POST-TRANSLATIONAL MODIFICATION OF NF-κB: REGULATION OF STABILITY AND GENE EXPRESSION

Transcription

1 POST-TRANSLATIONAL MODIFICATION OF NF-κB: REGULATION OF STABILITY AND GENE EXPRESSION DISSERTATION Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University By Erin Kathleen Hertlein The Ohio State University 2006 Dissertation Committee: Approved by Denis Guttridge, PhD, Advisor Tim Huang, PhD Christoph Plass, PhD Caroline Whitacre, PhD Advisor Integrated Biomedical Science Graduate Program

2 ABSTRACT NF-κB was discovered over 20 years ago, and while the knowledge of this transcription factor has been considerably expanded, it is still not completely understood how a single signaling pathway regulates such a diverse array of events in cells. One of the ways in which this transcription factor may target a variety of genes under different conditions is through post-translational modifications that regulate how the complex binds to gene promoters, as well as how it binds to other co-factors. This thesis is designed to investigate how these modifications, specifically phosphorylation, regulate gene expression and control NF-κB mediated events. Chapter one is a general introduction to NF-κB, highlighting the members of this family, the activation pathway, and the various modifications known to occur to members of the complex. This chapter also discusses some of the known NF-κB target genes as well as diseases that are known to occur, at least in part due to deregulated NF-κB activity. Chapter two investigates the effect of phosphorylation on the stability of IκBβ, an inhibitor of NF-κB. We have found that IκBβ is normally unstable in the absence of p65 protein, however a hyperphosphorylated form of IκBβ is resistant to proteasome mediated degradation, and is stable under these conditions. We further show that this regulation is important for maintenance of normal cell growth in mouse embryo fibroblasts. Chapters three and four ii

3 focus on a different aspect of phosphorylation. In contrast to how protein phosphorylation controls stability, these chapters are designed specifically to determine how this modification affects target gene transcription. The goal of this part of the thesis is to further elucidate whether specific phosphorylation at particular resides can differentially affect sub-sets of endogenous genes. Finally, chapter five discusses the attempts to determine the role of NF-κB post-translation modification in other cell types, as well as discuss the importance of increasing our current knowledge as to how this complicated transcription factor signals. Insight into how NF-κB differentially regulates subsets of genes may allow for development of specific therapeutic targets. Currently, drugs that inhibit NF-κB on a broad level are in use, however these types of treatments may have undesirable side effects due to non-specific inhibition of other beneficial pathways in the cell. The ability to develop more specific inhibitors that affect only a small number of genes important in a particular disease will allow for more efficient therapies. iii

4 DEDICATION Dedicated to my family and friends. iv

5 ACKNOWLEDGMENTS First, I would like to thank my advisor Denis Guttridge. From the day I started in Denis lab he has been a constant source of support though every obstacle. Denis told me that in research, on any given day, I am doing something, making some discovery, that potentially no one else on the entire planet is doing. This thought has been on my mind during the frustrations and complications that I have had to face during the last five years, and have always reminded me why I love what I do. But among the many scientific lessons I have learned from you Denis, which have been considerable, equally important have been the life lessons. You have been able to perfectly balance being a successful scientist, while at the same time being a devoted father and husband. I admire your earnest dedication to both the family and the science you so dearly love, and you have shown me that the only way to be truly successful in your career is to be equally successful in life. To all the members of the Guttridge lab, you have all in your own way helped me throughout these last five years. Kate, you have been my bench neighbor, lab DJ, and a good friend. Jingxin, you are one of the most brilliant people I have met, while at the same time the most humble; from you I have truly learned a lot. Jay, you started as a coworker and quickly became a close friend. I thank you for always being there for me. I have enjoyed having you in the lab and in my life. All other members of the lab, past and v

6 present: Nadine, Mike, Huating, Swarnali, and Lori, and Jeff, I thank you all for being you. You make me happy to come into work every day, and I am privileged to work in a place with such special people. There are so many others that I have encountered throughout my time in graduate school who have helped me get where I am today. Jeni and Krissy, as fellow students and close friends, you have been there through first year classes, candidacy exams, and various other stressors. You have also been there through many fun times as well, and I couldn t have done half of what I did without you. Laura Smith, I think you are a remarkable scientist, and an equally remarkable person. Thanks for always being there to make me feel like I am brilliant, even when it isn t true. I also thank Ralf Krahe, currently associate professor at MD Anderson Cancer Center in Texas. Ralf, you introduced me to the world of science, and I never left, and for that I thank you. I also want to thank Drs. Tim Huang, Christoph Plass, and Caroline Whitacre for taking time out of their busy schedules to serve on my committee. It really is an honor to be surrounded by such exceptional researchers. In addition to all of the people who have helped me scientifically, I have been lucky to have a huge support system at home as well. To my good friends David, Lynn and Jessica, I have known you all such a long time and your friendship means the world to me. I thank you for always being there for me when I have needed you. To Cari, I have known you my whole life, and while our friendship has taken on many changes throughout the years, it has always been one constant thing I can count on. I thank you for always being there, knowing when I need to talk, and never hesitating to listen. Finally I want to thank my entire family. Without all of you to share this with, nothing I vi

7 have done would be nearly as meaningful. I thank you for supporting me throughout my many years of school! To my little brother Tommy, I am amazed at the caring and thoughtful person you have become. Even though I chose to move away, I thank you for reminding me how good it feels to come home. And to my parents, you have supported me unconditionally in every way imaginable throughout my entire life. You always make me feel that you are proud, and pleased with the person I have become. Dad, even though your hopes for me pursuing agricultural genetics haven t worked out, I love that you are still proud enough to tell all of your friends about my work. Mom, you more than anyone have been there for me. I can call you upset, frustrated, tired, happy, excited, scared, worried or just plain lost, and you always know what to say to bring me back to reality. I love you very much, you are truly my best friend, and I would not be where I am today without all of you. vii

8 VITA April 9 th 1979 Born, Vandalia, Ohio, USA June Bachelor of Science in Molecular Genetics, The Ohio State University 2001 Present...PhD Candidate, Integrated Biomedical Science Graduate Program, The Ohio State University PUBLICATIONS 1. Hertlein, E., J. Wang, K. J. Ladner, N. Bakkar, and D. C. Guttridge RelA/p65 regulation of IκBβ. Mol Cell Biol 25: Sarkar, A., M. Duncan, J. Hart, E. Hertlein, D. C. Guttridge, and M. D. Wewers ASC directs NF-κB activation by regulating receptor interacting protein-2 (RIP2) caspase-1 interactions. J Immunol 176: FIELD OF STUDY Major Field: Integrated Biomedical Science Graduate Program Areas of Emphasis: Genetics viii

9 TABLE OF CONTENTS ABSTRACT... ii DEDICATION... iv ACKNOWLEDGMENTS... v VITA...viii LIST OF TABLES... xi LIST OF FIGURES... xii LIST OF ABBREVIATIONS... xiv CHAPTER INTRODUCTION The NF-κB Family NF-κB Activation NF-κB Post-Translational Modifications Phosphorylation Ubiquitination Acetylation Sumoylation NF-κB Mediated Gene Expression NF-κB in Disease CHAPTER RELA/P65 REGULATION OF IκBβ Introduction Material and Methods Results Discussion CHAPTER DESIGN OF NF-κB CUSTOM MICROARRAY Review of Microarray Technology Design of Custom Oligonucleotide Microarray Design of PCR Spotted Array CHAPTER ix

10 PHOSPHORYLATION OF p65 AND DIFFERENTIAL REGULATION OF GENE EXPRESSION Introduction Material and Methods Results and Discussion CHAPTER DISCUSSION Overview Future Directions Approaches to study serine phosphorylation in muscle cells Approaches to study serine phosphorylation in vivo by adoptive transfer Summary REFERENCES x

11 LIST OF TABLES Table 3.1: Genes on custom oligonucleotide microarray Table 3.2: p-values for custom oligonucleotide array Table 3.3: Genes on custom spotted nylon microarray Table 3.4: p-values for replicate spots on custom spotted nylon microarray Table 4.1: Primers to generate mutant p65 constructs xi

12 LIST OF FIGURES Figure 1.1: The NF-κB Family Figure 1.2: Diagram of NF-κB Activation Figure 1.3: p65 Phosphorylation Figure 2.1: IκBβ is specifically regulated by p65, but does not require NF-κB activity. 49 Figure 2.2: IκBβ is regulated by p65 in primary fibroblasts Figure 2.3: p65 regulation of IκBβ is not limited to fibroblasts Figure 2.4: p65 regulates various forms of IκBβ in postnatal development in a tissue specific manner Figure 2.5: IκBβ downregulation in p65 -/- MEFs occurs at the protein level Figure 2.6: IκBβ downregulation in p65 -/- MEFs is regulated by the proteasome independent of classical IKK signaling Figure 2.7: IκBβ stability is directly regulated by p Figure 2.8: IκBβ stability is regulated by the carboxyl terminus of p Figure 2.9: MEFs stably expressing IκBβ exhibit a growth defect Figure 2.10: Growth defect in MEFs stably expressing IκBβ due in part to increased apoptosis Figure 3.1: Image of custom oligonucleotide microarray Figure 3.2: Diagram of experimental design for microarray hybridization Figure 3.3: Comparison of Affymetrix microarray with custom oligonucleotide array Figure 3.4: Dilution of controls for custom spotted nylon microarray xii

13 Figure 4.1: Mutation of Serines 276 and 536 reduce p65-mediated transcription of IκBα Figure 4.2: Custom NF-κB Microarray Figure 4.3: Mutation of Serine 276 displays represses activation to a greater extent than mutation of serine 536 in response to TNFα Figure 4.4: Verification of array data Figure 4.5: Mutation of multiple serine residues located in TA domain of p65 does not impair transactivation Figure 4.6: Biphasic activation of NF-κB target genes Figure 4.7: Different cytokine signaling pathways alternatively regulate p65 mediated gene expression in a manner dependent on serine phosphorylation Figure 5.1: Diagram of MSCV construct Figure 5.2: Infection of fetal liver cells xiii

14 LIST OF ABBREVIATIONS AD....Alzheimer s disease ATP..adenosine triphosphate BAFF....B-cell activating factor Bp.....base pairs CBP......CREB binding protein cdna...complementary DNAs CK....casein kinase CLL......chronic lymphocytic leukemia CMV....cytomegalovirus CTP.....cytosine triphosphate CYLD...cylindromatosis DNA.deoxyribonucleic acid ds..double stranded EMSA.....electrophoretic mobility shift assay FLC.....fetal liver cells GAPS...gamma amino propyl silane GFP.....green fluorescent protein HAT....histone acetylase HDAC.. histone deacetylase xiv

15 HIV... human immunodeficiency virus HTLV... human T-cell lymphotrophic virus IBD... inflammatory bowel disease IFN... interferon Ig..immunoglobulin IHC..immunohistochemistry IκB...inhibitor of kappa B IKK..IkB kinase IL......interleukin IP. incontenentia pigmenti LOH.... loss of heterozygosity LPS...lipopolysaccharide LT-βR...lymphotoxin beta receptor MD......muscular dystrophy MEF....mouse embryo fibroblast MSCV.....murine stem cell virus MyHC.....myosin heavy chain NEMO..NF-κB essential modulator NES.... nuclear export signal NF-κB..nuclear factor kappa B NIK..NF-κB inducing kinase NK...natural killer NLS..nuclear localization signal xv

16 PAGE...polyacrylamide gel electrophoresis PCR..polymerase chain reaction PD....Parkinson s Disease PKA..protein kinase A PMA.phorbol myristate acetate RA....rheumatoid arthritis RHD.rel homology domain RIP receptor interacting protein RNA.ribonucleic acid RSK..ribosomal S6 kinase SNPs.single nucleotide polymorphisms SUMO..small ubiquitin-related modifier TAD.....transactivation domain TFII.....transcription factor for RNA polymerase II TNFα tumor necrosis factor TNFR tumor necrosis factor receptor TRAF TNF receptor associated factor UC.ulcerative colitis UTP..uracil triphosphate UV.ultraviolet WCE.whole cell extracts xvi

17 CHAPTER 1 INTRODUCTION 1.1 The NF-κB Family NF-κB is a family of transcription factors characterized by the presence of the Rel homology domain (RHD) (Diagram Figure 1.1). Five different subunits are included in this family, RelA/p65 (from here on referred to as p65), RelB, crel, p105/p50 and p100/p52. NF-κB was originally discovered as a factor in the nucleus of B cells that binds to immunoglobulin (Ig) kappa chain [1]. This binding occurs through recognition of a consensus DNA sequence (GGGRNNYYCC; R=G/A; Y=T/C; N = any nucleotide). Three of these family members, p65, RelB and crel, contain a transactivation domain (TAD) in their C terminus responsible for activation of target genes. p105 and p100 are the inactive precursor forms of the p50 and p52 subunits. Proteolytic cleavage at a 23 amino acid glycine rich residue and subsequent ubiquitin dependent degradation of the C terminal portion of these proteins generates the active subunits [2, 3]. p50 and p52 do not have transactivation domains, but they are able to bind DNA, and as a complex with other subunits play an important role in NF-κB mediated transcription. The study of NF-κB in various knockout mouse models has revealed both unique and overlapping roles for each of the subunits in this family. Although p50 is nearly 1

18 ubiquitously expressed, mice null for this subunit are phenotypically normal. They do however display a defect in their ability to generate a humoral immune response [4]. This is likely due to the accelerated apoptosis and rapid turnover of B cells in these mice [5]. Unlike p50, p52 is predominantly expressed in stomach epithelium and immune cells [6]. Mice null for p52 are also normal, with a slight defect in the splenic and lymph node architecture [7, 8]. crel null mice have a B cell defect similar to the p50 null mice in that B cells do not proliferate in response to mitogens due to a block in the cell cycle at the G1 stage, and display elevated levels of apoptosis [5]. Studies in crel -/- mice are also some of the first to indicate that regulation of target genes by NF-κB may be gene specific. In crel -/- T cells stimulated with lipopolysaccharide (LPS), the expression of GM-CSF, G-CSF and IL-6 were higher than wild type indicating that crel acts as a repressor of these NF-κB targets rather than the expected transcriptional activator [9]. This regulation is specific to the particular activation signal as well, as crel -/- T cells stimulated with PMA/ionomycin showed reduced levels of GM-CSF [10]. RelB is normally expressed in dendritic cells and lymphocytes, and its deletion causes defects in both acquired and innate immune response [11, 12]. RelB -/- mice actually lack certain populations of dendritic cells, unlike mice null for other NF-κB family members [13]. Mice deficient in p65 exhibit the most severe phenotype. These mice are lethal at embryonic day due to TNFα induced apoptosis in the fetal liver [14]. Furthermore, MEFs isolated from p65 -/- mice are also more sensitive to TNFα induced apoptosis [15], which may be due to reduced expression of the p65-dependent anti- 2

19 apoptotic genes A1, IEX-1L, and ciap, among others [16-18]. Additional evidence for liver apoptosis as the underlying cause of embryonic lethality in the p65 -/- mice is provided in that additional deletion of TNFα or the TNFR rescues the lethality, allowing p65 -/- pups to survive to birth [19, 20]. Overall conclusions from genetic studies in knockout mice of the immediate NF-κB family indicate that p65 is the major mediator of NF-κB activity, and likely the most relevant component in terms of regulation of cell homeostasis. The active form of NF-κB is that of a dimer of two subunits. Most of the NF-κB subunits are able to dimerize, the most common form being a heterodimer of p65 and p50. crel is also able to dimerize with p50 and p52, however the contribution of this complex to overall NF-κB activity in most cell types is not as significant as p65/p50 dimers. The exception is in B cells, where crel/p50 complexes are constitutively active [21-23]. It is however possible that in cells expressing little or no p65, crel/p50 complexes may still be able to compensate for reduced NF-κB function. RelB/p52 dimers function mainly in immune cells, and are activated through an alternative, or noncanonical activation pathway (discussed further in section 1.2). p50 homodimers also occur, and because these proteins lack a transactivation domain they function as repressors of NF-κB activation. p50/p50 complexes bind to DNA and compete with the active complexes for access to NF-κB binding. These homodimers can also recruit HDAC1 activity to the promoter, leading to decreased acetylation and access of transcriptional activity [24]. When p65/p50 complexes are activated, they are able to displace the inactive dimers at the promoter. 3

20 NF-κB is inactive in unstimulated cells due to its localization in the cytoplasm, which is maintained through binding to another family of proteins the inhibitor of kappa B, or IκB family containing IκBα, IκBβ, IκBε, IκBγ, and bcl3 (Diagram in Figure 1.1). These proteins contain a series of ankyrin repeats that bind to the RHD to block the nuclear localization sequence (NLS). Due to the presence of ankyrin repeats, NF-κB proteins p105 and p100 are considered members of the IκB family as well. The C terminal region of these proteins containing the ankyrin repeats are able to fold back onto the RHD in the N terminus, masking the NLS and thereby inhibiting their own nuclear localization [25]. When NF-κB is bound to the IκB proteins, the NLS on the p50 subunit is exposed, allowing NF-κB to shuttle in and out of the nucleus and bind DNA. Therefore there is a certain amount of basal transcription of NF-κB target genes even in the absence of a stimulus, however full function requires an activation signal. While members of the IκB family primarily function to inhibit NF-κB activation, bcl3 in complex with p52 has been shown to activate the transcription of certain target genes, such as Cyclin D1 [26]. Therefore these proteins can play rather diverse roles in the regulation of NF-κB. Since IκB family members prevent constitutive activation of NF- κb, a condition present in many disease states, it is expected that deletion of these proteins would considerably affect the organism. Indeed, mice null for IκBα have severe skin inflammation and granulocytosis and die 7-10 days after birth. This condition is due to persistent activation of NF-κB, as additional deletion of p50 partially rescues the lethality allowing the mice to survive 3-4 weeks post-natal before developing the same phenotype 4

21 as IκBα -/- mice [27, 28]. Deletion of IκBβ however, does not have nearly the same detrimental effect as loss of IκBα, and in fact, mice null for IκBβ are phenotypically normal (data not published, referenced in [29]). Interestingly, knock in experiments placing IκBβ under control of the IκBα promoter rescues the lethality witnessed in IκBα null mice, indicating that these proteins are actually functionally redundant, and it is the temporal and spatial expression of these proteins which is important in NF-κB regulation [29]. A third family of proteins associated with NF-κB is the IκB kinase, or IKK family of serine kinases responsible for phosphorylation of the IκB proteins and subsequent IκB degradation and NF-κB activation. This family contains three proteins: IKKα and IKKβ which are the catalytic components of the complex, and IKKγ or NEMO, which plays a regulatory role. The kd IKK complex was identified to consist of a dimer of IKKα and IKKβ, and a dimer or trimer of IKKγ/NEMO [30]. IKKα and IKKβ share a high degree of homology with one another, although studies in knockout mice indicate rather different functional activities. Deletion of IKKβ results in embryonic death at day E [31, 32], a phenotype similar to yet slightly more severe than the p65 -/- animals. In contrast, mice null for IKKα are viable, however they exhibit abnormal skeletal and craniofacial morphogenesis as well as impaired keratinocyte proliferation and differentiation and die shortly after birth [33-35]. IKKγ/NEMO does not have kinase activity, but is important in recruiting the IκB proteins to the active IKK complex [36]. Cells deficient in IKKγ have impaired IκBα phosphorylation, and reconstitution of IKKγ increases this phosphorylation in a dose 5

22 dependent manner [36]. Deficiency of the X-linked IKKγ genes in humans causes the skin disorder incontenentia pigmenti (IP) in heterozygous females, and is embryonic lethal in males [37]. A mouse model in which the NEMO gene is disrupted develops characteristics of the human IP disease, where deletion in males is embryonic lethal and females develop a skin disorder due to granoulocytosis and increased apoptosis of keratinocytes [38]. Since its discovery, NF-κB has shown to be important in various cellular processes such as immune response, cell growth and differentiation, and apoptosis [39-42]. Mouse knockout models have aimed to elucidate how the members of this family function together to regulate these pathways. However, in terms of disease progression, it is the activity of the complex, and not the expression of the proteins themselves that is disrupted. Therefore determining how NF-κB is activated, and what events occur to deregulate this activation (discussed below) is vital to our understanding of this signaling pathway. 1.2 NF-κB Activation As mentioned, NF-κB is unlike many transcription factors in that it is normally held in the cytoplasm in an inactive state, and in order to regulate transcription it needs to be translocated to the nucleus where it can bind DNA. This is accomplished through coordinated activation and modification of many different proteins in the NF-κB and related families discussed in chapter 1.1. A diagram of NF-κB activation is provided in Figure

23 NF-κB complexes are maintained in the cytoplasm through binding to the IκB family, by the interaction of the RHD with ankyrin repeats on IκB in such a way that masks the NLS [43]. NF-κB is activated by various stimuli such as proinflammatory cytokines (TNFα, IL-1β), bacterial products such as LPS, viruses such as HTLV, UV radiation, or dsrna. These stimuli cause phosphorylation of IκB at two conserved N terminal serine residues, (serines 32 and 36 in IκBα and 19 and 23 in IκBβ). This phosphorylation targets IκB for ubiquitination and subsequent degradation by the 26S proteasome [44, 45]. The kinase responsible for phosphorylating IκB was identified to be the IKK complex (introduced in section 1.1) [46, 47]. Serine to threonine mutations in IκB determined that the kinase activity regulating IκB phosphorylation was serine specific [48]. Activation of IKK by TNFα occurs through its recruitment to TNF-R1, mediated by the regulatory subunit IKKγ/NEMO [36], and forming a complex containing TRAF2 and RIP. This results in subsequent phosphorylation and activation of the catalytic subunits IKKα and IKKβ, which in turn phosphorylate IκB proteins. While the receptors and associated proteins differ in response to other stimuli, the downstream effect of IKK activation and IκB phosphorylation is the same. Once IκB has been degraded, the NLS of p65 is exposed and the complex is free to translocate to the nucleus where it can bind DNA and activate target gene transcription. Activation of the IKK complex requires phosphorylation at two serine resides in the N terminal activation loop. Mutation of these serine residues in IKKβ, serines 177 and 181, to a non-phosphorylatable alanine reside, results in the inactivation of NF-κB signaling. Alternatively, replacement of these serine residues with glutamate, which 7

24 mimic a constitutively phosphorylated state, causes persistent NF-κB activity [47, 49]. Mutation of corresponding serine residues in IKKα do not affect IκB degradation or NFκB signaling [49], and IκBα degradation and NF-κB nuclear localization were not affected in IKKα-/- fibroblasts [35]. This indicates that activation in response to cytokines, termed classical activation, requires IKKβ and IKKγ, but not IKKα. IKKα has been shown to be important in activation of RelB/p52 complexes through a non-classical pathway involving NF-κB inducing kinase (NIK) [50, 51]. This pathway results from activation via inducers such as LT-βR, BAFF and CD40L, and does not require IKKβ, IKKγ or IκB degradation. NIK activates IKKα, which directly phosphorylates p100 and targets it for proteolytic processing, freeing the active RelB/p52 dimer [51, 52]. Once the NF-κB complexes have entered the nucleus and bound to the promoter of target genes, the complex must recruit members of the basal transcriptional complex (RNA polymerase II and associated general transcription factors), and well as various other co-activators in order for transcription to occur. One of the ways in which NF-κB associates with co-factors is through interaction with the p65 subunit. p65 contains binding sites for the TAF(II)105 subunit of TFIID, and association of TAF(II)105 is important for regulation of a subset of NF-κB target genes [53]. The activator-recruited cofactor (ARC) complex also directly interacts with p65 and enhances activation of HIV LTR [54]. In addition, p65 has also been shown to recruit CBP/p300, which contains histone acetylase (HAT) activity and therefore facilitates formation of transcriptional complexes at NF-κB regulated promoters [55]. 8

25 Because persistent activation of NF-κB has been implicated in conditions such as chronic inflammation and cancer, it is important that it be carefully regulated. One of the ways in which this is accomplished is through a feedback mechanism in which NF-κB regulates the transcription of it own inhibitor, IκBα [56-58]. Once IκBα is resynthesized in the cytoplasm, it is able to enter the nucleus where it can bind to the p65 complex and remove it from DNA exporting it back into the cytoplasm in its inactive state [59-63]. A nuclear export signal (NES) present on IκBα is responsible for this CRM1 dependent export [64, 65]. In addition to re-synthesis of IκBα, other mechanisms exist by which NF-κB signaling is terminated. A20 is another NF-κB target gene that is able to regulate NF-κB activity. A20 can bind to and shut down signaling from the TNF receptor, therefore blocking IKK activation and NF-κB signaling [66, 67]. In addition, although phosphorylation of IKK in its activation loop is essential for its activation, autophosphorylation of IKK in a C terminal cluster of serines causes a conformational change inactivating the complex and therefore decreases downstream IκB degradation and NF-κB activation [49]. NF-κB activity is terminated by methods other than negative feedback loops. For example, phosphatase PP2Cβ can associate with the IKK complex, leading to dephosphorylation in the activation loop and subsequent down regulation of activity [68]. Phosphorylation of p65 is also required for full transcriptional activity of NF-κB, and mutations in p65 altering its ability to be phosphorylated at several serine residues lead to reduced NF-κB activity [69-73]. Therefore proteins such as protein phosphatase 2A 9

26 (PP2A), which are able to de-phosphorylate p65, may be responsible for decreasing its activity [74]. NF-κB signaling is also terminated by ubiquitin dependent degradation of p65 at the promoter [75]. Other binding partners of p65 can affect its activity. Pin1, a peptidyl-prolyl isomerase, which binds to phosphorylated p65 and prevents it from interacting with IκBα, maintains the complex in the nucleus in an active state [76]. Pin1 competes with SOCS1 for binding to the same site on p65. SOCS1 is a ubiquitin ligase which targets p65 for degradation, therefore decreasing activity. 1.3 NF-κB Post-Translational Modifications As was discussed in the previous section, activation of NF-κB is a complex signaling cascade consisting of many different proteins. Control of this signaling is mediated at several different levels one of which being transcriptional feedback loops (discussed in section 1.2). However, many different post-translational events have also been shown to regulate NF-κB activity, including phosphorylation, ubiquitination, sumolation, and acetylation. This section focuses on the various protein modifications known to modulate NF-κB signaling. Phosphorylation Phosphorylation of both IκB and NF-κB proteins coincides with nuclear localization and DNA binding [23], and it is clear that phosphorylation plays an important role in how NF-κB activity is regulated. In fact, activation of classical NF-κB activity requires both phosphorylation and ubiquitination at several steps to degrade IκB 10

27 proteins [77]. As discussed in chapter 1.2, upon NF-κB activation the IKK complex is phosphorylated at an N terminal activation loop located within the kinase domain. When bound to each other in a complex, IKKα and IKKβ are able to both phosphorylate each other as well as auto-phosphorylate. However, phosphorylation at a cluster of serines in the C terminus of these proteins causes a conformational change, which unlike phosphorylation in the N terminus, actually serves to inactivate the complex [49]. Therefore, the same modification can have either a postitive or negative effect on activity, depending on where the phosphorylation occurs. This feedback mechanism in which IKK is activated and subsequently terminated prevents long-term potentiation of NF-κB activity. The IKKγ/NEMO subunit of the IKK complex is also phosphorylated in response to TNFα or Tax protein, which positively regulates IKK activity [78]. Following activation of the IKK complex, the next step in NF-κB activation is modulation of the IκB proteins. The IκB proteins are inherently unstable, and are rapidly degraded when not bound to the NF-κB complex [79]. IκBα is phosphorylated by IKK at two conserved serine resides in the N terminus, targeting it for ubiquitination via the βtr-cp ubiquitin ligase. Phosphorylation in the C terminal PEST domain of IκBα also destabilizes the protein allowing degradation in response to UV-C, however this occurs through an IKK-independent mechanism [80]. Like IκBα, IκBβ is also phosphorylated, ubiquitinated and degraded in response to IKK activation. Interestingly however, as will be discussed in Chapter 2, our lab has identified a hyperphosphorylated form of IκBβ which is stabilized in the absence of p65, while the basally phosphorylated form is 11

28 rapidly degraded [81]. Therefore phosphorylation under certain conditions can serve to protect IκB from proteasome-mediated degradation, rather than induce it. In addition to IκBα and IκBβ, other members of the IκB family can be regulated by phosphorylation. p105 and p100 require phosphorylation for efficient processing and degradation to the active p50 and p52 subunits of NF-κB. This is particularly important for RelB/p52 dimers, as they are activated solely by this IκB independent mechanism. Phosphorylation of the p65 subunit is also required for full transcriptional activity of NF-κB. Serine 276 is phosphorylated in the cytoplasm in response to LPS by the catalytic subunit of PKA, a component of the p65/p50/iκb complex [69]. A conformational change occurs upon the degradation of IκB exposing serine 276 and facilitating phosphorylation, and this is important for recruiting CBP/p300 to the complex [55]. Once the active p65/p50/cbp/p300 complexes enter the nucleus, they are able to displace repressive p50/p50/hdac1 complexes at the promoter of target genes [24]. Serine 276 is also phosphorylated in response to TNFα by MSK1 [70], however this occurs in the nucleus rather then the cytoplasm. Serine 529 is also phosphorylated in response to TNFα and IL-1β via casein kinase II (CKII) [71, 82, 83]. Phosphorylation of p65 by IKKα or IKKβ also occurs at serine residue 536 in response to LPS, IL-1β, TNFα, and LT-βR [84-87], and more recently it has been shown that 536 is constitutively phosphorylated by IKKε/IKKi [88]. It is suggested that phosphorylation in this manner regulates the constitutive NF-κB activity evident in some types of cancer. While phosphorylation of p65 generally occurs in parallel with IκBα degradation, IκBαindependent p65 phosphorylation by ribosomal S6 kinase (RSK1) can occur in response 12

29 to some stimuli such as angiotensin II [89]. Phosphorylation of serine residues 276, 529 and 536 have been to date the most extensively studied. These residues and the upstream activating kinases and activating stimuli are shown in Figure 1.3. Phosphorylation at serine 311 by PKCδ is necessary for recruitment of members of the basal transcription complex, as mutation of serine 311 to alanine prevent recruitment of CBP and RNA polymerase II to the IL-6 promoter [90]. Phosphorylation at serine 468 in transactivation domain 2 of p65 is interesting given that it has different effects on transcription depending on the cellular signal. In response to TNFα, GSK-3β can phosphorylate serine 468, and this is required for at least a subset of NF-κB regulated genes [91]. However, in unstimulated cells, phosphorylation of 468 (also by GSK-3β) leads to transcriptional repression [92]. Phosphorylation of p65 at serine 468 can also occur via IKKβ in response to TNFα and IL-1. Under these conditions, mutation of serine 468 to alanine does not affect nuclear localization, but does enhance transcription of some NF-κB target genes indicating that serine 468, when phosphorylated by IKKβ, is inhibitory [93]. Regardless of the position of the serine residues, it is clear that phosphorylation of p65 is necessary for full NF-κB activation. Other members of the NF-κB family are phosphorylated, however less is known about their role in NF-κB mediated transactivation. Phosphorylation of RelB at threonine 84 and serine 552 are necessary for proteasome-mediated degradation of the protein, and subsequent reduction of NF-κB activation [94]. Whereas phosphorylation of serines 886 and 870 of p100 are critical for processing this protein generating active RelB/p52 dimers [52], phosphorylation at serine 368 stabilizes p100, therefore reducing the activity of 13

30 RelB/p52 complexes [95]. crel is also phosphorylated by an unknown kinase at an ERK consensus site at serine 451 [96]. Ubiquitination Ubiquitin is a reversible modification which occurs by coordinated activation of ubiquitin-activating enzyme (E1), ubiquitin-conjugating enzyme (E2), and ubiquitinprotein ligase (E3). There are several examples of ubiquitination of the IKK proteins. Mono-ubiquitination at lysine 163 is required for C terminal phosphorylation of IKK which leads to its degradation [49, 97]. Alternaively, ubiquitination of NEMO positively regulates IKK activity [98]. The tumor suppressor CYLD contains deubiquitinating activity and is able to decrease IKK activity by removing this modification from IKKγ [99]. However is has also been shown that IKKγ dependent phosphorylation of CYLD disables the DUB activity [100], indicating yet another feedback mechanism governing NF-κB activity. As mentioned in Chapter 1.2, the IκB proteins are ubiquitinated upon activation of NF-κB to signal proteasome-mediated degradation. In a similar manner, the p100 and p105 subunits of NF-κB are ubiquitinated to recruit proteasome which partially degrades the protein, removing the inhibitory C terminal portion and allowing activation of active p50 and p52 containing complexes [101]. p65 has also been shown to be ubiquitinated, leading to proteasome-dependent degradation at NF-κB promoters [75]. This degradation is important to reduce NF-κB 14

31 activity in the absence of IκBα which normally serves to terminate NF-κB signaling by removing active complexes from the promoter [59, 60, 102]. Acetylation Acetylation occurs on histone proteins causing a conformational change in chromatin allowing active transcriptional complexes to access the promoter. Acetylation can also occur on other types of proteins as well, and this modification can regulate the activity of transcription complexes in a variety of ways: recruitment of co-activators, regulation of DNA binding affinity, as well as affinity of a protein for other binding partners. In many cases, acetylation is preceded by other modifications. Phosphorylation at serine 276 and 536 of p65, and subsequent recruitment of CBP/p300 is a prerequisite for acetylation of p65 at lysine 310, which is important for increased transcriptional activity of p65 [103]. This increase in activity occurs without altering DNA binding of the complex. Alternatively, acetylation at lysine residue 221 and possibly 218 is important for increasing DNA binding by impairing p65 s interaction with IκBα, which facilitates nuclear export [104]. Interestingly, while acetylation of 310 appears to be more important for full activation of the p65/p50 complex, it does not increase DNA binding as does acetylation at 221. Therefore it is likely that 310 acetylation contributes to transcriptional activation by other methods such as recruitment of transcriptional machinery or other co-activators. While acetylation at lysine residues 218, 221 and 310 are clearly important in maintaining activity, acetylation at lysine 122 and 123 are inhibitory. Acetylation at these residues weakens the affinity of the p65/p50 complex for DNA, facilitating its removal and nuclear exportation by IκBα [105]. 15

32 Sumoylation Sumoylation of proteins is a modification similar to ubiquitination, where a small SUMO (small ubiquitin-related modifiers) molecule is attached enzymatically to the protein. SUMO modification has been shown to act both as an activator and a repressor of NF-κB activity. Sumoylation occurs on IκBα in the nucleus on the same lysine residues to which ubiquitin molecules attach. This prevents ubiquitination of the protein, thereby protecting IκBα from degradation [106] and repressing NF-κB mediated transcription. Sumoylation also occurs on IKKγ/NEMO in response to stress. However unlike SUMO modification of IκBα, which prevents ubiquitination, sumoylation of IKKγ is a prerequisite for its nuclear localization, subsequent ubiquitination, and activation of cytoplasmic IKK complexes [107]. 1.4 NF-κB Mediated Gene Expression NF-κB regulates hundreds of targets genes, either directly through binding to the promoter, or indirectly through the transcriptional regulation of intermediate factors. These targets include genes involved in immune response, cell stress response, cell survival and apoptosis, cell migration, and cellular differentiation. In this section, the main focus is to describe a portion of these target genes, which are later described in the design of a custom microarray (chapter 3). NF-κB mediated immune response is perhaps the most well known function of this complex, given that the transcription factor was first discovered in B cells. NF-κB is 16

33 activated by a variety of stimuli which elicit an immune response, such as bacterial products and viruses, most commonly LPS [108] and the Tax protein of HTLV [109, 110]. In addition, a number of proinflammatory cytokines produced during an immune response also activate NF-κB such as IL-1β and TNFα. In fact, not only is NF-κB activated by these cytokines, but the complex in turn transcriptionally regulates many of them as well. Some of the common chemokines and cytokines regulated by NF-κB are Groα, IP-10, MIP-2, MCP-1/CCL2 and TNFα. Most of these genes are known to play a role in autoimmune diseases such as rheumatoid arthritis or psoriasis, and some of them such as TNFα, have even been used as therapeutic targets for treatments of these diseases (discussed in section 1.5). IP-10 and MCP-1 are important for activation of immune cells such as monocytes, natural killer (NK) cells, and T-cells. C3 is important for activation of the complement system, and C3 deficiency correlates with higher risk of bacterial infection [111]. In addition to cytokines and other ligands, immunoreceptors regulated by NF-κB such as MHC class I and CD40 are also important in maintaining normal immune function. In addition to immune response, NF-κB also plays an important role in cell growth and survival. While most of NF-κB s activities favor cell survival, its regulation of genes such as FasR, FasL and A20 are pro-apoptotic, through the induction of TNF mediated cell death or inhibition of NF-κB signaling. On the other hand, pro-survival genes Bcl2, A1/Bfl1 and Bcl-xL and inhibitors of apoptosis such as IEX-1L, c-flip and ciap are also NF-κB regulated. CyclinD1 is a known NF-κB target, and fibroblasts lacking NF-κB activity exhibit reduced levels of cyclind1 as well as impaired re-entry 17

34 into the cell cycle, indicating regulation of cyclind1 by NF-κB is important in this process [112]. As mentioned above, NF-κB regulates genes involved in immune response as well as cell survival, both important aspects of NF-κB s role in cancer. However NF-κB also regulates genes involved in cellular migration, such as matrix metalloproteinases (Mmp s). These proteins play a role in cancer metastasis by breakdown of the extracellular matrix, allowing migration through the basement membrane and into the blood vessel [113]. 1.5 NF-κB in Disease As mentioned earlier in this introduction, constitutive NF-κB activity has been implicated in multiple diseases. Due to its role in these conditions, NF-κB activity has been targeted therapeutically, and depending on how NF-κB is involved in different diseases, specifically the genes it regulates, is an important aspect of how to design these pharmaceuticals. Due to NF-κB s role in immune response, this transcription factor has been identified in several different autoimmune diseases. Inflammatory bowel disease (IBD), a common name for a group of symptoms including Crohn s disease and ulcerative colitis (UC), is a condition of chronic inflammation of the intestinal tract. Patients with Crohn s have been show to display increased NFκB activity [114], and it is likely that the increased expression of NF-κB regulated cytokines such as IL-6 and TNFα contribute to the increased inflammation in these patients [ ]. There is currently no cure for 18

35 this disease, however there are drug treatments available to alleviate the inflammation and limit the disease progression. Some of these drugs, immunosuppressive and antiinflammatory agents, target the general immune response without specifically inhibiting NF-κB. However some drugs designed to directly block NF-κB [119] or its upstream activators such as TNFα (Remicade, Enbrel, or Humira ), have also been identified. Other diseases resulting from abnormal immune response and known to involve NF-κB are rheumatoid arthritis (RA) and muscular dystrophy (MD). Fibroblast-like synoviocytes taken from RA patients display increased DNA binding of p65 and p50, as well as high levels of IL-6 [120, 121]. This increased NF-κB activity protects cells from apoptosis, causing hyperplasia and exacerbating the inflammatory response [122]. In muscular dystrophies, progressive damage to the muscle eventually leads to the advanced form of the disease. This damage is due to a combination of events, in part due to mutations in genes necessary for protection during muscle contraction. However progression of the disease also involves an early increase in inflammation in muscle tissue, which indicates a possible role for NF-κB. Studies have shown that NF-κB binding activity is increased in the diaphragm muscle of mdx mice, the model for Duchenne Muscular Dystrophy (DMD). The level of NF-κB target genes TNFα and IL- 1β increase as well [123], and this increase in NF-κB regulated inflammatory cytokines occurs just prior to disease onset in this model. Increased NF-κB DNA binding activity was also evident in muscle samples from patients with DMD compared to normal controls [124]. 19

36 While NF-κB s role in inflammation is important for the conditions described above, its role in apoptosis and cell survival is equally important and has been implicated in other disorders. Neurological diseases such as Alzheimer s (AD) and Parkinson s (PD) are characterized by the death of nerve cells in the brain. NF-κB may actually play opposing roles in the pathogenesis of Alzheimer s, depending on the target genes and biological processes affected. β-amyloid, a gene involved in the progression and risk for AD, activates caspase mediated apoptosis in neuronal cells in AD. NF-κB mediated regulation of caspase inhibitors and other cell survival genes is important in preventing this death, allowing survival of these cells. NF-κB is reduced in β-amyloid plaques in AD patients [125, 126], which increases susceptibility of neuronal cells to apoptosis. On the other hand, NF-κB activity has been found to be elevated in AD patients, leading to increased inflammatory response and neuronal damage [127, 128]. NF-κB is similarly increased in neurons of Parkinson s patients as well [129]. NF-κB is also important in diseases such as atherosclerosis and heart disease. Two of the early steps in initiating atherosclerosis, modification of LDL and attraction of monocytes, are regulated by NF-κB target genes COX-2 and MCP-1, respectively [ ]. NF-κB also regulates selectins and cellular adhesion molecules which have been implicated in atherosclerosis mouse models [ ]. NF-κB is also activated in congestive heart failure [136]. Cancer is yet another disease involving inflammation, cell survival and apoptosis mediated by NF-κB. NF-κB can activate proto-oncogenes such as c-myc and cyclin D1 [112, 137], leading to cell growth and tumor formation. Extensive studies have provided 20

37 evidence that constitutive NF-κB activity is involved in the progression of several lymphomas and myelomas [ ] through mediating resistance to apoptosis and regulation of proliferation. Alternatively, decreased NF-κB activity can also lead to squamous cell carcinoma and epidermal neoplasia [145, 146] due to spontaneous hyperplasia and a lack of cell cycle arrest. The mechanism by which NF-κB is able to control such diverse pathological states as described here, as well as the fact that NF-κB can act as either a positive or a negative influence on disease, is likely due to the specific target genes regulated in each condition. This underscores the importance of further determining how this transcription factor differentially regulates gene expression, and how it can be more efficiently used a pharmacological inhibitor. 21

38 Figure 1.1: The NF-κB Family Diagram of the NF-κB and IκB families. The rel homology domain (RHD) and the transactivation domains (TAD) of the NF-κB family are shown in red and purple, respectively. The ankyrin repeats within the IκB family are in yellow. The location of the NF-κB nuclear localization signal (NLS) is also indicated. 22

39 Figure 1.2: Diagram of NF-κB Activation When NF-κB receives a stimulus such as pro-inflammatory cytokines, the IKK complex is activated. This causes phosphorylation of the IκB proteins, subsequent ubiquitination and degradation by the 26S proteasome and exposure of the NF-κB NLS. Once the NFκB complex enters the nucleus, it binds to DNA and activates target gene transcription. One of the target genes activated is the IκBα, which is resynthesized and serves to terminate NF-κB activation through nuclear export and inhibition of NFκB. 23

40 Figure 1.3: p65 Phosphorylation Diagram indicating the upstream signals and kinases known to induce phosphorylation of three common serine residues in p65. 24

41 CHAPTER 2 RELA/P65 REGULATION OF IκBβ 2.1 Introduction As discussed in Chapter 1, NF-κB is a vital regulator of cellular processes involved in immune response, cellular proliferation, differentiation and apoptosis [39-41, 147]. Constitutive activation of NF-κB is also thought to contribute to multiple pathophysiological conditions such as rheumatoid arthritis [148], inflammatory bowel disease [149], AIDS [150], and with ever increasing evidence, cancer [ ]. Each NF-κB subunit contains a Rel domain specifying DNA binding, protein dimerization, and nuclear localization, and although in vitro most NF-κB subunits possess the ability to homo or heterodimerize, in vivo, NF-κB primarily exists as a p50/p65 heterodimer. NF-κB is sequestered predominantly in the cytoplasm bound to IκB inhibitor proteins. Classical activation of NF-κB proceeds by the degradation of IκB proteins, which is mediated by the activity of the IKK [41, 155]. In response to a multitude of stimuli, IKK is activated leading to subsequent IκB degradation via the 26S proteasome [41, 156] and activation of gene expression [24, ]. Over the past decade there has been a concerted effort to understand the function by which IκB proteins regulate NF-κB activity. Earlier studies demonstrated that shortly 25

42 following NF-κB activation, IκBα is resynthesized in an NF-κB dependent manner [56-58]. Following this resynthesis, IκBα enters the nucleus by a yet to be confirmed mechanism where it then binds and removes NF-κB from the DNA and exports the complex back to the cytoplasm [59, 61-63]. By this autoregulatory mechanism NF-κB transcriptional activity remains transient, lasting between 1-4 hours in most cells, with the exception in mature spleenic B cells where the p50/c-rel complex is constitutively active [21, 22, 161]. Disruption of this regulatory loop via the deletion of IκBα expression leads to persistent p65 activity and postnatal lethality [27, 28]. More recently, structural analysis of IκBα in complex with the p50/p65 heterodimer revealed that IκBα ankyrin repeats 3-6 contribute the majority of p50/p65 binding while repeats 1 and 2 are more loosely associated with the NLS site of NF-κB [162, 163]. The p50 NLS remains exposed when bound to IκBα [162, 163], which is thought to allow for cytoplasmic to nuclear shuttling of the complex [164], while reverse shuttling is regulated by nuclear export signals located in the carboxyl and amino terminal halves of IκBα [60, 64, 65, 102], as well as in the TAD of p65 [165]. In contrast to this well described regulatory interplay between NF-κB and IκBα, the regulation of NF-κB by IκBβ and a potential feedback mechanism has not been described. Unlike IκBα the beta isoform is not rapidly degraded by classical NF-κB inducing signals, and following NF-κB activation IκBβ is also not resynthesized in an NF-κB-dependent manner. Depending on given cell type or stimulus IκBβ may instead undergo persistent degradation leading to constitutive NF-κB activity [ ]. Constitutive NF-κB activity is also regulated by a hypophosphorylated form of IκBβ that 26

43 is capable of competing with IκBα for NF-κB binding, but is incapable of dislodging NFκB from the DNA [150, 169]. Basal phosphorylation of IκBβ occurs in its carboxyl terminal PEST domain that functions to inhibit NF-κB DNA binding, and is thought to be primarily responsible for the formation of latent IκBβ/NF-κB complexes [170, 171]. However, unlike IκBα/NF-κB, IκBβ/NF-κB complexes do not undergo cytoplasmic to nuclear shuttling [172], due to the addition of a linker region between ankyrin repeats 3 and 4 in IκBβ that binds κb-ras to efficiently mask the second NLS in the NF-κB dimer complex [ ]. Also, unlike IκBα [27, 28], mice lacking IκBβ were noted to have a mild phenotype [29], suggesting at first that these proteins are functionally distinct. However, knockin expression of IκBβ under the control of the IκBα promoter rescued IκBα associated lethality [29]. This demonstrates that functional overlap between these proteins clearly exists, but given the overt phenotypic differences among respective IκB knockouts, it also points to the importance of spaciotemporal expression of IκB proteins in the regulation of NF-κB. We have made the observation that IκBβ levels are dramatically reduced in MEFs null for the p65 subunit of NF-κB. Although others have indirectly noted this phenomenon [27, 177], to date a detailed characterization of this regulation has not been performed. In this chapter, the specificity, mechanism, and physiological relevance of p65 regulation of IκBβ are now described. Our results reveal a remarkable dependence of IκBβ for p65 in most, but not all tissues. This dependence is mediated through the stabilization IκBβ protein by p65 and its carboxyl terminus encompassing the TADs. Our results also elucidate that IκBβ expression in p65 null MEFs has a severe impact on 27

44 cell growth and viability. Interestingly, although p65 is considered constitutively expressed, studies have reported that p65 expression is in fact low or even undetectable in early development [178, 179] and in selective cell types [180]. Based on such studies as well as our own current findings, we suggest that the destabilization of IκBβ in cells lacking p65 is a regulatory process that might have emerged to ensure proper cell growth and viability. 2.2 Material and Methods Materials. Murine TNFα was purchased from Roche Biochemicals (Indianapolis, IN). Antibodies to IκBα (C21), IκBβ (C20, N20), Bcl-3, IKKγ, and p100 were obtained from Santa Cruz Biotechnology (Santa Cruz, CA), myosin heavy chain and α-tubulin from Sigma (St. Louis, MO), p65 from Rockland Immunochemicals, Inc. (Gilbertsville, PA), and hemagglutin (HA) from Covance (Princeton, NJ). MG-132, ALLN and lactacystin were purchased from Calbiochem (San Diego, CA), cyclohexamide from Sigma (St. Louis, MO), and λ-phosphatase from New England Biolabs (Beverly, MA). [ 35 S] Easy Tag protein labeling mix was purchased from NEN (Boston, MA) and methionine/cysteine-free Dulbecco s modified Eagle s media (DMEM) from Invitrogen (Carlsbad, CA). Materials required for immunohistochemical analysis were obtained from Vector Laboratories (Burlingame, CA). Cell culture. All fibroblast cells were cultured in DMEM high glucose media containing 10% fetal bovine serum and antibiotics. For p50 and Bcl-3 MEFs, mice null for these proteins (Jackson Laboratories, Bar Harbor, MA) were crossed with their respective wild 28

45 type strains. Resulting heterozygotes were then bred and MEFs were prepared from embryos at day13.5 pc. In a similar manner, p65 MEFs were generated from E13.5 p65 +/+, +/-, and -/- embryos. C2C12 myoblasts were cultured as previously described [112]. Plasmids. Full length p65 (FL), and carboxyl truncation mutants, p65( 534), p65( 521) and p65( 313) were cloned into the pflag-cmv-2 expression plasmid as previously described [71]. For the generation of p65( 431), Flag tagged p65(fl) was used as a template and a DNA fragment corresponding to amino acids was amplified with primers, 5 -GATCAAGCTTGACGAACTGTTCCCCCTCATC and 3 - GATCGATATCTCAAGCCT GGGTGGGCTTGGGG. The DNA was subsequently cloned into the HindIII/EcoRV sites of pflag-cmv-2 plasmid. Construct p65( 319) was generated in a similar manner using primers 5 - GATCAAGCTTGACGAACTGTTCCCCCTCATC and 3 - CGATATCTCAGCTGAAAG GACTCTTCTTCATG. Retroviral expression constructs for p65 and IκBβ were created by RT-PCR to amplify the respective cdna from human fibroblasts and cloned into pbabepuro. The p65 construct was generated to produce the full-length protein. Wild type IκBβ and IκBβ-SR (deleted in the signal response region, amino acids 1-54) were generated with an N-terminal HA epitope tag. pbabeiκbα-sr was generated by excising the cdna of human IκBα-SR (mutated in Ser-32 and Ser-36 to alanines) from a pcmv4 expression plasmid (generously provided by D. Ballard, Vanderbilt University) with Bgl II/Sma I restriction enzymes and subcloned in 29

46 BamH1/SnaB sites in pbabepuro. pcmviκbβ(s19/23a) and pcmviκbβ(s19/23e) were generated by site directed mutagenesis from pcdna3.1 and phm6 expression plasmids containing IκBβ using the QuikChange Site-Directed Mutagenesis kit (Stratagene) with the following primers (mutated residues are underlined): S19/23A 5 - GAATGGTGCGACGCCGGCCTGGGCGCCCTGGGTCCG-3 and S19/23E 5 - CAGATGAATGGTGCGACGAAGGCCTGGGCGAGCTGGGTCCGGAC-3. All clones were confirmed by DNA sequencing and Western blot analysis. Immunoblotting, EMSA and kinase assay. Whole cell extracts from cultured cells and immunoblotting were prepared as previously described [112]. Extracts from mouse tissue were prepared by homogenization in lysis buffer (1% Triton-X, 150mM NaCl, 50mM Tris [ph 7.5], 1mM EDTA, 1mM PMSF, standard protease inhibitors). Protein detection was obtained by enhanced chemiluminescence (PerkinElmer Life Sciences, Boston, MA) and imaged either by using a Chemidoc gel documentation system (BioRad Laboratories, Hercules, CA) or by exposing blots to film. All quantitation was performed using the ImageJ software (NIH, Bethesda MD). Electromobility shift and IKK kinase assays were performed as described, respectively [46, 181]. [ 35 S]-Labeling and immunoprecipations. For labeling reactions, p65 +/+ and p65 -/- MEFs were cultured in DMEM media lacking methionine or cysteine for 2 hours and subsequently pulse labeled with [ 35 S] methionine and cysteine for up to 1 hour. Whole cell extracts were prepared in a standard RIPA buffer. For immunoprecipitation, 500 µg of protein was pre-cleared with rabbit IgG immunoglobulin for 2 hours and non-specific 30

47 complexes were precipitated by centrifugation using 25 µl of A/G agarose beads (Santa Cruz Biotechnology, Santa Cruz, CA). Supernatants were then incubated overnight at 4 o C with 1 µg of either IgG or an IκBβ antibody. The following day complexes were precipitated by centrifugation using 30 µl of A/G agarose beads, washed three times in lysis buffer, resuspended in gel loading buffer and fractionated by SDS PAGE. Gels were dried and visualized on X-OMAT film. Transfections and retrovirus infections. Typically, 75% confluent MEFs were transfected in low serum Opti-MEM media using Lipofectamine reagent (Invitrogen, Carlsbad, CA) according to manufacturer s recommendations. Retrovirus production and infections were performed as previously described [112]. Immunohistochemistry. E13.5 embryos either wild type or null for p65 were fixed in 10% formalin overnight at 4 o C and then dehydrated and paraffin embedded. Longitudinal sections of the entire embryo as well as cross sections through selective tissues were prepared. Sections were deparaffinized by heating at 60 0 C for one hour followed by treatment with xylene, and rehydration. Slides were treated with 3% hydrogen peroxide to block any endogenous peroxidase activity that could interfere with the detection reaction. Sections were steamed for 30 minutes in antigen retrieval solution, and then incubated in avidin and biotin solutions. Following blocking for one hour in 5% goat serum diluted in PBS, sections were incubated with primary antibody against p65 (for 1 hour; 1:5000 dilution in 5% goat serum), IκBβ (for 1 hour; 1:300 dilution in 5% goat serum) or myosin heavy chain (for 30 minutes, 1:2000 dilution in 31

48 MOM diluent). Sections were subsequently incubated with a biotinylated secondary antibody. For p65 and IκBβ, the secondary used was goat anti-rabbit (for 30 minutes, 1:250 in 5% goat serum) and for MyHC, goat anti-mouse (for 10 minutes, 1:250 in MOM diluent). Sections were developed by incubation with avidin/biotin complexed peroxidase to recognize secondary antibody (ABC Elite) and using 3,3 - diaminobenzidine as enzyme substrate (DAB kit). Real Time PCR. RNA was prepared in TRIzol reagent (Invitrogen, Carlsbad, CA) as recommended by the manufacturer and further purified using RNeasy affinity columns (Qiagen, Valencia, CA). cdna was generated from 2 µg total RNA by reverse transcription with Superscript II (Invitrogen, Carlsbad, CA) according to manufacturers specifications. 1 µl of cdna was used as template in a total reaction volume of 25 µl containing final concentrations of 1X iq SYBR Green Super mix (BioRad, Hercules, CA) and 0.5 µm each of forward and reverse primers. Primer sequences used: IκBα 5 - GGAGACTCGTTCCTGCACTTGG, 3 -AACAAGAGCGAAACCAGGTCAGG; IκBβ 5 -ACACAGCCCTGCACTTGGCTG, 3 -GGTATCTGAGGCATCTCTTGGG; internal control GAPDH 5 -GCAAATTCAACGGCACAGTCAAG, 3 - GTTCACACCCATCACAAACATGG. Data was read and collected on the BioRad icycler. Mice and genotyping. Animals were housed in the animal facility at the Ohio State University Heart and Lung Research Institute under super sterile conditions maintaining constant temperature and humidity, and fed a standard diet. Treatment of mice was in 32

49 accordance to institutional guidelines for Animal Care and Use Committee. p65 -/- ;TNFα - /- mice were generated as previously described [20]. Briefly, p65 +/- ;TNFα +/+ mice were crossed to p65 +/+ ;TNFα -/- mice (Jackson Laboratories, Bar Harbor, MA). From this cross, resulting p65 +/- ;TNFα +/- mice were crossed to obtain p65 +/+ ;TNFα -/- and p65 -/- ;TNFα -/- mice in the expected Mendelian ratios. Genotypes for p65, TNFα, NFKB1/p50, and Bcl-3 mice were confirmed by PCR analysis from prepared tail DNA. Growth Curves and Flow Cytometry. p65 -/- cells infected with pbabe vector, IκBβ- SR, or IκBα-SR were grown under puromyocin selection. 1 x 10 4 cells were plated in triplicate in 12-well cell culture plates and counted on indicated days. Apoptosis was evaluated using Annexin-V-FITC staining according to the manufacturer s specifications (Santa Cruz Biotechnology, Santa Cruz, CA). Briefly, 5 x 10 5 cells were washed with cold phosphate buffered saline and suspended in 50 µl of Annexin-V-FITC staining solution. After 15 min of incubation at room temperature, cells were fixed in 10% formaldehyde and subsequently analyzed on a FACS Calibur flow cytometer (Becton- Dickinson, Mountain View, CA, USA). Fluorescence data was analyzed using CellQuest Pro software. 2.3 Results Regulation of IκBβ is specific for the loss of p65. MEFs null for p65 were utilized in an interest to gain insight into the mechanisms of NF-κB-dependent transcription. During the course of this analysis we made the observation that IκBβ 33

50 levels were strikingly lower in p65 -/- fibroblasts compared to their wild type counterpart (Fig. 2.1A). Although IκBα is a transcriptional target of p65, its level of regulation was noticeably less than that for IκBβ. In addition, no detectable changes in IκBε, p100, or Bcl-3 were seen between p65 -/- and p65 +/+ cells, demonstrating that p65 does not function as a general regulator of IκB proteins. To determine the specificity of this regulation, levels of IκBβ were compared in MEFs lacking other subunits in either the NF-κB family or NF-κB signaling pathway. As opposed to the marked reduction of IκBβ in p65 -/- MEFs, no changes in IκBβ were observed in fibroblasts lacking c-rel, RelB, p50, p52, IKKα, IKKβ, IKKγ, IκBα, or Bcl-3 (Fig. 2.1B), implying that regulation of IκBβ is specific to p65. To determine whether regulation of IκBβ was due to the physical absence of p65 or simply loss of its activity, we verified IκBβ protein levels in myoblast cells devoid of NF-κB transactivation function due to the stable expression of the degradation resistant IκBα-SR mutant [182]. In these cells, lack of NF-κB activity maintained IκBβ protein levels (Fig. 2.1C) indicating that reduction of IκBβ results from the physical loss of p65. Since decreases in IκBβ were detected using established fibroblasts, it was important to determine whether this regulation was a consequence of the immortalization process due to the absence of p65. Mice heterozygous for p65 were therefore bred and E13.5 MEFs were prepared. In comparison to wild type cells, p65 +/- primary MEFs expressed approximately 50% less IκBβ, and strikingly, IκBβ was nearly undetectable in null cells (Fig. 2.2A and 2.2B). In contrast, IκBα expression was only slightly reduced in p65 +/- MEFs, while approximately 35% remained in cells lacking p65. These data 34

51 demonstrate that in fibroblasts, p65 regulation of IκBβ is not a phenomenon of cellular immortalization, and the degree to which p65 controls IκBβ expression is significantly higher than that for IκBα. p65 regulation of IκBβ is maintained in embryonic and postnatal development. To determine if this regulation occurred in cells other than MEFs, immunohistochemical analysis of IκBβ was performed in p65 +/+ and p65 -/- embryos at day 13.5, a time that precedes liver apoptosis and lethality of p65 null mice [14]. As expected, no overt morphological defects were observed at this developmental stage in embryos lacking p65 (Fig. 2.3A). Results revealed however that IκBβ expression was generally reduced in p65 -/- embryos, with more apparent regulation occurring in liver, lung, and brain (Fig. 2.3A and 2.3B). This implied that IκBβ regulation by p65 occurred in multiple cell types. To confirm that these findings were not due to staining artifacts, immunoblotting was performed from isolated fetal livers. In line with the immunohistochemistry data, IκBβ was found strongly repressed in p65 -/- liver cells (Fig. 2.3C). Immunoblots also revealed that, similar to embryonic fibroblasts, the downregulation of IκBα in p65 -/- livers was not nearly to the same extent as that for IκBβ, reaffirming the tight control of IκBβ expression by p65. Next we asked whether p65 regulation of IκBβ could be maintained in adult mice. Although p65 -/- mice die between E [14], liver apoptosis and embryonic lethality can be rescued with the additional deletion of TNFα [183]. Thus, p65 -/- ;TNFα -/- double knockouts were generated and at approximately 4 weeks of age, p65 +/+ ;TNFα -/- and p65 -/- ;TNFα -/- mice were sacrificed and tissues were processed for immunoblot 35

52 analysis. Except for a p65 reactive band that reproducibly appeared in brain, the complete absence of p65 staining in all remaining tissues confirmed the null phenotype of these mice (Fig. 2.4A). In agreement with immunohistochemical data, IκBβ was also found generally repressed in p65 -/- tissues suggesting that this regulation is maintained into adulthood (Fig. 2.4A). Of the p65 -/- tissues examined, spleen, thymus, and skin contained the least IκBβ, with less but still significant reduction observable in brain, liver and lung as compared to wild type mice (Fig. 2.4A and 2.4B). Although IκBα was also down regulated in tissues lacking p65, similar to cultured MEFs and liver cells, the qualitative difference in expression compared to wild type tissues was not as significant as that for IκBβ. IκBβ is expressed in alternative forms that are differentially regulated by p65. Interestingly, upon closer examination of immunohistochemical sections shown in Figure 2.3, we observed that in heart and skeletal muscle, levels of IκBβ expression appeared almost comparable between p65 +/+ and p65 -/- embryos (Figs. 2.3D and 2.3E; note the colocalization between IκBβ and the skeletal muscle marker, myosin heavy chain). This suggested that IκBβ regulation by p65 might also be tissue specific. Indeed, immunoblot analysis in Figure 2.4A showed significantly less reduction of IκBβ in skeletal muscle and heart as compared to other tissues lacking p65. IκBβ levels in these mice were also largely retained in testis, which is consistent with previous findings showing that IκBβ is particularly rich in this tissue [184]. Of further interest was the identification that IκBβ produced from skeletal muscle and heart migrated at a distinctly higher mobility compared to other tissues (Figs. 2.4A and 2.4B). In fact, as opposed to reports that 36

53 murine cells produce only one form of IκBβ [185, 186], under our standard gel fractionation conditions, at least four forms of IκBβ were clearly discernable, which we refer to as forms I, II, III and IV (Fig. 2.4B). IκBβ-I and IκBβ-IV represented the major and minor expressing forms in most tissues respectively, and both were sensitive to p65 regulation, although IκBβ-I appeared more so then IκBβ-IV (Fig. 2.4A). However, in skeletal muscle and heart, IκBβ-I and IV forms were less expressed, while IκBβ-II and IκBβ-III forms were readily detectable as compared to other tissues. In addition, similar to what we had observed with IκBβ-I in testis, IκBβ-II and IκBβ-III forms also remained largely expressed in muscle tissues deficient in p65. Since IκBβ is constitutively phosphorylated [167, 170, 171], tissue homogenates were phosphatase treated to further ascertain the characteristics of these various IκBβreactive polypeptides. This treatment caused a shift in IκBβ-I and IV forms promoting the appearance of a slightly faster migrating IκBβ protein (Fig. 2.4C, denoted by asterisks). Based on these results, IκBβ-I is likely to represent the basally phosphorylated 45kD form of IκBβ that is most commonly described, and whose stimulus-dependent hypophosphorylated state is associated with persistent activation of NF-κB [167, 169]. Although IκBβ-IV is a minor component, the data also demonstrate that IκBβ can exist in a hyperphosphorylated state. In comparison, IκBβ-II and IκBβ-III were completely resistant to phosphatase treatment, which suggests that these proteins either represent unique forms of IκBβ devoid of phosphorylation, or are IκBβ-like polypeptides that may have cross reacted with this IκBβ carboxyl terminal specific antibody used in immunoblot and immunohistochemical analyses. To make this 37

54 distinction, immunoblots were repeated with a second antibody generated instead to the amino terminus of IκBβ (referred to as N20). Consistent with our previous results, the N20 antibody also reacted with IκBβ forms I and IV, thus validating the expression of these IκBβ forms in murine tissues (Fig. 2.4D). However, in contrast to the carboxyl terminal antibody, the N20 antibody was clearly reactive with IκBβ-I, but less so with IκBβ-IV, in skeletal muscle and heart (Fig. 2.4D). In addition, this IκBβ antibody again recognized altered IκBβ forms in skeletal muscle and heart, but these forms did not migrate to the same mobility as forms II and III. Collectively, these data imply that skeletal muscle and heart may be capable of synthesizing distinct IκBβ-like proteins. This same logic could apply to p65 expression in brain, where a polypeptide migrating with an approximate mobility of 65kD, as described above, was reproducibly detected in p65 null mice (Fig. 2.4A). In any regard, it is clear that a more detailed investigation of IκBβ in muscle tissues will be required to determine whether such forms derive from post-translational modifications or alternative splicing events. Absence of p65 promotes IκBβ degradation by the 26S proteasome independent of IKK and phospho-acceptor serines 19 and 23. Next, we examined the mechanism by which p65 regulates IκBβ. Although a consensus NF-κB binding site is contained within the IκBβ promoter, over expression of p65 has been shown to be incapable of stimulating IκBβ transcription [184], which argued as others have before [167] that IκBβ is not a transcriptional target of p65. In line with these findings, we too could not detect any significant difference in steady state levels of IκBβ mrna between 38

55 p65 +/+ or p65 -/- fibroblasts treated with TNF, while as expected, IκBα mrna was readily induced by this cytokine in a p65 dependent manner (Fig. 2.5A). The above findings suggested that p65 regulation of IκBβ was not transcriptionally mediated and therefore is likely to occur at the protein level, affecting either the synthetic rate or stability of IκBβ. To address these possibilities, p65 +/+ or p65 - /- fibroblasts were metabolically labeled with [ 35 S] and IκBβ was subsequently analyzed by immunoprecipitation. Results showed similar expression levels of IκBβ following 10, 30, or 60 minutes of labeling (Fig. 2.5B and data not shown), indicating that loss of p65 does not affect the rate of IκBβ synthesis. To examine IκBβ stability, fibroblasts were treated with cyclohexamide and IκBβ was analyzed over time. During an 8h period, little destabilization of IκBβ was observed in p65 +/+ cells, whereas in p65 -/- cells, nearly a third of the protein was degraded after only 30 min of treatment (Fig. 2.5C). Further treatment of fibroblasts with MG-132 to inhibit proteasome activity demonstrated little increase in IκBβ stability in p65 +/+ cells, while levels of IκBβ protein increased steadily over time in cells lacking p65 (Fig. 2.6A). Similar results were obtained with additional proteasome inhibitors, ALLN and lactacysteine (Fig. 2.6A), demonstrating that absence of p65 leads to IκBβ destabilization mediated by the 26S proteasome complex. To further investigate the mechanism of IκBβ turnover, we asked whether this regulation by the proteasome was also dependent on IKK activity and Ser-19 and Ser-23 that are phosphorylated in response to a classical NF-κB inducing signal. Interestingly, kinase assays revealed that p65 -/- MEFs exhibited substantially higher basal IKK activity compared to wild type cells (Fig. 2.6B). However, transient over-expression of 39

56 catalytically inactive IKKα and IKKβ subunits (Fig. 2.6C), or treatment with IKK inhibitor compounds (Bay and PS1145) (data not shown) did not restore IκBβ levels in p65 -/- MEFs (data not shown). In addition, transient expression of HA-tagged IκBβ proteins mutated at Ser-19 and Ser-23 to alanine (S19/23A) or glutamic acid (S19/23E) residues had no significant affect on either the basal level or turnover rate of IκBβ in p65 -/- cells (Fig. 2.6D). Therefore, despite elevated levels IKK activity in p65 -/- MEFs, IκBβ proteolysis in these cells does not appear to be regulated by the classical IKK signaling pathway. The carboxyl terminus of p65 is required for IκBβ stability. To address whether p65 is a direct regulator of IκBβ stability, p65 was reconstituted in null fibroblasts. Results showed that IκBβ expression was indeed restored in these cells (Fig. 2.7A). To further test the specificity of this regulation, we also examined the levels of κb-ras, a Ras-like small GTPase recently shown to directly bind IκBβ and contribute to its stabilization in response to an NF-κB inducing signal [174, 175]. We considered the possibility that loss of IκBβ could be mediated by the preceding destabilization of κb- Ras resulting from the absence of p65. Our findings revealed however that κb-ras expression was generally unaltered in either p65 -/- MEFs or p65 -/- tissues (Fig. 2.7B and 2.7C). Similar results were obtained with another recently identified IκBβ stabilizing protein, β-arrestin [187] (Fig. 2.7B). Together, these findings support that IκBβ stability is directly mediated by p65. Next, p65 deletion mutants were generated (Fig. 2.8A) and subsequently expressed in p65 -/- MEFs in order to map the region in p65 responsible for IκBβ stability. 40

57 Since IκBβ binding is known to occur through the RHD of NF-κB monomers, intuitively we did not consider the possibility that amino acids carboxyl to the NLS of p65 would contribute to this stability. Although deletion of the first 17 amino acids from the carboxyl terminus ( 534) restored IκBβ to equivalent levels as that of wild type p65 (FL), to our astonishment, further deletion of the TA1 domain ( 521) was sufficient to cause a minor but reproducible reduction of IκBβ (Fig. 2.8B). MG-132 treatment of p65 ( 521) expressing cells restored IκBβ to wild type levels, indicating that the reduction in IκBβ observed in p65 ( 521) cells was due to IκBβ destabilization (data not shown). Additional destabilization of IκBβ occurred when residues mapping to the second TA domain of p65 were removed ( 431), and still further loss of IκBβ was observed upon deletion of residues lying just proximal to the RHD ( 319). However, further deletion of residues to the NLS ( 313) reproducibly had little to no further affect on IκBβ stability (Fig. 2.8B). Moreover, reconstitution of p65 -/- MEFs with only the carboxyl terminal residues of p65 was unable to restore IκBβ expression over that of vector control cells (data not shown) demonstrating that both the RHD and carboxyl residues of p65 are critical to sustain IκBβ expression. Also intriguing was the observation that destabilization of IκBβ due to carboxyl terminal deletions of p65, led to the increased expression of a higher molecular weight form of IκBβ that appeared similar to the IκBβ-IV form that we had earlier identified in mouse tissues. Direct comparison of IκBβ forms from tissue and p65 reconstituted fibroblasts showed that the higher molecular weight form of IκBβ in fibroblasts expressing p65 ( 313) migrated to the same apparent molecular weight as IκBβ-IV from 41

58 brain and testis (Fig. 2.8C), and like IκBβ-IV in tissues, was sensitive to phosphatase treatment (Fig. 2.8D). In addition, in contrast to IκBβ-I, MG-132 treatment was unable to further increase the expression of IκBβ-IV (Fig. 2.8E), suggesting that this form of IκBβ is resistant to 26S proteasome activity. This result is consistent with our previous observation that IκBβ-IV regulation appeared to be less dependent on p65 compared to IκBβ-I (Fig. 2.4A) and was unchanged in p65 -/- MEFs treated with various proteasome inhibitors (Fig. 2.6D). Taken together, these results demonstrate that IκBβ dependence on p65 occurs due to the stabilization of IκBβ protein mediated largely through the p65 carboxyl terminus encompassing the TADs, and that this regulation is specific to IκBβ-I. Expression of IκBβ but not IκBα causes defects in cellular growth and survival. In the final analysis of this study, an attempt was made to understand the physiological relevance underlying the regulation of IκBβ by p65. Although it is widely believed that p65 is ubiquitously expressed, evidence suggest that p65 expression can be quite low or even undetectable during pre-gastrula development [178, 179] or in specific cell types in late embryogenesis [180]. Based on these findings, as well as our current data, we asked if there was a reason why cells would need to degrade IκBβ under conditions where p65 expression is either low or absent. To address this question, an HA-tagged IκBβ retrovirus was generated in order to infect p65 -/- MEFs. Attempts to stably express wild type IκBβ or a S19/23A mutant by this system or by conventional CMV expression plasmids proved unsuccessful. MG-132 treatment partially restored recombinant IκBβ expression demonstrating that the inability to express IκBβ in p65 -/- cells resulted from on-going proteolysis (data not shown). We were however able to 42

59 readily express a truncated form of IκBβ lacking the first 54 amino acids (Fig. 2.9A). This suggested that other determinants in the N-terminus aside from Ser-19 and Ser-23 are required to mediate IκBβ proteasome-mediated degradation in p65 -/- cells. We refer to this non-degradable mutant as an IκBβ super repressor (pbabeiκbβ-sr). To address whether IκBβ-SR was functional, p65 -/- vector or IκBβ-SR expressing cells were treated with TNFα and NF-κB activity was monitored by EMSA. Results showed that even in the absence of p65, an NF-κB complex was induced by TNFα that by supershift analysis was found to contain p50 and c-rel subunits (Fig. 2.9B). Expression of the IκBβ-SR transgene reduced this activation, which confirmed the inhibitory property of the IκBβ protein. But despite this function IκBβ-SR expression was not found to increase the incidence of TNFα mediated killing over that of vector control cells, which suggested that IκBβ destabilization in p65 -/- cells does not simply result to allow compensatory anti-apoptotic function from c-rel containing complexes. Clearly observable however was that p65 -/- MEFs expressing IκBβ-SR, while under puromycin selection, grew at a considerably slower rate than vector control cells (Fig. 2.9C). Although not as evident, this growth defect was maintained even in the absence of antibiotic treatment (data not shown). To determine whether this effect was dependent on p65, IκBβ-SR retroviral infections were repeated in both p65 +/+ and p65 -/- MEFs. Over time, a growth reduction was also observed in p65 +/+ MEFs, but not to the same extent as in p65 -/- cells (Fig. 2.9D). To further examine the specificity of this phenotype, p65 -/- MEFs were infected with viruses expressing non-degradable versions of either IκBα or IκBβ, and growth rates were monitored compared to vector control cells. 43

60 Results showed little growth difference between control and IκBα-SR expressing cells, while again IκBβ-SR cells exhibited a clear growth defect (Fig. 2.10A). To address whether this defect was related to viability, cells were stained with Annexin V to monitor for apoptosis. Results showed that levels of apoptosis were nearly equivalent between vector control and IκBα-SR expressing cells. In contrast, apoptosis was increased approximately 65% in cells expressing IκBβ-SR (Fig. 2.10B). Collectively, these data demonstrate that IκBβ expression is linked to cellular growth defects, which provides at least one rationale to explain why IκBβ downregulation would be required in p65 deficient cells. 2.4 Discussion In contrast to IκBα, much less is known regarding the regulation of IκBβ by NFκB. The present study was performed based on the observation that p65 -/- MEFs contained dramatically lower levels of IκBβ. Although our group is not the first to note such a phenomenon [27, 177], a comprehensive study as to how p65 regulates IκBβ had yet to be undertaken. We have now performed such an analysis, which we believe provides fresh insight into the specificity, mechanism, and biological significance of this regulation. One question we sought to address was the specificity of IκBβ regulation by p65. By using fibroblasts null for various components of NF-κB and its signaling pathway, we determined that regulation of IκBβ only occurred in cells lacking p65. By this genetic criterion, the data strongly support that IκBβ regulation is specific to this NF-κB subunit. 44

61 In addition to p65, IκBβ can also associate with c-rel. It thus remains possible that the inability to detect IκBβ regulation in c-rel -/- MEFs may be due to an under representation of c-rel in mouse fibroblasts. Nonetheless, to the best of our knowledge, regulation of IκBβ in tissues from c-rel -/- mice has not been reported, which further supports the specificity of p65 in this regulation. Our findings also revealed the differences in the degree to which IκBα and IκBβ expression is dependent on p65. IκBα is the prototypical IκB protein whose basal and stimulated expression is controlled by p50/p65 DNA binding sites within its promoter [56-58]. From this perspective, it stands to reason why IκBα has long been considered to be the most highly regulated IκB protein by NF-κB, specifically by the p65 subunit. Our current data however challenge this thinking. We found that IκBα levels were indeed downregulated in p65 -/- primary MEFs and fetal liver cells, but only to about 60% that of wild type cells (Fig. 1). Such data support the role of p65 in regulating IκBα transcription and/or protein stability [188], but it also highlights the requirement for other factors in this regulation. In comparison, IκBβ expression was almost completely absent in these same cells, accentuating this proteins dependency on p65. These data provide compelling evidence that of these two IκB proteins, IκBβ is the one most tightly regulated by the p65 subunit of NF-κB. Analysis into the mechanism of IκBβ downregulation revealed the regulation by the proteasome complex. Unlike stimuli-induced activation of NF-κB that requires IKK dependent-phosphorylation of serine residues and subsequent polyubiquitination in the N-terminus, we found that proteasome-mediated degradation of IκBβ was independent of 45

62 both IKK and serines 19 and 23. However, deletion of the first 54 amino acids stabilized IκBβ, suggesting that other determinants within the signal response element are required for IκBβ turnover in cells lacking p65. This regulation is highly reminiscent of the mechanism controlling IκBα degradation in response to UV treatment, which also requires the N-terminus but not IKK activity or phosphorylation of N-terminal serines [189, 190]. Similar to UV-induced degradation of IκBα [189], it remains to be seen whether polyubiquitination in the signal response element of IκBβ is critical for its decay in p65 -/- cells. The fact that the expression of p65 could completely restore IκBβ in p65 -/- MEFs, and that κb-ras and β-arrestin levels were unaffected in these cells suggested that p65 was a direct regulator of IκBβ stabilization (Fig. 5). Data derived from the IκBβ/p65 crystal structure has shown that much of the binding from the IκBβ inhibitor ankyrin repeats occurs in the RHD of p65, between residues [176]. Because these structures lacked the TADs of p65, it has not been possible to formally conclude whether residues in the carboxyl half of the RHD participate in IκBβ binding. To our surprise however, it is precisely this region of p65, between residues , that was found to be required for maximal IκBβ stability. Based on these findings we predict that IκBβ stabilization is directly regulated by contributions from both the RHD and the carboxyl terminal half of p65. This carboxyl terminus is likely to adopt a particular conformation when p65 is activated and involved in interactions with the basal transcription machinery and transcriptional co-activators in the nucleus. As an inactive cytoplasmic complex, we foresee that the carboxyl half takes on a different structure as a result of direct or indirect 46

63 contacts with IκBβ. Interestingly, S. Ghosh and colleagues had previously proposed that p65 bound to IκBα in the cytoplasm exists as an intramolecular structure, that when phosphorylated on serine 276 undergoes a conformational change that is necessary for nuclear CBP/p300 interaction and p65 TAD function [55]. In an analogous fashion, we reason that the carboxyl terminal half of p65 within this intramolecular structure plays a critical role in mediating IκBβ stability. Our results also revealed that destabilization of IκBβ due to removal of the p65 carboxyl half gave way to increased expression of an IκBβ variant which co-migrated with IκBβ-IV in tissues and existed in a hyperphosphorylated state (Fig. 5). Intriguingly, unlike IκBβ-I which was stabilized by proteasome inhibition in p65 -/- MEFs, IκBβ-IV levels did not increase under these same conditions, nor were levels stabilized by proteasome inhibition in p65 -/- cells expressing carboxyl deletion mutants of p65. This implicates that hyperphosphorylation of IκBβ-IV may be a contributing factor in the stability of this IκBβ form. While phosphorylation is generally considered to target proteins for degradation by the 26S proteasome, studies have also demonstrated the requirement of phosphorylation for protection from the proteasome [191]. So perhaps deletion of the p65 carboxyl terminus can lead to IκBβ hyperphosphorylation resulting in a protein now refractory to 26S proteasome activity. It will be interesting to ascertain in future studies whether stability associated with IκBβ-IV is a result of phosphorylation, and whether this minor form of IκBβ that exists in tissues possesses a functional activity distinct from the predominantly expressed IκBβ-I form. 47

64 In the final aspect of our study, we attempted to address the physiological significance underlying the tight control of IκBβ by p65. Our data revealed that p65 -/- MEFs stably expressing a non-degradable form of IκBβ exhibited an extreme growth defect that could at least partially be attributed to apoptosis (Fig. 6). A similar but less severe phenotype was seen in wild type cells, which argues that this defect is p65 dependent, but also reveals the general susceptibility of cells to IκBβ expression. Importantly, our data demonstrated that this effect on growth and survival was specific to IκBβ since a comparable phenotype was not observed in cells expressing a nondegradable form of IκBα. Taken together, these results provide rationale for why IκBβ may undergo such strong proteolysis in cells lacking p65, and the more pronounced regulation of IκBβ compared to IκBα. Our results indicate that even in p65 containing cells, proper regulation of IκBβ turnover is likely to be required to circumvent growth defects associated with stable expression of this protein. Although p65 is widely considered to be a constitutively expressed transcription factor, studies have shown that certain development stages early in embryogenesis [178, 179], or specific cell compartments of the thymus [180], do in fact possess low levels of p65. Such regulation of p65 has also been observed in selective neurons throughout development and adulthood (K. Pahan, personal communication). Based on our findings, we predict that it is precisely under such conditions that IκBβ proteolysis would be required in order to allow the proper development and maintenance of tissue homeostasis. 48

65 Figure 2.1: IκBβ is specifically regulated by p65, but does not require NF-κB activity. A. Whole cell extracts (WCEs) were prepared from immortalized MEFs either wild type or null for the p65 subunit of NF-κB, and the levels of the indicated proteins were analyzed by Western blot. B. Western blots to probe for IκBα and IκBβ in MEFs either wild type or null for various components in the NF-κB signaling pathway. Genotypes for all cell lines were confirmed by PCR analysis. C. WCEs were prepared from vector or IκBα-SR C2C12 myoblasts, and Westerns were performed to probe for IκB proteins. 49

66 Figure 2.2: IκBβ is regulated by p65 in primary fibroblasts. A. Primary MEFs were isolated from E13.5 day embryos and genotyped for p65 by PCR. WCEs were prepared from p65 wild type, heterozygous, and null MEFs and Western blots were performed to probe for p65, IκBα, and IκBβ. α-tubulin was used as a loading control. B. Quantitation of IκB proteins from two Western blots performed with WCEs obtained from two independent litters as described in (A). Levels of IκB proteins were normalized to α-tubulin, and compared to the expression of IκB proteins obtained from wild type cells, which was set to a value of 100%. 50

67 Figure 2.3: p65 regulation of IκBβ is not limited to fibroblasts. A. Histological H&E and immunohistochemical stains of p65 and IκBβ from longitudinal sections of p65 +/+ (a, b, c) and p65 -/- (d, e, f) embryos (1x magnification). B. Immunohistochemical stains of IκBβ of liver (a, d), lung (b, e), and brain (c, f) from p65 +/+ and p65 -/- embryos (4x magnification). C. Primary fetal liver cells were prepared from E13.5 embryos, and after genotypes were confirmed, Western blots were performed to probe for p65, IκBα, and IκBβ. α-tubulin was used as a loading control. D. p65 and IκBβ immunohistochemical stains from heart of p65 +/+ and p65 -/- embryos (4x magnification). E. p65 and IκBβ immunohistochemical staining of forelimb from p65 +/+ and p65 -/- embryos. To confirm skeletal muscle staining, serial sections of forelimbs were separately stained for myosin heavy chain (MyHC, arrowheads denote muscle tracks; 4x magnification). 51

68 Figure 2.4: p65 regulates various forms of IκBβ in postnatal development in a tissue specific manner. p65 +/- ;TNFα -/- mice were bred to generate p65 +/+ ;TNFα -/- and p65 -/- ;TNFα -/- progeny. A. At 4 weeks of age mice were sacrificed and tissue homogenates were prepared (skmc, skeletal muscle). Western blots were then performed to probe for p65, IκBα, IκBβ, and α-tubulin. Western is representative of a total of 4 different blots derived from 2 sets of littermates. B. Western blot to probe for IκBβ in p65 +/+ ;TNFα -/- brain, skin, skeletal muscle (skmc) and heart tissues. Arrows denote IκBβ forms I through IV. C. Similar extracts as in (A) were either untreated or treated with phosphatase enzyme and Western blot was performed to probe for IκBβ. Arrows denote phosphorylated forms of IκBβ and asterisk (*) denotes the shifted (de-phosphorylated) forms of IκBβ. D. The N- terminal IκBβ antibody (N20) was used in Western analysis to verify IκBβ forms in p65 +/+ ;TNFα -/- tissues. 52

69 Figure 2.5: IκBβ downregulation in p65 -/- MEFs occurs at the protein level. A. MEFs wild type and null for p65 were treated with TNFα and at indicated time points Real Time PCR was used to measure IκBα and IκBβ RNA. B. p65 +/+ and p65 -/- MEFs were incubated for 2 hours in methionine and cysteine free DMEM medium, and subsequently labeled with [ 35 S] methionine/cysteine mix for an additional hour. WCE were prepared and IκBβ was immunoprecipitated with either IgG (control) or with an IκBβ specific antibody. Complexes were resolved by SDS-PAGE and visualized by exposing dried gels to film for up to three days (arrow denotes IκBβ). C. p65 +/+ and p65 -/- MEFs were treated with cyclohexamide (10 µg/ml), and at indicated times WCE were prepared and Westerns performed to probe for IκBβ. Levels of IκBβ were quantitated from an average of three Western blots. 53

70 Figure 2.6: IκBβ downregulation in p65 -/- MEFs is regulated by the proteasome independent of classical IKK signaling. A. Similar conditions were used to treat p65 +/+ and p65 -/- MEFs with MG-132, ALLN, and lactacysteine (all at 10 µm). B. Kinase assays (KA) were performed with IKKγ immunoprecipitates from p65 +/+ and p65 -/- MEFs incubated with GST-IκBα as a substrate. WCEs from p65 +/+ and p65 -/- MEFs used in kinase assay were analyzed by Western blots to probe for IKKα and p65. C. Western blots to probe for IκBβ and tubulin in p65 -/- MEFs transfected with vector control, and catalytically inactive IKKα or IKKβ proteins (upper panel). Inhibitory activity of IKK proteins was verified by luciferase NF-κB reporter assays (lower panel). D. p65 -/- MEFs were transfected with either HA-tagged wild type IκBβ or HA-tagged IκBβ mutated at serines 19 and 23 to either alanine (S19/23A) or glutamic acid (S19/23E) residues. After 24 hours, cells were treated with cyclohexamide (10 µg/ml) for indicated times. WCE were then prepared and Westerns performed to probe for HA and α-tubulin. 54

71 Figure 2.7: IκBβ stability is directly regulated by p65. A. p65 -/- MEFs were infected with either vector control retrovirus vector or virus expressing wild type p65. WCE were prepared from a mixed population of drug selection resistant cells, and Westerns were performed to probe for p65 and IκBβ. B. Western blots to probe for κb-ras and β-arrestin in p65 +/+ and p65 -/- MEFs. α-tubulin was used as a loading control. C. Western blots were performed with p65 +/+ ;TNFα -/- and p65 -/- ;TNFα -/- tissue homogenates to probe for κb-ras. 55

72 Figure 2.8: IκBβ stability is regulated by the carboxyl terminus of p65. A. Illustration of full length and C-terminal truncation mutants generated in p65. The mutants are named according to the fragment of p65 expressed. For example, p65( 534) denotes p65 containing amino acids B. Western blots for IκBβ (upper panel) and p65 (lower panel) in p65 -/- MEFs transfected with the indicated p65 truncation mutants. C. Western blot for IκBβ in p65( 313) expressing cells, brain, skeletal muscle (skmc), heart and testis from p65 +/+ ;TNFα -/- mouse. Arrows denote forms IκBβ-I and IκBβ-IV. D. Identical lysates as described in (C) were either treated or not treated with phosphatase, and Western blots were subsequently performed to probe for IκBβ. E. p65 - /- MEFs were transfected with p65( 431) or p65( 313) and subsequently treated or not treated with MG-132. Western blot was then performed to probe for IκBβ. 56

73 Figure 2.9: MEFs stably expressing IκBβ exhibit a growth defect. p65 -/- MEFs were infected with either a pbabe vector retrovirus or virus expressing a degradation resistant form of IκBβ tagged with an HA-epitope (pbabeiκbβ-sr). A. Extracts from p65 -/- expressing vector or HA-IκBβ-SR cells were prepared and Westerns were performed to probe for HA and α-tubulin. B. p65 -/- cells expressing either vector control or IκBβ-SR were treated with TNFα, and at the indicated times nuclear extracts were prepared for EMSA. Supershift EMSA was performed with antibodies against p50, p65 or c-rel to confirm the composition of NF-κB complexes (arrowheads denote supershifted subunits). C. p65 -/- MEFs were infected with vector control or IκBβ-SR expressing virus. Following 4 days of cell expansion under 1µg puromycin selection, cell number was determined in both cell lines. Cell number was normalized to vector control cells which was set to a value of 100% growing cells. D. Growth curves in p65 +/+ or p65 -/- cells expressing either vector control or IκBβ-SR. 57

74 Figure 2.10: Growth defect in MEFs stably expressing IκBβ due in part to increased apoptosis. A. Growth curves identical to those performed in (2.9D) from p65 -/- cells infected with vector control, IκBα-SR, or IκBβ-SR pbabe retroviruses. Cell number was normalized to vector which was set to a value of 100%. F. Cells infected as in (A) were expanded in puromycin selection for 2 days and subsequently stained for Annexin V. 58

75 CHAPTER 3 DESIGN OF NF-κB CUSTOM MICROARRAY 3.1 Review of Microarray Technology The development of DNA microarrays has facilitated the study of gene expression, and increased understanding as to exactly how, when and where genes are regulated. The concept of microarrays, or DNA chips, was developed in 1989 by a team of scientists at the Affamax Research Institute lead by Stephen Fodor [192]. In 1993, in order to mass-produce these chips, Dr. Fodor co-founded the company Affymetrix, which today is one of the leading producers. There are two major platforms commonly used for DNA microarray analysis: oligonucleotide arrays such as those produced by Affymetrix, and complementary DNA (cdna) spotted arrays. Oligonucleotide arrays consist of short oligos (20-25 bp), and are built directly onto the glass substrate using a process called photolithography, or produced by piezoelectric deposition. Because of the small size of the oligos, each gene on the array is represented multiple times, using sequences complementary to various regions throughout the mrna transcript. Alternatively, cdna arrays consist of significantly larger oligos (up to a few thousand bp) and are printed intact onto the substrate. 59

76 The first evidence of microarrays being used to measure gene expression was in 1995 [193]. These early arrays consisted of cdnas corresponding to 45 genes in the Arabidopsis genome spotted by robot onto a glass substrate. With advances in the technology of how the genes are spotted, arrays now contain as much as genes on one chip, and sets of arrays can be used to examine entire genomes. Affymetrix started production of their oligonucleotide array, termed GeneChip, in The arrays available from Affymetrix and several other companies such as GE Healthcare (Piscataway, NJ) or Aglient (Palo Alto, CA) are the oligonucletide array format. These types of arrays are often called single channel arrays, in that only one sample can be hybridized per array. In order to evaluate the difference between samples, multiple chips are hybridized with a single sample and then chips are compared to one another. In contrast to single channel arrays, dual channel arrays (such as spotted cdna arrays) utilize a 2-color fluorescent hybridization technique to determine the level of mrna transcripts in different samples by competitive hybridization on a single array. Eppendorf (Westbury, NY) supplies dual platform arrays of this type. In addition to the glass slide microarray technology, companies have developed systems in which cdna or PCR products are spotted onto nylon membranes (SuperArray, Frederick MD). Most of these arrays are designed in a 96 well plate format, and due to the small number of genes available on each array they are often specialized and contain only genes pertaining to a particular signaling pathway or disease. The advantage of arrays of this type is that the data is more manageable given the small number of genes, and the cost is significantly lower due the difference in the technology 60

77 from the glass slide microarrays. However, the disadvantage is that systems of this type are not as high throughput as the higher density glass arrays. Since the introduction of cdna and oligo arrays for analysis of gene expression, the technology has been expanded to address other biological questions as well. Genotyping arrays are available that can be used to detect single nucleotide polymorphisms (SNPs) in populations in order to determine if a particular mutation is indicative of risk for diseases such as cancer, diabetes, cardiovascular disease and others. SNP arrays can also be used to determine if a particular polymorphism or mutation confers more or less susceptibility to a particular drug treatment for a disease, allowing more personalized drug protocols for individuals based on their genetic composition. These types of arrays are also useful diagnostic tools as well, given that they can determine gene copy number, and risk for some diseases is known to involve loss of heterozygosity (LOH) at certain gene loci. For example, higher risk for Alzheimer s disease can be attributed to loss of one allele of Apolipoprotein E4 (ApoE4) [194, 195]. In addition to DNA based arrays, protein arrays are also now available. These arrays are used to study protein-protein interactions, protein-small molecule interactions and kinase-substrate interactions. In comparative protein arrays protein lysates from two samples are co-hybridized to arrays containing various antibodies immobilized on the glass support. These types of arrays are used to determine for example differences in cancerous versus normal tissue. Also, using a concept similar to SNP genotyping arrays, tissue arrays are used to profile histological changes in disease in a high throughput manner. In these types of arrays, multiple core samples (as small as 0.6 mm) from 61

78 various normal or disease tissue are paraffin embedded in a array pattern, sectioned, and placed on glass slides for analysis. Commercial microarrays have certainly advanced the analysis of target genes for transcriptional complexes such as NF-κB. However, a majority of the genes present on available arrays are not NF-κB dependent, and serve as extraneous data compounding the data analysis. Some studies have addressed the potential difficulty of microarray data analysis and the problem of managing such large amounts of data [196, 197]. And while some companies do provide specific custom arrays containing a smaller set of genes related to only one pathway, the target genes on these arrays still might not be ideal in some situations. SuperArray has an NF-κB custom array available containing 96 genes related to the NF-κB family. However many of these genes correspond to the NF-κB family members themselves, and only a small portion are actual target genes of this transcription factor. Generally it is the activity of the NF-κB complex, and not the level of the proteins, that influences NF-κB controlled processes. Therefore, it is more informative to assay for changes in target genes, rather than changes in NFκB family members, and arrays of this type, while useful, are not necessarily the most efficient way to assay for changes in NF-κB target gene expression. The analysis of NF-κB transcriptional activity is compounded by the fact that this complex is activated by many stimuli, at several different timepoints. In addition, there are many post-translational modifications that occur on NFκB which can modulate its activity. The studies outlined in this dissertation are designed to determine how different post-translational modifications regulate NF-κB. To study NF-κB under the many 62

79 different activating conditions would be rather complex and cost prohibitive using the currently available arrays. Therefore, in order to carry out the studies described in chapter 4 of this dissertation, we designed a customized array containing a small set of NF-κB targets. 3.2 Design of Custom Oligonucleotide Microarray In order to select which NF-κB target genes to include on the custom array, Affymetix arrays were used as an initial screening tool. RNA was isolated from p65 +/+ cells treated with TNF, and p65 -/- cells left unstimulated. These conditions were used to identify TNF induced, p65 dependent genes. Affymetrix analysis revealed approximately 850 genes that were induced by at least a factor of 2-fold in the p65 +/+ TNF 1hr cells versus the p65 -/- unstimulated cells. This list was further narrowed based on published literature, and the final list contained 50 known NF-κB regulated genes (see Table 3.1). 70-mer oligonucleotides corresponding to these genes were designed and synthesized by Operon (Huntsville, AL). The oligos were spotted onto glass slides obtained from Corning (Corning, NY). There are several types of slide surface chemistries available to spot oligos. Glass slides coated with epoxide facilitates covalent attachment of amine modified oligos. Slides may also be coated with gamma amino propyl silane (GAPS). This coating gives the slide an overall positive charge allowing ionic attachment of the oligo to the substrate. Another method to coat slides is using poly-lysine. The advantage to this method is that it can be performed in the lab so the cost is reduced because the slides do not need to be ordered from a company. However poly-lysine coating can disintegrate, leaving cracks in the surface that trap the fluorescent label during the 63

80 hybridization step and causing streaks on the slide. Slides coated with GAPS yielded the most reproducible results, and were therefore used for the arrays designed in this study (representative image of hybridized microarray shown in Figure 3.1). Lyophilized oligos were resuspended in water to obtain a stock solution (50 mm), and a working concentration for spotting (40 µm) was prepared in a 96 well plates in 3X SSC buffer (1X is 0.15M NaCl, 0.015M sodium citrate, ph 7.0). Slides were spotted in a dust free, humidity controlled environment using a Promedia Associate Microarray Printer PA-MP2002, and each oligo was spotted 8 times for reproducibility. After spotting, the slides were prepared for hybridization. The slides, DNA side down, were held over steam, and crosslinked at 600 mj. The slides were then again held over steam and dried on an 80 o C heat block. Slides were placed in blocking solution (succinic anhydride, n-methyl-pyrrilidone, NaBorate) for 15 minutes followed by the following washes: 0.1% SDS for 20 seconds, room temperature water for 20 seconds, 95 o C water for 4 minutes, and cold 100% ethanol for 1 minute. Slides were then dried and stored at room temperature until ready for hybridization. The slides were hybridized with Cy3 and Cy5 labeled cdna. The advantage to these oligonucleotide arrays is that the use of two fluorochromes allows for competitive hybridization on a single slide. 20 µg of RNA prepared in Trizol reagent was reverse transcribed in the presence of Cy3 or Cy5-dUTP (the two samples being compared on the array were labeled with different fluorochromes, for example, sample 1 - Cy3 and sample 2 - Cy5). The two reactions were pooled, ethanol precipitated, and resuspended in hybridization solution (SSPE, Denhardt s solution, formamide, SDS). The arrays were hybridized overnight (16-20 hours) at 50 o C followed by washes: brief dip in 1X 64

81 SSX/0.1% SDS, 10 minutes in 0.2X SSC/0.1% SDS, brief dip in 0.2X SSC, and 10 minutes in 0.2X SSC. The arrays were then dried and scanned using the Affymetrix Scanner GMS418. Fluorescence data was analyzed using Genepix software (Molecular Devices, Union City, CA). Cy5/Cy3 determines the fold change of sample 1 versus sample 2 (diagram Figure 3.2). Hybridization quality was determined by the reproducibility (p-value) among the 8 replicate spots on the slide. The fold change and p- values for a representative experiment comparing vector control and wild type reconstituted p65 -/- MEFs is shown in Table 3.2. The results from the custom oligonucleotide array were reproducible as determined by p-value among the replicates for a single gene, and correlated well with the data from Affymetrix arrays (Figure 3.3). However the cost of the Cy3 and Cy5 fluorescent dyes were still slightly prohibitive for the number of conditions we wished to investigate, and we were still dependent upon the microarray core facility to hybridize and scan arrays. In addition, while the two-channel competitive hybridization system is desirable because it allows for direct comparison, it also required a control or baseline sample to be prepared for each experiment, which increased the cost. A one-channel system would allow us to compare several different mutants or conditions to a single control array. We therefore developed a radioactivity-based system, and arrays that we were able to spot ourselves without the dependence on a core. The design of this custom array is described in detail in the following section. 65

82 3.3 Design of PCR Spotted Array The design of the spotted array was based on a publication describing CpG island arrays designed to study differential methylation of promoters [198]. The clones for spotting were designed by amplifying base pair products from cdna prepared from mouse muscle, liver or heart tissue using specific primers for the NF-κB target sequences. Due to the amount of labor required to prepare each clone, the initial number of genes included on the array was reduced from the 50 present on the oligonucleotide array. We chose 25 genes that responded well on the oligonucleotide array, and these genes along with the primers used to amplify them are listed in Table 3.3. Restriction sites were added on the 5` and 3` ends allowing cloning into the psk+ vector. 100 ng of Qiagen purified plasmid DNA was then used as template in a primary PCR reaction using the T3 and T7 primers in psk+ to amplify the NF-κB target genes. 0.5 µl of this primary PCR was used as template in a second round of amplification to generate product for spotting. Each gene was spotted in total nine times for reproducibility, and replicates were positioned in different regions throughout the array. The PCR was performed in a 96-well plate format, so that it may be spotted directly on the membrane. Control (actin and GAPDH) signals were very strong and were therefore diluted before spotting (Figure 3.4). A 96 metal pin replicator and guide system (V & P Scientific, San Diego, CA) was used to spot the product onto charged nylon membrane (Biodyne, Gelman Corporation, Ann Arbor MI). After spotting, arrays were denatured for 1 minute (0.5 M NaOH, 1.5 M NaCl ph 10-14), neutralized for 3 minutes (0.5 M Tris-HCl, 1.5 M NaCl ph 7-8), and washed for 3 minutes (5X SSC). The blots were crosslinked at 1200 mj, dried, and stored at 20 o C until ready for hybridization. 66

83 The arrays were pre-hybridized for 2 hours at 42 o C prior to addition of the probe. The probe was generated by reverse transcription using superscript II (Invitrogen) and [α- 32 P]dCTP. This radioactive cdna probe was then added to the hybridization solution on the arrays and hybridized overnight (16 hours) at 42 o C. Arrays were washed twice in 2X SSC/0.1% SDS at room temperature, twice in 0.1X SSC/0.1% SDS twice at 42 o C, and exposed overnight on a screen. Exposure and Quantitation was completed using the Typhoon phosphoimager and ImageQuant (GE Healthcare, Piscataway, NJ). Like the oligonucleotide arrays, reproducibility of the custom nylon arrays was verified by p-value (representative experiment shown in Table 3.4). The implementation of this system, and how it is used to investigate NF-κB mediated transactivation is discussed further in chapter 4. 67

84 Name GeneID 70-mer Oligo Sequence actin GGATTTAAAAACTGGAACGGTGAAGGCGACAGCAGTTGGTTGGAGCAAACATCCCCCAAAGTTCTACAA GAPDH TGGTCACCAGGGCTGCCATTTGCAGTGGCAAAGTGGAGATTGTTGCCATCAACGACCCCTTCATTGACC A1/Bfl TCTGAGCTCATGCATATCCACTCCCTGGCTGAGCACTACCTTCAGTATGTGCTACAGGTACCCGCCTTTG A TAAACAGCGCTGCCGGGCCCCTGCTTGTGATCACTTTGGCAATGCCAAGTGTAATGGTTACTGCAATGA Bcl ACCTGGATCCAGGATAACGGAGGCTGGGATGCCTTTGTGGAACTATATGGCCCCAGCATGCGACCTCTG BclX-L GGACTCCGGAATGTCAAGAATTTAGTTGTCCATAGGCTGGCCTAGGGCGGGGCAAGCACTTTGCTTTAC C ATGGGATGCTCAGCAAGCTGTGCCACAGTGAAATGTGCCGGTGTGCTGAAGAGAACTGCTTCATGCAAC Ccl CACAACCACCTCAAGCACTTCTGTAGGAGTGACCAGTGTGACAGTGAACTAGTGTGACTCGGACTGTGA CD TATACAATGGCATCTCAGAAACTCTAGCAGGTGGGGCAGAAAACAGGTAGTGGAATGATGGGTAGAGAA c-flip GCTTCGGTGAAAGGAGAATGAGCCCTACTCCTTGAAAGGTTGTAGTGCTTGGGAGAGCAGTCTGTACCT ciap ATCAAGGGGACTGTGCGCACATTTCTCTCATGAGTGAAGAATGGTCTGAAAGTATTGTTGGACATCAGA c-myc TCAGAGGAGGAACGAGCTGAAGCGCAGCTTTTTTGCCCTGCGTGACCAGATCCCTGAATTGGAAAACAA Cp AGAAGGTTGTGTATCGCCAGTTTACTGACAGCTCATTCAGAGAACAGGTGAAGAGACGAGCCGAAGAAG Cyclin D TCTGTTATCGATATTGTTACTTGTAGCGGCCTGTTGTGCATGCCACCATGCTGCTGGCCCGGAGGGATT Cyclin D AACTGGAAGCGTGGGAGCAGCCATCTGTGGGCTTCAGCAGGATGATGAAGTGAACACACTCACGTGTGA E2F TGTCCCGGGAATGAAGGTGAACACATCTGTATGTGTGCTGCAGACACATCCTGGTGTGTCCACATGTGT FasL GAGGTTGAACTACTGCACTACTGGACAGATATGGGCCCACAGCAGCTACCTGGGGGCAGTATTCAATCT FasR GCTGCTCCTGTGCTGGTACCAATCTCATGGGAAAGTGATGCATATCAAGATTTAATCAAGGGTCTCAA Gadd45β GAAAAGCCAAGGCTTGGTGGAGGTGGCCAGTTACTGTGAAGAGAGCAGAGGCAATAACCAATGGGTCCC Gdnf TTTATGTAGAAGTTAATGTATCACTCCCTTTGTGGCTGCTATCCCCCGCCCCCAGAGTAGGACAGATAG GM-CSF GAACCTCCTGGATGACATGCCTGTCACATTGAATGAAGAGGTAGAAGTCGTCTCTAACGAGTTCTCCTT Groα GAGACCACTAAGTGTCAACCACTGTGCTAGTAGAAGGGTGTTGTGCGAAAAGAAGTGCAGAGAGATAGA Gstt GACTGTCCCCCTGCTGACCTCATCATAAAGCAGAAGCTGATGCCCAGAGTGCTGACCATGATCCAGTGA ICAM GTCATGTCTGGACATGAGTGCCCAGGGAATATGCCCAAGCTATGCCTTGTCCTCTTGTCCTGTTTGCAT IEX-IL ACCATGGATGGGTACCTGGTGCGAGAGAACGTATCCCAAACTGGGATTTCTAAGGCAACGCTAACTCAG IκBα GTTGAAATGTGGGGCTGATGTCAACAGGGTAACCTACCAAGGCTACTCCCCCTACCAGCTTACCTGGGG IL GGAGACTTCACAGAGGATACCACTCCCAACAGACCTGTCTATACCACTTCACAAGTCGGAGGCTTAATT INFβ CAAAGGTACCTTAAACTCATGAAGTACAACAGCTACGCCTGGATGGTGGTCCGAGCAGAGATCTTCAGG inos AGAGCCAGAAACGTTATCATGAAGATATCTTCGGTGCAGTCTTTTCCTATGGGGCAAAAAAGGGCAGCG IP TGGTCACATCAGCTGCTACTCCTCCTGCAGGATGATGGTCAAGCCATGGTCCTGAGACAAAAGTAACTG JUN B CTGCAAAGAGACTGAATTCATATTGAATATAATATATTTGTGTATTTAACAGGGAGGGGAGAAGGGGGC MHC-class CTCTAGCATGAAGACAGCTGCCGGGTGTGGACTTGGTGACAATGTCTTCTCATATCTCCTGTGACATCCA MIP GATTTCAATGTAATGTTGTGAGTAACCCTTGGACATTTTATGTCTTCCTCGTAAGGCACAGTGCCTTGC Mmp AGATCCAGCTAAGACACAGCAAGCCAGAATAAAGACTGTGCCAGCTGGTCAGTCGCCCTTTTGAGACCA Mmp CATCCCCTGATGTCCTCGTGGTACCCACCAAGTCTAACTCTCTGGAACCTGAGACATCACCAATGTGCA Mmp GCTTCCCTCTGAATAAAGACGACATAGACGGCATCCAGTATCTGATGGTCGTGGCTCTAAGCCTGACC MnSOD GGGGACATATTAATCACACCATTTTCTGGACAAACCTGAGCCCTAAGGGTGGTGGAGAACCTAAAGGAG p ACAGAGTCTCTTACTGGAGACAGCCCACTGCTATCTCTGAACAAAATGCCCCACGGTTATGGGCAGGAA P ATGGGGACCCCAGTTGGGGTTCTCAGTGACTTCTCCCATTTCTTAGTAGCAGTTGTACAAGGAGCCAGG Ptx GCTACCACTGTAGAGATGGCCAAAAGTCACTCTGTTCCTGAGGGTGGACTCCTACAGATTGGCCAAGAA RANTES GCTTGCTCTAGTCCTAGCCAGCTTGGGGATGCCACTCCAGTAATCCCCTACTCCCACTCCGGTCCTGGG Saa TATTAGTTCAGAAGGCTGTGTTGGGGTCCTGAGGGTGGGGTCTGGGCTTCCTATCTAGGAACACTGAAG Saa CAGCCAAAGATGGGTCCAGTTCATGAAAGAAGCTGGTCAAGGGTCTAGAGACATGTGGCGAGCCTACTC Sele ACCAGAATGACTCAGTTTCCTCTTTCCTAGTTAGTGACAGTATGAAGAACATGCTGCCACAGAAGGTGG Serpine CAGCAACAACTGCAATCCTAATTGCAAGGTCATCACCTCCCTGGTTTATAGTAGACAGGCCTTTCCTGT TNFα CATCAAGGACTCAAATGGGCTTTCCGAATTCACTGGAGCCTCGAATGTCCATTCCTGAGTTCTGCAAAG TRAF GGTCAGGGCATAACTGGAAAAATGCCCCCATCTCTCTGTTCAGACTCAAAACTAGAACCACAGGGCAGA VCAM TTCTCTACATGGTACTGTACGGTACGGGGACCTGTTTCTTCATTGGGGTCTTGTGTAATGACTAACCTG VEGF GCAGACGCATTCCCGGGCAGGTGACCAAGCACGTGCCTCGTGGGACTGGATTCGCCATTTTCTTATATC xiap CATCCTGCTGTTTCCAAATGGAGACCAATGCTAACAGCACTGTTTCCGTCTAAACATTCAATTTCTGGA Table 3.1: Genes on custom oligonucleotide microarray. 68

85 Figure 3.1: Image of custom oligonucleotide microarray. 69

86 Figure 3.2: Diagram of experimental design for microarray hybridization. 70

87 Name Fold Induction p-value Name Fold Induction p-value actin IκBα GAPDH IL A1/Bfl IFNβ A inos Bcl IP Bclx-L JUN B C MHC-class Ccl MIP Cyclin D Mmp Cyclin D Mmp CD Mmp c-flip Mn SOD ciap p c-myc P Cp Ptx E2F RANTES FasL Saa FasR Saa Gadd45β Sele Gdnf Serpine GM-CSF TNFα Gro TRAF Gstt VCAM Icam VEGF IEX-IL xiap Table 3.2: p-values for custom oligonucleotide array. Microarray was co-hybridized with vector control (Cy3-labeled) and p65wt (Cy5- labeled) cdna. Fold induction was determined by the median ratio (Cy5/Cy3) of the 12 replicates spots for each gene. p-value was determined using Student s T-test (p-values < 0.05 considered significant). 71

88 Figure 3.3: Comparison of Affymetrix microarray with custom oligonucleotide array. 72

89 Gene GeneID Fragment Size 5` Primer 3` Primer 18S CAATACAGGACTCTTTCGAGGCC GATCGGATTCGTTGAGTCAAATTAAGCCGCAGGC actin GTTCGCCATGGATGACGATATCG GCTCGTTGCCAATAGTGATGACC GAPDH GCAAATTCAACGGCACAGTCAAGG GATCGGATCCCATACTTGGCAGGTTTCTCCAGG A GCCCAGGACTGTTACAGATATCC GATCGGATCCCTCCTGCACTTCATTGCAGTGG C CCATCAAGATTCCAGCCAGTAAGG GATCGGATCCCCACACCATCCTCAATCACTACG Ccl CGGAATTCACCAGCCAACTCTCACTGAAGC CGGGATCCCTGGTCACTCCTACAGAAGTGC CD CGGAATTCTATGGGGCTGCTTGTTGACAGC CGGGATCCGCGACTCTCTTTACCATCCTCC Cp GATCGGATCCCTGTCTACCTTGGAGAAAGGACC GTGCATTGTGAGGCCTTGTAGG Fas CGGAATTCCACTCTGCGATGAAGAGCATGG CGGGATCCCTCTTCATGGCTGGAACTGAGG GM-CSF GATGGATCCCTAAGGTCCTGAGGAGGATGTGG ATCGAATTCTCGTTTGTCTTCCGCTGTCCAAGC Groα CGGAATTCGCACCCAAACCGAAGTCATAGC CGGGATCCGTCCTTTGAACGTCTCTGTCCC ICAM GATGAATTCCCCCAAGGAGATCACATTCACG CTTCGTTTGTGATCCTCCGAGC IEX-1L CGGAATTCCAACCGAGGAACCCAACATTGC CGGGATCCGATTCCCAAGGGTCTTCCAAAGC IκBα GATGGATCCGGAGACTGCTTCCTGCACTTGG CCTCTGTGAATTCTGACTCCGTG IL GATCGGATCCCCGCTATGAAGTTCCTCTCTGC GATCGGATTCGCATAACGCACTAGGTTTGCCG inos GGAATGCCCCTCGCTGCATCG ATGAAGCTTGGAGCACAGCCACATTGATCTCC IP GATCGGATCCCACGTGTTGAGATCATTGCCACG GATCGATATCCTTGATAACCCCTTGGGAAGATGG JunB CGGAATTCGACGACCTGCACAAGATGAACC CGGGATCCGACATGGGTCATGACCTTCTGC MHC CGGAATTCCGATGAGGTATTTCGAGACCGC CGGGATCCGTGATGTCAGCAGGGTAGAAGC MIP GATCGGATCCGAACAAAGGCAAGGCTAACTGACC GATCGAATTCGTGTCCACTTCAAGACACGAAAAGG Mmp GATCGGATCCCAAAACACCAGAGAAGTGTGACCC CTCGTTCACAGAACTAAGCTCTCC Mmp GGAAATCCCACATCACCTACAGG GATCGGATTCCCTCATATGCAGCATCCATGTTGG Mmp CTGGTGATCTCTTCTAGAGACTGG GATCGGATTCCACAGTCTGACCTGAACCATAACG Ptx GCAATTCTCTTCCCAATGCGTTCG ATCGAATTCCTCCTTCATTCGTCTATTACGCACC Saa GATGATATCGATCTGCCCAGGAGACAGGAGC ATCGGATCCGTTCCTGTTTATTACCCTCTCCTCC Saa GATGAATTGGAGCTCGCAGCACGAGCAGG GCACATTGGGATGTTTAGGGATCC Sele GTGGTTGAATGTGAAGCTTTGACCC GAAAGGCACATGAGGACTTGTAGG VCAM GGATACTGTTTGCAGTCTCTCAAGC ATCGGATTCTTGTGGAGGGATGTACAGAGATCG Table 3.3: Genes on custom spotted nylon microarray. 73

90 actin GAPDH undiluted 1:10 1:20 1:50 1:100 1:500 Figure 3.4: Dilution of controls for custom spotted nylon microarray. 74

91 Gene Fold Induction p-value 18S actin GAPDH A C Ccl CD Cp Fas GMCSF Groa ICAM IEX-1L IkBa IL inos IP JunB MHC MIP Mmp Mmp Mmp Ptx Saa Saa Sele VCAM Table 3.4: p-values for replicate spots on custom spotted nylon microarray. Arrays were hybridized with vector control or p65wt cdna, and fold induction was determined by comparison of the two arrays. p-value was determined using Student s T- test (p-values < 0.05 considered significant). 75

92 CHAPTER 4 PHOSPHORYLATION OF p65 AND DIFFERENTIAL REGULATION OF GENE EXPRESSION 4.1 Introduction The NF-κB family of transcription factors, discussed extensively in chapter 1, is involved in various cellular processes such as immune response, cellular proliferation, differentiation, and apoptosis [39-42]. Given that constitutive NF-κB has been implicated in several diseases, activation is controlled by a negative feedback mechanism to ensure that the signaling remains transient. This is accomplished in part by NF-κB regulation of its own inhibitor, IκBα [56-58]. IκBα is then able to enter the nucleus and remove NF-κB complexes from the DNA and export them back to the cytoplasm [59, 61-63]. Similarly, NF-κB also regulates the transcription of A20, which is able to disrupt formation of the TNFR complex, and therefore terminate NF-κB signaling [66, 67]. In addition to transcriptional feedback mechanisms, NF-κB activity is also regulated by post-translational modification. crel, RelB, p65, IκBα, and p105 are all phosphorylated in response to activating stimuli, (reviewed in [199]), however it is phosphorylation of p65 that has been the most extensively studied. Particularly phosphorylation of serine residues 276, 529 and 536 are necessary for full transcriptional activity of NF-κB. In response to LPS stimulation, phosphorylation of p65 at serine

93 causes a conformational change that is important to allow binding of CBP/p300 [55, 69]. This phosphorylation dependent recruitment of CBP displaces HDAC1 [24], which is previously shown to bind p65 and negatively regulate NF-κB target genes [157]. Serine 276 can also be phosphorylated in response to TNFα by MSK1, which is necessary for p65 mediated transcription of target gene IL-6, but not required for nfkb2/p100 [70]. Serine 529 is phosphorylated by casein kinase II in response to TNFα, and mutation of this serine impairs activation of an NF-κB promoter construct [71, 83]. TNFα also induces IKKβ to phosphorylate p65 at serine 536 [86], while IKKα is required for phosphorylation of p65 at this same residue in response to TAX protein [73] or lymphotoxin-β receptor [85]. Serine 536 is also phosphorylated in response to IL-1β or DNA damage by various other kinases [84, ]. All of these studies demonstrate by mutational analysis and reporter assays that phosphorylation at serine 536 is necessary for full transcriptional activity of NF-κB, however little is known as to how these phosphorylation events regulate endogenous gene expression. Another post-translational modification of p65 that has been found to be important for NF-κB transcriptional activity is acetylation, and in most cases phosphorylation is a prerequisite for this event. For example, mutation of serine 536 to alanine impairs binding to p300 and subsequently reduces p65 acetylation [204]. Another study shows that p65 acetylation is reduced in cells expressing mutated forms of p65 that are unable to be phosphorylated at serines 276 or 536 [103]. Both of these reports show that acetylation mediated through p65 phosphorylation is necessary for transcriptional activity of NF-κB in response to TNFα. 77

94 The evidence provided thus far has demonstrated the requirement for p65 serine phosphorylation in NF-κB mediated transcription. However, it is possible that phosphorylation of these serines may not be necessary on a general transcription level, but rather play an important role in gene specific regulation. While some studies have demonstrated a gene specific effect by real time PCR, a large-scale analysis has yet to be undertaken. This type of analysis is complicated by the fact that NF-κB regulates hundreds of target genes, and to study these targets on a single gene level by real time or Northern analysis would be labor intensive. In addition, there are many different conditions to be considered in terms of NF-κB activation. The expression of a single gene likely depends on the activating signal, the kinase it activates, and the specific serine which is phosphorylated. This is further complicated by the fact that the expression of a gene under identical conditions may depend on the cell type, or the phase of NF-κB activation [168, 205, 206]. Therefore determining how p65 phosphorylation mediates differential expression has been a daunting task. Previous analysis was performed using reporter assays with a construct containing a synthetic promoter containing multiple copies of a consensus NF-κB binding site fused to a luciferase reporter gene. However, this approach does not elucidate the role of this phosphorylation with respect to endogenous gene expression nor does it distinguish any gene specific regulation based on a particular serine. However, given that there are hundreds of known NF-κB target genes techniques such as real time PCR and northern analysis still may not be the most efficient methods to study the role of serine phosphorylation under the many conditions known to activate NF-κB. While the use of 78

95 commercially available microarrays facilitate the analysis of known NF-κB targets, the majority of the genes on these arrays are not NF-κB dependent and only compound the data analysis. In addition, testing multiple mutations under various conditions and over several timepoints would become cost prohibitive using the currently available arrays. Therefore, in this chpater we describe a system of NF-κB transcriptional analysis using custom spotted microarrays containing a subset of NF-κB target genes. We will show that the effect of a particular serine mutation on endogenous target gene expression is dependent on the activating stimuli. We also demonstrate that while some genes are activated in a biphasic manner as previously described [168, 205, 206], this is not a general profile of all NF-κB targets, and phosphorylation of p65 may be more relevant in a particular phase depending on the target gene. In summary, we demonstrate that p65 serine phosphorylation can in fact mediate differential target gene expression, and to the best of our knowledge this is the first study to investigate this effect on endogenous expression on a large scale. 4.2 Material and Methods Materials. Murine TNFα was purchased from Roche Biochemicals (Indianapolis, IN), recombinant human IL-1β from Promega (Madison, WI) and IFN-γ from Invitrogen (Carlsbad, CA). Antibodies to IκBα (C21), crel (C) and p65 (A) were obtained from Santa Cruz Biotechnology (Santa Cruz, CA), p65-c terminal antibody from Rockland Immunochemicals, Inc. (Gilbertsville, PA), and α-tubulin (B-5) from Sigma (St. Louis, MO). 79

96 Plasmids. Retroviral expression constructs for full-length p65 and the truncated p65 (N) were created by RT-PCR to amplify the respective cdna from human fibroblasts and cloned into pbabepuro. Serine to alanine mutant p65 constructs were generated using the QuikChange Site-Directed Mutagenesis kit from Stratagene (La Jollla, CA) and indicated primers (Table 4.1). Cell culture and Retrovirus Infections. All fibroblast cells were cultured in DMEM high glucose media (Gibco/Invitrogen) containing 10% fetal bovine serum and antibiotics. Retrovirus production and infections were performed as previously described [112]. Northern Analysis and Hybridization. Northern analysis was performed as previously described. Briefly, 5 µg of total RNA prepared in TRIzol reagent (Invitrogen) was separated on 1.4% agarose gel, and the RNA was transferred overnight onto nylon membrane. RNA was UV cross-linked and pre-hybridized with QuikHyb (Stratagene). Probes were generated using HighPrime (Roche) and [α- 32 P]dCTP. Blots were hybridized for 1 hour at 68 o C, followed by two washes in 2X SSC (NaCl, sodium citrate, ph 7)/0.1% SDS at room temperature, and two washes in 0.1X SSC/0.1% SDS at 65 o C. Blots were exposed overnight on a screen, visualized and quantitated using the Typhoon phosphoimager and ImageQuant software (GE Healthcare, Piscataway, NJ). 80

97 Microarray Hybridization. The design of the custom printed nylon microarray is discussed in detail in chapter 3. Briefly, 25 NF-κB target genes and 3 controls (18S, actin and GAPDH) were PCR amplified and spotted onto nylon membranes. Arrays were pre-hybridized (50% formamide, 5X SSC, 1X PE) for two hours at 42 o C. To generate the probe, 10 µg of total RNA from untreated or cytokine induced wild type and mutant MEFs was reverse transcribed using SSII and [α- 32 P]dCTP. Resulting cdna was added to hybridization mixture on arrays and incubated overnight at 42 o C. Arrays were washed in 2X SSC/0.1% SDS twice at room temperature, and in 0.1X SSC/0.1% SDS twice at 42 o C. Arrays were exposed overnight on a screen, and visualized and quantitated using the Typhoon phosphoimager and ImageQuant. Immunoblotting and EMSA. Whole cell extracts from cultured cells were prepared and immunoblotting was performed as previously described [112]. Protein detection was performed using enhanced chemiluminescence (PerkinElmer Life Sciences, Boston, MA) and imaged either by using a Chemidoc gel documentation system (BioRad Laboratories, Hercules, CA) or by exposing blots to film. All quantitation was performed using the ImageJ software (NIH, Bethesda MD). Electromobility shift assay was performed as described [181]. 4.3 Results and Discussion Several serine residues in p65 have been previously shown to be important for NF-κB transcriptional activity, 276 in the Rel homology domain, and 529 and 536 in the transactivation domain. Each of these residues is important for enhancing 81

98 transactivation, in particular, phosphorylation at serine 276 is important in mediating protein-protein interactions (such as recruitment of CBP/p300) [55] and serine 536 is important to decrease the affinity of the complex for IκBα [200]. Most of the studies thus far to determine the importance of phosphorylation of these serine residues have been done using reporter assays, which do not reveal the role on endogenous gene expression. In order to address how phosphorylation affects the regulation of endogenous genes, we generated cell lines in which each of these serine residues is mutated to alanine and expressed in the p65 null (p65 -/- ) background using retroviral infection with pbabepuro expression vector. Each of the mutant p65 proteins was expressed to the same extent as endogenous p65 in wild type MEFs, and IκBα protein levels were not affected (Figure 4.1A). In addition, IκBα degradation occurred normally in the mutant cell lines in response to TNFα (Figure 4.1B), as well as DNA binding (Figure 4.1C), suggesting normal NF-κB signaling. In order to determine if p65 mediated transcription was affected by the mutations, northern analysis was performed for IκBα, a known NF-κB target gene. Both serine 276 and 536 were required for full IκBα induction in response to TNFα (Figure 4.1D). While mutation of serine 529 does not significantly impair the ability of p65 to activate IκBα transcription, it is clearly important for general NF-κB mediated transcription, as indicated previously by reporter assays [71]. As discussed in the introduction, NF-κB is phosphorylated at these serine residues by many different kinases in response to a variety of stimuli. Activation can also occur at various times post stimulation and in several different cell types. Therefore the particular conditions by which NF-κB is activated may determine which genes are 82

99 regulated, and it is possible that serine 529 may be important for other NF-κB targets distinct from IκBα. In order to establish if target genes are differentially regulated by these serine residues, we developed a custom microarray system to measure the transcription of multiple target genes simultaneously. Affymetrix microarrays (Mu74Av2) were used to initially screen for NF-κB targets to include on the custom array (data not shown). We selected from this screen genes induced by TNFα at least 4-fold in p65 +/+ vs. p65 -/- MEFs, and that were also shown previously in the literature to be regulated by NF-κB. This resulted in list of 25 genes with 3 internal controls (18S, actin, and GAPDH) (listed in Table 3.3). The spotting system used was that of printing PCR products, amplified using T3 and T7 primers from clones designed in psk+ vector, onto nylon membranes. This system was originally described for identifying methylated CpG islands, and modified slightly for our purposes in this study. Design and production of these arrays are discussed in more detail in Chapter 3. After spotting, SYBR green staining was used in order to establish that DNA is deposited evenly among the spots on the array (Figure 4.2A). There are areas where the PCR is not as efficient or the spotting is not as intense. However, each gene was represented nine times on the array, and the replicates were positioned at various regions throughout the array to prevent all the spots for a particular gene being affected by spotting variations. To ensure that the arrays are able to detect differences in gene expression, p65 +/+ MEFs treated with TNFα for 1 hour (TNF 1 ) were compared to untreated p65 -/- MEFs. There was a significantly greater amount of transcript in the TNF treated p65 +/+ MEFs 83

100 than in the p65 -/- MEFs, however the controls (actin and GAPDH) were equal on the two arrays indicating that the difference was not due to insufficient probe generation or hybridization (Figure 4.2B). Quantitation (p65 +/+ TNF 1 vs. p65 -/- untreated) of these arrays verified the induction (Figure 4.2C). Next, in order to determine the effect of serine mutation on gene expression, the wild type and S276A and S536A mutant cells lines were treated with TNFα for 1 hour and analyzed using the arrays. These two particular mutant lines were selected given that they showed the greatest difference on IκBα transcription by northern analysis (Figure 4.1D). While activation was overall reduced in both mutant cells lines, the effect was greater with the S276A mutant as evident on the array image itself (Figure 4.3A). Quantitation of the arrays revealed that in particular, Ccl2, Groα, IP-10, MIP-2, Mmp3, Ptx3, Saa2, Saa3 and Sele were significantly reduced in the S276A cell line (Figure 4.3B). The results from the array were verified by northern analysis for Ccl2 and IκBα, as well as MHC class I (MHC), a gene that was not particularly affected by the S276A mutation (Figure 4.4). Quantitation of the Northern blots correlated with the data from the array in that the same trend of expression was evident by both types of analysis. Given the regulation of chemokines and cytokines by S276, it would appear that phosphorylation at this residue is important in immune response. However, other immune related genes (GMCSF, C3, IL-6, CD40) were not as affected by mutation of S276. It is possible that regulation of these genes by S276 may be more relevant in response to other stimuli such as LPS or IL-1β. In addition, other cells types in which these genes play a more active role may require phosphorylation of S276 for full transcription (for example, GM-CSF in macrophages or CD40 in T-cells). 84

101 The S536A mutant (located in the TAD) did not display the same extent of decreased activation as S276A (located in the RHD). Therefore, we reasoned that other serine residues that may be phosphorylated in the activation domain of p65 might compensate for the function of 536. Using a program to predict residues with a high probability of phosphorylation, we identified several other serine residues in the transactivation domain of p65 likely to be phosphorylated (Figure 4.5A). We mutated these seven serine residues in the transactivation (TA) domain of p65, including S529 and S536, and have designated this construct p65 Super Trans-Activation Mutant (STAM). We also made a truncated form of p65 that completely lacks the TA domain (N) (Diagram in Figure 4.5A). We again verified equal expression of these mutant proteins (Figure 4.5B), and determined DNA binding ability by EMSA (Figure 4.5C). A faster migrating binding complex was identified in the N cell line, due to the smaller size of the active complex resulting from removal of the TA domain of p65. STAM bound to DNA with equal affinity as wild type, indicating that mutation of these C-terminal serines does not impede nuclear localization and does not cause a conformational change that is prohibitive to DNA binding. Interestingly, while IκBα transcript was significantly reduced in the truncated cell line as expected, the activation was actually slightly increased in the STAM line (Figure 4.5D). This could be due to alleviation of any repressive effect exerted by phosphorylation at one of these residues. Phosphorylation of S468 for example, under certain conditions, has been identified to have a negative effect on NF-κB mediated transcription [92, 93]. We therefore mutated A468 back to serine in the STAM cell line, which resulted in IκBα expression similar to wild type levels (data not shown). Therefore, depending on the particular serine residue modified, p65 85

102 phosphorylation can have a positive or a negative effect on transcription. The p65- STAM and p65-n mutants were analyzed on the array to determine whether the neutralizing result of mutation of the different serine residues was a general effect, or specific to certain genes. Mutation of the seven serine residues in the TA domain of p65 had no significant effect on transactivation, as most genes had around a 1-fold difference between wt and mutant (Figure 4.5E). Therefore while disrupting phosphorylation at a single residue impedes the expression of some genes, this same modification at other residues may actually increase transcription. NF-κB can directly repress transcription of anti-apoptotic genes in response to UV and doxorubicin [207]. While the TAD of the p65 subunit is necessary for this repression, it is not determined whether this phosphorylation occurs in the TAD. However our results support the idea that phosphorylation in the TAD can repress transcription, depending on which residue the phosphorylation occurs. Another way in which the transcription of NF-κB is regulated is that all serine resides might not be phosphorylated at the same time. Our lab had previously shown that NF-κB activation in response to TNFα occurs at two distinct phases, an early rapid induction at one hour, and a second wave of activation at hours [168]. It is not yet known whether two waves of induction is typical of all NF-κB target genes, or whether this is a gene specific effect. Therefore we used the array to determine which of our target genes displayed biphasic induction in MEFs treated with TNFα for 1, 6, or 12 hours. Many genes displayed activation at 1 and 12 hours while activation was slightly reduced at 6 hours, correlating with previous gel shift data [168] (Figure 4.6A). Interestingly some genes, such as Groα and IEX-1L, are only induced at one hour 86

103 whereas other genes (Saa2, Saa3 and Mmp3) are activated at the highest level at 12 hours. This suggests that while some genes are induced at two different phases, other genes are only induced once, and when this induction occurs depends on the target gene. In order to determine whether the effect of the serine mutations differs in the various phases of activation, the wild type and the S276A cells were treated with TNFα for 1 or 12 hours and analyzed on the array (Figure 4.6B). For Ccl2, a gene induced at both 1 and 12 hours, the mutation of serine 276 only reduced activation at the one hour time point. Alternatively, Saa2 and Saa3 exhibited more significant reduction at 12 hours, the point at which these genes were strongly activated in wild type MEFs. Many NF-κB target genes are rapidly activated in response to most stimuli, and therefore the transcriptional effect is often determined immediately following treatment. However if only the early time point were examined in this study, the effect of serine 276 on Saa2 and Saa3 may have been overlooked. This further highlights the importance of the array for the analysis of multiple target genes and multiple conditions. Our lab has shown that the activation of NF-κB activation in the second phase is necessary for the loss of skeletal muscle protein in response to cytokine, which may play a role in the muscle degenerative disease cachexia [168]. Additionally, our results from the biphasic analysis with the array indicate that the serum amyloids, members of a family of proteins that play a potential role in amyloid plaque formation in degenerative disease such as Alzheimer s, were affected by mutation of S276 more significantly at in the second phase. Therefore, genes regulated in the late phase of NF-κB activation may be important in degenerative disease, and it would be interesting to look at the expression of other genes involved in these conditions. 87

104 TNFα is only one of many cytokines to activate NF-κB. Therefore it is likely that a different set of target genes may be regulated by another stimuli, and this may be mediated by phosphorylation at a different serine residue. IL-1β for example can induce phosphorylation of serines 529 and 536, but not 276 [199]. We therefore reasoned that in response to IL-1β, serines 529 and 536 might play a more important role in regulation of gene expression. We found that in fact, serine 536 is more important than both 276 and 529 (Figure 4.7A and data not shown). Interestingly, for genes such as Ccl2, Ptx3, Saa2 and Saa3, where activation was reduced in the S276A mutant in response to TNFα, activation was not altered in either the S276A or S536A lines in response to IL-1β. Alternatively for IL-6 or CD40, target genes which were not affected in either mutant cell line in response to TNFα, activation was reduced by mutation of serine 536 in response to IL-1β. This data indicates that specific serine phosphorylation events mediated through cytokines can differentially affect target gene expression. Therefore it is not only the serine residue, but the stimulus as well that determines which targets genes are activated. This effect on transcription could be dependent on other modifications of p65 occurring downstream of phosphorylation in response to cytokine induction. For example, acetylation of p65 at lysine 310 requires phosphorylation at serines 276 or 536, as mutation of either of these residues reduces acetylation [103]. It has previously been shown that NF-κB activation is synergistically enhanced by the combination of two cytokines, TNFα and IFNγ [208]. We therefore wanted to determine the effect of these two cytokines on gene expression with respect to serine phosphorylation. We predicted that the effect would be similar to that observed with 88

105 TNFα treatment, and that addition of IFNγ may serve to further reduce the activation of target genes in the S276A cell line. However, we observed a pattern similar to that of IL- 1β treatment, where many genes on the array were more affected in the S536A mutant than S276A (Figure 4.7B), which is in contrast to results with treatment with TNFα alone. This difference could be due to the recruitment of other transcription factors, such as the STAT proteins, that are activated in response to IFNγ and cooperate with NF-κB at the promoter. inos and IP-10 for example are activated by TNFα alone, however this activation is enhanced by the combination of TNFα and IFNγ [208]. The fact that these genes require STAT for full activation could explain why at least for inos, no change in transcription was apparent upon treatment with TNFα alone, and was only evident upon additional treatment with IFNγ. Knowing which genes are regulated in response to a particular cytokine, and how p65 phosphorylation mediates this response may begin to elucidate a role for a specific serine residue in a pathway or disease. In our studies, we mainly observed differences in immune gene regulation. However, this was upon activation by proinflammatory cytokines known to initiate an immune response. Following treatment with other NF-κB inducers that initiate DNA damage such as UV, H 2 O 2, or etoposide, we may have witnessed little change in immune related genes while identifying a serine specific role on apoptotic or cell survival genes. Similarly, under conditions such as serum removal, p65 phosphorylation may affect cell growth and proliferation target genes. Given that NF-κB is activated in different disease states, knowing which genes are activated, and how they are activated (i.e. through phosphorylation at which serine 89

106 residue) in these cases is critical, as NF-κB inhibition has been targeted as a form of therapy. Therefore a better understanding of how this transcription factor regulates different genes may lead to better, or more specific, pharmaceutical development. 90

107 Mutant S276A 5` S276A 3` S529A 5` S529A 3` S536A 5` S536A 3` S335A 5` S335A 3` 5` primer CTGCGGCGGCCTGCCGACCGGGAGC GCTCCCGGTCGGCAGGCCGCCGCAG CCAATGGCCTCCTTGCAGGAGATGAAGACTTCTCC GGAGAAGTCTTCATCTCCTGCAAGGAGGCCATTGG GGAGATGAAGACTTCTCCGCCATTGCGGACATGG CCATGTCCGCAATGGCGGAGAAGTCTTCATCTCC CATTGCTGTGCCTGCCCGCAGCTCAGCTTCTG CAGAAGCTGAGCTGCGGGCAGGCACAGCAATG S338/340A 5` CTTCCCGCAGCGCAGCTGCTGTCCCCAAGCC S338/340A 3` GGCTTGGGGACAGCAGCTGCGCTGCGGGAAG S356A 5` S356A 3` S468A 5` S468A 3` GCCCTATCCCTTTACGTCATCCCTGGCCACCATCAACTATGATG CATCATAGTTGATGGTGGCCAGGGATGACGAAAGGGATAGGGC GTTCACAGACCTGGCAGCCGTCGACAACTCCG CGGAGTTGTCGACGGCTGCCAGGTCTGTGAAC Table 4.1: Primers to generate mutant p65 constructs. 91

108 Figure 4.1: Mutation of Serines 276 and 536 reduce p65-mediated transcription of IκBα. A. Whole cell extracts (WCEs) were prepared from p65 -/- immortalized MEFs reconstituted with the indicated mutant p65 retroviruses, and wild type MEFs were included as a control. The levels of the indicated proteins were analyzed by Western blot. α-tubulin was used as a loading control. B. WCEs treated with TNFα (10 ng/ml) were prepared at the indicated timepoints from p65 -/- immortalized MEFs reconstituted with the indicated mutant p65 retroviruses. Westerns blots were performed for the indicated proteins, with α-tubulin as a loading control. C. EMSA for nuclear proteins in the mutant cell lines treated with TNFα (5 ng/ml) for indicated timepoints. Supershifts were performed using antibodies to crel and p65. D. Northern analysis for IκBα in the indicated mutant cell lines untreated (-) or treated with TNFα (10 ng/ml) for 1 hour (+). 92

109 Figure 4.2: Custom NF-κB Microarray. A. SYBR green staining of spotted array, 9 replicates for IκBα are indicated (in boxes). B. Microarray analysis p65 +/+ MEFs treated with TNFα (10 ng/ml) for 1 hour, and untreated p65 -/- MEFs. C. Quantitation of p65 +/+ TNF 1 vs. p65 -/- untreated. Fold induction indicated on y-axis. 93

110 Figure 4.3: Mutation of Serine 276 displays represses activation to a greater extent than mutation of serine 536 in response to TNFα. A. Microarray analysis of wild type, S276A or S536A mutants treated with TNFα (10 ng/ml) for 1 hour. B. Quantitation of wild type TNF 1 vs. mutant TNF 1. Fold change indicated on y-axis, negative numbers indicate reduced activation in the mutant compared to wild type. 94

111 Figure 4.4: Verification of array data. Array data is verified by northern analysis. Quantitation of array (left) is comparable to quantitation by northern analysis (right). 95

112 Figure 4.5: Mutation of multiple serine residues located in TA domain of p65 does not impair transactivation. A. Diagram of p65 STAM mutant indicating the location of S/A mutations, and C- terminal truncation mutant. B. WCEs were prepared from p65 -/- immortalized MEFs reconstituted with the indicated mutant p65 retroviruses. Western blots were performed to probe for the indicated proteins. C. EMSA for nuclear proteins in the mutant cell lines treated with TNFα (5 ng/ml) for indicated timepoints. D. Northern analysis for IκBα in the indicated mutant cell lines untreated (-) or treated with TNFα (10 ng/ml) for 1 hour (+). E. Microarray analysis for STAM and p65n mutants compared to wild type. Fold change indicated on y-axis, positive numbers indicate increased activation, negative numbers indicate reduced activation, in the mutant compared to wild type. 96

113 Figure 4.5 (continued): 97

114 Figure 4.6: Biphasic activation of NF-κB target genes. A. Microarray analysis in the wild type MEFs treated with TNFα (10 ng/ml) for indicated timepoints to determine biphasic activity of NF-κB target genes. Fold induction (compared to untreated) indicated on y-axis cells. B. Microarray analysis for 1 and 12 hour timepoints in the wild type versus S276A mutant. Fold change indicated on y-axis, positive numbers indicate increased activation, negative numbers indicate reduced activation in the mutant compared to wild type. 98

115 Figure 4.7: Different cytokine signaling pathways alternatively regulate p65 mediated gene expression in a manner dependent on serine phosphorylation. A. Microarray analysis in wild type, S276A or S536A mutants treated with IL-1β (10 ng/ml) for 1 hour. Fold change indicated on y-axis, negative numbers indicate reduced activation in the mutant compared to wild type. B. Microarray analysis in wild type, S276A or S536A mutants treated with TNFα and IFNγ (5 ng/ml/50 U/ml) for 6 hours. Fold change indicated on y-axis, negative numbers indicate reduced activation in the mutant compared to wild type. 99

116 CHAPTER 5 DISCUSSION Overview As stated in the introduction, NF-κB was discovered over twenty years ago. And while our understanding of how this transcription factor regulates its targets has grown significantly in this time, there is still much that is unknown. One question that has remained is how a signaling pathway in response to a single stimulus is able to ultimately regulate distinct target genes. It has been suggested that interactions with other transcription factors and co-activators are responsible for this specificity. One way in which NF-κB interacts with other proteins is through various posttranslational modifications, which are discussed in depth in the introduction. How and when these modifications occur has been previously addressed in the literature, however mechanisms by which they differently regulate NF-κB target genes remains largely unknown. This dissertation investigates two different ways in which NF-κB members are modified, specifically through phosphorylation, and how this regulates NF-κB activity and down stream processes. In chapter two, a hyper-phosphorylated form of IκBβ is described, which is more stable than the basally phosphorylated form found in most cell types in the absence of p65, a state which normally causes degradation of IκBβ. This down regulation under 100

117 normal conditions is important given that over-expression of IκBβ has detrimental effects on cell growth. This study provided mechanism and biological relevance for an observation first evident in 1995 [14]. However, there are still many questions to address in this study. Specifically, what is the identity of the other forms of IκBβ (forms II and III) that are also stable in the absence of p65? Also, what is the mechanism by which IκBβ causes a growth defect? It was determined that the decreased cell number upon over-expression of IκBβ was at least partially due to an increase in apoptosis. However, the amount of apoptosis does not completely account for the decrease, suggesting that other mechanisms are in effect as well. It would be of interest to study the changes in gene expression in IκBβ over-expressing cells versus control to determine if there are specific cell growth or proliferation target genes affected. This regulation may also be important therapeutically. Drugs are currently available to regulate uncontrolled NF-κB mediated cell growth and proliferation in disease. Curcumin, a natural inhibitor of the NIK/IKK pathway, has been demonstrated to induce apoptosis in chronic lymphocytic leukemia (CLL) [209]. Anti-inflammatory and antioxidant based drugs are often used to treat NF-κB related diseases, however these drugs are very non-specific and affect pathways other than NF-κB. There are also several synthetic NF-κB inhibitors available, such as a peptide that binds to IKKγ/NEMO and inhibits active NF-κB complex formation [210]. These types of therapies are upstream of the IκB proteins, and likely affect many different processes regulated by NF-κB. Other specific inhibitors may be more useful in efficient treatment without harmful side effects, and this could be accomplished by targeting proteins further downstream in the NF-κB pathway such as 101

118 IκB. Data provided in chapter two suggest that over expression of IκBβ may be a method of inducing apoptotic cell death in diseased cells. Therapeutic approaches have shown that over expression of IκBα does sensitize cells to radiation-induced apoptosis [211]. It is possible that IκBβ over expression may function in a similar manner, and maybe even more specifically. Further knowledge as to how IκBβ induces cell death is necessary to determine if this type of treatment would be logical. One step in further understanding how NF-κB regulates pathways such as cellular proliferation or apoptosis, especially in terms of designing better therapeutic options using this complex, is gaining more information how it regulates its target genes. As part of this dissertation, we sought to determine an efficient method for the simultaneous analysis of multiple target genes. Chapter three describes the design of two types of custom microarrays, one using an oligonucleotide system, the other using a PCR product based system. The design of this custom array was necessary due to the fact the commercially available arrays are expensive and not specific to NF-κB. And although there are NF-κB arrays available, making our own arrays provided the control to modify the genes on the array when necessary. Chapter two focused on IκBβ, and how its phosphorylation regulates the stability of the protein and subsequently controls cell growth. This regulation by NF-κB through protein stability is different than the typical manner by which NF-κB affects cellular processes, which is through regulation of gene expression. Chapter four focused on this type of regulation, specifically how modification of p65 regulates gene expression. 102

119 Of the many modifications that occur to NF-κB proteins, this dissertation focused on phosphorylation, and previous reports have demonstrated that phosphorylation at serine residues 276, 529 and 536 is required for p65 mediated transactivation. Given that most of these studies were using luciferase assays, we sought to determine the effect of phosphorylation on endogenous genes using the microarray system designed in chapter 3. Even though it is reasonable to assume that not all NF-κB targets genes are regulated in the same way, a systematic approach to identify differential gene expression had yet to be undertaken. A better understanding of how phosohorylation at a particular serine regulates specific genes may lead to development of specialized therapies. For example, if phosphorylation of serine 276 is necessary for regulation of inflammatory genes but dispensable for genes involved in cell survival or differentiation, than this particular residue can be targeted for therapies of autoimmune diseases, such as rheumatoid arthritis. Importantly this type of drug treatment may be able to function in a similar manner as currently available drugs such as TNFα blockers, which have adverse side effects due to regulation of other factor rather than just NF-κB target genes. By more specifically targeting therapies, it may be possible to treat the disease while avoiding the side effects that occur with the current treatments, such as increased risk of infection. Results from chapter 4 showed that there is a gene specific regulation caused by phosphorylation at certain serines, and this response varies depending on the activating signal. We also show that some, but not all NF-κB target genes are activated in a biphasic manner. This is an important observation because the fact that different targets genes are activated at different time courses may affect NF-κB regulated processes. 103

120 Therefore knowing when genes are activated in response to stimuli increases our knowledge of how NF-κB elicits its response. The development of the microarray in chapter 3, and its implementation to study endogenous target genes under various activation conditions in chapter 4 has provided more insight into how NF-κB regulates gene expression. However there are still more questions to be answered and many areas in which this technology can be useful, and these are discussed in the future directions. Future Directions Approaches to study serine phosphorylation in muscle cells. The work in chapter 4 was completed solely in fibroblast cells lines. However, as mentioned in the introduction to Chapter 4, the expression profile in response to different cytokines, or phosphorylation of a particular serine, may vary depending on the cell type. The study of serine mutations in cell types other than fibroblasts is limited, due to the requirement of a p65 -/- cell that can be infected with retrovirus. Currently underway are efforts to isolate primary murine myoblasts from p65 -/- mice, which can be immortalized using SV40 large T antigen, and subsequently reconstituted with the serine mutants. NFκB is known to regulate myogenic genes such as MyoD [182], therefore it will be interesting to determine if there is a specific role for serine phosphorylation in muscle cells. Also, given that the design of the custom arrays allows us to change the target genes on the array, our lab has also modified this technology in order to study NF-κB regulation of muscle specific genes during differentiation, not necessarily just NF-κB 104

121 target genes, indicating yet another use of this technology to further increase our understanding of NF-κB signaling. Approaches to study serine phosphorylation in vivo by adoptive transfer. The role of phosphorylation and serine mutations has yet to be determined in vivo. A large reason for this is the complex nature of the available methods to approach this, such as construction of a knock-in mouse model. Also, it is unknown how mutation of these serine residues will affect the physiology of the mouse, therefore given that expression in the entire animal may be harmful, or even lethal, a conditional knock-in approach may be necessary. An alternative to knock-in models is a technique such as adoptive transfer, where cells from one animal are transplanted to another. This process allows reconstitution of a gene in a specific cell lineage, the hematopoietic system, without affecting the entire animal. These studies have been attempted in our lab, where host mice are irradiated to deplete the hematopoietic system, and are repopulated with fetal liver cells (the hematopoeitc precursors) from a donor mouse infected with retrovirus for the various serine mutations. The limiting factor of this approach has been infection of fetal liver cells. These cells do not proliferate well, and it is therefore difficult to incorporate retrovirus. Even using the murine stem cell virus (MSCV) construct, which is specifically designed for the infection of stem cells, very little expression of mutant is evident. The MSCV construct contains an internal ribosomal entry site, so that GFP is translated from the same transcript as the mutant allowing visualization of mutant expression (Figure 5.1). Flow cytometry for GFP verified that only about 12% of the FLCs were infected (Figure 5.2), therefore after FACS to isolate 105

122 only the infected cells, we were not able to obtain a sufficient number of cells for the adoptive transfer. Further attempts to increase the viral infection efficiency are required to implement this approach. Summary The studies outlined in this dissertation provide new evidence for gene specific regulation by NF-κB. While p65 phosphorylation has previously been shown to be required for full NF-κB activity, how this affects most NF-κB endogenous genes is still largely unknown. Here, the use of a custom designed microarray begins to investigate this regulation under several NF-κB activating conditions. This type of analysis has the potential to elucidate specific pathways downstream of NF-κB controlled by phosphorylation at particular serine residues. As has been discussed throughout this dissertation, perhaps the greatest relevance of this information may be the design of pharmacological inhibitors of NF-κB. In fact, it may be possible to design of agents that specifically target a single disease, rather than the more general inhibitors that are currently available. This subsequently would lead to more efficient treatment regimens and ultimately better patient care. 106

123 Figure 5.1: Diagram of MSCV construct. 107