 in a rat heterotopic tracheal transplant model using anti-mcp-1 and anti-rnates antibodies (184). These experiments suggest that some chemokines are upregulated as a part of an innate immune response and may function as cofactors of the alloimmune response and following OAD development. Belperio et al. demonstrated the role of MCP-1-CCR2 signalling in the trafficking of mononuclear phagocytes (185). Increased MCP-1 expression paralleled mononuclear cell recruitment and CCR2 expression whereas CCR2 / recipients or wild-type mice treated with neutralizing anti-mcp-1 antibody showed significantly reduced recruitment of neutrophils, mononuclear phagocytes, and OAD development. Interestingly, the lymphocyte infiltration was not reduced under these conditions, supporting the importance of innate immunity for alloimmune response and following OAD development. Belperio et al. also described the effects of CXCL9, CXCL10, and CXCL11 on the recruitment of CXCR3-expressing T cells, NK cells, mononuclear phagocytes, and subsequent OAD development (186). They also described the roles of the CXCR2/CXCR2 ligands pathway in early neutrophil infiltration using neutralizing anti-cxcr2 antibody and CXCR2 -/- mice (96). Interestingly, in the same study, they also demonstrated another role of this pathway in late vascular remodelling independent of neutrophil recruitment. b. Proinflammatory cytokines 42

62 Among proinflammatory cytokines, the role of TNF-α in OAD development has been well described. Blocking of TNF has been shown to delay epithelial damage significantly and OAD development in total MHC-mismatched allografts. Alho et al. demonstrated that preoperative treatment with neutralizing anti-tnf antibody attenuated inflammation and delayed epithelial loss and fibrosis formation at early time points. The beneficial effect, however, was lost by day 28 in a pig heterotopic airway model (187). Aris also showed that the blocking of TNF-α by its soluble receptor significantly delays epithelial injury and OAD in mouse heterotopic allografts, although the effect was gradually lost after completing the scheduled treatment by day 21 (188). Farivar demonstrated similar results using anti-tnf-α antibodies or a TNF-α translational inhibitor, RDP-58, in rat subcutaneous allografts (189). On the other hand, Smith et al. used anti-tnf-α antibody in a mouse model from HLA-A2 transgenic donor mice. Treated animals showed no evidence of OAD by day 60 and significantly delayed the development of anti-hla-a2 antibody whereas untreated controls showed full OAD development by day 28 (190). Smith et al. also showed a moderately protective effect of anti-il-1 antibody in a mouse subcutaneous model from HLA-A2 transgenic mice (190). Conversely, they did not observe any beneficial effects associated with anti-interferon-γ antibody treatment. Supporting this report, Batirel et al. demonstrated that the absence of interferon-γ does not protect the allograft from OAD when using interferon-γ -/- mice as the donor or recipient (191). Endothelin (ET)-1 was identified as a vasoconstrictive peptide (192), which can be released by various cell types in the lung including endothelial, epithelial and smooth muscle cells, in response to various inflammatory stimuli including proinflammatory cytokines such as TNF-α and IL-1β. ET-1 stimulates further production of inflammatory cytokines by macrophages and amplifies inflammatory and Th1 immune responses (193). Tikkanen et al. demonstrated that a nonselective ET-1 receptor antagonist, bosentan, significantly delays epithelial loss, reduces IL-1 β and IL-2 immunoreactive cells, and attenuates OAD in a rat heterotopic model, emphasizing the important role of ET-1 as an inflammatory 43

63 mediator in OAD (194). c. Complements Complements are circulating effector proteins of innate immunity that mediate acute inflammatory responses such as hyperacute allograft and xenograft rejection, and stimulate the humoral adaptive immune response. Complement receptor type 1 has the inhibitory potential to block C3 and C5 convertases of the classic and alternative pathways of the complement system. Recombinant human soluble complement receptor type 1 treatment started preoperatively and continued for 7 days in combination with suboptimal dose of cyclosporine was reported to decrease epithelial necrosis, neutrophil infiltration and OAD in a rat heterotopic model (195). d. Arachidonic acid derivatives Prostaglandins, thromboxanes and leukotrienes are arachidonic acid derivatives that can elicit a wide variety of physiologic reactions. Leukotriene B4 has been shown to have potent chemotactic activity for effector T lymphocytes mediated by its receptor, BLT1. Medoff et al. demonstrated the important role of leukotriene B4-BLT1-mediated T cell trafficking in OAD development in a mouse heterotopic model (196). In this experiment, allogenic tracheae transplanted into BLT1 -/- mice or wild-type mice receiving the BLT1 inhibitor CP 105,696 showed less collagen deposition and OAD development. Conversely, Paivaniemi et al. described the expression of cyclooxygenase-2, a cytokine-inducible isoform of the first enzyme in the pathway of arachidonic acid conversion into prostaglandins and thromboxanes. In pig subcutaneous allografts, cyclooxygenase-2 expression was observed in macrophages and fibroblasts during the inflammatory response and fibroblast proliferation, although the causal relationships are yet to be studied (197). e. Nitric oxide (NO) The role of NO in OAD development is controversial because of the multiple pathways and byproducts of NO formation. Endothelial NO synthase (enos) produces small amounts of NO and maintains vascular homeostasis by relaxing vascular smooth muscle and inhibiting neutrophil and 44

64 platelet adhesion (198) whereas the inducible form of NOS (inos) is induced by pathological stimuli such as acute rejection and infection it produces larger amounts of NO and contributes not only to the killing of exogenous organisms but also to vascular collapse in septic shock, impaired hypoxic vasoconstriction, and tissue injury (199). NO production also results in the generation of nitrogencentered free radicals, in particular the highly cytotoxic reactive oxygen species peroxynitrite (ONOO - ) (200). Kallio et al. suggested a protective role of NO in OAD by showing the association of L-arginine treatment with delayed OAD development and a marked shift from the Th1 to Th2 cytokine response in a rat heterotopic tracheal transplant model (201). Although this experiment did not elaborate upon the roles of inos and enos, or the role of peroxynitrite, Minamoto et al. clarified their roles in OAD development using transgenic mice. In a double lumen mouse tracheal transplant model, tracheal allografts transplanted into inos -/- recipients, but not enos -/- recipients, showed attenuation of the inflammatory response and resulting submucosal fibrosis. Treatment of wild-type recipients with a selective inos inhibitor showed similar results with inos -/- recipients (136). Salminen et al. (202) and Romanska et al. (203) observed inos and peroxynitrite formation in pig and rat heterotopic models, respectively. Both reported common observations, after the early post-transplant ischemic phase, that isografts showed inos expression but not peroxynitrite formation whereas allografts showed high expression of both inos and peroxynitrite in fibroblasts over the fibroproliferative process. Naidu et al. demonstrated the role of peroxynitrite in OAD development more directly using a peroxynitrite decomposition catalyst, FP-15, in a rat heterotopic model. Allograft recipients treated with FP-15 showed reduced nitrotyrosine formation, preservation of respiratory epithelium, limited inflammation, and reduced OAD (204). f. Anti-inflammatory mediators IL-10 inhibits the production of inflammatory cytokines and chemokines by macrophages, neutrophils, lymphocytes and other cells (205). IL-10 downregulates MHC II expression and costimulatory 45

65 molecules on APCs (206), inhibits the Th1 cytokine response and cellular immunity (207). Murine IL- 10 administered at day 5 in a recombinant form or by adenovirus-mediated intramuscular gene transfer led to reduced mononuclear cell infiltration and prevention of OAD, although treatment at day 0 or 10 was ineffective, emphasizing the relatively narrow therapeutic window period (208). Shoji et al. tested Sendai-virus-mediated murine IL-10 gene transfection targeting graft epithelium preserved by temporal immunosuppression in a mouse heterotopic model with successful prevention of OAD (209). In rat heterotopic allografts, neutralizing anti-il-10 antibody was reported to accelerate OAD whereas recombinant IL-10 delayed it (210). Heme oxygenase-1 (HO-1) is an intracellular enzyme that degrades heme into carbon monoxide (CO), biliverdin, and Fe 2+ (ferritin) (211). HO-1 is induced by oxidative and other stresses, and has antiinflammatory and anti-apoptotic effects, although the mechanism mediating HO-1-induced cytoprotection remains elusive. Visner et al. demonstrated the protective role of HO-1 in OAD development in a mouse heterotopic model (212). They first observed increased HO-1 staining in macrophages and mesenchymal cells at day 21. Removal of HO-1 activity by using a metalloporphyrin, tin protoporphyrin or with HO-1 -/- mice as donors or recipients accelerated epithelial loss and fibrosis development in allografts. Interestingly, the use of HO-1 -/- mice reduced IL-10 expression whereas IL- 10 -/- grafts showed reduced expression of HO-1, thus supporting recent reports of reciprocal regulation between IL-10 and HO-1 (213, 214). Minamoto and Harada et al. suggested that the cytoprotective effect of HO-1 is mediated by carbon monoxide (CO) and its regulation on inos-mediated NO production. Enhanced expression of HO-1, by cobalt protoporphyrin or sublethal levels of exogenous CO, decreased the binding activity of nuclear factor kappa B (NF-κB) to the inos promoter region and inos expression, and reduced OAD whereas zinc protoporphyrin, a competitive inhibitor of HO-1, increased OAD (215). 46

66 Alloantigen-independent pathways In parallel with a large body of clinical evidence demonstrating the contribution of alloantigenindependent factors to OB/BOS (e.g. infection and aspiration), animal models of OB have demonstrated the potential impacts of alloantigen-independent factors on the development of OAD. These factors are considered to elicit innate immunity and potentiate alloimmune responses through costimulatory molecules, inflammatory mediators and other pathways. a. Chemical injury Papaverine, the alkaloid in Sauropus plants and the chemical responsible for human toxicant-induced OB, was used to create a rat model of OB not related to the alloimmune response. Lee et al. demonstrated that non-specific airway inflammation caused by activated charcoal enhances OAD development in allografts using a rat orthotopic transplant model. Trans-tracheal injection of activated charcoal into native lungs resulted in slow, progressive airway injury whereas underimmunosuppressed allografts subjected to activated charcoal developed prominent OB (216). b. Brain-death related lung injury Peri-transplant lung injuries, such as donor brain death and ischemia-reperfusion injury, may increase the risk for OB/BOS. Using orthotopic lung transplantation from Fischer 344 to Wistar Kyoto rats, Zweers et al. demonstrated the significant impact of brain death on long-term lung function and histology (217). Six hours of brain death induced the influx of polymorphonuclear cells and macrophages and expression of vascular cell adhesion molecule-1 in the donor lungs on retrieval. At day 100 after transplantation, the lung function was significantly decreased compared with allografts from living donors, and chronic rejection was histologically confirmed. c. Viral infection Infection can induce innate and adaptive immune responses, both of which can theoretically enhance the alloimmune response. Animal studies have focused on virus infection and following OAD development. Using a rat orthotopic lung transplant model, Winter et al. observed the effect of a 47

67 respiratory viral infection (Sendai virus) on the chronic damage of iso- and allografts (218). Allografts with viral infection established after transplantation showed more severe characteristics of OB-like lesions than the allografts without infection whereas infected isografts did not show OB-like chronic pathologic changes despite some airway damage soon after infection. These results suggest that respiratory viral infection enhances rejection and OB-like lesions in allografts. Reichenspurner et al. demonstrated that allografts but not isografts are affected by systemic rat cytomegalovirus (CMV) infection in recipient rats and rapidly develop OAD in a rat heterotopic tracheal transplant model (219). Koskinen et al. also showed that both acute and chronic rat CMV infection in recipients accelerates OAD in allografts (220). In this experiment, they also demonstrated the enhanced expression of the MHC II molecule, PDGF, and its receptor in epithelial cells that appeared to mediate the alloimmune response and the subsequent fibrotic process. Tikkanen et al. demonstrated the significant role of PDGF in CMV-enhanced OAD development using a selective PDGF receptor tyrosine kinase inhibitor, CGP (221). Using the same models of acute and chronic rat-cmv infection in rats, they further demonstrated that prophylactic treatment of CMV infection before transplantation reduced the impact of CMV infection on OAD development more effectively than anti-cmv treatment at day 5 (222). CMV infection also enhanced expression of cytokines like TNF-α, IFN-γ and IL-2, which are known to enhance the alloimmune response Anti-inflammatory strategies In addition to the anti-inflammatory agents or neutralizing antibodies tested as part of the blocking experiments discussed earlier, several new anti-inflammatory agents have been tested in OAD. Naidu et al. demonstrated that a broad-spectrum chemokine inhibitor, NR , delays epithelial destruction and OAD development by day 14 in a heterotopic rat model (223). Activation of p38 kinase, one of the major mitogen-activated protein kinases, has been shown to be directly responsible for the NF-κB-dependent transcriptional activation of cytokine, chemokine, and adhesion molecule genes in multiple models of inflammatory injury. Farivar et al. demonstrated that the specific inhibition of p38, FR167653, which inhibits the nuclear translocation of NF-κB, reduces TNF-α transcription and delays 48

68 OAD by day 14 in a rat heterotopic model (224). Poly (ADP)-ribose synthetase (PARS) is a proteinmodifying and nucleotide-polymerizing nuclear enzyme. Overwhelming oxidative and nitrosative stress induces PARS over-activity and culminates in cellular death via apoptotic mechanisms. Moreover, Farivar et al. demonstrated that a PARS inhibitor, INO-1001, reduces TNF-α gene expression and cellular death by day 14 in a rat heterotopic model (225). Phosphodiesterase-4 inhibitors have been established for anti-inflammatory treatment, particularly of pulmonary diseases. Using a rat heterotopic model, Roth-Eichhorn et al. tested the effect of the phosphodiesterase-4 inhibitor, rolipram. Compared with cyclosporine, rolipram was less effective in protecting the epithelium and preventing OAD, but was more effective in reducing cell proliferation. Combination of rolipram and cyclosporine exerted the most efficient graft protection by day 60 (171). Because OB is acknowledged as a multifactorial disorder in which alloantigen-independent pathway is also important, effective anti-inflammatory strategies may well improve the outcome of lung transplantation. The clinical effectiveness of azithromycin, anti-reflux surgery, and statins support the importance of anti-inflammatory strategies (see Chapter 1). Interestingly, airway neutrophilia and upregulated IL-8 have been demonstrated to be associated with several clinically and biologically important findings including bile-acid aspiration (33), the effectiveness of azithromycin (70), upregulation of IL-17 (71), bacterial colonization in airways (32), and upregulation of MMPs in airways (35, 226, 227). Thus, neutrophils, chemokines recruiting neutrophils (e.g. IL-8) and biological products of neutrophils (e.g. reactive oxygen species and proteinases) are attractive therapeutic targets necessary for further investigation Epithelial injury The existence of intact epithelial lining has been considered to be an important regulator in airway fibrosis. On the other hand, recent studies have revealed that damaged or stimulated epithelium has significant impact on mesenchymal overgrowth through various signalling pathways beyond the simple barrier-against-fibrosis concept. 49

69 a. Loss of epithelium The important regulatory role of the epithelial lining in airway fibrosis has been successfully demonstrated in a series of studies. Tracheal isografts denuded of epithelium by protease digestion develop OAD similar to allografts in a rat heterotopic tracheal transplant model whereas reseeding of epithelial cells significantly reduced airway obliteration (44, 228). Retransplantation of rat tracheal allografts into the donor strain or donor-recipient F1 cross demonstrated the existence of a point of no return between days 7 and 14 after transplantation. At this point, donor-derived epithelium is lost due to the alloimmune response and is unable to prevent OAD thereafter even if immune reactivity is removed by retransplantation (229). This result was confirmed by another group (230). These experiments have successfully demonstrated that epithelial loss alone is enough to develop OAD independent of the alloimmune response. Orthotopic tracheal transplant models have also demonstrated the importance of epithelial re-growth in OAD prevention. Orthotopically transplanted rat allograft tracheae did not show obliteration by day 60, although subepithelial infiltration of T-cells and myofibroblasts were observed (133). Similarly, orthotopically transplanted mice allograft tracheae were not obliterated and repopulated with epithelium (134), which was later confirmed to be of recipient origin (137) and the repopulation of which is facilitated by immunosuppression (231). Fernandez et al. demonstrated that airway epithelium is the primary target of rejection and OAD development. BALB/c donor tracheae were first transplanted orthotopically into C57BL/6 recipients. The MHC-mismatched allografts were harvested at day 28, when the BALB/c-derived grafts were covered by C57BL/6-derived epithelium. The grafts subcutaneously retransplanted into BALB/c rejected C57BL/6-derived epithelium and developed OAD whereas those transplanted into C57BL/6 did not develop OAD (232). This result is also supported by a similar study reported by Genden et al (233). b. Epithelial-mesenchymal crosstalk Cross-talk between the airway epithelium and mesenchymal fibroblasts potentially play an important 50

70 role in the development of OAD and OB. Studies on airway remodelling in asthma have demonstrated important regulatory functions of airway epithelium on the underlying tissue including fibroblasts, forming the epithelial-mesenchymal trophic unit (234). Prostaglandin E 2 secreted from airway epithelial cells is a potent inhibitor of fibroblast proliferation (235, 236) and TGF-β-dependent differentiation into myofibroblasts (43). Conversely, injured human bronchial epithelial cells produce growth factors such as TGF-β, PDGF, fibroblast growth factor (FGF) and ET-1, which could potentially activate fibroblasts (237). Interestingly, antibody-mediated epithelial stimulation and growth factor production has been suggested to be a potential mechanism underlying fibroproliferation of OB (18). MMP-2 from bronchial epithelial cells has also been demonstrated to stimulate the proliferation of subepithelial fibroblasts (238). Moreover, recent reports have suggested that damaged or stimulated airway epithelial cells could transform into fibroblast-like cells (epithelial-mesenchymal transition) and contribute to fibrosis including OB after lung transplantation (49, 239, 240). The fibrogenic potential of cyclosporine has been extensively studied in kidney transplant recipients. In vivo and in vitro studies indicate that cyclosporine might have a stronger fibrogenic effect than another calcineurin inhibitor, tacrolimus (241, 242). In lung transplantation, a similar fibrogenic potential of cyclosporine has been suggested in vitro. Cyclosporine is reported to decrease the proliferation and viability of human tracheobronchial epithelial cells in a dose-dependent manner and increase the production of TGF-β (45), induce the production of the proinflammatory cytokines IL-6 and IL-8 in cultured airway epithelial cells in a dose dependent manner (243) and enhance fibroblast proliferation in vitro (244). Hence, immunosuppression itself might modulate the epithelial-mesenchymal interaction in vivo Epithelial protection and regeneration Many studies have reported protection of epithelium as a consequence of anti-inflammatory or immunosuppressive treatment. However, few studies have focused on epithelial protection or regeneration. Our preliminary study suggested that anti-apoptotic agent treatment using Z-BAD can 51

71 attenuate OAD in a mouse heterotopic model (unpublished data). We have recently described bonemarrow-cell mediated modulated epithelial recovery (245), although the detailed mechanism is still under investigation. It has been recently reported that chemokine, CXCL12-mediated recruitment of circulating progenitor epithelial cells is necessary for the re-establishment of a normal pseudostratified epithelium after airway injury in a mouse subcutaneous syngenic tracheal transplantation (92). Our group has recently demonstrated the potential in using mesenchymal stem cells that contain epithelial progenitor cells to facilitate epithelial regeneration in airways in a naphthalene-induced mouse airway injury model (93). It is likely that strategies to regenerate airway epithelium could become a potent therapeutic arm in OB in future Fibroproliferation Fibroproliferative responses in OAD are the result of a complex process involving fibroblast activation, myofibroblast differentiation, excessive extracellular matrix (ECM) production and angioneogenesis. This process is mediated by numerous soluble and insoluble factors including cytokines and growth factors. Aris et al. observed that potentially fibrogenic growth factors including TGF-β, PDGF, FGF, insulin-like growth factor-1 and ET-1 reach their peaks 2 to 6 weeks after transplantation in mouse heterotopic allografts (188). These profibrotic factors are likely released from various sources including macrophages, lymphocytes and damaged epithelial cells (237) during inflammatory and immune responses and contribute to the development of OAD synergistically with the loss of epithelium. Recent studies have also revealed the involvement of the angiotensin system and proteolytic enzymes such as MMPs in the fibroproliferation of OAD. a. Profibrotic growth factors TGF-β is a potent immunosuppressive and fibrogenic cytokine ( ). We have demonstrated that TGF-β1 is expressed mainly by mononuclear cells in rat heterotopic allografts at day 7 and the blocking of TGF-β signalling by soluble TGF-β receptor type III started at day 5 after transplantation prevented OAD (249). Smad3 is an essential component of TGF-β-mediated signal transduction (250). 52

72 Ramirez et al. observed TGF-β1 expression in fibroblasts and accumulated connective tissue at later stages in a mouse heterotopic model. Disruption of Smad3 gene expression using Smad3 -/- mice as recipients resulted in failed organization of intralumenal fibrosis despite fibroblast accumulation (251). Growth factors such as PDGF and FGF are known to stimulate the proliferation of fibroblasts (252). Al-Dossari et al. demonstrated that mouse tracheal isografts can develop OAD independent of the alloimmune response if PDGF or basic FGF is locally delivered from day 0 after operation (253). The authors implied the possibility that the growth factor treatment started before re-epithelialization of isografts induced OAD in this model. The work of Kallio et al. supports the role of PDGF in fibroproliferation. A protein tyrosine kinase inhibitor, CGP 53716, selective for PDGF receptor, markedly inhibited fibroproliferation and airway occlusion in a rat heterotopic model. However, CGP had no effect on inflammatory cell infiltration and proliferation, or on the intensity of immune activation, suggesting that blocking PDGF by CGP is not immunosuppressive but inhibitory to fibroproliferation (254). ET-1 may have fibroproliferative as well as proinflammatory effects. Takeda et al. demonstrated that transfection of the ET-1 encoding gene in untransplanted rat airways can induce hyperplastic connective tissue plaque in the alveolar duct and small conducting airway, indicating histologically distinctive OB (255). Macrophages are considered to play a number of important roles in OAD. Macrophages are a major source of cytokines, growth factors and other inflammatory mediators. Phagocytosis of apoptotic cells by macrophages induces TGF-ß1 (256), which is involved in the resolution of inflammation as well as fibrosis. Macrophages can function as APCs bridging the innate and adaptive immune response. Macrophage infiltration in airway grafts usually peak from early inflammation to the initiation of fibrosis (139). Depletion of recipient macrophages by gadolinium chloride (GdCl3) was reported to reduce OAD significantly in rat subcutaneous allografts (257). 53

73 b. Th2 cytokines When inflammation is sustained during tissue remodelling, the Th2 cytokine response might be important in promoting excessive fibroproliferation. For examine, IL-13 is known to have potent effects on fibroproliferation and collagen production (258), and myofibroblasts in human OB lesions have been demonstrated to express IL-13 receptor α1 (40). IL-13 has been suggested to play a critical role in human and experimental OB (40, 41). In contrast to the destructive Th1-mediated immune response, the Th2 response is likely to promote fibroproliferative inflammation (258). Current immunosuppression is generally considered to target the Th1 response more effectively than the Th2 response (83). Given that the Th2 as well as the Th1 response and alloantigen-independent inflammation are important in the pathogenesis of OB, immunomodulatory therapy, such as using IL- 10, that potentially regulates them would be an attractive therapeutic option (110, 258). c. Angiogenic factors promoting fibrosis Vascular endothelial growth factor (VEGF) is a growth factor involved in vascular remodelling and other various physiological phenomena such as inflammation and tissue repair. Krebs et al. demonstrated dual roles of VEGF in OAD. Intra-graft VEGF overexpression by adenoviral VEGF gene transfer increased early epithelial cell proliferation and regeneration in allografts but increased microvascular remodelling, lymphangiogenesis and lumenal occlusion in a rat heterotopic model (259). Conversely, VEGF receptor inhibition decreased early epithelial regeneration, microvascular remodelling, lymphangiogenesis, intra-graft trafficking of CD4 + and CD8 + T cells and the degree of lumenal occlusion. Interestingly, a combination of VEGF overexpression and inhibition of another fibrogenic growth factor, PDGF, preserved epithelium and prevented OAD in allografts. Although chemokines have been identified as chemotactic cytokines that recruit leukocytes in response to inflammatory stimuli, their roles are not limited to the inflammatory response. In OAD, chemokines may contribute to late tissue remodelling and fibrosis. Belperio et al. demonstrated that the CXCR2/CXCR2 ligand axis induces vascular remodelling and angiogenesis independent of the 54

74 CXCR2-mediated neutrophil recruitment in a mouse heterotopic model (96). Furthermore, in other fibrosis models, chemokines have been reported to contribute to fibrosis through the recruitment of fibroblasts (260) and circulating fibrocytes (47), differentiation into myofibroblasts (261), and epithelial-mesenchymal crosstalk (236). It is possible that chemokines bridge the immune response and fibroproliferation as well as innate and adaptive immunity. d. The angiotensin system Increasing evidence suggests that local renin angiotensin system plays a prominent role in tissue repair and fibrosis. This system has been found to promote fibroblast proliferation and extracellular matrix deposition after acute lung injury partly through TGF-β (262, 263). It has been demonstrated that treatment with the angiotensin converting enzyme (ACE) inhibitor captopril, started 5 days before or 1 day after transplantation, prevents OAD without preserving epithelium by day 21 in a rat heterotopic model, identifying an important effect on the fibroproliferative response (264). Although captopril treatment 5 days post-transplantation was not effective, further preliminary study using the angiotensin-1 receptor blocker losartan, started on post-operative day 1 or 5, showed OAD prevention only when losartan was administered on day 5 (265). This paradox might be partly explained by the complex function of TGF-β1, which possesses immunosuppressive properties during early inflammation but later exerts fibroproliferative effects. e. The MMP-TIMP systems MMPs are a major group of enzymes targeting virtually all structural extracellular matrix proteins, chemotactic molecules, latent growth factors, cell surface receptors and adhesion molecules ( ). Tissue inhibitors of metalloproteinase (TIMPs) are endogenous regulators of MMPs. Involvement of the MMP-TIMP system has been implicated in a variety of diseases. Fernandez et al. has demonstrated the importance of MMP9, but not MMP2, in fibrosis development using MMP2 -/- and MMP9 -/- mice in a mouse heterotopic tracheal transplant model. In this study, they demonstrated the preventive effect of doxycycline, a weak broad-spectrum MMP inhibitor, on OAD (269). Inaki et al. suggested an 55

75 important role of MMPs 2 and 14 in the destruction of epithelium and fibrosis development in a rat heterotopic model (270). Conversely, absence of TIMP-1 in either the donor trachea or the allograft recipient was reported to reduce OAD at day 28 in a mouse heterotopic model (271). These results demonstrate that the relationship between fibroproliferation and the MMP-TIMP system is not simply explained by the balance of TIMPs over MMPs as advocated previously (Figure 1-9, left). Instead, the MMP-TIMP system is now considered to regulate tissue remodelling including fibroproliferation and reversal of fibrosis in a complex way (Figure 1-9). Research in this area will be particularly important in helping to understand the mechanisms of tissue remodelling in OB after lung transplantation. 56

76 MMP<TIMP Decreased ECM degradation MMP>TIMP Independent of Increased ECM degradation ECM degradation GF-BP MMP Matrikines GF MMP GF activation by GF- BP degradation MMP Matricrine MMP Inactivation of GF/CK by degradation Proliferation of interstitial cells Increased ECM production F I B R O S I S Figure 1-9. Theories explaining the relationships between fibrosis and MMP-TIMP balance. Overbalanced effects of TIMPs on MMPs were believed to induce fibrosis (left). However, recent studies demonstrate that the overbalanced activities of MMPs can also induce fibrosis (right). 57

77 Anti-fibroproliferative strategies In addition to the strategies inhibiting or manipulating growth factors, the angiotensin system and the MMP-TIMP system (as discussed earlier), new anti-fibrotic agents have been tested. Tranilast, an anti-allergic agent, is known to inhibit the release of histamine, IL-1β, TGF-β1, and PDGF from various cells and is currently used to treat keloids and hypertrophic scars. The benefit of tranilast in intractable granular airway obliteration has been reported (272). Tranilast treatment was also reported to significantly decrease fibroproliferative airway changes in rat heterotopic allografts by day 28 (273). Pirfenidone (5 methyl-1-phenyl-2-(1h)-pyridone) is a novel anti-fibrotic agent that causes no immunosuppression, downregulates the production of TGF-β and collagen in vitro, and inhibits fibroblast proliferation. Pirfenidone treatment started at the time of transplantation prevented OAD with mild change of epithelium in allografts by the end of observation at day 28 in a mouse subcutaneous model, although delayed treatment was less effective (274). Conversely, similar pirfenidone treatment did not prevent OAD by day 28 in rat heterotopic allografts (275). It is encouraging that there are studies underway that target effector points other than the alloimmune response in OB, which could synergistically prevent OB with current immunosuppression and might treat fibrosis beyond the point of no return (229) beyond which immunosuppression is usually ineffective. Most reported strategies, however, have limited therapeutic window periods that are usually within 10 days after transplantation in heterotopic tracheal transplant models for various reasons (249, 264). In other words, they are effective only if initiated before fibrosis becomes apparent in the histology. Anti-fibroproliferative strategies targeting existing fibrosis are novel, requiring significant investigation in future. 58

78 1.3. The concept of tissue remodelling integrating repair, regeneration and aberrant remodelling: the rationale Tissue remodelling integrating repair, regeneration and aberrant remodelling Tissue remodelling: not equivalent to fibrosis but a dynamic process in tissue Over the last two decades, lung transplantation has become the mainstay of treatment for terminal pulmonary disorders. However, in contrast to increased success in early post-transplant management, chronic graft dysfunction remains a major challenge. Chronic graft dysfunction after lung transplantation is represented by obliterative bronchiolitis (OB), a pathological finding, and its clinical correlate, bronchiolitis obliterans syndrome (BOS). Pathologically, OB/BOS is characterized by inflammation and fibroproliferative obliteration of small airways (86, 276). The purpose of this chapter is to highlight tissue remodelling as an important paradigm in the mechanism determining the fate of transplanted pulmonary allografts: whether they maintain the normal tissue architecture and function or succumb to OB. The concept of tissue remodelling rationalizes the following studies in this dissertation. Tissue remodelling indicates replacement of cellular and extracellular components of a tissue with the same or different types of cells and matrix. The terms tissue remodelling and fibrosis have often been used almost interchangeably, to describe a range of processes in the lung where the architecture is altered up to, and including, the formation of permanent, fixed scar tissue (277). However, tissue remodelling is originally a concept proposed in bone where there is continuous replacement of the osseous matrix to maintain its homeostasis. As similar homeostatic replacement of the tissue components also occur in the lung, the term tissue remodelling also includes such a homeostatic process after lung transplantation. Tissue remodelling is also an important event in normal wound healing. After tissue injury either alloantigen-dependent or alloantigen-independent injury damaged cells and matrix must be removed and replaced by the pre-existing cell types and matrix to regenerate the original tissue architecture. 59

79 Since transplanted lungs are constantly exposed to aerogenous injurious stimuli, including transplantrelated dangers (e.g. ischemia, opportunistic infection, and rejection), the appropriate repair and regeneration mechanisms are highly important for maintaining the normal tissue architecture and function. Tissue remodelling could become aberrant for many reasons as will be discussed in the following sections. Resulting abnormal tissue architecture (e.g. airway fibrosis) has deteriorating effects on the function of the organ (e.g. obstructive pulmonary dysfunction). Moreover, the abnormal tissue architecture and function can exacerbate ongoing tissue injury (see Chapter 1). This type of tissue remodelling should be described as aberrant tissue remodelling as the original definition of tissue remodelling encompasses other types of physiological remodelling processes. Finally, tissue remodelling may be bidirectional and potentially include fibrosis regeneration and reestablishment of the pre-existing tissue architecture. In this dissertation, we will discuss the regression of fibrosis after its establishment. Studies in the liver have clearly demonstrated that regression of fibrosis occurs under certain conditions (278, 279). A similar paradigm is emerging in pulmonary fibrosis (277). Although fibrosis regression has not been demonstrated in allograft airway fibrosis of OB after lung transplantation, it is theoretically possible. We herein propose fibrosis regression as a potentially important component of tissue remodelling after lung transplantation. To summarize, tissue remodelling after lung transplantation includes 1) homeostatic tissue remodelling; 2) tissue repair and regeneration after injury; 3) aberrant tissue remodelling; and 4) fibrosis regression (tissue regeneration after aberrant tissue remodelling). In lung transplantation, repair, regeneration, and aberrant remodelling are important issues. The mechanisms of tissue remodelling will be discussed Transition from injury to remodelling Alloantigen-dependent rejection and alloantigen-independent insults (e.g. infection, aspiration, etc.) are 60

80 major mechanisms of tissue injury in transplanted lungs. Because episodes of acute rejection and viral infection are generally controlled by specific pharmacological interventions, the process of tissue remodelling then initiates upon resolution of inflammation. Rejection and alloantigen-independent insults such as chronic bacterial infection and silent aspiration may progress insidiously without obvious clinical episodes. In such cases, tissue injury and tissue remodelling likely occur simultaneously. Removal of apoptotic and necrotic cells by scavenging macrophages is one of the critical initial steps in tissue remodelling. On ingestion of apoptotic cells, macrophages secrete growth factors that are important in tissue remodelling, such as TGF-β (256). Damaged or activated epithelial cells and fibroblasts secrete proteinases such as MMPs, subsequently contributing to degradation of the existing extracellular matrix (ECM). These cells also secrete growth factors and cytokines which collectively create a microenvironment for tissue remodelling (88). In parallel with the clearance of apoptotic/necrotic tissue, an initial fibrotic reaction plays an important role in tissue remodelling. It is important to note that this fibrotic reaction is not necessarily pathological, but an integral step in the recovery of normal tissue architecture. Subepithelial fibroblasts that are usually inhibited by epithelial cells through prostaglandin E2 (235) transform into myofibroblasts under the influence of growth factors including TGF-β. The myofibroblast is an active phenotype of fibroblasts expressing alpha smooth muscle actin (α-sma) and is capable of producing large amounts of ECM molecules (e.g. fibronectin, collagens) (280). As tissue injury often accompanies destruction of the subepithelial basement membrane and derangement of subepithelial ECM, the provisional ECM serves as a scaffold for the ensuing migration of epithelial cells (88). Upon the provisional ECM scaffold, epithelial cells start to migrate, demonstrating squamous metaplasia, which is considered to be a transitional condition leading to epithelial regeneration (88). Airway epithelial progenitor cells such as basal (and perhaps ciliated) cells de-differentiate into a flat morphology and contribute to squamous metaplasia (88). In general, MMP-7 and MMP-9 play 61

81 important roles in the detachment of airway epithelial cells from the basement membrane, allowing for their migration (281). Epithelial progenitor cells may also be recruited from the systemic circulation (92), which is supported by the observation of epithelial chimerism in lung-transplant recipients (138) Regeneration vs. aberrant remodelling Tissue remodelling reaches an important juncture wherein the tissue may undergo successful regeneration to recover its normal architecture or aberrant remodelling resulting in abnormal architecture. During tissue regeneration, epithelial cells that show squamous metaplasia transform into or are replaced by a pseudostratified columnar phenotype (Figure 1-10A), while in aberrant remodelling, squamous metaplasia persists (Figure 1-10B). If recovery of the epithelium is completed, subepithelial myofibroblasts would disappear presumably through apoptosis. If epithelial recovery is incomplete, or if other myofibroblast-stimulating factors continue to exist, myofibroblasts persist and keep secreting ECM, leading to the development of subepithelial fibrosis protruding into the airway lumen (compare Figure 1-10 A and B). Subepithelial fibrosis may eventually result in total obliteration of the airway lumen. In many cases, myofibroblasts persist even after the completion of lumenal obliteration. 62

82 A Tissue injury Airway epithelium Regeneration - crosstalk Squamous metaplasia Epithelial (temporary) progenitor Fibrotic reaction Fibroblast Myofibroblast ECM reorganization Myofibroblast Apoptosis B - Aberrant remodeling Factors impairing epithelial regeneration Tissue injury Repetitive injury Immunosuppression Epithelial progenitors Abnormal GF, cytokine, MMP profile Ischemia (IR injury, chronic vascular rejection?) crosstalk Fibrotic reaction Factors contributors to excessive fibrotic reaction Repetitive tissue injury Myofibroblast resistance to apoptosis Myofibroblast progenitors (e.g. fibrocytes, EMT) Abnormal GF, cytokine, MMP profile (Th1 Th2 cytokine shift) Angiogenesis Ongoing injury Abnormal Epithelium Mechanical barrier Protective protein Infection Ciliary clearance Aspiration Airway obstruction Epithelial denudation Tissue Injury Myofibroblasts persist Aberrant remodeling Autoantigen exposure (e.g. collagen V) + Persistent immune attack Antibody-mediated rejection Memory lymphocytes Angiogenesis Lymphoid neogenesis Figure Normal tissue repair and regeneration vs. aberrant tissue remodelling after injury. (A) Tissue remodelling leads to regeneration and recovery of normal tissue architecture. Stable epithelial cells maintain airway homeostasis in part by regulating submucosal fibroblast activation. After epithelial injury, fibroblasts differentiate into myofibroblasts and support epithelial migration by reorganizing the scaffold of the extracellular matrix (fibrotic reaction). Epithelial cells temporarily exhibit squamous metaplasia for migration. Epithelial cells and myofibroblasts interact through cytokines, growth factors, chemokines, adhesion molecules and matrix metalloproteinases (epithelialmesenchymal crosstalk). Then, epithelial cells re-differentiate into a pseudostratified columnar phenotype (epithelial regeneration). Circulating epithelial progenitors may contribute to epithelial regeneration. Upon epithelial regeneration, myofibroblasts undergo apoptosis and matrix reorganization is completed. (B) Tissue remodelling leads to aberrant remodelling, abnormal tissue architecture and further injury. Epithelial regeneration is impaired by various factors, which may result in epithelial denudation. Fibrotic reaction becomes excessive under the influence of various factors and myofibroblasts persist. Fibrous tissue is supported by angiogenesis, while continuous recruitment of inflammatory and immune cells may result in lymphoid neogenesis. Abnormal tissue architecture is vulnerable to further tissue injury mediated through various pathways such as abnormal epithelium, aberrant remodelling, and persistent immune attacks. ECM, extracellular matrix; GF, growth factor; MMP, matrix metalloproteinase; IR, ischemia-reperfusion; EMT, epithelial-to-mesenchymal transition. 63

83 Reversibility of aberrant remodelling and required conditions The pathological paradigm has been that tissue fibrosis or architectural changes in the tissue after aberrant remodelling is the terminal, fixed result of tissue damage that is not capable of regeneration and is fixed with no possibility of return to the pre-existing structure (282). However, accumulating evidence suggests that this traditional perspective may not always be true. Reversibility of fibrosis has been best demonstrated in the liver. In animal experiments, automatic regression or resolution has been demonstrated in liver fibrosis induced by chemical injuries (e.g. CCl 4 ) (283). In clinical settings, spontaneous resolution of alcoholic liver fibrosis has been known to occur. Moreover, advance in the treatment of hepatitis virus B and C resulted in regression of fibrosis in some patients (278, 279). These phenomena may well reflect the potent regenerative capacity of the liver, which is capable of recovering its size even after 80% of its total volume is resected. Interestingly, fibrosis regression in the lung has also been suggested by some authors (284). It has been widely acknowledged that cryptogenic organizing pneumonia (bronchiolitis obliterans organizing pneumonia) accompanies fibrotic changes in peripheral airways and alveoli and that the condition responds to steroid treatment (277). Fibrosis of OB is a disorder that affects lung transplant recipients so selectively that the clinical experience is much more limited compared with that of pulmonary fibrosis. Moreover, the patchy distribution of OB lesions in airways makes transbronchial biopsies much less reliable than that of pulmonary fibrosis and consequently, reversal of fibrosis has not been histologically confirmed in posttransplant OB. However, some reversibility in declined pulmonary function has been demonstrated in association with recent treatments with azithromycin and anti-reflux surgery (69, 74). In contrast to the conventional paradigm in pathology, these observations of histological or functional reversibility of fibrotic conditions are highly encouraging. If the mechanisms of tissue remodelling underlying the reversal of fibrosis are revealed, we might be able to harness the regenerative process and develop novel therapeutic strategies for OB treatment after lung transplantation. 64

84 Potential mechanisms of tissue remodelling that promote or resist fibrosis regression: the rationale of the study Although the mechanisms of tissue remodelling in OB after lung transplantation are largely unknown, in particular the reversal aspect, studies in the liver and lung have revealed more or less common mechanisms. Wallace et al. have recently suggested that the resolution of fibrosis in the lung could occur when alveolar structure is retained, fibroblasts are inactivated, and lumenal collagen is reabsorbed. Cryptogenic organizing pneumonia, acute hypersensitive pneumonitis and resolving acute respiratory distress syndrome are examples of reversible changes in the alveolar structure. However, fibrosis persists and becomes irreversible scar tissue when alveolar structure is lost, fibroblasts are continuously activated, and progressive matrix deposition occurs (277). Interestingly, resistance of myofibroblasts to apoptosis has been reported in non-resolving acute respiratory distress syndrome (ARDS) (285). In OB lesions after lung transplantation, the tissue structure or framework of airways appears to be relatively well preserved (86, 286). On the other hand, fibroblast activity represented by the existence of myofibroblasts appears to be continuous in OB lesions (36, 286). Since myofibroblasts are generally capable of producing an excessive amount of collagen and other ECM molecules, and contribute to fibroproliferative tissue remodelling (37), persistence of these cell types and their ECM production may well be an important factor in the tissue remodelling of OB after lung transplantation. In the resolution of fibrosis, the role of MMPs and their endogenous inhibitors, TIMPs have been investigated extensively in the liver. Although the roles of these molecules in fibrosis are complex and sometimes contradictory (287) (also see Figure 1-9), the involvement of the MMP-TIMP system appears to have a significant impact on tissue remodelling. While there are many reports showing upregulation of MMPs in fibrosis including pulmonary fibrosis (288, 289) and OB after lung transplantation (226, 227), spontaneous upregulation of MMPs (290) and gene transfer of collagenolytic MMPs (291, 292) have been reported to lead to fibrosis regression in the liver. Activity 65

85 of some MMPs has been reported to induce apoptosis in myofibroblasts in the liver (293) and the lung (294). Thus, modulation of the MMP system (either upregulation or downregulation) is an important angle to consider when investigating tissue remodelling. Another important factor might be the existence of continuous underlying injury that might make the tissue remodelling process more resistant to fibrosis regression. In the liver, fibrosis regression is always associated with the treatment or resolution of injurious processes (e.g. viral hepatitis, alcohol intake) (278, 279). However, the detailed mechanisms that link chronic fibrosis and ongoing tissue remodelling are not as clearly understood as those at the initiation of the fibrotic process triggered by tissue injury. One of the advantages of transplantation in the research of fibrosis is the relatively obvious cause of fibrosis (i.e. transplantation) in contrast to many idiopathic fibrous disorders. In our animal models, like the heterotopic tracheal transplantation, allograft airway fibrosis occurs reliably in association with alloimmune-mediated tissue injury, which is theoretically continuous as alloantigen persists long-term after transplantation. Thus, our animal models provide an excellent opportunity to investigate the relationship between tissue remodelling and underlying continuous injury associated with fibrosis Models to be used to investigate tissue remodelling and immune responses after lung transplantation Based on the rationale, we will investigate the mechanisms of obliterative bronchiolitis after lung transplantation, with specific focus on tissue remodelling. As mentioned earlier, heterotopic tracheal transplant models are ideal in their reproducibility of fibrosis and ongoing alloimmune responses. In addition, in most of the experiments in the present study, we take advantage of intrapulmonary tracheal transplantation. Because the lung is continuously exposed to aerogenous antigen stimuli, the lung appears to have evolved unique innate and adaptive immune systems, particularly at the mucosal surface of airways somewhat analogous to that of the intestine (295, 296). As such, alloimmune responses in the lung might be qualitatively and quantitatively different from that in other anatomical 66

86 places such as the subcutaneous tissue. Moreover, the components of tissue remodelling, like fibroblasts, may well be different between subcutaneous tissue and the lung (297). Thus, we consider a rat intrapulmonary tracheal transplant model to be an appropriate animal model to test our hypotheses and to answer our specific questions as described next Hypothesis and Study Aims Hypothesis Tissue remodelling plays important roles in obliterative bronchiolitis after lung transplantation. Continuous immune responses are important mechanisms that support ongoing tissue remodelling Specific study aims We test the hypothesis along the following specific study aims: 1. To investigate the process of tissue remodelling at different stages of obliterative airway disease in the lung. 2. To examine the roles of matrix metalloproteinases in the migration of myofibroblast precursors contributing to tissue remodelling and fibrosis development. 3. To examine the mechanisms of tissue remodelling in established allograft airway fibrosis. 4. To investigate the contribution of chronic allograft rejection to the maintenance of fibrosis. 5. To investigate remodelling in the lymphoid tissue in the lung as a possible mechanism of chronic allograft rejection. 6. To investigate the effector function of lymphoid tissue in the lung in allograft airway rejection. 67

87 Chapter 2 Materials and Methods 68

88 2.1. Human tissue samples of BOS lungs and normal control lungs Human tissue samples of BOS lungs were obtained from patients during their follow-up at the Toronto General Hospital. Thirteen lung grafts were obtained from patients with established diagnoses of BOS at the time of retransplantation (4 males and 9 females with a mean age of 38.0 ± 3.8 years; a mean transplantation period of 52.9 ± 10.4 months; original diagnosis of 3 pulmonary fibrosis (including one secondary to scleroderma), 3 cystic fibrosis, 3 bronchiectasis, 3 bronchiolitis obliterans secondary to bone-marrow transplantation, and 1 chronic obstructive pulmonary disease). Fifteen normal lung tissue samples were obtained from patients with early stage lung cancer. The human study is approved by the Research Ethical Board of University Health Network, University of Toronto Animal Models Animals Male Brown-Norway (MHC class I, RT1A n ) and Lewis rats (RT1A l ) were purchased from Charles River (Wilmington, MA), and male Wistar-Furth rats (RT1A u ) were purchased from Harlan Sprague Dawley (Indianapolis, IN). F1 rats were bred from male Brown-Norway and female Lewis rats in the Toronto General Research Institute. All animals received care in compliance with the Guide to the Care and Use of Experimental Animals formulated by the Canadian Council on Animal Care. The experimental protocol was approved by the Animal Care Committee of the Toronto General Research Institute Tracheal transplant models of OB and combination with orthotopic lung transplantation Intrapulmonary tracheal transplantation was conducted as described previously (127). When combined with concurrent orthotopic lung transplantation in Chapter 4, the left lung of a Lewis rat was transplanted into a male F1 rat as described previously (298), and then, after reperfusion, intrapulmonary transplantation of a trachea of Wistar-Furth rat in the Lewis-derived left lung was conducted. 69

89 When combined with subsequent orthotopic lung transplantation in Chapter 7, after 7 or 28 days, the left lung of the intrapulmonary tracheal transplant recipient was used as a donor lung of orthotopic left lung transplantation. The recipient animals of orthotopic lung transplantation were sacrificed 28 days after lung transplantation (i.e. day 35 or 56 after the initial intrapulmonary tracheal transplantation, n = 4 for each group). In another experiment in Chapter 7, following intrapulmonary tracheal transplantation, the left lung of the initial intrapulmonary tracheal transplant recipient was used as a donor lung of orthotopic left lung transplantation at day 28. After reperfusion of the left lung graft, a secondary tracheal graft was implanted in the left lung adjacent to the first intrapulmonary tracheal graft using the same technique as regular intrapulmonary tracheal transplantation. As a control, a tracheal graft from the same strain as the secondary intrapulmonary tracheal graft was implanted in the back through a subcutaneous tunnel created from the site of thoracotomy. The recipient animals of orthotopic lung transplantation were sacrificed 28 days after the concurrent transplantation (i.e. 56 after the initial intrapulmonary tracheal transplantation, n = 4 for each group) Epithelial denudation of a tracheal graft To compare obliterative fibrosis that developed in isografts and allografts in Chapter 6, we modified the techniques of enzymatic epithelial denudation reported by Qu et al (228). The optimal concentration of protease (6 U/mL of Pritease steptomyces griseus, p8811, Sigma-Aldrich) dissolved in Dulbecco s Minimum Essential Media (DMEM, Invitrogen) was applied to the tracheal lumen of either a Brown-Norway or Lewis rat. Both ends of the trachea were tied, and then the trachea was incubated in DMEM for 1.5 hours at 37ºC, followed by flushing the lumen of the graft for 5 times with 0.9% saline. After the procedure, airway epithelial cells were removed leaving the subepithelial mucosa. The trachea was subjected to intrapulmonary tracheal transplantation into a Lewis rat either as an isograft or allograft In vivo drug Treatment 70

90 Broad-spectrum MMP inhibitors, SC080: N-hydroxy-1-(2-methoxyethyl)-4-[4-[4- (trifluoromethoxy)phenoxy]phenyl]sulfonyl]-4-piperidinecarboxamide monohydrochloride, (a gift from Dr. Teresa Sunyer, Pfizer, Inc., St. Louis, MO) and MMI270: N-hydroxy-2(R)-[(4- methoxysulfonyl)(3-picolyl)-amino]-3-methylbutaneamide hydrochloride monohydrate (a gift from Novartis Pharmaceuticals Canada Inc., Mississauga, Canada) were used. MMP inhibitors were administered under direct visualization of the vocal cords and pharynx, avoiding intubation of the trachea. An immunosuppressant, cyclosporine A (10 mg/kg/d, Novartis) was subcutaneously injected Cell culture Fibrocyte isolation and culture In Chapter 4, peripheral blood mononuclear cells (PBMCs) from Lewis rat blood underwent centrifugation over Lympholyte (Cedarlane laboratories, Burlington, Canada) and 2.5x10 6 cells were plated on a 24-well BD Biocoat TM fibronectin-coated cell culture plate (BD Biosciences, Mississauga, Canada) and cultured in Dulbecco's Modified Eagle's Medium (Invitrogen), supplemented with 20% fetal bovine serum, L-glutamine, penicillin, and streptomycin (Sigma). Non-attached cells were removed at day 2 and continuously cultured until day MMP inhibitor treatment for fibrocytes and wound healing assay Migration of fibrocytes was examined via the standard wound healing assay in Chapter 4 by scraping a monolayer of fibrocytes cultured on fibronectin for 10 days using a pipet tip. To reduce the influence of cell proliferation, fibrocytes were pre-treated with mitomycin C (10 μg/ml, Sigma) for 30 min, followed by a scratch with or without additional treatment with an MMP inhibitor, SC080 or MMI270 (10 nm and 100 nm). At the beginning of the experiment and after 24 hours, images were acquired with a Nikon Eclipse TE200 microscope from the same spots of culture wells and the distance of cell migration was measured with Image J, 1.30v (Wayne Rasbamd, NIH, USA) Fibroblast culture and induction of the myofibroblast phenotype Cultured myofibroblasts were used in Chapters 5 and 6. Primary rat pulmonary fibroblasts were 71

91 cultured from adult Lewis rat lung that was flushed with normal saline, excised, and minced in DMEM with 10% FBS. The lung tissue was incubated with collagenase IV (0.5 mg/ml, Sigma) at 37 C for 1 h. After removing tissue debris through a cell strainer and washing with DMEM with 10% FBS, the single cell suspension was plated on a 6-well plate (BD Biosciences, Mississauga, Canada). After incubating at 37 C for 1 h, cells that were not adherent to the plate were gently removed and adherent cells were continuously cultured in DMEM with 10% FBS. Growth of fibroblasts was confirmed by morphology and collagen I staining (Abcam). Fibroblasts were lifted using trypsin/edta (Invitrogen) and transferred. Fibroblasts from passage 3 to 6 were used for the following experiments SC080 treatment for myofibroblasts Cultured myofibroblasts were treated with SC080 in Chapter 5. The rat pulmonary fibroblasts were lifted and 50,000 cells were transferred into a well of a 12-well plate (BD Biosciences) and cultured on a glass cover slip. The phenotype of the myofibroblast was induced by stimulating cells with 10 ng/ml of human recombinant transforming growth factor beta 1 (hrtgf-β1) for 48 hours. The myofibroblasts were then cultured in DMEM with 2% FBS with filter-sterilized SC080 solution (4 μm) or its vehicle for 48 hours. Culture medium was replaced every 12 hours Myofibroblast culture and co-culture with T lymphocytes Myofibroblasts were co-cultured with PBMC-derived T cells in Chapter 6. Primary culture of pulmonary fibroblasts and induction of the myofibroblast phenotype were conducted as described above. Peripheral blood mononuclear cells (PBMCs) were isolated from normal rat peripheral blood using Lympholite (Cedarlane Laboratories, Burlington, Canada), and T cells were isolated using MACS cell separation (anti-t-cell OX52 MicroBeads, Miltenyi Biotech, Auburn, CA). Myofibroblasts (5 x 10 4 cells) plated on a 12-well plate were continuously cultured at the bottom of a Transwell (0.4 μm-pored, BD Biosciences). Isolated T cells (20 x 10 5 cells) were added to the top or bottom of the Transwell, or myofibroblasts were continuously cultured alone. Cells were cultured in RPMI1640 with 1% FBS for 24 hours for gene expression analysis or for 96 hours for cell 72

92 viability/differentiation analysis PBMC labelling and injection into allograft recipient animals PBMC-derived myofibroblast differentiation was tracked in Chapter 4. After isolating PBMCs from normal Lewis rats as described above and incubating over night in a 6-well plate (1 x 10 7 cells/well), cells that were not sticking to culture wells were gently removed. The cells remaining were lifted by Trypsin/EDTA, which yielded about 1% of cells. These cells were labelled with PKH26 (Sigma) following manufacturer s instructions. After the staining procedure, more than 95% of the cells were viable in trypan blue staining. These stained cells were administered to Lewis rats that had received intrapulmonary transplantation of a BN trachea. Each allograft recipient animal received PKH26- labelled cells at post-operative day 7, 10, and 13 (3 x 10 5 cells/animal/time point) (n = 3). Labelled cells that were not injected into animals were continuously cultured for 14 days as a control of cell differentiation in vitro. The recipient animals were sacrificed at day Histological, immunohistochemical, and immunofluorescence assessments Histology The middle section of a graft trachea was fixed in 10% formalin, and the other sections were snap frozen. Five micron paraffin sections were used for standard haematoxylin and eosin (H&E) staining and picrosirius red (PSR) staining for collagen. PSR staining was also observed under polarized light to depict thick type I collagen and thin type III collagen as yellow-orange fibers and green-white fibers, respectively (299). Quantification of total collagen (300) and computerized morphometry for occlusive fibrosis (249) were performed as previously described. To quantify the occlusion ratio of a tracheal graft lumen in animal models of OB, the outlines of the occlusive lesion and the whole tracheal lumen interior to the cartilage were traced using Photoshop 6.0 (Adobe, San Jose, CA). To calculate epithelial preservation ratio, total epithelial lining and preserved epithelial lining were traced using Photoshop. The binary images were then morphometrically quantified using Image J, 1.30v (Wayne Rasbamd, NIH, USA). 73

93 Immunohistochemistry and Immunofluorescence for tissue Frozen sections (embedded in Tissue-Tec O.C.T. compound, Sakura Finetek USA, Torrance, CA) of 6 μm thickness and paraffin-embedded sections of 5 μm thickness were used, respectively. For paraffin sections, deparaffinization was followed by antigen retrieval in boiling 0.01 M citrate buffer (ph 6) for 20 minutes. All sections were blocked and incubated overnight at 4 C with primary antibodies listed in Table 2-1. For immunohistochemistry, VECTASTAIN ABC Kits (Burlingame, CA) were used following the manufacturer's instructions. For immunofluorescence labelling, FITC-conjugated donkey anti-goat antibody (1:50, Jackson ImmunoResearch Laboratories, West Grove, PA) or Alexa-Fluor 488- or 555-conjugated goat anti-rabbit antibody (1:300, Invitrogen Canada Inc., Burlington, Canada) were used as secondary antibodies. For α-smooth muscle actin (α-sma), sections were incubated in Cy3-conjugated mouse anti-human α-sma antibody (1:100, Sigma-Aldrich, St. Louis, MO). For double labelling, the first staining of procollagen or MMPs was followed by α-sma staining. Immunofluorescence labelling was followed by Hoechst nuclear counter stain (1:1000, Invitrogen). Images were acquired with a Nikon Eclipse TE200 microscope or an Olympus FluoView 1000 Laser Scanning Confocal Microscope (Olympus Canada Inc., Markham, Canada) Quantification of lymphoid tissue in the lung In animal experiments in Chapter 7, the size of peribronchiolar lymphocyte aggregates were evaluated using the image of double staining for T cells and B cells. The number of lymphocyte aggregates was manually counted. Lymphocyte aggregates adjacent to the graft were excluded from the calculation. To quantify the size of peribronchiolar and peri-graft lymphocyte aggregates, the outline of a lymphocyte aggregate was traced on an image of immunofluorescence labelling for CD3 and CD79a, and then the binary image was quantified using Image J. To standardize the number and size of peribronchiolar lymphocyte aggregates, the whole lung area was calculated by scanning a whole lung section containing the graft trachea using SNAPSCAN e50 (AGFA Corp. Ridgefield Park, NJ) Immunofluorescence labelling for cells 74

94 For immunofluorescence labelling of cultured cells, cells were fixed with 4% paraformaldehyde for 60 min, permeabilized with 0.1% Triton-X, blocked with 6% goat serum in PBS, and then incubated with primary antibodies of anti-collagen I (Abcam, Cambridge, MA) or α-sma (Sigma), and secondary antibody in the same manner as slide labelling. Cells were washed 3 times with PBS between steps Immunofluorescence with terminal deoxynucleotidyl transferase-mediated dutp nick end labelling (TUNEL) To detect apoptosis of fibroblasts and myofibroblasts, we used immunofluorescence labelling for TUNEL in combination with staining for vimentin or α-sma. Paraffin sections were permiabilized using a microwave in 0.1 M citrate buffer (ph 6.0) for 7 min, or cells cultured on a cover slip were fixed and permeabilized as described above, followed by incubation for 1 hour at 37 C using the In Situ Cell Death Detection Kit fluorescein (Roche Applied Science, Indianapolis, IN) and immunofluorescence staining using Cy3-conjugated anti-vimentin or anti-α-sma antibodies (Sigma) Morphometric myofibroblast viability/differentiation analysis Myofibroblasts cultured with or without T cells were stained for α-sma in Chapter 6. In some experiments, double staining was conducted for α-sma and TUNEL or CD3. In quantification of viable differentiated myofibroblasts, an image for red fluorescence (α-sma) was first binalized with a predetermined optimized threshold using Image J software. After merging this binalized image, the corresponding images for green (CD3) and blue (nucleus) nuclei positive for CD3 (i.e. T cells) and those negative for α-sma were manually erased. Some cells were clumped to show high red fluorescence although they do not show intact morphology of myofibroblasts. To remove these cells out of counting, the binalized red channel was replaced by the original image for α-sma staining and nuclei corresponding to the clumped cells were erased. Finally, the blue channel for nuclei was binalized with a predetermined threshold and the number of nuclei was counted. 75

95 Antigen Host Species Cross Reactivity Company Application Frozen/Paraffin/cell CD3 rabbit human/rat DAKO Frozen/Paraffin CD68 mouse rat Serotec Paraffin Procollagen α1(i) goat rat Santa Cruz Paraffin MMP-2 rabbit human/rat Biomol Paraffin MMP-3 rabbit rat Biomol Paraffin MMP-9 rabbit human DAKO Paraffin MMP-9 rabbit rat Biomol Frozen MMP-13 rabbit rat Biomol Paraffin MMP-14 rabbit human/rat Abcam Paraffin Pan MHC I (OX18) mouse rat Serotec Frozen RT1A n (OX27) mouse rat Serotec Frozen Ki-67 mouse rat DAKO Paraffin α-sma mouse human/rat DAKO Frozen/Paraffin/cell collagen I rabbit rat Abcam Cell CD20 mouse human DAKO Paraffin CD79a mouse human/rat Abcam Paraffin PNAd rat human BD Paraffin vwf rabbit human/rat polyclonal Millipore Paraffin CCR7 rabbit human Novus Paraffin MadCAM-1 mouse rat BD Frozen Table 2-1. Antibodies used in immunohistochemical and immunofluorescence assessments. PNAd, peripheral-node addressin; vwf, von Willbrand factor; MadCAM-1, Mucosal addressin celladhesion molecule 1. 76

96 2.5. Flow Cytometry Analysis The following antibodies and appropriate IgG isotype controls were purchased from BD PharMingen (San Diego, CA): FITC-labelled CD3 (clone G4.18, mouse IgG3), PECy5-labelled CD4 (OX-35, mouse IgG2a), PE-labelled CD8a (OX-8, mouse IgG1), Biotin-CD25 (OX-39, mouse IgG1), Streptavidin PE-Cy7. PE-labelled CD45RC (OX22), and PE-labelled or unconjugated RT1A u (OX27) were purchased from Serotec. Single cell suspensions were obtained from fresh tissue specimens by enzymatic digestion using collagenase type IV (0.5 mg/ml, Sigma) and DNAse I (250 U/ml, Sigma) for 1.5 hr at 37 C. After counting, cells were incubated with antibodies for 30 minutes at 4 C as described previously (85).. Cells were analyzed using an FC-500 flow cytometer (Beckmann Coulter Canada Inc., Mississauga, ON) Quantitative Reverse Transcriptase Polymerase Chain Reaction (RT-PCR) Analysis Frozen pieces of tracheae were embedded in Tissue-Tek OCT compound (Sakura Finetek USA. Inc. Torrance, CA) and 20 μm-thickness sections were cut onto glass slides and counterstained lightly with haematoxylin. The tissue interior to the tracheal cartilage was micro-dissected using an 18G needle (Figure 2-1) under stereo microscopic visualization (Leica MZ7.5, Leica Microsystems Inc. Bannockburn, IL) and immediately put in the appropriate buffer from the RNeasy Fibrous Tissue Mini Kit (QIAGEN Inc., Mississauga, Canada). RNA was extracted from sections for each sample following manufacturer s instructions. ABI Reverse transcriptase reaction kit (Applied Biosystems, Foster City, CA) was used for RT reaction and real-time PCR was run on ABI-7900HT (Applied Biosystems). Data were normalized using the method of Vandesompele (301), in which ACTB and GAPD were selected as the most stable set of house keeping genes among 8 tested candidate genes using GeNorm program (free software, Primer sequences are summarized in Table

97 Figure 2-1. A technique of microdissection. (a) A frozen section of an allograft trachea obliterated with fibrous tissue lightly stained with Haematoxylin. (b) The fibrous tissue is stripped off from the tracheal cartilage with an 18G needle. (c) The trachea left behind after dissection and retrieval of the intra-lumenal fibrous tissue. (d) The fibrous tissue retrieved by an 18G needle. 78

98 Gene Forward/Reverse Sequence (5' 3') ACTB Forward AAGGCCAACCGTGAAAAGATG Reverse CACAGCCTGGATGGCTACGT GAPD Forward TCGGTGTGAACGGATTTGG Reverse CAATGTCCACTTTGTCACAAGAGAA α-sma Forward GTGGATCAGCAAACAGGAGTATGA Reverse GAGGGTGGTGGCGTGACTTA COL1a(I) Forward GGGCCTCTCCTCCATTGC Reverse TGGGACGATTCAGCATTGC COL1a(III) Forward GTGGACATTGGCCCTGTTTG Reverse GTTGACGAGATTAAAGCAAGAGGAA MMP-2 Forward TTCCAGGGCACCTCCTACAA Reverse TCCATATTTCTTATCCCGGTCATAG MMP-3 Forward TCCCAGGAAAATAGCTGAGAACTT Reverse ACTGTGAAGATCCGCTGAAGAAG MMP-8 Forward GATTCAGAAGAAACGTGGACTCAA Reverse TGGAGTGAGAGAGTCCCAAGGA MMP-9 Forward ACGACAGCTGACTACGACACAGA Reverse TGGAAATACGCAGGGTTTGC MMP-13 Forward CACCCCAAAACACCAGAGAAG Reverse GGGTGCAGACGCCAGAAG MMP-14 Forward CCCAACGGGAAGACCTACTTC Reverse CCTTCCCAGACTTTGATGTTTTTG TIMP-2 Forward CGTTTTGCAATGCAGACGTAGT Reverse CCCGGAATCCACCTCCTT IFN-γ Forward ATCGCACCTGATCACTAACTTCTTC Reverse TGTGCTGGATCTGTGGGTTGT IL-1β Forward CTAATGCCTTCCCCAGGACAT Reverse TGGCTCTGAGAGACCTG MMP-12 Forward AACACATTTCGTCTCTCTGCTGAT Reverse TTGTCAAGGATGGGTTTTTCACT TIMP-1 Forward, CCACCTTATACCAGCGTTATGAGA Reverse CCGGAAACCTGTGGCATTT CD45 Forward CCCGGGATGAGACAGTTGAT Reverse TGCACACTTGTTCCTGTTTCCT CD3 Forward ATCTTCGCTGAAATCATCAGCAT Reverse GTCTGCTTGTCTGAAGCTCTTGAC CD19 Forward CTATGGCGGCTTTTCTCTATTTTC Reverse GGGAGGCGTCACTTTGAAGA 18S rrna Forward CCTTTAACGAGGATCCATTGGA Reverse TACGAGCTTTTTAACTGCAGCAACT Table 2-2. PCR primers used in this study. 79

99 2.7. Assessment of proteinase activities Assessment of gelatinolytic MMPs We assessed the activity of gelatinolytic MMPs with SDS-PAGE zymography (302) using Novex Tris- Glycine gels containing 0.1% gelatin (Invitrogen), and in situ zymography (303) using MMP in situ Zymo-Film (Fuji Photo Film, Tokyo, Japan) as previously described. In situ zymography was also conducted using FITC-conjugated DQ TM -gelatine (Invitrogen). Fresh frozen sections (8 μm) were first stained with Cy3-conjugated α-sma (Sigma) and Hoechst for 15 min without fixation, and then washed with PBS. The tissue was incubated with or without MMP inhibitor (SC080, 20 μm; EDTA 25 mm; or 1,10-phenanthroline, 1 mm) at room temperature for 1 hour. After removing the solution, DQgelatine mixed with reaction buffer (0.05 M Tris-Hcl, 0.15 M NaCl, 5 mm CaCl2, 200 μm sodium azide, ph 7.6) was applied. Images were visualized in 3 hours Collagenolytic activity assay EnzChek FITC-collagen (Invitrogen Canada, Burlington, Canada) was used following the manufacturer s instructions. Frozen tracheal samples were homogenized in 50 mm Tris-HCl buffer ph 7.6, containing 1.5 mm NaCl, 0.5 mm CaCl 2, 1 μm ZnCl 2, 0.01% Brij 35 and 0.25% (v/v) Triton X After incubation at 37 C for 1 hour, and centrifugation at 4,000 rpm at 4 C for 10 min, the supernatant of the tissue lysate was recovered. After mixing the tissue lysate with DQ-collagen and 6 hours of incubation with or without 1 mm of 1,10-phenanthroline (pan MMP inhibitor, Sigma-Aldrich, St. Louis, MO) using a black half-volume 96-well plate (Nalge Nunc International, Rochester, NY) at 37 C, fluorescence was read using a Gemini XPS (Molecular devices, Sunnyvale, CA) at the emission wavelength 413/488 nm T-cell Proliferation Assay and Mixed Lymphocyte Reaction (MLR) Assay In a T-cell proliferation assay, cells (1x10 5 ) from draining lymph nodes of non-transplanted Lewis rats were stimulated with plate-bound anti-rat CD3 (G4.18; 10 mg/ml) and anti-rat CD28 (JJ319; 2 mg/ml) with doses of indicated drugs (for SC080: 0.5 mm, 5 mm, 50 mm; for cyclosporine A (CsA): 50 ng/ml, 80

100 500 ng/ml). [ 3 H] thymidine was added for the last 18 hours of a 4 day culture. In MLR assays, allograft recipient spleens and mediastinal lymph nodes were harvested, minced and processed for MLR. Spleens cells from Brown-Norway rats (stimulators, 0-4x10 5 cells) were irradiated with 20 cgy and mixed with 1x10 5 responder cells from transplanted Lewis rats (mediastinal lymph nodes or spleen) that received intrapulmonary transplantation of a Brown-Norway trachea 10 days prior with or without SC080 treatment (2.5 mg/kg/d). Cells were cultured in 96-well round bottom plates for 4 days in RPMI media. T-cell proliferation was measured on day 4 by means of an 18 hour pulse with 3 H-thymidine (1 µci) ELISA for TGF-β1 After treatment with SC080, the supernatant of the cell culture was examined for active and latent forms of TGF-β1 using a second generation TGF-β1 ELISA kit (R&D System, Minneapolis, MN) following manufacturer s instructions Statistics To compare two groups, the Student s t-test was used when sample distribution was normal, while the Mann-Whitney U test was used when the sample distribution was not normal. When comparing more than two groups or different time points, one-way analysis of variance (ANOVA) was followed by post hoc Tukey tests. All statistical analyses were performed using JMP 5.0 (SAS Institute Inc., Cary, NC). P values < 0.05 were considered to be significant. 81

101 Chapter 3 Allograft airway fibrosis in the pulmonary milieu: A disorder of tissue remodelling The content of this chapter was published in the American Journal of Transplantation 2008;8(3):

102 3.1. Abstract Obliterative bronchiolitis (OB) is thought to be a form of chronic allograft rejection. However, immunosuppressive therapy is not effective once fibrosis has developed. We hypothesize that disordered tissue remodelling is a mechanism for the pathogenesis of OB. We examined allograft airway fibrosis in an intrapulmonary tracheal transplant model of OB. Allograft airways were completely obliterated at day 21 by fibrotic tissue; however, tissue remodelling continued thereafter, as demonstrated by the change of collagen deposition density, shift from type I to type III collagen, shift from fibroblasts to myofibroblasts, and shift of expression profiles and activities of matrix metalloproteinases (MMPs). We then used a broad-spectrum MMP inhibitor, SC080, to attempt to manipulate tissue remodelling. Administration of the MMP inhibitor from day 0 to day 28 reduced airway obliteration, without inhibiting T cell activation. MMP inhibition from day 14 to day 28 showed similar effects on airway obliteration. MMP inhibition from day 21 to day 35 did not reverse the airway obliteration, but significantly reduced the collagen deposition, type III collagen, and myofibroblasts in the lumen. We conclude that tissue remodelling plays a critical role in the development and maintenance of fibrosis after transplantation. 83

103 3.2. Introduction Obliterative bronchiolitis (OB) is the most common cause of morbidity and mortality in long-term survivors of lung transplantation (304). Although OB is generally accepted to be a form of chronic rejection (305), it is well known that immunosuppressive therapy generally appears ineffective once the fibroproliferative tissue process is initiated (82). A search for other cellular and molecular mechanisms is required. Tissue remodelling is a dynamic process in which extracellular matrix and other tissue components undergo continuous modification. Physiological remodelling is important to maintain tissue homeostasis, whereas pathological remodelling could result in unfavourable architectural changes in tissue and eventual malfunction of the organ. Fibroproliferative processes are typical of pathological tissue remodelling (288). In OB, after the total obliteration of the airway lumen is established, the fibrosis has generally been considered to be static and irreversible (126). In the present study, we hypothesize that the development and maintenance of fibrotic obliteration of small airways after lung transplantation is a dynamic and potentially plastic process. Evidence to support this concept may lead to the development of novel therapies for OB. Matrix metalloproteinases (MMPs) comprise a group of enzymes that are important in physiological and pathological tissue remodelling such as seen in various forms of fibroproliferative disorders (306, 307), including pulmonary fibrosis (288, 289, 308). In order to investigate our hypothesis, we used a broad-spectrum MMP inhibitor, SC080, to interfere with tissue remodelling. Administration of this inhibitor was started at different stages of fibrotic obliteration in an intrapulmonary tracheal transplantation model in rats. Heterotopic transplantation of the grafts has been employed in most reported studies. We previously described a novel intrapulmonary tracheal transplantation model of OB in rats, and demonstrated the connection of the grafts with the pulmonary circulation (127). Although obliterative fibrosis develops with similar kinetics to previously reported tracheal subcutaneous transplant models, the advantage of this model is that the graft is in the pulmonary milieu, 84

104 which may have a significant relevance to the accommodation of implanted lung tissue (128). The pulmonary milieu could also be important for airway remodelling. We found that a broad-spectrum MMP inhibitor is able to modulate tissue remodelling at various time periods in the intrapulmonary tracheal transplant model of OB. The significance of tissue remodelling in the development and maintenance of allograft airway fibrosis is discussed Study Design For the time-course observation, four animals each from the iso- and allograft groups were sacrificed on days 7, 14, 21, and 35. Allograft recipients were also sacrificed on days 49 and 98 (n = 4). The MMP inhibitor study was designed to examine MMPs in different pathological stages of allografts observed in the time-course study (Figure 3-1). Vehicle or a broad-spectrum MMP inhibitor, SC080 (2.5 mg/kg/day; gift from Dr. Teresa Sunyer, Pfizer, Inc., St. Louis, MO) was administered via an orogastric tube to allograft recipients on days 0-7, or 0-28 to cover the initiation phase of fibrosis and after; days for developing fibrosis; or days for established fibrosis, (n = 6 for each group). 85

105 Figure 3-1. Schematics of the study design using an MMP inhibitor. Dynamic changes of cell population, matrix metalloproteinases (MMPs), lumenal obliteration and stages of fibrosis in allografts are summarized. Interventional experiments using a broad-spectrum MMP inhibitor, SC080, were planned to target early inflammation, developing fibrosis, and established fibrosis. 86

106 3.4. Results Establishment of fibrosis in allografts by day 21 To determine the effects of the alloimmune response in airway fibrosis and remodelling, we first histologically compared allografts and isografts in the rat intrapulmonary tracheal transplant model. At post-transplant day 7, isografts showed partially denuded epithelium with little lumenal cell infiltration, whereas allografts showed both denuded epithelium and infiltration of polymorphonuclear cells (Figure 3-2A). At day 14, isografts showed recovery of the epithelium, whereas the epithelium was completely lost with infiltration of fibroblast-like cells and partial luminal obliteration in allografts (Figure 3-2AFigure 3-). At day 21, the allograft lumen was completely obliterated with fibrous tissue. In contrast, the airway epithelium was completely repaired in the isograft (Figure 3-2A). The infiltrating cells in the lumen of allografts were mainly T cells and macrophages, as determined with immunohistochemistry staining with cellular markers (Figure 3-2B). Morphometric quantification demonstrated a significant difference in lumenal obliteration between allografts and isografts after day 14 (Figure 3-2C). Obliterative fibrosis in allografts persisted to day 98 (Figure 3-2D). 87

107 Figure 3-2. Fibrotic obliteration of allograft airways in the lung. (A) In isografts, the epithelial lining was lost at day 7 and recovered by day 21. The airway was not obliterated by fibrotic tissue. In allografts, the epithelium was lost, and the airway was completely obliterated by fibrosis at day 21. (B) Immunohistochemistry for CD3+ T cells and CD68+ macrophages (brown staining). Large numbers of T cells and macrophages in allograft lumen at days 7 decreased over time. (C) Morphometric quantification of obliterative fibrosis demonstrated significant differences between isografts and allografts after day 14 (*p < 0.05; data are expressed as mean ± SEM; n = 4). (D) Fibrosis established in allografts persisted until day 98. Scale bars = 50 μm. Original magnification of insets was 40x. 88

108 Dynamic tissue remodelling in allografts after complete fibrotic obliteration in the airway We then examined the collagen deposition in the lumen of allografts using PSR staining. Observation of PSR staining under regular light revealed collagen at day 14 that further increased gradually thereafter (Figure 3-3A, upper panel). Although the airway was occluded completely after day 21 (Figure 3-2A), the collagen deposition continued to increase and reached a plateau after day 35 (Figure 3-3C). Observation of PSR staining under polarized light is a special technique to demonstrate type I (thick, orange-to-red) and type III (thin, green-to-white) collagens. A shift from type I to type III collagens over time was observed in the lumen of allografts (Figure 3-3A, lower panel). The gene expression of procollagen α1(i) reached a peak at day 21, while that of procollagen α1(iii) peaked at day 35 and remained at high levels thereafter, as determined by quantitative RT-PCR (Figure 3-3D). Both fibroblasts and myofibroblasts are able to produce collagen (37). We used procollagen α1(i) as a marker for fibroblasts (309) and α-sma as a marker for myofibroblasts (37) to determine their expression in the obliterated lumen of allografts. The immunofluorescence staining showed that the cell population switched from procollagen α1(i) positive (green, fibroblasts) dominant at day 14 to α- SMA positive (red, myofibroblasts) dominant at day 21 and thereafter (Figure 3-3B). Semi-quantitative analysis revealed a peak in myofibroblast number at day 21 and sustained high levels thereafter (Figure 3-3E). This persistence of myofibroblasts in allograft airway fibrosis is in contrast to what is seen in normal wound healing where myofibroblasts disappear over time (37). 89

109 Figure 3-3. Dynamic changes of collagen, fibroblasts, and myofibroblasts in allografts. (A) picrosirius red (PSR) stain under regular light demonstrated increases in collagen deposition (red) in the allograft lumen by day 35, which appeared to stabilize thereafter. PSR stain under polarized light demonstrated a mixture of type I (thick, orange-to-red) and type III (thin, green-to-white) collagen fibers. Thin fibers were dominant after day 49. Arrows at day 14 indicate connective tissue in a subepithelial area. (B) Immunofluorescent staining of procollagen α1(i) positive fibroblasts (green) and alpha-smooth muscle actin (α-sma) positive myofibroblasts (red). A shift from fibroblast dominant to myofibroblast dominant staining was noted. (C) Quantification of PSR positive areas demonstrated a significant increase in total interstitial collagen that plateaued at day 35. (D) Real-time RT-PCR for microdissected tissues from allograft lumen demonstrated the peak of procollagen α1(i) gene expression at day 21, whereas gene expression of procollagen α1(iii) was sustained at high levels after day 21. (E) Semi-quantification of myofibroblasts demonstrated a significant increase in the number of myofibroblasts that peaked at day 21. Scale bars = 50 μm. Significant differences (p < 0.05) according to post hoc Tukey tests are represented by * (vs. days 7, 14, and 21), (vs. days 7 and 14), (vs. days 7, 14, and 49), and (vs. day 7). Data are expressed as mean ± SEM; n = 4. 90

110 Dynamic changes of MMPs in allografts We examined MMPs as candidate molecules for mediating tissue remodelling of allograft airways. The SDS-PAGE gelatin zymogram of whole tracheal lysates showed an increase in pro-mmp-9 at day 7 followed by a decrease (Figure 3-4A). Real-time RT-PCR analysis performed on the micro-dissected intra-lumenal tissue of allografts demonstrated peaks in MMP-3, MMP-9, and MMP-13 gene expression at day 7 (Figure 3-4B, top panels). Immunofluorescent labelling demonstrated expression of MMP-3, MMP-9, and MMP-13 in polymorphonuclear cells and other infiltrating cells in the lumen at day 7 (Figure 3-4C), but not after day 14 (data not shown). On the other hand, an increase in active MMP-2 was observed at days 7 and 14 compared to normal trachea and a further increase was observed after day 21 in the SDS-PAGE zymogram (Figure 3-4A). The gene expression of MMP-2 and MMP-14 remained upregulated from days 21 to 35 (Figure 3-4B). TIMP-2, which forms an MMP-2-activating complex with MMP-14 (310), was also elevated in established fibrosis (Figure 3-4B). Immunofluorescent staining localized MMP-2 and MMP-14 mainly in myofibroblasts at day 21 (Figure 3-4C) and thereafter (data not shown). Taken together, these time course studies demonstrated dynamic tissue remodelling before and after airway obliteration, with shift of types of collagens, types of fibroblasts and types of MMPs. We then tested whether these changes are related to the fibrosis by using a broad-spectrum MMP inhibitor. 91

111 Figure 3-4. Dynamic changes in activity and expression of matrix metalloproteinase (MMP) in allografts. (A) SDS-PAGE gelatin zymography using whole trachea lysates demonstrated upregulation of pro-mmp-9 at day 7 and increases in pro- and active MMP-2 after day 21. (B) Real-time RT-PCR analysis of gene expression in microdissected allograft lumen tissues demonstrated higher levels of MMP-3, MMP-9, and MMP-13 at day 7 compared to the other time points (p < 0.001). MMP-2 and TIMP-2 was highly expressed in established fibrosis (p = and p < 0.001, respectively). MMP-14 did not change significantly over time but showed a pattern similar to MMP-2. Significant differences based on post hoc Tukey tests are expressed as (vs. days 7 and 14), (vs. days 7, 14, and 49), (vs. day 7), and (vs. all the other time points). Data are expressed as mean ± SEM; n = 4. (C) MMP-3, MMP-9, and MMP-13 protein was highly expressed by polymorphonuclear and other infiltrating cells in graft lumen at day 7. At day 21, MMP-2 and MMP-14 were mostly localized to α-smooth muscle actin (α-sma) positive myofibroblasts. Scale bars = 50 μm. 92

112 MMP inhibition reduces allograft airway fibrosis but does not block T cell activation In the MMP inhibition studies, drug delivery was confirmed by measuring plasma levels of the inhibitor, SC080. Drug levels were maintained in the therapeutic range (Cmax = 4 ± 0.5 μm; Cmin = 0.05 ± 0.02 μm, mean ± SEM). Serious side effects, such as joint stiffness or reduced mobility, were not observed during the treatment period. Weight gain in SC080-treated groups was higher at day 28 (vehicle, 2.88 ± 0.88 % (increase from day 0); SC080, 4.52 ± 1.31 % (p = 0.02, paired t-test). To examine the role of tissue remodelling in the initiation phase of allograft airway fibrosis, SC080 was administered from day 0 to day 28. SC080-treated allografts demonstrated partial preservation of graft lumen patency at day 28 (Figure 3-5A, B) despite not recovering the epithelium (Figure 3-5A, insets of the left panels). The cells observed in the lumen of SC080-treated allografts were mainly procollagen α1(i) positive, and the α-sma staining was much weaker in comparison with that in vehicle-treated allografts (Figure 3-5C, right panels). The basement membrane was better preserved in SC080-treated allografts (Figure 3-5C, arrow). Semi-quantification of fibroblasts and myofibroblasts in the partially or totally obliterated allograft lumen demonstrated a reduced number of myofibroblasts, but not α-sma (-) fibroblasts Figure 3-5D (i). The ratio of myofibroblasts/fibroblasts in SC080-treated allografts was reduced (Figure 3-5D (ii)). These results indicate that SC080 treatment may affect differentiation or persistence of myofibroblasts in fibrous tissue. 93

113 Figure 3-5. SC080 treatment from day 0 to day 28 prevents allograft airway fibrosis. (A) SC080 treatment reduced obliterative fibrosis at day 28 (original, 40x). SC080 treatment did not recover epithelium (insets; original, 400x). (B) Morphometric quantification demonstrated significantly decreased lumenal obliteration in SC080-treated allografts compared to vehicle-treated allografts (* p = 0.015). Data are expressed as mean ± SEM; n = 6. (C) SC080 treatment reduced the number of myofibroblasts (left panel; original, 200x) and preserved basement membrane (arrow, autofluorescence of basement membrane). SC080 treatment reduced expression of α-smooth muscle actin (α-sma) in fibroblasts (right panel; original, 1000x). Scale bars = 50 μm. (D) Semi-quantification of myofibroblasts and α-sma negative fibroblasts in the obliterated area of the allograft lumen demonstrated a reduced number of myofibroblasts (P = 0.001), and a reduced myofibroblast/fibroblast ratio (P = 0.017) in SC080-treated allografts. 94

114 We further determined the effect of MMP inhibition on allogenic immune responses. In the graft lumen at day 7 (Figure 3-6A), SC080-treated allografts showed significantly reduced inflammatory cell infiltration. Quantification of lumenal cell infiltration confirmed this observation (Figure 3-6B). In the subepithelial area of the allograft trachea, however, SC080 treatment did not reduce infiltration of macrophages and T cells (Figure 3-6C). In flow cytometry analyses of whole tracheal tissue, the numbers of CD3 + T cells, CD4 + and CD8 + T-cell subsets (Figure 3-6D(a)), and the CD8 + /CD4 + ratio (Figure 3-6D(b)) were similar in allografts treated with or without SC080. These T cells were significantly less in the isografts. Activation markers for T cells, CD25 and CD134 (data not shown), also did not show significant differences between vehicle- and SC080-treated allografts either among total CD3+ T cells, or CD4 + and CD8 + T-cell subsets. In a T-cell proliferation assay, SC080 did not significantly affect T-cell proliferation within a range of concentrations similar to that detected in the plasma, in contrast to the potent inhibitory effect seen with cyclosporine A (Figure 3-6E(i)). In an MLR assay for lymphocytes from spleen (Figure 3-6E(ii)) or mediastinal lymph nodes (data not shown), SC080 did not significantly affect T-cell proliferation either. In a real-time RT-PCR analysis of post-transplant allograft trachea (Figure 3-6F, left), SC080 treated allografts did not show differences in tumor necrosis factor α, whereas interferon-γ was significantly increased compared to control allografts. Although the mechanism of increased interferon-γ gene expression is not clear, this finding is consistent with the abnormal cytokine profile observed in MMP-9 knockout mice (269). Conversely, gene expression of interleukin-1β was down regulated by SC080 treatment (Figure 3-6F, right). Since interleukin (IL)-1β is a potent cytokine that stimulates fibroblasts (311), this downregulating effect of SC080 on IL-1β may in part explain its effect on tissue remodelling. These results indicate that the effect of SC080 on obliterative fibrosis is not likely mediated by suppression of T-cell alloreactivity. 95

115 Figure 3-6. SC080 treatment from day 0 to day 28 inhibits lumenal leukocyte infiltration, but not T cell activation in allografts. (A) SC080 treatment prevented cell infiltration in graft lumen. Original magnification of insets was 40x. (B) Quantification of lumenal cell infiltration demonstrated significantly less numbers of cells in the SC080-treated allografts at day 7 (*p < 0.001). Data are expressed as mean ± SEM. (C) At day 7, in the subepithelial area SC080 treatment did not reduce total cell infiltration (hematoxylin and eosin stain), CD68 + macrophages, or CD3 + T cells. Arrow heads indicate epithelium or the location that should contain epithelium. (D) Flow cytometry analysis at day 10 using whole graft extract. (a) SC080 treatment did not reduce infiltration of total T cell number (CD3 + ) and T cell subsets (CD4 + and CD8 + T cells) in allografts. The number of cells was significantly less in isografts. (b) SC080 treatment did not reduce the CD8/CD4 ratio in allografts, suggesting an active cellular immune response in allografts either with or without SC080 treatment. (E) (i) T-cell proliferation assay. Lymphocytes (1x10 5 ) from draining lymph nodes of non-transplanted Lewis rats stimulated with plate-bound anti-rat CD3 and anti-rat CD28 with doses of indicated drugs. (ii) Mixed lymphocyte reaction by transplant recipients. Cells (1x10 5 ) from the spleen of SC080-treated vs. nontreated transplanted (Tx, post-operative day 10) animals were used as responders against indicated numbers of irradiated allo-splenocytes (Brown-Norway) and cultured for 4 days. (F) Real-time RT- PCR analysis for tracheal allografts at day 7 with or without SC080 treatment. Gene expression of interferon (IFN)-γ was significantly increased by SC080 whereas IL-1β was significantly decreased by SC080. (*represents significant differences compared to other groups, p < 0.05, Tukey test; Scale bars = 50 μm.) 96

116 MMP inhibition reduces the progression of developing fibrosis To elucidate the role of tissue remodelling in developing fibrosis, SC080 was administered from day 14 to day 28, when fibroproliferative tissue remodelling had been initiated, but not completely obliterated the lumen yet. Despite the delayed initiation of MMP inhibition and relatively shorter treatment, SC080-treated allografts demonstrated better graft lumen patency at day 28 (Figure 3-7A, B), compared with allografts treated with SC080 continuously for four weeks (Figure 3-6A, B). Again, the epithelium was not recovered by SC080-treatment (Figure 3-7A, insets). Immunofluorescence labelling demonstrated less staining of α-sma in the lumen of SC080-treated allografts at day 28 (Figure 3-7C). Semi-quantification of fibroblasts and myofibroblasts in the partially or totally obliterated allograft lumen demonstrated a reduced number of myofibroblasts (P = 0.004). The number of α-sma (-) fibroblasts showed a trend toward a decrease with SC080 treatment (P = 0.064) and the ratio of myofibroblasts/fibroblasts was not significantly different (Figure 3-7D). 97

117 Figure 3-7. SC080 treatment from day 14 to day 28 reduced airway obliteration. (A) SC080 treatment partially preserved the opening of the allograft lumen at day 28 (original, 40x). SC080 treatment did not recover epithelium (insets; original, 400x). (B) Morphometric quantification demonstrated significantly decreased lumenal obliteration in SC080-treated allografts compared to vehicle-treated allografts (* p < 0.001). Data are expressed as mean ± SEM; n = 6. (C) SC080 treatment reduced total number of fibroblasts (myofibroblasts and α-sma negative fibroblasts) in the graft lumen (left panel; original, 200x). SC080 treatment reduced myofibroblast (α-sma positive; red) to fibroblast (procollagen α1 positive, green) ratio (right panel; original, 1000x). Scale bars = 50 μm. (D) Semi-quantification of myofibroblasts and fibroblasts in the obliterated area of the allograft lumen demonstrated a reduced number of myofibroblasts (P = 0.004) in SC080-treated allografts. The number of fibroblasts and the myofibroblast/fibroblast ratio did not reach statistic significance. 98

118 Established fibrosis is modulated by MMP inhibition To determine whether the ongoing tissue remodelling in established fibrosis can be further modulated, SC080 was administered from day 21 to day 35 (Figure 3-1), that is, when the allograft lumen has been completely obliterated by fibrotic tissue. SC080 treatment did not reverse the obliteration of the airway lumen; however, it did reduce the density of fibrotic tissue with less extracellular matrix and cellularity at day 35 (Figure 3-8A). SC080 treatment reduced MMP-mediated gelatinolytic activity in allograft lumen as assessed by in situ zymography (Figure 3-8B). Immunofluorescence staining demonstrated less number of myofibroblasts in the SC080-treated allograft lumen (Figure 3-8C). PSR staining indicated reduced total collagen deposition with regular light microscopy (Figure 3-8D). Under polarized light, less number of type III collagen fibers were observed in SC080-treated allografts (Figure 3-8D). 99

119 Figure 3-8. SC080 treatment from day 21 to day 35 modulated airway fibrosis. (A) Reduced extracellular matrix accumulation and cells in the graft lumen seen at day 35 in the SC080-treated group. (B) In situ gelatin zymogram. (a) Serial frozen sections were processed for H&E staining and in situ gelatin zymogram. Gelatinolysis on the gelatin-coated film appears blue to white color. SC080 treatment inhibited gelatinolytic MMP activity in the fibrotic area. Original magnification was 40x. (b) Morphometric quantification of the gelatinolytic area demonstrated significant difference (*P = 0.001). (C) Immunofluorescence labeling for myofibroblasts with antibody against α-smooth muscle actin (α- SMA; red). (a) Myofibroblasts were reduced in the lumen of SC080-treated allografts at day 35. (b) Semi-quantification demonstrated significant difference (*p = 0.012). (D) Picrosirius red (PSR) staining for fibrosis area. (a) PSR staining demonstrated reduced interstitial collagen in the lumen of SC080-treated allografts (left panel; observed under regular light), particularly thick type I collagen fibers that appear orange-to-red under polarized light (right panel) at day 35. (b) Quantification of PSR positive areas demonstrated significantly reduced total collagen in SC080-treated allografts (*p = 0.028). Scale bars = 50 μm. Data are expressed as mean ± SEM; n =

120 3.5. Discussion Tissue remodelling is a continuous process after fibrosis establishment This study shows that the alloimmune response promotes the initiation and progression of allograft airway fibrosis and maintains established fibrosis in the pulmonary milieu. Tissue remodelling continues even after complete airway obliteration, with dynamic changes in collagen deposition and composition, phenotypes of fibroblasts and expression and activation of different types of MMPs. Importantly, using a pharmacological approach we demonstrated that this dynamic tissue remodelling can be manipulated by MMP inhibition, suggesting potential for novel therapies for obliterative bronchiolitis (OB). The first important finding of the present study is that after total obliteration of the allograft lumen, established obstructive fibrosis still undergoes dynamic remodelling. We have noted three features of tissue remodelling that continue after airway obliteration is completed: changes in collagen types, cell types and MMP profiles. Using PSR staining and polarized microscopy, not only did we find that the density of deposited collagens continues to increase after complete airway obliteration, but we also observed that type III collagen becomes dominant over type I collagen, which was also supported by the changes seen in procollagen gene expression profiles. Consistent with our finding, an increase in the ratio of type III to type I collagen in the airways of lung allografts has been reported in patients with chronic lung allograft dysfunction related to OB (312). Switching from type I to type III collagen has been suggested to represent a maturation process of allograft fibrosis (313). The dynamic changes of collagen might be related to the change from fibroblasts to myofibroblasts, as demonstrated with immunofluorescence staining. Again, similar to our findings, established lesions of human post-lung transplant OB are often accompanied by a large number of myofibroblasts (36). Persistence of myofibroblasts may prevent epithelial repair and regeneration (314). The contractile property of myofibroblasts may also contribute to altered tissue structure (37). Continuous production of growth factors by myofibroblasts may contribute to progression of the tissue remodelling in an autocrine or paracrine manner (315, 316). Continuous production of extracellular matrix molecules by these cells 101

121 may prevent the resolution of fibrosis. We also observed dynamic alterations of MMPs, which appear to play a role in regulation of the fibrotic process, changing the quality and quantity of collagen fibers. The continuous tissue remodelling in allograft airway fibrosis could be clinically very important, partially explaining the ineffectiveness of augmented immunosuppression. Although the alloimmune response is the cause for the initiation and continuation of fibrosis in the airway, the process of tissue remodelling could be regulated relatively independently. Suppression of the alloimmune response alone may have little impact on this pathway Tissue remodelling is manipulatable Another important finding of the present study is that after complete obliteration of the allograft lumen, the established fibrosis can still be manipulated by MMP inhibition. This finding is indeed striking, considering the general assumption that established fibrosis is static and irreversible (126) far beyond the point of no return (229). When the MMP inhibitor was used immediately after transplantation, it partially reduced airway obliteration and lumenal infiltration of inflammatory cells. Gelatinolytic MMPs such as MMP-9 have been reported to facilitate cell migration into inflammatory sites by degrading type IV collagen in the basement membrane (317). The role of MMP-9 in the initiation of fibrosis has been demonstrated in a mouse model of OB (269). Furthermore, in patients with OB, a high ratio of MMP-9 to TIMP-1 has been detected in sputum (226) and bronchoalveolar lavage fluid (227). Since MMP-3, MMP-9, and MMP-13 have common regulatory elements in their promoter regions responding to inflammatory stimuli (267) and compose a post-transcriptional proteolytic cascade (318, 319), these MMPs might promote the initial phase of tissue remodelling. The blocking effects of SC080 in the early phase of fibrosis may work mainly through these MMPs. In our previous studies, interventions directed towards preventing fibrosis with adenoviral mediated gene delivery of TGF-β receptor 3 (249), or inhibitors for angiotensin converting enzyme (264), were started immediately after allograft transplantation. Both approaches reduced fibrotic obliteration in the allografts. A novel finding, however, is that the MMP inhibitor did not affect T cell activation, cellular 102

122 infiltration in the sub-epithelial area, or early fibrous obliteration by day 7, suggesting that MMP inhibition could slow down the fibrosis even if the alloimmune response still takes place. MMP inhibition also did not improve epithelial repair and regeneration, suggesting that the epithelial lesion is likely primarily mediated by the alloimmune process MMP inhibition is effective even after the initiation of the fibrotic process Interestingly, we found that administration of the MMP inhibitor during the developing phase of fibrosis (days 14-28), showed even better effects to reduce the airway obliteration. This result needs to be further confirmed and the underlying mechanism is unknown. However, we did notice that the protein expression and activation of MMP-2 was elevated after day 14 and persisted afterwards. The gene expression of MMP-14 was also higher after day 14 of allograft transplantation. MMP-2 has been shown to stimulate the proliferation of subepithelial fibroblasts in airways (238). MMP-2 can also release and activate transforming growth factor-β1 (320), which in turn promotes fibroblast migration (321) and differentiation into myofibroblasts (322). MMP-14 activates pro-mmp-2 (310) and also directly regulates the microenvironment around myofibroblasts by digesting intact type-i and type-iii collagen fibers as well as other extracellular matrix components (323, 324). It is possible that although SC080 is not specific to certain MMPs, administration of this inhibitor during the developing phase of fibrosis mainly blocks the activities of MMP-2 and MMP-14. The most intriguing observation in the present study is that when SC080 was administrated after the completion of airway obliteration, it reduced the density of fibrotic tissue, gelatinolytic activity of MMPs, the number of myofibroblasts and the collagen deposition. These results suggest that the fibrosis in the airway can be modulated; thus, it may ultimately be possible to develop novel therapies to reverse the lesion. This observation may lead to a revolutionary concept in the management of OB in the future. 103

123 Limitations of the study and future directions to modify tissue remodelling in OB The purpose of the study was not to reveal a single pathway or molecule that may be responsible for the development of OB, but to re-evaluate the concept of OB. The significant impact of general MMP inhibition on advanced fibrosis indicates that we should investigate specific MMPs in tissue remodelling at different phases of OB. Refining MMP inhibition might impact more significantly on established allograft airway fibrosis. Careful evaluation of the respective roles of MMPs and collagen may further reveal possible mechanisms by which modulation of tissue remodelling may attenuate developing and established fibrosis. Although we emphasize the importance of tissue remodelling as a new mechanism of obliterative airway disease in lung transplantation, the interaction between tissue remodelling and alloimmune response is yet to be addressed. Particularly, in the initiation phase, these two processes are not easily dissectible: expression of some MMPs depends on inflammation (267); tissue remodelling (e.g. basement membrane destruction) facilitates inflammatory cell infiltration (317). Furthermore, recent evidence suggests possible crosstalk between leukocytes (e.g. lymphocytes) and stromal cells (e.g. fibroblasts) through chemokines, growth factors (325), or cell-cell contact (100, 326) in chronic inflammation (327). It is possible that rejection and tissue remodelling are different, but reciprocally dependent processes in the pathophysiology of chronic allograft airway fibrosis. The novel intra-pulmonary tracheal transplant model appears to be an improved model for the study of mechanisms of OB, reflecting the influences of the milieu of the lung. This model has provided us an opportunity to examine novel hypotheses regarding tissue remodelling in allograft airways. However, a recognized limitation of this model includes the airway epithelium and supporting structures that may not be exactly analogous to the distal airways affected by OB in clinical lung transplantation. Thus, it would be important to further examine the significance of tissue remodelling in changes of distal airways after orthotopic lung transplantation in future Conclusion 104

124 In conclusion, we have demonstrated continuous tissue remodelling in allograft airway fibrosis in the pulmonary milieu. These results serve to enhance our understanding of the mechanisms underlying the development of OB after lung transplantation from a paradigm of pure rejection to a multifactorial disorder, of which tissue remodelling is an important component. Importantly, the fact that the process is dynamic and that it can be altered in the advanced stages of fibrosis indicates a novel potential therapeutic avenue to pursue in developing strategies to prevent and treat OB after lung transplantation. 105

125 Chapter 4 Metalloproteinase-dependent migration of myofibroblast progenitors of extra-pulmonary origin contributes to posttransplant allograft airway fibrosis in the lung The content of the chapter was submitted to the American Journal of Transplantation. A portion of this manuscript was presented at the International Society for Heart and Lung Transplantation 26 th annual meeting and scientific sessions (April 7, 2006, Madrid, Spain). The abstract was published in the Journal of Heart and Lung Transplantation, 2006;25(2):S

126 4.1. Abstract Fibroproliferative airway remodelling mediated by myofibroblasts is an important mechanism of obliterative bronchiolitis (OB) after lung transplantation and is a potential therapeutic target. We developed a novel animal model of OB by concurrently conducting orthotopic left lung transplantation from Lewis (RT1A l ) to F1 (Brown-Norway (RT1A n ) x Lewis) and intrapulmonary tracheal transplantation from a Wister-Furth rat (RT1A u ) into the transplanted Lewis-derived lung. Analyzing MHC class I molecule expressed by myofibroblasts using immunohistochemistry, we demonstrated that myofibroblasts in allograft airway fibrosis were almost exclusively of extra-pulmonary origin in this model. Cultured fibrocytes derived from peripheral blood mononuclear cells showed high levels of MMP-9 and MMP-12 expression as well as type I collagen and α-smooth muscle actin compared to total mononuclear cells at day 14. Migration of these fibroblast-like cells was significantly inhibited by general MMP inhibitors, SC080 and MMI270 in a standard wound healing assay. The original bloodborne mononuclear cells were cultured overnight, labelled with PKH26, and then intravenously injected into a recipient rat of intrapulmonary allogenic tracheal transplantation (Brown-Norway to Lewis) at post-operative day 7, 10, and 13. At day 28, PKH26-labelled myofibroblasts were demonstrated in the fibrous tissue. Using the same animal model, we demonstrated that the general MMP inhibitors can inhibit an increase in myofibroblasts in allograft airways and reduce obliterative fibrosis without suppressing alloimmune responses. MMP inhibitor treatment did not change the ratio of Ki-67 positive proliferating cells or TUNEL positive apoptotic cells either in vimentin positive fibroblasts or α-sma positive myofibroblasts. These results collectively demonstrate that MMPdependent migration of myofibroblast precursors of extra-pulmonary origin is an important mechanism of allograft airway fibrosis. 107

127 4.2. Introduction Obliterative bronchiolitis (OB) is a fibroproliferative airway obliteration complicating transplanted pulmonary allografts (276). OB and its clinical correlate, bronchiolitis obliterans syndrome (BOS) affect about 50 % of lung transplant recipients in 5 years (3). Although OB/BOS has been accepted to be a form of chronic rejection, immunosuppression is usually ineffective after the initiation of the fibrotic process (82). A novel therapeutic intervention is required to target the fibrotic process at a different level from immunosuppression. The myofibroblast plays a central role in fibroproliferative tissue remodelling (37) including OB after lung transplantation by producing excessive amounts of extracellular matrix. Studies have demonstrated that myofibroblasts are probably of multiple origins (328) and one of the most important sources is circulating fibrocytes, blood-borne cells expressing haematopoietic cell markers (e.g. CD34, CD45) as well as markers for fibroblasts such as type I collagen (329). After tissue injury, circulating fibrocytes migrate into the injured tissue in a chemokine-dependent manner and differentiate into a myofibroblast phenotype, contributing to wound healing (330). Circulating fibrocytes have been demonstrated to contribute to pathological fibrotic processes in the lung such as in animal models of asthma (331) and pulmonary fibrosis (47). It has recently been demonstrated that up to 30% of myofibroblasts in OB lesions in human pulmonary allografts are of recipient origin (36). Thus, migration of myofibroblast precursors from the systemic circulation is an important mechanism of the fibrotic process of OB after lung transplantation and is a potential therapeutic target of OB. Matrix metalloproteinases (MMPs) are a group of enzymes that exert various biological effects in cell migration, matrix degradation, as well as inflammatory and immune responses (268). We have recently demonstrated that broad-spectrum MMP inhibition blocked the fibrotic process at a different level from immunosuppression in a rat intrapulmonary tracheal transplant model of OB (85). However, the mechanism whereby MMP inhibition reduces the number of myofibroblasts and the degree of airway obliteration has not been elucidated. Since circulating fibrocytes have been demonstrated to express 108

128 higher levels of MMP-9 (332), it is possible that circulating fibrocytes contribute to allograft airway fibrosis of OB through MMP-mediated migration into allograft airways. In the present study, we examine the origin of myofibroblasts in allograft airway fibrosis using a novel concurrent transplantation model of an orthotopic lung and an intrapulmonary trachea. We also confirm the results by transfusing fluorescently labelled fibrocyte progenitor cells in transplant recipient rats of intrapulmonary tracheal transplantation and demonstrate the contribution of blood-borne cells to the myofibroblast population of allograft airway fibrosis. Moreover, we culture fibrocytes from rat peripheral blood mononuclear cells and demonstrate their expression of MMPs and the effect of MMP inhibition on their migration in vitro. Lastly, we treat allograft transplant recipients with MMP inhibitors and examine the mechanisms whereby MMP inhibition reduces airway fibrosis in vivo Results Myofibroblasts of extra-pulmonary origin in allograft airway fibrosis in the lung The intrapulmonary tracheal transplant model of OB is an improved model, sharing the simplicity and reliable reproducibility of other heterotopic tracheal transplant models, while reflecting the pulmonary milieu that may have significant impact on immune responses and tissue remodelling (85). To investigate the origin of myofibroblasts of allograft airway fibrosis, however, this model is limited by the fact that the graft is surrounded by recipient lung tissue. On the other hand, rat orthotopic lung transplant models do not necessarily develop OB in a reliable and reproducible manner (115). To overcome these limitations, we sequentially conducted orthotopic lung transplantation and intrapulmonary tracheal transplantation. The left lung of a Lewis rat was transplanted into a F1 rat (Lewis x Brown-Norway) and then a trachea of a Wister-Furth rat was transplanted into the orthotopically transplanted Lewis-derived lung (Figure 4-1A(d)). In this novel model, an F1 rat has MHC class I molecules of both Lewis-type RT1A l and Brown-Norway-type RT1A u, while F1-derived lymphocytes as well as passenger lymphocytes in the Lewis-derived lung theoretically reject a tracheal graft derived from a Wister-Furth rat that has RT1A n. The result was compared with three control 109

129 groups of intrapulmonary tracheal transplantation: (a) Lewis to Brown-Norway, (b) Brown-Norway to Lewis, and (c) Wister-Furth to F1 (Lewis x Brown-Norway) (Figure 4-1A-a, b, and c, respectively). Tracheal grafts of the four groups showed total obliteration of the lumen by day 28. The obliterative fibrosis was morphologically indistinguishable so that these changes were considered to have occurred due to similar alloimmune responses due to MHC mismatch. Major histocompatibility complex (MHC) class I molecules are known to be expressed by almost all the nucleated cells of the body. Different inbred rat strains may express different MHC class I molecules such as RT1A l of Lewis rats. Thus, the strain variance has been utilized to detect the origin of cells in transplant settings using immunohistochemical detection (333) and PCR for genomic DNA (333, 334). We examined MHC class I molecules expressed by myofibroblasts in the fibrous tissue of obliterative fibrosis. Immunofluorescence labelling for pan-mhc class I (OX-18) demonstrated that myofibroblasts in allograft airway fibrosis of the four groups express MHC class I molecules at a similar level (data not shown). On the other hand, labelling for RT1A n (OX-27) for Brown-Norwaytype MHC class I demonstrated a variable staining pattern among groups. RT1A n was the most intensively labelled in myofibroblasts of the intrapulmonary tracheal transplantation group from Lewis to Brown-Norway (Figure 4-1B(a)), and was the least intensive in myofibroblasts of Brown-Norway to Lewis (Figure 4-1B(b)). These results demonstrate that myofibroblasts of allograft airway fibrosis that develop after intrapulmonary tracheal transplantation of an MHC-mismatched allograft trachea are of recipient origin. In the intrapulmonary tracheal transplantation from Wister-Furth to F1 rats, myofibroblasts were positively labelled for RT1A n (Figure 4-1(c)). Interestingly, however, the staining intensity was not as intensive as that of Lewis-to-Brown-Norway (Figure 4-1(a)). This may be due to the difference in the number of MHC class I molecules expressed by cells derived from a Brown-Norway rat or those derived from an F1 rat, the latter of which express both RT1A n and RT1A n so that the theoretical number of RT1A n molecules expressed is half of Brown-Norway-derived cells. 110

130 Strikingly, in the combined orthotopic lung and intrapulmonary tracheal transplant model, immunofluorescence labelling for RT1A u positively stained myofibroblasts to a similar degree as those in an intrapulmonary tracheal transplantation from Wister-Furth to F1 rats (FIgure 4-1(d), compared with Figure 4-1(c)). This result demonstrates that myofibroblasts in the allograft airway fibrosis originate from the F1 rat. Since the lung in which the tracheal graft was implanted was derived from a Lewis rat, myofibroblasts in allograft airway fibrosis are considered to be of extra-pulmonary origin. 111

131 Figure 4-1. Myofibroblasts in allograft airway fibrosis are of extra-pulmonary origin. (A) (a-c) Intrapulmonary tracheal transplantation. (d) Sequential orthotopic lung transplantation from Lewis to F1 (Lewis x BN) and intrapulmonary tracheal transplantation from Wister-Furth to the transplanted Lewis lung. (B) OX27 staining for RT1A n MHC class I molecules demonstrates high levels of RT1A n expression in myofibroblasts of Lewis-to-BN intrapulmonary transplantation (a), intermediate levels of RT1A n expression in Wister-Furth to F1 intrapulmonary transplantation, (c) sequential Wister-Furth to Lewis to F1 transplantation (d), and negative staining in BN to Lewis intrapulmonary transplantation (b). Note that BN rats have homozygous RT1A n while F1 (Lewis x BN) rats have both RT1A l and RT1A n. Lewis and Wister-Furth rats have RT1A l and RT1A u, respectively. Confocal microscopy was used to demonstrate co-localization of MHC I molecules and α-sma. 112

132 Cultured circulating fibrocytes can differentiate into myofibroblasts Recent evidence suggests that circulating fibrocytes are an important source of myofibroblasts in animal models of asthma (331) and pulmonary fibrosis (47). Thus, we investigated PBMC-derived fibrocytes next. Continuous culturing of PBMCs adherent to fibronectin-coated plates resulted in differentiation of round-shaped PBMCs (Figure 4-2A(a)) into spindle-shaped fibroblast-like cells by day 14 (Figure 4-2A(b)). Consistent with previous reports, these cells were mostly positive for collagen-i (Figure 4-2A(d)) and approximately 10% of these cells showed α-sma expression (Figure 4-2A(e)). Stimulation of cells from days 12 to 14 with transforming growth factor β1 (TGF- β1) induced a more evident myofibroblast phenotype (Figure 4-2A(f)). At day 28, cultured fibrocytes showed a swollen morphology and decreased in their number (Figure 4-2A(c)). In real-time RT-PCR, gene expression of procollagen α1(i) and α-sma peaked at day 14, while CD45 expression was maintained by day 28 (Figure 4-2B). Because the cultured cell population was not pure, the relative increase in CD45 and decrease in procollagen α1(i) and α-sma at day 28 might be explained by selective deletion of fibrocytes and survival of undifferentiated cells. Collectively, the results demonstrate that a portion of PBMCs could differentiate into fibroblast-like cells and that these blood-borne fibrocytes can further differentiate into mature myofibroblasts. 113

133 A B Relative gene expression COL1 * D1 2 D14 D30 28 Time (Days) α-sma * D1 2 D14 D30 28 Time (Days) CD45 D1 2 D14 D30 28 Time (Days) Figure 4-2. Cultured blood-borne fibrocytes and their gene expression. (A) (a) Bright field observation of PBMCs at day 2 shows round cells adherent to a fibronectin-coated culture plate. (b) PBMCs differentiate into spindle-shaped fibroblast-like cells by day 14. (c) Many of the fibroblast-like cells show a swollen morphology around day 21 and the number starts to decline. Fluorescence labelling of cultured PBMCs demonstrates collagen I expression in the spindle-shaped cells (d) and α- SMA expression in approximately 10% of the collagen I positive cells (e). Insets indicate IgG negative control for collagen I and α-sma, respectively. (f) Stimulation of cells with transforming growth factor beta 1 (10 ng/ml) induces a more evident myofibroblast phenotype with intracellular stress fiber formation. (B) Real-time PCR analysis for procollagen α1(i), α-sma, and CD45 gene expression. Both procollagen α1(i) and α-sma gene expression peak at day 14, and decrease by day 28. *P < 0.05 compared with the other groups. 114

134 PBMC-derived cells migrate to allograft airways and differentiate into myofibroblasts in vivo To examine whether PBMC-derived fibrocytes or their progenitor mononuclear cells differentiate into myofibroblasts in vivo, we labelled syngenic PBMCs with PKH26 (red fluorescent dye) and injected them into recipient animals of intrapulmonary allograft tracheal transplantation at post-operative days 7, 10, and 13 (Figure 4-3A). PBMCs were incubated overnight after isolation and only cells adherent to the culture well were labelled and injected. Control cells that were continuously cultured in vitro showed a fibroblast-like phenotype with expression of collagen I at day 14. Almost all of these cells showed intensive intracellular PKH26 signalling. The red fluorescence of PKH26 appeared to be granular and heterogeneous in the cytoplasm (Figure 4-3C). Allograft recipient animals that were injected with PKH26-labelled cells were sacrificed at postoperative day 28 and the tracheal allograft in the lung was examined. In the fibrous tissue, α-smapositive PKH26-positive myofibroblasts were observed only in a third of tissue sections, where these cells tended to be clustered (Figure 4-3B). With this double labelling, we recognized that PKH26 disperses if the tissue section is fixed in acetone or ethanol. Consequently, we used unfixed tissue sections so that the α-sma staining does not appear as clear as conventional immunofluorescence labelling but was sufficient to examine colocalization of PKH26 labelling to myofibroblasts under confocal microscopy. The staining pattern of PKH26 was granular and heterogeneous in the cytoplasm. By 3-dimensional reconstruction of the confocal images, the localization of PKH26 in the cytoplasm of myofibroblasts was clearly demonstrated (Figure 4-3B, right panel). The relatively small number of PKH26-positive myofibroblasts may be explained by the number of injected cells containing all adherent PBMCs, of which only a portion is considered to differentiate into fibrocytes (47). 115

135 Figure 4-3. PKH26-labelled PBMCs differentiate into myofibroblasts in allograft airway fibrosis. (A) Study design of PKH26-labeled PBMC injection. *Freshly harvested PBMCs were cultured over night to select cells adherent to plastic wells and then the adherent cells (3 x 10 5 cells) were harvested, labeled with PKH26 and administered intravenously to an allograft recipient. This procedure was repeated three times at post-transplant days 7, 10, and 13. (B) Confocal microscopic analysis colocalizes PKH26 signalling (red fluorescence) with α-sma-positive myofibroblasts in allograft airway fibrosis at day 28. Yellow arrows indicate myofibroblasts positive for PKH26. Scale bar = 5μm. (C) Cells have differentiated into collagen I-positive fibroblast-like cells by day 14 in vitro. The labeling pattern of PKH26 is granular and heterogeneous in the cytoplasm of cells similar to PKH26-positive myofibroblasts in vivo. 116

136 MMP expression and MMP-dependent migration of fibrocytes in vitro Given the contribution of PBMC-derived fibrocytes to the myofibroblast population of allograft airway fibrosis in the present model, we examined expression of MMPs, an important group of enzymes involved in cell migration of various cell types (268) next. Using real-time RT-PCR analysis, gene expression of MMP-9 and MMP-12 as well as TIMP-1 were observed to be significantly higher in fibrocytes cultured for 14 days compared with that of original PBMCs (Figure 4-4A). We then examined whether MMPs play important roles in fibrocyte migration in vitro using a standard wound healing assay. After scratching a monolayer of fibrocytes cultured on a fibronectin-coated well, a wound of approximately 60 μm in width was almost completely closed by fibrocytes within 48 hours without any additional treatment. To distinguish the effect of cell proliferation, fibrocytes were pretreated with mitomycin C, and then a wound was created by a pipet tip. When comparing fibrocyte migration between MMP inhibitor-treated groups and vehicle control groups, we observed that both MMP inhibitors significantly reduced the distance of fibrocyte migration compared with control (Figure 4-4B). 117

137 A Relative gene expression MMP-9 * D1 2 D14 D30 28 Time (Days) MMP-12 * * D1 2 D14 D30 28 Time (Days) TIMP-1 * D1 2 D14 D30 28 Time (Days) B Cell migration (arbitrary unit) * 0 Ve 10 nm 100 nm SC080 2 Cell migration (arbitrary unit) * Ve 10 nm 100 nm MMI270 Figure 4-4. MMP expression and MMP-dependent migration of cultured fibrocytes. (A) Realtime RT-PCR analysis demonstrates increased gene expression of MMP-9, MMP-12, and TIMP-1 at day 14 of cultured fibrocytes. *P < 0.05 compared with the other groups. (B) A standard wound healing assay was conducted using cultured fibrocytes. Cells were treated with general MMP inhibitors, SC080 or MMI270. Distance of cell migration was evaluated 24 hours after scratching a cell layer. *P < 0.05 compared with control

138 MMP inhibition reduces obliterative fibrosis even after the initiation of fibrotic process while cyclosporine does not To examine the contribution of MMP-dependent migration of myofibroblast precursors to allograft airway fibrosis, we subsequently treated recipient animals of intrapulmonary allograft tracheal transplantation with 1) a broad-spectrum MMP inhibitor, SC mg/kg/d, oral administration, bid; 2) a broad-spectrum MMP inhibitor, MMI270 (gift from Novartis Pharmaceuticals Canada Inc., Canada), 30 mg/kg/d; 3) cyclosporine (Novartis Pharmaceuticals Canada Inc.), 10 mg/kg/d, subcutaneous injection, qd; 4) only a vehicle for SC080 as a control (oral administration), qd. In the first set of animals, drug treatment was initiated from the time of transplantation (i.e. day 0). At day 28, the lumen of vehicle-treated control allografts was almost completely obliterated by fibrous tissue; the airway epithelium was not observed. Infiltration of T cells was observed in the lumen as well as in the perigraft area beside a graft cartilage (Figure 4-5A). Cyclosporine treatment initiated at the time of transplantation prevented obliterative fibrosis and preserved the airway epithelium by day 28. T-cell infiltration was not observed in the lumen of allografts treated with cyclosporine from day 0 and the number of T cells in subepithelial area was minimal compared to that of control (Figure 4-5A). Treatment with SC080 or MMI270 did not protect the airway epithelium at day 28, and the allograft lumen remained partially open (Figure 4-5A). Treatment groups of SC080 and showed similar numbers of infiltrating T cells to that of control in the lumen at day 28. Immunosuppression is usually ineffective in reducing allograft airway fibrosis once the fibrotic process is initiated (82). To examine whether MMP inhibition can still reduce fibrosis at this stage, we administered SC080, MMI270, or cyclosporine from day 14. Cyclosporine had no effect on allograft airway obliteration (Figure 4-5B). Cyclosporine treatment from day 14 reduced T-cell infiltration in the lumen compared with that of control (Figure 4-5B), suggesting that development of fibrosis became more independent of alloimmune responses after the fibrotic process is initiated. Interestingly, MMP inhibitors started from day 14 still reduced lumenal obliteration by day 28, while the airway epithelium was totally lost (Figure 4-5B). These results suggest that the effect of MMP inhibition on allograft 119

139 airway fibrosis is unlikely to be mediated by immunosuppression, although T-cell infiltration appeared to be reduced to some extent in these groups. Morphometric quantification of the lumenal obliteration demonstrated significant differences among groups (Figure 4-5C, P = 0.001). A post-hoc Tukey test indicated significant attenuation of obliterative fibrosis with cyclosporine treatment [day 0-28], SC080 [day 0-28], SC080 [day 14-28], and MMI270 [day 14-28] when compared with control groups and cyclosporine [day 14-28]. Semi-quantification of T cell infiltration in the allograft demonstrated significant differences among groups (Figure 4-5D, P < 0.001) as well. A post-hoc Tukey test indicated a significantly smaller number of T cells in the graft lumen with cyclosporine [day 0-28], cyclosporine [day 14-28], SC080 [day 14-28], and MMI270 [day 14-28] treatment, when compared with that of controls. In addition to the effects of cyclosporine on T-cell infiltration, these results imply that there are complex effects of MMI inhibitors on lymphocyte trafficking. We previously demonstrated an increase in subepithelial infiltration of T cells and macrophages in allograft airways after MMP inhibitor treatment (85). Similar lymphocyte accumulation and activation has been reported in transplant settings (335, 336). Thus, enhancement or accumulation of lymphocyte infiltration might be a common side effect of general MMP inhibitors. 120

140 Figure 4-5. The effect of cyclosporine, SC080, and MMI270 on obliterative allograft airway fibrosis. (A) Representative pictures show an overview of allograft lumen from day 0 to day 28 (top panels) with preserved epithelial lining after cyclosporine treatment, while the epithelium is lost in all the other groups (bottom panels). Immunofluorescence labelling for T cells demonstrate reduced T-cell infiltration in cyclosporine treated allografts at day 28, while other groups show T-cell infiltration. (B) The effect of cyclosporine, SC080, and MMI270 (started at day 14) on obliterative allograft airway fibrosis and T-cell infiltration. Representative pictures show an overview of allograft lumen treated from day 14 to day 28. Note the ineffectiveness of cyclosporine in graft obliteration, while MMPinhibitor treatment shows preserved graft lumen at day 28. Original 20x (top panel), 200x (bottom panels). (C) Morphometric quantification demonstrates significantly reduced lumenal obliteration in the treatment groups: cyclosporine (day 0-28), SC080 (day 0-28 and day 14-28), and MMI270 (day 14-28). *P < 0.05 compared with vehicle controls. (D) Semi-quantification demonstrates significantly reduced number of T cells in the treatment groups: cyclosporine (day 0-28 and day 14-28), SC080 (day 14-28), and MMI270 (day 14-28). *P < 0.05 compared with vehicle controls. 121

141 MMP inhibitors reduce myofibroblasts without significantly changing the fibroblastmyofibroblast ratio, proliferation, and cell death The next step was to elucidate the mechanisms whereby MMP inhibitors reduce obliterative airway fibrosis. The number of myofibroblasts in allograft lumen at day 28 was examined in all the treatment and control groups both from day 0 and day 14. Both myofibroblasts and total fibroblasts including α- SMA-positive and -negative cells appeared to be decreased in MMP inhibitor treatment groups, while cyclosporine treatment showed less effects on fibroblasts and myofibroblasts when compared with MMP inhibitors (Figure 4-6A). Semi-quantification of α-sma-positive cells in the allograft lumen demonstrated a significantly smaller number of myofibroblasts in MMP inhibitor treatment groups (Figure 4-6B, P < 0.01). Semi-quantification of the number of vimentin-positive total fibroblasts was also decreased in MMP inhibitor treatment groups (Figure 4-6B, P < 0.01). The effect of MMP inhibitors on total fibroblasts was mainly due to reduced myofibroblasts and the number of α-smanegative vimentin-positive cells did not vary among groups. Notably, the semi-quantification was conducted within the obliterated area of the graft lumen so that the total number of fibroblasts and myofibroblasts in the graft lumen should be even smaller in MMP inhibitor treatment groups than that of the cyclosporine group due to different obliteration ratio of the graft lumen. Myofibroblasts are considered to play important roles in OB by producing excessive amounts of extracellular matrix including interstitial collagen (85). Indeed, the reduced number of myofibroblasts by MMP inhibitors and, to a lesser degree, cyclosporine [day 14-28] translated into a reduced amount of collagen accumulation in the allograft lumen. Observation of PSR staining for collagen demonstrates reduced total (intact and denatured) collagen in the treatment groups under regular light microscopy (Figure 4-6C, left pictures), while observation under a polarized microscope demonstrates reduced intact collagen fibrils in the treatment groups (Figure 4-6C, right pictures). Morphometric quantification of PSR staining under a polarized microscope demonstrated significantly reduced fibrilar collagen in the treatment groups of cyclosporine [day 14-28], SC080 [day 14-28], and MMI280 [day 14-28] (Figure 4-6D). Notably, the quantification was conducted within the obliterated area of the graft 122

142 lumen so that the total amount of collagen in the graft lumen should be smaller in MMP inhibitor treatment groups than that of the cyclosporine group due to different obliteration ratio of the graft lumen. The reduced number of myofibroblasts could be explained by several mechanisms including 1) reduced migration of myofibroblasts or their precursors, 2) reduced differentiation of myofibroblast precursors into myofibroblasts, 3) reduced proliferation of myofibroblasts or their precursors, and 4) increased cell death in myofibroblasts or their precursors. Since examination of MHC class I molecules expressed by myofibroblasts already demonstrated that these cells migrate from recipients in intrapulmonary tracheal transplantation, we examined the contribution of the other three factors to the decrease in the number of myofibroblasts after MMP inhibitor treatment. Firstly, the fibroblast-myofibroblast ratio was not significantly different among groups (Figure 4-6B, Table 4-1), suggesting that differentiation into myofibroblasts is not a significant factor. Secondly, cell proliferation that was examined using Ki-67 did not show significant difference among groups either in total fibroblasts or in myofibroblasts (Figure 4-7, Table 4-1), suggesting that cell proliferation is not a major contributing factor to the effect of MMP inhibitors. Thirdly, cell death that was assessed by TUNEL staining was not significantly different among groups either in total fibroblasts or in myofibroblasts (Figure 4-7, Table 4-1), indicating that increased cell death does not explain the effect of MMP inhibitors. 123

143 C D PSR positive area in obliterated area (%) Day 0-28 Day * * * 0 Vehicle CsA SC080 MMI270 Figure 4-6. Fibroblasts, myofibroblasts, and collagen in allograft fibrosis after treatment. (A) Double immunofluorescence labeling for α-sma and vimentin demonstrates allograft airway fibrosis in vehicle control that is highly populated with α-sma-positive myofibroblasts. The allograft lumen treated with SC080 and MMI270 from day demonstrates the most reduced numbers of myofibroblasts, Original, 200x. (B) Semi-quantification of fibroblasts and myofibroblasts in allograft airway fibrosis. A significantly reduced number of myofibroblasts was demonstrated in the treatment groups: CsA [14-28], SC080 [0-28], SC080 [14-28], and MMI270 [14-28] as compared with vehicle controls (*P < 0.05 in a post-hoc Tukey test). Total vimentin+ fibroblasts also reflect similar results, but this is due to the difference in myofibroblasts. Only the obliterated area in the lumen was compared. CsA [0-28] was not included in the study due to the minimal obliteration of the lumen. (C) PSR collagen staining demonstrates quantitative and qualitative differences in interstitial collagen accumulated in the lumen of allografts. MMP inhibitors reduced total collagen observed under a regular microscope (left). MMP inhibitors also reduced type-i thick collagen (red color) observed under a polarized microscope (right). (D) Significantly reduced amounts of fibrilar collagen was demonstrated in the treatment groups of CsA [14-28], SC080 [14-28], and MMI270 [14-28] as compared with vehicle controls (*P < 0.05 in a post-hoc Tukey test). Only the obliterated area in the lumen was compared. CsA [0-28] was not included in the study due to the minimal obliteration of the lumen. 124

144 α-sma Ki-67 Hoechst Vimentin Ki-67 Hoechst A Vehicle SC080 B α-sma TUNEL Hoechst Vimentin TUNEL Hoechst Vehicle SC080 Figure 4-7. Double immunofluorescence of Ki67 + or TUNEL and vimentin or α-sma. (A) Double immunofluorescence labelling for Ki67 + and vimentin or α-sma demonstrates that a small fraction of fibroblasts and myofibroblasts in allograft airway fibrosis are proliferating. Arrows indicate Ki67+ myofibroblasts (left) or fibroblasts (right). Only vehicle and SC080 (day 14-28) treatment groups are shown. (B) Double immunofluorescence labelling for TUNEL and vimentin or α-sma demonstrates that a small fraction of fibroblasts and myofibroblasts in allograft airway fibrosis are undergoing cell death. Arrows indicate TUNEL + myofibroblasts (left) or fibroblasts (right). Only vehicle and SC080 (day 14-28) treatment groups are shown. Original magnification, 200x. 125

145 Treatment Treatment Differentiation Proliferating Proliferating Apoptotic Apoptotic Group [Days] Ratio (%) fibroblast (%) myofibroblast (%) fibroblast (%) myofibroblast (%) Vehicle ± ± ± ± ± ± ± ± ± ±0.26 Cyclosporine 0-28 * * * * * ± ± ± ± ± 0.21 SC ± ± ± ± ± ± ± ± ± ± 0.19 MMI ± ± ± ± ± ± ± ± ± ± Statistic significance N.S. N.S. N.S. N.S. N.S. Table 4-1. MMP inhibition does not change the ratio of myofibroblast differentiation, fibroblast/myofibroblast proliferation or apoptosis. Differentiation ratio is the ratio of α-sma+ myofibroblasts among vimentin+ general fibroblasts including myofibroblasts. Cell proliferation and apoptosis were detected by Ki67 and TUNEL staining, respectively. *The cyclosporine [day 0-28] group was excluded from the analysis due to the lack of fibrous tissue in the allograft lumen. N.S. = not significant. 126

146 4.4. Discussion Fibroproliferative tissue remodelling is an important therapeutic target of OB to overcome the limitations of current immunosuppression Understanding the migration mechanism of myofibroblast progenitors provides a novel opportunity for intervention in fibroproliferative tissue remodelling of allograft airway fibrosis after lung transplantation. The comparison between cyclosporine and two general MMP inhibitors clearly demonstrates that the mechanisms whereby these agents modulate allograft airway fibrosis are different (Figure 4-5). In this interventional study, the best result was obtained in cyclosporine treatment from the time of transplantation. However, in clinical lung transplantation, such a protective effect of immunosuppression is easily overcome for multiple reasons such as chronic bile-acid aspiration (33) and bacterial colonization (32). These alloantigen-independent factors could directly induce tissue injury and/or augment alloimmune responses. Although recent therapeutic approaches including azithromycin (70) and anti-reflux surgery (73) target these alloantigen-independent factors, they are not perfect in preventing OB. Once excessive tissue injury occurs, a fibroproliferation process is launched and immunosuppression and other treatments become less effective. Thus, it is important to understand the mechanisms of fibroproliferative tissue remodelling and to develop novel therapeutic strategies targeting this final common pathway of OB Concurrent intrapulmonary tracheal and orthotopic lung transplantation enables the direct evaluation of myofibroblast progenitor migration of extra-pulmonary origin into the lung It has been demonstrated that circulating fibrocytes contribute to various fibrotic processes in the lung as well as normal wound healing (47, 330, 331). A similar mechanism has been suggested by the chimerism of myofibroblasts in human OB lesions after lung transplantation (36). In the present study, we started to address the question of whether circulating fibrocytes contribute to allograft airway fibrosis in the lung. For this purpose, we conducted two experiments: sequential orthotopic lung transplantation and intrapulmonary tracheal transplantation; and infusion of fluorescence-labelled PBMCs containing fibrocyte progenitors into recipient animals of allograft tracheal transplantation. 127

147 These two experiments overcame several limitations of previous studies. Firstly, the present study using sequential orthotopic and intrapulmonary transplantation provided direct evidence supporting the migration of myofibroblast progenitors from the systemic circulation. Although donor-recipient chimerism of myofibroblasts in human OB lesions (36) supports the contribution of circulation-derived cells to the myofibroblast population in vivo, it is hardly distinguishable whether the direct migration of circulating fibrocytes or the indirect contribution of bone marrow-derived cells (that once resided in the lung as mesenchymal stem cells and then differentiated into multiple lineages including myofibroblast-like cells) (337) are important. Bonemarrow transplantation is a potential experimental strategy to address this question. It is important to note that after bone-marrow transplantation, allograft airway or orthotopic lung transplantation cannot be performed until after transplanted haematopoietic stem cells home in to the host bone marrow and the host recovers from lethal irradiation. The time lag between bone marrow transplantation and secondary allograft airway or orthotopic lung transplantation will obscure the distinction of direct and indirect cell migration to fibrous tissue. Moreover, lethal irradiation may well cause pulmonary injury and facilitate migration of bone marrow-derived progenitor cells to the lung. Secondly, in experiments using labelled cells, stages of cell differentiation in vitro is a matter of concern. In an asthma model, already differentiated fibrocytes were amplified, labelled and infused to demonstrate the contribution of circulating fibrocytes to airway myofibroblasts (331). Although this strategy is convenient, recent studies indicate that many CD14 + monocytes migrate into peripheral tissue and then differentiate into myofibroblasts (46). Moreover, the monocyte-fibrocyte differentiation appears to be affected by a number of local factors including a direct interaction with T cells (330), Th1 and Th2 cytokines produced by T cells and dendritic cells (46), growth factors such as TGF-β (330) and platelet derived growth factor (46) produced by resident macrophages, epithelial cells, and endothelial cells. As such, if fibrocyte differentiation was induced in vitro first and then injected, the migration into allograft airways might not reflect the microenvironment that really affects fibrocyte 128

148 differentiation. Instead, injection of premature PBMCs into the allograft recipient was a more relevant approach even though the number of cells necessary was an obvious disadvantage and likely resulted in a relatively small but demonstrable number of labelled myofibroblasts in allograft airway fibrosis (Figure 4-3) Fibrocytes depend on MMPs for their migration Next, we demonstrated that MMPs play important roles in the migration of fibrocytes in vitro (Figure 5-9). This result was speculated in light of the high levels of expression of MMP-9 by fibrocytes, which was demonstrated previously (332); the general importance of MMP-9 in mobilization of bonemarrow-derived stem cells (338); and the general importance of MMPs in cell migration including fibroblasts (266). Moreover, the high expression of MMP-12 by fibrocytes was the novel finding in this study (Figure 4-2). MMP-12 has multiple interesting properties in tissue remodelling of airways. Its elastase activity has been considered to be important to tissue destruction of pulmonary emphysema (339). In addition, MMP-12 is known to augment inflammation, for example, by releasing tumor necrosis factor alpha from macrophages (339). Since general MMP inhibitors have side effects, including augmented immune responses (85, 335, 336), selective MMP inhibitors for MMP-9 and MMP-12 (340) might be beneficial for clinical application MMP inhibition could have effects other than inhibition of fibrocyte migration Our next course of action was to apply the knowledge of MMP-mediated fibrocyte migration in vivo. In this experiment, we demonstrated the importance of MMP-dependent fibrocyte migration by excluding changes in myofibroblast differentiation, proliferation, and cell death. However, MMPs have such versatile biological effects that it is important to note that the beneficial effect of MMP inhibition could also result from other effects. For example, we previously demonstrated that a low-dose of SC080 can decrease IL-1β expression in allograft airways (85) this may indirectly affect fibrocyte differentiation or myofibroblast activation in vivo (311). We have also demonstrated that MMP inhibition could modulate remodelling of the extracellular matrix (85). The myofibroblast is an 129

149 important cell type involved in allograft remodelling, and thus it is important to note that the effect of MMP inhibition on tissue remodelling is hardly dissectable from that of the migration of myofibroblast precursors. It is possible that the net effect of MMP inhibition on allograft airway fibrosis is multiple and complex, and that the effect on fibrocyte migration is an important part of this net effect. Further investigation is necessary to elucidate the relationship between MMPs and inflammation, immune responses, and matrix remodelling. Notably, we did not use immunosuppression to elucidate the contribution of circulating fibrocytes to allograft airway fibrosis (i.e. the initial sets of animal experiments). Because alloimmune responses appear to eradicate resident donor-derived subepithelial fibroblasts in this model, it is likely that we overestimate circulation-derived myofibroblasts in fibrous lesions. Indeed, in human OB lesions, only 30% of myofibroblasts were of recipient origin (36). Although this is still a significant number, the discrepancy from our finding (almost exclusively circulation-derived cells) might reflect the effect of different experimental settings Conclusion In conclusion, we have demonstrated the contribution of myofibroblast progenitors of extra-pulmonary origin to allograft airway fibrosis in the lung. We have also elucidated that the migration process is MMP-dependent and that MMP inhibition could block or delay the accumulation of myofibroblasts following obliterative fibrosis. These novel and important findings may lead to the development of novel therapeutic strategies for OB after lung transplantation that will directly target fibroproliferative tissue remodelling. 130

150 Chapter 5 The mechanisms of myofibroblast-mediated metalloproteinase-dependent tissue remodelling after lung transplantation A part of this manuscript was presented at the International Society for Heart and Lung Transplantation 27 th annual meeting and scientific sessions (April 27, 2008, Boston, MA). The abstract was published in the Journal of Heart and Lung Transplantation, 2007;26(2):S

151 5.1. Abstract Obliterative bronchiolitis after lung transplantation is a fibroproliferative obstruction of small airways, in which myofibroblasts play an important role in matrix production and degradation. The tissue remodelling in OB might be, however, dynamic, bidirectional, and potentially reversible. The mechanism of myofibroblast-mediated tissue remodelling was investigated, focusing on matrix metalloproteinases (MMPs). In post-transplant human BOS lungs, MMP-2 and MMP-14 were localized to myofibroblasts in airways, vasculature, and peripheral lung tissue; while MMP-9 was localized to leukocytes infiltrating major airways. A rat intrapulmonary tracheal transplant model of obliterative bronchiolitis was used for further investigation. MMP-2 and MMP-14 were localized to myofibroblasts in allograft airways and pericellular gelatinolytic activity was demonstrated by in situ zymography. Intensive MMP inhibition (SC080, 5 mg/kg bid) for allograft recipients after establishment of airway fibrosis down-regulated procollagen gene expression without changing net collagen deposition, inducing myofibroblast cell death by day 35. In vitro, SC080 (4 μm) induced significant cell death in cultured myofibroblasts. Conversely, low-dose SC080 (2.5 mg/kg, qd) treatment in vivo increased MMP-2 gene expression and total MMP-dependent collagenolytic activity in allograft airways, decreasing total collagen and the number of myofibroblasts. In conclusion, the balance of MMP activity appears critical for myofibroblast persistence and continuous fibroproliferation in allograft fibrosis. Modification of the MMP-mediated tissue remodelling represents a possibility to induce regression and reversal of fibrosis. 132

152 5.2. Introduction Chronic allograft dysfunction after lung transplantation is manifested by obliterative bronchiolitis (OB), a fibroproliferative lesion in small airways associated with chronic rejection and variable alloantigenindependent factors (1, 341). OB and its clinical correlate BOS affects about 50% of lung transplant recipients within 5 years, limiting long-term success of lung transplantation (3). Once the fibrotic process of OB is initiated, immunosuppression and conventional treatments are usually ineffective (82). The traditional pathological perspective is that fibrosis is the end result of damage in a tissue not capable of regeneration and is fixed with no possibility of return to the pre-existing structure (282). However, accumulating evidence suggests that liver fibrosis is a dynamic, bidirectional process, wherein recovery with remodelling of scar tissue is possible (278, 279). In pulmonary fibrosis, a similar concept of bidirectional tissue remodelling and possible resolution of fibrosis is evolving (277). Using an intrapulmonary tracheal transplant model of OB, we have recently demonstrated that allograft airway fibrosis is also a dynamic process of tissue remodelling, in which collagen turnover and changes in cellular component of fibrosis continue after the establishment of the lumenal obliteration (Chapter 3). Moreover, we found that the dynamic remodelling process is pharmacologically manipulatable by a general inhibitor of matrix metalloproteinases (MMPs), which reduced total collagen and the number of myofibroblasts even when the treatment was initiated after the lumenal obliteration is established (Chapter 3). Further mechanistic understanding of tissue remodelling might lead to the development of novel strategies that induce fibrosis regression, the ideal treatment for OB after lung transplantation. MMPs are a group of key enzymes modulating tissue remodelling likely in a bidirectional way (287). A bulk of evidence suggests that MMPs are involved in aberrant remodelling such as tissue destruction in pulmonary emphysema (342), while the activity of MMPs could also be important to fibroproliferation through activation of profibrotic growth factors such as transforming growth factor (TGF)-β1 (320) and modulation of matrix microenvironment necessary for fibroblasts (268). 133

153 Conversely, studies of liver fibrosis indicate that MMPs are necessary for fibrosis regression through degradation of accumulated extracellular matrix (ECM). Upregulation of collagenolytic MMPs, MMP- 2 and MMP-14 (290, 343) have been associated with the spontaneous resolution of liver fibrosis. Gene transfection of collagenolytic MMPs, MMP-8 (291) and human MMP-1 (292) has been successfully tested to induce fibrosis regression in the liver. MMPs might also be important to the activity of myofibroblasts, a differentiated phenotype of fibroblasts that produces an excessive amount of collagen and other ECM molecules (328). In vitro studies suggest that the activity and phenotype of myofibroblasts are dependent on the matrix microenvironment (37), so that modulation of matrix molecules by MMPs may well alter the activity of myofibroblasts. Moreover, the local milieu of cytokines and growth factors could have significant impact on the relationship between MMPs and myofibroblasts, and subsequent tissue remodelling (294, 344, 345). Our previous study demonstrated that myofibroblasts in established allograft airway fibrosis express MMP-2 and MMP-14 (Chapter 3). Interestingly, these MMPs expressed by hepatic myofibroblasts appear to be involved in the resolution of liver fibrosis (343). Thus, we hypothesize that myofibroblast-mediated MMP-dependent tissue remodelling is an important process in OB after lung transplantation. In the present study, we examine myofibroblasts and their MMP expression in the lung tissue of BOS patients. Using an animal model of OB and cultured myofibroblasts, we further explore the role of myofibroblast-mediated MMP-dependent tissue remodelling and possible strategies to intervene in the remodelling process. 134

154 5.3. Results MMP-2 and MMP-14 are localized to myofibroblasts in BOS lungs We then examined the localization of MMP-2, MMP-9, and MMP-14 in the lung affected by BOS. MMP-9 is reported to be highly expressed in bronchoalveolar lavage of early stage BOS patients (227, 346, 347), while MMP-2 and MMP-14 are highly expressed in advanced fibrosis of intrapulmonary tracheal transplant of OB and mainly localized to myofibroblasts (Chapter 3). Double immunofluorescence labelling of BOS lungs localized both MMP-2 and MMP-14 to myofibroblasts in OB lesions (Figure 5-1). Conversely, MMP-9 was localized to other cells in the alveolar space (Figure 5-1), the airway lumen and the blood flow not to myofibroblasts in OB lesions. These cells positive for MMP-9 appeared to be leukocytes like neutrophils. These results indicate the involvement of MMP-2 and MMP-14 in myofibroblast-mediated MMP-dependent remodelling. 135

155 H&E Trichrome Figure 5-1. Localization of MMP-2, MMP-9, and MMP-14 in human OB after lung transplantation. Confocal microscopic examination demonstrates that MMP-2 and MMP-14 were mainly localized to myofibroblasts in OB lesions, while MMP-9 is expressed by cells in the alveoli (top, right). 136

156 Localization of MMP-2, MMP-14, and matrix degradation to myofibroblasts in experimental OB To investigate the mechanism of MMP-mediated tissue remodelling, we used a rat intrapulmonary tracheal transplant model of OB. Consistent with observations from our previous experiments (85), an isograft trachea from Lewis rat implanted in the lung parenchyma of a Lewis rat showed complete recovery of the epithelium with no evidence of rejection and obliterative fibrosis at post-operative day 21, while an allograft trachea from a Brown-Norway rat implanted in the lung parenchyma of a Lewis rat with full MHC mismatch resulted in complete obliteration of the graft with fibrous tissue (Figure 5-2A). The fibrous tissue in the allograft lumen was occupied with α-sma positive myofibroblasts and, like human OB lesions, MMP-2 and MMP-14 were mainly localized to these myofibroblasts in the allograft airway fibrosis at day 21 (Figure 5-2B). We used in situ gelatine zymography to further test whether these MMPs are involved in tissue remodelling through matrix degradation. DQ TM gelatin is gelatin that is so heavily labelled with fluorescein that the fluorescence is quenched. This substrate is efficiently degraded by gelatinases (e.g. MMP-2 and MMP-9) to yield highly fluorescent peptides. Since the increase in fluorescence is proportional to proteolytic activity, gelatinolytic activity can be visualized on cultured cells or frozen tissue sections using the technique of in situ zymography. A frozen tissue section of allograft airway fibrosis that had been stained with α-sma was incubated with DQ TM -gelatine to demonstrate that myofibroblasts in allograft airway fibrosis degrade extracellular gelatin (Figure 5-2C). Because MMP inhibitors eradicated the ECM degradation capacity (Figure 5-2C, b and c), degradation of gelatin is likely mediated by MMPs. SDS-gelatin zymography demonstrates increased active MMP-2 in allograft airways at day 21 (Figure 5-2D), suggesting that MMP-2 mediates the major gelatinolytic activity in allograft airway fibrosis. The matrix-degrading activity was mainly localized to the peri-cellular area of myofibroblasts, indicating the involvement of MMP-2 that forms a complex with MMP-14 on the cell membrane (348). 137

157 A Isograft Allograft B C D Figure 5-2. Localization of MMPs to myofibroblasts and peri-cellular matrix degradation. (A) Rat intrapulmonary tracheal transplantation was conducted from a Lewis donor to a Lewis recipient (isograft) or from a Brown-Norway donor to a Lewis recipient (MHC-full-mismatched allograft). In H&E staining, an isograft showed complete recovery of the epithelium with the open lumen in an isograft (left), while an allograft showed complete lumenal obliteration with fibrous tissue at postoperative day 21. Original magnification, 20x (top panels) and 200x (bottom panels). (B) Localization of MMP-2 and MMP-14 in myofibroblasts of established allograft airway fibrosis in an intrapulmonary tracheal transplant model of OB at day 21. Immunofluorescence labelling for α-sma and MMP-2 or MMP-14 demonstrates the localization of these MMPs mainly to myofibroblasts in established experimental OB lesions at post-operative day 21. Original 400x. (C) In situ zymography for an established experimental OB lesion. (a) In situ gelatin zymography using fluorrescence-quenched DQ TM -gelatin in combination with immunofluorescence labelling for α-sma in a frozen tissue section demonstrates active degradation of the extracellular matrix by myofibroblasts. Note the pericellular green fluorescence indicating active matrix degradation at the periphery of the cells. Negative controls that chelate Zn 2+ necessary for MMP activity (b) or directly inactivate MMPs (c) demonstrate the gelatinolytic activity is mediated by MMPs. Original magnification, 400x. (D) SDS-gelatin zymography for intrapulmonary tracheal grafts. Gelatin zymography demonstrates increased levels of active MMP-2 as well as pro-mmp-2 in allograft airways at day 21 as compared with those of isografts. MMP-9 is highly expressed in allograft airways at day 7. *An archival sample from our previous study (Chapter 3) is used as a positive control for pro-mmp

158 MMP inhibitor treatment for animals From day 21 to 35, five different regiments of the broad-spectrum MMP inhibitor, SC080 were tested: 2.5, 5.0, or 10 mg/kg/d, qd; or 5.0 or 10 mg/kg/d, bid with a treatment period and compared with vehicle-treated controls. All the treated animals were killed at day 35, 12 hours after the final drug administration. In general, plasma concentrations of SC080 peaked 1 hour after administration, followed by a rapid decline by 50-75% in the next 2 hours. Among the 5 regimens, the highest dose (HD) of 10 mg/kg/d, qd showed the most stable and highest plasma concentration: Cmax = 5.3 ± 0.69 μm, Cmin = 1.1 ± 0.09 μm, and AUC = The lowest dose (LD) of 2.5 mg/kg/d, qd that was selected based on our previous study (Chapter 4) showed the lowest plasma drug levels: Cmax = 2.0 ± 0.25 μm (mean ± SEM), Cmin = 0.05 ± 0.00 μm, AUC = 16.9 (Figure 5-3A) MMP inhibitor treatment modulates cellular and matrix components of established fibrosis SC080 treatment was initiated at day 21, when allograft obliteration was completed. By day 35, different regimens of SC080 treatment dramatically changed the fibrous tissue in the lumen (Figure 5-3B). The lumens of control allografts were occupied with dense deposition of ECM with a number of spindle-shaped myofibroblast-like cells and mononuclear cells. PSR collagen staining showed dense depositions and organized alignment of collagen fibers (Figure 5-3B(c)). Surprisingly, allografts of LD showed the lowest cellularity and loose density in ECM deposition in the lumen (Figure 5-3B(f)). The amount of total collagen was smaller than that of control (Figure 5-3B(g)). Conversely, allografts of HD showed the highest cellularity even when compared with control allografts (Figure 5-3B(v)). In PSR staining, allografts treated with the HD showed a large amount of total collagen (Figure 5-3B(w)). Intermediate doses between LD and HD showed moderate changes in cellularity and matrix deposition when compared with LD and HD. Morphometric quantification of PSR collagen staining demonstrated significantly small amounts of collagen in LD only when compared to that of control (Figure 5-3C). This result shows that only the LD has a significant net effect on total collagen accumulation. Since total collagen is determined by the balance between the production and degradation of collagen, we further examined procollagen gene expression and collagen degradation to elucidate the mechanisms. 139

159 A C B Figure 5-3. SC080 treatment for established allograft airway fibrosis. (A) Pharmacokinetics of SC080. SC080 was administered to allograft recipients from day 21 to 35 with 5 different treatment regimens and the plasma concentration of SC080 was measured towards the end of experiment. The plasma concentration of SC080 shows dose and timing dependent kinetics with the most stable concentration at 5 mg/kg b.i.d. and the lowest concentration at 2.5 mg/kg q.d. (B) The lowest dose of SC080 most effectively reduces extracellular matrix and cellularity in fibrosis. After treatment with 5 regimens of SC080 ( mg/kg/d, qd or bid) administered from day 21 to day 35, allograft tracheae were histologically examined at day 35 using H&E staining. All the treated and control allografts show lumenal obliteration (left panel, original 40x). However, fibrosis treated with the lowest dose of SC080 (LD, 2.5 mg/kg/d, qd) shows the lowest amount of ECM component and the least cellularity. The highest dose with intensive administration of SC080 (HD, 10 mg/kg/d, bid) shows increased cellularity (right panel, original 200x). (C) Morphometric quantification of PSR staining under polarized light. Morphometric quantification of PSR staining under polarized light demonstrates significantly reduced total collagen in LD only (*P < 0.05 in Tukey test (vs. vehicle control), n = 5 for each group). 140

160 Dose-dependent decrease in procollagen gene expression by SC080 treatment Gene expression analysis of procollagen α1(i) and α1(iii), precursors of type-i and type-iii collagen respectively, demonstrated a significant dose-dependent decrease in the gene expression of procollagen α1(i) and α1(iii) (Figure 5-4A, P < 0.01). This result suggests that the reduced total collagen at LD is not explained by changes in procollagen gene expression Low-dose SC080 upregulates gene expression of MMP-2 and increases collagenolytic activity On the other hand, collagenolytic activity analysis using DQ TM collagen (a formulated type-i collagen that provides fluorescence on degradation) in allografts treated with LD showed significantly higher collagenolytic activity than that of controls (Figure 5-4B, P < 0.05). The decreased fluorescence by additional pan-mmp inhibitor, 1,10-phenanthronine, demonstrated that most of the collagen degradation depends on MMP activity. An increase in MMP-dependent collagenolytic activity contributes to reducing total collagen at LD. We then examined gene expression of MMPs that are involved in collagenolytic activity. In general, collagen degradation mostly depends on MMPs (349), and only limited MMP members including MMPs-1 (not in rodents), -2, -8, -13 and -14 (323, 350) can degrade interstitial collagen. In gene expression analysis using real-time RT-PCR, only MMP-2 showed significant upregulation at LD. Gene expression of MMP-14 was maintained in LD, while gene expression of MMP-8 and MMP-13 was significantly lower at LD than other groups (Figure 5-4C). Immunofluorescence staining localized expression of MMP-2 and MMP-14 to myofibroblasts mainly similar to the results in controls (Figure 5-4C). These results suggest that MMP-2 and/or MMP-14 expressed by myofibroblasts mediated collagenolytic activity at LD (Figure 5-4D). 141

161 A Relative gene expression B Collagenolytic activity (AFU) 0 Vehicle ,10- Phenanthronin C Procollagen a1(i) qd qd bid qd qd 5 qd 5 bid 10 qd 10 bid VehicleC [Total dose (mg/kg/d)] * * 1 bid Vehicle control * LD HD SC080 D Vehicle Procollagen a1(iii) qd qd bid qd bid [Total dose (mg/kg/d)] C qd 5 qd 5 bid 10 qd 10 bid Figure 5-4. Low-dose SC080 increases MMP-2 gene expression and collagen degradation. (A) SC080 decreases procollagen gene expression dose-dependently. Quantitative real-time RT-PCR analysis for gene expression of procollagen α1(i) and α1(iii), precursors for type-i and type-iii collagen, respectively. Both genes showed dose-dependent decreases in expression (P < 0.01 for α1(i), P = for α1(iii) in regression analysis). Collagen degradation in the tissue is increased in low dose SC080 treatment. (B) Collagenolytic activity in tissue homogenates of allograft tracheae for control, LD (SC080, 2.5 mg/kg/d, qd) and HD (SC080, 10 mg/kg/d, bid). Collagenolytic activity was significantly high in LD (*P < 0.05 in Tukey test (vs. HD), n = 5 for each group). (C) Gene expression analysis of MMPs by real-time RT-PCR. Gene expression of MMP-2 was upregulated at LD and HD (P < 0.05). MMP-14 was upregulated at HD (P < 0.05) but maintained at LD. Gene expression of MMP-8 and MMP-13 showed low expression at LD (MMP-8, P < 0.05; MMP-13, P < 0.05). Localization of MMP-2 and MMP-14 to myofibroblasts at low-dose SC080 treatment. (D) Double immunofluorescence for α-sma and MMP-2 or MMP-14 demonstrates localization of these MMPs to myofibroblasts after low-dose SC080 treatment. Original, 200x 142

162 These results are paradoxical in that a broad-spectrum MMP inhibitor increased MMP activity. This might be explained by biological feedback. Ttransgenic deletion and chemical inhibition of MMPs are known to induce feedback upregulation of collagenolytic MMPs ( ). Unlike HD that maintained stable plasma concentration of SC080 through time, the low plasma concentration of SC080 in LD may allow for the upregulated MMPs to degrade ECM with a higher ratio than that without any MMP inhibition. Thus, the net effects of LD on MMPs and collagen are more likely to be the result of biological feedback than the direct pharmacological inhibition on MMPs MMP-inhibitor treatment decreases myofibroblasts in fibrous tissue We next examined whether modulation of MMPs by SC080 treatment affects myofibroblasts. Surprisingly, allografts at HD and LD showed a smaller number of myofibroblasts than that of control (Figure 5-4A). Semi-quantification of myofibroblasts confirmed this finding (Figure 5-4B, P < 0.01), suggesting that modulation of MMPs has a significant impact on the persistence of myofibroblasts in allograft airway fibrosis. Apoptosis was a speculated mechanism decreasing the number of myofibroblasts as apoptosis of vascular smooth muscle cells was induced by high-dose SC080 treatment in a pulmonary hypertension model (303). A small number of TUNEL-positive myofibroblasts were observed at HD (less than 1 cell/hpf average, 3-6 cells/tracheal graft in total). In addition to the direct effect of SC080 on myofibroblasts, a possible mechanism leading to apoptosis induction may involve the effect of cytotoxic lymphocytes that express Fas ligand. The massive inflammatory infiltration in HD-treated allografts (Figure 5-3A(q)) was indeed found to show significant T-cell as well as macrophage infiltration (Figure 5-6A). On the other hand, no TUNEL-positive myofibroblasts were observed at LD or control (Figure 5-5A). Mechanisms other than apoptosis are suspected to be involved in the reduction of myofibroblasts at LD. 143

163 Figure 5-5. The effect of SC080 treatment on myofibroblasts in vivo. (A) Immunofluorescence labelling for α-sma and TUNEL staining. The allograft lumen at LD (SC080, 2.5 mg/kg/d, qd) and HD (SC080, 10 mg/kg/d, bid) show reduced number of myofibroblasts. Myofibroblasts at LD show a shrunken morphology, whereas myofibroblasts at HD appear stretched. Double immunofluorescence for α-sma and TUNEL. Apoptotic myofibroblasts were exclusively observed at HD (arrow), Original, 400x. (B) Semi-quantification analysis of the numbers of myofibroblasts in fibrous tissue. The number of myofibroblasts in the allograft lumen was significantly decreased at LD and HD (*P < 0.01 versus control, n = 5 for each group). 144

164 Figure 5-6. Accumulation of macrophages and T cells in fibrous tissue after high-dose SC080 treatment. (A) Immunofluorescence labelling for macrophages (CD68) and myofibroblasts (left panels) demonstrates a large number of macrophages in the fibrous tissue at HD (SC080, 10 mg/kg/d, bid) than that in control or LD (SC080, 2.5 mg/kg/d, qd). Immunofluorescence labelling for T cells (CD3) and B cells (CD79a) demonstrates a large number of T cells at HD and control compared with that of LD. The number of B cells infiltrating in the fibrous tissue was small in all the groups. Original, 200x. (B) The number of macrophages was significantly increased at HD when compared with other groups (*P < 0.05 compared with the other groups, n = 5 in each group). (C) The number of T cells at HD was significantly larger than that at LD ( P < 0.05 compared with LD, n = 5 in each group). 145

165 SC080 treatment induces myofibroblast apoptosis in vitro Myofibroblast apoptosis in vivo after HD could be directly induced by SC080 or induced by infiltrating cytotoxic T cells. To dissect the mechanisms, we further conducted an in vitro experiment free of T cells or macrophages. Primary cultured rat pulmonary fibroblasts were stimulated by hrtgfβ1 (10 ng/ml) for 48 hours and the myofibroblast phenotype with evident expression of α-sma stress fibers was induced (Figure 5-7A). The percentage of myofibroblasts was significantly higher in the treatment group of hrtgf-β1 (10 ng/ml) (Figure 5-7B). Myofibroblasts were then cultured with vehicle or SC080 (4 μm). After 48 hours, the number of myofibroblasts was significantly reduced in the SC080-treated groups (Figure 5-8A, B(i)) and a larger number of myofibroblasts treated with SC080 were stained positive for TUNEL than those treated with the vehicle (Figure 5-8A, B(ii)). The result demonstrates that SC080 can directly induce apoptosis in myofibroblasts. To examine the MMP inhibition-dependent apoptosis of myofibroblasts, we measured active and latent TGF-β1 in cell culture medium. Although activation of TGF-β1 can be mediated by MMP-2 and/or MMP-9 (320) and play important roles in myofibroblast survival (345), the active form of endogenous TGF-β1 was not significantly different between groups (Figure 5-8C). The result suggests that the effect of MMP inhibition on myofibroblasts is mediated by a mechanism other than inhibition of TGFβ1 activation. 146

166 A B α-sma positive cell (%) * 0 Ctrl 5 ng/ml 10 ng/ml hrtgf-β1 (ng/ml) Figure 5-7. Induction of the myofibroblast phenotype in primary cultured pulmonary fibroblasts. (A) After culturing pulmonary fibroblasts (as indicated in Methods), human recombinant TGF-β1 was added to the culture medium. After 48 hours, the phenotype of myofibroblasts was induced with evident α-sma stress fiber (right bottom) as well as increased collagen type I production (right top). Original, 200x. (B) Human recombinant TGF-β1 induces myofibroblasts. Human recombinant TGF-β1 at 10 ng/ml induced a significant percentage of myofibroblasts compared with controls and 5 ng/ml treatment (*P < 0.05, n = 4). 147

167 A Vehicle SC080 a-sma TUNEL Hoechst B (i) (ii) C (i) (ii) Figure 5-8. SC080 induces apoptosis in myofibroblasts in vitro. (A) After induction of the myofibroblast phenotype, myofibroblasts were cultured with SC080 (4 μm) or vehicle for 48 hours. Double immunofluorescence labelling for α-sma and TUNEL demonstrates evident apoptosis in myofibroblasts induced by SC080. (B) Semi-quantitative analysis of the number of myofibroblasts and ratio of TUNEL positive myofibroblasts. The number of myofibroblasts was significantly reduced after SC080 treatment (*P < 0.05). Semi-quantification of TUNEL positive myofibroblasts demonstrated a significantly higher ratio of apoptosis in myofibroblasts treated with SC080 (*P < 0.05). (C) SC080 treatment has no effect on total and active TGF-β1 levels in vitro. ELISA for total and active TGF-β1 in culture medium of myofibroblasts after 48 hours of incubation with vehicle or SC080 demonstrated no significant difference between groups (P > 0.05, n = 6). 148

168 5.4. Discussion MMP-mediated tissue remodelling of allograft airway fibrosis is potentially bidirectional The traditional pathological perspective of fibrosis is that fibrosis is the end result of damage in a tissue not capable of regeneration and is fixed with no possibility of return to the pre-existing structure (282). However, fibrosis is suggested to be a dynamic, bidirectional process, wherein recovery with remodelling of scar tissue might be possible ( ). Since MMPs are the key enzymes involved in various physiological and pathological remodelling processes and myofibroblasts are the central cells that express MMPs in established fibrosis, we hypothesized that myofibroblast-mediated MMPdependent tissue remodelling is an important process in OB after lung transplantation. The most important finding in this study is bidirectional MMP-mediated tissue remodelling, the modulation of which could drive the process toward fibrosis regression or apoptosis of myofibroblasts. MMP-2 and MMP-14 expressed by myofibroblasts appears to be essential to the survival of myofibroblasts, while upregulation of these MMPs could facilitate degradation of interstitial collagen. In other words, the apparently fixed fibrosis containing myofibroblasts is likely undergoing incessant tissue remodelling depending on balanced MMP activity. Modulation of the balance of the MMP activity could be a novel therapeutic strategy for advanced stages of OB after lung transplantation MMP-2 and MMP-14 expressed by myofibroblasts are the key MMPs involved in tissue remodelling of OB Even though this major finding was obtained in an animal model of OB, we also found that it could be translated to human OB lesions, in which myofibroblasts express MMP-2 and MMP-14, too (Figure 5-1). In contrast, MMP-9 was most likely to be localized to inflammatory cells such as neutrophils consistent with reports in early human BOS patients (35, 227) as well as our previous animal work (Chapter 3). In addition to the role of MMP-9 in airway neutrophilia and inflammation, we have recently demonstrated that MMP-9 is highly expressed by circulating fibrocytes that contribute to the myofibroblast population in allograft airways (Chapter 4). The information on different MMPs is 149

169 important for the design of MMP modulation strategies. For example, MMP-9 may be targeted by an MMP inhibitor to attenuate the continuous underlying inflammation and recruitment of circulating fibrocytes, while it may be beneficial to preserve or even enhance the activity of MMP-2 and MMP-14, which appears to degrade interstitial collagen and other ECM (Figure 5-2). Although we examined these three MMPs in human tissue samples, there are more than 20 MMPs and other proteinases involved in ECM turnover (354). We selected these MMPs because they showed different expression patterns in our previous study. However, to refine MMP modulation strategies, further investigation is necessary to evaluate other MMPs Upregulation of collagenolytic MMPs is important to fibrosis regression In the present study, enhanced expression of MMP-2 and increased collagenolytic activity were obtained by low-dose SC080 treatment. This effect appears to be due to biological feedback often seen in transgenic deletion or pharmacological inhibition of MMPs ( ). An important indication of this finding is that upregulation of collagenolytic MMPs can promote degradation of interstitial collagen and potentially drive tissue remodelling towards fibrosis regression. Such collagenolytic properties of some MMPs have been applied in experimental liver fibrosis. Gene transfection of collagenolytic MMPs, MMP-8 (291) and human MMP-1 (292) has been demonstrated to induce fibrosis regression in the liver Treatment of underlying tissue injury may be necessary to realize fibrosis regression Unlike liver fibrosis models, however, we did not observe actual regeneration of the airway architecture despite reduced collagen in the graft lumen after the low-dose SC080 treatment. Compared with autonomous or gene therapy-mediated fibrosis regression in the liver, there are two limitations in this study. Firstly, underlying inflammatory and immune responses were not treated in this study. In the liver, fibrosis regression is usually observed after attenuating underlying injurious mechanisms such as treatment of alcoholic and viral hepatitis (278, 279) and self-resolving chemical injury in animal models (290, 343). Similarly, bleomycin-induced pulmonary fibrosis appears to recover 150

170 spontaneously to some extent for its temporal tissue injury (284). In contrast, in human OB lesions and an OB animal model, alloantigen persists long term after total obliteration of the airway lumen we observed continuous T-cell infiltration in the lumen (Figure 5-6). Thus, the MMP modulation strategy might be more effective by combining immunosuppression and/or anti-inflammatory treatments. Secondly, we did not observe epithelial regeneration in this study. We speculate this is partly because we started MMP inhibitor treatment at post-operative day 21, when the epithelium is also completely lost, and partly because we did not treat underlying alloimmune responses that would inhibit epithelial regeneration even if the epithelial progenitor cells survived. Further concept-proving studies are necessary to demonstrate the possibility of airway regeneration in OB after transplantation Induction of myofibroblast apoptosis is another therapeutic niche for MMP modulation Novel therapeutic potential also lies in targeting myofibroblasts through modulation of the MMP system. Intensive MMP inhibition induced myofibroblast apoptosis in vivo (Figure 5-5) and this appears to be a direct effect of the MMP inhibitor on myofibroblasts as suggested in vitro (Figure 5-8). Although the effect of cytotoxic T cells on myofibroblasts was not excluded in this study, a recent study has demonstrated that myofibroblasts could escape the immune surveillance and selectively survive in tissue fibrosis (355). The mechanism appears to be independent of autocrine production of TGF-β1 by myofibroblasts and the detailed mechanism remains elusive. We speculate the effect of MMPs on myofibroblast apoptosis is mediated by the modulation of cell adhesion. In general, myofibroblast survival is considered to be regulated by growth factors (mainly TGF-β) and signalling through a focal adhesion complex including cell surface integrin, a receptor for ECM (356, 357). It is possible that excessive stabilization of the ECM may decrease differentiation and survival signals to myofibroblasts, for example by decreasing signalling dependent on focal adhesion kinase (358) or down-regulating myofibroblast-activating fibronectin (e.g. ED-A fibronectin) (359). This hypothesis is apparently contradictory to a previous report, in which apoptosis of pulmonary myofibroblasts was induced by hepatocyte growth factor-dependent upregulation of MMP-2 and MMP inhibition rescued myofibroblast from apoptosis (294). However, once again, the contradictory effects of MMPs on 151

171 myofibroblasts could represent their bidirectional effect in tissue remodelling. Further investigation is necessary at the cell biology level to explore the mechanisms that regulate myofibroblast survival in association with MMPs Conclusion In conclusion, we demonstrated that tissue remodelling of allograft airway fibrosis is potentially bidirectional and is regulated by MMPs, particularly MMP-2 and MMP-14 expressed by myofibroblasts. Our finding indicates two important therapeutic directions to treat advanced allograft airway fibrosis: enhanced collagenolytic MMPs to induce fibrosis regression and targeting MMPdependent survival signals for myofibroblasts. Further investigation is necessary to understand the detailed molecular mechanisms and design of potential therapeutic strategies. 152

172 Chapter 6 Myofibroblast-T-cell interaction supporting persistent post-transplant allograft airway fibrosis A part of this manuscript was presented at the International Society for Heart and Lung Transplantation 27 th annual meeting and scientific sessions (April 28, 2007, San Francisco, CA). The abstract was published in the Journal of Heart and Lung Transplantation, 2007;26(2):S

173 6.1. Abstract Obliterative bronchiolitis (OB) is an obliterative airway fibrosis associated with chronic rejection after lung transplantation. We hypothesize that continuous alloimmune responses are the underlying mechanisms that support myofibroblast activation and counteract therapeutic strategies. A rat intrapulmonary tracheal transplant mode of OB was used for a translational experiment. Fibrosis was created in isografts by enzymatic epithelial denudation and compared with epithelium-denuded allografts and untreated allografts at day 28. Despite complete obliteration of isografts, myofibroblasts and collagen deposition were significantly less than those of allografts. Cultured myofibroblasts maintained their phenotype only when directly co-cultured with T cells, demonstrating the importance of direct myofibroblast-t-cell contact. Next, established allograft airway fibrosis of animal OB at day 28 was treated with cyclosporine, or an MMP inhibitor, SC080, or a combination of cyclosporine and SC080 for 2 weeks. Strikingly, the combined treatment induced partial fibrosis regression (P < 0.01) in association with significant reduction in T-cell infiltration and the number of myofibroblasts. We previously demonstrated that SC080 modulates tissue remodelling of established allograft airway fibrosis. Thus, the present study demonstrates that modulation of tissue remodelling in combination with treatment of underlying ongoing tissue injury is essential to reverse apparently established allograft airway fibrosis. 154

174 6.2. Introduction Chronic allograft dysfunction after lung transplantation is manifested by obliterative bronchiolitis (OB), fibroproliferative changes affecting small airways. OB and its clinical correlate, BOS affects about 50% of lung transplant recipients within 5 years, limiting the long-term success of lung transplantation (3). OB has been considered to be an end result of tissue damage due to chronic rejection and variable alloantigen-independent factors (1, 341). Indeed, immunosuppression and other conventional treatments are usually ineffective once the diagnosis of OB/BOS is established (82). However, the concept of bidirectional tissue remodelling and possible resolution of fibrosis is evolving in the liver and lung fibrosis ( ). Using an intrapulmonary tracheal transplant model of OB, we have recently demonstrated that allograft airway fibrosis is not a static but a dynamic process of potentially bidirectional tissue remodelling (Chapter 3). Matrix metalloproteinases (MMPs) are a group of key enzymes modulating tissue remodelling likely in a bidirectional way (287). Low-doses of a general inhibitor of matrix metalloproteinases (MMPs) upregulated expression of MMP-2, increased tissue collagenolytic activity, and reduced total collagen even after the allograft lumen was completely obliterated (Chapter 5). The result suggested that tissue remodelling is modulated toward regression. Moreover, we found MMP-2 is expressed by myofibroblasts both in human OB lesions and an OB animal model (Chapter 5). As such, the strategy of modulating tissue remodelling by using low-dose MMP inhibition appears to be a clinically relevant approach. In our previous study, however, fibrosis regression (re-canalization of the airway lumen) and regeneration of normal tissue architecture in transplanted allograft airways was not achieved by modulating tissue remodelling using a general MMP inhibitor (Chapters 3, 5). We speculated that this was because of immune-mediated injuries that are continuous after establishment of fibrosis. Indeed, in fibrosis of the liver and the lung, resolution of fibrosis always accompanies cessation of underlying injurious mechanisms (e.g. successful treatment of viral hepatitis) (278, 279, 284, 290, 343). Thus, we hypothesize that ongoing immune-mediated tissue injury is an important factor preventing resolution of 155

175 allograft airway fibrosis after its establishment. Firstly, to examine the impact of continuous immune responses on allograft airway fibrosis, we compare human lungs affected by OB after transplantation with those affected by transplant-unrelated OB, which is free of alloimmune responses. We also compare allograft fibrosis of intrapulmonary tracheal transplantation with fibrosis in isografts that develops after enzymatic epithelial denudation, which is known to induce similar obliterative fibrosis without alloimmune responses (44, 228). We then examine whether immunosuppression has an additive effect on tissue remodelling. We combine a conventional immunosuppressor, cyclosporine, with a low-dose MMP inhibitor that was tested in our previous study (Chapter 3). We herein for the first time demonstrate partial regression of established allograft airway fibrosis after transplantation Results Alloimmune-related animal OB lesions accompany continuous T-cell infiltration, but isograft OB lesions do not We examined the impact of alloimmune responses on tissue remodelling using an intrapulmonary tracheal transplant model of OB. In subcutaneous tracheal transplantation models of OB, syngenic grafts have been demonstrated to develop fibrosis similar to that of allografts if the tracheal epithelium is denuded by enzymatic treatment before transplantation (44, 228). This technique was originally developed to demonstrate the importance of the epithelium in protecting graft airways from fibrosis. We applied this method to examine differences between allograft airway fibrosis and isograft airway fibrosis that is free of alloimmune responses. We used an intrapulmonary tracheal transplant model of OB (127), which is considered to reflect tissue remodelling processes under the influence of the pulmonary milieu; as well as a more relevant model to human OB/BOS after lung transplantation (Chapter 3). We compared epithelium-denuded isografts with epithelium-denuded and untreated allografts. Epithelium-denuded isografts developed obliterative fibrosis similar to epithelium-denuded or 156

176 untreated allografts by day 28 (Figure 6-1A). In all the groups, epithelium was lost and the graft lumen was almost completely obliterated. Morphometric quantification of lumenal obliteration did not show significant difference among groups (Figure 6-1C). Since the epithelium is not the only allogenic component in tracheal transplantation and some donorderived tissues, such as cartilage, survive after epithelial denudation and fibrosis development, alloimmune responses are likely to persist in allografts. We then examined whether the difference in the number of myofibroblasts at day 28 derive from the difference in the initial alloimmune-mediated myofibroblast activation process or from the difference in alloimmune responses which may prevent myofibroblasts from disappearing in fibrous tissue. Immunofluorescence labelling demonstrated a larger number of T cells in the graft lumen of untreated allografts and epithelium-denuded allografts compared with that of epithelium-denuded isografts (Figure 6-1B). Semi-quantification of CD3 + T cells in the lumen demonstrated a significantly larger number of T cells in untreated allograft controls and epithelium-denuded allografts at day 28 (Figure 6-1D), suggesting the persistence of alloimmune responses at day

177 A H&E (20x) H&E (200x) B CD3 CD68 Hoechst (Fibrosis) CD3 CD68 Hoechst (Perigraft) Figure 6-1. The effect of epithelial denudation on graft obliteration and T-cell infiltration. (A) Tracheal grafts of all the groups show almost total obliteration of the airway lumen (top panels, original 20x). Higher magnification demonstrates eosinophilic spindle-shaped myofibroblast-like cells and infiltrating round mononuclear lymphocyte-like cells in untreated allografts (bottom panles, left and middle), while epithelium-denuded isografts do not show myofibroblast-like cells or lymphocytelike cells (bottom panel, right, original 400x). (B) Allograft airway fibrosis is associated with T-cell and macrophage infiltration in the fibrous tissue (top panels) and in the peri-graft area (bottom, left and middle). Original, 400x. (C) Lumenal obliteration ratio after intrapulmonary tracheal transplantation at day 28. Morphometric quantification demonstrated no significant difference in the obliteration ratio of the graft lumen among groups at day 28. (D) Semi-quantification of T cells demonstrates a significantly smaller number of T cells in the lumen of epithelium-denuded isografts as compared with the other groups (*P < 0.001). 158

178 Alloimmune-related animal OB lesions are associated with increased myofibroblasts and collagen deposition compared with alloimmune-free animal OB lesions The quality of fibrosis the cellular and extracellular matrix components, however, appeared to be different in the isograft fibrosis, which showed less myofibroblast-like and lymphocyte-like cells and looser fibrosis (Figure 6-1A, bottom panels). We then examined myofibroblasts using immunofluorescence labelling for α-sma. In the lumen of untreated allografts and epithelium-denuded allografts, a number of myofibroblasts were observed at day 28, whereas in the lumen of epitheliumdenuded isografts, myofibroblasts were not observed frequently despite almost total obliteration of the lumen with fibrous tissue (Figure 6-2A). Semi-quantification of the number of myofibroblasts demonstrated a significantly smaller number of myofibroblasts in epithelium-denuded isografts when compared to the other two groups (Figure 6-2C, P < 0.01). These results demonstrate that fibrosis development and an increase in myofibroblasts are regulated at different levels. Alloimmune responses are likely to be critical to an abnormal myofibroblast increase in allograft airways. Myofibroblasts are the major cells producing extracellular matrix in fibrosis (37). In PSR staining for collagen, obliterative fibrosis in epithelium-denuded isografts showed a smaller amount of total collagen (Figure 6-2B). Morphometric quantification demonstrated significantly decreased amounts of collagen in epithelium-denuded isografts (Figure 6-2D, P < 0.05). The result demonstrates that alloimmune responses have a significant effect on the amount of collagen accumulating in the lumen of graft airways. 159

179 A Vimentin α-sma Hoechst B PSR Figure 6-2. The effect of alloimmune responses on myofibroblasts and collagen. (A) Allograft fibrosis is associated with a large number of myofibroblasts. Immunofluorescence labelling for vimentin (fibroblast) and α-sma (myofibroblast) demonstrates a large number of myofibroblasts in the lumen of untreated allografts (left) and epithelium-denuded allografts (middle); while in epitheliumdenuded isografts, the number of myofibroblasts is small. The number of total fibroblasts (myofibroblasts and α-sma negative fibroblasts) appears to be similar among groups. Original, 400x. (B) Allograft fibrosis is associated with a large amount of collagen deposition in the lumen. PSR collagen staining demonstrates a large amount of collagen in the lumen of untreated and epitheliumdenuded allografts at day 28, while epithelium-denuded isografts show a small amount of collagen deposition in the lumen. Original, 400x. (C) Allograft fibrosis is associated with a large number of myofibroblasts. Semi-quantification of myofibroblasts in the graft lumen demonstrates a significantly larger number of myofibroblasts in the epithelium-denuded isograft group as compared with the other groups at day 28 (*P < 0.01). (D) Morphometric quantification of PSR collagen staining. Morphometric quantification of PSR collagen staining demonstrates a significantly smaller amount of collagen deposition in the lumen of the epithelium-denuded isografts as compared with untreated and epithelium-denuded allografts at day 28 (*P < 0.05). 160

180 Although myofibroblasts might persist independent of ongoing alloimmune responses, our previous study using the same intrapulmonary allograft tracheal transplant model demonstrated that cyclosporine treatment from day 14 significantly reduces the number of myofibroblasts by day 28 when compared with that of control allografts (Chapter 4). In this model, the airway epithelium is totally lost and the fibrotic process in allografts is initiated by day 14 (85). After this time point, immunosuppression does not effectively prevent lumenal obliteration (Chapter 4), indicating that the effect of cyclosporine on the number of myofibroblasts after day 14 is not due to protection of the airway epithelium but is more likely to be due to the effect on the interaction between immune responses and myofibroblasts. Collectively, our present and previous data suggest that ongoing alloimmune responses are important to the persistence of myofibroblasts even after the initiation of fibrotic process Cytokines and chemokines did not show significant difference between isograft OB and allograft OB lesions We speculated that molecules involved in the interaction between myofibroblasts and T cells are important to the persistence of myofibroblasts. We conducted real-time RT-PCR analysis for microdissected tissue of the graft lumen (i.e. fibrous tissue). Interestingly, however, this analysis demonstrated no significant difference among groups in potentially pro-fibrotic cytokines and growth factors (IL-1b, TNF-a, IL-13, and TGF-β1) and lymphoid chemokines (CXCL12 and CXCL13) (Figure 6-3). Although these factors are considered to be involved in the communication between mesenchymal cells, such as fibroblasts/myofibroblasts and immune responsive cells such as T cells (327), the result suggests that these soluble factors are not important to continuous activation of myofibroblasts in vivo. We then speculated that direct T-cell-myofibroblast interaction might be an important mechanism for myofibroblast persistence in the allograft lumen. To further examine the mechanisms whereby ongoing alloimmune responses support myofibroblasts in fibrous tissue, we conducted in vitro studies. 161

181 2 IL-1β 2 TNF-α Untreated allograft Denuced alloraft Denuded isograft 0 Untreated allograft Denuced alloraft Denuded isograft Relative gene expression Mean IFN-γ Untreated allograft Denuced alloraft Denuded isograft TGF-β1 Untreated allograft Denuced alloraft Denuded isograft 2.4 CXCL CXCL Untreated allograft Denuced alloraft Denuded isograft 0 Untreated allograft Denuced alloraft Denuded isograft Figure 6-3. Cytokines and chemokines are not different between isograft fibrosis and allograft fibrosis. Real-time RT-PCR analysis. None of the examined soluble factors that could affect the interaction between immune responses and myofibroblast activation showed significant difference among groups. 162

182 Direct contact with T cells facilitate persistence of the myofibroblast phenotype in vitro To examine whether T cells have a significant impact on the persistence of already differentiated myofibroblasts, we used an in vitro co-culture system of myofibroblasts and T cells with Transwell with 0.4 μm pores that can physically separate these cells but allow for communication through soluble factors. In a flow cytometric analysis, T cells isolated from PBMCs were positive for CD3 in 98% of cells, and among CD3 + T cells, 70% and 30% were positive for CD4 and CD8, respectively, and 60% of CD4 + T cells were negative for CD45RC showing the memory T-cell phenotype. The phenotype of myofibroblasts was first induced by stimulating primary cultured pulmonary fibroblasts with TGF-β1. The myofibroblasts were then cultured in a 12-well plate at the bottom of the Transwell. Myofibroblasts were cultured alone, directly co-cultured with T-cells at the bottom of the Transwell, or separately cultured with T-cells in the top chamber. After 96 hours of culture under optimised conditions (1% FBS), many myofibroblasts cultured alone were clumped and lost α-sma expression, losing the myofibrobast phenotype (Figure 6-4A). These myofibroblasts were, however, almost exclusively negative for TUNEL staining. Myofibroblasts separately cultured with T-cells also showed a clumped morphology, while myofibroblasts directly co-cultured with T-cells maintained their phenotype, and their number appeared to be larger than those cultured alone (Figure 6-4B). Morphometric quantification demonstrated that a significantly larger number of myofibroblasts persisted when directly co-cultured with T cells (Figure 6-4C, P < 0.01), demonstrating the importance of direct contact with T cells in maintaining the phenotype of myofibroblasts. 163

183 C Figure 6-4. Direct co-culture of myofibroblasts with T cells facilitates the persistence of the myofibroblast phenotype in vitro. (A) Study design of a myofibroblast co-culture experiment with PBMC-derived T cells. (B) Myofibroblasts cultured in 1% FBS for 96 hours alone or with T cells across a 0.4 μm-pore Transwell show clumping and loss of α-sma expression, while myofibroblasts cultured with T cells show preserved phenotype and their number.direct co-culture of myofibroblasts with T cells facilitates the persistence of the myofibroblast phenotype. (C) Cells with α-sma expression without clumping were morphometrically quantified. Myofibroblasts directly co-cultured with T cells show a significantly larger number than those cultured without T cells (*P < 0.01). 164

184 Modification of tissue remodelling in combination with immunosuppression induces regression of established obliterative fibrosis Given the importance of T cells in the persistence of myofibroblasts, the cells responsible for excessive production of ECM molecules in pathological fibrosis, it is possible that immunosuppression in combination with a treatment that modulates tissue remodelling may synergistically modify the dynamic process of tissue remodelling in allograft airway fibrosis and drive it towards fibrosis regression. We subsequently treated animals that had received allograft airways with immunosuppression with a low-dose of an MMP inhibitor, SC080, after the lumenal obliteration was already established. Twenty-eight days after intrapulmonary tracheal transplantation, allograft recipients were randomized into day-28 controls, treatment groups of cyclosporine alone, SC080 alone, a cyclosporine/sc080 combination, and vehicle controls. Indeed, the day-28 control animals showed near total obliteration of the allograft lumen (Figure 6-5A). Strikingly, at day 42, the allografts treated with the combination of cyclosporine and SC080 showed partial opening in the graft lumen (Figure 6-5A). Cyclosporine alone or SC080 alone did not show partial opening in the lumen of allografts. Morphometric quantification demonstrated significantly reduced obliteration in the combined treatment group, even in comparison with day 28 controls (Figure 6-5B, P < 0.01). These results suggest that combination of immunosuppression and MMP inhibition induced fibrosis regression after the establishment of allograft airway fibrosis. 165

185 Figure 7-1. Allograft airways after treatment with vehicle, SC080, cyclosporine, and a Figure 6-5. Combination of SC080 and cyclosporine induces fibrosis regression in vivo. (A) Treatment was initiated at day 28, when allograft airway fibrosis is established in untreated control animals. Only allografts treated with the combination of SC080 and cyclosporine showed partial opening at day 42. (B) Morphometric quantification of lumenal obliteration in allografts. Combined treatment of SC080 and cyclosporine resulted in a significant decrease in the lumenal obliteration ratio (*P < 0.01) as compared with other treatment groups as well as untreated controls at day

186 Immunosuppression reduces the number of myofibroblasts and changes their morphology Immunofluorescence labelling for α-sma demonstrated a reduced number of myofibroblasts in the treatment groups of SC080, cyclosporine, and their combination (Figure 6-6A, top panel). Morphologically, myofibroblasts in the combined treatment group were small and only a small portion of the α-sma signals had identifiable nuclei (Figure 6-6A, bottom panel). Myofibroblasts in the cyclosporine group and SC080 group also showed similar morphology. Semi-quantification of the number of myofibroblasts confirmed significantly smaller numbers of myofibroblasts in SC080 and cyclosporine treatment groups compared with the vehicle control (Figure 6-6B). Two-way ANOVA demonstrated the significant contribution of both cyclosporine and SC080 to the reduced number of myofibroblasts (P < 0.01). To examine the mechanisms by which the number of myofibroblasts was changed, we used double immunofluorescence labelling for α-sma and TUNEL. A small number of TUNEL positive signals were localized to myofibroblasts in the combined treatment group (Figure 6-6C, right) as well as treatment groups of cyclosporine and SC080 alone. Interestingly, in the SC080 treatment group, most TUNEL positive signals were localized to clustered cells negative for α-sma and the TUNEL positive cells were demonstrated to be T cells by double staining for CD3 and TUNEL (Figure 6-6C, left). Counting of all the TUNEL positive signals and its co-localization in α-sma positive cells revealed a significant difference in total TUNEL signals (Figure 6-6D(i)). TUNEL positive myofibroblasts were not observed in the control group, although the difference was not significant (Figure 6-6D(ii)). The results imply that induction of apoptosis is a possible mechanism in reducing myofibroblasts in cyclosporine treatment and its combination with SC080. However, the majority of myofibroblasts seems to disappear without obvious cell-death signals, possibly due to losing the phenotype of myofibroblasts by down-regulating α-sma expression as was observed in vitro (Figure 6-4). 167

187 Figure 6-6. The effect of CsA, SC080, and their combination on myofibroblasts. (A) Treatment with cyclosporine, SC080, and their combination reduced the number of myofibroblasts. (B) The number of myofibroblasts in obliterative fibrosis of allograft airways. Treatment with cyclosporine, SC080, and their combination from day 28 to 42 significantly reduced the number of myofibroblasts in obliterative fibrosis of allograft airways (*P < 0.01). (C) Apoptosis of myofibroblasts and T cells. (i) Double staining for TUNEL and α-sma demonstrates TUNEL positive myofibroblasts in the allograft fibrosis of the cyclosporine + SC080 combined treatment group. (ii) Double staining for TUNEL and CD3 demonstrates clustered cell death in T cells in the SC080 treatment group. (D) Semiquantification of general cell death and cell death in myofibroblasts. (i) Counting of TUNEL positive cells demonstrates significantly greater cell death in the treatment groups of SC080, cyclosporine, and their combination when compared with that of controls. *P < 0.05 compared with vehicle control. (ii) Semi-quantification of double positive cells for TUNEL and α-sma demonstrates a small number of apoptotic myofibroblasts in the treatment groups of SC080, cyclosporine, and their combination the difference was not significant when compared with that of controls. 168

188 SC080 treatment affects T-cell apoptosis while cyclosporine reduces collagen deposition Clustered T-cell apoptosis is a novel and interesting finding of SC080 treatment. Thus, we also examined T-cells infiltrating in the allograft lumen in each treatment group. Interestingly, not only cyclosporine and combined treatment groups but also the SC080 treatment group demonstrated reduced numbers of T cells (Figure 6-7A). Semi-quantification demonstrated reduced numbers of T- cells in these three treatment groups as compared with that of vehicle controls (Figure 6-7B, P < 0.01). Two-way ANOVA demonstrated the significant contribution of both cyclosporine (P < 0.001) and SC080 (P < 0.05) to the reduced number of T cells. Although the mechanism whereby low-dose SC080 treatment reduces T cell infiltration is unclear, this might indicate a complex relationship between tissue remodelling and T cells in allograft airway fibrosis. We also examined the effect of the combined treatment on interstitial collagen deposition. PSR staining for collagen demonstrated reduced amounts of collagen in the cyclosporine-treated group and combined treatment group (Figure 6-7C, top panel). On the other hand, the result of the SC080 treatment group was variable and appeared to be not as effective as the same dose tested from day 21 to 35 as tested before (Chapter 5). Morphometric quantification of collagen deposition using PSR staining did not reach statistical significance at day 42 when comparing the 3 groups (Figure 6-7D, P = 0.072). However, the treatment groups showed a trend of reduced collagen deposition, suggesting the impact of immunosuppression as well as MMP inhibition on collagen turnover and tissue remodelling in established fibrosis. 169

189 A B 500 Number of T cells (/10hpf) * * * 0 C Vehicle CsA SC080 Comb CsA+SC080 Treatment day C Vehicl SC080 D CsA CsA+SC080 PSR positive area (%) Ctrl d42 Vehicle CsA SC080 Combo CsA+SC080 Treatment day Figure 6-7. The effect of SC080, CsA, and their combination on T cells and collagen deposition. (A) Immunofluorescence labelling for T cells demonstrates a smaller number of T cells in the graft lumen in the treatment groups of SC080, cyclosporine, and their combination as compared with the control group. Original, 200x. (B) SC080, cyclosporine, and their combination reduce T-cell infiltration in the lumen. Semi-quantification of T cells in the graft lumen demonstrates a significantly smaller number of T-cells in the treatment groups as compared with that of vehicle controls (*P < 0.01). (C) PSR staining for collagen in the allograft lumen. PSR collagen staining demonstrates a reduced amount of collagen by cyclosporine and combined treatment of cyclosporine and SC080 when compared with controls (top panels). SC080 did not obviously decrease the amount of collagen in the allograft lumen. Original, 200x. (D) Morphometric quantification of PSR staining for collagen. Morphometric quantification of PSR staining did not demonstrate a significant difference among groups. The clear trend of reduced collagen in cyclosporine and combined treatment of cyclosporine and SC080 was observed (P = 0.072). 170

190 The therapeutic effect of combined treatment is reduced in the long term Expecting further regression of fibrosis, we extended the treatment period to up to 4 weeks (i.e. from day 28 to day 56). The therapeutic effects of combined treatment were, however, reduced by day 56 when compared with the treatment for 2 weeks. In H&E staining, most treated grafts showed almost complete obliteration (Figure 6-8A) with no statistically significant difference in the lumenal obliteration ratio (Figure 6-8B). The number of myofibroblasts did not show a significant difference either. T-cell infiltration was significantly reduced in all the therapeutic treatment groups; particularly cyclosporine alone and the combined treatment by day 56 (Figure 6-8C). It appears that the immunosuppressive effects of these agents are still maintained at day

191 A B 100 NS Lumenal obliteration (%) C Vehicle d56 Ctrl CsA SC080 CsA+SC080 Combo Treatment day Number of T cells (/10hpf) * * * 0 Ctrl-56 Vehicle CsA-56 SC-56 SC080 W-56 CsA+SC080 Treatment day Figure 6-8. Allograft airways at day 56 after 4 weeks of treatment of CsA, SC080, and their combination. (A) H&E staining demonstrates the reduced therapeutic effect of combined treatment at day 56. The allograft lumen was almost completely obliterated in all the treatment groups at day 56 after 4 weeks of treatment. Original, 20x. (B) Morphometric quantification of allograft airway obliteration at day 56 demonstrates a reduced effect of the combined treatment of cyclosporine and SC080, which are no longer significantly different from other groups. (C) Semi-quantification of T- cells infiltrating in the allograft lumen demonstrates their significantly reduced number in cyclosporine, SC080, and combined treatment groups as compared with that of vehicle controls at day

192 6.4. Discussion Treatment of underlying injury is essential to reverse fibrosis Most importantly, the present study for the first time demonstrated regression of fibrosis in established allograft airway fibrosis. The concept of bidirectional tissue remodelling and possible resolution of fibrosis is evolving in liver and lung fibrosis ( ). The result of the present study further supports possible reversal of tissue remodelling, encouraging future investigation targeting advanced airway fibrosis. Since induction of fibrosis regression in fibrosis of the liver and the lung always accompanies attenuation or resolution of underlying injurious mechanisms (278, 279, 284, 290, 343), the result of the present study emphasizes the importance of suppressing underlying injurious mechanisms to treat advanced fibrosis after lung transplantation. In the present study, single cyclosporine treatment had a significant effect on T-cell infiltration. In clinical settings, however, recent studies suggest that there are multiple mechanisms overcoming conventional immunosuppression in the long term such as: antibody-mediated alloimmune responses (360), development of memory lymphocytes (181), development of tertiary lymphoid organ forming unresolving chronic inflammatory milieu (102, 327), and augmentation of alloimmune responses by alloantigen-independent tissue injury (32, 33), innate immune responses (28, 29) and autoimmune mechanisms (20, 21, 24). Thus, overcoming these underlying injurious mechanisms is mandatory to realize a therapy for reversing established fibrosis Modification of tissue remodelling is necessary to reverse allograft airway fibrosis Conversely, the observation of epithelial-denuded isografts suggests that treatment of underlying injury alone is probably not sufficient to reverse fibrosis once it is established. The apparent irreversibility of fibrosis has long been recognized and indeed this led to the assumption that fibrosis is static and irreversible. The liver has been considered to be exceptional in its extremely high regenerative capacity (278). However, our recent studies demonstrate that apparently established fibrosis is still dynamic and manipulatable (Chapter 3 and 5). Low-dose MMP inhibition that modifies tissue remodelling and directs fibrosis to matrix degradation appears to be one of the possible approaches (Chapter 5). The 173

193 effect of MMP inhibition is still under investigation and possible side effects including musculoskeletal syndrome (349) and activation of the immune system (85, 336) need to be overcome. Novel antifibrotic agents are emerging and their efficacy has been suggested (273, 275). Although these agents have only been tested to prevent fibrosis of animal models, further investigation on fibrosis regression is encouraged Direct myofibroblast-t-cell contact is one of the pathways that keeps activating myofibroblasts in pathological fibrosis An important mechanistic discovery in this study is the importance of direct contact of myofibroblasts with T-cells to maintain the phenotype of myofibroblasts. Myofibroblasts are differentiated cells of multiple origins, including resident fibroblasts, and contribute to fibroproliferative tissue remodelling by producing an excessive amount of extracellular matrix (37, 328). Soluble cytokines and growth factors derived from inflammatory cells or injured resident cells (e.g. epithelial cells) have been recognized to induce the phenotype of myofibroblasts (40, 41, 249, 251). In parallel, direct T-cellfibroblast interaction through adhesion molecules such as ICAM-1/LFA-1 and CD40/CD40L have been demonstrated to facilitate fibroblast activation (361). However, the role of ongoing tissue injury in maintaining the phenotype of myofibroblasts has not been well investigated. The present study demonstrated that direct cell-cell contact of myofibroblasts with T-cells facilitates the maintenance of the myofibroblast phenotype in vitro. Myofibroblasts have been considered to stimulate themselves to preserve their fibrogenic activity through multiple pathways including autocrine/paracrine production of TGF-β1 (315, 316). However, in normal wound healing, myofibroblasts usually undergo apoptosis and disappear without causing excessive fibroproliferation (37). Thus, some external stimuli should keep activating myofibroblasts in chronic fibrosis and direct contact with T cells may be one of the mechanisms. We have not determined actual adhesion molecules involved in the cell-cell communication. Exploration of the mechanisms might lead to a more direct approach for interrupting the crosstalk between immune responses and fibrosis, and eventual fibrosis regression. 174

194 Regeneration of the epithelium is missing in the present approach to fibrosis Unfortunately, the therapeutic effect (i.e. partial regression of fibrosis) of combined treatment of cyclosporine and SC080 was temporal. We speculate that the strategy is still limited by a lack of regeneration of airway epithelium. The experiment using enzymatic epithelial denudation clearly demonstrated that the epithelium is a critical component in preventing fibrosis development. Thus, airways lacking in the epithelial barrier are likely to remain vulnerable to fibroproliferation. In clinical settings, many OB lesions still harbour the epithelial lining so that the current approach might be effective to some extent. To treat completely obliterated fibrosis, however, more regenerative approaches using epithelial stem cells might be beneficial (92, 93). OB lesions after lung transplantation are obviously heterogeneous in their pathological stages and possibly in their underlying mechanisms. Thus, it would be extremely important to determine an appropriate therapeutic strategy to the specific disease condition of a patient. Phenotyping of BOS is currently planned at the International Society for Heart and Lung Transplantation and this might be helpful to individualize therapies in the future Conclusion In conclusion, we demonstrated the significant effect of ongoing immune responses on myofibroblast activation in animal models and an important mechanism of the effect is attributed to direct T-cellmyofibroblast interaction. Based on the novel insight, we demonstrated temporal but significant regression of established allograft airway fibrosis for the first time by combining immunosuppression and a low-dose MMP inhibitor treatment. This concept-proving study encourages investigation of advanced fibrosis to realize fibrosis regression, an ideal therapeutic strategy for OB after lung transplantation and likely for other fibroproliferative disorders. 175

195 Chapter 7 The role of intrapulmonary de novo lymphoid tissue in obliterative bronchiolitis after lung transplantation The content of the chapter was submitted to the Journal of Clinical Investigation. 176

196 7.1. Abstract Chronic rejection after lung transplantation is manifested as obliterative bronchiolitis (OB). The development of de novo lymphoid tissue (lymphoid neogenesis) may contribute to local immune responses in small airways. Compared with normal lungs, the lung tissue of 13 lung transplant recipients who developed OB demonstrated a significantly larger number of small airway-associated peripheral-node addressin positive (PNAd+) high endothelial venules (HEVs) unique to lymphoid tissue (P < 0.001). HEVs were most abundant in lesions of lymphocytic bronchiolitis and active OB infiltrated by lymphocytes, compared with those of inactive OB. T cells in lymphocytic bronchiolitis and active OB were predominantly of the CD45RO+ CCR7- effector memory phenotype. Similar lymphoid tissue was also observed in the lung after intrapulmonary transplantation of allograft trachea (Brown-Norway to Lewis), but not after isograft transplantation. Subsequent orthotopic transplantation of the recipient Lewis lung, containing a Brown-Norway trachea, into an F1 (Lewis x Brown-Norway) rat demonstrated stable homing of Lewis-derived T cells in the lung and their antigen-specific effector function against the secondary intrapulmonary Brown-Norway trachea. In conclusion, we found de novo lymphoid tissue in the lung composed of effector memory T cells and HEVs but lacking germinal center formation. This de novo lymphoid tissue may play a critical role in chronic local immune responses after lung transplantation. 177

197 7.2. Introduction De novo lymphoid tissue formation or lymphoid neogenesis has recently been recognized as a potentially important mechanism that supports chronic inflammation in peripheral tissue (102). The best characterized de novo lymphoid tissue is the tertiary lymphoid organ (TLO), a structure resembling lymph nodes that include high endothelial venules (HEVs), well-defined T-cell and B-cell zones, and a germinal center accompanying follicular dendritic cells (87). TLO has been reported in various chronic inflammatory conditions such as autoimmune (362, 363) and chronic infectious diseases (364, 365). Chronic rejection after organ transplantation is a condition in which alloantigen persists and chronic inflammation occurs. Indeed, development of TLO has been reported in various tissues and organs after transplantation including human cardiac and renal allografts (104) and animal models of cardiac (103), skin (366), and vascular (104) allografts. However, formation of TLO after lung transplantation has not been reported and is controversial. Chronic rejection after lung transplantation is manifested by obliterative bronchiolitis (OB), a chronic inflammatory and fibroproliferative condition in small airways, and its clinical correlate, bronchiolitis obliterans syndrome (BOS) (367). In the lung, TLO is considered to be represented by an inducible form of bronchus associated lymphoid tissue (BALT), namely ibalt (102, 368). Formation of ibalt has been reported in various chronic inflammatory lung diseases: emphysema (369), pulmonary fibrosis (370), and chronic hypersensitive pneumonitis (371). In contrast, Hasegawa et al. demonstrated a negative result regarding the relationship between OB/BOS and BALT, which should theoretically include ibalt (372). Although the study using transbronchial biopsies was limited in its sensitivity to detect OB lesions and (i)balt, it is possible that de novo lymphoid tissue after lung transplantation does not form the conventional structure of TLO with HEVs, T and B cell zones, and germinal center formation. Interestingly, an evolving concept proposed by van Panhuys et al. indicates the existence of effector 178

198 lymphoid tissue (ELT) that exerts effector function by collecting effector and effector memory T cells, but does not necessarily take the conventional anatomical form of TLO or ibalt (373). Increasing evidence suggests that the lung might be an important reservoir of effector and effector memory T cells (374). In a subcutaneous tracheal transplant model of OB, preferential localization of effector memory CD4+ and CD8+ T cells to the lung parenchyma and airways has been demonstrated (375, 376). Thus, the question remains as to whether formation of de novo lymphoid tissue occurs after lung transplantation in association with the development of OB. If it does occur, does it take the form of conventional TLO or ibalt, or an unidentified form of lymphoid tissue? Is it associated with collection of effector and effector memory T cells? To address these questions, we conducted careful histological and immunohistochemical analyses of the lungs that were explanted at the time of retransplantation due to OB/BOS after the initial lung transplantation. Furthermore, we have developed a novel animal model that combines a rat intrapulmonary tracheal transplant model of OB (127) and orthotopic lung transplantation to examine the time course and functional assessment of intrapulmonary de novo lymphoid tissue after transplantation. 179

199 7.3. Results Effector memory T cells in small airways of BOS lungs We examined 13 human lungs from patients that developed BOS after lung transplantation and compared them with 15 normal lung controls. Patient demographics are shown in Table 1. Histologically, all the post-transplant lungs contained multiple OB lesions and 11 of them also contained airway inflammation without obliteration (lymphocytic bronchiolitis). Among 164 OB lesions examined, 65 (39.6%) were classified as active lesions while the others were classified as inactive based on H&E, elastic trichrome, and CD3/CD20 staining as described in Materials and Methods. In immunofluorescence labelling for T cells and B cells, we observed a number of T cells and a relatively small number of B cells in lymphocytic bronchiolitis and active OB lesions, while inactive OB lesions and normal lungs contained only a small number of T cells and B cells (Figure 7-1A). In lymphocytic bronchiolitis and active OB lesions, the memory T-cell marker CD45RO was largely colocalized to CD3+ T cells, demonstrating that T cells in active inflammatory lesions in BOS lungs are mostly memory T cells (Figure 7-1B). Furthermore, CCR7, a chemokine receptor expressed by naïve and central memory T cells, but not by effector and effector memory T cells (377), was not localized to CD45RO+ cells, demonstrating that memory T cells in BOS lungs are of an effector memory phenotype (Figure 7-1C). 180

200 # Sex age at Time from 1st Primary Examined lung Lymphocytic bronchiolitis Active OB Inactive OB 2nd Tx Tx (months) diagnosis area (cm2) (number/area [cm2]) (number/area [cm2]) (number/area [cm2]) 1 M CF (2.6) 2 (0.5) 2 (0.5) 2 F CF (2.0) 6 (1.0) 0 (0.0) 3 F CF (0.0) 1 (0.9) 0 (0.0) 4 M BE (0.4) 3 (1.6) 7 (3.6) 5 F IPF (4.0) 7 (0.8) 12 (1.4) 6 M COPD (5.8) 4 (0.3) 10 (0.8) 7 F IPF (0.0) 0 (0.0) 23 (1.9) 8 M OB (3.3) 2 (0.5) 0 (0.0) 9 F IPF (2.1) 2 (0.5) 4 (1.1) 10 F OB (0.0) 1 (0.1) 0 (0.0) 11 F BE (2.1) 5 (1.3) 0 (0.0) 12 F OB (2.2) 9 (1.4) 11 (1.7) 13 F 39* 18.6* BE (0.0) 1 (1.2) 7 (2.1) Average ± ± ± 1.9 ± SEM ± 3.8 ± 10.4 ± 1.4 (1.89 ± 0.5) (0.77 ± 0.1) (1.00 ± 0.3) Table 7-1. Demographics of lung transplant recipients who were diagnosed with OB/BOS *The patient underwent open lung biopsy. The age at 2nd Tx indicates the age at the time of open lung biopsy. The other 12 patients underwent retransplantation for diagnosed OB/BOS after the initial lung transplantation. Tx, transplantation; CF, cystic fibrosis; BE, bronchiectasis; COPD, chronic obstructive pulmonary disorder; IPF, idiopathic pulmonary fibrosis; OB, obliterative bronchiolitis (idiopathic or secondary to haematopoietic stem cell transplantation) 181

201 Figure 7-1. Effector memory T cells in airways after lung transplantation. (A) H&E staining of a normal airway, and lesions of lymphocytic bronchiolitis, active OB, and inactive OB (top panels) and corresponding immunofluorescence labeling for T-cell (CD3) and B-cell (CD20) staining (bottom panels). (B) Immunofluorescence labeling for CD3 and CD45RO, a marker of memory T cells. Pictures were taken from the same active OB lesions as in A. (C) Immunofluorescence labeling for CD45RO and CCR7, a marker for naïve and central memory T cells. Scale bar = 50 µm. 182

202 Development of high endothelial venules in small airways after lung transplantation With respect to the definition of ibalt (87, 102), the aggregates of effector memory T cells in BOS lungs do not meet the anatomical criteria since they do not include segregated T-cell and B-cell zones (Figure 7-1A) or B-cell follicles positive for CD21+ follicular dendritic cells (data not shown). High endothelial venules (HEVs) are specialized endothelial lined vessels that exist uniquely in lymphoid tissue and play critical roles in lymphocyte trafficking (378). Interestingly, immunofluorescence labeling for peripheral-node addressin (PNAd), a marker specific for HEVs demonstrated a large number HEVs composed of characteristic cuboidal endothelial cells in the airways of BOS lungs, while bronchioles in normal lungs did not show HEVs (Figure 7-2). Because HEVs positive for PNAd are considered to be unique to secondary and tertiary lymphoid organs, a similar finding in a non-lymphoid organ was surprising to us. Through quantitative evaluation, we found that HEVs existed in almost all the lesions of lymphocytic bronchiolitis and active OB lesions in the bronchiolar wall (Figure 7-2A), while a portion of inactive OB lesions also were accompanied by a small number of HEVs (Figure 7-3A). In some active OB lesions, HEVs were also observed in the lumen of obliterated airways, suggesting that induction of HEVs can occur through angioneogenesis. The percentages of lymphocytic bronchiolitis, active OB lesions, and inactive OB lesions that are accompanied by HEVs is shown in Figure 7-3A. The number of bronchioles with accompanying HEVs per unit lung area was significantly larger in BOS lungs than normal lungs (Figure 7-3B, P < 0.01). The small number of HEVs observed in normal lungs is considered to represent constitutive bronchus associated lymphoid tissue (BALT). 183

203 Figure 7-2. Development of high endothelial venules (HEVs) in small airways after lung transplantation. (A) H&E staining, trichrome staining, and double immunofluorescence labeling for CD3 and PNAd, a marker of high endothelial venules. Representative pictures of a normal bronchiole, and lesions of lymphocytic bronchiolitis, active OB with lymphocyte infiltration in the lumen, active OB with lumenal fibrosis and peri-bronchiolar lymphocyte aggregates, and two inactive OB lesions (only one of which shows positive PNAd staining). (B) Higher magnification of a PNAd+ HEV in an active OB lesion. Scale bar = 100 µm. 184

204 Figure 7-3. HEVs are associated with lymphocytic bronchiolitis and active OB. (A) Quantification of the number of bronchioles positive for PNAd demonstrates that most of the lesions of lymphocytic bronchiolitis and active OB accompany HEVs. (B) Quantification of the number of bronchioles positive for PNAd demonstrates a significantly larger number of bronchioles accompanying HEVs in the lung affected by BOS compared with that of normal lung control (P < 0.01; n = 15 and 13 in control and BOS, respectively; the mean values are indicated by black bars). 185

205 Alloantigen-dependent ELT formation after rat intrapulmonary tracheal transplantation A rat intrapulmonary tracheal transplant model of OB is an animal model of OB that reflects the influences of the pulmonary milieu (85, 127). This model enables examination of the pulmonary immune system to allogenic stimuli. In the present study, we examined whether lymphoid tissue also develops in the lung in MHC full-mismatched intrapulmonary tracheal transplantation (Brown-Norway donors to Lewis recipients) and compared them with syngenic intrapulmonary tracheal transplantation (Lewis to Lewis). In normal lungs of Lewis rats (i.e. without transplantation), we observed a small number of lymphocyte aggregates, which are consistent with reported constitutive bronchus associated lymphoid tissue (BALT) in rats (379). After intrapulmonary tracheal transplantation, the recipient lung of both isografts and allografts showed lymphocyte aggregates around bronchioles at day 7 (Figure 7-4A). By day 28, the lymphocyte aggregates were diminished in isograft recipients, while allograft recipient lungs exhibited dense lymphocyte aggregates composed of T cells as well as B cells (Figure 7-4B). At day 28, isografts showed complete recovery of the epithelial cells with no evidence of obliterative fibrosis, while allografts showed total loss of the epithelium with obliterative airway fibrosis (Figure 7-4B, insets). The lymphocyte aggregates in allograft recipient lungs persisted until the end of observation at day 120 (Figure 7-4C). Morphometric quantification demonstrated a significantly larger number of lymphocyte aggregates in allograft recipients at day 28 and after (Figure 7-4D). These results demonstrate that persistence of lymphocyte aggregates in the lung is an alloantigen-dependent process. In rats, loss of the CD45RC expression is considered to represent maturation into memory T cells equivalent to CD45RO expression in humans (377). Flow cytometric analysis of allograft recipient lungs at day 28 demonstrated that more than 90% of T cells in the lung tissue are of the CD45RCmemory phenotype after allograft tracheal transplantation at day 28, while mediastinal lymph nodes and spleen tissue showed smaller numbers of memory T cells (Figure 7-4E). Although demonstration of HEVs in rat lungs was limited by the unavailability of an anti-rat PNAd antibody, we used 186

206 MadCAM-1, another marker of HEVs, in combination with von Willebrand factor, a marker of vascular endothelial cells. Indeed, we observed MadCAM-1+ von Willebrand factor+ HEVs in welldeveloped lymphocyte aggregates (Figure 7-4F). These observations indicate that lymphocyte aggregates that develop after intrapulmonary allograft tracheal transplantation are likely to be de novo lymphoid tissue similar to those observed in transplanted human lungs. 187

207 Figure 7-4. Development of effector lymphoid tissue in airways after intrapulmonary tracheal transplantation. (A, B) H&E staining (top panel) and double immunofluorescence labeling for CD3 (T cell) and CD79a (B cell). Representative pictures of recipient lungs of intrapulmonary isograft and allograft tracheal transplantation at days 7 and 28 are shown. Insets in B are corresponding tracheal grafts (original, 40x). (C) H&E staining of recipient lungs of intrapulmonary isograft and allograft tracheal transplantation at days 120. Arrows indicate peri-bronchiolar lymphocyte aggregates. (D) Morphometric quantification of lymphocyte aggregates in the recipient lungs that received an isograft and allograft. Allograft recipient lungs demonstrated a significantly larger number of lymphocyte aggregates than those of isograft recipient lungs at day 28 and after (P > 0.05). n = 4-5, mean ± SEM. (E) Flow cytometric analysis of CD4 and CD45RC expression on CD3+ T cells in the spleen, mediastinal lymph nodes, and the left lung at day 28 after intrapulmonary allograft tracheal transplantation. Cells were first gated by CD3 positivity. (F) Double immunofluorescence labeling for von Willebrand factor (vwf) and MadCAM-1 (top panel) and CD3 and CD79a (bottom panel) taken from a similar area of a well-developed lymphocyte aggregates at day 56 after intrapulmonary allograft tracheal transplantation. The arrow indicates an HEV-like vessel observed in the aggregate of T cells. Scale bar = 100 µm. 188

208 Early lymphocyte aggregates are insufficient for allograft airway rejection We subsequently addressed the question regarding whether or not the lymphocyte aggregates become a stable structure in association with stable homing of memory T cells. Following initial intrapulmonary allograft tracheal transplantation (BN to Lewis), we conducted orthotopic transplantation of the Lewis lung containing a BN tracheal graft into an F1 (BN x Lewis) rat at day 7 or day 28 (Figure 7-5A). Since F1 rats should accept both Lewis-derived and BN-derived grafts, the only components that could reject BN-derived grafts are Lewis-derived lymphocytes that may persist in the lung after orthotopic lung transplantation into an F1 rat. In the first set of experiments wherein orthotopic lung transplantation was conducted at day 7, peribronchiolar lymphocyte aggregates in the Lewis lung disappeared and the BN graft showed recovery of the epithelium with little subepithelial fibrosis at day 35 (Figure 7-5B). The negative control of Lewis-Lewis-F1 transplantation also demonstrated minimum lymphocyte aggregates and complete recovery of the tracheal graft, in contrast to the positive control of BN-Lewis-Lewis transplantation, in which graft rejection should continue and indeed lymphocyte aggregates were still observed in the lung with obliterated BN trachea (Figure 7-5B). Morphometric quantification demonstrated significantly larger lymphocyte aggregates in BN-Lewis-Lewis transplantation at day 35 compared with the other two groups (Figure 7-5C). Thus, the intrapulmonary lymphocyte aggregates at day 7 are still immature and incapable of stably harbouring memory T cells. 189

209 Figure 7-5. Lymphocyte aggregates are immature at day 7 after intrapulmonary allograft tracheal transplantation. (A) A study design of combined intrapulmonary tracheal transplantation (red) and orthotopic lung transplantation (blue), which is conducted at day 7 or day 28. (B) A result of intrapulmonary tracheal transplantation followed by orthotopic lung transplantation at day 7. Representative pictures (H&E staining) of a recipient lung (top panels) and the corresponding intrapulmonary tracheal graft (bottom panel) at day 35 are shown. Scale bar = 500 µm. (C) Morphometric quantification at day 35 demonstrates a significantly larger number of peribronchiolar lymphocyte aggregates in the recipient lung of BN-Lewis-Lewis transplantation as compared with the other two groups (P < 0.05). n = 4, mean ± SEM. 190

210 Conversely, the second set of experiments wherein orthotopic lung transplantation was conducted at day 28 demonstrated stable homing of memory T cells. In BN-Lewis-F1 transplantation, the lymphocyte aggregates in the peribronchiolar tissue was obviously enlarged compared with the other two groups, and extended into the perigraft tissue at day 56 (i.e. 28 days after orthotopic lung transplantation) (Figure 7-6A). The negative control of Lewis-Lewis-F1 transplantation showed the small size of lymphocyte aggregates, while those of the positive control of BN-Lewis-Lewis transplantation were similar to untreated allograft transplantation control. Morphometric quantification confirmed the significantly larger size of lymphocyte aggregates in the group of BN-Lewis-F1 transplantation (P < 0.05, Figure 7-6B). Flow cytometric analyses demonstrated that a significant percentage of Lewis-derived RT1A u (-) CD4+ T cells and CD8+ T cells exist in the left lung in BN- Lewis-F1 transplantation as compared with the other groups (Figure 7-7A and 7-7B). Immunofluorescence labeling for RT1A u in the lymphocyte aggregates in the lung after BN-Lew-F1 transplantation demonstrated chimerism of F1-derived RT1A u (+) cells and Lewis-derived RT1A u (-) cells (Figure 7-7C). Moreover, the majority of CD4+ T cells in the lung of the BN-Lewis-F1 transplantation group demonstrated a higher ratio of the CD45RC low memory phenotype as compared with the other groups (Figure 7-7D). This ratio was also higher than the opposite lung, mediastinal lymph nodes, and spleen (< 60%), indicating that memory T cells dominantly exist in the orthotopically transplanted lungs in association with lymphocyte aggregates. This overexpansion of lymphocyte aggregates, which was even greater than that of the BN-Lewis- Lewis positive control transplantation, was somewhat surprising to us. We speculate that Lewisderived RT1A u (-) lymphocytes homing to the lung reacted to F1-derived RT1A u (+) lymphocytes. The Lewis lung containing BN trachea was orthotopically transplanted into an F1 host because an F1 rat should not reject either Lewis or BN-derived tissue; however, the opposite is not true and, if there were significant Lewis-derived lymphocytes, they could theoretically be reactive to the F1 host-derived cells through a mechanism similar to graft-versus-host disease, which can lead to OB after haematopoietic stem cell transplantation (380). 191

211 Figure 7-6. Lymphocyte aggregates become mature by day 28 after intrapulmonary allograft tracheal transplantation. (A) A result of intrapulmonary tracheal transplantation followed by orthotopic lung transplantation at day 28. Representative pictures (H&E staining) of a recipient lung (top panels) and the corresponding intrapulmonary tracheal graft (bottom panel) at day 56 are shown. Scale bar = 500 µm. (C) Morphometric quantification at day 56 demonstrates a significantly larger number of peribronchiolar lymphocyte aggregates in the recipient lung of BN-Lewis-F1 transplantation as compared with the other two groups (P < 0.05). n = 4, mean ± SEM. 192

212 Figure 7-7. Stable homing of T cells derived from the initial tracheal transplant recipient and memory T cells in the lung. (A) Representative results of flow cytometric analysis for the expression of RT1A u (expressed by BN and F1) by CD3 + CD4 + cells (Top panel) and CD3 + CD8 + cells (bottom panel) in the lung. (B) Immunofluorescence labeling for RT1A u demonstrating chimerism of peribronchiolar lymphocyte aggregates at day 56. Scale bar = 100 µm. (C) A representative result of flow cytometric analysis for the expression of CD3 + CD4 + CD45RC low memory T cells in the lung. These results were obtained at day 56 from the left lung after intrapulmonary tracheal transplantation (day 0) followed by orthotopic left lung transplantation (day 28). 193

213 Memory lymphocytes homing to the lung exert effector function The previous experiment could not demonstrate effector function of intrapulmonary lymphocytes against the airway graft because the allograft trachea was already obliterated when orthotopic lung transplantation was conducted at day 28. Thus, we added a second intrapulmonary tracheal transplantation at the time of orthotopic lung transplantation at day 28 (Figure 7-8A). At day 56 after the initial intrapulmonary tracheal transplantation (i.e. 28 days after orthotopic lung transplantation), the first intrapulmonary BN-derived tracheal grafts showed total loss of the epithelium and lumenal obliteration with fibrous tissue in both groups (Figure 7-8B). The histological characteristics of the fibrous tissue of the first BN grafts were similar to those of untreated control allografts (Figure 7-3A). On the other hand, the second intrapulmonary BN-derived tracheal grafts showed massive infiltration of T cells with total loss of the epithelium (Figure 7-8B). Conversely, another BN graft implanted in the subcutaneous tissue at the time of orthotopic lung transplantation showed complete recovery of the epithelium with no evidence of T-cell infiltration (Figure 7-8C). Thus, the effector function of the intrapulmonary de novo lymphoid tissue has a significant impact on the local immune response against allogenic airways. 194

214 Figure 7-8. The effector function of lymphocytes in the lung. (A) A study design of intrapulmonary tracheal transplantation followed by concurrent orthotopic lung transplantation, secondary intrapulmonary tracheal transplantation, and subcutaneous tracheal transplantation. (B) A representative picture of the first and second intrapulmonary BN-derived tracheal grafts in the Lewisderived lung implanted in an F1 host (top panel) and a corresponding high-magnification picture of the secondary graft. (C) The subcutaneous BN-derived tracheal graft from the same experiment. M indicates mucus produced by the airway epithelium indicated by the arrow. The insets show CD3 staining, demonstrating infiltration of T cells in the second intrapulmonary tracheal graft. 195

215 7.4. Discussion A novel type of de novo lymphoid tissue in the lung In the present study, we have identified a novel type of de novo lymphoid tissue in the lung composed of effector memory T cells and HEVs but lacking germinal center formation. The result of the present study demonstrates that de novo lymphoid tissue formation is not necessarily restricted to the form of conventional TLO that is defined as a structure containing HEVs, organized T-cell and B-cell zones, and a germinal center (102), but can take other anatomical forms to mediate adaptive immune responses. Moreover, the T-cell component of the de novo lymphoid tissue in this study shows critical differences from conventional TLO. In general, secondary and tertiary lymphoid tissues have been characterized by recruitment of naïve lymphocytes including CD45RO- CCR7+ naïve T cells and their activation through antigen presentation (102). In contrast, the lymphoid tissue we found in the present study is almost exclusively composed of CD45RO+ CCR7- effector memory T cells. Interestingly, in lung tissue affected by pulmonary fibrosis, Marchal-Somme et al. has recently reported another type of de novo lymphoid tissue that is composed mainly of the CD45RO+ CCR7- effector memory T cells, while the lymphoid tissue also contained an organized B cell zone with frequent follicular formation (381). Although their finding is clearly different from ours in the B-cell component, their and our observations both support the recently proposed concept of effector lymphoid tissue (ELT) that is defined by its collection of effector and effector memory T cells and their function (373). Furthermore, an animal experiment has demonstrated that effector memory T cells can reside in the lung tissue including ibalt long after the initial antigen exposure (105). These pieces of evidence are of great importance in the implication that de novo lymphoid tissue - whether or not in the form of conventional TLO is not necessarily a duplication of secondary lymphoid tissue (e.g. lymph nodes) but a functionally distinct effector structure organized in peripheral organs in the context of chronic inflammation. 196

216 The role of de novo lymphoid tissue in obliterative bronchiolitis after lung transplantaiton The findings of the present study also suggest a novel mechanism of OB after lung transplantation. Firstly, stable formation of intrapulmonary lymphoid tissue may be important to the transition from an alloantigen-independent (or innate) immune response to alloantigen-dependent (or adaptive) immune responses. We found that intrapulmonary transplantation of syngenic as well as allogenic tracheae induced lymphocyte aggregates (Figure 4A), suggesting the role of alloantigen-independent process in the development of lymphocyte aggregates. On the other hand, stable formation of intrapulmonary lymphoid tissue was observed only in allograft transplantation (Figure 7-4). Moreover, lymphocyte aggregates at day 7 after intrapulmonary allograft transplantation are still immature and reversible (Figure 7-5), while the lymphocyte aggregates become more stable after day 28 (Figure 7-6, 7). In general, activation of resident cells including endothelial cells is considered to be mediated by antigenindependent inflammatory stimuli (e.g. cytokines and other innate danger signals ) (99, 382), while persistence of the antigen (e.g. microorganisms, autoantigens) is essential for antigen-specific lymphocytes to reside in the tissue and augment the inflammatory microenvironment (325, 327). As such, both innate and adaptive processes appear to be important to the development and maturation of intrapulmonary de novo lymphoid tissue. Since the role of innate immune responses associated with chronic bacterial colonization in airways (32, 383), bile/acid aspiration (26, 33), and associated airway neutrophilia (32, 33, 384) has been of great interest in the pathogenesis of OB (341), initiation of intrapulmonary lymphoid tissue formation in such alloantigen-independent inflammatory conditions is a potential novel mechanism in the pathogenesis of OB after lung transplantation The role of de novo lymphoid tissue in the transition from acute to chronic rejection Formation of de novo lymphoid tissue may also be important in the transition from reversible acute rejection of airways called lymphocytic bronchiolitis (385) to the recalcitrant condition of OB. Lymphocytic bronchiolitis has been considered to be a precursor lesion of OB (126) and its severity has recently been demonstrated to correlate with the incidence of OB/BOS (386). Considering the striking numbers of HEVs in the lesions of lymphocytic bronchiolitis that we observed (Figure 7-2), de 197

217 novo lymphoid tissue formation may well precede the development of OB. Recent studies on chronic inflammatory disorders including chronic infectious diseases and autoimmune diseases (e.g. rheumatoid arthritis) suggest that the critical switch from acute, resolvable inflammation to persistent, poorly resolvable inflammation lies in the phenotypical alterations of resident cells including endothelial cells (99, 325). These cells begin to express adhesion molecules and ectopic lymphoid chemokines in response to inflammatory stimuli (327) and the altered tissue microenvironment allows for large influx and survival of antigen-specific lymphocytes in peripheral tissue (327). Indeed, the PNAd and MadCAM-1 we observed in human and rat lungs respectively in allograft transplantation (Figure 2 and 4) are both adhesion molecules, the ligand of which is expressed by lymphocytes (102). Moreover, induction of the HEV phenotype in endothelial cells is known to be induced by inflammatory cytokines such as tumor necrosis factor (387). Thus, the development of HEVs may reflect inflammatory stimuli in small airways and de novo lymphoid tissue formation is likely to contribute to the formation of a chronic (i.e. hard to reverse) inflammatory microenvironment in small airways and to the poor reversibility of OB lesions Questions remaining in de novo lymphoid tissue in the lung after transplantation We acknowledge multiple limitations in the present study. Firstly, our human study lacks chronological observation; namely, how soon the process of de novo lymphoid tissue formation is initiated. Although we took advantage of the abundant lung tissue available at the time of retransplantation in this study, the strategy was limited in longitudinal observation. Secondly, the role of HEVs in the de novo lymphoid tissue remains unclear. In considering the role of HEVs in the trafficking of naïve and central memory T cells and the contradictory effector memory T-cell phenotype, an important question is whether the HEVs continue the recruitment of naïve lymphocytes for priming and activation as conventional secondary or tertiary lymphoid organs do. Thirdly, the mechanism whereby chronic immune responses result in different types of de novo lymphoid tissue, namely conventional TLO versus the de novo lymphoid tissue after lung transplantation remains unclear. Multiple factors such as lung-specific immunological properties related to mucosal immunity 198

218 and the existence of immunosuppression after transplantation may be associated. Lastly, the animal model we used has multiple limitations including the initial intrapulmonary tracheal transplantation (not transplantation of the entire lung) and the lack of immunosuppression. Difference among species is another important limitation of an animal experiment, particularly in the organization of intrapulmonary lymphoid tissue such as BALT (388, 389). Despite these limitations, we emphasize that the animal experiments in the present study complemented the limitations in the human study in many ways and provided important insights regarding the role of intrapulmonary de novo lymphoid tissue in OB after lung transplantation Conclusion In conclusion, we found de novo lymphoid tissue in the lung composed of effector memory T cells and HEVs but lacking germinal center formation. The de novo lymphoid tissue likely plays a critical role in chronic local immune responses after lung transplantation. Further studies are obviously necessary to improve our understanding of this novel form of lymphoid tissue and its contribution to the development OB after lung transplantation. 199

219 Chapter 8 Discussion and Future Directions 200

220 8.1. Discussion The concept of tissue remodelling in OB after lung transplantation Obliterative bronchiolitis (OB) describes fibroproliferative changes in bronchioles frequently observed after lung transplantation in association with chronic allograft rejection. The clinical correlate of OB, bronchiolitis obliterans syndrome (BOS) is manifested by deteriorating pulmonary function with increasing shortness of breath that relentlessly progresses to death. The reported incidence of OB/BOS is approximately 50% within 5 years after transplantation. OB/BOS remains the greatest cause of morbidity and mortality in long-term survivors of lung transplantation. OB/BOS has long been accepted to be a form of chronic rejection (1, 276). Although there is robust evidence demonstrating the role of rejection in OB/BOS, this does not necessarily mean OB/BOS is a manifestation of pure rejection. Today, a bulk of evidence suggests its association with alloimmuneindependent tissue injury that may be caused by aspiration and infection (341). In the present thesis, we further hypothesized that tissue remodelling is an important mechanism of obliterative bronchiolitis (OB) after lung transplantation. Tissue remodelling can be a physiological phenomenon that maintains tissue homeostasis and recovers normal tissue architecture and function through repair and regeneration after injury. Nevertheless, tissue remodelling could become aberrant, leading to abnormalities in tissue architecture and function. Conversely, tissue remodelling is also important in reversing the process of aberrant tissue remodelling. Induction of fibrosis regression or reversal of aberrant tissue remodelling is theoretically the ideal treatment of fibrous disorders including OB after lung transplantation. The present work establishes an important foundation in the tissue-remodelling concept in the pathophysiology of OB/BOS, integrating aberrant remodelling, repair, regeneration, and reversal of fibrosis. Moreover, we found that tissue remodelling is always linked to underlying tissue injury such as ongoing alloimmune responses. Since the lung is equipped with special properties to develop lymphoid tissue that harbours effector memory T cells, these properties are likely to contribute to 201

221 continuous local alloimmune responses and render the lung vulnerable to OB/BOS after lung transplantation. The putative mechanisms of OB/BOS after lung transplantation are summarized in Figure

222 Figure 8-1. Overview of the mechanisms of tissue remodelling. Alloimmune-dependent and independent injury in small airways induces high endothelial venules (HEVs) that support the influx of lymphocytes, amplify local immune responses, and form effector lymphoid tissue. Excessive tissue injury leads to aberrant remodelling (fibroproliferation) through recruitment and activation of myofibroblast progenitors (e.g. circulating fibrocytes). Although the most active immune responses subside as fibrosis develops, established fibrosis is still related to underlying continuous immune responses, making the whole process refractory to spontaneous reversal or any current treatment. Novel strategies such as chronic immunomodulation, interruption of the T-cell-myofibroblast interaction, and modification of tissue remodelling, might realize an effective therapy for fibrosis in the future. 203

223 Tissue remodelling, a novel therapeutic niche in lung transplantation Using a novel animal model of intrapulmonary tracheal transplantation of OB, we demonstrated that allograft airway fibrosis in the pulmonary milieu is not a static but a dynamic process, in which cellular and matrix components are incessantly changing (Chapter 3). We found that MMP-2 and MMP-14 produced by myofibroblasts play important roles in the dynamic tissue remodelling in an animal model of OB (Chapter 3). Importantly, we found analogous MMP expression in human OB lesions after lung transplantation (Chapter 5). Furthermore, detailed analysis of matrix turnover in established allograft airway fibrosis suggested that the direction of tissue remodelling is potentially bidirectional (Chapter 5) as indicated in liver fibrosis, in which fibrosis regression does occur under certain conditions (278, 279). By combining MMP manipulation with immunosuppression, we demonstrated that established allograft airway fibrosis can even undergo regression (Chapter 6). We found MMP-dependent migration of circulating fibrocytes contributing to fibroproliferative tissue remodelling in allograft airways in the lung as viable potential therapeutic targets (Chapter 4). We also found that intensive MMP inhibition can induce apoptosis in myofibroblasts in vivo and in vitro (Chapter 5). Thus, tissue remodelling is now considered to be an important target of therapy for OB/BOS after lung transplantation. Despite the proof of concept, strategies to modulate tissue remodelling using a broad-spectrum MMP inhibitor are in part limited due to safety concerns. The clinical application of general MMP inhibitors have been limited by a systemic side effect called musculoskeletal syndrome (349). Moreover, we found that MMP inhibitors are able to activate the immune system and accumulate lymphocytes in allografts. Similar findings have also been reported by other groups recently in transplant settings (335, 336). As a next step in this particular area, it is urgent to seek out and test strategies that could be clinically applicable in the near future Ongoing immune responses, T-cell-myofibroblast crosstalk, and reversibility of fibrosis 204

224 In the studies presented in this thesis, we also identified the importance of regulating immune responses in preventing fibrosis in OB as well as reversing fibrosis after its establishment. In human and animal studies, we demonstrated that ongoing alloimmune responses are associated with qualitative differences in fibrosis such as in the number of myofibroblasts and the amount of collagen that accumulates in the tissue (Chapter 6). Moreover, we found that direct co-culture of myofibroblasts with T-cells helps myofibroblasts to maintain their phenotype (Chapter 6). These findings led us to hypothesize that underlying alloimmune responses may be an important factor to support fibroproliferative tissue remodelling in allograft airway fibrosis even after its establishment. Strikingly, we observed partial fibrosis regression of established allograft airway fibrosis in an experimental model of OB when the process of tissue remodelling is modified by a low-dose of an MMP inhibitor in combination with immunosuppression (Chapter 6). These encouraging results suggest that targeting tissue remodelling and alloimmune responses, and/or their interaction is an attractive therapeutic strategy for advanced OB/BOS after lung transplantation. Currently, we are facing two limitations in developing novel therapeutic strategies in addition to the safety issues associated with MMP inhibitors as mentioned earlier. Firstly, we could not reveal the detailed mechanism of myofibroblast-t-cell crosstalk. Further mechanistic studies are necessary to reveal the critical link between tissue remodelling and alloimmune responses after lung transplantation. Secondly, we found that the therapeutic effect of combined immunosuppression and tissue remodelling modulation was lost in the long-term, likely due to the lack of epithelial regeneration in the present animal model (at least in part). Currently, strategies to regenerate the epithelial lining are enthusiastically investigated in Dr. T. Waddell s group in our Thoracic Surgery Laboratories (93). Epithelial regeneration in combination with immunosuppression and tissue remodelling strategies could be a future therapy for very advanced OB after lung transplantation. It is important to note, however, that we found that approximately half of OB lesions are partially open with preserved epithelium even in advanced BOS cases for which retransplantation was necessary (Chapter 7). In lung 205

225 transplant recipients affected by OB/BOS, all the airways are not equally obliterated but are partially open and distributed across the lung. Thus, these partially open airways with preserved epithelial lining might be a practical therapeutic target at this moment. Further investigation is necessary to demonstrate the concept of fibrosis regression in an animal model that simulates such a clinical setting of partial airway obliteration De novo lymphoid tissue formation in the lung after transplantation We suspect that the lung is particularly vulnerable to chronic graft dysfunction among solid transplanted organs (82). We have also recognized that small airways are highly and selectively affected by OB for unknown reasons. Strikingly, in the present study, we found that small airways per se become lymphoid tissue after lung transplantation by developing HEVs (specialized postcapillary venules unique to lymphoid organs) and harbouring effector memory T cells (Chapter 7). By examining the lung tissue after intrapulmonary tracheal transplantation of allografts and isografts, we found that allograft rejection occurs only in the lung and that the rejection process is associated with formation of lymphoid tissue in small airways (Chapter 7). We also demonstrated effector function of lymphoid tissue in the lung mediated by effector memory T cells homing to the lung (Chapter 7). These studies have commenced a new area of research in lung transplant immunology. Not surprisingly, there remain many questions to be answered. An important question is whether the de novo lymphoid tissue in the lung is always an ominous manifestation. Detrimental effects of tertiary lymphoid organs (TLOs) in organ transplantation have been suggested in renal and cardiac transplantation (104); however, their effector function has not been demonstrated yet (87). We have experienced a similar dilemma in the analysis of the lymphoid tissue in pulmonary allografts using archival tissue specimens (Chapter 7). On the other hand, a study has demonstrated that cardiac allografts with Quilty lesions (a form of TLO) survive better than those without these lesions (390). Hasegawa et al. demonstrated that bronchus-associated lymphoid tissue (BALT) is observed in stable lung transplant recipients as well as in those with OB/BOS, and speculated that BALT represents an evolution of immunological tolerance 206

226 (i.e. BALT is protective) (372). Since BALT, per se, is not observed in healthy adult human lungs (388), BALT observed in lung transplant recipients is considered to be the result of de novo lymphoid tissue formation. Interestingly, it is possible that regulatory T cells home in to ectopic lymphoid tissue to regulate local immune responses (391). Thus, de novo lymphoid tissue in the lung may play binary roles, namely, protective and detrimental roles. Moreover, the potential regulatory role of de novo lymphoid tissue emphasizes the importance of regulating local immune responses in chronic immune responses in contrast to current systemic immunosuppression. Further investigation is necessary to reveal immunological roles of de novo lymphoid tissue after transplantation Future directions Hypothesis and specific aims of the future study To answer the questions raised in the present study, we will test the hypothesis along specific study aims as follows. Hypothesis: De novo lymphoid tissue in bronchioles is an important regulator of local chronic immune responses that may lead to and maintain obliterative bronchiolitis after lung transplantation. Interaction between activated memory T cells homing to the intrapulmonary lymphoid tissue and myofibroblasts in fibrosis is an important mechanism that prevents established fibrosis from resolving through tissue remodelling. This hypothesis will be tested with the following specific study aims: Study Aim 1. To investigate the role of lymphoid tissue in the lung after transplantation. Study Aim 2. To investigate the long-term immunoregulatory effects of IL-10, including the effect on lymphoid tissue formation after transplantation. Study Aim 3. To investigate strategies to modulate tissue remodelling of allograft airway fibrosis. 207

227 Study Aim 4. To investigate the interaction between the immune system and tissue remodelling in allograft airway fibrosis To investigate the role of lymphoid tissue in the lung after transplantation a. To investigate human intrapulmonary lymphoid tissue in stable lung transplant recipients and in those who developed OB/BOS. We have found de novo lymphoid tissue formation in the lung after human lung transplantation and in an animal model of OB. The results we obtained so far strongly suggest their deteriorating roles in chronic alloimmune responses. However, the role of de novo lymphoid tissue in the lung and other organs after transplantation remains controversial (87, 372). Studies on Quilty lesions after cardiac transplantation (390) and BALT after lung transplantation (372) imply protective roles of the lymphocyte aggregates in allografts. It is possible that lymphocyte aggregates in allografts have dual roles depending on the phenotype of the lymphocytes (e.g. effector vs. regulatory T cells), and depending on the conditions after transplantation. If de novo lymphoid tissue could play binary roles (detrimental vs. protective) in pulmonary allografts, the protective or non-detrimental component might be explained by at least the following two factors in the lymphoid tissue: homing of regulatory T cells and formation of donor-recipient chimerism. Because it is possible that regulatory T cells home in to ectopic lymphoid tissue to regulate local immune responses (391), markers for regulatory T-cells such as FOXP3 and IL-10 will be examined immunohistochemically in archival tissue samples. Another possibility is that the degree of chimerism in lymphoid tissue affects local immune responses in the lung. It has been reported that the lung of stable transplant recipients harbour a large number of donor-derived leukocytes including lymphocytes, forming chimerism in the lung (392). It is generally accepted that formation of stable chimerism of donor-derived and recipient-derived lymphocytes 208

228 induce the most robust tolerogenic conditions in organ transplantation (181). We may detect the degree of chimerism in lymphoid tissue using microdissection and spectrotyping, as used to detect chimerism of myofibroblasts in human OB lesions (36). In addition to microdissection, spectrotyping has been conducted to examine the repertoire of T-cell receptors in rat lung lymphoid tissues (Figure 8-2), not included in the study of this thesis). Similar techniques are applicable by using primers for genomic hypervariable regions (36). These retrospective studies will be conducted for archival biopsy samples (transbronchial). The study is primarily a continuation of the human histological studies conducted in this thesis with great assistance from Dr. David Hwang, at the Department of Pathology (also a principal investigator in our group). As the number of cases is a potential limitation in these studies, collaboration with the pathology department of another large transplant center, such as the Pittsburgh group, is one of our options. b. To investigate dendritic cells involved in de novo lymphoid tissue in the lung after transplantation Another controversy associated with de novo lymphoid tissue in the lung related to BOS is whether the lymphoid tissue only has effector function to amplify the immune responses initiated at different sites (e.g. secondary lymphoid organs) or the lymphoid tissue could start to prime naïve lymphocytes and play independent roles in peripheral organs. In the lung affected by BOS, T-cells are predominantly of the CCR7 - CD45RO + effector memory phenotype (Chapter 7), indicating the role of de novo lymphoid tissue in amplifying immune responses initiated at different sites. Such a role of lymphoid tissue in peripheral organs is compatible with the originally proposed hypothesis of effector lymphoid tissue (373). However, since the ligand of PNAd (a marker of HEV) is CD62L (a marker of naïve or central memory T cells), the emergence of HEVs in the lymphocyte aggregates may suggest attraction of CD62L hi naïve and central memory T cells. If the lymphoid tissue in bronchioles is also important in priming naïve T-cells or re-activating central memory T cells, antigen presenting cells such as dendritic cells in the tissue should play important roles. In general, dendritic cells that survey in peripheral tissue 209

229 are immature and capable of capturing antigens. Once they capture antigens, they are stimulated under inflammatory conditions such as through what is called danger signals (160). Dendritic cells mature and migrate to secondary lymphoid organs for antigen presentation (393, 394). Thus, if the priming of naïve T cells or reactivation of central memory T cells occurs in peripheral tissue, such as in the de novo lymphoid tissue in bronchioles, dendritic cells in the tissue must be mature and fully capable of presenting antigens as those in secondary lymphoid organs. We will examine archival tissue samples using several dendritic cell markers. Immature dendritic cells will be detected by CD1a, CD1c; maturing dendritic cells will be detected by DC-SIGN; and fully matured dendritic cells will be detected by CD83, CD86, and DC-LAMP. A study of similar design has recently been reported for pulmonary fibrosis, demonstrating immature dendritic cells distributed in fibrous tissue, while mature dendritic cells were observed in de novo lymphoid follicules or ibalt (395). In pulmonary allografts with chronic rejection, lymphoid tissue is not anatomically as organized as that of pulmonary fibrosis. Thus, distribution of dendritic cells at different maturation stages will provide clues about the function of the de novo lymphoid tissue after lung transplantation. c. To determine the role of secondary and ectopic lymphoid tissue in allograft airway rejection using transgenic mice lacking secondary lymphoid tissue In an animal experiment, we will try to dissect the role of de novo intrapulmonary lymphoid tissue from secondary lymph nodes in alloimmune responses. In animal experiments of the present study (Chapter 7), the development of de novo lymphoid tissue appeared to be dependent on secondary (mediastinal) lymph nodes. However, the roles of the secondary lymphoid organ and de novo lymphoid tissue are controversial, depending on the locales of the immune responses. Transgenic lymphotoxin knockout (LTα -/- ) mice that do not have the secondary lymphoid organ did not reject skin allografts, demonstrating the important role of secondary lymphoid organ (366). The same strain of mice was, however, capable of eradicating the influenza virus efficiently by organizing inducible BALT in the lung and demonstrating the likely role of de novo lymphoid tissue as an independent lymphoid tissue 210

230 (368). Such a discrepancy might represent an important feature of the pulmonary immune system. We may be able to clarify this point by using LTα -/- mice. We are developing a novel mouse intrapulmonary tracheal transplant model. A preliminary result of this mouse intrapulmonary tracheal transplant model of OB, similar to that of rat, is shown in Figure 8-2D. We will use this model in LTα -/- mice and compare the result with wild-type mice and with the subcutaneous tracheal transplant model. 211

231 D Figure 8-2. Microdissection and specrotyping of rat intrapulmonary lymphoid tissue. Rat intrapulmonary lymphoid tissues after tracheal transplantation were microdissected and RNA was extracted (A). Using specific primers for 20 T-cell receptor V regions, RT-PCR was conducted (B). The PCR product was further amplified using fluorescence-conjugated primers and the amplicon was examined using spectrotyping (C). In this example, preferential monoclonal expansion was shown in the right sample, while the left sample shows more polyclonal characteristics. The same technique may be used for genomic DNA and typing of polymorphisms. (D) Preliminary results of mouse intrapulmonary tracheal transplantation. The isograft shows recovery of the epithelium at day 7. The arrow indicates regenerating squamous-metaplastic epithelial cells. The allograft is rejected and obliterated at day 14. The lung tissue shows lymphocyte aggregates around a bronchiole (lower panel). 212

232 To investigate the long-term immunoregulatory effect of IL-10, including its effect on lymphoid tissue formation after transplantation For the last decade, our laboratory has investigated the role of IL-10 in preventing primary graft failure (ischemia-reperfusion injury) after lung transplantation. Adenovirus-mediated gene therapy at the time of transplantation has been demonstrated to be a promising strategy to attenuate early graft dysfunction after lung transplantation, by attenuating multiple transplant-related lung injuries. We have also determined multiple factors in the early immune response in BO. We found that transplant-induced fibroproliferative airway lesions are associated with the upregulation of T-helper-1 cytokines and chemokines (182). Based on these observations, we then tested tracheal allografts with recombinant interleukin-10 (IL-10), an immunoregulatory cytokine, and later developed adenoviral-mediated IL-10 gene transfer. Both strategies significantly inhibited the development of obliterative lesions of allograft airways in a rat subcutaneous tracheal transplant model (208). However, previous experiment settings had multiple limitations. Firstly, subcutaneous tracheal transplant models do not reflect the pulmonary milieu and do not involve lymphoid tissue formation as we have demonstrated in human and animal studies (Chapter 7). Secondly, previous studies targeted an acute-type of rejection that occurs within 10 days after transplantation more than the chronic-type of rejection that is associated with OB in human transplantation. As such, the short-term experiment (up to day 21) might not be relevant to prevention and treatment of OB in clinical settings. Thirdly, adenovirus-mediated gene expression is limited to the short-term. Since it is hard to overcome all the problems at once, we will approach IL-10-mediated chronic immunomodulation in the following two ways. a. Adenovirus-mediated IL-10 gene therapy in an intrapulmonary tracheal transplant model Using an IL-10 encoding adenovirus, we will examine whether IL-10 can modulate de novo lymphoid tissue formation and chronic local immune responses. Our preliminary result suggests that IL-10 alone started at the time of transplantation does not replace conventional immunosuppression in an 213

233 intrapulmonary tracheal transplant model. The discrepancy of the result from that of the previous subcutaneous model (208) might reflect the difference of aggressive innate and adaptive immune responses in the lung. On the other hand, we have found that the initial use of immunosuppression (cyclosporine mg/kg/d) up to day 14 restores the airway epithelium in allografts, overcoming the first transplant-related injury and acute-type rejection (Figure 8-3). Testing an immunomodulatory treatment, such as adenovirus-mediated IL-10 gene therapy to the lung (the recipient lung in the intrapulmonary tracheal transplant model), from this time point will allow us to assess the effect of IL- 10-mediated immunomodulation targeting a chronic-type immune response in the pulmonary milieu. Taking advantage of the intrapulmonary model, we can also examine how IL-10 can modulate de novo lymphoid tissue formation in the lung in this study. Since induction of HEVs appears to be dependent on inflammatory cytokines (e.g. TNF-α) and persistent alloantigens (378), HEVs and lymphoid neogenesis are an interesting potential therapeutic targets of IL

234 A Cyclosporine ( mg/kg/d) Treatment X (e.g. IL-10) Transplant Day 14 Transplant-related injury + Acute-type rejection B No treatment ctrl Day 14 Cyclosporine (7.5 mg/kg/d) Figure 8-3. A preliminary result of initial immunosuppression after rat intrapulmonary tracheal transplantation. (A) A study design of initial immunosuppression after intrapulmonary tracheal transplantation. A rat intrapulmonary tracheal transplantation is conducted from Brown-Norway to Lewis rats and cyclosporine treatment (7.5 mg/kg/d) is initiated at the same time. Another treatment (treatment X) can be started at day 14, when transplant-related injury and acute rejection are overcome. (B) After 14 days, the immunosuppression restored the allograft epithelium and inhibited the development of fibrous obliteration, suppressing acute-type alloimmune responses that occur around day 10 after transplantation if immunosuppression is not used. The epithelium is lost in untreated control allografts and fibrous tissue is developed in an untreated allograft ad day

235 b. Lentivirus-mediated long-term IL-10 gene transfer Since the gene expression of IL-10 is limited to the short-term (usually up to 4 weeks), the experiment using adenovirus-il-10 cannot help to assess the long-term immunomodulatory effect. This limitation could be overcome by using lentiviral vectors. Lentivirus is a retrovirus that transfect cells without inducing major immune responses and realizes long-term gene expression. We have initiated testing lentivirus-mediated gene modification in collaboration with Dr. Jeffrey Medin at the Ontario Cancer Institute and Dr. Paul McCray at the Iowa University Vector Core. Dr. Medin s laboratory has demonstrated lentivirus-mediated gene transfer in mouse lungs lasting more than a year (396). However, at this moment, the amount of vectors available is too small to transfect whole lungs in many animals, particularly in rats. Thus, we will conduct a proof-of-concept study using a simple mouse subcutaneous tracheal transplant model and examine how IL-10 modulates alloimmune responses to allografts in the long-term. An interesting possibility is that, under less-inflammatory circumstances, exposure of IL-10 to T-cells along with antigen presentation might induce IL-10 producing regulatory T-cells (397, 398). Indeed, this regulatory mechanism is considered to be important in suppressing overactivation in the mucosal surface of the lung which is continuously exposed to aerogenous antigens (394). In the lung, dendritic cells play important roles in downregulating T-cell activity by producing IL-10 (394). Although the experimental setting is not in the pulmonary milieu, it is innovative and intriguing to create a local immunoregulatory environment using IL-10 gene transfer in airway allografts and to test how the immune system responds to it. In this study, we will also need to use short-course immunosuppression to overcome the initial acutetype rejection. We will then transfect IL-10-encoding lentivirus to the regenerated epithelial cells in an allograft and observe alloimmune responses over a significant time period: months to a year. Although a similar study has been conducted using a Sendai-virus in a mouse subcutaneous tracheal transplant model (209), the observation period was limited to a month. Long-term immunomodulation using IL- 10 is a novel approach to local chronic immune responses that may lead to potential benefits in the management of lung transplant recipients. 216

236 Following this proof-of-concept study, the next step would be to transfer the knowledge to an intrapulmonary model, in which the long-term IL-10 gene expression occurs in the pulmonary milieu To investigate strategies to modulate tissue remodelling in allograft airway fibrosis For the last few years, we have investigated the role of tissue remodelling in allograft airway fibrosis, mainly using general MMP inhibitors to modulate tissue remodelling, and ultimately to reverse established fibrosis. Although we have demonstrated partial regression of fibrosis (Chapter 6), the therapeutic effect of our current strategy (combination of low-dose MMP inhibitor, SC080 and immunosuppression) appears to be limited in part due to the lack of epithelial regeneration. Since the epithelium is considered to be a critical component to prevent fibrosis, partially open airways with preserved epithelial lining (as often seen in human OB lesions) might be a better therapeutic target than completely obliterated OB lesions with scar tissue. To simulate this condition of OB, we tested a range of suboptimal doses of cyclosporine. We found that 5 mg/kg/d to 7.5 mg/kg/d of cyclosporine is sufficient to: (i) preserve epithelial cells and (ii) not completely block alloimmune responses and subepithelial fibroproliferation in allografts by day 28 in a rat intrapulmonary tracheal transplant model (Figure 8-4A). Because acute aggressive rejection has been one of the limitations of heterotopic tracheal transplant models of OB in studying chronic immune responses, the use of suboptimal immunosuppression appears to be another way to improve the model of OB. Based on these new models, we will further test strategies to modulate tissue remodelling. General MMP inhibitors have limitations including a systemic side effect called musculoskeletal syndrome (349) as well as activation of immune responses (335, 336). On the other hand, there are more selective MMP inhibitors emerging and available, including Tolylsam, a selective inhibitor for MMP-2, MMP-9 and MMP-12 (Shionogi) (399, 400), and PHA-785, a selective inhibitor that preserves collagenolytic MMPs, MMP-1, MMP-8, and MMP-14 (Pfizer). Tolylsam is attractive in its inhibitory effect on MMP-9 and MMP-12, which we found are highly expressed in cultured fibrocytes (Chapter 4). Moreover, MMP-2 that is highly expressed by 217

237 myofibroblasts (Chapter 3 and Chapter 5) might also be an important target, especially for the initiation of fibrosis. Intensive inhibition of MMPs induced apoptosis in myofibroblasts in vivo when myofibroblasts express significantly high levels of MMP-2 (Chapter 5). Since inhibition of MMP-2 could retard collagen turnover in the long-term (Chapter 5), it might be safe to limit its clinical application to the short-term. Theoretically, a good clinical target is the initiation of the fibrotic process at the time of inflammatory and rejection episodes, as an adjuvant of antibiotics or augmented immunosuppression. The inhibitory spectrum PHA-785 is attractive in its preservation of collagenolytic MMPs. PHA-785 particularly preserves MMP-8 and MMP-14, as well as MMP-1 in humans. This inhibitor might preserve collagen degradation under MMP inhibition and divert the tissue remodelling process toward fibrosis regression. Indeed, our preliminary study demonstrates reduced total collagen in established allograft airway fibrosis similar to low-dose SC080 (Figure 8-4B). It would be interesting to test PHA- 785 in a refined OB model, in which the epithelium is preserved. Another interesting anti-fibrotic agent is tranilast. Tranilast is an antibiotic-derived agent that was developed and has been used to prevent and treat excessive cutaneous scar formation in Japan. Currently, tranilast is orally available and the systemic administration has been suggested to be effective in various tissue remodelling disorders such as chronic cardiac matrix deposition and dysfunction as well as renal fibrosis ( ). Also, I myself used this agent in a clinical case of refractory tracheal stenosis of fibrogranulation and reported its beneficial effect (272). Tranilast has been demonstrated to reduce obliterative fibrosis in a subcutaneous tracheal transplant model of OB (273). Mechanistically, tranilast has been demonstrated to reduce the production of collagen type-i and III as well as TGF-β-induced procollagen gene expression in human fibroblasts (404). Tranilast has been demonstrated to reduce production of MMP-7, MMP-8 and MMP-9 by neutrophils (405) and production of MMP-2 by nasal fibroblasts (406). Conversely, tranilast promotes expression of MMP-1, which is an important collagenolytic MMP in humans, and suppresses TGF-β production by fibroblasts 218

238 (407). Tranilast has been demonstrated to inhibit cytokine-induced NF-κB activation and upregulation of adhesion molecules in endothelial cells (408). Considering the mechanistic analyses of tissue remodelling in allograft airway fibrosis, these characteristics of tranilast suggests that it may have potentially beneficial effects. We are currently promoting a collaboration with Dr. Robert Gilbert at St. Michael s Hospital, Toronto, whose laboratory has reported promising anti-fibrotic effects of tranilast in cardiac and kidney diseases ( ). Importantly, unlike general MMP inhibitors, the safety of this agent has been well-documented in clinical cases in Japan. Once the therapeutic effect is demonstrated, this agent may be translated to application in clinical practice relatively easily. It is an exciting next step for us to test this agent in a clinically relevant model of tissue remodelling such as intrapulmonary tracheal transplant model of OB in combination with initial cyclosporine or continuous cyclosporine treatment (as tested in Chapter 6) and to examine whether this agent is capable of reversing established fibrosis of allograft airways. 219

239 A Suboptimal cyclosporine (5-7.5 mg/kg/d) Treatment X Transplant Day 28 No treatment ctrl Day 28 Cyclosporine (5.0 mg/kg/d) Cyclosporine (7.5 mg/kg/d) B Figure 8-4 Preliminary results of modified intrapulmonary tracheal transplant model and the effect of a novel MMP inhibitor. (A) Suboptimal cyclosporine (5 7.5 mg/kg/d) allows for the development of fibrosis, while preserving the airway epithelium by day 28. This partially obliterative fibrosis is similar to that seen in approximately 50% of human OB lesions in the lungs of BOS patients. Since epithelial recovery and regeneration is considered to be an important component to reverse fibrosis, this type of OB lesion might be a realistic therapeutic target of anti-fibrotic treatments. A novel therapeutic strategy can be tested from day 28 (Treatment X in the study design panel). (B) A preliminary study using selective MMP inhibitor, PHA-785. An MMP inhibitor preserving the activity of a collagenolytic MMP, MMP-14, was started at day 21, when allograft airway fibrosis is established in an intrapulmonary tracheal transplant model. The number of myofibroblasts and the amount of collagen deposition was reduced by day