TWO NOVEL MECHANISMS OF MHC CLASS I DOWN- REGULATION IN HUMAN CANCER: ACCELERATED DEGRADATION OF TAP-1 mrna AND DISRUPTION OF TAP-1 PROTEIN FUNCTION

Transcription

1 TWO NOVEL MECHANISMS OF MHC CLASS I DOWN- REGULATION IN HUMAN CANCER: ACCELERATED DEGRADATION OF TAP-1 mrna AND DISRUPTION OF TAP-1 PROTEIN FUNCTION DISSERTATION Presented in Partial Fulfillment of the Requirement for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University by Tianyu Yang * * * * * The Ohio State University 2004 Dissertation Committee: Professor Pan Zheng, Adviser Approved by Professor Yang Liu Professor Joan Durbin Professor Xuefeng Bai Professor William Lafuse Adviser Department of Pathology

2 ABSTRACT Both viruses and tumors evade cytotoxic T lymphocyte (CTL)-mediated host immunity by down-regulation of major histocompatibility complex (MHC) class I antigen presentation machinery. The transporter associated with antigen processing (TAP), a heterodimer of TAP-1 and TAP-2, plays essential role in the MHC class I- restricted antigen presentation pathway by translocating antigenic peptides from cytosol into the ER lumen, where the assembly of MHC class I complex takes place. TAP deficiency is a frequent observation in human cancers, which is one of the causes of MHC class I down-regulation. However, the underlying molecular mechanism has been limited. In search for novel mechanisms of TAP deficiency and MHC class I downregulation in human cancer cells, we characterized an MHC class I-deficient melanoma cell line SK-MEL-19. The expression of TAP-1 mrna was found deficient in this cell line even after interferon-gamma (IFN-γ) stimulation, despite its active transcription. This abnormality is caused by a single nucleotide deletion at position of the TAP- 1 gene, which results in rapid degradation of the TAP-1 mrna. Subsequently, using this TAP-1 deficient cell line, we studied the function of TAP-1 variants containing amino acid changes within or close to the signature motif that exist in the normal population or were generated by site-directed mutagenesis. All the TAP-1 variants showed decreased ii

3 function compared to the wild type, which supports an essential role of the signature motif in TAP transporter function. One of these variants, TAP-1 R648Q, with the alteration immediately C-terminal to the signature motif, was found to occur 17.5 times more frequently in HLA - colon cancers than those with normal HLA levels. Functional analysis revealed that the Q648 variant retained only about 20% peptide translocation activity compared to TAP-1 (R648). To our knowledge, the two mechanisms we identified in this work, which lead to abrogation of TAP-1 expression and decreased TAP transporter function, respectively, have not been addressed before. The determination of transporter activity of a natural TAP-1 variant, TAP-1 R648Q, and its increased presence in HLA - colon cancer samples may help to develop diagnostic and therapeutic methods for colon cancer immunotherapy. iii

4 DEDICATION Dedicated to my parents. iv

5 ACKNOWLEDGMENTS I wish to thank my adviser, Dr. Pan Zheng for her support and encouragement, and for her patience in correcting my stylistic and scientific errors. I thank Dr. Yang Liu for his valuable advice. Without them, this thesis would not be possible. I thank Philip E. Lapinski and Haotian Zhao for their contribution to this work. I m in debted to Dr. Soledad Fernandez for the statistical analysis and Dr. Qunmin Zhou for sharing normal population blood DNA samples. I thank Dr. Jian-Xin Gao, Dr. Huiming Zhang, Dr. Kurtis Yearsley and Paul Rangel for technical assistance. I am grateful to Dr. Soldano Ferrone for providing us the SK-MEL-19 cell line, Dr. Malini Raghavan for providing the peptide and technical assistance for the peptide translocation assay and Dr. Peter Cresswell for rabbit antiserum against TAP1 and tapasin. This work was supported by NIH grants CA82355, CA58033 and CA69091, Department of Defense grant DAMD and the Ohio State University Comprehensive Cancer Center. v

6 VITA November 8, 1974 Born, Hebei, People s Republic of China Bachelor in Medicine and Master in Medicine, China medical University present Graduate Research Assistant, The Ohio State University PUBLICATIONS Research Publication 1. T. Yang, B.A. McNally, S. Ferrone, Y. Liu, and P. Zheng, A single-nucleotide deletion leads to rapid degradation of TAP-1 mrna in a melanoma cell line. J Biol Chem, 278(17), , (2003). 2. Y. Guo, T. Yang, X. Liu, S. Lu, J. Wen, J.E. Durbin, Y. Liu, and P. Zheng, Cis elements for transporter associated with antigen-processing-2 transcription: two new promoters and an essential role of the IFN response factor binding element in IFNgamma-mediated activation of the transcription initiator. Int Immunol, 14(2), , (2002). Major Field: Immunology. FIELDS OF STUDY vi

7 TABLE OF CONTENTS Page ABSTRACT... ii DEDICATION...iv ACKNOWLEDGMENTS...v VITA...vi LIST OF TABLES...ix LIST OF FIGURES...x 1. INTRODUCTION Major histocompatibility complex (MHC) class I-restricted antigen processing and presentation Presentation of exogenous antigens by the class I MHC molecules TAP-1 gene expression and protein function, Impaired MHC Class I expression in viral infection MHC Class I down-regulation in cancer MATERIALS AND METHODS Cell lines Antibodies Flow cytometry Northern blot Southern blot Generation of TAP-1 cdna constructs and stable transfection Generation of TAP-1 constructs and Tet-Off SK-MEL-19 cell lines for RNA stability analyses Generation of luciferase reporter constructs and assay for promoter activity Nuclear run-on assay Restriction fragment length polymorphism (RFLP) RNase protection assay Immunofluorescence microscopy Immunoprecipitation Microsome-based peptide translocation assay Colon cancer tissue sections, immunohistochemistry and statistical analysis Isolation of genomic DNA and genotyping A SINGLE-NUCLEOTIDE DELETION LEADS TO RAPID DEGRADATION OF TAP-1 MRNA IN A MELANOMA CELL LINE Introduction Results and discussion Down-regulation of TAP-1 mrna by a posttranscriptional mechanism in melanoma cell line SK-MEL A single-nucleotide deletion leads to accelerated decay of TAP-1 mrna...47 vii

8 4. FUNCTIONAL CONSERVATION BETWEEN THE SIGNATURE MOTIF OF TAP-1 AND CYSTIC FIBROSIS TRANSMEMBRANE REGULATOR PROTEIN: A ROLE OF ITS POLYMORPHISM IN HLA DOWN-REGULATION IN CANCER CELLS Introduction Results CF-like mutations in TAP-1 signature motif abrogate TAP-1 activity TAP-1 G646D is expressed in the ER and remains associated with the peptideloading complex TAP-1 G646D failed to transport peptide across the ER membrane Increase of the G1943A allele frequency in HLA low colon cancer tissues Function of TAP-1 R648Q Discussion CONCLUSION BIBILIOGRAPHY...99 viii

9 LIST OF TABLES Table Page Table 4.1 Nucleotide changes and amino acid substitutions generated by the site directed mutagenesis Table 4.2 TAP-1G1943A genotyping in different types of human cancer...80 ix

10 LIST OF FIGURES Figure Page Figure 1.1 Conventional MHC class I antigen processing and presentation pathway...26 Figure 1.2 Genomic localization and promoter structure of human TAP-1 gene Figure 1.3 Alignment of the nucleotide-binding domains (NBDs) of ABC transporters Figure 3.1 Deficiency of surface HLA class I expression in melanoma cell line SK-MEL-19 was due to the TAP-1 down-regulation...52 Figure 3.2 Posttranscriptional mechanisms are responsible for poor accumulation of TAP-1 mrna...55 Figure 3.3 TAP-1 mrna level in SK-MEL-19 cells was increased by CHX...58 Figure 3.4 A homozygous single-nucleotide deletion was identified in the TAP-1 gene at position that resulted in premature termination codons Figure 3.5 The single-nucleotide deletion D1489 in the TAP-1 gene results in accelerated decay of the TAP-1 mrna by a non-nmd-related mechanism...62 Figure 4.1 Sequence alignment of the NBDs of ABC transporters Figure 4.2 Expression of TAP-1 mutant proteins and cell surface MHC class I in cells stably transfected with wild type or mutant TAP Figure 4.3 TAP-1 G646D was expressed in the ER and associated with TAP-2, tapasin, and MHC class I heavy chain Figure 4.4 TAP-1 G646D failed to transport peptide across ER membrane Figure 4.5 Immunohistochemical staining of colon cancer with HC-10 mab Figure 4.6 Defective activity of TAP-1 R648Q x

11 CHAPTER 1 INTRODUCTION 1.1 Major histocompatibility complex (MHC) class I-restricted antigen processing and presentation. CD8 + cytotoxic T lymphocytes (CTLs) are the major players of immune system to fight against tumor cells and viral infections. CTLs recognize antigenic peptides presented by class I MHC molecules on the target cell surface. The interaction between T cell receptors (TCRs) and peptide-mhc complexes is a fundamental event for the adaptive immune system. It mediates the positive and negative selections during the thymus development of CD8 + T cells. It is indispensable for the survival and activation of naïve T cells, effector function of activated T cells and their transition to memory cells [1, 2]. Blocking surface expression of MHC class I in host cells is one of the strategies certain viruses use to escape immune surveillance by CTLs [3-5]. Down-regulation of MHC class I on tumor cells has also been frequently observed in a variety of cancers, which is thought to facilitate tumor cells to evade immune surveillance by CTLs and poses a major obstacle to the T-cell based cancer immunotherapy [6, 7]. 1

12 MHC class I molecules are expressed by virtually all the nucleated cells except those of the immune-privileged organs, such as brain, testis and eyes. The MHC class I complex expressed on the cell surface consists of two non-covalently associated polypeptide chains, an α or heavy chain (HC) and a β2-microglobulin (β2m) or light chain, as well as the antigenic peptide. The gene encoding for the class I HC is located in the MHC region on the short arm of human chromosome 6 and the corresponding position of murine chromosome 17. The HC of class I MHC is a trans-membrane glycoprotein, whose extra-cellular part folds into three domains: α1, α2 and α3. The α3 domain interacts with the non-membrane spanning β2m and the folded structure they form closely resembles that of the immunoglobulin constant domain. The α1 and α2 domains, which are distal to the membrane, form the peptide-binding platform and make direct contact with TCR [8]. Like other membrane proteins, MHC class I heavy chain is synthesized on the ER membrane with its extra-cellular domain exposed to the ER lumen. It is inside the ER where HC associates with β2m and is subsequently loaded with antigenic peptides. The generation of antigenic peptides from intact proteins, their assembly with class I HC and β2m, as well as their final display on the cell surface have been studied extensively in the past two decades [9-11]. Figure 1.1 shows a simplified representation of the conventional class I antigen processing and presentation pathway. To briefly describe, antigenic peptides are generated from endogenous proteins by the proteasome, and then transported into the ER lumen by the transporters associated with antigen 2

13 processing (TAP). Inside the ER, some of the antigenic peptides will be loaded onto the empty class I HC-β2m heterodimer with the help of a so-called peptide-loading complex, which is comprised of tapasin, calreticulin, ERp57 and TAP. Class I restricted antigenic peptides are mostly generated in the cytosol from endogenous cellular proteins or intracellular pathogen derived proteins, mostly from virus. Proteasome, the major proteolytic machinery in eukaryotic cells that degrades ubiquitin-conjugated proteins or polypeptides, provides the vast majority of antigenic peptides for the class I restricted presentation [12-14]. The 20S proteasome is composed of two heptameric outer rings of α subunits and two hexameric inner rings of β subunits arranged in a hollow-cylinder shape. Upon cytokine stimulation, such as gammainterferon (IFN-γ), three active β subunits of the 20S proteasome, X, Y and Z, are replaced by the multicatalytic endopeptidase complex-like 1 (MECL1), low molecular weight proteasome subunit (LMP)-2 and LMP-7 [15-18]. The LMP-2/7-containing proteasome exhibits enhanced trypsin- and chymotrypsin- like activity, which generates peptides with a basic or hydrophobic C-terminal residue that favors their subsequent binding to the TAP transporter and MHC class I molecules [19-22]. Although LMP-2 and LMP-7 are dispensable for the overall class I antigen presentation in some studies [23-25], decreased cell surface class I expression and defects in class I-restricted T cell response have been observed with the LMP-2 or LMP-7 deficient mice [26, 27]. The 20S proteasome can function independently or with regulatory subunits associated at both ends, which greatly enhances the proteolytic activity of the 20S proteasome. Such regulatory subunits include the 19S (PA700) subunit [28], which is 3

14 comprised of multiple subunits including six ATPases, and the 11S (PA28; REG) regulatory complex [29], which consists of two IFN-γ inducible subunits PA28α and PA28β. Together with the 20S proteasome core, they form the 26S proteasome (19S-20S- 19S), the PA28-proteasome complex (11S-20S-11S) or the 19S-20S-PA28 complex, which are all involved in the generation of antigenic peptides for MHC class I-restricted presentation. However, proteasome is not the only source of antigenic peptides presented by class I MHC. Recently, cytosolic peptidase that can generate antigenic peptides independent of proteasome system has been identified, such as tripeptidyl peptidase II (TPPII) [30]. In human, some of the signal peptides produced in the ER during protein synthesis can bind to the MHC class I molecules directly [31]. The carboxyl termini of antigenic peptides generated by the proteasome are hydrophobic or charged, which favors their bindings to MHC [32]. The amino termini, however, are mostly extended and vary greatly. Some cytosolic aminopeptidases have been identified that are involved in trimming the amino termini of these peptides, which include tripeptidyl peptidase II, puromycin-sensitive aminopeptidase (PSA), bleomycin hydrolase (BH) and the IFN-γ inducible leucyl aminopeptidase (LAP) [33, 34]. In the cytosol, antigenic peptides are subjected to the attack by other peptidases, such as thimet oligopeptidase (TOP), and some may be destroyed completely [35]. In fact, most peptides are lost at this stage before being presented by class I MHC molecules [36]. 4

15 It is not clear yet whether the peptides that survive are chaperoned in the cytosol while being modified or protected on their way to the ER. Cytosolic heat shock proteins (HSPs), such as Hsp70 and Hsp90, are the possible candidates. The important role of cytosolic HSPs in class I antigen presentation pathway is supported by studies by Binder et al [37], in which they treated cells with deoxyspergualin, a drug that binds hsp70 and hsp90, and observed abrogation of MHC I antigen presentation, which was restored by introduction of additional hsp70 into the cytosol. In another study, overexpression of hsp72 in the B16 melanoma cell line leads to enhanced antigen presentation by class I MHC on the cell surface [38]. Although physical association of HSPs with proteasomes and TAP transporters has been detected, respectively [39, 40], the association of antigenic peptides or their precursors with HSPs or other chaperone proteins in the cytosol has not been directly detected. The peptides that manage to reach the ER membrane may bind to the TAP transporters on the ER membrane. TAP transporters are capable of translocating the peptides from cytosol into the ER lumen utilizing energy released from ATP hydrolysis. Inside the ER, the newly synthesized MHC class I HC binds to ER chaperone protein calnexin or immunoglobulin binding protein (BiP), which facilitates its proper folding and assembly with β2m [41]. The empty HC-β2m heterodimer will subsequently associate with multiple ER-resident proteins to form a peptide-loading complex, which consists of: calreticulin, another ER chaperone protein [42, 43]; ERp57, a thio-reductase involved in disulfide bond formation of MHC class I molecules [44-46]; tapasin, a transmembrane glycoprotein that mediates the association of HC-β2m heterodimer with 5

16 TAP transporter [42, 47]; and the TAP transporter, which forms the core of the loading complex and provides antigenic peptides generated in the cytosol. Tapasin is the key component in this complex. Unlike other chaperone proteins, such as calnexin and calreticulin, which are dispensable for MHC class I expression, tapasin is specific for this process and its deficiency results in decreased cell surface MHC class I level [42, 47]. Tapasin is capable of interacting with both TAP-1 and TAP-2 through its C-terminal region and with MHC class I HC via the 50 residues on its N-terminus [48]. The interaction between tapasin and TAP is sufficient to stabilize TAP proteins and increase their expression levels without significantly affecting the intrinsic translocation activity [48, 49]. Tapasin is also linked to ERp57 via disulfide-bond formation, which has been shown required for the complete oxidation of class I HC in the loading complex [50]. However, the interaction between tapasin-mhc complex and TAP is not necessary for the peptide loading as a soluble form of tapasin, which is unable to bind TAP, still supports the loading of MHC class I molecules with antigenic peptides [51, 52]. It is thought that formation of this multimolecular peptide-loading complex may allow the assembly of MHC class I molecules with their peptide cargo to optimize over time. The antigenic peptides transported by the TAP transporters are usually 8-15 amino acid long, which is longer than those presented by the class I MHC molecules (8-10 amino-acid long). As mentioned above, although the carboxyl termini of these peptides are hydrophobic or charged, which favor their bindings to MHC, their amino termini can be varied and extended, which are subjected to further trimming by the ER aminopeptidases until the best fitters are selected by the empty HC-β2m dimers. Such an 6

17 ER resident aminopeptidases has been identified recently, named ER aminopeptidase I (ERAPI). The expression of ERAPI can be induced by IFN-γ and its deficiency results in decreased cell surface expression of class I MHC complexes in IFN-treated cells [53-55]. ERAPI was shown capable of cleaving peptide precursors of 10-residue or longer, generated in cytosol or ER, to 8-9 residues that are fit for presentation by class I MHC molecules [53-55]. Peptides of 8-residue or shorter are spared by ERAPI. It destroys about half of the 9-residue peptides in vitro, which may limit antigen presentation and explain the phenomenon that without IFN-γ treatment, lack of ERAPI increases class I surface expression [54, 55]. After the MHC class I complexes are loaded with proper antigenic peptides, they will dissociate from the peptide-loading complex, be released from ER and go to the cell surface via the secretory expression pathway. Those that fail to bind any peptides will be retained in the ER and eventually translocated out of the ER via the translocon and subsequently degraded by the proteasome in the cytoplasm [56, 57]. Only under temperature below physiological level, the unstable empty MHC class I molecules can be expressed on the cell surface [58]. 1.2 Presentation of exogenous antigens by the class I MHC molecules. The conventional class I antigen presentation pathway described above is the machinery by which endogenous antigens are processed and displayed on the cell surface, which then can be scrutinized by CD8 + T lymphocytes, the major players of immune system to fight against viral infection and cancer cells. However, the naïve CD8 + 7

18 T cells cannot be activated unless two conditions are fulfilled: 1) two stimulatory signals are required [59], one delivered through the engagement of MHC class I/antigen and T cell receptor/cd8, the other through the interaction between the co-stimulatory molecules on T cells (such as CD28) and their ligands on the antigen presenting cells (APCs) (such as B7.1 and B7.2). Although MHC class I complexes are universally expressed, ligands for the costimulatory molecules are only expressed by the professional APCs, including dendritic cells (DCs), macrophages and B cells. 2) The antigens have to be seen by the naïve CD8 + T cells, which, however, circulate in the blood and lymphatics after their maturation in thymus and do not go to the peripheral tissues [60]. Lymphoid organs, especially lymph nodes (LNs), are the most likely sites of T cell activation [61]. Therefore, when tissue cells other then the professional APCs are infected with virus or undergoing malignant transformation, they are unable to alert the CD8 + T cells by themselves. The immune system has evolved a way to activate the naïve CD8 + T cells in such a situation that the antigens are picked up by the professional APCs that are billeted in the peripheral tissues, especially the bone marrow-derived DCs, which if turning activated and mature, will migrate to the draining LN and present the anithgens to the naïve CTLs [62-64]. The process that exogenous antigens are processed and presented by the class I MHC molecules on the professional APCs is termed cross-presentation. Traditionally, exogenous antigens are presented by the class II MHC complexes, which, unlike the class I MHC, are expressed only by the professional APCs on their surface. Since the emphasis of this thesis is on the class I antigen presentation machinery, the class II-restricted antigen presentation pathway is summarized here in an over- 8

19 simplified way. The exogenous antigens are taken up by the professional APCs via endocytosis or phagocytosis, and then degraded by proteases in the acidic endocytic compartments. The class II molecules are synthesized in the ER, where they associate with the invariant chain (Ii) molecule, which not only prevents the binding of antigenic peptides in the ER, but also targets the complexes to a specialized endosomal compartment, MHC class II-containing compartment (MIIC), where they will be loaded with the antigenic peptides derived from the exogenous antigens and then expressed on the cell surface. The presentation of exogenous antigens by class I MHC molecules is distinct from both the conventional class I antigen presentation and the class II antigen presentation machinery. Up to date, although not fully understood, two cross-presentation pathways have been proposed, the phagosome-to-cytosol pathway and the post-golgi loading pathway or vacuolar pathway. In both pathways, the exogenous antigens are internalized by phagocytosis or endocytosis. The internalization process is much more efficient when the exogenous antigens are associated with antibodies, which can bind to the Fc receptors on the surface of APCs, or when the exogenous antigens are chaperoned by the HSPs, which APCs also express surface receptors for, such as CD91 [65, 66]. In the phagosome-to-cytosol pathway, upon entry into the cells, phagosomes containing exogenous antigens may fuse with the ER and thereby acquire the empty class I MHC HC-β2m dimers as well as other ER resident proteins, including those involved in the conventional class I restricted antigen presentation pathway [67, 68]. The exogenous antigens are then translocated out 9

20 of the phagosomes via a translocon Sec61 (or possibly by other mechanisms) and ubiquitinated by the ubiquitin-conjugating enzymes (UBCs) nearby [67, 68]. The antigens are subsequently cleaved by the proteasomes that are associated with the phagosome membrane. Or they may diffuse through the cytosol, which subjects them to degradation by the free proteasomes and cytosolic proteases. The antigenic peptides with proper length and sequence will be transported back into the ER-fused phagosomes or ER by the TAP transporters and then follow the conventional class I antigen presentation pathway as described above. In the vacuolar pathway, the exogenous antigens remain in the phagosomes and are degraded in situ by the proteases present in the phagosomes. It is not clear yet where the MHC class I molecules come from. One possibility is that the phagosomes fuse with the ER, as in the phagosome-to-cytosol pathway, and thus acquire ER proteins including the empty MHC class I HC-β2m dimers and tapasin. The other possibility is that the cell surface MHC class I molecules have been taken into the phagosomes together with the exogenous antigens. In the latter case, the MHC class I molecules are very likely to be peptide-loaded and a peptide exchange process might occur in the endosomal compartments, probably with the help of tapasin [69]. After loaded with antigenic peptides, these MHC class I complexes could recycle to the cell surface. In support of the existence of such a vacuolar pathway, peptide exchange on class I molecules at low ph condition has been observed in vitro [70]. Mice that express transgenic GPI-linked MHC class I molecules, which can not be recycled, fail to generate CTL response against 10

21 particular antigenic epitope, which indicates that MHC class I recycling is required for presentation of certain antigenic epitopes [71]. However, the vacuolar pathway is rather speculative and needs more experimental evidence. 1.3 TAP-1 gene expression and protein function, As described above, TAP transporter plays pivotal role in both the conventional class I MHC restricted antigen presentation and the cross-presentation pathway. It is the only known transporter on the ER membrane providing access for the antigenic peptides generated in the cytosol to the ER lumen. The importance of TAP transporter to the class I restricted-antigen presentation was first noticed in studies on several mutant cell lines that were mutated and then selected for the loss of class I expression on the cell surface. These cell lines include the derivatives of a human B-lymphoblastoid cell line B-LCL 721, and , as well the mutant RMA-S cell line generated from murine lymphoma RMA cells [72, 73]. In these mutant cell lines, MHC class I HC and β2m are expressed but fail to bind antigenic peptides, and thus unable to be presented on the cell surface under physiological condition. The surface expression of MHC class I molecules can be restored when peptides are given exogenously [74]. The genetic defects were later mapped to the MHC class II gene locus, from where two genes were cloned, now known as TAP-1 and TAP-2 [75-78]. The class I expression defects in the mutant cell lines were corrected when TAP-1 and/or TAP-2 were expressed [79-81]. Subsequently biochemical experiments show that TAP-1 and TAP-2 polypeptides form a heterodimeric transporter 11

22 on the ER membrane and transport antigenic peptides across the ER membrane in an ATP-dependent manner [82-84]. The necessary role of TAP in the class I antigen presentation pathway was further supported by the TAP-1 knockout mice, which showed severely decreased surface MHC class I expression and impaired CD8 + T cell development [85]. Genes that encode for TAP-1 and TAP-2 are clustered with those for the proteasome subunits, LMP-2 and LMP-7 (Figure 1.2). TAP-1 and LMP-2 genes share a bi-directional promoter of 593 base-pair long [86]. Although there is no TATA box homology at either end, a prevalence of GC boxes is prominent [86]. The region proximal to the TAP-1 gene has been shown to control the maximal basal level expressions of both TAP-1 and LMP-2 and several cis elements have been identified in this region, including an Sp1 element, an NF-kappa B binding site, the partially overlapping IFN consensus sequence (ICS)-1/2 and a gamma activating sequence (GAS) [86]. The Sp1 site is required for the basal expression of both genes and the NF-kappa B binding site is essential for the induction of both genes by tumor necrosis factor-alpha (TNFα) [86]. The rapid induction of the promoter by IFN-γ is mediated by STAT-1 that binds to the GAS element [87, 88]. The binding of IRF-1 to the ICS/GAS region supports constitutive expression of both genes and augments the induction by IFN-γ [89]. In the IRF-1- deficient mice, expressions of TAP-1 and LMP-2, as well as the surface class I MHC are greatly reduced, which leads to the defective CD8 + T cell development [89, 90]. In human melanoma cells, TAP-1 is constitutively expressed, the expression of LMP-2 12

23 gene, however, requires IFN-γ stimulation [91]. Further analysis show that whereas binding of either IRF-1 or STAT-1 is sufficient for the transcription of TAP-1 gene, binding of both factors is required for LMP-2 gene transcription [91]. TAP-1 and TAP-2 polypeptides belong to the ATP-binding-cassette (ABC) superfamily, which is comprised of a diverse class of proteins including not only the transmembrane transporters that carry various substances, ranging from ions to proteins, across membrane, but also the non-transporter enzymes that are involved in DNA repair or structural maintenance of chromosomes [92, 93]. A prototype ABC transporter is composed of two paired membrane integral domains and two paired nucleotide-binding domains (NBDs). For some ABC transporters, such as CFTR, these domains are fused as a single polypeptide chain, while for others, like TAP, two polypeptides are separately encoded and each consists of one transmembrane domain and one NBD. In ABC transporters, the paired hydrophobic membrane-spanning regions are thought to form the transmembrane pore through which substrates get across the membrane. The precise membrane topology of TAP-1 and TAP-2 is still controversial. It is predicted that the pore-forming domains of TAP-1 and TAP-2 assume a head-head/tailtail orientation to each other [94]. Both N- and C- termini of TAP-1 reside in the cytoplasm, whereas for TAP-2, the N-terminus is facing the ER lumen and the C- terminus remains in the cytosol [94-97]. At least 8 and 7 transmembrane fragments have been proposed for TAP-1 and TAP-2, respectively [94-97]. Despite the lack of precise 13

24 structure, the membrane-spanning regions of TAP-1 and TAP-2 have been shown important for their ER retention, dimerization, interaction with tapasin, peptide binding, as well as peptide translocation [94-97]. The substrate binding is not conserved among ABC family members. The antigenic peptide binding sites of TAP have been mapped to the regions of the cytosol- ER membrane boundaries of the two transmembrane segments close to the ATP binding site of NBDs [98-100]. TAP preferentially transports peptides 8-15 residue long [101]. Peptides longer or shorter can still be translocated, but with much reduced efficiency. The amino acid composition of peptide, especially on the carboxyl terminus, is also critical for TAP-mediated translocation [ ]. Human TAP has been shown to efficiently transport peptides with hydrophobic or basic C-termini, which are characteristic of those produced by the IFN-induced proteasome [ ]. In addition, the N-termini of peptides may also limit their access to the ER lumen [105]. The NBDs of ABC transporters, including TAP, are paired and serve as an ATPase that binds and hydrolyzes ATP molecules to provide energy for the substrate translocation process. Like other members of the ABC superfamily, TAP-1 and TAP-2 contain several conserved sequence motifs in their NBDs, including Walker A and Walker B, signature motif or C loop, Q loop, D loop and switch region (Figure 1.3). The Walker A and Walker B domains are required for the ATP binding and ATP hydrolysis, whereas roles of the other motifs are largely unknown. In the past ten years, the peptide translocation process of TAP has begun to be elucidated. By disrupting the Walker domains of TAP-1 and TAP-2 individually, it has been shown that nucleotide binding by 14

25 TAP precedes peptide binding [106]. TAP-1 and TAP-2 play distinct roles in this process that nucleotide binding and hydrolysis by TAP-2 is required for the binding of peptide, whereas hydrolysis by TAP-1 is not required [ ]. Successful peptide binding then induces both the ATPase activity of TAP and the conformational changes associated with peptide translocation [110, 111], for which the integrity of the Walker domains of both TAP subunits is necessary [107, 108]. It has been shown that the NBDs of ABC transporters, including TAP, interact with each other, which is thought to mediate the domain communication and conformational transmission in the substrate translocation process [112]. The signature motif with consensus LSGGQ/E sequence has been shown important for the NBD interaction [ ]. Signature motif is conserved among all the ABC transporters and is specific for the ABC superfamily. Recent studies on the crystal structure of several ABC ATPase have shown that in the biologically relevant dimer of NBDs, the signature motif of one NBD forms a composite ATP-binding site at the dimer interface with the Walker motifs of the other NBD [113, 116]. The two most conserved residues of signature motif, the serine and second glycine, form specific hydrogen bonds to the phosphate oxygens of the ATP molecule [113, 116]. In this way, the signature motif may act as a γ-phosphate sensor and mediate the ATP-driven conformational changes during the substrate translocation process. The important role of signature motif in the ABC transporter function has been reflected by the fact that its mutations led to loss or impairment of the transporter function of several ABC transporters, including the human cystic fibrosis transmembrane 15

26 conductance regulator (CFTR) [ ], yeast cadmium factor (Ycf1p)[121], yeast a- factor transporter Ste6p[122, 123], the bacterial maltose-transport system ATPase subunit MalK [124], and TAP [125]. However, it is not known whether the function of signature motif is conserved among these ABC family members. In humans, mutations in the signature motif of CFTR are among the frequent causes of cystic fibrosis (CF) [126]. Whether the signature motifs of other ABC transporters are also prone to mutations has not been addressed. Although natural variants of both human TAP-1 and TAP-2 have been identified in the normal population, the amino acid changes in these variants are not in the signature motif. Since congenital TAP deficiency in humans results in lack of cell surface MHC class I expression, the type 1 bare lymphocyte syndrome [127], and TAP deficiency has been frequently abserved in a variety of human cancers, which causes surface MHC class I down-regulation of cancer cells, it will be interesting to examine whether the signature motif of TAP are frequently mutated in these disease conditions. 1.4 Impaired MHC Class I expression in viral infection. CTLs are the soldiers that fight against viral infection in our body. They need to see the targets, the antigenic peptides presented by the cell surface MHC class I molecules, to fulfill their job. To disguise themselves, the intruding viruses have evolved complicated strategies to block the expression of MHC class I on their host cells. Every step in the class I antigen processing and presentation pathway can be interfered by these sophisticated mechanisms. For example, the Epstein-Barr virus encoded nuclear antigen 16

27 (EBNA)-1 contains internal glycine-alanine repeats that inhibit its proteasomal degradation, therefore the antigenic peptides cannot be produced [128]. The Herpes simplex virus (HSV)-1 and HSV-2 express an infected cell protein (ICP)-47. ICP-47 inhibits translocation of antigenic peptides from cytosol into the ER by binding to the TAP transporters and thus preventing other peptides from binding to TAP [ ]. The human cytomegalovirus (HCMV) genomic unique short (US) region encodes a family of homologous genes essential for the inhibition of MHC class I-mediated antigen presentation during viral infection. Among them, US6, a transmembrane protein on the ER membrane, interacts with TAP transporter via its ER luminal domain [132]. Unlike ICP-47, US6 blocks the peptide translocation activity of TAP without affecting peptide binding [133]. The MHC class I molecules themselves can also become the targets of viruses. US2 and US11, glycoproteins expressed by HCMV, bind to and dislocate nascent MHC class I HC from the ER into the cytosol through translocon Sec61 [56, 134, 135]. The class I HC subsequently will be deglycosylated and degraded by the proteasome [56, 134, 135]. The mk3 protein of γ2-herpesvirus 68 also targets the degradation of nascent class I molecules via the ubiquitin/proteasome pathway [ ]. However, the mk3 protein has to be associated with members of the peptide-loading complex (TAP/tapasin) [139]. The association is not only critical for stable mk3 expression, but also required by mk3 to specifically recognize and ubiquitinate class I MHC molecules [139]. Adenovirus type 2 and 5 express a 19KD glycoprotein, E3/19K, that has two ways to interfere with class I surface expression: 1) it binds to TAP and acts as a tapasin inhibitor, thus preventing 17

28 class I/TAP association, which delays the maturation of class I MHC complexes [140]; 2) it retains the peptide-loaded MHC class I molecules in the ER/cis-Golgi compartment [141, 142]. Another viral product that can cause retention of peptide-associated MHC class I molecules in the ER is US3, an ER resident glycoprotein expressed by HCMV [143, 144]. After being expressed on the cell surface, MHC class I complexes can still be down-regulated. For example, the Nef protein of human immunodeficiency virus (HIV) is able to target surface class I MHC to accelerated endocytosis [145]. Above all, all the key steps in the class I restricted antigen presentation pathway can be targeted by viruses to decrease the surface presentation of viral antigens by the host cells. One virus may utilize different escape strategies at different times during the infectious cycle, such as the US proteins of HCMV. 1.5 MHC Class I down-regulation in cancer. MHC class I down-regulation has been frequently encountered in a variety of cancers, which is thought to facilitate tumor cell escape from immune surveillance by CTLs [7, 146, 147]. Loss of HLA class I in human cancers can be at different levels, i.e. total loss of class I, loss of expression of one locus or one haplotype, or even one specific allele [148]. Loss of class I expression is observed more frequently in metastatic lesions than in primary lesions from studies of various of cancers, such as breast cancer, squamous cell carcinoma of the head and neck, cervical carcinoma, prostate cancer, colon cancer, etc [ ]. It has been correlated with disease progression and may serve as a marker of poor prognosis for certain human cancer [153, 157, 158]. In a mouse model, 18

29 mice that develop breast carcinoma spontaneously due to the expression of HER-2/Neu oncogene showed down-regulation of MHC class I expression as the tumor progressed [159]. In transgenic mice that develop melanoma as a result of the expression of SV40 large T antigen in melanocytes, MHC class I expression was also found to decrease with tumor growth [160]. It has been shown that tumor cells that have lost surface MHC class I grow in the presence of pre-existing strong anti-tumor immune response, which provides direct evidence that loss of MHC class I expression may result in immune evasion of tumor cells [161]. In the past two decades, research on tumor immunology and human tumor immunotherapy has made tremendous progress [162]. Since antibody treatments are not effective except those using antibodies against growth factor receptors, tumor immunotherapy in human has been focused on generating tumor specific cellular immune responses, which can be achieved by active immunization or passive transfer and further enhanced by cytokine treatment, such as interleukin (IL)-2 [162]. As more and more class I-restricted tumor associated antigens (TAAs) are identified, both the active and the passive approaches have been improved [163]. It is now feasible to generate and expand enough tumor specific CD8 + T cells in vitro or in vivo to kill tumor cells in patients [162, 164]. Though complete and partial remission has been reported in cancer patients receiving T cell-based immunotherapy in clinical trials, relapse is not uncommon [165, 166]. Factors that limit the effectiveness of anti-tumor immune response include, among others, lack of help from CD4 + T cells, active tolerization of T cells due to absence of co- 19

30 stimulatory molecules on the tumor cells, apoptosis of T cells induced by interacting with tumor targets, loss of surface MHC class I expression or loss of tumor antigen expression in tumor cells [7]. In patients, relapse due to loss of class I expression after effective immunotherapy has been observed [167]. Apparently, lack of class I HLA on the surface makes tumor cells invisible to the tumor specific CD8 + T lymphocytes. The number of HLA molecules present on the surface of tumor cells has been shown to affect their lysis by CTL quantitatively in situations with borderline amounts of peptide and/or MHC [168]. As class I down-regulation is frequently encountered in a variety of cancers, it has become one of the major obstacles to the T cell based tumor immunotherapy [169]. In some cases, the decreased surface expression of MHC class I complex can be restored by treatment with cytokines, such as IFN-γ [ ]. For those that do not respond to the cytokine treatment, elucidation of the underlying mechanisms for loss of class I expression in cancer cells may provide information for developing effective diagnostic or therapeutic methods to improve cancer immunotherapy. However, unlike viral infection, the causes of MHC class I down-regulation in cancer can vary greatly due to the genetic instability and rapid growth of cancer cells. Reports on causes of MHC class I downregulation in tumor cells have been diverse, which are summarized here. As described above, antigen presentation by MHC class I molecules requires coordinated expression of class I HC, β2m, proteasome and its inducible subunits (MECL, LMP-2/7), the ER aminopeptidases, the TAP transporter, tapasin and other ER chaperone proteins. The deficiency of any of these molecules may affect the class I- 20

31 restricted antigen processing and presentation quantitatively or qualitatively. In some cases, these factors are coordinated down-regulated [171, 172, 174, 175]. However, selective down-regulation of individual components of the class I antigen presentation machinery is more often reported. Lack of functional β2m causes total loss of class I expression as the class I HC is unstable without its association, which, though rare, has been found in metastatic melanomas and lung cancers [167, 176]. Defects in the expression of tapasin, LMP-2/7 or TAP-1/2 were found to cause down-regulation of total level of class I HLA on the surface of human cancer cells [173, ]. Decreased PA28 expression in tumor cells leads to impaired presentation of a human TRP2 tumor antigen [29]. Some of the underlying molecular mechanisms have been identified, though limited, for the class I antigen presentation machinery defects. Loss of heterozygosity (LOH) is an important mechanism that leads to loss of gene expression in cancer cells. It contributes to the defective MHC class I antigen presentation as well. LOH occurs at frequencies of 25%, 29% and 28%, respectively, in the MHC region on chromosome 6p, β2m gene on chromosome 15q and the putative HLA class I modifier of methylation gene (MEMO)-1 on chromosome 1p in breast cancer patients [183]. LOH causes β2m inactivation and thus total loss of MHC class I in microsatelite instability (MSI)-positive colorectal cancer cells [184]. HLA deficiency has been attributed to LOH in MHC class I region in melanoma and laryngeal carcinomas [ ]. The allelic imbalances on 21

32 chromosome 6p has been frequently observed in primary mediastinal B-cell lymphoma (PMBL) [189]. In one case, loss of HLA-A2 melanoma cell surface expression is associated with a complex rearrangement of the short arm of chromosome 6 [190]. Transcriptional dysregulation of MHC class I gene is another frequent cause of MHC class I down-regulation in cancer. The human CCAAT displacement protein CDP/cut has been shown as a locus-specific repressor of MHC class I gene transcription in human tumor cells [191]. Lack of transcription factor binding, such as NFκB, to the class I enhancer sequence has been observed in tumor cells [192] and in tumorigenic adenovirus transformed cells [193]. In murine insulinoma model, the decreased expression of MHC class I antigen was found controlled by the R1 element of the H-2 enhancer [194]. Reduced co-expression of the c-fos and c-jun proto-oncogenes has been correlated with the down regulation of class I MHC in high-metastatic cell lines derived from the Lewis lung carcinoma, B16 melanoma and the K1735 melanoma [195]. Overexpression of certain oncogenes, such as n-myc and c-myc, has been shown to downregulate MHC class I expression on the transcription level [196, 197]. Class I MHC expression can also be suppressed by cytokine IL-10, which act through NKκB inhibition in the short term and may repress TAP-1/2 transcription in the long term [ ]. Hypermethylation is another cause of transcription inactivation, which has been found in the class I MHC gene region in both human [201, 202] and murine tumor cells [203, 204]. 22

33 Post-transcriptional defects also contribute to loss of class I expression in tumor cells, reports about which, however, have been limited. Aberrant pre-mrna splicing causes selective HLA-A2 loss and β2m deficiency in melanoma cells [205, 206]. Premature stop codons resulting from nonsense mutation or frame-shifting insertion or deletion mutation have been shown to cause allelic loss of MHC class I expression in cervical cancer [207]. Due to its essential role in the MHC class I antigen processing and presentation, deficiency of TAP may cause loss of MHC class I surface expression in tumor cells [208]. In fact, loss of TAP expression has been frequently observed in a variety of cancers [208]. In melanoma and cervical cancer, TAP-1 expression in primary lesions may represent an independent prognostic marker [209, 210]. Presentation of specific tumor associated antigenic epitope has been found limited by the level of TAP expression in tumor cells [211]. In mouse model, TAP-deficient tumor cells show higher tumorigenicity compared to the wild type cells in the immune competent host but not in the athymic nude mice [212]. However, the underlying molecular mechanisms of TAP deficiency in tumor cells have been limited. In a renal cell carcinoma cell line, defect in the earliest steps of the IFN-γ signaling pathway causes loss of IFN-γ inducibility of TAP-1 and LMP-2 genes [213]. Expression of a defective TAP-1 protein variant (R659Q) results in loss of MHC class I surface expression in a human small cell lung cancer (SCLC) cell line [214]. 23

34 In search for new mechanisms that contribute to the down-regulation of MHC class I expression in human cancers, we started this work by analyzing class I-deficient tumor cell lines. A melanoma cell line, SK-MEL-19, which had been shown lack surface expression of MHC class I, was found deficient in TAP-1 mrna expression even after IFN-γ treatment. A single nucleotide deletion (D1489) was then identified in the exon 7, which causes accelerated degradation of TAP-1 mrna. To our knowledge, this is the first example of down-regulation of MHC class I antigen presentation machinery by decreasing mrna stability. Considering that although deficiency in TAP expression is a common observation in human cancers, TAP transporter activity can be disrupted without being downregulated on the expression level., we then looked for mutations that abrogate TAP protein function and may contribute to MHC class I down-regultion in human cancers. Since the signature motif has been known as a mutation hot spot in the CFTR protein of CF patients, we introduced the analogous CF-mutations into TAP-1 protein and characterized the function of these mutants in the SK-MEL-19 cells. Though all the CF-mutations abrogate TAP function, they are not present in the 228 human cancer samples we screened, 66 of which have down-regulated MHC class I expression. However, a rare allele of TAP-1, R648Q (G1943A in DNA sequence), which harbors a base pair replacement just C-terminal to the signature motif, occurs at a significantly higher frequency in the class I MHC - colon cancer samples than those with normal class I expression level. The functional analysis showed that the Q648 TAP-1 has only about 20% transporter activity of the R648 TAP-1. 24

35 Above all, we identified two mechanisms in this work that have not been addressed before, which down-regulate the TAP-1 mrna level and disrupt TAP-1 protein function, respectively, and as a result cause the loss of MHC class I expression in human cancer cells. 25

36 Figure 1.1 Conventional MHC class I antigen processing and presentation pathway. The antigenic peptides presented by class I MHC molecules are mostly derived from the endogenous cellular proteins or viral proteins, which are degraded by the cytosolic proteasome. TAP transporters on the ER membrane are responsible for transporting the antigenic peptides from cytosol to the ER lumen, where they associate with the empty MHC class I HC-β2m heterodimers with the help of the peptide-loading complex that consists of tapasin, calreticulin and ERp57. 26

37 ~ ~ Golgi Cellular or viral protein Proteasome ~ ~ ~ peptides ~ ~ ~ ER Nucleus MHC class I HC β2m tapasin calnexin ERp57 ER aminopeptidase calreticulin TAP1/2 27

38 MHC Class II MHC Class I DP DN DO DQ2 DQ1 DR B C LMP2 TAP1 LMP7 TAP2 DO DQ ICS-1/2 GAS kb SP Figure 1.2 Genomic localization and promoter structure of human TAP-1 gene. The genes that encode for TAP-1/2 and LMP-2/7 are clustered in the MHC class II locus on the short arm of human chromosome 6. TAP-1 and LMP-2 share a bi-directional promoter of 593 nucleotides, which harbors cis elements including the IFN consensus sequence (ICS-1/2), a gamma activated sequence (GAS), an NFκB binding site and an Sp1 site. The four transcription start sites for TAP-1 gene are shown as arrowheads with position labeled accordingly (the first ATG codon for TAP-1 as +1). 28

39 Walker A (529) TAP1(493) LLTPLHLEGLVQFQDVSFAYPNRPDVLVLQGLTFTLRPGEVTALVGPNGSGKSTVAALLQNLYQPTGGQLLLDGKPLPQ TAP2(458) TLAPTTLQGVVKFQDVSFAYPNRPDRPVLKGLTFTLRPGEVTALVGPNGSGKSTVAALLQNLYQPTGGQVLLDEKPISQ CFTR-N(413) KQNNNNRKTSNGDDSLFFSNFSLLGTPVLKDINFKIERGQLLAVAGSTGAGKTSLLMMIMGELEPSEGKIKHSGR---- CFTR-C(500) KDDIWPSGGQMTVKDLTAKYTEG-GNAILENISFSISPGQRVGLLGRTGSGKSTLLSAFLRLLN-TEGEIQIDGVSWDS MDR1-N(382) GHKPDNIKGNLEFRNVHFSYPSRKEVKILKGLNLKVQSGQTVALVGNSGCGKSTTVQLMQRLYDPTEGMVSVDGQDIRT MDR1-C(365) GLMPNTLEGNVTFGEVVFNYPTRPDIPVLQGLSLEVKKGQTLALVGSSGCGKSTVVQLLERFYDPLAGKVLLDGKEIKR Consensus(529) L P L GNV F DVSFAYPSRPDVPVLKGLSFTIRPGQVLALVG SGSGKSTVLALLQRLYQPTEG VLLDGK I Q loop C loop/signature motif (608) TAP1(572) YEHRYLHRQVAAVGQEPQVFGRSLQENIAYGLTQK-PTMEEITAAAVKSGAHSFISGLPQGYDTEVDEAGSQLSGGQRQ TAP2(537) YEHCYLHSQVVSVGQEPVLFSGSVRNNIAYGLQS--CEDDKVMAAAQAAHADDFIQEMEHGIYTDVGEKGSQLAAGQKQ CFTR-N(488) ISFCS-QFSWIMPGTIKENIIFGVSYD---EYRYRSVIKACQLEEDISKFAEKDNIVLGEGGITLSGGQRA CFTR-C(577) ITLQQWRKAFGVIPQKVFIFSGTFRKNLDPYEQWS---DQEIWKVADEVGLRSVIEQFPGKLDFVLVDGGCVLSHGHKQ MDR1-N(461) INVRFLREIIGVVSQEPVLFATTIAENIRYGREN--VTMDEIEKAVKEANAYDFIMKLPHKFDTLVGERGAQLSGGQKQ MDR1-C(444) LNVQWLRAHLGIVSQEPILFDCSIAENIAYGDNSRVVSQEEIVRAAKEANIHAFIESLPNKYSTKVGDKGTQLSGGQKQ Consensus(608) I V YLR VGVV QEPVLFSGSIRENIAYGLQ TDDEI KAAKEANAHSFI LP KYDTVVGEKGSQLSGGQKQ Walker B D loop Switch (687) TAP1(650) AVALARALIRKPCVLILDDATSALDANSQLQVEQLLYESPERYSRSVLLITQHLSLVEQADHILFLEGGAIREGGTHQQ TAP2(614) RLAIARALVRDPRVLILDEATSALDVQCEQAKTLWKFMIF CFTR-N(555) RISLARAVYKDADLYLLDSPFGYLDVLTEKEIFESCVCKLMA-NKTRILVTSKMEHLKKADKILILHEGSSYFYGTFSE CFTR-C(653) LMCLARSVLSKAKILLLDEPSAHLDPVTYQIIRRTLKQAFA--DCTVILCEHRIEAMLECQQFLVIEENKVRQYDSIQK MDR1-N(538) RIAIARALVRNPKILLLDEATSALDTESEAVVQVALDKARK--GRTTIVIAHRLSTVRNADVIAGFDDGVIVEKGNHDE MDR1-C(523) RIAIARALVRQPHILLLDEATSALDTESEKVVQEALDKARE--GRTCIVIAHRLSTIQNADLIVVFQNGRVKEHGTHQQ Consensus(687) RIAIARALVR PKILLLDEATSALD SE VV AL A RT ILI HRLS V NAD ILVLEEG IREYGTHQ Figure 1.3 Alignment of the nucleotide-binding domains (NBDs) of ABC transporters. The NBDs of selective ABC transporters, including human CFTR, human MDR1, TAP-1/2 and bacterial HisP, are compared. The conserved sequence motifs, Walker A, Walker B, C loop/signature motif, D loop and switch region, are marked with rectangles. 29

40 CHAPTER 2 MATERIALS AND METHODS 2.1 Cell lines Human melanoma cell lines 1195, 1102, and SK-MEL-19 were cultured as described previously [215]. The breast cancer cell line SK-BR-3 was obtained from ATCC (HTB-30; ATCC, Manassas, VA). All cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% heat-inactivated fetal calf serum, 2 mm L-glutamine, 100IU/ml penicillin, and 100µg/ml streptomycin (Invitrogen Inc. Carlsbad, CA). For induction of HLA class I expression, cells were cultured in medium supplemented with recombinant human IFN-γ (R&D Systems, Minneapolis, MN) at 1000 units/ml. 2.2 Antibodies. The following antibodies were used: R.RING4C, a rabbit anti-peptide antibody to the C-terminus of TAP-1 (a gift from Dr. Peter Cresswell); HC10, a mab recognizing free class I heavy chain; W6/32, a β2m-dependent anti-class I heavy chain mab; R.gp48N, a rabbit anti-peptide Ab to the N-terminal regions of tapasin (a gift from Dr. 30

41 Peter Cresswell); a rabbit anti-human TAP-1 peptide ( ) Ab (Calbiochem- Novabiochem corporation); an anti-human calnexin mab (Affinity Bioreagents, Golden, CO); an anti-human TAP-2 mab (BD PharMingen, San Diego, CA). 2.3 Flow cytometry Cell surface HLA class I expression was examined by flow cytometry as described previously [174]. Briefly, viable cells were incubated with PE-conjugated mouse IgG1 (BD Pharmingen), as isotype control, and PE-conjugated anti-hla-a, -B, and -C antibody (BD Pharmingen), respectively, at 4 C for 2 h. After three washes with phosphate-buffered saline (PBS) containing 1% fetal calf serum, cells were fixed with 1% paraformaldehyde and examined by flow cytometry. 2.4 Northern blot Cells were either treated with IFN-γ (R&D Systems) at 1000 units/ml for 48 h or left untreated. For cycloheximide (CHX; Sigma-Aldrich Corp. St. Louis, MO) treatment, SK-MEL-19 cells were cultured with IFN-γ at 1000 units/ml for 48 h, and then CHX was added to the cells for a final concentration of either 5 or 10 µg/ml, respectively, for up to 16 h. Total RNA was isolated using TRIzol reagent (Invitrogen). Hybridization conditions followed the instructions of the Northern hybridization kit (Eppendorf Scientific, Westbury, NY). The cdna probes for TAP-1, TAP-2, LMP-2, LMP-7, HLA class I heavy chain, and β 2 M were made from PCR products using primers listed previously [161]. A human splenocyte cdna library from Invitrogen was used as the template for PCR 31

42 reactions. All PCR products were subcloned into pbluescript vector (Stratagene, La Jolla, CA) and sequenced and confirmed to be identical to published sequences. The probes were labeled with [α- 32 P]dCTP (PerkinElmer Life Sciences, Boston MA) using the DECAprime TM II kit (Ambion, Austin, TX). 2.5 Southern blot Genomic DNA was isolated from SK-MEL-19 cells, SK-BR-3 cells, and HeLa cells. Genomic DNA (20 µg) was digested with Afl III (Invitrogen) and separated in 0.8% agarose gel. The TAP-1 promoter probe was made by PCR from normal human lymphocyte genomic DNA with sense primer 5'- TCCCGCCTCGAGCATCCCTGCAAGGCA-3' and antisense primer 5'- TGCAGTAGCCTGGTGCTATCCG-3'. Probes were labeled with [α- 32 P]dCTP (PerkinElmer Life Sciences) using the DECAprime TM II kit (Ambion). 2.6 Generation of TAP-1 cdna constructs and stable transfection The human small cell carcinoma H146 cell line (provided by Dr. N. P. Restifo; National Cancer Institute, Bethesda, MD) was incubated with IFN-γ at 1000 units/ml for 48 h. Total RNA was isolated as described above. Reverse transcription was done using the SUPERSCRIPT First-Strand cdna Synthesis System (Invitrogen). TAP-1 cdna was amplified by PCR in three fragments. Primers were as follows: htap.f1, 5'- GCGGCCGCTTTCGATTTCGCTTTC-3'; htap.r1, 5'- TGCAGTAGCCTGGTGCTATCCG-3'; htap.f2, 5'- 32

43 CTTGCCTTGTTCCGAGAGCTGA-3'; htap.r2, 5'-CTCGTTGGCAAAGCTTCGAAC- 3'; htap.f3, 5'-CGGCCATGCCTACAGTTCGAAG-3'; and htap.r3, 5'- ATAAATATCAAGAACCTACAGGG-3'. The three fragments were cloned into pbs-ks vector (Stratagene) at NotI/SmaI, SmaI/HindIII, and HindIII/XhoI sites, respectively, and sequenced to confirm that the cdna has a wild type sequence. When SK-MEL-19 cells grew to 70% confluence, 0.2 µg of pcdna3.1/hyg(+) vector (Invitrogen) and pcdna3.1/hyg(+) vector with TAP-1 cdna insert were respectively transfected into each well of a 24-well plate, using 6 µl of FuGENE 6 transfection reagent (Roche Diagnostics Corporation, Indianapolis, IN) according to the manual. 48 h later, the transfected cells were replated into 96-well plates and cultured in the presence of 0.5mg/ml hygromycin B (Invitrogen). Single cell clones were selected for further culture and analyzed for HLA class I antigen expression. Constructs for all the TAP-1 signature motif mutants and TAP-1 R648Q were made by overlapping PCR and expressed in the pcdna3.1/hyg(+) vector (Invitrogen). Nucleotide changes and amino acid changes are summarized in Table 2.1. To ensure that any functional defects of the TAP-1 mutants were due to the mutations we generated, all the constructs were sequenced before further analyses. Stable cell clones were established as described above. Expression of TAP-1 protein was examined by Western blotting using a rabbit anti-human TAP-1 peptide ( ) Ab (Calbiochem, EMD Biosciences Inc., San Diego, CA). Cell clones expressing TAP-1 protein were selected for flow cytometric analysis of cell surface MHC class I. 33

44 2.7 Generation of TAP-1 constructs and Tet-Off SK-MEL-19 cell lines for RNA stability analyses TAP-1 D1489 was generated by PCR using the total cdna from the SK-MEL-19 cells as the template. Site-directed mutagenesis by overlapping PCR was performed to make TAP-1 Del3 cdna. After confirming their sequences, all three TAP-1 cdnas were inserted into the multiple cloning site of the pbi-egfp vector (BD Bioscience, Clontech, Palo Alto, CA) by blunt-end ligation. The three constructs, pbi-egfp-tap-1, ptet-off (Clontech), and pcdna3.1/hyg(+) (Invitrogen), were co-transfected into SK-MEL-19 cells using FuGENE 6 transfection reagent (Roche). Stably transfected cell clones were selected in 96-well plates in the ordinary DMEM culture medium supplemented with 0.5mg/ml hygromycin (Invitrogen). To confirm the efficiency of the Tet-Off construct in the tumor cell line, the green fluorescence protein-positive cell clones were treated with 1 µg/ml tetracycline (Roche) for 24 h. The total RNA was extracted from the cell clones with or without tetracycline treatment. The cell clones in which TAP-1 mrna expression was inhibited by at least 95% by tetracycline were used for the study of mrna stability. 2.8 Generation of luciferase reporter constructs and assay for promoter activity The TAP-1 promoter was amplified from SK-MEL-19 cell genomic DNA by PCR using the following primers: htap-1.pr1, 5'- GCTCTAGATGGCACTCGGACGCCGTC-3'; and hlmp2.pf1, 5'- GCTCTAGACCCTGCAAGGCACCGCTC-3'. The PCR products were subcloned using the Zero Blunt TOPO PCR cloning kit (Invitrogen) and then cloned into pgl2-basic 34

45 vector (Promega, Madison, WI) at XhoI and HindIII sites. All constructs were confirmed by DNA sequencing. Expression level of the firefly luciferase from the pgl2 constructs (basic, SV40, ptap-1/t, and ptap-1/g) was normalized to the internal control prl- SV40 Renilla luciferase level. Results were shown as the fold increase compared with the pgl2-basic. The dual luciferase assay was carried out according to the manufacturer's instructions (Promega). 2.9 Nuclear run-on assay The assay was performed as described elsewhere [216]. Briefly, nuclei were extracted from 10 7 to 10 8 SK-MEL-19 cells treated with or without 1000 units/ml IFN-γ. The transcripts were labeled in vitro with 40 nm biotin-16-utp (Roche) in the presence of 3.75 mm ATP, GTP, and CTP; 25 mm Tris-HCl; 12.5 mm MgCl 2 ; and 750 mm KCl. cdna fragments of LMP-2 and GAPDH were amplified by PCR from cloned cdna constructs [161]. TAP-1 cdna fragment was amplified by PCR from cloned cdna constructs using primers htap-1.f1 and htap-1.r1 described above. The pcdna3.1/hyg(+) vector was linearized with HindIII. All the DNA was immobilized on nitrocellulose membrane using S&S Minifold II slot blot apparatus according to the manual (Schleicher & Schüll Bioscience Inc., Keene NH). Hybridization conditions were as described previously [216], and the biotin-labeled transcripts were detected using streptavidin-alkaline phosphatase conjugate (Roche) and CDP-star Ready-To-Use with Nitro-Block-II reagent (Tropix, Bedford, MA). 35

46 2.10 Restriction fragment length polymorphism (RFLP) To detect the D1489 mutation in TAP-1 exon 7, primers htap-1ce7.f (5'- GCACCCCTCGCTGCCTACCCAGTGGTCT-3') and htap-1ce7.r (5'- TACAGGGAGTGGTAGGTTGTACCTG-3') were used to amplify from genomic DNA the region of TAP-1 exon 7 where the single-nucleotide deletion resides. The region was also amplified from cdna using primers htap-1ce7.f and htap-1ce7.r PCR products were separated by gel electrophoresis and purified using Qiagen gel extraction kit (Qiagen Inc., Valencia, CA). The purified PCR products were incubated with BslI (New England Biolabs Inc., Beverly, MA) at 55 C overnight and then separated in 5% agarose gel. To screen the G1943A polymorphism in normal population, a forward primer, E10 Ahd.f 5 -ACCGTTCTCATCTTGGCCCTTTGCTCTG-3, was designed to generate an Ahd I site in the G allele but not in the A allele. Together, with the reverse primer, E10 Ahd.r 5 - ATCAATGCTCGGGCCAACGCGACTGCCT-3, TAP-1 exon 10 was amplified from the genomic DNA. PCR products were then separated by gel electrophoresis and purified using QIAGEN gel extraction kit (Qiagen). The purified PCR products were incubated with Ahd I (New England Biolabs) at 37 o C overnight and then separated in 5% agarose gel RNase protection assay The TAP-1 cdna construct used to make the TAP-1 antisense transcript was the same as the one used for the Northern blot. The GAPDH cdna fragment was generated by PCR using the primers hgapdh.f (5'-TGAGAACGGGAAGCTTGTCATCAA-3') 36

47 and hgapdh.r (5'-CAGCCTTCTAGATGGTGGTGAAGA-3'). The EGFP cdna fragment was also generated by PCR using the primers EGFP.f (5'- TCCAGCAGGATCCTGTGATCGCGCT-3') and EGFP.r (5'- ACCTACGGCCTCGAGTGCTTCAGCC-3'). The antisense probes were made using the Riboprobe in vitro transcription system (Promega). The RNase protection assay was conducted with the RPA III ribonuclease protection assay kit (Ambion) according to the instruction manual. After separation of protected fragments on a 6% sequencing gel, signals were quantified by phosphorimaging (Amersham Biosciences Inc., Piscataway, NJ). The TAP-1 signal intensity was normalized by the signal intensity of GAPDH, which served as a loading control. The percentages of the amount of remaining TAP-1 mrna at different time points after tetracycline was added compared with time 0 were calculated Immunofluorescence microscopy Cells were plated on coverslips in a 24-well plate and were allowed to grow overnight. Cells were fixed for 15 min in a freshly prepared solution of 4% formaldehyde in PBS, ph 7.4, and then permeabilized in washing buffer (PBS with 0.2% Triton X-100) for 5 minutes at room temperature. After three washes, the primary antibodies were added, rabbit anti-human TAP-1 (Calbiochem) and mouse anti-human calnexin (Affinity Bioreagents, Golden, CO) or anti-human TAP-2 mab (BD Pharmingen), which were diluted in 200 µl of dilution buffer (PBS containing 3% bovine serum and 0.2% Triton X- 100). After incubation for 1 hour at room temperature, coverslips were washed three times in washing buffer and then incubated with the secondary antibodies, fluorescein- 37

48 conjugated donkey anti-rabbit Ig (Amersham) for TAP-1 and Texas red dye-linked goat anti-mouse Ig (Jackson ImmunoResearch Laboratories Inc., West Grove, PA) for calnexin or TAP-2, for 1 hour at room temperature. After three washes, the coverslips were dipped in water and then mounted on glass slides in the mounting media with DAPI (Vector Laboratories, Burlingame, CA) Immunoprecipitation Immunoprecipitation procedures described [42] were followed. Cells were extracted for 30 min on ice at cells per ml in 1% digitonin (Sigma) in TBS (10 mm Tris, 150 mm NaCl [ph 7.4]) containing 0.5 mm phenylmethyl sulforyl fluoride (Sigma), 0.1 mm N-tosyl-L-lysine chloromethylketone (Sigma), and 5 mm iodoacetamide (Sigma). Postnuclear supernatant was precleared overnight at 4 C with 5 µl normal rabbit serum (Invitrogen) and 50 µl Zysorbin (Zymed Laboratories, Inc., San Francisco, CA) per ml of extract. Aliquots were then incubated with the appropriate antibodies, R.RING4C, a rabbit anti-peptide antibody to the C-terminus of TAP-1 (a gift from Dr. Peter Cresswell), W6/32, a mab detecting the β2m-associated class I heavy chain and an anti-human TAP-2 mab (BD PharMingen), and protein A Sepharose for 1 hr each at 4 C. Protein G Sepharose (Amersham) was used with anti-tap-2 mab and W6/32. The immunoprecipitates were washed twice with TBS, 0.1% digitonin and once with distilled water before being separated by SDS PAGE. TAP-1, TAP-2 and tapasin were detected by Western blotting using the rabbit anti-human TAP-1 peptide ( ) 38

49 Ab (Calbiochem), the anti-human TAP-2 mab (BD PharMingen), and R.gp48N, a polyclonal rabbit Ab against the N-terminal regions of tapasin (a gift from Dr. Peter Cresswell). The same antibodies were used for the following Western blotting analyses Microsome-based peptide translocation assay Microsomes were prepared as described previously [108]. Briefly, SK- MEL-19 cell clones that were stably transfected with wild-type TAP-1, TAP-1 G646D, TAP-1 R648Q and pcdna3.1/hyg vector, respectively, were harvested. Cells were washed once in ice-cold PBS, and then resuspended in 800 µl of ice-cold cavitation buffer (250 mm sucrose, 25 mm potassium acetate, 5 mm magnesium acetate, 0.5 mm calcium acetate, 50 mm Tris, ph 7.4) supplemented with a protease inhibitor mixture (Sigma). Cells were lysed by repeatedly drawing the suspension through a 26-gauge needle at least 10 times. After centrifugation at ~500 g for 5 min, the supernatant was thoroughly mixed with 5 ml of 2.5 M sucrose in gradient buffer (2.5 M sucrose, 150 mm potassium acetate, 5 mm magnesium acetate, 50 mm Tris, ph 7.4), which was then overlaid with 3 ml each of 2.0 M sucrose in gradient buffer, then 1.3 M sucrose in gradient buffer, and finally with 800 µl of cavitation buffer. This was centrifuged overnight at 80,000 g, and the microsomal fraction at the interface of the 2.0 and 1.3 M sucrose layers was collected. This fraction was diluted into 5 ml of PBS, 1 mm dithiothreitol, and centrifuged for 1 h at 100,000 g. The pellet was resuspended in 39

50 100µl of PBS, 1 mm dithiothreitol and frozen in aliquots at -70 C. The total protein content was determined by a BCA assay (Pierce Biotechnology Inc., Rockford, IL). The expressions of TAP-1 and TAP-2 proteins were examined by Western blotting. Translocation assays were carried out as previously described [108]. An iodinated model peptide, RRYNASTEL, with a specific activity of µci/µg was used for the study. Briefly, 30 µg of microsomes containing wild-type TAP-1, TAP-1 G646D, TAP-1 R648Q or pcdna3.1/hyg vector were added to 150 µl of assay buffer (PBS, 0.1% bovine serum albumin, 1 mm dithiothreitol, ph 7.3) containing 10 mm MgCl 2 and 5 mm ATP (+ATP samples) and then incubated with the radioiodinated RRYNASTEL peptide at 37 C for 15 min. The samples were then centrifuged at 4 C at 8,800 g and washed once with 250 µl of the assay buffer. The pellets were subsequently resuspended in 250 µl of lysis buffer (50 mm Tris, 150 mm NaCl, 1% Nonidet P-40, ph 7.4) and incubated on ice for 1 h. After centrifugation at 8,800 g for 5 min at 4 C, the supernatants were transferred to ConA-Sepharose (Amersham) beads and incubated for 2 h at 4 C with agitation. Beads were washed twice with the assay buffer before the radioactivity was determined using a Beckman 5500 gamma-counter Colon cancer tissue sections, immunohistochemistry and statistical analysis Colon cancer tissue microarray slides were acquired from National Cancer Institute (NCI) and Zymed Laboratories Incorporation. All specimens contained in the array were formalin-fixed and paraffin-embedded. Mouse monoclonal antibody HC-10 that detects HLA-A,B,C heavy chain was used in immunohistochemical staining. Tissue 40

51 samples were graded as MHC class I negative when at least 75% of the tumor cells were not stained or weakly stained. The comparisons of the allelic frequencies were tested using two-tailed z-tests and an alpha level of The p values are reported Isolation of genomic DNA and genotyping The genomic DNA from blood of normal population was extracted as described elsewhere [217] and then subjected to the restriction fragment length polymorphism (RFLP) analysis as described above. Tumor tissues were collected either by laser capture microdissection (LCM) or manually using sterile tweezers. DNA was then isolated using a lysis buffer containing 1 high fidelity PCR buffer (Roche Diagnostics Corporation, Indianapolis, IN), 1% tween-20 and 4mg/ml proteinase K and incubated at 55 o C for 72 hours. Nested PCR was designed using the following primers to amplify the exon 10 region of TAP-1 gene: oue10.f 5 -GTTCTCATCTTGGCCCTTTGCTCTG-3 and ine10.r1: 5 - AGAAGATGACTGCCTCACCTGTAAC-3 for the first-round amplification; E10.F 5 -CCTTTGCTCTGCAGAGGTAGACGAG-3 and ine10.r2 5 - TGCCTCACCTGTAACTGGCTGTTTG -3 for the second round. The final PCR products were purified with QIAGEN PCR purification kit (Qiagen) before sequencing. 41

52 CHAPTER 3 A SINGLE-NUCLEOTIDE DELETION LEADS TO RAPID DEGRADATION OF TAP-1 MRNA IN A MELANOMA CELL LINE 3.1 Introduction. Recent studies demonstrate that patients with malignant melanoma often have a high number of cytotoxic T lymphocytes specific for melanoma-associated antigens [218, 219]. The co-existence of T cells and tumor cells even in the draining lymph nodes suggests that the tumors were able to evade destruction by host cytolytic T lymphocytes. Accumulating evidence supports the notion that both malfunction of T cells and downregulation of antigen presentation machinery in tumors can be responsible for tumor evasion of host immunity [148, 161, 174, ]. In fact, a high proportion of malignant tumors, including melanoma, have severely depressed cell surface expression of class I HLA antigens [221], the target molecules that present tumor antigenic peptide to cytolytic T lymphocytes. Understanding the mechanisms underlying the T-cell malfunction or antigen presentation defects may thus provide insight for immunotherapy of melanoma and other cancers. 42

53 Optimal cell surface expression of HLA molecules requires the coordinated expression of several genes, such as transporters associated with antigen processing (TAP)-1/2, low molecular weight peptide (LMP)-2/7, and tapasin, as well as HLA class I heavy chain and β 2 -microglobulin (β 2 M). In cases of both tumorigenesis and viral infection, expression of these genes and the function of the encoded proteins are often impaired [ ]. The mechanisms for such down-regulation have been studied extensively. Theoretically, gene expression can be modulated by transcriptional, posttranscriptional, translational, and posttranslational mechanisms. The mechanisms that have been shown to underlie the antigen presentation abnormalities are transcriptional suppression of antigen presentation genes and/or functional inactivation of their gene products, either by missense mutation or by protein-protein interactions [200, 213, 214, 225]. Here we show that actively transcribed TAP-1 mrna in the melanoma cell line SK-MEL-19 is rapidly degraded even after stimulation with IFN-γ. Cloning and sequencing analysis have revealed a single-nucleotide deletion at position This mutation results in substantial reduction of the stability of TAP-1 mrna by mechanisms unrelated to nonsense-mediated mrna decay (NMD). These results reveal a new potential mechanism for tumor evasion of host T-cell recognition. 43

54 3.2 Results and discussion Down-regulation of TAP-1 mrna by a posttranscriptional mechanism in melanoma cell line SK-MEL-19 Three human melanoma cell lines (1102, 1195, and SK-MEL-19) were examined by flow cytometry for their cell surface HLA class I expression with or without IFN-γ stimulation. A PE-conjugated anti-human HLA-A, -B, and -C antibody was used to detect all HLA class I alleles, and a PE-conjugated mouse IgG1 was used as isotype control. As shown in Figure 3.1a, 1102 and 1195 cells had significant HLA class I that was further up-regulated by incubation with 1000 units/ml IFN-γ for 3 days. Confirming previous studies[215], we found that SK-MEL-19 cells had no cell surface HLA. Surprisingly, whereas other melanoma cell lines up-regulated their cell surface HLA in response to IFN-γ, very little HLA class I antigen could be found on the SK-MEL-19 cells even after IFN-γ-treatment. Because optimal cell surface HLA class I expression requires the coordinated expression of multiple genes, including TAP-1/2, LMP-2/7, and β 2 M as well as HLA class I heavy chain, a Northern blot analysis was performed to detect the expression of these genes (Figure 3.1b). In 1102 and 1195 cells, all six genes were expressed at low but detectable levels. IFN-γ treatment drastically induced expression of all six genes. Interestingly, in the SK-MEL-19 cells, whereas β 2 M, HLA heavy chain, LMP-2, LMP-7, 44

55 and TAP-2 were present at low levels without induction, no TAP-1 mrna was detected. After IFN-γ treatment, β 2 M, HLA heavy chain, TAP-2, LMP-2, and LMP-7 were expressed at high levels, yet TAP-1 was still expressed at low levels. It has been known that TAP-deficient cells can express HLA class I after transfection with the TAP-1 or TAP-2 gene [79, 226, 227]. To test whether the lack of TAP-1 expression was responsible for the barely detectable expression of HLA class I antigen on the surface of SK-MEL-19 cells, we transfected the cells with TAP-1 cdna. As shown in Figure 3.1c, the TAP-1 cdna-transfected SK-MEL-19 cells expressed significant levels of HLA class I antigen even before IFN-γ-treatment. Moreover, the TAP-1 transfectants were as responsive to IFN-γ as the other melanoma cell lines. Based on these results, it is likely that the primary defect of antigen presentation in SK-MEL-19 cells is attributable to defects in TAP-1 expression. Because the TAP-1 expression was low at the mrna level, we hypothesized that the TAP-1 down-regulation was caused by defective transcription or malfunction in RNA metabolism. The TAP-1 expression is under the control of a bidirectional promoter, as characterized by Wright et al. [86]. We cloned and sequenced the 593-bp TAP-1 promoter from SK-MEL-19 cells. In comparison with the published sequence [86], a singlenucleotide G ->T replacement was identified at position -446 (the first ATG of the TAP-1 gene is designated as +1), which is close to the first transcription start site at -427 [86] (Figure 3.2a). Because the T allele results in a loss of restriction site AflIII, we did a Southern blot hybridization using AflIII to confirm the mismatch. As shown in Figure 3.2a, whereas the HeLa cell line contained homozygous G alleles as described previously 45

56 [86], both SK-MEL-19 and the breast cancer cell line SK-BR-3 were homozygous for T alleles that lack the restriction site for AflIII. To test whether this single-nucleotide replacement results in reduced promoter activity, both alleles of the TAP-1 promoter were cloned into the pgl2-basic vector that contains the luciferase reporter gene. As shown in Figure 3.2b, the T allele TAP-1 promoter retained 50% of the promoter activity compared with the G allele. However, given the significant variation in transient transfection and luciferase assays, it is unclear whether the G ->T change has a significant effect on TAP-1 transcription. However, both promoters were equally efficiently induced by IFN-γ treatment, whereas the TAP-1 mrna in the original SK-MEL-19 cell line was not induced by IFN-γ treatment (Figure 1.1b). Moreover, our analysis of normal human peripheral blood lymphocyte samples revealed that both alleles were present at a high frequency, and individuals that carry either G or T alleles have equivalent cell surface HLA class I antigen expression (data not shown). We therefore performed a nuclear runon assay to directly evaluate the transcription of the TAP-1 gene. LMP-2 transcription, which is under the control of the same bidirectional promoter, was also evaluated. As shown in Figure 3.2c, TAP-1 was transcribed at high levels in SK-MEL-19 cells under basal conditions, although IFN-γ appeared to up-regulate TAP-1 transcription somewhat. In contrast, LMP-2 was transcribed at an undetectable level but was induced to high levels by IFN-γ (Figure 3.2c). The lack of LMP-2 transcription at basal condition may reflect the IFN-γ-inducible expression pattern of this gene. These results demonstrate that lack of TAP-1 mrna in SK-MEL-19 cells was not due to defective transcription. Taken together, the results demonstrate that a posttranscriptional defect is responsible for poor 46

57 TAP-1 expression in SK-MEL-19 cells even after IFN-γ stimulation. Numerous studies have revealed defective TAP-1 expression among tumor cells [214, 221]. To our knowledge, however, this is the first example of a posttranscriptional defect of TAP-1 expression A single-nucleotide deletion leads to accelerated decay of TAP-1 mrna A major mechanism responsible for posttranscriptional regulation of mrna is RNA degradation, which can be prevented by CHX, a protein synthesis inhibitor of mammalian cells. It is well established that the turnover of mrna is closely linked to the translation process and that blocking translation can stabilize mrna, especially those mrna with short half-lives [ ]. To test whether accelerated RNA degradation is responsible for the lack of TAP-1 in SK-MEL-19 cells, we treated the SK-MEL-19 and control HeLa cells with CHX after incubation with or without IFN-γ (1000 units/ml) for 48 h at 37 C. At different time points after the CHX was added to the cell culture, cells were harvested, and the total cellular RNA was analyzed for TAP-1 mrna. The intensity of each band was quantified using ImageQuant 5.0 software (Amersham Biosciences) after exposure to a phosphorimaging screen (Figure 3.3). For a better comparison, TAP-1 mrna levels were normalized to the endogenous housekeeping gene GAPDH level, and the fold increase compared with non-chx-treated cells was calculated. Under basal conditions, the TAP-1 mrna was up-regulated 4.8-fold in SK-MEL-19 cells. After IFN-γ induction, the TAP-1 mrna was up-regulated 25.7-fold. In comparison, CHX caused a less significant increase of TAP-1 mrna in both IFN-γ-treated and untreated HeLa cells. 47

58 Taken together, the lack of TAP-1 mrna, the normal transcription of TAP-1, and the rescue of TAP-1 mrna by CHX treatment suggest that the TAP-1 mrna was rapidly degraded in the SK-MEL-19 cells. It is noteworthy that in SK-MEL-19 cells, the effect of CHX was significantly stronger when used in combination with IFN-γ. This finding cannot be fully explained by the fact the IFN-γ is a transcriptional activator for antigen presentation genes, because its effect on TAP-1 transcription is not so obvious in SK- MEL-19 cells as shown in Figure 3.2c. It is likely that IFN-γ stabilized mrna in SK- MEL-19 cells, although this possibility remains to be tested formally. The rapid degradation of TAP-1 mrna can be due to a genetic lesion in the TAP- 1 gene. Alternatively, it is possible that the tumor cell line expresses factors that can cause TAP-1 mrna degradation. The successful rescue of cell surface HLA class I antigen expression by wild type TAP-1 in SK-MEL-19 cells favors the first hypothesis because a wild type cdna can be expressed in the tumor cell line. As the first step to test this hypothesis, we cloned TAP-1 cdna from SK-MEL-19 cells that were treated with both IFN-γ and CHX as described above. All of the three clones sequenced showed a single-nucleotide deletion at position (Figure 3.4a), which resides in exon 7 in the TAP-1 gene. Further analysis showed that multiple downstream premature termination codons (the closest one is at position +1555) were present due to this nucleotide deletion (two of them are shown in Figure 3.4c). To confirm that the mutation was in the TAP-1 gene, we amplified exon 7 of the TAP-1 gene from the SK-MEL-19 cells by PCR. The PCR products were digested with BslI because this restriction enzyme recognizes the deletion mutant but does not recognize the wild type exon 7. Because complete digestion 48

59 was obtained, it appears that the SK-MEL-19 cells are homozygous for the frameshift mutation (Figure 3.4b), even though the cytogenetic analysis revealed that there are four copies of chromosome 6 present in the SK-MEL-19 cells (data not shown). We subsequently amplified exon 7 of the TAP-1 gene from 50 normal human peripheral lymphocyte genomic DNA samples by PCR and subjected the PCR products to BstI digestion. Because none of the PCR products from the 50 samples could be digested by BslI, it is most likely that the single-nucleotide deletion in SK-MEL-19 cells resulted from a somatic mutation (data not shown) and that the apparent homozygosity of the TAP-1 locus caused aneuploidy. Premature termination codons resulting from frameshift mutation or nonsense mutation have been shown to interfere with the metabolism of many different mrnas in mammalian cells, leading to nonsense-mediated altered RNA splicing, such as exon skipping and intron retention and/or NMD [ ]. Alternatively, a mutation may disrupt a cis-element that is necessary for mrna stability and thereby cause RNA decay. To differentiate the two possible mechanisms, we designed another TAP-1 mutant (TAP- 1 Del3) that has an in-frame 3-nucleotide deletion at the same position as TAP-1 D1489 (Figure 3.5a) and compared the mrna half-lives of TAP-1 WT, TAP-1 D1489, and TAP-1 Del3. If the degradation of TAP-1 D1489 is via the NMD pathway, then the halflife of the TAP-1 Del3 message should be comparable with that of TAP-1 WT. Otherwise, if the message of TAP-1 Del3 is comparable with or even more unstable than that of the TAP-1 D1489, then it is the deletion itself, but not the resulting premature termination codons, that leads to the accelerated decay of TAP-1 mrna in SK-MEL-19 49

60 cells. The Tet-Off gene expression system was adapted (Figure 3.5a) to compare the stabilities of various TAP-1 mrnas. We first selected the SK-MEL-19 transfectants with an induction ratio (Tet 0/Tet 24 h) of 20 or greater to test mrna half-lives. This ensured that more than 95% of transcription was blocked by the tetracycline after 24 h. The selected cell clones were then treated with 1 µg/ml tetracycline for different lengths of time before the total RNA was harvested. The amounts of the mrna were quantified by phosphorimaging. In addition, the intensity of the TAP-1 signal was normalized to that of GAPDH before the percentage of remaining mrna was calculated. Four experiments were conducted, and an average percentage value was used to derive the mrna halflives (t 1/2 ) via regression analysis. As shown in Figure 3.5, b and c, with a t 1/2 of 7.2 h, the TAP-1 WT mrna is considerably more stable than that of the TAP-1 D1489 (t 1/2 = 3.5 h). However, the mrna derived from TAP-1 Del3 (t 1/2 = 2.7 h), which has an in-frame 3- nucleotide deletion, was degraded at least as fast as that of TAP-1 D1489. Because the TAP-1 Del3 mrna has no premature termination codon downstream to the deletion, the accelerated decay in mutant TAP-1 mrna is most likely through mechanisms other than NMD. It is more likely that the mutation disrupts a cis-element critical for the stability of TAP-1 mrna. Whereas few cis-elements that help to stabilize mrna have been reported, at least two have been reported by others [231, 232] and our group [233]. Preliminary analysis showed no similarity between the region surrounding the mutation and the previously reported cis-element. 50

61 Nevertheless, the NMD is a well-conserved cellular surveillance mechanism. Whereas our work revealed a non-nmd mechanism for TAP-1 mrna degradation, it is still possible that mrna derived from the endogenous mutant TAP-1 gene can also be degraded by NMD. It has been shown recently that several criteria have to be met for the pathway to degrade a premature termination codon-containing message. First, at least one downstream spliceable intron is required for optimal NMD [ ]. The intron is thought to help recruit NMD factors, such as hupf3, to the mrna via the spliceosome [ ]. Second, the premature termination codon should be at least nucleotides away from the next spliceable intron [234, 235]. The lack of introns in our constructs may have prevented us from revealing NMD in TAP-1 mrna degradation. However, intronless premature termination codon-containing HEXA mrna was shown to be subject to NMD, although at a lower efficiency than that seen when multiple downstream introns are present [240]. In preliminary studies, when we made pbi-egfp/tap-1 constructs with intron 7 or 8, the results also failed to support a role for NMD in degradation of the mutated TAP-1 mrna (data not shown). Posttranscriptional regulations of other genes involved in antigen presentation have been reported previously [241]. The increased turnover of HLA-C heavy chain mrna has been suggested to contribute to the low level of HLA-C surface expression [241]. Our work shows that mutations in the TAP-1 gene in a tumor cell line can modulate its mrna stability. This mechanism may be exploited by tumors to evade host immunity. 51

62 Figure 3.1 Deficiency of surface HLA class I expression in melanoma cell line SK- MEL-19 was due to the TAP-1 down-regulation. a, HLA class I expression in three melanoma cell lines, SK-MEL-19, 1102, and Bold black lines depict the staining by PE-conjugated anti-human HLA-A, -B, and -C antibody in untreated cells; dotted lines represent the staining by PE-conjugated mouse IgG1 as isotype control; and red lines represent anti-hla-a, -B, and -C antibody staining after stimulation with 1000 units/ml IFN-γ for 72 h. b, expression of HLA class I heavy chain (MHC I), β 2 M, TAP-1, TAP-2, LMP-2, and LMP-7 in each cell line with or without IFN-γ induction (1000 units/ml for 72 h). Total RNA loading to each well was shown as 28 S rrna (28S) and 18 S rrna (18S). c, transfection with wild type TAP-1, but not vector alone, restored HLA class I expression in the SK-MEL-19 cells. SK-MEL- 19 cells were transfected with either vector alone (top panels) or vector with TAP-1 cdna insert (bottom panels). These stable clones from each group were stimulated with or without IFN-γ and analyzed for cell surface HLA-A, -B, and -C, as detailed in a. 52

63 Figure 3.1 a SK-MEL Ctrl Cell numbers -IFNγ +IFNγ log Fluorescence (PE-anti-human HLA-A, B, C) b SK-MEL IFNγ β 2 M MHC I LMP-2 LMP-7 TAP-1 TAP-2 28S 18S 53 (continued)

64 Figure 3.1 (continued) c. Cell numbers F2 D9 E6 Ctrl -IFNγ +IFNγ 5G5 5C6 3A2 vector TAP-1 log Fluorescence (PE-anti-human HLA-A, B, C) 54

65 Figure 3.2 Posttranscriptional mechanisms are responsible for poor accumulation of TAP-1 mrna. a, a single-nucleotide polymorphism, adjacent to the first transcription start site (-427), was identified at -446 in the bidirectional promoter shared by the TAP-1 and LMP-2 genes. The G-> T change results in the loss of the AflIII restriction site. Southern blot hybridization was performed using AflIII and detected by a DNA probe that encompasses the downstream region of the polymorphism site. SK-MEL-19 cells showed one 5.6-kb band that represents homozygous T allele, as did the breast cancer cell line SK-BR-3, which has significant cell surface HLA class I surface expression (data not shown). HeLa cells, in contrast, are homozygous for the G allele. b, activities of T and G alleles of TAP- 1 promoter (ptap1/t and ptap1/g, respectively) in SK-MEL-19 cells. The two allelic forms of TAP-1 promoter were cloned into pgl2-basic vector (basic) that did not contain any promoter or enhancer but encoded firefly luciferase. The pgl2-sv40 construct (SV40) that had both SV40 promoter and SV40 enhancer as well as the firefly luciferase reporter gene was used as positive control. After transfection, IFN-γ was added to the cell culture at 1000 units/ml. Cells were lysed 48 h after transfection, and luciferase expression was tested using a luminometer. Data shown are representative of at least five independent experiments. c, the TAP-1 gene was actively transcribed in SK-MEL-19 cells in the presence and absence of IFN-γ in nuclear run-on assay. Endogenous GAPDH expression was used as a positive control, and the pcdna3.1/hyg(+) vector was used as a negative control. The run-on experiments were repeated three times with similar results. 55

66 Figure 3.2 a LMP-2 G/T TAP-1 AflIII 2.9kb AflIII 2.7kb AflIII probe SK-MEL-19 HeLa SK-BR-3 5.6kb 2.7kb (continued) 56

67 Figure 3.2 (continued) b. ptap1/g G LUC ptap1/t T LUC SV40 SV40 promoter LUC SV40 enhancer basic LUC IFN-r(-) IFN-r(+) Basic SV40 ptap/t ptap/g c. SK-MEL-19 -IFNγ +IFN γ LMP-2 TAP-1 pcdna3.1/hyg(+) GAPDH 57

68 Figure 3.3 TAP-1 mrna level in SK-MEL-19 cells was increased by CHX. The protein synthesis inhibitor CHX was added to the SK-MEL-19 cells and HeLa cells that had normal TAP-1 and HLA class I expression. Total RNA was isolated from both cells at different time points and subjected to Northern blot hybridization to detect TAP-1 expression. The blot was exposed to a PhosphorImager, and the signal intensity was quantified using ImageQuant 5.0 software (Amersham). After normalization of TAP-1 signal to endogenous GAPDH signal in each sample, the signals in the CHX-treated group were compared with those that received no CHX treatment. These signals were quantitated as fold of those in untreated cells. 58

69 Figure 3.3 SK-MEL-19 + IFNγ -IFNγ CHX 16hr 8hr 4hr 2hr - 16hr 8hr 4hr 2hr - Folds TAP1 GAPDH Hela + IFNγ -IFNγ CHX 16hr 8hr 4hr 2hr - 16hr 8hr 4hr 2hr - Folds TAP1 GAPDH 59

70 Figure 3.4 A homozygous single-nucleotide deletion was identified in the TAP-1 gene at position that resulted in premature termination codons. a, sequencing chromogram. The arrow points to the deletion site. b, primers htap1e7.f and htap1e7.r (arrows) were used to amplify the deletion region in TAP-1 exon 7 (E7). The arrowhead points to the position of the deletion resulting in a new BslI site. The PCR products of genomic DNA were purified and digested with BslI. Gel electrophoresis data showed that SK-MEL-19 cells were homozygous for the deletion. U, uncut; B, BslI-digested; M, molecular weight. c, the sequence of the deletion region is shown, and the downstream premature termination codons (X) are underlined. 60

71 Figure 3.4 a. b. E SK-MEL-19 M U B U B 500bp 200bp 100bp c. T P Y T W R A L S S S K M S P Mu: 1483 ACTCCC-TACACTTGGAGGGCCTTGTCCAGTTCCAAGATGTCTCCT 1528 WT: 1483 ACTCCCTTACACTTGGAGGGCCTTGTCCAGTTCCAAGATGTCTCCT 1528 T P L H L E G L V Q F Q D V S L P T Q T A Q M S X C Y R G X Mu: 1529 TTGCCTACCCAAACCGCCCAGATGTCTTAGTGCTACAGGGGCTGA 1572 WT: 1529 TTGCCTACCCAAACCGCCCAGATGTCTTAGTGCTACAGGGGCTGA 1573 F A Y P N R P D V L V L Q G L 61

72 Figure 3.5 The single-nucleotide deletion D1489 in the TAP-1 gene results in accelerated decay of the TAP-1 mrna by a non-nmd-related mechanism. RNase protection assay of three Tet-Off SK-MEL-19 cell clones stably transfected with TAP-1 WT, TAP-1 D1489, and TAP-1 Del3 cdna constructs was conducted using the antisense transcripts as probe. a, depiction of the Tet-Off system and the pbi-egfp/tap- 1 constructs containing TAP-1 WT, TAP-1 D1489, and TAP-1 Del3 used to measure the mrna half-lives. b, The nucleotide sequences around the mutation site in the cdna constructs are denoted. c, RNase protection assay to demonstrate the accelerated decay of TAP-1 D1489 and TAP-1 Del3 message compared with that of the wild type TAP-1. The antisense probe protects a 345-nucleotide TAP-1 mrna (labled as B). The protected message for EGFP and GAPDH are labeled as A and C, respectively. One representative assay for each TAP-1 construct is shown. d, decay kinetics of the TAP-1 WT, TAP-1 D1489, and TAP-1 Del3 mrna. Means ± S.D. of four experiments are presented. The half-lives of the mrna were determined by manual best-fit regression analysis. 62

73 Figure 3.5 a. b. (continued) 63

74 Figure 3.5 (continued) c. Ctrl RNase (-) Ctrl RNase (+) Tet. (Hrs) Ctrl RNase (-) Ctrl RNase (+) Tet. (Hrs) Ctrl RNase (-) Ctrl RNase (+) Tet. (Hrs) A B A B A B C C C WT D1489 Del3 d. 10 RNA remaining % WT( t 1/2 =7.2 hours) D1489 (t 1/2 =3.5 hour) Del3 (t 1/2 =2.7 hours) Hours after tetracycline 64