Novel Roles for the Transcriptional Repressor PRDM1 in Human Natural Killer Cells and Identification of an Inhibitor of Its Interacting Methyltransferase G9a

University of South Florida Scholar Commons

Graduate Theses and Dissertations Graduate School

2011

Novel Roles for the Transcriptional Repressor PRDM1 in Human Natural Killer Cells and Identification of an Inhibitor of its Interacting Methyltransferase G9a

Matthew Adams Smith University of South Florida, [email protected]

Follow this and additional works at: https://scholarcommons.usf.edu/etd

Part of the American Studies Commons, Immunology and Infectious Disease Commons, and the Molecular Biology Commons

Scholar Commons Citation Smith, Matthew Adams, "Novel Roles for the Transcriptional Repressor PRDM1 in Human Natural Killer Cells and Identification of an Inhibitor of its Interacting Methyltransferase G9a" (2011). Graduate Theses and Dissertations. https://scholarcommons.usf.edu/etd/3354

This Dissertation is brought to you for free and open access by the Graduate School at Scholar Commons. It has been accepted for inclusion in Graduate Theses and Dissertations by an authorized administrator of Scholar Commons. For more information, please contact [email protected].

Novel Roles for the Transcriptional Repressor PRDM1 in Human Natural Killer Cells

and Identification of an Inhibitor of its Interacting Methyltransferase G9a

Matthew Adams Smith

A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Molecular Medicine College of Medicine University of South Florida

Major Professor: Kenneth L. Wright, Ph.D. Amer Beg, Ph.D. George Blanck, Ph.D. John Koomen, Ph.D Sheng Wei, M.D.

Date of Approval: June 24, 2011

Keywords: Blimp-1, Chromatin, Innate Immunity, Cytokines, Histone Methylation

DEDICATION

I would like to dedicate this dissertation to Colleen Jewel Smith, my beautiful baby girl. May life bring you challenges, excitement and fulfillment.

ACKNOWLEDGEMENTS

Science is a collaborative effort and progress invariably requires the cumulative effort of multiple individuals. Were it not for the contributions of many people I would not be where I am today. First and foremost, I would like to thank my mentor Ken

Wright, Ph.D., who has allowed me to learn and grow as a scientist over the past six years. He has provided insight on countless occasions over a wide variety of topics and I have truly valued our mentor-trainee relationship. I would also like to thank Lillian

Stark, Ph.D., M.P.H. who took a chance on hiring me at a transitional stage in my life and helped inspire me to pursue a career in science. I am indebted to my parents who have helped in so many ways over the years and although they often didn’t understand exactly what I was doing in the lab, were always interested. Members of my committee members

(John Koomen, Ph.D., Gloria Ferreira, Ph.D., George Blanck, Ph.D., Amer Beg, Ph.D.

and Sheng Wei, M.D.) have provided insight and guidance along the way. I am also

grateful to John Coligan, Ph.D., to agree to serve as my outside chairperson

I would especially like to thank the current and past members of the Wright

laboratory. Michelle Maurin has been a great friend, colleague and “cloning magician.”

Having joined the lab at roughly the same time, we have learned a lot together through

the years. Former lab members Shruti Desai, Ph.D., and Sophie Bolick, Ph.D., were

good friends and (at times) co-commiserates. I have enjoyed working with current lab

members Tahseen Ismail, Lou Ella Alexander and Iris Koopmans and wish them well in

their future studies. Gabriëla Wright performed the ChIP-chip experiments described here and her technical skills have helped the lab tremendously. I have enjoyed working with her and have learned a great deal from her.

I would also like to thank members of other labs who have provided assistance in several projects. Harshani Lawrence, Ph.D., Nick Lawrence, Ph.D., and Said Sebti,

Ph.D., provided useful insight with the G9a inhibitor project. Michael Caligiuri, M.D.,

Ph.D., provided FACS-sorted samples that were crucial to this work. Esteban Celis,

M.D., Ph.D., and Cho Hyun Il, Ph.D., provided mouse splenocytes and performed polyI:C injections. Staff members of the Genomics, Flow Cytometry, Molecular

Modeling and High Throughput Screening core facilities have been instrumental in several aspects of this work. Interactions with many friends and colleagues at both USF

College of Medicine and Moffitt Cancer Center have been a critical component to my growth as a scientist.

Finally, I would like to thank my wife, Susan, for everything she has done for me over the past twelve years. I owe her big for her patience on “20 minute +/- 3 hour” weekend trips to the lab, tolerance for my rants about technical problems related to “cells, gels or pellets” and a thousand other things. I hope to spend many years repaying this debt.

TABLE OF CONTENTS

LIST OF TABLES ...... iv

LIST OF FIGURES ...... v

LIST OF ABBREVIATIONS ...... ix

ABSTRACT ...... x

CHAPTER I: INTRODUCTION ...... 1 Part I: PRDM1 Structure and Function ...... 1 Initial Characterization ...... 1 Genomic Organization, Protein Structure and Functional Domains ...... 2 Mechanisms of Action ...... 4 PR Family Members ...... 5 Roles in Embryonic Development and Non-Immune Lineages ...... 7 Roles in B Lymphocytes ...... 8 Role as a Tumor Suppressor ...... 10 Roles in T Lymphocytes ...... 13 Roles in Myeloid Lineage ...... 15 Part II: Natural Killer Cells ...... 16 Initial Characterization ...... 16 Development and Subset Diversity ...... 17 Mechanisms of NK Activation and Effector Functions ...... 20 Part III: Histone Methylation in Transcriptional Regulation ...... 25 Nucleosome Organization ...... 25 The Histone Code Hypothesis ...... 26 Methylation in Gene Silencing...... 28 G9a Structure and Function ...... 29

CHAPTER II: MATERIALS & METHODS ...... 33 Cell Lines ...... 33 Primary Cell Isolation ...... 33 Mice ...... 34 Flow Cytometry ...... 34 Immunoblotting ...... 35 RNA Extraction and cDNA Synthesis ...... 35 Microarray Hybridization and Analysis ...... 36

Enzyme-Linked Immunosorbent Assay ...... 36 51Chromium Release Assay ...... 36 siRNA-Mediated Knockdown ...... 37 Transient Transfection and Luciferase Assay ...... 37 Adenoviral Constructs and Transduction ...... 38 Real-Time Quantitative PCR ...... 39 Chromatin Immunoprecipitation...... 41 ChIP-chip...... 42 ChIP-chip Data Analysis ...... 43 DNA Constructs ...... 44 Recombinant G9a Expression and Purification ...... 45 Radioactive In Vitro Histone Methyltransferase Assay ...... 45 MALDI-TOF Mass Spectrometry Analysis ...... 46 Small Molecule Screening to Identify Methyltransferase Inhibitors ...... 47 Statistical Analyses ...... 47

CHAPTER III: PRDM1 REGULATES EFFECTOR CYTOKINE PRODUCTION IN HUMAN NATURAL KILLER CELLS ...... 48 Introduction ...... 48 Results ...... 50 Human NK cells alter expression of multiple effector molecules and transcription factors in response to cytokine stimulation ...... 59 Cytokine stimulation induces multiple PRDM1 isoforms in human NK cells ...... 67 PRDM1 expression is associated with NK activation ...... 68 PRDM1 is not involved in perforin-mediated cytotoxicity ...... 71 PRDM1 binds promoters in multiple target genes ...... 75 Blockade of cytokine-induced PRDM1 expression increases effector cytokine production...... 79 Over-expression of PRDM1 mediates repression of activation- induced expression of IFNG and TNF ...... 82 PRDM1 mediates repression of IFNG via elements in the proximal promoter ...... 83 PRDM1 accumulates during LAK expansion and associates with functional exhaustion ...... 84 Discussion...... 85

CHAPTER IV: IDENTIFICATION OF NOVEL PRDM1 TARGET GENES IN NATURAL KILLER CELLS ...... 92 Introduction ...... 92 Results ...... 93 PRDM1 is constitutively-bound to numerous target genes in the NKL cell line ...... 96 Expression changes associated with PRDM1 ...... 99 Identification of PRDM1-regulated genes by Microarray Analysis ...... 102 Discussion...... 108

CHAPTER V: PRDM1-MEDIATED REGULATION OF FOCAL ADHESION KINASE ...... 113 Introduction ...... 113 Results ...... 115 PRDM1 binds to the PTK2 promoter ...... 115 PRDM1 regulates FAK expression ...... 116 PRDM1 suppresses PTK2 promoter activity ...... 120 Discussion...... 121

CHAPTER VI: HIGH-THROUGHPUT SCREENING IDENTIFICATION OF A SMALL MOLECULE INHIBITOR OF G9A...... 125 Introduction ...... 125 Results ...... 127 G9a Assay Development ...... 127 High-Throughput Screening ...... 130 In vitro inhibitors of G9a ...... 133 G9a inhibitors fail to abrogate cellular functions ...... 138 Discussion...... 140

CHAPTER VII: DISCUSSION AND SCIENTIFIC SIGNIFICANCE ...... 144

REFERENCES ...... 152

APPENDICES ...... 185 Appendix I: Genes changed 4-fold in stimulated NK cells ...... 186 Appendix II: Expression changes of Chip-Chip Target genes ...... 202 Appendix III: Genes changed 1.3-fold in NK cells with PRDM1 knockdown ...... 208

ABOUT THE AUTHOR ...... END PAGE

iii

LIST OF TABLES

Table 1: Oligonucleotide Primers...... 39

Table 2: Changed ChIP-chip target genes in response to stimulation (Top 30) ...... 100

Table 3: Genes increased > 1.3 fold with PRDM1 Knockdown (Top 30) ...... 104

Table 4: Direct PRDM1 target genes: increased > 1.3 fold ...... 105

LIST OF FIGURES

Figure 1: PRDM1 exon organization, functional domains and isoforms ...... 2

Figure 2: PRDM1 functions in plasma cell differentiation ...... 8

Figure 3: Minimal 6q21 deletion found in B and NK-derived lymphomas ...... 12

Figure 4: PRDM1 functions in CD4+ T cells ...... 14

Figure 5: PRDM1 functions in CD8+ T cells ...... 15

Figure 6: Cytokine receptors and signaling pathways utilized by NK cells ...... 22

Figure 7: NK cells mediate extensive crosstalk between innate and adaptive arms of the immune system ...... 25

Figure 8: Histone methyltransferase structure ...... 32

Figure 9: Cells utilized for microarray analysis display NK surface phenotype and minimal donor-to-donor variability ...... 51

Figure 10: Top 75 up- and down-regulated transcripts upon stimulation ...... 52

Figure 11: Up- and down-regulated effector molecules in response to stimulation ...... 53

Figure 12: Up- and down-regulated cytokine receptor molecules in response to stimulation ...... 54

Figure 13: Up- and down-regulated NK receptors in response to stimulation ...... 55

Figure 14: Up- and down-regulated SLAM family members in response to stimulation ...... 56

Figure 15: Up- and down-regulated transcription factors in response to stimulation ...... 57

Figure 16: Up- and down-regulated IRF Family Members in response to stimulation ...... 59

Figure 17: Up- and down-regulated STAT Family Members in response to stimulation ...... 59

Figure 18: Cytokine stimulation induces PRDM1 isoforms in human NK cells ...... 61

Figure 19: Human NK cells lines do not recapitulate dynamics of cytokine- induced expression observed in primary NK cells ...... 62

Figure 20: Identification of a PRDM1 splice variant lacking exon 6 in NKL cell ...... 63

Figure 21: PRDM1 is preferentially expressed in the CD56dim NK population ...... 64

Figure 22: NK-specific expression of PRDM1 isoforms ...... 65

Figure 23: PRDM1 induction is transient and is regulated via a post- transcriptional mechanism ...... 66

Figure 24: Plasma cell transcription factor expression diverges in NK cells ...... 67

Figure 25: XBP1 splicing is not associated with NK activation ...... 68

Figure 26: PRDM1 induction is correlated with degree of activation...... 69

Figure 27: Autocrine and or paracrine feedback loops are not responsible for PRDM1 induction ...... 70

Figure 28: PRDM1 is induced in vivo ...... 71

Figure 29: Perforin-mediated cytotoxicity is not affected by PRDM1 knockdown ...... 72

Figure 30: Clustering phenotype observed upon cytokine activation is not diminished with PRDM1 knockdown ...... 74

Figure 31: PRDM1 binds CIITApIV, but not other B-cell target genes ...... 75

Figure 32: PRDM1 binds known regulatory loci within the IFNG locus ...... 77

Figure 33: PRDM1 binds known regulatory loci within the TNF locus ...... 78

Figure 34: Knockdown of PRDM1 increases levels of effector cytokine mRNA ...... 80

Figure 35: Knockdown of PRDM1 increases levels of secreted effector cytokines ...... 81

Figure 36: Time course of cytokine mRNA and PRDM1 protein levels ...... 81

Figure 37: Over-expression of PRDM1 suppresses activation of target genes...... 82

Figure 38: Promoter-proximal elements in the IFNG promoter are required for PRDM1-mediated repression ...... 83

Figure 39: PRDM1 expression and function in LAK Cells ...... 85

Figure 40: CisGenome consistently identifies more PRDM1-enriched peaks ...... 96

Figure 41: Distribution of PRDM1 binding locations across the genome ...... 97

Figure 42: Confirmation of Selected ChIP-chip target genes ...... 98

Figure 43: Motif analysis of associated peaks ...... 99

Figure 44: Immunoblots of PRDM1 knockdown used for microarray analysis ...... 102

Figure 45: Knockdown of PRDM1 increases mRNA levels of selected PRDM1 target genes ...... 107

Figure 46: PRDM1 occupies the distal PTK2 promoter ...... 116

Figure 47: Knockdown of PRDM1 increases PTK2 mRNA expression ...... 117

Figure 48: Knockdown of PRDM1 increases FAK protein expression ...... 118

Figure 49: Time-dependent reduction in PTK2 mRNA and FAK protein levels ...... 119

Figure 50: Over-expression of PRDM1 decreases FAK expression ...... 120

Figure 51: PRDM1 mediates repression through regulatory elements upstream of PTK2 ...... 121

Figure 52: MALDI-TOF MS analysis to monitor methylation ...... 129

Figure 53: Optimization of HMT assay conditions ...... 130

Figure 54: Raw data from four selected plates...... 131

Figure 55: Analyzed data from four selected plates ...... 132

Figure 56: Chemical structure of HLM-2414 ...... 134

Figure 57: IC50 values for various derivatives of HLM-2414 (YL1-132) ...... 135

Figure 58: In silico docking of YL1-148-1 to H3 peptide binding cleft ...... 136

vii

Figure 59: YL-148-1 inhibits in vitro methylation of core histones ...... 137

Figure 60: YL1-148-1 fails to inhibit G9a-dependent methylation in cell-based assays...... 139

Figure 61: Model of PRDM1 function in NK cells ...... 152

viii

LIST OF ABBREVIATIONS

Blimp1 B-lymphocyte induced maturation protein 1 BMP bone morphogenic protein bp base pairs CD cell determinant (marker) ChIP chromatin immune-precipitation CLP common lymphoid precursor DLBCL diffuse large B cell lymphoma DNA deoxyribonucleic acid ELISA enzyme linked immunosorbent assay EDTA ethylenediaminetetraacetic Acid ELP early lymphoid precursor HDAC histone deacetylase HMT histone methyltransferase HSC hematopoietic stem cell IFN interferon IL interleukin LTA lymphotoxin alpha LTB lymphotoxin beta MALDI matrix assisted laser desorption ionization miR micro RNA MS mass spectrometry NK natural killer cell NKP natural killer (cell) precursor PBS phosphate-buffered saline PRDM1 positive regulatory domain containing 1 qPCR quantitative polymerase chain reaction qRT-PCR quantitative reverse transcription – polymerase chain reaction RIPA radio-immuno-precipitation assay RNA ribonucleic acid SAM s-adenosyl methionine SD standard deviation SEM standard error of the mean SDS sodium dodecyl sulfate TNF tumor necrosis factor TOF time of flight TES transcriptional end site TSS transcriptional start site

ABSTRACT

The studies presented within this dissertation provide the first description of

PRDM1 (also known as Blimp-1 or PRDI-BF1) function in natural killer cells. NK cells are major effectors of the innate immune response via antigen-independent cytotoxicity and link to the adaptive immune response through cytokine release. Molecular mechanisms mediating NK activation are relatively well-studied; however, much less is known about the mechanisms that restrain activation.

In the first study, the transcriptional repressor PRDM1 is shown to be a critical negative regulator of NK function. Microarray analysis was used to characterize transcriptional changes associated with cytokine-mediated activation. PRDM1 is expressed at low levels in resting NK cells and three distinct PRDM1 isoforms are

dim selectively induced in the CD56 NK population in response to activation. PRDM1

coordinately suppresses the production of IFNγ, TNFα and TNFβ through direct binding

to multiple conserved regulatory regions. Ablation of PRDM1 expression leads to

enhanced production of IFNγ and TNFα but does not alter cytotoxicity, whereas over-

expression blocks cytokine production. PRDM1 response elements are defined at the

IFNG and TNF loci.

To further delineate the targets of PRDM1-mediated regulation in NK cells,

global approaches were utilized. Experiments utilizing chromatin immunoprecipitation

coupled to promoter tiling arrays identified 292 novel direct targets of PRDM1 binding.

These studies revealed widespread binding of PRDM1 to the genome, which was not limited to proximal promoter regions. Furthermore, microarray analysis of stimulated

NK cells combined with PRDM1 knockdown has enabled the identification of genes responsive to PRDM1 knockdown using primary cells. Collectively, these experiments identify both direct and indirect targets of PRDM1 regulation and help define a PRDM1- centered gene regulatory network in NK cells.

Data presented in the final chapter pertains to an independent project aimed at

identifying small molecule inhibitors of the methyltransferase G9a, which is recruited by

PRDM1 and is required for silencing of target genes. A mass spectrometry-based assay

was developed and used to screen a small molecule library. Several hits were identified

and combinatorial chemistry yielded several compounds with < 20µM IC50 values. In

cell-based assays, however, treatment with the small molecules had limited efficacy,

indicating additional chemical modifications are necessary to yield bioactive compounds.

The data presented here demonstrate a key role for PRDM1 in the negative regulation of NK activation and position PRDM1 as a common regulator of the adaptive

and innate immune response.

CHAPTER I

INTRODUCTION

Part I: PRDM1 Structure and Function

Initial Characterization

The transcriptional repressor Positive Regulatory Domain containing protein 1,

with Zinc Finger Domain (PRDM1) is the founding member of a 17-member protein

family characterized by the presence of an amino-terminal PR domain. It was first

identified in 1991 by Tom Maniatis and colleagues as Positive Regulatory Domain I

Binding Factor 1 (PRDI-BF1) by screening a cDNA expression library with a probe

containing multiple PRDI sites (1). This protein was shown to be virus-inducible, capable of binding to the PRDI DNA element of the IFNB promoter in footprinting experiments and sufficient for repression of a reporter gene under the control of the PRDI element. Thus, PRDM1 was established as a negative regulator of interferon-β in virally-

infected cells.

Several years later the murine homolog was isolated in Mark Davis’s laboratory by using a subtractive cloning approach to identify genes associated with the acquisition of plasma cell phenotypic properties (2). The authors named the isolated gene Blimp1

(B-lymphocyte induction of maturation protein 1) and provided initial evidence that it may play a crucial role in the differentiation of antibody-producing plasma cells.

Interestingly, these groups reporting independently did not realize they were analyzing the same gene product, albeit in two different experimental systems (3). Thus, pleiotropic functions of PRDM1 were established early after its initial description.

Genomic Organization, Protein Structure and Functional Domains

Human PRDM1 is a large, multi-domain protein containing 789 amino acids; murine Blimp1/Prdm1 contains 859 amino acids. Not including the murine 70 amino acid extension on the amino terminus, the two orthologs are 90% identical. Human

PRDM1 is located on chromosome 6q21 and spans approximately 23,600bp. Seven exons are spliced to form the full-length protein (Figure 1). Exon usage is identical between mouse and human with the exception of an extra amino-terminal exon present only in the mouse (4). A truncated isoform was identified in human myeloma cells and is

transcribed from an independent promoter, resulting in the deletion of 100 amino acids

Figure 1: PRDM1 exon organization, functional domains and isoforms. Three isoforms of PRDM1 are encoded by seven exons. Acidic regions are present in both the amino- and carboxy-terminal portions of the molecule. Five zinc fingers are present between residues 575-707. Proline-reach regions are clustered near the center of the molecule. The PR domain is located near the amino terminus and is disrupted in the β isoform.

and a disrupted PR domain (5). This amino-terminal truncation has been termed

PRDM1β, while the full-length protein is referred to as PRDM1α. In both human and mouse a deletion of the penultimate exon (PRDM1Δ6 or Prdm1Δ7, respectively) has also

been observed (4, 6).

The PR domain is located near the amino-terminal of the PRDM1 protein. This domain has been suggested to mediate protein-protein interactions in vitro (7), however,

its in vivo functions have not been experimentally demonstrated. The region has high

homology to the Su(var)3-9, Enhancer-of-Zeste, Trithorax (SET) domain, originally

defined as an epigenetic modifier in Drosophila (8). This domain is commonly found in

lysine-specific histone methyltransferases and its presence defines the PR-family of

proteins. Despite the high degree of homology to SET, the PR-domain lacks intrinsic

methyltransferase activity. An NMR structure of the closely-related PRDM2 has been

determined and reveals that the highly-conserved PR domain diverges significantly from

that of other lysine methyltransferases. It has a disordered post-SET domain and lacks

the characteristic Zn-Cys ion coordinating motif present in other lysine

methyltransferases (9). The truncated PRDM1β isoform is transcribed from an alternate

promoter present within the third intron and produces an alternate exon 155bp upstream

of exon four (5). This exon contains a novel translational start site resulting in a protein

with three novel amino acids (methionine-glutamate-lysine) and a disrupted PR domain.

The N-terminal acidic region may be important for stability. Mutations in proline

48 have recently been shown to impart increased susceptibility to degradation (10).

Furthermore, the interacting co-repressor Reptin52 fails to interact with PRDM1β,

suggesting that it may be required for protein-protein interactions (Desai, unpublished

observation). The central proline-rich domain (PRD) may also be important for protein-

protein interactions as specific structural motifs generated by proline residues can provide

docking sites for PRD-binding proteins (11). Indeed, this region facilitates the interaction

with several partners such as Groucho, LSD1 and HDAC2 (12)

Mechanisms of Action

PRDM1 is a classic zinc finger DNA-binding transcription factor. The five zinc fingers recognize the consensus sequence (A/C)AG(T/C)GAAAG(T/C)(G/T), found in the proximal region of target genes (13). This sequence is closely related to the recognition sequences of the interferon regulatory factors (IRFs). IRF transcription factors function as both activators and repressors and regulate numerous aspect of immune function. Thus, one mechanism of PRDM1-mediated repression is the direct competition for regulatory sites, resulting in displacement of IRFs and a decrease in transcription. Some degree of functional redundancy exists within the zinc finger domain because only two of the five fingers are required to maintain DNA-binding activity (14).

Furthermore, these residues are required for nuclear localization because deletion results in cytosolic accumulation of the mutant PRDM1 protein (Győry, unpublished observation).

In addition to passive repression via competition with activating factors, PRDM1 also exhibits active repression mediated through specific recruitment of co-repressors and histone-modifying enzymes. Interactions with HDAC2 are required for the repression of c-myc and HDAC inhibitors decrease the repressive ability of PRDM1 in reporter gene assays (15, 16). HDAC recruitment does not require a functional PR domain as

PRDM1α and PRDM1β both co-purify HDAC activity and mediate TSA-sensitive repression (5).

PRDM1 also functionally interacts with several molecules involved in the

regulation of histone methylation. Recruitment of G9a requires a region encompassing

the first two zinc fingers and results in increased di-methylation of histone 3 lysine 9

(H3K9) at the IFNB promoter to mediate silencing (15). PRDM1 recruitment of the

arginine methyltransferase PRMT5 results in symmetrical di-methylation of argine 3 in

both H2A and H4, silencing Dhx38 in primordial germ cells (17). Additionally, an

interaction with PRMT7 has been shown to occur in T-cam2 cells, a cell line derived

from a germ cell tumor (18). The H3K4-specific lysine specific demethylase 1 (LSD1) has also been shown to interact with PRDM1 in murine B-cells and occurs through the proline-rich domain of PRDM1 (19). LSD1 demethylates H3K4, a histone mark associated with gene activation. Thus, the removal of this histone modification likely enhances repressive ability. PRDM1-mediated repression is further reinforced via the recruitment of a variety of co-repressors such as members of the Groucho family (20) and

Reptin52 (Desai, unpublished observation).

PR Family Members

The PR domain defines a protein family which currently includes 17 members.

PRDM1, RIZ1 (PRDM2) and MDS1-EVI1 (PRDM3) represent the founding members of the family and are the best characterized. RIZ1 was identified as a retinoblastoma- interacting protein and was shown to be a transcriptional repressor that binds to Sp1 elements (21, 22). RIZ1 functions as a tumor suppressor and is inactivated in gastric

tumors (23). MDS1-EVI1 is a fusion protein resulting from a MDS1 and EVII

translocation and is frequently found in patients suffering from myelodysplastic syndrome. Interestingly PRDM1, RIZ1 and MDS1-EVI1 are all characterized by the presence of alternate isoforms with disrupted PR domains. These alternate isoforms have frequently been observed in malignant tissues and have been suggested to be oncogenic, lacking the tumor suppressive functions of the full-length proteins (24).

PR family members have been named PFM1-PFM14 starting from PRDM4, however, members are officially named PRDM_ and this convention is most common in the literature. All family members contain the characteristic PR domain and Krüppel-like

zinc fingers and appear to act as DNA-binding transcription factors, with the exception of

PRDM11 (PFM8) which lacks zinc fingers. Due to extensive rearrangements, PR family

members vary widely in amino acid length and number of zinc fingers. For example,

PRDM8 (PFM5) and PRDM12 (PFM9) have only three zinc fingers, while PRDM15

(PFM12) has 17 zinc fingers. Orthologs have been found in all metazoans and a

generally positive correlation exists between number of PR family members within a

species and organism complexity (25).

Expansion of the PR family occurred primarily due to gene duplication events and

has resulted in a large number of functionally diverse transcription factors with tissue-

and temporally-restricted expression patterns. For example, PRDM16 (PFM13, MEL1)

controls brown fat adipogenesis through a PPARγ-dependent mechanism (26, 27).

PRDM14 (PFM11), on the other hand, is required for primordial germ cell specification

(28). PRDM9 (PFM6) is associated with “hotspots” of meiotic recombination and its function may have contributed to speciation events, driving genetic divergence (29-33).

Currently, many of the remaining PR family members remain poorly characterized, but likely mediate unique functions.

Roles in Embryonic Development and Non-Immune Lineages

Roles for PRDM1 in early embryonic development have been established in model organisms. Germline knockout of Prdm1 in mice results in early embryonic lethality, confirming a requirement for its activity in early development (34). PRDM1 is expressed as early as embryonic day seven in a variety of mesoderm-derived tissue precursors in a spatially and temporally restricted manner indicating important roles in early cell-fate decisions (35). In zebrafish, the PRDM1 homolog Ubo specifies the location of the neural crest in response to BMP signaling and is required for sensory neuron formation (36). Early formation and delineation of bone patterning is also regulated by Ubo as mutations induce severe defects in pectoral fin structure (37).

PRDM1 dose-dependently regulates the formation of caudal pharyngeal arches,

secondary heart field and sensory vibrissae in mice (38).

Recently, PRDM1 has been shown to be involved in terminal differentiation of

several diverse tissues. PRDM1 is required for osteoclast differentiation in a calcium-

dependent mechanism that is mediated by NFATc1 (39). PRDM1 favors osteoclast differentiation, while BCL6 suppresses the process (40). PRDM1 is also involved in the differentiation of epidermal cells. Its function is required for keratinocytes to transition from the granular to cornified layer via the regulation of approximately 250 genes (41).

This diverse repertoire of PRDM1 functions is consistent with its conservation across

many metazoan species.

Roles in B lymphocytes

The initial subtractive cloning approach of the Davis Laboratory suggested that

Blimp1 was a key regulator of plasma cell terminal differentiation (2). [Note: the use of

Blimp1 or PRDM1 reflects naming conventions in the literature, Blimp1 typically refers to murine work, while PRDM1 denotes human. For clarity, PRDM1 is used exclusively here]. Indeed, Prdm1 was identified as a unique cDNA clone induced in BCL1 cells upon stimulation with IL-2 and IL-5, which initiates a plasmacytic differentiation

program. In both stable transformants and transient transfection assays, the Prdm1 gene product increased surface expression of syndecan 1 (CD138), induced immunoglobin secretion, and increased cell size and granularity (2). Thus, introduction of Prdm1 was sufficient to induce phenoptypic changes associated with plasma cells (Figure 2).

Figure 2: PRDM1 functions in plasma cell differentiation. Upon antigen encounter, naïve mature B cells undergo a differentiation process in order to functionally mature into antibody-secreting plasma cells. PRDM1 and several other transcription factors, such as IRF4 and XBP-1, are required for this phenotypic transition.

Exit from cell cycle is a hallmark of plasma cell differentiation and is

accompanied by a reduction in the cell cycle regulator c-myc. PRDM1 represses c-myc

upon induction of plasma cell differentiation and leads to either differentiation or

apoptosis, depending on the cell lines used (42). Induction of apoptosis occurs

preferentially in immature B-cells with ectopic PRDM1 expression, suggesting a

“gatekeeper” role, with apoptosis resulting when cells are not “primed” to differentiate

(43). This observation has clinical correlatives in terms of B cell lymphomas and

predicts that induction of PRDM1 in lymphomas may provide therapeutic benefit.

Subsequent studies identified target genes of PRDM1 in the context of plasma cell

differentiation. PRDM1 was shown to regulate the Class II transactivator (CIITA), which

is the master regulator of MHC class II expression, providing a mechanism for the down-

regulation of antigen presentation function that accompanies plasma cell differentiation

(44). PRDM1 also directly represses Pax5, which is required for B cell maturation, but is

silenced in plasma cells (45). Microarray analyses cataloged transcriptional changes

associated with PRDM1 demonstrating the existence of an extensive gene regulatory

network. PRDM1 directly regulates a number of transcription factors such as Spi-B and

Id3 (46). In a second study, functional domains of PRDM1 were also shown to mediate

differential gene expression patterns. Of 376 genes differentially expressed with

enforced PRDM1 expression, disruption of the PR domain abrogated the regulation of 31,

while disruption of the carboxy-terminus abrogated the regulation of 12 (47). Together, these microarray analyses identified nearly 600 unique PRDM1-dependent transcripts and demonstrate modular nature of PRDM1-mediated repression.

Through the regulation of key transcription factors, PRDM1 is the master regulator of plasma cell differentiation. Induction of XBP1 and IRF4 expression and loss of PAX5 expression are all required for the generation of plasma cells (48-50). Loss of

PAX5 expression initiates the commitment to plasma cell generation and PRDM1 acts downstream of PAX5 to regulate the transition (51). PRDM1 is also required for the maintenance of long-lived plasma cells populations within the bone marrow. Conditional ablation of Prdm1 in pre-formed plasma cells results in decreased numbers of plasma cells and reduced Ig secretion upon secondary challenge (52). Consistent with this,

Prdm1 knockdown in plasma cells initiates apoptosis (53). Thus, PRDM1 expression must be maintained subsequent to differentiation.

Role as a Tumor Suppressor

Given the importance of PRDM1 in the terminal differentiation of B lymphocytes it is not surprising that it has been investigated as a tumor suppressor. Deletions of the genomic region surrounding PRDM1 (6q21-22) are relatively common in lymphomas, which suggests a tumor suppressive role. Indeed, portions of the 6q arm are lost in ~25% of lymphoid malignancies and 6q21-23 deletions are one of the most frequently encountered genetic lesions (54). In fact, one study found 6q21 deletions in 85% of samples analyzed (55). Furthermore, inactivating mutations have been identified in diffuse large B cell lymphomas (DLBCL) by several groups (56-58). DLBCL are derived from the late stages of B cell development. Consequently, PRDM1 mutations have not been found in germinal center-derived lymphomas which arise from early stage

B cells that are not dependent upon PRDM1. Recent data has shown that in vivo

inactivation of PRDM1 causes increased incidence of various lymphocyte-derived

malignancies and this incidence is significantly enhanced when NFκB is constitutively active (10, 59). Thus, PRDM1 is a bona fide tumor suppressor gene in B cell lymphomas.

NK-derived malignancies represent a relatively rare fraction of lymphoid-derived malignancies in the United States and Europe, but exhibit increased incidence in Asia and

South America. NK malignancies are classified as either extranodal NK/T-cell

lymphoma (nasal type) or aggressive NK-cell leukemia, are uniformly CD3-CD56+ and

are typically EBV-infected (60). Deletion of 6q21-23 is the most common genetic lesion

found in NK lymphomas (61). Various studies have reported 6q21 deletions with

frequencies between 20 and 100% of samples analyzed (62). A recent study used high

resolution array comparative genomic hybridization (aCGH) analysis to map

chromosomal abnormalities in NK-derived tumor cell lines and patient samples. The

authors identified a 2Mbp minimal region of 6q21 (containing PRDM1 and two other

known genes, ATG5 and AIM1) that was deleted in 60% of the samples analyzed

(Figure 3) (62). Thus, PRDM1 may also function as a tumor suppressor in NK-derived

lymphomas.

Figure 3: Minimal 6q21 deletion found in B and NK-derived Lymphomas. A minimal, commonly deleted region found in NK lymphomas has been described in a significant fraction of NK lymphomas. Deletion of this region is also frequently associated with B cell lymphomas.

Interestingly, the tumor suppressive nature of PRDM1 may not be universal.

Studies in breast cancer cells demonstrate that PRDM1 is expressed at high levels in estrogen receptor α (ERα) negative tumors (63). PRDM1 is activated by RelB, suppresses ERα promoter activity and is associated with increased migratory capacity

(63). Thus, PRDM1 functions as an oncogene in breast cancer cells because its expression is mechanistically linked with both treatment resistance and metastasis. Thus,

PRDM1 exhibits pluripotent effects on tumor growth, which are very likely context and cell-type specific.

Roles in T-lymphocytes

Recent studies have demonstrated that PRDM1 mediates unique, non-overlapping

functions within the T cell compartment, positioning PRDM1 as a crucial regulator of

lymphocyte biology. Germline deletion of Prdm1 results in embryonic lethality,

however, conditional strategies have allowed the dissection of PRDM1 function. Crosses

of mice bearing a floxed Prdm1 allele to Lck-Cre mice yield progeny with a severe

inflammatory phenotype that die at an early age from colitis (64). Conditional knock out

T cells from these mice are hyper-responsive. They produce higher levels of IL-2 and

IFNγ and are increased in number in peripheral circulation. An alternate genetic ablation

strategy whereby Prdm1-deficient fetal liver cells are used to reconstitute Rag1-/- mice also demonstrates profound T cell-mediated inflammatory pathology (65).

PRDM1 is up-regulated in CD4+ T cells and regulates the differentiation of

effector subsets (Figure 4). It abrogates Th1 polarization via suppression of Ifng, tbx21,

and bcl6 (66). A negative feedback loop exists whereby PRDM1 is induced in response

to IL-2 and then silences both Il2 and its activator Fos resulting in decreased proliferative

capacity in vivo (67, 68). The inverse relationship between PRDM1 expression and

BCL6 observed in B cells is maintained in CD4+ T cells because PRDM1 prevents the polarization into T follicular helper cells by suppressing BCL6 (69). Very recently,

PRDM1 has been shown to be required for IL-10 production and tissue homeostasis of

CD4+CD25+ T regulatory cells in an IRF4-dependent manner (70).

Figure 4: PRDM1 functions in CD4+ T cells. PRDM1 contributes to the polarization of CD4 subsets. PRDM1 negatively regulates Th1 and Tfh subsets, while promoting the polarization of Th2 and Treg subsets.

In CD8+ T cells PRDM1 controls effector functions and is required for terminal

differentiation of CD8+ T cells (Figure 5). Prdm1-deficient CD8+ T cells fail to

efficiently migrate to the lung in response to influenza infection, in part due to

diminished expression of the chemokine receptor Ccr7 (71). Prdm1-deficient CD8+ T cells exhibited higher levels of TNFα and IFNγ in response to antigen stimulation, but lower levels of transcripts associated with cytotoxic granules Prf1, Gzmb and Gzmk (71).

PRDM1 expression is highest in terminally-differentiated effector cells, limits the

proliferative capacity of CD8+ T cells and restricts the potential of effector cells to

differentiate into memory cells (72). During chronic infection, PRDM1 expression is

highest in exhausted subsets and is correlated with up-regulation of inhibitory receptors

such as PD-1 and LAG-3 (73). Thus, PRDM1 regulates multiple facets of CD8+ T cell effector function.

Figure 5: PRDM1 functions in CD8+ T cells. In CD8+ T cells PRDM1 expression accumulates as naïve cells move toward exhaustion. Its expression is required for efficient recall responses and restricts the potential of effector cells to develop into memory cells.

Roles in Myeloid Lineage

PRDM1 functions not only within the lymphocyte lineage, but in myeloid-derived

cells as well. Expression is elevated in human granulocytes relative to monocytes and

over-expression drives terminal differentiation of the promonocytic U937 cell line (74).

In dendritic cells, PRDM1 is associated with maturation and is highly induced in

response to maturation-inducing stimuli, such as LPS, TNFα and CD40L (75). PRDM1

competes with IRF8 for binding to EICE sites present in the myeloid-specific class II

transactivator promoter (CIITApI). Down-regulation of CIITApI requires G9a and

HDAC2 and provides a mechanism for the attenuation of class II transcription that accompanies DC maturation. Furthermore, in mice PRDM1 represses pro-inflammatory cytokines (Ccl6, Il6) and knockdown diminishes DC maturation in vitro and in vivo (76).

Thus, PRDM1 mediates critical functions associated with myeloid differentiation.

Part II: Natural Killer Cells

Initial Characterization

Natural killer (NK) cells are a subset of lymphocytes representing 10-15% of

circulating peripheral blood mononuclear cells. As circulating lymphocytes these cells

localize to a variety of organs and can be found in the spleen, lymph nodes and liver. NK

cells were first identified in the late 1970’s by several independent groups via their ability

to lyse tumor cells via MHC-independent mechanisms (77, 78). Indeed, non MHC- restricted cytotoxicity is a defining hallmark of NK cells, but NK cells are also potent producers of cytokines and mediate antibody-dependent cytotoxicity. NK cells are

identified on the basis of cell surface phenotype: CD3-CD56+ in humans and CD3-DX5+

in mice. NK cells function in tumor surveillance, graft rejection and response to viral and

bacterial pathogens through direct killing and cytokine production. Thus, NK cells are

critical intermediates between the adaptive and innate immune systems.

Development and Subset Diversity

A detailed understanding of NK development has only recently begun to emerge.

Early studies revealed that bone marrow ablation using either 89Strontium or estradiol

treatment significantly reduced NK cell numbers, while not affecting B and T cell

numbers to the same extent, presumably because they develop external to the bone

marrow (79, 80). Transplantation studies also revealed that a functional bone marrow

niche was critical for NK development, suggesting that extrinsic signals from the bone

marrow provide key signaling events to drive NK development (81-83). Current models

postulate that NK progenitors originate and develop primarily within the bone marrow,

but may undergo subsequent functional maturation in secondary lymphoid organs such as

the spleen and lymph nodes (84).

In the earliest stage of NK development within the bone marrow, CD34+

hematopoietic stem cells (HSC) develop into early lymphoid progenitors (ELP)

expressing c-Kit and FLT3, both of which are abundant within the stromal niche. These

ELP have diminished propensity to develop into myeloid cells, but are not irreversibly

committed to the lymphoid lineage (85). Upon further differentiation, ELP become

common lymphoid progenitors (CLPs) expressing IL-7Rα with surface phenotype Lin-

(CD3-CD19-Ter119-Gr-)c-KitdimSca-1dim (86). Commitment to the NK lineage occurs

when CLP become NK progenitors (NKP) via acquisition of surface IL-2/15Rβ

expression (87, 88). This receptor along with the common γ chain imparts the ability to

respond to IL-15, which is critical for development of mature NK cells in vivo (89, 90).

This is supported by observations that IL-15 and/or IL-2 support in vitro NK development from bone marrow progenitors in both mice and humans (87, 88).

Furthermore, IL-2Rβ-deficient mice have significantly decreased numbers of circulating

NK cells, which also do not exhibit cytotoxicity in vitro (91).

In the mouse, these NKP do not exhibit cytotoxicity nor do they express mature

NK markers such as NK1.1 or DX5. NKP express several transcription factors such as

Ets1, PU.1 and Id2, which are all known regulators of early hematopoiesis (92). Defects in these transcription factors cause severe reductions in bone marrow-derived NK cells and NKP (93-96). Functional maturation of NKP is characterized by extensive

remodeling of surface receptor expression. It is at this stage where surface expression of

mature NK markers such as NK1.1 and DX5 are acquired, however, intermediates

between NKP and mature NK cells have been identified that are DX5-NK1.1+ (92).

In humans, HSC present in the bone marrow with surface phenotype

dim bright CD34 CD45RA(+)β7 have been identified as bona fide NK precursors in humans

(97). This fraction is also highly enriched in human lymph nodes, which has been suggested to be a site of functional maturation. Indeed, CD56brightCD16dim/- are highly

enriched in lymph nodes and are currently thought to be the first stage of NK maturity.

These cells have longer telomeres, increased proliferative capacity and can be

differentiated into CD56dim in vitro (98). These cells represent only ten percent of circulating NK cells and possess the ability to produce large amounts of cytokines, but a diminished capacity for cytolysis. Conversely, CD56dimCD16+ NK cells are the

predominant population in circulation, are highly cytotoxic and secrete decreased

quantities of cytokines (99).

Important differences exist between the CD56dim and CD56bright subsets. Gene

expression profiling has identified hundreds of differentially-expressed transcripts that

generate phenotypic diversity (100, 101). CD56bright NK cells express IL-2Rα, which facilitates the formation of the trimeric, high affinity IL-2 receptor imparting the ability to respond to picomolar concentrations of IL-2 (102). They also express L-selectin

(CD62L), which may facilitate differential homing and distribution among secondary lymphoid organs (103). CD56dim cells express higher levels of KIR transcripts, which

likely impart regulatory control over the enhanced cytotoxic capacity of these cells.

Furthermore, expression of CD16 (the low-affinity FcγIII receptor) is responsible for the

enhanced capacity for antibody-dependent cellular cytotoxicity (ADCC) observed in

CD56dim NK cells.

Additional subsets have been described which appear to be functionally discrete from the peripheral CD56bright and CD56dim subsets. Unique among these are the

decidual NK (dNK) cells which are found in the uterine lining during pregnancy. Gene

expression profiling demonstrates that these cells clearly represent a third subset and

differ markedly from both CD56bright and CD56dim NK cells (104). dNK are largely

uncharacterized, but may have evolved to perform specialized protective functions while

maintaining maternal-fetal tolerance. Very recently, functional intermediates directly between CD56bright and CD56dim have been described based on surface expression of

NKG2A and KIR (105). These cells develop from CD56bright NK cells and secrete

intermediate levels of IFNγ in response to stimulation. Furthermore, intermediate

populations based on surface expression of CD94 with similar functional overlap have

been described (106). These functional subsets are not entirely conserved between mice

and human, however, surface density of CD27 expression has recently been proposed to

segregate functional murine NK populations (107, 108).

Mechanisms of NK Activation and Effector Function

Upon maturation, NK cells actively patrol in the peripheral circulation and can become rapidly activated in response to a variety of environmental stimuli. NK activation is mediated through a variety of cell surface receptors that respond to pathogens, “stressed” cells or cytokines. NK cells, unlike B and T cells, do not undergo somatic hyper-mutation and clonal expansion in response to receptor stimulation.

Consequently, NK function is unperturbed in Rag-deficient mice. Instead, NK cells express a broad range of both activating and inhibitory receptors, as well as numerous cytokine receptors. NK cells are a relatively heterogeneous population and individual cells possess discrete receptor expression profiles, often referred to as the receptor repertoire.

The heterogeneity of receptor expression provides a source of functional diversity and imparts the ability to balance the activating and inhibitory signals. This balance forms the basis of the “missing self” hypothesis, originally proposed by Klaus Kärre in the 1980’s. The original hypothesis stated that NK activity was directed against “self” molecules that are altered in a manner that leaves them susceptible to NK-mediated cytotoxicity and predicted the existence of inhibitory receptors. Experimental evidence over the last several decades has largely born this out. It is now well-established that killer cell immunoglobin recpetors (KIRs) and other receptors present on the surface of

NK cells recognize MHC class I molecules to mediate inhibitory signals. Conversely, activating receptors, such as NKG2D, recognize stress-induced ligands and mediate activating stimuli culminating with release of cytotoxic granules and increased secretion of effector cytokines.

Three structurally distinct families of NK receptors are expressed on human NK

cells: KIRs, C-type lectins and natural cytotoxicity receptors. Several KIR family

members were originally cloned based on their ability to interact with HLA-B and HLA-

C to mediate an inhibitory signal (109). The KIR family contains 15 members which show a high level of polymorphism among the population and differ in the number of immunoglobin domains (two or three) and length of intracellular tail (short or long).

Although identified as inhibitory receptors, some short-tailed KIRs (e.g. KIR2DS) mediate a DAP12-dependent activation signal through intracellular tyrosine activation motifs (ITAM). Members of the C-type lectin family, typified by NKG2A, share homology to receptors for carbohydrates, but do not contain calcium binding domains.

These receptors can be either activating or inhibitory and recognize MHC Class I molecules. Nearly all of them require heterodimerization with CD94 for functionality

(110). The distantly-related NKG2D molecule functions as a homodimeric activating receptor that recognizes several stress-inducible, HLA-related molecules such as MICA and MICB and signals through the adaptor DAP10 (111, 112). The natural cytotoxicity receptors (NKp30, NKp44 and NKp46) are members of the Ig-superfamily and are expressed exclusively on NK cells. These receptors trigger activation upon recognition of their viral protein ligands, influenza hemagluttin and cytomegalovirus pp65 (113).

NK cells are also characterized by the expression of specific cytokine receptors.

All NK cell express the IL-2 receptor, with the CD56bright population preferentially

expressing the high affinity heterotrimeric form (102). NK cells also express heterodimeric receptors for IL-12 and IL-18 (Figure 6). These cytokines can be produced by mature DCs at sites of inflammation and within lymph nodes and are major

mediators of NK activation (114). NK cells also express a variety of chemokine

receptors and other ligands associated with mobility. Significant differences in the

expression of molecules such as CCR7 and L-selectin (CD62L) likely account for the

differential localization of CD56bright and CD56dim NK cells in vivo (99, 103).

Figure 6: Cytokine receptors and signaling pathways utilized by NK cells. NK cells express receptors for a variety of cytokines including, but not limited to: IL-2/IL-15, IL- 12, IL-18 and type I interferons. These molecules propagate signals through a variety of distinct and converging pathways such as MAP kinase, JAK/STAT and NFκB. Additionally, CD16 is capable of propagating activation signals upon interaction with the Fc portion of antibodies. Currently, no signaling potential of CD56 has been described and ligands have not been identified.

In response to activating stimuli NK cells proliferate, increase cytotoxicity and

produce cytokines such as IFNγ, TNFα and GM-CSF (99). IL-2 up-regulates the

expression of effector molecules and enhances natural cytotoxicity against a variety of targets (115, 116). IL-2 modulates the expression of hundreds of transcripts in distinct

temporal waves, likely indicative of direct and indirect signaling modules (117). IL-2

appears to exert differential functions in NK cells and T cells. For instance, JAK2 and

STAT4 become phosphorylated in response to IL-2 in NK cells, but not in T cells (118).

Furthermore, the IL-2 “activation signature” in NK cells as determined by gene

expression profiling is largely divergent from that obtained with CD8+ cytotoxic T

lymphocytes (117). Long term culture of NK cells with IL-2 generates lymphokine

activated killer (LAK) cells that exhibit increased cytotoxicity and enhanced proliferative

capacity.

IL-15 exhibits significant functional overlap with IL-2. The molecules are highly

homologous and both signal through the common γc receptor to control proliferation. IL-

15 can maintain NK survival in vitro (119) and appears to be required for NK cells in vivo (89, 120). This is supported by earlier observations that IL-2 deficient mice are not

devoid of NK cells (121, 122).

IL-12 and IL-18 were originally identified as cytokines capable of activating

lymphocytes. IL-12 was cloned as “NK-stimulating factor” and IL-18 as “interferon

gamma inducing factor” (123, 124). These structurally-unrelated cytokines signal

through distinct heterodimeric receptor complexes to elicit increases in IFNγ and

cytotoxicity via several mechanisms, including increased transcription, message stability

and nuclear retention (125-127). They are produced by activated macrophages, DCs and

liver-residing Kupffer cells. STAT4 is required for IL-12 signaling and NK cells derived

from Stat4-/- mice have impaired IFNγ production and display no enhancement of

cytotoxicity in response to IL-12 (128). IL-18 induces nuclear localization of NF-κB

p50/p65 which, cooperatively with AP-1, increases IFNγ and cytotoxicity (129).

Synergistic increases in cytotoxicity and IFNγ production are observed in response to co-

stimulation with IL-12 and IL-18 (130, 131).

Upon activation NK cells participate in a variety of reciprocal interactions with immune

cells. Interactions between NK cells and DCs have been extensively studied and this

interaction allows NK-derived GM-CSF and TNFα to enhance DC maturation and DC-

derived IL-12 and IL-15 to further stimulate NK activation (Figure 7) (114). The

interaction requires cell-to-cell contact and is critically-dependent upon trans-

presentation of IL-15 by DCs to NK cells (132). Similarly, NK-derived IFNγ can drive

macrophage activation, thus increasing IL-12 and IL-18 production. NK cells can also

provide early IFNγ to shape CD4+ Th1 polarization, while in turn responding to T cell-

derived IL-2 and IL-21. Thus, the early NK-dominated innate responses can be

“communicated” via cytokine stimulation to antigen-specific adaptive cells as they

expand. Virally-infected cells produce interferons which are known activators of NK

cells (133). Viral infection also leads to up-regulation of activating ligands such as

MICA and loss of MHC class I, rendering these cells more susceptible to NK receptor-

mediated cytolysis. Thus, NK cells are critical regulators of both the adaptive and innate

arms of the immune response.

Figure 7: NK cells mediate extensive crosstalk between innate and adaptive arms of the immune system. Through the release of cytokines, NK cells serve as crucial regulators of both innate and adaptive immune cells. Reciprocally, NK cells respond to cytokines and other soluble factors produced by these cells.

Part III: Histone Methylation in Transcriptional Regulation

Nucleosome Organization

The compaction of DNA into chromatin within eukaryotic nuclei represents a

remarkably elegant solution that has evolved in order to overcome the spatial constraints

associated with packaging an enormous amount of material into a relatively tiny space.

Higher order structure of DNA is regulated at numerous levels, but the primary unit of

chromatin organization present in interphase cells is the nucleosome (134, 135). Each

nucleosome consists of 146bp of DNA wrapped approximately 1.6 times around a

histone core (136). This core octamer consists of eight histone molecules, two copies

each of histones H2A, H2B, H3 and H4. Nucleosomes are separated from one another by

20-80bp with the “linker” histone H1 and other non-histone proteins interacting with the

DNA with varying affinities (137).

Structural data suggests that the nucleosome-DNA interaction has co-evolved to

impart important functionalities. The affinity between DNA and the core histone octamer

is likely dependent not only upon primary sequence motifs in the DNA, but also on an

“induced-fit” mode of interaction, which is supported by the observed deviations in

helical periodicity between free and histone-bound DNA (136). Importantly, the crystal structure verified that the N-terminal tail of the core histone molecules H3 and H4 were relatively un-structured and protruded from the histone core. Thus, these N-terminal tails are available substrates for post-translational modification within the nucleus. It had been known as early as the 1960’s that histones were subject to both acetylation and methylation (138). The identification of enzymes catalyzing these post-translational modifications in the 1990’s and early 2000’s and the recognition that these molecules were frequently present in complexes that regulate transcription led to the development of the histone code hypothesis (139).

The Histone Code Hypothesis

According to the histone code hypothesis, specific modifications of histones are associated with discrete biological outcomes. For example, acetylated histones are generally associated with transcriptional activation. The addition of acetyl groups to

histone tails effectively neutralizes the basic, positively-charged character of the histone,

resulting in a decreased affinity for the negatively-charged phosphate backbone of DNA.

This results in an increase in the accessibility of DNA to polymerases and general

transcription factors, thus histone acetylation is positively correlated with transcriptional

activity. Indeed, histone acetyltransferases co-purify with transcription activating

complexes and actively transcribed regions of the genome display increased levels of

acetylation. Conversely, histone deacetylase (HDAC) enzymes function as

transcriptional repressors and co-purify with repressive complexes such as N-CoR, SIN3,

NuRD and Co-REST (140, 141).

Distinct effects on transcriptional outcomes are associated with methylation of specific amino acid residues of histones. Unlike acetylation, the addition of a methyl group(s) does not result in an alteration of the electronegativity of the histone, but rather

provides novel docking sites for interacting effector proteins. For example, methylation of H3K4 at promoter regions results in transcriptional activation via interactions with positive regulators of transcription, while methylation of H3K27 results in recruitment of the repressive polycomb complex PRC2/3 (142, 143). Furthermore, methyl groups can be covalently attached to both lysine and arginine residues and multiple methyl moieties can exist on individual residues; up to three per lysine, two per arginine (144).

Differential methylation has distinct functional consequences. H3K9 mono-methylation, for instance, is permissive to transcription, while di- and tri-methylation are associated with transcriptional repression.

Phosphorylation of histones is often cell cycle dependent. Phosphorylation of

H3S10 by Aurora kinase increases during cell cycle progression and levels are highest in

cells undergoing mitosis (145). Phosphorylation of H3S10 blocks methylation of H3K9, imparting a further level of specificity to the histone code (146). Importantly, cell-cycle dependent abrogation of methylation (via phosphorylation) provides a mechanism for cell-cycle dependent transcriptional control. Numerous other modifications have been documented included H2A/H2B ubiquitination, ADP-ribosylation and proline isomerization. The functional consequences of multiple modifications and the effects of cross-talk among different modifications are still being elucidated as the histone code hypothesis is tested experimentally.

Methylation in Gene Silencing

The classic example of histone-mediated methylation leading to gene silencing is the phenomenon of position-effect variegation, originally described in Drosophila melanogaster. This phenotype results in variegated expression of a non-mutated allele of the white gene, yielding eye color which is variegated among red and white pigments.

This process is considered epigenetic because it is both heritable and independent of changes in the nucleotide sequence of the white gene. It is now known that both the methyltransferase Su(var)3-9 and HP1, which binds methylated histones is required for this process. Indeed, mutations within HP1 abrogate position effect variegation (147) and structural evidence confirms that HP1 specifically interacts with tri-methylated H3K9 via its chromodomain (148, 149). HP1 self-associates with other HP1 molecules and leads to

spreading of HP1 across large regions of the genome (150). Thus, methylation-specific recognition of H3K9 mediates the formation of heterochromatin, which is highly compact and remains untranscribed throughout cell divisions.

Methylation of histones can also contribute to silencing of genes in euchromatic

regions. Indeed, the localization of Su(var)3-9 and HP1 correlate with heterochromatic

foci, while G9a and other histone lysine methyltransferases do not (151). Genome-wide

analysis of H3K9me2 indicates that this mark associates with regions of transcriptional

silencing at non-heterochromatic loci (152). Thus, several levels of silencing are

mediated via histone methylation.

G9a Structure and Function

The histone methyltransferase G9a was originally cloned as an ankyrin-repeat containing protein of unknown function from the MHC class III locus (153). The G9a protein was later shown to contain 1,210 amino acids and exhibit primarily nuclear localization (154). Although it was observed to share a region of homology with the

Drosophila SET domain-containing protein trithorax, its role in transcriptional regulation was not initially appreciated. Only later was G9a shown to preferentially di-methylate histones H3 (K9 and K27) and H1 (155). Furthermore, it plays critical roles in vivo as

G9a-/- embryos are not viable and G9a-/- ES cells display global hypomethylation of

H3K9 (156). G9a interacts with the highly-related G9a-like protein (GLP) to form heterodimers, which forms the predominant, catalytically active species in vivo. GLP knockout animals also display reduced H3K9me2 levels and double knockout animals display no further abrogation, suggesting that G9a and GLP must act together in vivo

(157).

G9a interacts with a variety of molecules to mediate its functions. It is recruited by PRDM1 in order to suppress IFNB and other genes as described earlier. Interaction

with C2H2-type zinc finger transcription factors is a common modality as G9a interacts with Gfi1, NRSF, PRISM/PRDM6, ZNF200, and ZNF217 (158). Another zinc-finger containing molecule, Wiz, forms a complex with G9a-GLP and regulates the stability of the heterocomplex. Knockdown of Wiz causes a concomitant reduction in G9a protein levels (159). Thus, interactions with accessory proteins not only target G9a to specific genomic regions, but also function in a regulatory capacity.

In yeast, the histone methylation and DNA methylation machinery are functionally linked. An interesting connection also exists between G9a and DNA methylation in mammalian cells. Knockdown of DNMT1 mediates a reduction in

H3K9me2 levels in both HeLa and HCT116 tumor cell lines (160, 161). Conversely, knockdown of G9a results in global hypomethylation of DNA. However, catalytic activity is not required to restore DNA methylation levels, suggesting that the functional effect is mediated through the formation of distinct complexes (162). DNMT1 requires

UHRF1 for targeting to hemi-methylated DNA. Indeed, G9a interacts with UHRF1, which is capable of recognizing H3K9me2 (163). Thus, this bridging interaction may functionally link the histone and DNA methylation systems in mammalian cells.

Knockout studies have demonstrated crucial roles for G9a. Germline G9a- deficient embryos are not viable and die via apoptosis by day 13.5 (156). Through the use of conditional knockout strategies, numerous lineage-specific G9a functions have been identified. G9a-deficient CD4+ T cells produce diminished levels of Th2 cytokines

IL-4, IL-5 and IL-13 and CD4 conditional knockout mice succumb to helminth infections

(164). In B cells, G9a regulates the composition of the B cell receptor through inhibition of germline transcription, IgG Lambda light chain usage and V(D)J recombination (165).

Conditional knockout results in a slight reductions in both proliferative capacity of

mature B cells and plasma cell generation in response to stimulation with IL-4 and LPS

(166). Recently, G9a was shown to mediate cocaine-induced transcriptional alterations

within the nucleus accumbens in response to repeated administration of cocaine (167).

Thus, G9a exerts control over a wide array of physiological functions.

The SET domain is the catalytic domain of the molecule and is found in nearly all histone methyltransferases. G9a contains a carboxy-terminal SET domain, which is flanked by “pre” and “post” SET domains, which are enriched in Cys residues and bind zinc atoms (168). The first H3K9-specific methyltransferase to be crystallized was DIM-

5 from Neurospora crassa (169, 170). DIM-5 shares significant homology with G9a and

the structure reveals a conserved catalytic core of residues provides an aromatic cage

where H3K9 fits into a cleft and can be methylated using S-adenosyl-methionine (SAM)

which is bound to a separate module (Figure 8). Surrounding residues define the

specificity of G9a for lysine 9 of the H3 tail. Within the catalytic domain, a “Phe/Tyr

switch”, controls methylation multiplicity. Presence of phenylalanine is required for di-

and tri-methylation and mutation to tyrosine renders G9a capable of only mono-

methylation (171). Ankryin repeats present within the central portion of the molecule are

required for interaction with methylated H3 and likely serve crucial tethering functions

(172). This tethering may contribute to H3K9me2 spreading and the establishment of

large regions of histone methylation, a mechanism analogous to HP1 spreading across

heterochromatic regions.

Figure 8: Histone methyltransferase structure. The crystal structure of DIM5 reveals conserved catalytic residues required for activity. The H3 peptide fits into a binding cleft and methyl groups are transferred to the amino terminal of Lys 9 (shown in yellow) via co-factor SAM, which binds on the rear-facing portion of the molecule. (PDB: 1PEG)

CHAPTER II

MATERIALS & METHODS

Cell Lines

NKL cells were grown in RPMI, supplemented with 20% fetal bovine serum,

1mM sodium pyruvate, 1X non-essential amino acids and 100U/m IL-2. NK-92 were

grown in Alpha-MEM, supplemented with 20% fetal bovine serum, 1mM sodium

pyruvate, 1X non-essential amino acids and 100U/m IL-2. YT cells were grown in

RPMI, supplemented with 10% fetal bovine serum. NKL, NK-92 and YT cells were split every 2 days. Jurkat and THP-1 were grown in RPMI, supplemented with 10% fetal bovine serum. Jurkat and THP-1 cells were split approximately every 3 days to a density of 2 x 105/ml. U2OS cells were grown in DMEM, supplemented with 10% fetal bovine

serum and split approximately every 3 days at a 1:10 ratio. All cell lines were grown in a

humidified incubator at 37ºC and were supplemented with 1% penicillin/streptomycin.

Primary Cell Isolation

Primary human NK cells were isolated via negative selection using the human

EZSep kit (StemCell Technologies) according to the manufacturer’s instructions. Purity

was verified by flow cytometry and routinely found to be 90 – 95% CD3-CD56+CD16+.

Cells were maintained in RPMI1640 (Gibco), supplemented with 10% FBS and 1%

penicillin-steptomycin. For siRNA experiments, cells were grown in Accell Delivery

Media (Dharmacon) supplemented with 2% FBS. For stimulations, recombinant human cytokines were used at the following concentrations: IL-2 100U/mL (Peprotech), IL-12

10ng/mL (Peprotech), IL-18 100ng/mL (MBL), TNFα 20−100ng/ml (BD Biosciences),

IFNγ 20−100ng/ml (BD Biosciences), IFN-α 10−50ng/ml (Sigma).

Mice

Mice were housed in accordance with Moffitt Cancer Center guidelines by the laboratory of Esteban Celis, MD PhD. C57BL/6 mice (n=4) were immunized i.v. with

either 250 μg poly-IC (Oncovir, Inc., Washington, DC) or PBS. Mice were sacrificed at

48h post-injection and single-cell suspensions were prepared from pooled splenocytes.

Murine NK cells were isolated via negative selection using the Murine NK Enrichment

Kit (Stemcell Technologies) and maintained in RPMI1640, supplemented with 10% FBS

and 1% penicillin-steptomycin, when appropriate. Lysates were prepared from both

purified NK cells and total splenocytes and analyzed by immunoblot.

Flow Cytometry

Cells were collected and washed once with PBS. Cells (100,000/sample) were

resuspended in Flow Buffer (PBS, plus 0.5% BSA, mM EDTA). Antibody (5µl/test) was

added as per the manufacturer’s suggestion. Cells were incubated for 15 min. Samples

were collected on a Becton Dickinson FACs Calibur instrument using CellQuest software. Data was analyzed either within CellQuest or using FlowJo.

Immunoblotting

Whole cell lysates from 5 x 105 cells were prepared in lysis buffer (50mM tris pH

7.2, 150mM NaCl, 1% NP-40, 1% Na-deoxycholate, 0.1% SDS, 2mM EDTA) on ice for

20 min. Samples were then sonicated for 8 cycles (30 sec. “on”, 30 sec. “off”) in a

Diagenode Bioruptor. Lysates were centrifuged at 13,000rpm for 10 min. at 4ºC.

Soluble lysates were mixed with Laemmeli dye and heated to 85ºC for 2 min. before

separation via 8% SDS-PAGE. Gels were transferred to PVDF, blocked with 5% skim

milk and probed using antibodies directed against PRDM1 (PRDI-BF1) (Cell Signaling)

1:1000, PARP1 (Cell Signaling) 1:1000 or β-actin (Sigma) 1:10,000. Secondary

antibodies conjugated to HRP were used for detection: anti-rabbit (GE) 1:2000 and anti- mouse (GE) 1:10,000.

RNA Extraction and cDNA Synthesis

RNA was isolated using Qiagen RNeasy kit according to the manufacturer’s instructions. On-column DNase treatment was performed after the initial binding to the column. RNA was eluted in 30μl RNase-free H2O and quantified using a Nanodrop UV spectrophotometer. cDNA synthesis was carried using the qScript cDNA Synthesis kit

Quanta Biosciences per the manufacturer’s instructions. Within all experiments, starting quantities were the same across all samples and varied between 0.2μg and 1μg depending upon yield.

Microarray Hybridization and Analysis

Two micrograms of total RNA served as the mRNA source for microarray

analysis. The poly(A) RNA was specifically converted to cDNA and then amplified and

labeled with biotin as described (173). Hybridization with the biotin labeled RNA,

staining, and scanning of the chips followed the prescribed procedure outlined in the

Affymetrix technical manual. Scanned output files were visually inspected for

hybridization artifacts and then analyzed using Affymetrix GeneChip Operating Software

(GCOS). Heatmaps were generated using Heatmap Builder v1.1 using either signal

intensity or fold-change values for differentially expressed transcripts.

Enzyme Linked Immunosorbent Assay

Supernatants were collected from cells growing at 1 x 106/mL and stored at -20ºC

until use. ELISA assay was performed using 100μl volumes in triplicate using

commercial kits for IFNγ and TNFα according to the manufacturer’s instructions (E-

biosciences). For some experiments, supernatants were diluted between 1:5 and 1:1000

(depending on experimental conditions and analyte) in order to obtain wavelength

readings within the dynamic range of the assay. Plates were read at 450nm on a

Molecular Devices Spectramax Plus plate reader.

51Chromium Release Assay

NK cells isolated by negative selection were cultured for 72h prior to the assay.

K562 cells were labeled with 100μCi per 1 x106 cells for 1h, then were washed 3X with

warm PBS. NK cells were incubated with 5,000 K562 cells at various effector:target

ratios in triplicate in a total volume of 100μl for 4h. Supernatants were harvested and

c.p.m. determined using a Perkin Elmer 1470 automatic gamma counter. Percent

cytotoxicity was calculated using the following formula: (experimental – spontaneous

release)/(total – spontaneous release) x 100.

siRNA-Mediated Knockdown

The Dharmacon Accell siRNA system was used for all knockdown experiments.

Accell siRNAs are modified via a proprietary mechanism that enables passive diffusion

into target cells, eliminating the requirement for transfection. Two PRDM1-specific

siRNAs (A-0093222-16 targeting CUCUCGACAGCAAAUGGUU and A-009322-18

targeting CAAUGGAUCCUUAAGAUUU) were used at a final concentration of 500nM

each. A non-targeting control siRNA (D-001910-01) was always used in tandem at a

final concentration of 1µM. Delivery was accomplished via incubation of cells in Accell

Delivery Media, supplemented with 2% FBS and 1% pen/strep at a density of 1-2 x106

cells/ mL. Knockdown experiments were conducted at 24, 48 and 72h time points.

Transient Transfection and Luciferase Assay

Transfections were performed using either 20μg of total plasmid into 107 cells or

10ug of total plasmid into 3 x 106 Jurkat T cells. Electroporation was conducted using a

BioRad Gene Pulser II at 250V, 1070µF in 300μl RPMI1640 containing 10% FBS without antibiotics. Cells were cultured 36h at a density of 106/ml. Typically, ~60% of

cells remained viable after electorporation Luciferase activity assays were conducted

using the Promega DLR kit. Lysates were prepared from 10 x 106 or 3 x 106 cells in

volumes of 500µl or 100µl, respectively. Cells were collected and washed once in cold

PBS before lysis in passive lysis buffer. Lysates were rotated at 4ºC for 15 min., then

centrifuged at 13,000G for 10 min. Firefly luciferase values from 10µ lysate, assayed

with 50µl LARI. Stop & Glo (50µl) was then immediately added to obtain renilla

luciferase values. All readings were performed on a Turner 20/20n luminometer with 10

sec. collection times for both firefly and renilla. Data was expressed as “relative

luciferase activity” and was calculated by Firefly/renilla. Vector-only controls were set

to 100%.

Adenoviral Constructs and Transduction

Adenoviral constructs were created in the Ad5/F35 vector from previously described constructs (15). This replication-deficient adenovirus utilizes the ubiquitously-

expressed CD46 molecule to mediate high level transduction efficiency in a variety of

hematopoietic cells (174). Briefly, PRDM1a in the AdTrack vector was recombined with

Ad5/F35 to generate a bi-cistronic viral vector containing PRDM1a and GFP. Purified

stocks were obtained by infecting Ad293T cells for 48 hr. and concentrating via CsCl

banding according to standard protocols. Viral titers were calculated using the CellBio

Labs QuickTiter Adenovirus Quantitation Kit. For transduction experiments, Jurkat cells

were transduced using an MOI of 500 at a density of 1x106/ml for 44h before stimulation with PMA (1μg/ml) and PHA (10ng/ml) for 4h.

Real-time Quantitative PCR

RNA was isolated using Qiagen RNeasy kit according to the manufacturer’s instructions. Eluted RNA was DNase-treated, converted to cDNA and 1/20th of the cDNA reaction was analyzed by real-time PCR in duplicate using the BioRad iCycler (40 cycles, with primer-specific annealing temperatures between 55ºC and 60ºC). For expression analysis, data was analyzed by the ΔΔCt method, with normalization to either

18S or GAPDH. Primers were quality checked for single curve on melt curve and efficiencies between 90% and 110%. Sequences of primers and Tm values are provided below (Table 1).

Table 1: Oligonucleotide Primer Sets Gene FWD REV Tm Application BCL6 AACCTCTCGCTCCCTTTTGT GGTCTGGGGCTAATTCCTTC 60 ChIP CCR7 ATGTGTCAGTGCCTCCCATT TGAGAGAAGTGTCCCCATCC 60 ChIP CIIAp3 TCCCAACTGGTGACTGGTTA CAAGGATGCCTTCGGATG 60 ChIP CIITAp4 GGCCACAGTAGGTGCTTGGT CTCGTCCGCTGGTCATCCT 55 ChIP FOS CCACAGGGAGAGTGCAAGT CTGGGGATCCAAAAGTGAAA 55 ChIP GZMK CCTGAAAGTCCCCAAACTGA ACCACAGGTGCTCTAGGGGT 55 ChIP IBRDC3 AGGAGACAACCAGCAGCATC GTCCTTGCAGTGTGTAAACG 60 ChIP ID3 CCCATCTGTGCCTCCATATT ACCCTCTGCCCCATAAACTC 60 ChIP IFNB TGCTCTGGCACAACAGGTAG CAGGAGAGCAATTTGGAGGA 60 ChIP IFNG I CACCTGGGGTGACAAGAAAT GCCAATCACAAGAGGGAAAA 55 ChIP IFNG II CTTCCGTAGGTTTGGCTTG GGCGTAGATTTCTTCTTCCACTC 55 ChIP IFNG III TGGTGTGAAGTAAAAGTGCCTTCA CGATGAGACAGACCCATTATGC 60 ChIP IFNG IV GACTGGGTGAGGGAGATTG GGGAGTGACAGGTAGGGAGA 55 ChIP IL12RB1 SA Biosciences (part# GPH1020516(-)01A 60 ChIP

IL18BP TGGCCAGAGGGGCTAGGATGA CCCAGGCTACCCACTCCCCC 55 ChIP IL2 CCACGTGTGTCATACCCCCTGC GGGATGGGGGTTGCCCTTTGA 55 ChIP ITGAM CTTGGGTGAAGAGCTTGAGG AGGTGTGAGCCACTGCATC 55 ChIP KIR2DL4 GGTGAGAGTGGGCATTCTTG AGCCCTCAACAAGCTAAGCA 55 ChIP KLRF1 AGGGGCATAATTTCCATTCC AAAAGTTGCTTTCCCATGACA 55 ChIP KLRG1 CATTTTGTTTGTTTTGCCTTCA CATTGCCAAAAAGAGTCATCAA 55 ChIP LMO2 TGGTGACTGCTGTGGGTAAG GCCCACTCACTCTTGCTTTC 55 ChIP LRMP GCCAACGGCTGGAGGATGTGG TGGTTTGCAGGGCATCTGAGGA 55 ChIP

LTB IV TCCATCCCACAAGATTCCAG AGGTTTCCAATGTGGTTTGC 55 ChIP MKI67 TACGGAAGTCTGGAAGGAAC CTGGGTTTACAGGCGGTGA 55 ChIP MB exon2 Diagenode (part# pp-1006-050) 60 ChIP

PAX5 GATTTGGGCGAGAACAGGAC GAAGGCACCGTGAAATGATTA 55 ChIP PCNA CTGAGGAGCCACCATAAAGC CGCCTCTTTGACTCCTGAAC 55 ChIP PTK2 GCTAACCACAAGCCAGGAAG GCTCCTAATGCAATGACG 55 ChIP SELL GGTGGGGAAAGAGGAAAGAG GCTGTGCTGCAGGTAGACTG 60 ChIP SLAMF7 GGCCAGGAAAGTGAAACAGA TCAGAGCTTAAGTTTGCCATGT 55 ChIP SOCS2 GGCTCACCTGGGCCACACAC CCACCACCACACACAGCACCC 55 ChIP SOX4 TGTTTGGGCTATGCAGGATT GAAGGCAGGGAAGAGGACTT 60 ChIP STAMBP AGGCACACTGGGGCCTCAACT TGGTTGATGGGCAGAAGCCCT 55 ChIP STAT6 CGTGTTCCCCCACTCGGCAC GCCCGCTGTTTCCGGCTTCT 55 ChIP TNF TACCGCTTCCTCCAGATGAG TGCTGGCTGGGTGTGCCAA 55 ChIP TNF I CAACCTCCGAAACTGAGA AGGAAGGAAGGGTAAGACTG 55 ChIP TNF II TACCGCTTCCTCCAGATGAG TGCTGGCTGGGTGTGCCAA 60 ChIP TNF III TGCATCTGCCTCTTCTTGTG GGTCCAGGGATCTTAAAGC 55 ChIP TNFRSF10B GGGGCGTTCTGTCCCCGTTG AACCCCGCAATCTCTGCGCC 55 ChIP UBR2 AGCCGTGTGCTCCTCCGACT GCCTTCTTGGTGCCTGGGGC ChIP

VAV3 ACAGTTCCGAGCCTGACAGT TGAGTGAGAGTGCGTGTGTG 55 ChIP 18S GTAACCCGTTGAACCCCATT CCATCCAATCGGTAGTAGCG 55 exp DAP10 GGCTGCAGCTCAGACGAC AGGAGCGGCAGAGAGG 58 exp IFNG GAAAAGCTGACTAATTATTCGGTAAC GTTCAGCCATCACTTGGATGA 55 exp LTA CTGTCTGGCTGAGGATTTCA CCCTCTCTCCATCCTCCATA 58 exp PRDM1a TCAAACTCAGCCTCTGTCCA TCCAGCACTGTGAGGTTTCA 55 exp PRDM1b CCCGAACATGAAAAGACGAT ATAGCGCATCCAGTTGCTTT 55 exp PTK2 GAGCCTTCTCTTCCTCCAG TTGTCCTCCTACCCTCTACA 55 exp TNF TCCTTCAGACACCCTCAACC AGGCCCCAGTTTGAATTCTT 58 exp STAT6 ACTTTTTCTGGGGGCATCTT AGAAGACAGCAGAGGGGTTG exp

TNFRSF10B CTCTCTGCTGGGGAGCTAGG AAGACCCTTGTGCTCGTTGT exp

BCL10 TTGCACGTAGATGATCAAAATGT CGAGGAGGACCTCACTGAA exp

SELL CTAATTGCGTGCCACTCTGA CCCCCTCACACACTTTGTCT exp

LRMP GTTGTTGAATTGCATCTGGC GCACTTCAGCAGAAGACAAGG exp JAK2 CCATTCCCATGCAGAGTCTT CAGGCAACAGGAACAAGA exp ITGA4 TGGCTGTCTGGAAAGTGTGA AGACGTGCGAACAGCTCC exp IL18BP AGGCCACACAGGATAAGCTC CAGCAGCTAAGCAGTGTCCA exp KLRD1 CGGTGTGCTCCTCACTGTAA AGCCTGCTTCAGCTTCAAAA exp IBRDC3 CTCACAAGTTAGCTTCGGGC TACCTAGCCTCGGACCCC exp FBXO9 CAGACGCCATATTTCAGGGT TCCGATGGGTGGTGTCTAGT exp UBR2 GCACTCCATGCACAAAACAC CCTTCTCATCTTTGTGGTCGT exp DDX17 AAGTGGTCCGGTAACGAGAA GAGGCCAATCAGGCTATCAA exp

Chromatin Immunoprecipitation

Cells (20 x 106) were crosslinked with 1% formaldehyde for 10 min. at room

temperature, then quenched via addition of 5M glycine to a final concentration of 0.125M

for 5 min. Cells were pelleted and washed in PBS. Nuclei were isolated in TX100/NP40 buffer (10mM Tris pH8.1, 10mM EDTA, 0.5mM EGTA, 0.25% Triton-X-100, 0.5%

NP40, 1mM PMSF, 0.5X Roche protease inhibitor cocktail) and washed in salt wash buffer (10mM Tris pH8.1, 1mM EDTA, 0.5mM EGTA, 200nM NaCl, 1mM PMSF,

0.25X Roche protease inhibitor cocktail). After 10 min. incubation at 4ºC, nuclei were pelleted and resuspended in 600µL (2 x 300µL) sonication buffer (10mM Tris pH8.1,

1mM EDTA, 0.5mM EGTA, 1% SDS, 1mM PMSF, 1X Roche protease inhibitor cocktail) and sonicated for 8 cycles (30 sec. “on”, 30 sec. “off”) in a Diagenode

Bioruptor. Samples were centrifuged at 13,000rpm for 10 min. at 4ºC, then snap frozen

6 in LN2 and stored at -80ºC. 4.5 x 10 cell equivalents (135µl) were used for each

immunoprecipitation reaction. Primary antibodies were used at 0.5μg in each 900μl

reaction and incubated overnight. Antibodies used were: PRDM1 (PRDI-BF1) (Cell

Signaling) and normal rabbit IgG (Upstate). Immune complexes were captured with protein A/G beads (Santa Cruz) and washed five successive times. First and second washes were TSE150 (20mM Tris pH8.1, 150mM NaCl, 2mM EDTA, 0.1% SDS, 1%

Triton-X-100); third and fourth washes were TSE500 (20mM Tris pH8.1, 500mM NaCl,

2mM EDTA, 0.1% SDS, 1% Triton-X-100); fifth wash was LiCL250 (10mM Tris pH8.1,

250mM LiCl, 1% NP40, 1% sodium deoxycholate, 1mM EDTA). DNA was eluted from beads using two successive elutions of 150µl elution buffer (10mM tris pH8.1, 1% SDS,

1mM EDTA). Eluted DNA was column purified (Qiagen) after reversal of crosslinks (4

hr. 65ºC), RNase treatment and proteinase K treatment. PCR was performed using 1.5μl of eluted DNA (~1/40th) in duplicate. Primers designed against the second exon of

Myoglobin (MB) (Diagenode) were used as a negative control locus. Percent input was

calculated by linearization of ΔCt (CtIP – Ct1%input) for both specific and IgG samples.

ChIP-chip

For each of three independent experiments, chromatin was prepared as described above from 100 x 106 NKL cells at a density of ~5 x 105/ml. Ten immunoprecipitations

for PRDM1 were performed, with one IgG control immunoprecipitation and

immunoprecipitated DNA obtained as described above. Quality was verified by

performing qPCR for PCNA and CIITApIV as positive controls and as positive controls

and MB and HLA-DR as negative controls. PRDM1 immunoprecipitations were pooled,

quantified using a Nanodrop 1000 spectrophotometer and amplified using the Sigma

Whole Genome Amplification kits according to the manufacturer’s instructions. Ten ng

were used to perform amplification in two steps using the WGA2 and WGA3 kits. In

both amplifications, only 14 cycles were used to prevent non-linear amplification. For

the second amplification, 20 ng were amplified in duplicate using dNTP mix containing

dUTP (provided by Moffitt Microarray Core Faciliy). An additional quality check was

performed using a 1:100 dilution of the amplified DNA (using primers to PCNA,

CIITApIV, MB and HLA-DR) to verify >10-fold enrichment. Six µg DNA for both

PRDM1 immunoprecipitation and input samples were submitted to the Moffitt

Microarray Core Facility for hybridization to Affymetrix GeneChip Promoter 1.0R Tiling

Arrays. Separate arrays were run for each sample. Raw hybridization data was provided

as .cel files by the Core Facility.

ChIP-chip Data Analysis

Raw hybridization contained in .cel files was analyzed using two methods:

CisGenome and Model-based Analysis of Tiling Arrays (MAT) (175, 176). Both

CisGenome and MAT analysis programs are freely available for download

(http://www.biostat.jhsph.edu/~hji/cisgenome/ and http://liulab.dfci.harvard.edu/MAT/).

For Cisgenome, .cel files were added and normalized via quantile normalization. Peaks

were annotated using default settings (PRDM1 > input; maximum gap < 300bp, peak >

100bp and peak contains > 5 continuous probes) and mapped to human genome version

18 (hg_18). Nearest single genes associated with peaks of enrichment were identified

within 15kb upstream of the transcription start site (TSS), 15kb downstream of the

transcription end site (TES) and within each gene. Based on these criteria, up to three

genes could be identified for each peak. For MAT, .cel files were first converted to .txt

files, input into the Linux-based MAT platform and peaks were annotated using settings

identical to those used for CisGenome. This analysis generated a .bed file containing the

annotated peaks which was then converted to a CisGenome file (.cod) and used to

identify the nearest genes as stated above. A total of six analyses were performed, two

for each of the three experiments.

DNA Constructs

Flag-PRDM1

The full-length cDNA encoding PRDM1α was PCR cloned into pcDNA3.1,

modified to contain a FLAG epitope sequence. This plasmid has been previously

described (5).

IFNG Promoter Luciferase Constructs

A fragment of the human IFNG promoter (-507 - +120) was PCR cloned from

human genomic DNA into pcr2.1. The fragment was then subcloned into pGL3basic

using the HindIII and SmaI restrictions sites to generate pGL3-IFNγ_WT. pGL3-

IFNγ_mut was obtained via site-directed mutagenesis of the -254 site, changing residues

AAAAGT to TCTAGA, which created a novel XbaI site (Mutagenex, Inc). pGL3-

IFNγ_∆ was obtained by ligation of an XbaI-XbaI fragment obtained from pGL3-

IFNγ_mut with the XbaI-NheI fragment of the pGL3-Basic vector.

PTK2 Promoter Constructs

Plasmids encoding luciferase under control of the PTK2 promoter were obtained from William Cance (Roswell Park, Buffalo, NY). Three plasmids were obtained with progressive deletions of the promoter (spanning from +47 to -564, -1020 and -1173 realtive to the transcriptional start site) cloned upstream of luciferase in pGL3 as described (177). An additional construct pGL3-PTK2-1400 was created by PCR amplifying a genomic 230bp and subcloning into the 1173.

Recombinant G9a Expression and Purification

Using a portion of the murine G9a coding sequence cloned in-frame into pGEX3, a catalytically-active fragment (amino acids 1 -439) was purified from bacterial lysates.

E.coli BL21 cells were inoculated and grown in large scale cultures (500ml) for 2-3h to an OD450 of 0.5 – 0.6. Cultures were induced using 0.5 mM IPTG and incubated at 30ºC for 90 min. Cells were collected, washed with cold PBS and lysed in bacterial RIPA

(20mM Tris pH 7.5, 500mM NaCl, 5mM EDTA, 1% NP-40, 0.5% Na-desoxycholate, 1X

Roche protease inhibitor cocktail, 5mg/ml lysozyme, 1mM PMSF) at 10mL per L of

culture. Viscosity was disrupted via 2 freeze/thaw cycles using a L2 bath and 6 pulses of

30s sonication using a Branson Sonifer set at 30% in the cold room. Lysates were

cleared by centrifugation at 13,000xG. Supernatants were incubated with anti- glutathione beads (800µl/10mL) and rotated for 2h at 4ºC. HMT-GST fusion proteins were eluted in 4 fractions with glutathione elution buffer (50mM Tris pH 8.0, 150mM

NaCl, 15mM glutathione, 1X Roche protease inhibitor cocktail, 1mM PMSF). Fractions were stored in 15% glycerol at -20ºC.

Radioactive In Vitro Histone Methyltransferase Assay

Recombinant G9a was incubated with histone substrate (4µg core histones, isolated from calf thymus) and 1.5 3H-SAM (55-85 Ci/mmol) in a 40µl reaction containing the following: 50mM Tris pH 9.0, 5mM Mg Cl2, 0.5mM DTT, 50µM PMSF.

Reactions were incubated at 30ºC for 1h and stopped by addition of 10µl 5X Laemmeli

dye. Reaction products were resolved on 16% SDS-PAGE gels and transferred to PVDF

membranes. Membranes were exposed to high sensitivity Kodak MS film at -80ºC for 2-

14 days using an enhancer screen. Equal loading was monitored by amido black staining of membranes.

MALDI-TOF Mass Spectrometry Analysis

Recombinant G9a was incubated with 1uM synthetic human histone H3 peptide

(N-terminal residues 1-20, ARTKQTARKSTGGKAPRKQL) in a 50µl reaction containing the following: 50mM Tris pH 9.0, 5mM Mg Cl2, 0.5mM DTT, 50µM PMSF and 1mM SAM. Reactions were initiated via addition of two equally buffered “master” mixes, the first containing SAM and enzyme (40µl), the second containing peptide

(10µl). Reactions were carried out in 96-well plates and incubated 30 min. to 2h at room temperature with slow rotation. Reactions were stopped via the addition of tri-fluoro- acetic acid to a final concentration of 0.25%. Aliquots (0.2µl) were spotted on 192- sample analytical plates. Dried analyte spots were overlaid with 0.7µl α-cyano-4- hydroxycinnamic acid matrix in 55% aqueous acetonitrile. Matrix-assisted laser desorption ionization – time-of-flight mass spectrometry (MALDI-TOF MS) was performed with a tandem time-of-flight instrument (4700, Applied Biosystems,

Framingham, MA) in positive ion mode. Methylation was detected via shifts in the m/z spectrum for the H3 peptide. Fragments corresponding to Δm/z 2183 (unmodified), Δm/z

2197 (mono-methylation), Δm/z 2211 (di-methylation) and Δm/z 2225 (tri-methylation) were each quantified and used to assess the degree of modification based on signal-to- noise ratio for each peak. MALDI-TOF MS/MS was performed initially to verify that

Lysine at position 9 was the target of methylation.

Small Molecule Screening to Identify Methyltransferase Inhibitors

Methyltransferase activity assays were carried out as described above using small

compound libraries arrayed across 96-well plates in 1µl DMSO at 5mM. Reaction volumes of 49µl generated a final small molecule concentration of 100µM. Each plate contained 80 compounds with the first and last columns used for control wells. For each plate, 8 wells (A1, B1, C1, D1, E12, F12, G12, H12) contained no enzyme controls and 8 wells (E1, F1, G1, H1, A12, B12, C12, D12) contained DMSO-only controls. A

Beckman Biomek FX liquid handler was used to robotically pipette reaction components into 96-well plates. Reactions were set up as described above, with 39µl containing the enzyme added first, followed by addition of 10µl of solution containing the peptide substrate. Thus, the enzyme was exposed to the small molecule before addition of the peptide substrate. After each addition a “mixing” step was added, whereby 5µl was pipetted up and down 3 times. Reactions were stopped by the addition of TFA as above, bringing the total volume to 100µl. 0.2µl aliquots were analyzed by MALDI-TOF MS as described above.

Average degree of methylation for DMSO-only controls (n =8) was considered

baseline methylation for each plate. Inhibitors were identified as “hit” if they resulted in

a decrease of > 2 standard deviations in degree of methylation.

Statistical Analyses

For statistical analyses, two-tailed paired t-tests were used. P-values less than

0.05 were considered significant. All calculations were performed in Microsoft Excel.

CHAPTER III

PRDM1 REGULATES EFFECTOR CYTOKINE PRODCUTION IN NK CELLS

Introduction

NK cells play critical functions in both innate and adaptive immunity. Although

these lymphocytes were initially identified by their ability to lyse leukemia cells in a non-

MHC restricted manner, subsequent studies have highlighted their role in cytokine

production. In response to activating stimuli, NK cells proliferate, increase cytotoxicity

and produce cytokines such as IFNγ, TNFα and GM-CSF (99). IL-2 up-regulates the expression of effector molecules and enhances natural cytotoxicity against a variety of targets. Furthermore, IL-2 and IL-15 both signal through the common γc receptor to

control proliferation, with IL-15 being uniquely required for survival in vivo (119). IL-12

and IL-18 signal through distinct heterodimeric receptor complexes to elicit increases in

IFNγ via several mechanisms, including increased transcription, message stability and

nuclear retention (125-127). Synergistic increases in cytotoxicity and IFNγ production

are observed in response to co-stimulation with IL-12 and IL-18 (130, 131).

Cytokine-mediated activation of NK cells proceeds through several well-

characterized nuclear transcription factors, many of which are functionally conserved

between T and NK lineages (178). However, relatively few negative regulators of activation-induced transcription have been identified in NK cells. ATF3 was recently

shown to down-regulate IFNγ levels and ATF-/- mice exhibit increased resistance to

MCMV infection (179). The transcription factor H2.0-like homeobox (HLX) negatively regulates IFNγ production, primarily through degradation of phosphorylated STAT4, not direct DNA-binding activity (180).

PR Domain containing 1, with Zinc Finger Domain 1 (PRDM1, also BLIMP1 or

PRDI-BF1) is a transcriptional repressor encoded by the PRDM1 gene on chromosome

6q21. It was originally identified as a post-induction suppressor of IFNB in virally- infected osteosarcoma cells (1). Subsequent work revealed a pivotal role in the terminal differentiation of antibody-producing plasma cells (2). PRDM1 exerts its repressive functions through recruitment of histone-modifying enzymes (HDAC2, G9a, PRMT5 and

LSD1) and Groucho corepressors (15, 19, 181). Through silencing of direct (cMyc,

CIITA, Pax5) and indirect targets, PRDM1 is a master regulator of terminal differentiation of B-lymphocytes, mediating cell cycle exit, repression of early B-cell factors and induction of immunoglobulin secretion (46, 47).

More recently, a role for PRDM1 in T lymphocytes has emerged. PRDM1 is expressed in both CD4+ and CD8+ T cell lineages and is critical for maintenance of

homeostasis. Conditional knockout in T lymphocytes leads to increased effector

populations, resulting in severe colitis (64, 65). Upon activation, an auto-regulatory loop

exists whereby IL-2 induces PRDM1 expression which in turn negatively regulates IL-2

transcription (67, 68). During CD4 polarization, PRDM1 is preferentially expressed in

Th2 cells and reinforces commitment to this lineage through repression of Ifng, cfos and

tbx21 (66, 67). Within the CD8 lineage, PRDM1 is expressed at higher levels in

exhausted subsets and promotes acquisition of the effector phenotype through

suppr ession of memory potential (71-73). Thus, in addition to well-characterized B cell specific functions, PRDM1 is also a critical regulator of T lymphocytes. The work presented herein provides the first functional description of PRDM1 in NK cells.

Results

Human NK cells alter expression of multiple effector molecules and transcription factors in response to cytokine stimulation

Natural killer cells are well known for their ability to up-regulate both the production of effector cytokines and cytotoxic potential in response to IL-2 and other cytokines such as IL-12, IL-15, and IL-18. However, the roles of sequence-specific,

DNA-binding transcription factors in the modulation of NK activity are incompletely characterized. To directly address this, global gene expression profiling was conducted using microarray analysis of RNA from NK cells isolated from peripheral blood of two

healthy donors either immediately upon isolation or after 24h stimulation in the presence

of IL-2, IL-12 and IL-18. Purity of the NK population was determined by flow

cytometry to be greater than 92% in ungated populations and was further confirmed by

the absence of significant signals for transcripts associated with B cells, T cells, and

monocytes in the microarray (Figure 9A and 9B). Biological variance among donors

was quite low, yielding R2 value of .967 and .970 between donors in freshly isolated and stimulated samples, respectively (Figure 9C). In total, 541 genes were increased 3-fold

or more in both donors, while 609 genes were decreased 3-fold or more following 24h

stimulation (Figure 10 and Appendix I).

Figure 9: Cells utilized for microarray analysis display NK surface phenotype and minimal donor-to-donor variability. A) FACS analysis of negatively-selected NK cells obtained from PBMCs of two healthy donors. B) Average signal intensity obtained from microarray analysis of RNA obtained from freshly isolated samples demonstrating low to undetectable levels of transcripts associated with non-NK immune lineages. C) Regression analysis of signal intensity values of resting and stimulated conditions for each donor. R2 values indicate high correlation between donors.

Figure 10: Top 75 up- and down-regulated transcripts upon stimulation. Heat-map depicting top 75 genes sorted by signal intensity found to be up- or down-regulated at least 3-fold in response to 24h stimulation with IL-2 (100U/mL), IL-12 (10ng/mL) and IL-18 (100ng/mL) relative to time 0h. Pixel density (highest values in each are pure black, lowest are white) represents average hybridization signal intensity from two donors after and before stimulation for increased and decreased genes, respectively. Fold changes are shown in parentheses. Presentation of expression data in this manner allows appreciation of both magnitude of expression and increases/decreases in response to stimulation.

In response to stimulation, large transcriptional changes were observed in genes

encoding effector molecules. As previously reported, IFNG mRNA is detectable in

freshly isolated NK cells and is dramatically up-regulated upon stimulation (Figure 11).

Transcripts encoding the secreted cytokines GM-CSF and TNFα were also highly up- regulated. Importantly, transcripts for cytokines not produced by NK cells (i.e. IL2) were

not detected, further indicating purity of cells used in the analysis. Decreased levels of

TGFB1, which is constitutively expressed in NK cells, were also observed in response to

stimulation, consistent with its role as a negative regulator of activation/proliferation.

Although NK cells are known to produce IL-10, transcripts of IL10 were not detected.

Figure 11: Up- and down-regulated effector molecules in response to stimulation. Signal intensities were averaged from each donor before and after 24h stimulation with IL-2 (100U/mL), IL-12 (10ng/mL) and IL-18 (100ng/mL). Shown are selected secreted cytokines and tumor necrosis factor family members.

Numerous TNF family members were also found to be transcriptionally up-

regulated in response to stimulation. When present on the cell surface these molecules

facilitate induction of apoptotic programs upon engagement of ligands expressed on

target cells. FAS and TRAIL are both capable of mediating apoptosis via ligation with

their receptors, present on target cells. 4-1BB (CD137) is a co-stimulatory molecule whose expression increases upon activation and mediates activating signals upon

engagement with its cognate ligand (182). Similarly, OX40-OX40L interactions control effector function via interactions with T cells (183). Thus, stimulation alters the functional capacity of NK cells via transcriptional up-regulation of secreted cytokines and surface-bound effector molecules.

In order to carry out effector function, NK cells must be able to respond to signals. NK cells respond to a variety of cytokines produced by T cells, DCs and macrophages. IL-2 and IL-15 are both crucial regulators of proliferation, which signal through shared components of a heterodimeric or trimeric receptor. In response to stimulation, NK cells up-regulate transcription of the IL-2 receptor α (encoded by IL2RA)

(Figure 12). The trimeric, high affinity IL-2 receptor is composed of the α, β and γ subunits and imparts

Figure 12: Up- and down-regulated cytokine receptor molecules in response to stimulation. Signal intensities were averaged from each donor before and after 24h stimulation with IL-2 (100U/mL), IL-12 (10ng/mL) and IL-18 (100ng/mL). Shown are selected cytokine receptor molecules.

the ability to respond to very low qua ntities of IL-2. Thus, in response to stimulation NK cells quickly acquire increased sensitivity to IL-2 and are poised to rapidly proliferate.

Similarly, the IL-15 receptor is up-regulated and increases the ability to respond to IL-15, which is often presented in trans by DCs (184). IFNγ has been reported to increase NK cytotoxicity (185), yet the gene encoding one half of the dimeric IFNγ receptor (IFNGR1) is significantly down-regulated in response to stimulation. Such a mechanism likely prevents over-stimulation via an autocrine feedback loop involving NK-derived IFNγ.

NK activity is largely controlled by the repertoire of activating and inhibitory receptors present on the cell surface. NK receptor families are broadly defined by their shared structural motifs and propagate activating or inhibitory signals based upon the presence of specific intracellular motifs. Transcripts for the killer cell immunoglobin-like receptors (KIRs) are uniformly down-regulated in response to stimulation (Figure 13).

Figure 13: Up- and down-regulated NK receptors in response to stimulation. Signal intensities were averaged from each donor before and after 24h stimulation with IL-2 (100U/mL), IL-12 (10ng/mL) and IL-18 (100ng/mL). Shown are selected members of the KIR family, the killer-type lectins and the natural cytotoxicity receptors. KLRA1 = psuedogene, KLRB1 = CD161, KLRC1 = NKG2A, KLRC3 = NKG2E, KLRC4 = NKG2F, KLRD1 = CD94, KLRF1 = NKp80, KLRG1 = MAFA-L, KLRK1 = NKG2D, NCR1 = NKp46, NCR2 = NKp44, NCR3 = NKp30.

The killer lectin-like (KLRs) and natural cytotoxicity-triggering receptors (NCRs) remain

largely unchanged with the exception of KLRF1, which encodes NKp80 and is

significantly down-regulated upon activation. The inhibitory receptor NKG2A, which is

encoded by KLRC1, is the only NK receptor that is transcriptionally up-regulated in response to stimulation.

The SLAM family consists of nine members all of which share homology to CD2.

These molecules are expressed on a variety of lymphocytes. In NK cells these molecules

are variably expressed with CD48 (SLAMF2) and CRACC (SLAMF7) strongly up-

regulated in response to stimulation (Figure 14). CD48 is a ligand for 2B4 (SLAMF4),

which is highly expressed in resting NK cells, but decreases upon stimulation.

Figure 14: Up- and down-regulated SLAM family members in response to stimulation. Signal intensities were averaged from each donor before and after 24h stimulation with IL-2 (100U/mL), IL-12 (10ng/mL) and IL-18 (100ng/mL). Shown are SLAM family members. SLAM = signaling lymphocyte activation molecule. CD48 = SLAMF2, LY-9 = SLAMF3, 2B4 = SLAMF4, CD84 = SLAMF5, NTB-A = SLAMF6, CRACC = SLAMF7, BLAME = SLAMF8, CD84-H1 = SLAMF9.

Modulation of these molecules has the potential to propagate activating signals via cell- contact. Interestingly, CRACC has been shown to function as an activating receptor in

NK cells, while playing an inhibitory role in CD4+ T cells (186). It should be noted that

surface level protein expression may not directly correlate with transcriptional changes.

Thus, conclusions about function must be made cautiously.

Transcriptional control of NK effector function is poorly defined relative to other

lymphocyte lineages. Unlike B and T cells, very few NK-specific transcription factors

have been defined. Of the top 150 transcripts found to be modulated after stimulation,

nearly 10% were sequence-specific DNA-binding transcription factors (Figure 15A). A

variety of transcription factors associated with lymphocyte function are induced upon

Figure 15: Up- and down-regulated transcription factors in response to stimulation. A) Heat-map depicting top 15 up- and down-regulated DNA-binding transcription factors as in Figure 10. B) Confirmation of up-regulation of six transcription factors in response to IL-2 (100U/mL), IL-12 (10ng/mL) and IL-18 (100ng/mL) by qRT-PCR. Data was normalized using 18S as control gene, error bars represent s.d. from three independent donors. C) Confirmation of NFκB (p105/p50) and STAT5 protein induction in response to IL-2, IL-12 and IL-18 by immunoblot. D) Confirmation of IRF8 protein induction in response to IL-2, IL-12 and IL-18 by immunoblot.

stimulation, however, no data exist regarding their function in NK cells. BATF encodes the B cell Activating Transcription Factor, which dimerizes with JunB, but lacks transactivation activity. Thus, this factor whose expression is limited to hematopoietic cells has the potential to abrogate AP-1 function via competition for Fos (187, 188).

EGR2, a zinc-finger transcription factor, is required for both positive selection of

+ - + thymocytes and differentiation of IL-10 producing CD4 CD25 LAG3 Treg cells (189,

190). EOMES encodes the eomesodermin transcription factor which is critical for

effector function of CD8+ T cells (191). More recently, Eomes has been found to play a

role in murine uterine NK cells, which play crucial IFNγ-dependent functions during

pregnancy (192). NFKB1 encodes the p105 subunit of NFκB. In response to IL-12 and

IL-18, the p105 subunit is cleaved to form the p50 subunit, which dimerizes and enters

the nucleus. Thus, in response to stimulation a positive feedback exists where p50 levels

are increased via up-regulation of total p105 (Figure 15C). Similarly, STAT5A is up-

regulated in response to stimulation, resulting in higher levels of total STAT5, a major

mediator of IL-2 signaling. Finally, IRF8 levels are markedly increased in response to

stimulation (Figure 15D). This factor plays critical roles within the myeloid lineage, yet

has not been described in NK cells. Interestingly, the strong up-regulation observed for

STAT5A and IRF8 are quite specific, because other family members do not change

nearly as dramatically (Figures 16 and 17). Thus, NK cells utilize a variety of

transcription factors to modulate effector function.

Figure 16: Up- and down-regulated IRF family members in response to stimulation. Signal intensities were averaged from each donor before and after 24h stimulation with IL-2 (100U/mL), IL-12 (10ng/mL) and IL-18 (100ng/mL). Shown are IRF family members (IRF1-IRF8).

Figure 17: Up- and down-regulated STAT family members in response to stimulation. Signal intensities were averaged from each donor before and after 24 hr stimulation with IL-2 (100U/mL), IL-12 (10ng/mL) and IL-18 (100ng/mL). Shown are STAT family members (STAT1-STAT6).

Cytokine stimulation induces multiple PRDM1 isoforms in human NK cells

Among the up-regulated transcription factors, the transcriptional repressor

PRDM1 was increased in response to stimulation. PRDM1 has not been previously identified in NK cells, however, within the immune system PRDM1 plays crucial roles in both cell fate decisions and regulation of homeostasis. The role of PRDM1 in the terminal differentiation of mature B cells into immunoglobin-secreting CD38+ plasma

cells is well established. More recently, roles for PRDM1 in the maintenance of

homeostasis and effector versus memory lineage commitment in T lymphocytes have

been reported (71-73). The presence of PRDM1 and up-regulation upon activation

suggest that it may also have key functional roles in NK cells.

In order to directly establish PRDM1 activation, human NK cells were isolated by

negative selection and stimulated with multiple combinations of IL-2, IL-12 and IL-18.

Immunoblot analysis reveals that freshly isolated and unstimulated NK cells have barely

detectable levels of PRDM1 protein (Figure 18A). Stimulation with IL-2 or a combination of IL-12 and IL-18 results in increased levels of PRDM1 protein, which is markedly enhanced by co-stimulation with all three cytokines. Cell lines derived from malignant NK tumors constitutively express PRDM1 and exhibit distinct isoform expression patterns (Figure 18B). YT and NK92 cell lines exclusively express the larger

α-isoform, while NKL most closely resemble the expression pattern observed in primary human NKs. Consistent with immunobloting experiments, analysis of the α and β isoforms mRNA levels in primary NK cells indicates that cytokine stimulation up- regulates both isoforms and that PRDM1β mRNA is present at approximately 20-fold

higher levels than PRDM1α (Figure 18C).

Figure 18: Cytokine stimulation induces PRDM1 isoforms in human NK cells. A) Western blot analysis of purified human NK cells stimulated for 24h with IL-2 (100U/mL), IL-12 (10ng/mL) and IL-18 (100ng/mL) or combination. Blot shown is representative of numerous experiments. B) Immunoblot analysis of lysates prepared from human NK cell lines demonstrates differential expression of PRDM1 isoforms. C) RT-PCR analysis of cDNA synthesized from freshly isolated or 24h stimulated NK cells. Expression values were calculated using the ΔΔCt method with 18S as the control gene. Upon linearization of Ct values, average Time 0 PRDM1α was arbitrarily set to “1” to account for basal differences among donors and the consistently higher expression levels of PRDM1β. Error bars represent SD from at least three biologically independent samples isolated from different donors at different times. Paired two-tailed t-tests were conducted, with (∗) representing p < .05 p-values were 0.016 and 0.023 for PRDM1α and PRDM1β, respectively.

Human NK cell lines constitutively express high levels of PRDM1. In order to

assess changes in response to cytokine stimulation, two NK cell lines were stimulated for

24h or 48h with combinations of IL-2, IL-12 and IL-18 (Figure 19). NKL cells exhibit relatively static expression of PRDM1, although stimulation with IL-2, IL-12 and IL-18 results in a moderate increase in PRDM1β. PRDM1 expression in NK-92 cells, unlike

NKL cells, is highly dependent on IL-2. In the absence of IL-2, PRDM1 expression is

significantly diminished by 24h and is nearly absent by 48h. These data indicate that NK

cell lines do not recapitulate the cytokine responsiveness observed in primary NK cells.

In terms of PRDM1 expression, NKL and NK-92 cells differ markedly in basal isoforms patterns and response to IL-2 withdrawal.

Figure 19: Human NK cells lines do not recapitulate dynamics of cytokine-induced expression observed in primary NK cells. NKL cells (top) or NK-92 (bottom) were stimulated with various combinations of IL-2 (100U/mL), IL-12 (10ng/mL) and IL-18 (100ng/mL) for 24h or 48h. Immunoblot analysis was performed for PRDM1.

Primary NK cells express three distinct molecular weight forms of PRDM1. The largest molecular weight band corresponds to the full-length PRDM1α isoform, however this protein is significantly underrepresented compared to the smaller PRDM1 proteins.

The predominant and smallest form corresponds to the PRDM1β originally characterized in multiple myeloma tumor cells and shown to have partially reduced repressive ability in reporter assays (5). In addition, immunoblot analysis detected an intermediate-sized

isoform in primary NK cells. This protein migrating at ~85kD likely represents the

human homologue of a previously-described murine splice variant, prdm1Δexon7 (6).

Indeed, RT-PCR analysis using primers spanning the analogous exon in humans (exon 6)

confirms the presence of this splice variant (Figure 20).

Figure 20: Identification of a PRDM1 splice variant lacking exon 6 in NKL cells. NKL cells express a PRDM1 transcript variant corresponding to deletion of exon 6. RT- PCR was carried out using primers specific for the fifth and seventh exons of the human PRDM1 gene. Over-expression of cDNA obtained from either U2OS cells expressing a flag-tagged PRDM1 construct or NK-92 cells yielded the expected 311bp product, while cDNA obtained from NKL cells yielded both a 311bp and 182bp products.

Human NK cells can be divided into two subsets based on the surface density of

CD56 and the presence of CD16. The CD56brightCD16dim/- subset represents ~10% of the

human peripheral NK cell compartment and has been suggested to play a immuno-

modulatory role based on the increased ability to produce cytokines; the CD56dimCD16+

subset represents ~90% of human peripheral NK cells and is considered the primary

cytotoxic subset. In order to determine whether PRDM1 is restricted to specific subsets

or if it is broadly expressed, human NK cells were isolated from healthy donors via negative selection followed by flow cytometric sorting into CD56bright and CD56dim

subsets. These cells were then stimulated with IL-12 and IL-18 for 24h before analysis.

Immunoblot and qRT-PCR analysis reveals that PRDM1 is preferentially expressed in

the CD56dim subset in response to stimulation (Figure 21). Furthermore, PRDM1β is the

predominant isoform present.

Figure 21: PRDM1 is preferentially expressed in the CD56dim NK population. A) CD56dim and CD56bright cells were obtained by FACs sorting of negatively-selected human NK cells as described (193). Cells were stimulated for 24h with IL-12 (10ng/mL) and IL-18 (100ng/mL) then analyzed by RT-PCR for expression of PRDM1 isoforms. GAPDH was used as a housekeeping gene. Results are expressed as fold over U266 mRNA which expresses high levels of both PRDM1α and PRDM1β. B) Immunoblot analysis of lysates prepared from CD56dim and CD56bright subsets obtained as in panel A, but from different donors.

PRDM1 expression in NK cells appears to differ markedly from other immune

lineages. The characteristic isoform pattern is not observed in other cell types and

stimulations that induce PRDM1 are distinct. In B cells, using either CD19+ purified

lymphocytes or patient-derived CLL cells, IL-21 is involved in upregulation of PRDM1 protein expression (Figure 22A, B). In both experiments, the full-length PRDM1α is the predominant isoform. Similarly, during LPS-induced maturation of dendritic cells,

PRDM1α is induced with no detectable PRDM1β (Figure 22C). Furthermore,

Bortezomib treatment induces only PRDM1α in mantle cell lymphoma cell lines (Figure

22D). Conversely, multiple isoforms are consistently observed in NK cells.

Interestingly, IL-21 does not activate PRDM1 alone nor does it synergize with IL-12 or

IL-18 (Figure 22E). Thus, novel mechanisms of PRDM1 induction exist in NK cells.

Figure 22: NK-specific expression of PRDM1 isoforms. Immunoblot analysis of PRDM1 induced in various lineages. A) CD19+ B cells isolated from two healthy donors +/- 7 day stimulation with IL-2, IL-4, IL-10, IL-21 and trimeric CD40 ligand as described (194). B) Patient-derived CLL cells +/- 5 day stimulation with IL-21 (100ng/mL). C) Human monocyte-derived DCs induced to mature with LPS as described (75). D) Mino and Jeko cells +/- 5nM Bortezomib as described (195). E) Negatively-selected NK cells treated for 2 days with combinations of IL-2, IL-12, IL-18 and IL-21.

Several lines of evidence suggest that the induction of PRDM1 in NK cells is not associated with a terminal differentiation event as in B cells. First of all, NK cells in peripheral circulation are mature cells, yet lack detectable PRDM1 protein expression in the absence of stimulation. Constant cytokine stimulation is also required to maintain

PRDM1 expression, as indicated by “washout” experiments. In the absence of sustained stimulation, PRDM1 protein and mRNA levels diminish to near background levels

(Figure 23A). Post-transcriptional mechanisms are involved in maintaining high levels

Figure 23: PRDM1 induction is transient and is regulated via a post-transcriptional mechanism. A) “Washout” experiments were conducted with primary NK cells stimulated for 4h with IL-2 (100U/mL), IL-12 (10ng/mL) and IL-18 (100ng/mL). Cells were collected, washed in PBS and resuspended in fresh RPMI (w/o) or left under stimulation (stim.) for up to 4 days. Lysates were collected at various time points and analyzed by immunoblot for PRDM1 expression. B) Freshly isolated NK cells from two healthy donors were stimulated for 4h in the presence of either DMSO, cyclohexamide or actinomycin D. RT-PCR analysis was conducted of both PRDM1α and PRDM1β mRNA levels, relative to 18S. C) LAK cells were either stimulated or left un-stimulated for 4h and analyzed as in B. D) Model depicting putative labile repressor that regulates PRDM1 mRNA levels. This repressor has a high turnover rate and in the absence of new protein synthesis, PRDM1 mRNA levels accumulate to high levels.

of PRDM1 expression in NK cells. When protein synthesis is inhibited via cyclohexamide, cytokine-stimulated NK cells express markedly higher levels of PRDM1 after 4h stimulation (Figure 23B, C). This effect is enhanced with cytokine stimulation and indicates the presence of a labile negative regulator of PRDM1 mRNA. Furthermore, expression profiles of transcription factors associated with plasma cell differentiation are different in NK cells. In B cells, PAX5 and BCL6 decrease, while IRF4 and XBP1

increase concomitantly with induction of PRDM1 during plasma cell differentiation. In

NK cells PAX5 is not expressed, BCL6 does not decrease and IRF4 does not increase

(Figure 24). Finally, when NK cells are stimulated to induce PRDM1, XBP1 is also up-

regulated, however, no splicing of XBP1 mRNA is observed in NK cells (Figure 25). In

B cells, XBP1 splicing accompanies PRDM1 induction during plasma cell differentiation

and is required for the proper regulation of the secretory pathway (48). Thus,

stimulation-induced PRDM1 expression in NK cells is functionally distinct from B cells.

Figure 24: Plasma cell transcription factor expression diverges in NK cells. Signal intensities were averaged from each donor before and after 24h stimulation with IL-2 (100U/mL), IL-12 (10ng/mL) and IL-18 (100ng/mL). Shown are major transcription factors associated with plasma cell differentiation.

Figure 25: XBP1 splicing is not associated with NK activation. RT-PCR analysis was conducted for splicing of XBP mRNA. cDNA was prepared from RNA obtained from two healthy donors at time of isolation or after 24h stimulation with IL-2 (100U/mL), IL- 12 (10ng/mL) and IL-18 (100ng/mL).

PRDM1 Expression is associated with NK activation

Given that PRDM1 is maximally induced in response to co-stimulation with IL-2,

IL-12 and IL-18, it is likely that levels of PRDM1 correlate with degree of activation.

Thus, mRNA levels of the effector cytokines IFNγ and TNFα were profiled in purified

human NK cells after 24h stimulation with various combinations of cytokines.

Quantitative RT-PCR analysis indicates that IFNγ and TNFα mRNA are synergistically

increased upon stimulation with IL-2, IL-12 and IL-18 stimulation (Figure 26A).

Consistent with this, IL-2, IL-12 or IL-18 alone each minimally alter PRDM1 expression

(Figure 26B). These data suggest that PRDM1 levels correlate with effector cytokine transcription.

Figure 26: PRDM1 induction is correlated with degree of activation. A) RT-PCR analysis of cDNA synthesized from NK cells at time of isolation or after 24h stimulation with IL-2 (100U/mL), IL-12 (10ng/mL) and IL-18 (100ng/mL), alone or in combination. Expression values were calculated using the ΔΔCt method with 18S as the control gene. Upon linearization of Ct values, average Time 0 for each donor was arbitrarily set to “1” to account for donor-to-donor variability of basal mRNA levels. Error bars represent standard deviation from at least three biologically independent samples isolated from different donors at different times. B) Immunoblot analysis of purified human NK cells stimulated as in panel A.

As NK cells produce high levels of effector cytokines upon activation, the ability of these effector cytokines to induce PRDM1 through autocrine or paracrine feedback mechanisms was assessed. Treatment with either recombinant IFNγ or TNFα fails to induce PRDM1. Stimulation with α-interferon is also insufficient to induce PRDM1, suggesting that induction results primarily from cytokine receptor-mediated signaling and likely requires multiple signaling events (Figure 27A, B). Ligation of CD16, the low affinity Fc receptor, initiates activation of NK cells, resulting in ADCC and increased

IFNγ production (196). However, no induction of PRDM1 was observed when surface

CD16 was activating with anti-CD16 antibodies (Figure 27C). Furthermore, although a

minor dose-dependent increase in PRDM1 expression was observed, IL-2 alone is

insufficient to drive high level PRDM1 expression in freshly isolated NK cells (Figure

27D). Thus, PRDM1 expression is not a “pan-activation” marker and NK cells require

stimulation with multiple cytokines to maximally induce PRDM1.

Figure 27: Autocrine and or paracrine feedback loops are not responsible for PRDM1 induction. A) Immunoblot analysis of lysates prepared from purified NK cells treated with recombinant human TNFα (20ng/mL or 100ng/mL), IFNγ (10ng/mL or 50ng/mL) for 24h. B) NK cells treated with recombinant human IFNα (10ng/mL or 50ng/mL). C) NK cells incubated with plate-bound anti-CD16. D) NK cells treated with increasing concentrations of IL-2. In all experiments, stimulation with IL-2 (100U/mL), IL-12 (10ng/mL) and IL-18 (100ng/mL) was used as a positive control for PRDM1 induction.

To determine if PRDM1 is induced in vivo, murine NK cells were activated by

tail vein polyI:C injections into C57/BL6 mice. PolyI:C is known to activate NK cells

indirectly through TLR3-mediated release of cytokines from dendritic cells and other accessory cells. NK cells were purified by negative selection from the spleens of naïve

and polyI:C treated mice. PRDM1 is robustly induced specifically in NK cells upon 48h

treatment (Figure 28). PRDM1 expression was also detected in total splenocytes and the

level did not change with polyI:C treatment. This is likely due to the presence of

PRDM1 in T cells and differentiating plasma cells within the spleen. Interestingly, only the PRDM1α homologous isoform was detectable in murine NK cells. Although an inability of the antibody to cross-react with other murine isoforms cannot be excluded, it is possible that the expression and/or function of PRDM1 isoforms in murine cells may be differentially regulated.

Figure 28: PRDM1 is induced in vivo. C57/BL6 mice were treated with 100μg polyI:C via tail vein injection for 48h. NK cells from total splenocytes (right panel) or purified via negative selection (left panel) and analyzed by immunoblot analysis.

PRDM1 is not involved in perforin-mediated cytotoxicity

The ability to lyse target cells in an antigen-independent manner is a hallmark of

NK cells. Cytotoxicity against heterologous target cells proceeds through the release of

perforin and granzymes and is markedly increased upon treatment with IL-2, IL-12 or IL-

18 (99, 197). To assess whether PRDM1 regulates cytotoxicity, NK cells were isolated

from healthy donors and stimulated with either IL-2 and IL-12 or the combination of IL-

2, IL-12 and IL-18 in the presence of either a non-targeting control or a PRDM1-specific siRNA for 72h. Cytotoxicity against the K562 leukemia cell line was assessed in a 4h

51Chromium release assay. As expected, cytokine-activated NK cells exhibit significant cytotoxicity against K562 targets (Figure 29A). Knockdown of PRDM1 expression did

Figure 29: Perforin-mediated cytotoxicity is not affected by PRDM1 knockdown. A) 4h 51Chromium release assays were performed in triplicate from two donors. Purified NK cells were incubated for 72h in the presence of either a non-targeting (NT) or PRDM1-specific (KD) siRNA and stimulated with combinations of IL-2 (100U/mL) and IL-12 (10ng/mL), with or without IL-18 (100ng/mL). % 51Chromium release was calculated as described in Materials & Methods. Error bars represent SD of triplicate wells. B) Western blot analysis to confirm knockdown in cells used in cytotoxicity assay. IL-2-only sample was not assayed in cytotoxicity assay and is only included to avoid cropping of gel. C) Purified NK cells were incubated for 48h or 72h in the presence of either a non-targeting (NT) or PRDM1-specific (KD) siRNA. Cells were stimulated with combinations of IL-2 (100U/mL), IL-12 (10ng/mL) and IL-18 (100ng/mL) and analyzed by western blot for PRDM1 induction and PARP cleavage.

not alter cytolytic activity across several effector:target ratios. Distinct from experiments that revealed increases in IFNG and PRDM1 mRNA levels with the addition of IL-18 to

IL-2 and IL-12, no additive effect of IL-18 was observed in cytotoxicity assays, further

demonstrating that cytotoxicity and PRDM1 levels are not directly linked. Knockdown

of PRDM1 protein was confirmed by immunoblotting to be highly efficient and had no

effect on viability as assessed by trypan blue staining (Figure 29B , data not shown). No

activation-induced cell death was detected in stimulated NK cells as measured by PARP cleavage (Figure 29C). Furthermore, no defects in cytokine-induced clustering were

observed in response to PRDM1 knockdown (Figure 30). Collectively, these results

show that PRDM1 does not have a significant role in regulating perforin-mediated cytotoxicity in NK cells.

Figure 30: Clustering phenotype observed upon cytokine activation is not diminished with PRDM1 knockdown. Primary NK cells were stimulated with either IL-2 (100U/mL) and IL-12 (10ng/mL) or IL-2, IL-12 and IL-18 (100ng/mL) in the presence of either a non-targeting or PRDM1-specific siRNA. Cells were analyzed by light microscopy at 24h, 48h and 72h.

PRDM1 binds promoters in multiple target genes

As PRDM1 is well-documented as a DNA-binding transcriptional repressor we

sought to characterize its function in NK cells by identifying DNA elements to which

PRDM1 specifically binds. To this end, chromatin immunoprecipitation (ChIP) experiments were performed using primary NK cells isolated from healthy donors. Cells were stimulated for 24h with IL-2, IL-12 and IL-18 to induce PRDM1 expression prior to isolation of chromatin and immunoprecipitation with either anti-IgG or anti-PRDM1 antibodies. Initially, promoter regions of genes known to be regulated by PRDM1 were analyzed (Figure 31). A well-documented PRDM1 target gene is CIITA, the master regulator of MHC class II expression which is transcriptionally silenced as mature B cells

Figure 31: PRDM1 binds CIITApIV, but not other B cell target genes. Chromatin immunoprecipitation experiments were performed with chromatin isolated purified NK cells stimulated for 24h with IL-2 (100U/mL), IL-12 (10ng/mL) and IL-18 (100ng/mL). Induction of PRDM1 was confirmed by western blot. Percent input was calculated as described in Materials & Methods. Error bars represent SD from three biologically independent experiments performed at different times with chromatin preparations from different donors. Two-tailed t-tests between IgG and PRDM1 immunoprecipitated samples, * p-value = 0.035.

differentiate into antibody-producing plasma cells. PRDM1 binding was observed at the

IFNγ-inducible CIITApIV promoter, but not the lymphoid-specific CIITApIII promoter.

Although human NK cells can increase surface MHC Class II expression in response to activation (198) and have been reported to exhibit antigen presentation capacity (199),

they express low to undetectable levels of CIITA. Binding of PRDM1 to the CIITApIV

promoter may function to reinforce this repressed state and prevent IFNγ-mediated autocrine induction of CIITA.

No binding was detected at the promoters of PAX5 and IFNB, both of which have previously been shown to be directly regulated via binding of PRDM1 to promoters in B cells and Sendai virus-infected osteosarcoma cell lines, respectively. Furthermore, no binding was detected at the promoter of SLAMF7, which encodes the NK activating receptor CRACC and contains a potential PRDM1 binding motif within its proximal promoter. This indicates that PRDM1 binds target gene promoters in a selective and cell type specific manner. As expected, no binding was detected using primers to the second exon of Myoglobin, which was used as a negative control locus.

In murine T cells, Blimp-1 has been shown to bind to a distal, conserved regulatory site within the Ifng locus in in vitro polarized CD4+ Th1 lymphocytes (66). As

NK cells are potent producers of IFNγ during the early innate immune response and

induction is correlated with PRDM1 expression, its binding to the human IFNG locus in

cytokine-stimulated primary NK was evaluated. Four genomic locations were assayed

across the IFNG locus, which are highly conserved between rodents and humans and

have been demonstrated to regulate IFNG expression (200). Binding was detected at the

distal regulatory site, which has been demarcated as CNS -22 based on the genomic

distances relative to the transcriptional start site in the mouse (Figure 32). This site is

analogous to the PRDM1 binding site detected in T cells. Additionally, through

Figure 32: PRDM1 binds known regulatory loci within the IFNG locus. A) Chromatin immunoprecipitation experiments were performed as in Figure 31 using primers across several known regulatory regions of the IFNG locus. Error bars represent SD from three biologically independent experiments performed at different times with chromatin preparations from different donors. Two-tailed t-tests between IgG and PRDM1 immunoprecipitated samples, ** p-value = 0.0009. B) Locations of the primer sets, relative to a portion of the IFNG locus region of chromosome 12.

bioinformatic analysis two potential PRDM1 binding sites were identified within the

minimal promoter, located 370bp and 254bp upstream of the IFNG transcriptional start site. Robust PRDM1 binding was detected in this proximal promoter region which has not been identified in T cells. Consistent with the absence of potential PRDM1 binding sites, no binding was detected at CNS -6 or CNS +18-20. Collectively, these data

demonstrate that PRDM1 binds both proximally and distally to multiple sites across the

IFNG locus in human NK cells.

TNF is coordinately induced with IFNG upon cytokine stimulation of NK cells.

Thus, we sought to determine if this locus was also bound by PRDM1. To this end, we assayed four distinct locations across the ~12kB TNF locus by chromatin

immunoprecipitation. The TNF locus contains three genes, each of which contains four

exons (Figure 33). LTA and TNF are separated by ~1kB and transcribed from the same strand while LTB is transcribed from the opposite strand and separated from TNF by

Figure 33: PRDM1 binds known regulatory loci within the TNF locus. A) Chromatin immunoprecipitation experiments were performed as in Figure 31 using primers across several known regulatory regions of the TNF locus. Error bars represent SD from three biologically independent experiments performed at different times with chromatin preparations from different donors. B) Locations of the primer sets, relative to a portion of the TNF locus region of chromosome 6.

~3kB. PRDM1 binding was measured at the proximal promoter regions of LTB and

TNF, the intergenic region between TNF and LTB and a recently identified regulatory site located ~3.5kB upstream of LTA. This upstream enhancer element coordinately regulates the LTA and TNF genes but not the opposing LTB gene (201, 202). PRDM1 binding was clearly detected at this upstream regulatory site, but not at either proximal promoter or the intergenic region. Furthermore, bioinformatic analysis detected a consensus PRDM1 recognition sequence in the bound enhancer element. These data indicate that PRDM1 specifically associates with defined regulatory sequences of the TNF locus.

Blockade of cytokine-induced PRDM1 expression increases effector cytokine production

PRDM1 is coordinately induced with effector cytokines upon stimulation and occupies specific regulatory regions within the IFNG and TNF loci. To investigate the functional effects of PRDM1 in NK cells, gene expression knockdown experiments using primary human NK cells were performed. Cells were stimulated with IL-2 and IL-12 in the presence of either a non-targeting control or PRDM1-specific siRNA. mRNA was isolated after 48h and analyzed by real-time RT-qPCR. Significantly higher mRNA levels of IFNG, TNF, and LTA were observed when PRDM1 induction was abrogated via siRNA, yet no significant differences were found for DAP10 (Figure 34).

Figure 34: Knockdown of PRDM1 increases levels of effector cytokine mRNA: Purified NK cells were incubated for 48h in the presence of either a non-targeting (NT) or PRDM1-specific (KD) siRNA and stimulated with IL-2 (100U/mL) and IL-12 (10ng/mL). RT-PCR was conducted for several genes on cDNA synthesized from RNA isolated from three different donors at different times using 18S control gene. For each donor, NT was normalized to “1”. Data are presented as the mean, with error bars representing SEM from three biologically independent experiments.

In order to determine whether increases in mRNA levels observed with PRDM1 knockdown resulted in detectable changes in protein expression, ELISA experiments were performed. Secreted IFNγ and TNFα were consistently increased in multiple donors

in response to PRDM1 expression knockdown (Figure 35). Consistent with this

silencing, PRDM1 expression remains elevated over a four day post-stimulation time course, while IFNG and TNF levels decline (Figure 36). Together, these data indicate that PRDM1 negatively regulates production of these cytokines in response to NK cell activation.

Figure 35: Knockdown of PRDM1 increases levels of secreted effector cytokines: ELISA assay was conducted on supernatants harvested at 48h from samples in Figure 34 above. Data represent the mean of three independent wells. For each analyte, two-tailed paired t-tests demonstrated p-value < 0.05 among all three donors for NT vs. KD.

Figure 36: Time course of cytokine mRNA and PRDM1 protein levels. A) Primary NK cells stimulated with IL-2 and IL-12 for up to 4 days. RT-PCR was conducted for several genes on cDNA synthesized from RNA isolated from different times using 18S control gene. B) Representative immunoblot for PRDM1 expression at various timepoints.

Over-expression of PRDM1 mediates repression of activation-induced expression of

IFNG and TNF

Because PRDM1 occupancy at effector cytokine loci and increased production in

the context of PRDM1 knockdown were observed, the ability of PRDM1 to block

activation-induced transcription of IFNG and TNF was assessed. The Jurkat T cell line was transduced with adenovirus expressing either GFP alone or GFP and PRDM1α prior to stimulation. After PMA/PHA stimulation, RT-qPCR analysis was performed to assess induction of IFNG and TNF. GFP-transduced cells showed significant up-regulation of

IFNG and TNF upon stimulation, which was nearly abolished in PRDM1α-transduced cells (Figure 37). Thus, introduction of PRDM1 prior to stimulation was sufficient to block stimulation-induced transcription of IFNG and TNF, providing further support that

PRDM1 is a negative regulator of effector cytokine production.

Figure 37: Over-expression of PRDM1 suppresses activation of target genes. Jurkat cells were transduced for 40h, and then stimulated with PMA (50 ng/mL) and PHA (1 µg/mL) for 4h prior to RNA isolation. After correction using 18S, relative expression for IFNG and TNF were determined by assigning a value of 100% to the stimulated GFP- transduced treatment. Data are presented as mean of three independent experiments, with error bars representing SD, * p-value .0002, ** p-value .0005.

PRDM1 mediates repression of IFNG via elements in the proximal promoter

To directly assess the functionality of PRDM1 binding to the newly identified elements within the proximal promoter of the IFNG gene, we cloned the region comprising -507 to +121 of the human IFNG gene and inserted it upstream of a luciferase reporter (pGL3-IFNγ_WT). This promoter region contains the two potential PRDM1 binding sites, located at -370 and -254. Co-transfection of PRDM1 was sufficient to repress luciferase activity driven by the wild type IFNG promoter by approximately 50%

(Figure 38). To assess the relative contributions of the two potential PRDM1 binding motifs, a deletion of approximately 300 base pairs encompassing both of these sites

Figure 38: Promoter-proximal elements in the IFNG promoter are required for PRDM1-mediated repression. Luciferase assays were conducted in Jurkat cells, 36h post-transfection. Data are presented as the mean of four independent transfections +/- SD, with each reporter transfected with empty vector set to 100%. Two-tailed paired t- tests showed p-value of .005 for wild-type, indicated by *.

(pGL3-IFNγ_Δ) was created. This deletion eliminated the ability of PRDM1 to repress transcription from the construct. This repression was similarly eliminated by point mutations at the -254 site (pGL3-IFNγ_mut) demonstrating that this location is critical for PRDM1-mediated repression. Thus, PRDM1 represses IFNG through binding to this

previously uncharacterized motif.

PRDM1 accumulates during LAK expansion and associates with functional exhaustion

PRDM1 expression is associated with terminal differentiation in a variety of cell

types. Recent evidence demonstrates that PRDM1 accumulation in CD8+ effector T cells

is associated with exhaustion. After an initial lag phase, human NK cells can be induced to proliferate robustly in vitro for several weeks (in the presence of IL-2 and activating stimuli such as anti-CD335 beads) before reaching exhaustion. These stimuli do not induce PRDM1 expression after 48h in culture, but high levels are observed by day 19

(Figure 39). Interestingly, the full-length PRDM1α is the predominant isoform induced, unlike short-term stimulation with IL-2, IL-12 and IL-18, which induce the truncated

PRDM1β and the putative PRDM1αΔexon6 isoform more abundantly.

Figure 39: PRDM1 expression and function in LAK cells. A) LAK cells were generated from three healthy donors in the presence of IL-2 (500U/mL) and activating beads (anti-NKp46, anti-CD2). Immunoblot analysis at various timepoints. B) Primary NK cells from three donors were isolated and stimulated for 48h with IL-2, IL-12 and IL- 18. C) ChIP analysis of the IFNG locus in LAK cells derived from one donor.

Discussion

The data presented here provides the first functional description of PRDM1 in

natural killer cells. PRDM1 accumulates upon cytokine-mediated activation and acts as a

negative regulator of activation, attenuating inflammatory cytokine production. Such a

negative feedback loop has important implications in the context of inflammation and

immune homeostasis.

Using global gene expression profiling, PRDM1 was identified as a highly

expressed transcription factor in stimulated NK cells. This data complements previous

studies (100, 101, 104, 117) providing the first global gene expression profile of NK cells

utilizing a physiologically-relevant combination of cytokines to achieve activation.

During the early phases of the innate immune response IL-12 and IL-18 are produced by

monocytes and epithelial cells, while IL-2 is primarily produced by activated T cells, typically a later event. Another early monocyte-derived cytokine, IL-15, signals through the shared IL-2 receptor and can substitute for IL-2 to induce PRDM1 in the presence of

IL-12 and IL-18 (unpublished data). Consistent with this, maximal induction of murine

NK cells in response to Salmonella-infected macrophages has recently been shown to

require: 1) IL-2 and/or IL-15, 2) macrophage-derived IL-12 and IL-18 and 3) direct NK- macrophage cell contact (130). Thus, the exposure of freshly isolated NK cells to these cytokines ex vivo can mimic the milieu present in vivo during the early innate immune response and can provide insight into the physiological gene expression changes occurring.

The ability of NK cells to produce a variety of cytokines in response to stimulation provides a crucial mechanism of cross-regulation between the innate and adaptive arms of the immune system. In order to maintain homeostasis, regulatory control must be exerted not only upon the activation phase, but also during the recovery

phase to restrain the degree of activation. While activation events have been thoroughly

investigated, molecular events regulating dampening or recovery remain poorly

characterized. The studies herein now show that PRDM1 is an important mediator of this

phase. PRDM1 attenuates production of multiple effector cytokines in a coordinate

manner via direct binding to specific DNA sequences in known regulatory regions. At

the TNF locus, PRDM1 binds specifically to a highly conserved regulatory element

~3.5kB upstream of the transcriptional start site of LTA. This region exhibits DNaseI

hypersensitivity and contains binding sites for inflammatory activators such as NFκB and

NFAT (201). Furthermore, this site has been shown to be crucial for activation-induced looping (202). This three-dimensional looped chromatin structure mediates interaction

between distally- and proximally-bound NFAT at the TNF promoter forming an

enhanceosome resulting in transcriptional activation of TNF and (to a lesser extent) LTA

in response to stimulation. Consistent with the notion that this region imparts specific,

localized control at the TNF locus, knockdown of PRDM1 results in increases in both

LTA and TNF, but not LTB which is located further downstream and encoded on the

opposite DNA strand.

The IFNG locus is also subject to PRDM1-mediated regulation in NK cells.

Studies in the T cell compartment identified an upstream regulatory region in murine

CD4+ T cells (CNS-22) with enhancer activity which contributed to Tbet-dependent Ifng expression in in vitro polarized Th1 cells (200). This site was later shown to be bound by the murine homologue Blimp1 (66). The functionality of this distal site was confirmed in human NK cells. In addition to this distal element, we have identified a novel promoter- proximal element within the IFNG promoter to which PRDM1 binds. This novel proximal site is required for silencing of a reporter gene driven by the human IFNG promoter. Freshly isolated, resting NK cells differ substantially from T cells with respect to the kinetics of IFNG mRNA induction. Indeed, NK cells induce IFNG rapidly upon stimulation without the requirement for chromatin modifications because the entire locus is poised in a hyperacetylated state under basal conditions (203). Thus, PRDM1 binding directly upstream of the transcriptional start site may facilitate potent silencing in NK cells without necessarily requiring long-range chromatin modifications.

Human NK cells exist in at least two functionally divergent subsets based on the

surface density of CD56 and CD16. Several groups have demonstrated that

CD56dimCD16+ NK cells are the major population in peripheral blood, while

CD56brightCD16-/dim NK cells represent less than 10% of peripheral NK cells and primarily localize to lymph nodes (204). The CD56bright population has been described as

a “regulator” population. These cells express the high-affinity IL-2 receptor, are highly

proliferative in response to stimuli and localize at inflammatory sites. Accumulating

evidence suggests that CD56bright cells may, in fact, be developmental precursors to the

more mature CD56dim “effector” population (106, 205). Conversely, the CD56dim

“effector” population exhibits increased natural cytotoxicity, but produces lower levels of

inflammatory cytokines relative to the CD56bright population. Consistent with its role as a

negative regulator of cytokine production, PRDM1 is preferentially expressed in the

CD56dim population. Thus, PRDM1-mediated transcriptional repression likely

contributes to the functional divergence observed in these populations.

NK cells exhibit a unique pattern of PRDM1 protein isoform expression. In

addition to the full-length PRDM1α, NK cells express high levels of two smaller

molecular weight species. Expression of the PRDM1β-isoform has previously only been documented in myeloma cell lines and samples from myeloma and T cell lymphoma patients, and is always present at lower levels than the full-length protein (5, 206). NK cells represent the first non-transformed cell type with PRDM1β expression; furthermore, it is consistently observed at higher levels than the larger PRDM1α. The β-isoform is transcribed from an alternate promoter and utilizes a distinct transcriptional start site present in the third intron of the full-length protein. The resulting protein has a disruption

of the highly conserved PR-domain. It was previously shown that the β-isoform localizes

to the nucleus and maintains capacity for DNA-binding, albeit, its repressive activity is dampened relative to PRDM1α. All of the cytokine combinations tested activate transcription of both isoforms in NK cells, although there remains the potential that isoform-specific activation signals may exist. An intermediate-sized isoform

(PRDM1αΔ6) corresponding to a deletion of the amino acids encoded by exon 6 is also consistently observed. This isoform is generated via splicing to exclude exon 6 resulting in a deletion of the 2nd zinc finger as well as disruption of a portion of the 1st and 3rd zinc

finger. An analogous splice variant was recently described in naïve CD19+ murine B

cells, although it was present at very low levels (6). The first two zinc fingers of PRDM1

have a role in DNA binding and are required for recruitment of the histone

methyltransferase G9a, but are dispensable for interaction with HDAC2 (14, 15).

Interestingly, no induction of apoptosis or evidence of cell cycle exit concomitant with

PRDM1 induction in NK cells was observed. Furthermore, IL-2-expanded LAK cells

continue to proliferate despite acquisition of high level PRDM1 expression (unpublished

data). Thus, the presence of these lower molecular weight isoforms which have altered or

impaired activity may provide a cytoprotective effect, whereby NK cells are protected

from the anti-proliferative and apoptotic effect of PRDM1, but retain the ability to repress

selective target genes. Characterization of the contribution of each isoform will be an

important area of future investigation.

The PRDM1 allele is encoded on chromosome six within a region that is known

to be frequently associated with B-, T- and NK-derived malignancies. A recent report

used high resolution comparative genomic hybridization arrays to precisely map regions

deleted within NK lymphomas and identified a minimal common region spanning

approximately 2Mb of the 6q21 deletion, which is present in roughly 40% of cases (62).

Within this region, the authors identified three genes which were down-regulated in tumor specimens harboring the 6q21 deletion: ATG5, PRDM1 and AIM1. More recently,

PRDM1 expression levels were shown to be highly variable and found to be independent

of the 6q21 deletion in an independent study of six primary NK lymphoma patient

samples (207). This data clearly establishes a functional role for PRDM1 in normal NK

cell function. Thus, while PRDM1 can act as a tumor suppressor in Diffuse Large B Cell

lymphoma, its contribution to the transformed phenotype in NK-derived malignancies

remains controversial.

The experiments herein add NK cells to the expanding list of immune lineages in

which PRDM1 has a key functional role. Until very recently, PRDM1 has been

described in plasma cells, multiple T cell subsets, dendritic cells and myeloid cells.

Subsequent to the publication of these studies, a second group working in a murine system has published complementary data, confirming most of the findings presented here (208). Although lineage-specific regulation of target genes exists, a significant commonality is that PRDM1 regulates final effector function. For instance, negative regulation of Il2 and tbx21 by Blimp1 promotes Th2 polarization in CD4+ T cell

differentiation, yet differential expression of these genes does not play a role in either B

or NK effector function. Coordinate regulation of IFNG and TNF, on the other hand, is

critical to both T and NK effector function. Characterization of overlapping and non- overlapping functions of PRDM1 across multiple lineages will be an important area of focus in future experiments. For example, Lanier and colleagues recently provided

evidence for a “memory” phenotype in murine NK cells using an MCMV infection model wherein they demonstrated that antigen-specific Ly49H+ NK cells underwent a

prolonged contraction period, but were subsequently activated to higher levels upon re-

stimulation (209). PRDM1-mediated restriction of memory potential has recently been

described in CD8+ T-cells (72). Given that PRDM1 is induced in vivo during NK activation, it will be interesting to ascertain if PRDM1 has such a role in NK memory commitment.

In summary, the data suggest that PRDM1 plays a crucial role in the post-

activation phenotype of NK cells by negatively regulating cytokine transcription in a

coordinate manner, without compromising perforin-mediated cytotoxicity or inducing

exit from cell-cycle. Such a mechanism may have important implications in innate

immunity and tumor surveillance.

CHAPTER IV

IDENTIFICATION OF PRDM1 TARGET GENES IN NK CELLS

Introduction

Transcriptional control of cellular function is mediated through the combined action of transcriptional activators and repressors. PRDM1 is a master regulator of the terminal differentiation of activated B cells into antibody-producing plasma cells.

PRDM1 mediates this phenotypic change via direct binding to regulatory sequences of target genes, which are subsequently repressed. CIITA, PAX5, and cMYC are major targets of PRDM1-mediated repression in B cells, while PRDM1 regulates a distinct set of target genes including FOS and IL2 in T cells (181). In NK cells, the IFNG and TNF loci were found to be targets of coordinate repression via a candidate ChIP approach.

However, this candidate approach was not ideal to discover novel genomic loci to which

PRMD1 binds.

Recent advances in hybridization technologies have facilitated the de novo discovery of target genes. Using traditional ChIP coupled to promoter-tiled microarrays, all genes bound by a particular transcription factor can be identified. This approach, termed ChIP-chip or ChIP-on-chip was pioneered by the laboratory of Peggy Farnham and has recently become a widely-used technique (210). Affymetrix is one of several companies that now manufacture promoter tiling arrays which can be utilized by

microarray core facilities at a reasonable cost to investigators. More recently, ChIP-seq

protocols have been developed whereby immunoprecipitated DNA is analyzed by high-

throughput sequencing platforms. However, ChIP-seq procedures require equipment and

computer capabilities that remain cost prohibitive to many investigators.

In order to define the interaction network of PRDM1 in NK cells, two microarray

approaches were utilized. First, ChIP-chip experiments were conducted using the NKL

cell line to identify genomic regions bound by PRDM1. Next, mRNA expression arrays

were used to assess transcriptional changes associated with knockdown of PRDM1 in

primary NK cells. These data were used to define both the direct and indirect targets of

PRDM1 in NK cells. To date, no ChIP-chip or ChIP-seq analyses have been reported in

NK cells. Thus, these experiments are the first to use global approaches to transcription factor analysis in NK cells.

Results

PRDM1 is constitutively-bound to numerous target genes in the NKL cell line

Through the use of a candidate approach with traditional chromatin immunoprecipitation (ChIP) experiments several bona fide PRDM1 targets were identified,

several of which were described in the previous section. This approach, however, is

associated with a low “success” rate because relatively few potential PRDM1 binding

motifs were found to be bound in vivo. Less than five percent of in silico-identified

PRDM1 binding sites exhibited binding when assayed by ChIP. Thus, the presence of

the motif within the promoter region alone was a poor predictor of promoter occupancy.

Furthermore, PRDM1 occupancy was not routinely found at the promoters of known B

cell target genes indicating that occupancy of target genes may vary widely among cell

types.

In order to circumvent the low discovery rate associated with a candidate

approach, ChIP-chip experiments were performed. This technique expands upon

traditional ChIP and utilizes hybridization to microarrays containing tiled genomic DNA

sequences. A screening approach can then be taken facilitating the de novo identification

of target genes. The Affymetrix Promoter 1.0 array contains over 4.6 million probes tiled

over 25,500 human promoters. Typical coverage for each promoter is approximately

7.5kb upstream and 2.5kb downstream from the transcription start site (TSS), while promoter coverage for nearly 1,300 cancer-associated genes extends to 10kb upstream.

The requirement for large amounts of starting material (10ng immunoprecipitated DNA) precluded the use primary NK cells for ChIP-chip analysis. Thus, the NKL cell line had to be used. These cells constitutively express PRDM1 and exhibit an isoform expression pattern that is highly similar to activated primary NK cells as described earlier.

Three independent chromatin preparations were used to perform ChIP-chip experiments. Multiple immunoprecipitations were performed and analyzed by traditional

ChIP using primers for two positive loci (CIITApIV and PCNA) and two negative loci

(MYOB and HLA-DR) to verify ChIP efficiency (data not shown). Immunoprecipitated

DNA from each chromatin preparation was then separately pooled and subjected to

whole genome amplification in order to obtain sufficient quantity for hybridization. A

second efficiency check was performed on amplified DNA to ensure that negative and

positive loci remained easily distinguishable. Three hybridizations were performed using the amplified DNA and regions of enrichment for PRDM1 binding were identified using

two recently-described analytical algorithms: Cisgenome and Model-based Analysis of

Tiling-arrays (MAT) (175, 176).

Because ChIP-chip datasets are characteristically large and heterogenous, accurate determination of transcription factor binding to DNA can be challenging. In order to separate signal from noise (referred to as “peak calling”), CisGenome and MAT make use of distinct statistical algorithms in order to calculate enrichment over background. Cisgenome utilizes TileMapv2 to call peaks after quantile normalization.

This algorithm relies on a Bayesian shrinkage to analyze variance in hybridization values between specific and control samples, then applies either a Hidden Markov Model or a moving average to score for enrichment (175, 211). MAT analysis differs in the method used for initial normalization of hybridization intensities and it also applies a model- based statistical algorithm to correct for non-specific hybridization. This is based on the assumption that much of the observed “signal” is actually “noise” due to intrinsic properties of specific probe sets on the array (176). Importantly, each program allows the user to define the minimum peak lengths, number of probe sets required to define a peak and tolerable gap sizes so that only the underlying algorithms are different when analyzing an individual experiment each way.

Peak lists were created for each PRDM1 ChIP-chip experiment with both analysis methods using raw hybridization data for both PRDM1 immunoprecipitation and input

(.cel files). MAT analysis consistently produced a lower number of peaks per experiment

(mean = 100) than CisGenome (mean = 483) using identical parameters: a minimum peak size of 100bp, a minimum of five probe sets per peak as thresholds for detection and a separation of > 300bp between individual peaks (Figure 40). In order to identify genes

associated with these peaks (potential targets of PRDM1 regulation), peak lists generated

by both MAT and CisGenome were annotated with CisGenome using settings such that

Figure 40: CisGenome consistently identifies more PRDM1-enriched peaks. Graph plotting number of peaks identified in each experiment using both CisGenome and MAT analysis algorithms.

the single closest gene within 15kb of either the transcription start site (TSS) or transcription end site (TES) was identified. Thus, depending on the local gene architecture of the genome, it is possible to have multiple genes identified for each peak.

Consequently, a total of 1,695 unique annotated genes were identified in the six analyses

(three experiments, analyzed two ways). The PRDM1-enriched peaks were localized upstream, within and downstream of target genes, with approximately half of the peaks mapping upstream of the TSS (Figure 41A). Interestingly, for the 892 genes to which

PRDM1 was bound upstream of the TSS, peaks were equally represented across the probes (sized 7,500bp) with no enrichment observed at the minimal promoter regions

(Figure 41B).

Figure 41: Distribution of PRDM1 binding locations across the genome. A) 1,695 unique genes were identified within 15kb of PRDM1-enriched peaks in three experiments. Approximately half were located upstream of the gene, while the other half were located both within the gene and downstream. B) Binning analysis of the 895 genes for which PRDM1 was located upstream of the TSS. Equal distribution was observed across the 7,500bp represented by most of the probe sets.

Due to the potentially high false discovery rate associated with genome-level

binding experiments, a subset of annotated genes was evaluated further. Using the

criteria that a gene must be identified in at least two analyses, a total of 292 unique genes

were identified. Validation testing with independently-prepared chromatin yielded a confirmation rate of 83% (19/23) using percent input of PRDM1 immunoprecipitation relative to IgG immunoprecipitation (Figure 42). Genes such as STAT6, PTK2 and

LMO2 gave robust signal, while genes such as BCL10, IBRDC3 and IL2 demonstrated

more modest enrichment over control, consistent with variability in peak magnitude

observed by ChIP-chip. Of the four genomic loci that did not validate, all were identified

by both CisGenome and MAT analysis indicating that neither algorithm is solely

responsible for the false positives.

Figure 42: Confirmation of selected ChIP-chip target genes. Chromatin was prepared from NKL cells and analyzed by ChIP for PRDM1 occupancy at 23 genomic loci identified by ChIP-chip. Gene names represent the nearest gene within 15kb of the region. Confirmation rate was 83% (19/23).

Motif analyses were conducted to evaluate whether the identified genomic regions

were enriched for putative binding sites for PRDM1. Two approaches were utilized to

accomplish this. First, de novo motif searches were conducted for each of the six data

sets (three MAT, three CisGenome) to identify significantly enriched sequence motifs

present within the regions of PRDM1 binding. A PRDM1 motif was identified as highly

enriched sequence in both MAT and CisGenome data sets in two of three experiments

(Figure 43A). Next, a PRDM1 sequence matrix (MAGYGAAAGYK) was used to

search for the presence of this motif within each peak region. This motif was found in

the peaks associated with 84 of the 292 genes (29%). The frequency of this motif was significantly higher than a scrambled motif (p-value = .009) (Figure 43B). Of the 23

genes used for validation testing, 13 contained a PRDM1-binding site and all 13 could be

confirmed to be bound by PRDM1. This suggests that the presence of a defined binding

motif is not absolutely required for binding, but is associated with a lower false discovery

rate.

Figure 43: Motif analysis of associated peaks. A) The Gibbs motif sampler module of CisGenome was used to identify statistically-enriched motifs present in the sequences of annotated peaks for both the MAT and CisGenome data sets. The motif was found in two out of three experiments using both data sets. B) A known PRDM1-binding motif (MAGYGAAAGYK) or a scrambled motif was used to identify the presence of the motif in all sequences associated with PRDM1-enriched peaks. The frequency of the scrambled motif was significantly lower (p-value = .009). Although motifs were found in both MAT and CisGenome in experiment three, they were not identified as statistically- enriched, thus they do not show up in A).

Expression changes associated with ChIP-chip target genes

Of the 292 genes identified, expression data was available for 278 genes using the

24h IL-2, IL-12 and IL-18 stimulation microarray described in Chapter III. Using a

cutoff of 1.3 fold in both donors, 161 were found to be differentially expressed with

cytokine stimulation (76 decreased, 85 increased) (Table 2 and Appendix II). Genes

that decrease coordinately with PRDM1 induction are likely early targets of PRDM1-

mediated direct repression. Several of these down-regulated genes are intracellular

signaling adaptors likely important for NK function, including ZAP70, BCL10 and

STAMBP. ZAP70 is a cytoplasmic tyrosine kinase that is activated in response to CD16 ligation and therefore contributes to ADCC (212). In response to activating stimuli such as NKG2D ligation, BCL10 signals in cooperation with CARMA1 to induce NF-κB resulting in up-regulation of effector cytokine production (213). STAMBP binds to

STAM (signal transduction associated molecule) and abrogates JAK-STAT signaling downstream of the IL-2 receptor (214). Repression of these signaling molecules is consistent with PRDM1’s roles as a post-activation attenuator of NK activation.

Table 2: Changed ChIP-chip target genes in response to stimulation (Top 30) Decreased Increased Gene ID Donor 1 Donor 2 Gene ID D onor 1 D onor 2 LRMP -42.8 -77.0 LTA 4213.9 1686.3 GCHFR -20.7 -24.3 HAS2 815.2 63.2 ZAP70 -11.4 -10.9 ZNRF1 39.2 367.5 WHDC1L1 -9.4 -17.0 MTHFD1L 38.7 26.7 ATM -7.7 -14.9 CTLA4 12.9 3.3 NR4A3 -7.4 -15.1 CDGAP 10.4 7.1 AK1 -6.8 -5.8 RAB6IP1 9.5 11.1 RGS14 -5.9 -41.1 NFE2L1 8.2 12.2 NEK1 -5.2 -6.9 BIRC3 6.0 4.0 SREBF1 -4.2 -2.5 RAB10 6.0 8.2 BCL10 -4.2 -6.6 STX3A 5.3 3.9 MAPK7 -4.1 -2.3 GBP2 5.1 3.2 BTF3 -3.7 -2.6 WDR59 4.9 2.6 C20orf32 -3.6 -14.9 STK6 (AURKA) 4.9 6.3 C14orf43 -3.6 -3.3 STCH 4.8 3.9 PPP1R9A -3.5 -2.4 IBRDC3 4.7 6.3 SSBP1 -3.3 -4.4 BCL2L11 4.3 1.7 BRD3 -3.2 -2.4 FARSLA 4.1 4.8 SMARCA2 -3.2 -2.6 TTK 4.1 4.8 STAMBP -3.2 -2.4 BIRC4 3.9 3.2 C9orf55 -3.2 -11.1 EIF3S1 3.7 4.3 NCOA2 -3.0 -5.7 C1orf77 3.7 3.6 CRAMP1L -3.0 -5.6 CEP76 3.6 3.0 PCSK7 -2.9 -1.7 LUZP1 3.6 3.2 CDKN1B -2.9 -3.7 ZNF672 3.4 3.1 RUNX1 -2.7 -4.2 ZNF146 3.1 2.7 IL18BP -2.7 -3.1 ZC3H12A 3.1 4.7 HECTD1 -2.7 -4.4 KPNA4 3.0 1.7 BTN3A3 -2.6 -3.6 RMI1 3.0 3.5 EIF2B5 -2.6 -2.2 ANKRD12 2.9 2.9

100

For the genes that increased with stimulation, it is likely that the 24h time point used in the microarray represents a window that is too early for repression to be observed.

Indeed, LTA which encodes TNFβ, is one of the most highly up-regulated transcripts at

24h, but PRDM1-dependent repression is observed at 48h as previously revealed by knockdown experiments (Figure 34). Other molecules associated with NK activation and bound by PRDM1 likely behave in a similar manner. IBRDC3 (also known as NK- lytic associated molecule) is a RING domain-containing E3 ligase that is expressed at low levels in resting NK cells, but induced upon activation (215-217).

Numerous transcripts show either no change in expression levels or are not expressed in NK cells. For instance, IL2 has been documented as a PRDM1 target gene in T cells (67), but is not expressed in NK cells. PRDM1 binding to regulatory regions of the IL2 locus may serve to reinforce its silencing in NK cells. Another target gene that is expressed at very low levels in NK cells is LMO2. This gene encodes the Lim-only domain protein which is normally down-regulated during the transition of germinal center B cells to plasma cells and expression levels are predictive of survival in diffuse large B cell lymphoma patients (218).

Numerous genes that are targets of PRDM1 regulation in other lymphoid lineages

(such as PAX5, IFNB and SPIB) are not bound by PRDM1 in NK cells. A recent ChIP- chip analysis in the U266 myeloma cell line identified 282 unique target genes (219), of which only 13 (~5%) were found to overlap with the NK target genes identified in this study. This study utilized a different immunoprecipitating antibody and a different peak detection algorithm, which likely contribute to the low degree of overlap. However,

101

despite these experimental differences, binding patterns are likely highly divergent across

cellular lineages.

Identification of PRDM1-regulated genes by microarray analysis

ChIP-chip analysis is useful for the de novo identification of PRDM1 binding

sites, but does not facilitate characterization of the associated changes in gene expression

of the target genes. In order to identify genes that are responsive to PRDM1, a second

gene expression profiling microarray analysis was conducted. NK cells were isolated

from two healthy donors and treated with either IL-2 and IL-12 or IL-2, IL-12 and IL-18

for 48h. For both sets of stimulations, cells were incubated with either a PRDM1-specific

siRNA or a non-targeting siRNA. Samples containing IL-2 alone were also included for

each donor for a total of ten samples. Immunoblotting was performed on an aliquot of

each sample to confirm PRDM1 induction and knockdown efficiency. As expected, IL-2

and IL-12 induced PRDM1, with the addition of IL-18 further increasing PRDM1

expression and the presence of PRDM1-specific siRNA significantly down-regulated

protein expression in both conditions (Figure 44).

Figure 44: Immunoblots of PRDM1 knockdown used for microarray analysis. NK cells isolated from two donors were stimulated with either IL-2, IL-2 + IL-12 or IL-2 + IL-12 + IL-18 for 48 h. For IL-2 + IL-12 and IL-12 + IL-12 + IL-18, samples were treated with either a non-targeting or PRDM1-specific siRNAs. PRDM1 expression was analyzed by immunoblot.

102

A large number of genes were found to be differentially expressed concomitantly

with PRDM1 knockdown. Using IL-2 and IL-12 stimulation, a total of 56 unique

transcripts were up-regulated, while 77 were down-regulated more than 1.3 fold in both donors. Of the up-regulated transcripts, three (5.4%) were also ChIP-chip targets (Table

3 and Appendix III). When IL-2, IL-12 and IL-18 were used for stimulation, a larger number of transcripts were differentially expressed. A total of 174 transcripts were up- regulated, while 82 were down-regulated 1.3 fold or more in both donors with PRDM1 knockdown. Of the up-regulated transcripts, 13 (7.5%) were found to be ChIP-chip targets. The three transcripts changed with IL-2 + IL-12 treatment were also up-regulated in response to PRDM1 knockdown with IL-2, IL-12 and IL-18 treatment. Many of the expression changes associated with PRDM1 knockdown are not associated with ChIP- chip regions and likely result from indirect mechanisms. This is consistent with microarray experiments conducted in plasma cells, where only a small fraction of the hundreds of differentially expressed transcripts have been demonstrated to be direct targets of PRDM1 binding (46, 47).

103

Table 3: Target genes increased > 1.3 fold with PRDM1 knockdown (Top 30) 2+12 Up 2+12+18 Up

Gene ID Donor 1 Donor 2 Gene ID Donor 1 Donor 2

FLJ25791 4.72 1.78 C10orf97 8.64 1.31

HBA1 3.29 1.31 HIPK3 5.82 1.91

CXCL10 2.69 1.61 RAB8B 5.31 1.70

IL1R1 2.52 2.10 LRAP 4.96 2.04

DLL1 2.19 1.35 PSMA3 4.70 1.45

FAM26F 2.07 1.54 WHDC1L1 4.61 2.29

PTK2 2.03 2.12 RINZF 4.05 1.44

CD160 1.99 2.78 CXCL11 4.02 2.30

IPO11 1.93 1.65 SCOC 3.92 1.84

VNN2 1.87 1.53 SP1 3.91 1.42

DDX17 1.75 3.22 SGPP1 3.88 1.81

SLC16A7 1.73 1.50 MAF 3.88 1.37

DMRTA2 1.71 1.42 MLSTD2 3.85 1.92

HBA2 1.66 1.41 ZNF230 3.77 1.49

TNFRSF10B 1.64 1.88 NR4A3 3.70 1.32

CBX3 1.64 1.67 PDCD4 3.47 1.71

CLK4 1.57 1.35 FLI1 3.43 1.57

BCAR3 1.53 1.54 MBNL1 3.42 2.00

VEGF 1.51 1.49 MST4 3.33 1.33

JAK2 1.50 1.31 ITGA4 3.09 1.89

GART 1.50 1.38 NEDD1 3.08 1.43

ZAK 1.49 1.76 CAPRIN1 3.06 1.33

PHOSPHO2 1.46 1.47 B4GALT6 3.05 1.70

ZNF238 1.46 1.54 JAK1 3.03 1.41

FBXO8 1.45 1.37 MGC33214 2.98 1.59

FLJ37034 1.45 1.53 PSAT1 2.94 1.55

SOCS1 1.44 1.32 EIF4A2 2.92 1.35

TRFP 1.43 1.33 NCK1 2.89 1.57

IDH1 1.43 1.45 IGJ 2.86 1.84

TFAM 1.43 1.32 MCL1 2.79 1.44

This analysis has led to the identification of a variety of direct targets of PRDM1- mediated repression that likely contribute to NK function. Importantly, levels of these transcripts are altered in the absence of PRDM1 (Table 4). KLRD1, better known as

CD94, heterodimerizes with members of the NKG2 family via disulfide bonding to form biologically-active killer cell lectin-like receptors. CD94 can participate in both activating and inhibitory signaling depending upon the NKG2 family member present

104

(110). For instance, CD94:NKG forms an activating receptor, while CD94:NKG2A

forms an inhibitory receptor (220, 221). BCL2L11 encodes the pro-apoptotic molecule

Table 4: Direct PRDM1 target genes: increased > 1.3 fold Gene ID Donor 1 Donor 2 Protein Function BCL2L11 2.10 1.82 Bim; pro-apoptotic molecule ETV6 1.73 1.40 Ets family transcription factor; hematopoiesis FLCN 1.68 1.35 folliculin; tumor suppressor KLRD1 1.87 1.44 CD94; heterodimerization with NKG2 family members LRMP 1.70 1.71 lymphoid-restricted membrane protein; endosomal sorting, recycling MBNL1 3.42 2.00 RNA processing NR4A3 3.70 1.32 NOR1; nuclear receptor; tumor suppressor, hematopoiesis NUMA1 1.89 1.56 mitotic spindle assemble PTK2 1.80 1.46 FAK; response to integrin signaling RAB6IP1 1.70 1.54 modification of guanine exchange factor activity SH2B3 2.34 1.59 LNK; adaptor molecule, negative regulation of JAK-STAT signaling STAT6 2.68 1.76 IL-4-induced IFNg production in NK WHDC1L1 4.61 2.29 homologous to WASp; cytoskeletal rearrangement * bold genes were also changed with PRDM1 knockdown using IL-2 and IL-12 stimulation

Bim, which has been shown to be up-regulated during apoptosis driven by IL-15

withdrawal in murine NK cells (222). BCL2L11 is up-regulated in response to IL-2, IL-

12 and IL-18 signaling, therefore, post-induction repression by PRDM1 may prevent stimulation-induced apoptosis. STAT6 is required for IL-4 induced IFNγ production in vitro and in vivo (223, 224). PTK2 encodes FAK, which mediates up-regulation of cytotoxicity via cytoskeletal rearrangements in response to integrin signaling. Finally,

WHDCL1 is a poorly-characterized gene, however, it shares significant homology with the Wiskott-Aldrich syndrome protein (WASp) which functions at the NK immune

synapse and is critical for F-actin accumulation (225). Thus, PRDM1 modulates NK

function through the transcriptional modulation of extracellular receptors, signaling

molecules and regulatory components of the cytoskeleton.

A variety of other direct target genes have not been attributed directly to NK

function, yet their regulation may be crucial in order to maintain homeostasis. LRMP

105

encodes an ER-resident protein that is involved in TAP-independent MHC Class I

processing and is normally silenced as activated germinal center B cells differentiate into

plasma cells (226, 227). LNK, encoded by SH2B3, is a lymphoid-specific signaling

adaptor molecule that negatively regulates proliferation in response to a variety of

cytokines (228, 229). Both NR4A3 and ETV6 encode transcription factors that are

required for hematopoiesis. NR4A3 is also induced in response to LPS in activated

macrophages, suggesting a role in inflammatory responses (230). ETV6 is an ETS-

family transcription factor that normally functions in hematopoiesis and vascular

development. Also known as TEL1, ETV6 is an oncogene and numerous fusion proteins

(eg. ETV6-RUNX1, ETV6-PAX5, ETV6-ABL) have been identified in both lymphoid

and myeloid-derived malignancies (231).

In order to confirm the transcriptional changes associated with PRDM1

knockdown, qRT-PCR analysis was conducted for genes which changed with PRDM1

knockdown or were ChIP-chip targets. For these experiments, NK cells obtained from

three additional donors were stimulated with either IL-2 and IL-12 or IL-2, IL-12 and IL-

18 and treated with either non-targeting or PRDM1-specific siRNAs for 48h exactly as in

the knockdown microarray experiment (Figure 45). Immunoblotting confirmed efficient knockdown in all donors using both stimulations. Consistent with results obtained in the microarray experiment, a larger number of genes was differentially expressed when IL-18

was present. In both experiments, CXCL10, TNFRSF10B and PTK2 were up-regulated in

response to PRDM1 knockdown. Based upon microarray results comparing freshly

isolated to 24h stimulated NK cells, expression of CXCL10 is strongly up-regulated,

while the expression of its receptor CXCR3 is down-regulated. Thus, PRDM1 may help

106

Figure 45: Knockdown of PRDM1 increases mRNA levels of selected PRDM1 target genes. A) NK cells from three donors were stimulated for 48h in the presence of IL-2 and IL-12, treated with either PRDM1-specific siRNA or non-targeting siRNA. Relative levels of target gene mRNA were determined by qRT-PCR, with normalization to 18S. Error bars represent SD. B) Same analysis using IL-2, IL-12 and IL-18 for stimulation.

restore the CXCL10:CXCR3 ratio to attenuate stimulation-induced migration. Indeed, sustained CXCL10 expression by NK cells is associated with chronic colitis (232).

TNFRSF10B encodes a TRAIL receptor, also known as DR5 or KILLER, which is

capable of transducing a pro-apoptotic signal due to the presence of an intracellular death

domain upon ligand engagement. Down-regulation of this molecule may be a mechanism to prevent fratricidal killing of NK cells via TRAIL-DR5 interaction. PTK2 was also up-regulated and is covered in more detail in the next chapter. Up-regulation of

SELL, which encodes L-selectin, was observed with IL-2, IL-12 and IL-18 stimulation and may have functional consequences in terms of migratory capacity. BCL10, LRMP,

107

KLRD1, ITGA4 and IBRDC3 were also all slightly up-regulated in these experiments.

Thus, knockdown of PRDM1 leads to differential expression of a variety of genes

involved in modulating NK activity.

Discussion

The experiments conducted in this chapter aimed to define the global interactions

of PRDM1 in NK cells and examine the transcriptional changes associated with PRDM1

knockdown. Numerous novel PRDM1 target genes were identified, which modulate

function through well-characterized mechanisms such as NK receptors, cytokine

signaling and regulation of apoptosis. Additionally, many PRDM1 target genes may

regulate NK biology through mechanisms not previously associated with NK function.

Using ChIP-chip in NKL cells to identify PRDM1 target genes has several

caveats that must be taken into consideration. The NKL cell line is considered to

phenotypically mimic activated NK cells and does express the PRDM1 isoform pattern

observed with primary cells. However, NKL cells are an immortalized tumor cell line

derived from a patient with LGL-leukemia, have numerous karyotype abnormalities and are “addicted” to IL-2 signaling (233). NKL cells may have adapted novel mechanisms to regulate gene expression to maintain continuous growth. Furthermore, the response to stimulation with IL-12 and IL-18 is vastly attenuated in NKL cells. For example, up- regulation of IFNG transcription as measured by qRT-PCR is > 1,000 fold lower in NKL

cells when compared to primary cells. Consistent with this, PRDM1 is not found

associated with chromatin at the IFNG promoter in NKL cells. Thus, dynamic changes in

PRDM1 target changes may not be recapitulated in culture.

108

The binding of a transcription factor to specific regions near genes can have several potential outcomes. Generally, one would expect that a transcriptional repressor, such as PRDM1, would be associated with gene silencing and that removal of the factor

(via knockdown) should result in increased expression. At least four functional outcomes for target gene expression in response to PRDM1 binding exist: 1) silencing of genes expressed in the resting state, 2) post-induction silencing, 3) repression of non-NK genes

(maintenance) and 4) non-productive binding (no silencing). Over a quarter of the ChIP- chip target genes (27.6%; 77/278 genes, for which expression data was available) decreased upon stimulation, consistent with PRDM1 silencing genes normally expressed in resting NK cells. A similar number of PRDM1 target genes were increased upon stimulation (30.6%). Among these, LTA is one of the most highly induced genes at 24h stimulation. However, at 48h expression differences were observed when PRDM1 was abrogated via siRNA knockdown. IFNG also behaves in a similar manner, although it is only bound by PRDM1 in primary NK cells. IL2 and LMO2 are two examples of genes that are known PRDM1 targets in other lineages, but are not expressed in NK cells.

The observation that only 4.6% (13/278) of the ChIP-chip target genes exhibited increased expression in the absence of PRDM1 is consistent with many published reports

(210). Indeed, only a subset of PRDM1 target genes were dynamically expressed in response to PRDM1 knockdown in U266 cells (219). A variety of technical and biological factors may be involved in this phenomenon. On the technical side, it is possible that the remaining fraction of PRDM1 after knockdown is sufficient to maintain biological activity. Another possibility is that despite the presence of a robust band on immunoblots, PRDM1 may only be induced in a subset of NK cells. Thus, RNA lysates

109

from the total population may be biased by the non-PRDM1 expressing fraction. It may be possible to address this issue by the development of a FACS-based assay utilizing intra-cellular staining for PRDM1. Co-staining for effector cytokine targets IFNγ and

TNFα could be performed simultaneously using this approach. Finally, because the microarray experiments were conducted using purified primary NK cells, some experimental variability is unavoidable due to inherent differences existing among donors.

In terms of biological explanations for lack of response to PRDM1 knockdown, there are several possibilities. First of all, removal of a repressor may simply be insufficient to re-activate transcription. For example, PRDM1 binds the ZAP70 promoter and its expression decreases as PRDM1 is activated. However, ZAP70 levels do not change in response to PRDM1 knockdown. Another factor, such as NFAT or AP-1, may be required in order to activate expression. It has also been suggested that IRFs compete for PRDM1 binding sites (13, 219) and the presence of IRFs may abrogate silencing by

PRDM1. NK cells specifically up-regulate IRF8 in response to IL-2, IL-12 and IL-18 stimulation. In dendritic cells, PRDM1 and IRF8 directly compete for regulatory sites present within the myeloid-specific CIITApI (75). Whether or not such a mechanism exists in NK cells remains to be determined as no data is available for IRF8 function in

NK cells. Finally, cytokine stimulation is required to induce PRDM1 expression, but also leads to massive changes in the global NK transcriptional profile. This adds a complicating factor and may mask PRDM1-mediated changes. Consistent with this notion, in the PRDM1 knockdown microarray far fewer genes change with IL-2 and IL-

12 stimulation when compared to IL-2, IL-12 and IL-18 stimulation.

110

In the ENCODE project, a large portion of sequence-specific transcription factors

were found outside of what would typically be referred to as promoter regions (234).

This promiscuity in binding has several important correlates with potential biological

functions. First of all, the classical minimal promoter models in which activators and

repressors bind to discrete regulatory elements present < 1,000bp upstream of the

transcriptional start site (TSS) are most certainly oversimplified. Binding of transcription

factors to DNA is likely a relatively widespread phenomenon. Even using an array that

contains almost exclusively promoter sequences, PRDM1 binding shows nearly uniform

binding across the entire tiled region with no increased enrichment at locations < 1kb

upstream of the TSS. Similarly widespread binding has recently been reported for ChIP- chip studies in Drosophila with the heat shock factor HSF wherein the authors observed that only 11% of the 454 identified regions were found within 1,250bp of the TSS (235).

Additionally, no correlation exists between genes that were changed in response to

PRDM1 knockdown and distance of binding peak from the TSS because only three of the

thirteen changed genes demonstrated enrichment in the region between the TSS and -1kb

upstream. The presence of a PRDM1 binding motif (defined as MAGYGAAAGYK)

may not be required for binding as many of the validated PRDM1 target genes did not

contain this consensus sequence. However, eight of the thirteen genes differentially

expressed upon knockdown did contain matches to the motif.

In NK cells, PRDM1 appears to act as a modulator of effector response rather

than a master regulator of a differentiation program. Thus, PRDM1 function in NK cells

appears to be more similar to its role in T cells, than B cells (181). PRDM1 target genes

are not generally fully silenced in NK cells. This may be explained by the fact that

111

multiple isoforms are uniquely present in NK cells and knockdown experiments often

result in no more than a two- to three-fold increase in expression of the target gene in the

absence of PRDM1. The truncated PRDM1β-isoform is known to exhibit somewhat diminished repressive capacity and the PRDM1Δexon6 is predicted to have a disrupted zinc finger domain. Thus, genes may not be completely silenced due to these isoforms competing with the full-length PRDM1α, the only isoform observed in B cells. It is also possible that the disruption of the zinc fingers in PRDM1Δexon6 alters the affinity for specific recognition motifs, contributing to the large divergence in target genes between lymphoid lineages.

Modulation of expression levels of a large number of target genes through both direct and indirect mechanisms may be a key feature of PRDM1 in NK cells. In previous chapters, effector cytokines were identified as targets of regulation and here a key NK receptor (KLRD1), signaling adaptors (ZAP70, SH2B3, STAT6, BCL10) and a variety of other factors (eg. IBRDC3, LRMP) were identified. Diminished expression of a number of these genes may allow PRDM1 to serve as a rheostat to fine-tune the activated phenotype. In murine cells (which only express the full-length PRDM1α), ablation of

PRDM1 expression was recently shown to increase proliferative capacity and increase

NK-mediated clearance of RMAS target cells in vivo (236). Given the prevalence of the

6q21 deletion in NK lymphomas, it will be important to determine the mechanisms by which PRDM1 suppresses growth potential. Silencing of genes with oncogenic potential may be an important function of PRDM1 in NK cells.

112

CHAPTER V:

PRDM1-MEDIATED REGULATION OF FOCAL ADHESION KINASE

Introduction

Integrin signaling plays a critical role in numerous cellular processes such as

proliferation, migration and differentiation. Integrins on the cell surface function as receptors for various components of the extra cellular matrix (ECM). Fibronectin, a major component of the ECM, specifically interacts with α and β integrin receptors via the Arg-Gly-Asp sequence present in the central portion of the molecule. Human NK cells are functionally responsive to fibronectin-mediated signaling because they express

β1 and α4/α5 integrin receptors (237). Thus, integrin signaling has the potential to modulate NK function.

Focal adhesion kinase (FAK) is a tyrosine kinase that was originally identified as a 125kD protein which was phosphorylated in response to fibronectin (238, 239). FAK was later shown to integrate multiple signal transduction pathways in a variety of cell types. FAK is encoded by the PTK2 gene on chromosome eight. FAK is phosphorylated on several tyrosine residues including Tyr397, Tyr576/577 and Tyr925 in response to integrin signaling (240). These phosphorylation events increase enzymatic activity and create

novel binding sites for SH2 domain-containing proteins. FAK interacts with a variety of

adaptor molecules that eventually lead to the cytoskeletal rearrangements via actin-

113

mobilization through several alternative pathways. Several important intermediate molecules in the FAK signaling cascade include: SRC, p130CAS, calpain and paxillin.

FAK signaling is crucial for adhesion, migration and survival. Consequently, it is

associated with metastatic potential in a variety of tumors and FAK is over-expressed in a variety of solid tumors (240). Knockdown of FAK expression disrupts lamellipodia forming the “leading edge” resulting in defects in directional migration (241). FAK also mediates survival in response to inflammatory stimuli. This process requires both NF-κB and ERK signaling and FAK-/- fibroblasts display increased sensitivity to TNFα-induced apoptosis (242).

FAK activity also mediates a variety of functions within the adaptive immune

system. In T cells, FAK signaling in response to β1 integrin ligation increases migration,

but does not affect adherence to fibronectin (243). FAK signaling is independent from

CD3/TCR signaling, although some downstream substrates such as CAS-L are shared

among these pathways (244). In B cells, FAK is phosphorylated in response to integrin

signaling downstream of Rap GTPases. This phosphorylation is enhanced by BCR

signaling and is required for actin-dependent morphological changes associated with

activation (245). FAK likely also plays critical functions in early development of both T

and B lymphocytes via α4β1 intregrin:fibronectin interactions in the thymus and bone

marrow (246, 247).

In NK cells, FAK has been shown to increase cytotoxic capacity and has been

implicated in survival. Exposure of NK cells to full-length fibronectin or “active”

fragments containing integrin-interacting domains result in the phosphorylation of paxillin and several higher molecular weight proteins, one of which was later identified

114

as FAK (248). Fibronectin exposure increases the association of FAK with the intracellular kinases Fyn and ZAP70 suggesting the existence of an active signal transduction pathway (249). Furthermore, fibronectin dose-dependently increases NK cytotoxicity against a variety of target cells (250). Incubation with fibronectin also maintains the survival of murine NK cells in the absence of IL-2 or IL-15 (251).

Collectively, these data suggest that fibronectin signaling mediated through integrins and

FAK may be an important regulator of NK activity and survival. Mechanisms of FAK expression in NK cells have not been elucidated. Here, a role for PRDM1 in the negative regulation of PTK2 mRNA and FAK protein levels is established.

Results

PRDM1 binds to the FAK promoter

By ChIP-chip analysis PRDM1 was found to bind to a distal region of the PTK2 promoter at approximately 1,200bp upstream of the TSS. (Figure 46A). The defined peak region spans approximately 400bp and contains two potential PRDM1 binding sites at –1120 to –1110 and -1290 to -1279 upstream of the TSS, respectively. Both of these motifs are closely-related to the MAGYGAAAGYK consensus sequence used to identify potential PRDM1 target genes. Primers spanning this region were designed and used in

ChIP analysis in several types of NK cells. PRDM1 was found to occupy this region in the NKL cell line using independent chromatin preparations, thus confirming the ChIP- chip results (Figure 46D). Furthermore, PRDM1 was found to bind this region in both primary NK cells stimulated for 24h to induce PRDM1 and LAK cells derived from three

115

different donors (Figure 46B, C). Thus, PTK2 is a PRDM1 target gene within the NK lineage.

Figure 46: PRDM1 occupies the distal PTK2 promoter. A) Representative ChIP-chip result for PRDM1 binding to regions upstream of PTK2. B) Chromatin was prepared from primary NK cells isolated from three healthy donors after 24h stimulation with IL-2 (100U/mL), IL-12 (10ng/mL) and IL-18 (100ng/mL). ChIP analysis was performed using primers amplifying a region approximately 1,200bp upstream of the transcriptional start site. C) Similar analysis in LAK cells generated from three healthy donors. D) Similar analysis in triplicate from the NKL cell line.

PRDM1 regulates FAK expression

Given that PRDM1 binds the PTK2 promoter in vivo as revealed by ChIP,

knockdown experiments were undertaken to evaluate whether PTK2 mRNA and FAK

protein levels were elevated in the absence of PRDM1. Primary NK cells were isolated

and treated with either IL-2 and IL-12 or IL-2, IL-12 and IL-18 for 48h in the presence of

116

either a PRDM1-specific or non-targeting siRNA. During a four day culture period,

PTK2 mRNA levels are increased concomitantly with PRDM1 knockdown, consistent

with PRDM1’s role as a negative transcriptional regulator (Figure 47).

Figure 47: Knockdown of PRDM1 increases PTK2 mRNA expression. Purified NK cells from multiple donors were incubated for 48h in the presence of either a non- targeting (NT) or PRDM1-specific (KD) siRNA. Cells were stimulated with IL-2 (100U/mL) and IL-12 (10ng/mL) with or without IL-18 (100ng/mL). RNA was isolated and analyzed by RT-CR. Data was normalized to 18S and fold change calculated by the ΔΔCt method, with the non-targeting sample for each donor set to “1”. Data represent the mean of four (2+12) or three (2+12+18) donors +/- SD.

The observed increase in PTK2 mRNA with PRDM1 knockdown was greater

with IL-2 and IL-12 stimulation than for the combination of IL-2, IL-12 and IL-18.

Consistent with these results, protein levels for FAK were also increased when NK cells

were subjected to siRNA-mediated knockdown of PRDM1 during IL-2 and IL-12

stimulation (Figure 48). In several experiments, FAK protein levels showed minimal

response to PRDM1 knockdown when IL-2, IL-12 and IL-18 was used for stimulation

(data not shown). It is likely that the addition of IL-18 results in substantially higher levels of NF-κB activation. NF-κB has been shown to positively regulate FAK promoter activity (177) and thus may override the repressive potential of PRDM1.

117

Figure 48: Knockdown of PRDM1 increases FAK protein expression. Purified NK cells from three donors were incubated for 48h in the presence of either a non-targeting (NT) or PRDM1-specific (KD) siRNA. Cells were stimulated with IL-2 (100U/mL) and IL-12 (10ng/mL) and analyzed by immunoblot for PRDM1 and FAK expression. β-actin served as loading control.

Time course experiments were conducted to analyze the kinetics of PTK2 expression during ex vivo stimulation. NK cells were isolated and stimulated for up to four days with IL-2 and IL-12. qRT-PCR was conducted on samples harvested at 24h,

48h, 72h and 96h. In multiple donors, PTK2 mRNA levels decreased in a time- dependent manner (Figure 49A). Consistent with changes at the mRNA levels, FAK protein levels also diminish while PRDM1 levels remain elevated (Figure 49B). In order to support these findings, over-expression studies were carried out in the Jurkat T cell line. Transduction for 48h with PRDM1-GFP adenovirus diminishes both PTK2 mRNA and FAK protein levels, relative to GFP-only adenovirus (Figure 50). Consistent with these findings, FAK is not expressed in LAK cells after two to three weeks in culture

(data not shown). Collectively, these data confirm an inverse relationship between

PRDM1 and FAK, further supporting PRDM1’s role as a negative regulator of FAK.

118

Figure 49: Time-dependent reduction in PTK2 mRNA and FAK protein levels. A) Primary NK cells stimulated with IL-2 (100U/mL) and IL-12 (10ng/mL) for four days. RT-PCR was conducted for PTK2 mRNA on cDNA synthesized from RNA isolated at different time points using 18S control gene. B) Representative immunoblot for PRDM1 expression at various timepoints.

119

Figure 50: Over-expression of PRDM1 decreases FAK expression. Jurkat cells were transduced for 48h with adenovirus expressing either PRDM1 and GFP or GFP alone. A) RT-PCR analysis of PTK2 mRNA levels. B) Representative immunoblot analysis demonstrating that FAK levels are diminished in the presence of PRDM1.

PRDM1 suppresses PTK2 promoter activity

In order to directly assess the ability of PRDM1 to mediate transcriptional control

over FAK levels, luciferase experiments were carried out. Based on the presence of two

potential PRDM1 binding sites upstream of the PTK2 TSS, constructs with progressive

deletions of the 5’ promoter region cloned upstream of luciferase were utilized to monitor

PRDM1-mediated repression. Co-transfection of PRDM1 resulted in ~30% reduction in

luciferase activity from the pGL3-PTK2-1400 construct (Figure 51). This level of

repression was consistent with repression observed with a known PRDM1 target gene,

CIITA. Deletion of 227bp to generate pGL3-PTK2-1173 diminished, but did not

eliminate, the repressive ability of PRDM1. Further deletion of an additional 153 bp to generate pGL3-PTK2-1070 completely blocked repression by PRDM1. Thus, multiple elements likely contribute to PRDM1-mediated suppression because step-wise deletion of

120

these regions results in additive loss of repressive ability. As expected, no effects of

PRDM1 were observed using pGL3-PTK2-564. Importantly, luciferase assay results

confirm the functionality of the region identified by ChIP-chip.

Figure 51: PRDM1 mediates repression through regulatory elements upstream of PTK2. Deletion constructs were used in luciferase experiments. Jurkat T cells were transfected with either empty vector or PRDM1, harvested at 36h and analyzed by luciferase assay. Potential PRDM1-binding sites are shown as dashes. Data shown are averages of three independent experiments +/- SD.

Discussion

Data presented in this chapter define PTK2 as a direct target of PRDM1 regulation in human NK cells. PRDM1 is bound to two sites in the distal promoter in NK cell lines, stimulated primary NK cells and LAK cells. Furthermore, reduction of

PRDM1 via siRNA results increased PTK2 mRNA and FAK protein, while over

121

expression represses PTK2 mRNA and FAK protein. Finally, luciferase assays establish

that PRDM1’s repressive effects are mediated through two regions in the distal promoter.

Repression of FAK by PRDM1 in response to stimulation is consistent with

PRDM1’s role as a negative regulator of activation. In response to stimulation, PRDM1

is induced leading to reduction in FAK levels. Consistent with this, LAK cells which

acquire high levels of PRDM1 lose FAK expression during culture. A higher level of repression is observed when IL-18 is not included in the stimulation cocktail likely because IL-18 activates NFκB, a known activator of PTK2 (177). Thus, it is possible that

PRDM1-mediated repression of FAK may be over-ridden via an NFκB-dependent pathway. Nonetheless, such a mechanism may still serve to diminish responsiveness to integrin signaling during the attenuation phase of activation under specific stimulation conditions.

It has been difficult to experimentally elucidate the functional consequences of

FAK down-regulation in human NK cells. Initial functional experiments using adenoviral-mediated transduction of PRDM1 failed to demonstrate a difference in fibronectin-mediated survival as assessed by flow cytometry. Low transduction efficiency and non-specific cytotoxicity associated with adenoviral infection represent significant technical hurdles. It may be possible to utilize a FAK-specific siRNA or a

FAK-specific inhibitor to assess functional changes; however, these approaches circumvent PRDM1. It is also possible that NK cells can functionally compensate for diminished FAK expression via PYK2/FAK2 encoded by PTK2B. Indeed, this closely- related kinase is known to be activated in response to integrin signaling (252).

Furthermore, by microarray analysis PTK2B is expressed at considerably higher levels

122

than PTK2 and is further up-regulated two- to three-fold in response to stimulation.

Thus, the functional role of FAK in NK cells requires further investigation.

Interestingly, a microRNA (miR-151) resides within intron 22 of PTK2. This microRNA has not been extensively studied, but appears to be associated with malignant progression. In hepatocellular carcinoma (HCC), miR-151 mediates increases in

migratory capacity and metastatic potential (253). Levels of PTK2 mRNA correlate with levels of miR-151, indicating that both gene products are co-regulated. Consistent with

this, FAK over-expression is associated with metatastasis and poor survival rates in HCC

patients (254, 255). miR-151 targets the Rho GTP disassociation inhibitor RhoGDIA,

which blocks GTPase activity via disruption of nucleotide exchange activity and sub-

cellular localization. Knockdown of miR-151 increases RhoGDIA mRNA and protein,

resulting in significantly fewer metastatic nodules within the liver in vivo (253).

Suppression of miR-151 by PRDM1 may be an important tumor suppressive

mechanism. miR-151 is expressed in acute lymphoblastic leukemia (ALL) and is

specifically down-regulated in patient samples of T cell origin relative to those of B cell

origin. Among the several thousand potential target genes, miR-151 is predicted to target

ITK, ZAP70 and several interferon-related genes (256). Other potential targets identified

in MIRBase include: IL2Rγ, IL12Rβ2, RELA and KIR2DL1. It will be interesting to determine the levels of miR-151 in NK-derived malignancies, as well as in normal NK cells in response to stimulation. It will also be important to determine whether a correlation exists between 6q21 deletion (PRDM1 loss) and over-expression of miR-151

in patient samples of various hematological malignancies. Thus, in addition to down-

123

regulation of protein-coding genes, PRDM1 may regulate miR-151 (and other microRNAs) to modulate cellular functions.

124

CHAPTER VI

HIGH-THROUGHPUT SCREENING IDENTIFICATION OF SMALL

MOLECULE INHIBITORS OF THE G9A HISTONE METHYLTRANSFERASE

Introduction

A substantial portion of the human genome is devoted to transcriptional regulation and an estimated 1,400 genes encoding proteins that function as sequence- specific transcription factors have been identified (210). The ability to directly regulate transcriptional activity through the use of small molecule inhibitors is a long-standing goal of the pharmaceutical industry, yet transcription factors are largely considered to be

“undruggable” targets. It is now widely accepted that transcription factors do not function in isolation and frequently require the recruitment of histone-modifying enzymes to mediate both transcriptional activation and repression. Thus, by targeting enzymes which are recruited by transcription factors it may be possible to modulate transcription via disruption of the associated enzymatic activity with small molecule inhibitors.

The development of inhibitors of DNA methyltransferases and histone deacetylases has established that “epigenetic” therapy is a viable therapeutic stategy. The nucleoside analog 5’-Aza-cytidine (Decatabaine), a competitive inhibitor of DNA methyltransferase I (DNMT1), has been evaluated for efficacy in a variety of hematological disorders and is FDA-approved for the treatment of myelodysplastic

125

syndrome (257). Suberoylanilide hydroxamic acid (SAHA) is a histone deacetylase inhibitor and was approved by the FDA in 2006 for the treatment of cutaneous T cell lymphoma (258). SAHA treatment leads to growth arrest and apoptosis of transformed cells through several mechanisms, including up-regulation of pro-apoptotic proteins of the Bcl-2 family (259). Novel histone deacetylase inhibitors, some of which exhibit isoform selectivity, are in development for the treatment of a variety of disorders and numerous clinical trials are ongoing.

Inhibition of histone methyltransferases represents an attractive target for epigenetic therapy. However, relatively few small molecule inhibitors of methyltransferases have been described. At the outset of this project, the only described methyltransferase inhibitor was a compound identified from a natural product library, chaetocin (260). This compound preferentially inhibits the Suv(var)3-9 enzyme, which tri-methylates histone H3K9. However, this compound exhibits high toxicity in mammalian cells and is difficult to synthesize due to its complex chemical structure.

Thus, a need exists for the identification of novel inhibitors of histone methyltransferases.

Alterations in epigenetic patterns normally maintained through histone and DNA methylation are characteristic of human malignancies. Therapeutic approaches which modulate the activity histone methylating enzymes may, alone or in combination with other epigenetic modifiers, contribute to phenotypic reversion (261). G9a is the major enzyme responsible for H3K9me2, is functionally linked with the DNA methylation machinery and is over-expressed in a variety of tumors. siRNA-mediated knockdown of

G9a reduces proliferation, diminishes anchorage-independent growth and induces apoptosis in a lung cancer cell culture model (262). G9a is also required to maintain the

126

malignant phenotype in vitro and in vivo (263). In breast cancer cells, 5’-Aza-cytidine treatment reduces G9a levels and decreases global H3K9me2 levels leading to reactivation of the tumor suppressor gene MASPIN (264). In colorectal cancer cells,

H3K9me2 is a crucial component of silencing the tumor suppressor MHL1 and loss of

H3K9me2 is critical to achieve re-activation in response to 5’-Aza-cytidine.

Given that G9a plays a critical role in cancer progression and is known to be recruited by PRDM1 to silence its target genes, a screen was developed to identify novel small molecule inhibitors of G9a activity. Data presented here describe the development of a high throughput screening assay and the identification of a novel in vitro inhibitor.

Results

G9a assay development

PRDM1 recruits the histone methyltransferase G9a to mediate silencing of target

genes. Identification of a G9a-specific inhibitor could potentially provide a means to

selectively regulate gene expression due to its specific targeting to promoters via

transcription factors such as PRDM1. Typically, methylation of histones has been

monitored in vitro using incorporation of radio-labeled S-adenosyl methionine (SAM)

into core histone substrates (265). These assays are time-consuming, poorly quantitative

and incompatible with large sample numbers. Thus, a mass spectrometry approach was

developed to assay histone methylation with the aim of conducting a small molecule

inhibitor screen.

The histone methyltransferase G9a preferentially catalyzes the di-methylation of

histone H3 lysine 9 (H3K9) and to a lesser extent H3K27 (156). Bacterially-expressed

127

recombinant G9a was used to methylate a synthetic 20-mer peptide, representing the amino-terminal tail of human histone H3 (ARTKQTARKSTGGKAPRKQL). By

utilizing residues 1-20 as the substrate, the preferred methylation site is approximately centered within the peptide and the K27 is not present. Although other lysine residues

are present within the peptide (K4, K14, K18), they are not targets of G9a-mediated methylation as confirmed by MS/MS analysis (data not shown). As the methylation reaction proceeds, the H3 peptide (m/z 2183) initially accumulates mono-methylation

(m/z 2197), then di-methylation (m/z 2211) and finally tri-methylation (m/z 2225) if incubated for extended time periods (Figure 52). Using signal-to-noise ratios, each species can be quantified and relative methylation levels determined. Mono-methylation is the first product formed. However, mono-methylated peptide levels were never observed to accumulate above approximately 40% of the total peptide signal.

Components of the reaction were optimized in titration experiments to identify optimal concentrations of each (Figure 53). Optimal conditions were determined and used for scaling up the assay to 96-well plate format. Importantly, both SAM and peptide substrates were used at saturating concentrations.

128

Figure 52: MALDI-TOF MS analysis to monitor methylation. A) Example of MS analysis of HMT reaction. Reaction was quenched via addition of TFA after 2h incubation and analyzed by MALDI-TOF MS. MALDI-TOF mass spectrum indicates accumulation of methylated species. B) Same reaction, incubated for 24h.

129

Figure 53: Optimization of HMT assay conditions. Titration experiments were performed for individual reaction components. MALDI-TOF analysis was conducted to determine the degree of methylation. These values were used to optimize conditions for high throughput screening.

High-throughput screening

Using the reaction conditions established above, screening was carried out at the

High-throughput Screening Core Facility using the ChemDiv Chemical Diversity Set small molecule library. This library consisted of approximately 20,000 small molecules, representing hundreds of chemical backbones. The components of this library were resuspended in DMSO and arrayed across 96-well plates. Using robotic pipetting, individual histone methyltransferase reactions were conducted in the presence of small molecules and analyzed by MALDI-TOF MS to assess the degree of inhibition. To account for experimental variability, data were analyzed using normalization for each

130

plate. Several representative plate results are shown below (Figures 54 and 55). Each

data series consists of three points, showing the percent contribution to the overall MS

spectra for the unmodified, mono-methylated and di-methylated peptide species. Each

plate contains 96 samples (80 candidate compounds, 8 DMSO-only controls, 4 no enzyme controls and 4 no SAM controls).

Figure 54: Raw data from four selected plates. Blue lines represent un-modified peptide. No enzyme controls were included in the first four and last four wells of all plates and typically show nearly exclusive unmodified peptide. Magenta and yellow represent mono- and di-methylation, respectively. Loss of mono- and di-methyaltion signals and gain of unmodified signal indicates enzymatic inhibition.

131

Figure 55: Analyzed data from four selected plates. Data from the four plates above was analyzed in Excel. SD was determined for DMSO-only control wells on each plate (n = 8). Experimental samples were then graphed as deviations from control SD, expressed as the absolute value of the fold change. Large shifts (e.g. Sample A10, on plate 85) were considered “hits”.

After the initial rounds of screening and analysis, 8,320 small molecules had been

analyzed and numerous potential G9a inhibitors were identified. A total of 59 small

molecules were found to result in > 50% reduction in levels of di-methylation,

representing a 0.7% “hit” rate. In order to streamline the process of screening the

remainder of the library (11,680 small molecules), in silico docking studies were carried

out by the Molecular Modeling Core. Because the crystal structure of G9a was

unavailable, a homology model was constructed based on the crystal structure of the

closely-related SET domain-containing methyltransferase, DIM5, from Neurospora

crassa. Chemical structures of the 20,000 compounds in the library were docked into the

132

homology model and a docking score was generated. This value was an approximation of the free energy associated with interactions between the small molecule and the enzyme, with higher scores predicted to indicate higher affinity. These docking scores were ranked and used as a predictor of inhibition. Using the top 20% of these scores,

4,000 small molecules were identified. Of these, 2,316 were from the remaining unscreened portion of the library, while 1,684 had been previously screened. 1,600 of these 2,316 small molecules (representing 13.8% of the remaining library) were robotically spotted across 20 plates and screened at 50µM for inhibitory activity. No additional “hits” (as defined by > 50% inhibition) were found using this “cherry picking”

approach. A total of 9,920 small molecules were screened; 8,320 via an unbiased

approach and 1,600 using a structure-guided approach.

In vitro inhibitors of G9a

Of the 59 “hit” compounds identified, 35 were ordered from ChemDiv for further

testing. IC50 values were determined by MALDI-TOF analysis, using di-methylation levels as an endpoint; IC50 values were defined as the concentration at which dim-

methylation levels were half of the DMSO-only control levels. A relatively small

percentage of these compounds demonstrated significant inhibition when assayed at

concentrations lower than 100µM. In fact, only seven showed activity at 50µM. One

compound that did exhibit reproducible inhibition was HLM-02414. This compound is a

sulfamethazine derivative with a difluoro-substituted phenyl moiety on the opposite end

of the molecule (Figure 56). This chemical backbone is a suitable pharmacophore

133

because sulfamethazine derivatives have been used previously as antibacterials in a

variety of formulations.

Figure 56: Chemical structure of HLM-2414. Chemical structure of 4-(4-(2,4- difluorophenyl)-thiazol-2-ylamino)-N-(4,6-dimethylpyrimidin-2-yl)benezensulfonamide. The left portion of the molecule represents the sulfamethazine backbone.

HLM-2414 exhibited an initial IC50 value in the range of 50µM by MALDI-TOF

MS assay using a synthetic peptide substrate, even though it was not one of the strongest

inhibitors identified in the screen. However, inhibitory activity was significantly higher

using radioactive assays monitoring 3H-incorporation into core histones, a more physiologically-relevant substrate. Also, the chemical structure was found to be highly amenable to combinatorial chemistry approaches. Therefore, in collaboration with the

Chemistry Core Facility, a focused library of chemical derivatives was created based on this chemical backbone with the goal of improving inhibitory activity. Initial experiments demonstrated that the sulfamethazine backbone alone had no inhibitory activity (data not shown). Therefore, chemical modifications were targeted to the phenyl- thiazol portion of the molecule. A total of 17 derivatives were synthesized and IC50

values were determined using MS analysis of the original compound (synthesized “in-

134

house” as YL1-132) and the derivatives. The recently-published G9a inhibitor BIX-

01294 (synthesized as YG2-018) was also included for comparison (Figure 57).

Figure 57: IC50 values for various derivatives of HLM-2414 (YL1-132). Small molecules are shown above in concentric circles representing IC50 values. The outermost ring represents compounds with IC50 > 50µM, the middle ring represents compounds with IC50 between 25µM and 50µM, while the innermost ring represents compounds with IC50 < 25µM.

Modifications of the phenyl thiazol portion of the molecule resulted in significant

improvement of inhibitory activity. Substitution of the 2,4-difluorophenyl with 3,4-

dichlorphenyl (YL1-148-1) significantly improved inhibitory activity, resulting in a

compound with an IC50 value as low as 15µM. Alterations to the thiazol portion of the

molecule by addition of a methyl moiety (YL2-157) or removal of methyl groups from the pyrimidine ring of the sulfamethazine (YG2-101) did not abrogate inhibitory activity.

135

Other structure-function relationships, however, were difficult to elucidate. For example,

it was surprising that the ortho diphenyl substitution (YL1-116-1) maintained good

inhibitory activity and suggested that the elctronegativity associated with the dichlorophenyl moiety was not critical for inhibition. Importantly, five of the derivative molecules consistently inhibited G9a enzymatic activity to a greater extent than the BIX-

01294 compound. In silico docking experiments suggested that YL1-148-1 interacted strongly with the peptide-binding cleft, indicating a potential for substrate inhibition

(Figure 58). Furthermore, in methyltransferase assays YL1-148-1 inhibited HA-purified

Figure 58: In silico docking of YL1-148-1 to H3 peptide binding cleft. YL1-148-1 was “docked” into a homology model of G9a based on the crystal structure of Dim5 with the assistance of the Moffitt Molecular Modeling Core Facility. Docking suggests that YL1-148-1 may lie at the top of the catalytic cleft where the Lys residue (H3K9) must be inserted in order to receive the methyl group from the SAM co-factor. Binding at this site would displace the H3 N-terminal tail, consistent with YL1-148-1 acting as a substrate competitor.

136

human, full-length G9a with an IC50 of approximately 3-5µM as detected by radioactive

incorporation into core histones and immuoblotting for di-methylation of GST-H3 tails

(Figure 59).

Figure 59: YL-148-1 inhibits in vitro methylation of core histones. A) Histone methylation was monitored in vitro to determine IC50. Recombinant G9a was incubated with core histones (4µg) and a 3H-labeled SAM substrate. Reaction products were separated via SDS-PAGE and exposed to radiographic film. Results shown are from three independent experiments. B) Similar assays were conducted, using non-radioactive SAM and GST-H3 tails as substrates. Reactions were incubated with wither YL-148-1 or sulfamethazine, which lacks the pharmacophore. H3 tails were resolved by SDS-PAGE and analyzed by immunoblot using an antibody specific for H3K9me2.

137

G9a inhibitors fail to abrogate cellular functions

In order to further characterize the efficacy of these small molecule in vitro

inhibitors, cell-based assays were carried out. Several approaches were attempted to

demonstrate ability of the small molecules to abrogate G9a-dependent processes within

cells. First of all, reporter assays were employed in which G9a was targeted upstream of

luciferase via insertion of Gal4 binding sites upstream of the thymidine kinase promoter

(266). Co-transfection of a G9a-Gal4 fusion protein strongly represses luciferase activity

using this system. Significant repression was observed with co-transfection of a G9a-

Gal4 fusion protein, yet no de-repression was observed at doses significantly higher than

the in vitro determined IC50 (Figure 60A). Next, histones were acid-extracted from

U2OS cells that had been treated with increasing doses of YL1-148-1. These histones were analyzed via immunoblotting in order to monitor alterations in H3K9me2 levels in

response to treatment. By pre-treating U2OS cells for 24h with SAM-deficient media, a

reduction in H3K9me2 levels was observed, but treatment with YL-148-1 did not abrogate the recovery of H3K9me2 levels (Figure 60B).

138

Figure 60: YL-148-1 fails to inhibit G9a-dependent methylation in cell-based assays. A) U2OS cells were co-transfected with a G9a-repsonsive luciferase reporter with GAL4- binding sites upstream of the tk promoter. Luciferase assays were conducted at 24h post- transfection. B) U2OS cells were incubated 24h with SAM-deficient media, then media was replaced in the presence of increasing concentrations of YL1-148-1 for 24h. Histones were acid-extracted and H3K9me2 levels assessed by immunoblot. C) U2OS cells were induced to express IFNB by electroporation with 50µg pI:C. Samples were analyzed at 5h and 24h by qRT-PCR with normalization to 18S. D) THP-1 were “tolerized” by 16h treatment with LPS (1µg/ml) or left “responsive” by treatment with PBS. After 1h secondary LPS stimulation (1µg/ml), samples were analyzed by qRT- PCR. Experiments in A-D are representative of two to three independent experiments.

Given that G9a is required for silencing of numerous genes, two endogenous reporter systems were utilized to assess the ability of YL1-148-1 to interfere with G9a- mediated repression. G9a is required for post-induction silencing IFNB in response to polyI:C stimulation (15). Thus, inhibition of G9a enzymatic via a small molecule should result in the sustained expression of IFNB. When U2OS cells were treated with polyI:C,

IFNB were markedly elevated at early timepoints and diminished by 24 h, however, treatment with YL1-148-1 resulted in only marginal de-repression of IFNB (Figure 60C).

139

Similarly, THP-1 cells become “tolerized” to LPS stimulation and fail to induce TNF in response to secondary stimulation via a mechanism that requires G9a and histone methylation (267). However, no significant abrogation of tolerance was observed in the presence of YL1-148-1 (Figure 60D). Collectively, these data suggest that although

YL1-148-1 inhibits G9a-mediated methylation in vitro, it does not inhibit histone methylation within a cellular context.

Discussion

Data presented here describe the development of a mass spectrometry-based screening assay of G9a activity, which was subsequently used to identify in vitro inhibitors of enzymatic activity. The use of mass spectrometry allowed a relatively quick and sensitive readout of methylation levels in a format that was amenable to processing a large number of samples. Using this approach, nearly 10,000 small molecules were evaluated for their ability to inhibit G9a activity at 100µM. From this screen an in vitro inhibitor of G9a was identified with an IC50 of approximately 3-5µM. Furthermore, this assay was utilized to functionally validate an interaction of MDM2 with an H3K9- specifiic methyltransferase activity (268).

While small molecule inhibitors of histone deacetylases and DNA methyltransferases have made their way into clinical practice, the development of histone methyltransferase inhibitors has proceeded at a slower pace, making this an attractive area to conduct screening for novel small molecule inhibitors. When this project began the only known selective inhibitor of histone methylation was the natural product chaetocin. This fungal metabolite selectively inhibited drosophila SU(VAR)3-9 with an

140

IC50 of 0.6µM, but exhibited extremely high toxicity in mammalian cell culture

experiments (260). This molecule has a complex chemical structure and has only

recently been synthesized (269).

During the course of this work, a series of several small molecule inhibitors of

G9a was reported IC50 values < 10µM (270). BIX-0194 was the most potent molecule

and was reported to decrease levels of H3K9me2 at several G9a target genes, although

expression levels were not modulated for all target genes analyzed. BIX-0194 has been

shown to abrogate some G9a-dependent cellular processes (267), but controversy exists

regarding BIX specificity in vivo. Furthermore, structural and in vitro kinetic analysis

indicates that G9a-like protein GLP likely is inhibited to a greater extent than G9a. Other

groups have also reported that BIX-0194 inhibition is markedly diminished when total histones are used as substrates in place of H3 tails, further fueling the controversy (158).

Additional structural analogs of BIX-0194 with improved activity and in vitro-

determined IC50 values in the low nM range have recently been reported, but have not

been tested in cells (269). In the experimental conditions described here, YL1-148-1

inhibits G9a more potently in vitro. However, cellular-based assays failed to demonstrate specific inhibition of G9a using a variety of approaches for either compound.

There are several potential explanations as to why YL1-148-1 failed to mediate detectable changes in H3K9me2 levels and G9a-dependent functions within cells. First of all, the possibility exists that the compound either did not enter cells or was quickly metabolized. Both appear to be unlikely because treatment at doses of 50µM were sufficient to alter cellular morphology and growth rate. However, this defect was not overcome by ectopic expression of G9a in MTT assays, suggesting that the effects were

141

not directly attributable to inhibition of G9a. It is also possible that the in vivo interaction of G9a-GLP heterodimers alters the ability of YL1-148-1 to abrogate enzymatic activity.

G9a immunoprecipitated from cells is inhibited in vitro, however, indicating that co- precipitating proteins do not abrogate YL1-148-1 activity.

It is quite likely that YL1-141 acts as a substrate competitor and that the extreme abundance of histone molecules within cells significantly dilutes activity of the small molecule. Assuming two copies of histone H3 per 200bp of DNA based on the known structure of the nucleosome unit, each human cell will contain approximately 3 x 107 histone H3 molecules. This number will be even higher in transformed cell lines such as

U2OS, which display complex karyotypes and multi-nucleation. There is some evidence for this competition effect. In silico docking indicated that the compound likely interacts with the portion of the enzyme involved with substrate recognition. Furthermore, competition experiments with increasing concentrations of peptide, but not SAM, abrogate YL1-148-1 activity.

Although YL1-148-1 in its current formulation is not effective in eliciting changes in G9a-dependent processes within cells, the identification of this small molecule has several potential applications. It may be useful as a chemical probe, by enabling in vitro inhibition of G9a in a variety of experimental settings (271). For example, blocking

G9a enzymatic activity in nuclear extracts may be beneficial in deciphering the role of

G9a in DNA methylation and for examining the functional interdependence of other histone modifications. It may also be possible to further modify this scaffold, which clearly has activity against G9a. Modifications to improve bioavailability, stability or affinity to G9a may result in more potent compounds. Since the completion of these

142

studies, the crystal structures of both G9a and GLP have been deposited in the Protein

Data Bank, which will facilitate structure-guided modification strategies in the future.

The approach taken here could also be applied to future screens. These efforts may focus on larger compound libraries or structure-guided libraries in order to identify not only inhibitors, but also small molecules capable of altering substrate and/or product specificity for a variety of histone modifying enzymes.

143

CHAPTER VII

DISCUSSION AND SCIENTIFIC SIGNIFICANCE

The studies presented in this dissertation have led to the discovery of novel

functions for PRDM1 in human NK cells and described the identification of an in vitro

inhibitor of G9a. PRDM1 is a zinc finger transcriptional repressor that mediates its

repressive effects via the recruitment of G9a methyltransferase and other histone

modifying enzymes. PRDM1 is highly conserved throughout metazoan species and has

evolved to carry out a wide array of functions in embryonic development, lineage

commitment and effector regulation of differentiated cells. Its critical ability to both

restrain cell growth and limit effector function is highlighted by its role as a tumor

suppressor. Indeed, it is the founding member of a large family of proteins characterized

by the presence of a PR domain, many of which have been implicated in differentiation

events and tumor suppression.

PRDM1 is induced coordinately with NK activation. Knockdown experiments

performed at 48h time points demonstrate that PRDM1 mediates a post-induction repression of IFNG, TNF and LTA, but cytotoxicity against NK-sensitive targets remains

unaffected. Consistent with this, ChIP data confirm binding to regulatory loci and over-

expression suppresses activation of IFNG and TNF (272). Recent data published after

completion of this work largely confirm the findings presented here. In murine NK cells,

144

combinations of IL-12 and IL-18 induce only a single isoform of PRDM1 and mRNA

expression was significantly higher in CD56dim from human samples (236). The authors demonstrate that in vivo tumor clearance is enhanced in NK cells derived from lethally irradiated Rag2-/- mice reconstituted with fetal liver cells derived from Prdm1-deficient embryos and that this effect was due to enhanced proliferation of Prdm1-deficient NK cells. However, no negative regulation of Ifng was observed. This may be explained in part due to technical variations in experimental set-up (different stimulating cytokine cocktails, different culture times, etc.) or species-specific regulation differences between murine and human NK cells. Compensatory mechanisms may also be in play as total absence of Prdm1 in the murine system is achievable, while human NK cells require stimulation to induce PRDM1 followed by subsequent siRNA-mediated knockdown.

PRDM1 exerts both overlapping and non-overlapping functions in NK cells and other immune lineages. Significant similarities exist between PRDM1’s role in NK cells and T cells. Regulation of IFNG appears to be conserved as PRDM1 binding was detected at the distal CNS-22 site, previously identified in murine CD4+ T cells (66).

Binding of PRDM1 to the promoter of IL2 is detected in NK cells as in T cells, but this

cytokine is not expressed by NK cells. In Prdm1-deficient CD8+ T cells, higher levels of

IFNγ, IL-2 and TNFα are produced in response to influenza challenge, consistent with

PRDM1’s role as a negative regulator of effector cytokines (71). However, in CD8+ T

cells, Prdm1-deficient cells have a defect in cytotoxicity due to diminished expression of

cytolytic molecules Gzmb, Gzmk and Prf1, which is not observed in NK cells. The

ability of PRDM1 to restrict memory potential in NK cells may also be conserved.

Recent data has implicated NK cells in the maintenance of immunological memory using

145

both hypersensitivity and viral infection models (209, 273). In response to hapten-

mediated contact hypersensitivity, hepatic NK cells exhibit an increased “memory-like”

phenotype relative to splenic NK cells (274). It would be feasible to test a role whether

these hepatic NK cell “memory” populations express lower levels of Prdm1 and/or have

altered patterns of PRDM1 induction in response to hapten exposure. However, it is

currently still unclear if human NK cells exhibit this memory potential as all studies to

date have been performed exclusively in murine systems (274).

PRDM1 was identified as one of several transcription factors up-regulated in NK cells in response to IL-2, IL-12 and IL-18. It will be of interest to determine how

PRDM1 interacts with other immune-regulatory transcription factors such as EOMES,

IRF8 and BATF which are also strongly induced with stimulation. Conversely, transcripts encoding a number of transcription factors are down-regulated with stimulation, including: RUNX3, GATA3, GILZ and PLZF. RUNX3 and GATA3 are well- known for their roles in CD4+ T cell differentiation, while PLZF has recently emerged as

the master regulator of NKT effector function (275, 276). GILZ is a glucocorticoid-

responsive transcription factor that exhibits dual functionality; it negatively regulates T

cell proliferation, but enhances NK proliferation in response to IL-15 and hydrocortisone

treatment (277). Virtually nothing is known about the expression and function of these

factors within the NK lineage. Furthermore, published data (236) and microarray data

presented here suggest that the transcriptional network in activated NK cells is markedly

different than the well-defined transcriptional cascade that accompanies plasma cell

differentiation. It is likely that these transcriptional changes represent a stimulation-

146

specific activation signature and cooperate to shape the NK effector phenotype. Furhter

characterization of these factors may provide important insight into NK function.

In NK cells, PRDM1 is induced in response to activating stimuli and its levels at

early time points correlate with mRNA levels of effector cytokines IFNG and TNF.

PRDM1 mRNA is readily detectable in freshly isolated NK cells and increases upon

stimulation with IL-2, IL-12 and IL-18. PRDM1 protein, however, is expressed only

very weakly in freshly isolated cells, but is markedly up-regulated with stimulation. The requirements for induction appear to be relatively specific because CD16 ligation or treatment with recombinant cytokines (IFNγ, IFNα or TNFα) fails to induce PRDM1.

Interestingly, when new protein synthesis is inhibited via cycloheximide concomitantly

with stimulation, PRDM1 mRNA levels increase dramatically, analogous to the well-

established phenomenon of “super-induction” observed for IFNγ (278). This effect is

potentially mediated by a negative regulator that is itself labile and subject to constant

turnover. When de novo protein synthesis is blocked, the factor is rapidly degraded

resulting in a diminished repression of PRDM1. It is currently unclear whether this factor

actively represses transcription of PRDM1 or acts at a post-transcriptional level to control

message stability. The 3’ untranslated region (UTR) of PRDM1 contains numerous AU-

rich elements and has a half life of less than one hour, one of the shortest of all known

genes (279). This suggests that post-transcriptional mechanisms are critical for PRDM1

protein expression.

NK cells are unique in the expression of multiple PRDM1 isoforms. To date, NK

cells are the first example of a non-transformed cell that expresses high levels of

PRDM1β. Also, an intermediate isoform PRDM1αΔ6 is consistently detected at similar

147

levels. Expression levels of these two lower molecular weight isoforms are always equal

to or greater than the full-length PRDM1β. Surprisingly, protein species corresponding

to PRDM1βΔ6 are not detected, despite the fact that, if present, they should be observed with the antibody used in immunobloting experiments. Previously, PRDM1β has only been detected in myeloma cell lines and samples obtained from myeloma and lymphoma patients. Although lacking an intact N-terminal PR domain, PRDM1β still localizes to the nucleus and can mediate transcriptional repression (albeit this activity is diminished relative to that of PRDM1α) (5). PRDM1αΔ6, however, likely has an altered capacity for

DNA-binding due to disruptions in the zinc finger domain. It is possible that this isoform either does not interact with DNA or recognizes distinct sequence motifs.

Collectively, the two lower molecular weight isoforms may exhibit endogenous

“dominant negative” functionality. B cells express predominantly PRDM1α as they

terminally differentiate into plasma cells and exit the cell cycle. NK cells, on the other

hand, induce multiple isoforms in response to stimulation. This may allow NK cells to

continue to proliferate, but at a reduced rate. Consistent with this, PRDM1 induction in

NK cells is not accompanied by an increase in apoptosis as is observed in mantle cell

lymphoma cells, which induce only full-length PRDM1α in response to Bortezomib

(195). Furthermore, PRDM1 expression in LAK cells increases during a two to three

week culture period and is characterized by an increased PRDM1α:PRDM1β ratio as

cells approach exhaustion. Similarly, PRDM1 accumulates in murine CD8+ T cells as

they become exhausted (73). In lymphoma cells, treatment with the chemotherapeutic

agent doxorubicin in combination with Rituximib leads to a decrease in PRDM1β levels,

which is correlated with their survival (280). Thus, an increased PRDM1α:PRDM1β

148

ratio is associated with decreased proliferation and apoptosis. NK cells have likely struck

an evolutionary balance in terms of PRDM1 expression, sacrificing some potency in the

ability to repress target genes, but avoiding full repression of proliferation.

These studies identify numerous novel targets of PRDM1-mediated regulation in

NK cells. In fact, relatively few ChIP-based experiments have been performed in NK cells to date and there are no published ChIP-chip or ChIP-seq data sets for any transcription factors in NK cells. The target gene list identified shares some limited overlap with known targets of PRDM1 in other lineages, but is substantially divergent overall. In these studies, a criterion was applied whereby an annotated gene had to appear in two different analyses to be considered a true target gene. This relatively arbitrary approach represents a “best-fit” approximation to describe the NK-specific

“PRDM1 target interactome”. Indeed, genes such as PCNA and CIITApIV are consistently strongly positive in qPCR-based ChIP, but were not identified by ChIP-chip.

Technical differences between detection by PCR-based amplification versus hybridization of post-amplified material surely accounts for some of this divergence.

Furthermore, many genes were identified in only one experiment, while a small group was identified in five or more analyses, suggesting that some targets may be constitutively-bound, while others are more transient in nature.

PRDM1’s function in NK cells is mediated via the binding of numerous target genes. PRDM1 affects expression of genes that are constitutively expressed in resting

NK cells and also functions as a post-induction suppressor of genes that are induced upon stimulation. By regulating the expression of a large number of target genes, PRDM1

shapes the effector phenotype, functioning as a rheostat to control the degree of NK

149

activation. For example, regulation of ZAP70, KLRD1, STAT6 and others allows for a diminished response to certain activating stimuli, while leaving other signaling pathways largely intact. Other target genes such as LRMP and BCL2L11 may facilitate receptor recycling and protection from stimulation-induced apoptosis, respectively. Importantly,

PRDM1 binding is not limited to proximal regions of the promoter as binding was not enriched at or near the TSS of genes. PRDM1 also mediates its control over effector function via indirect mechanisms as transcript levels for a large number of genes that are not direct targets of PRDM1 binding are altered upon knockdown of PRDM1.

In order to repress its target genes PRDM1 recruits several co-repressors and histone modifying enzymes which mediate changes in the chromatin structure. PRDM1 mediates epigenetic silencing via increases in di-methylation of H3K9 (H3K9me2) and decreases in histone acetylation at target promoters. Thus, the use of epigenetic modifying agents has the potential to modulate transcription of specific genes based upon

selective recruitment of histone modifying enzymes to target genes. In NK cells, the use of HDAC inhibitors leads to increased lysis of a variety of target cells via up-regulation of NKG2D ligands, MICA and MICB (281-284). Yet, relatively few data exist concerning the NK-intrinsic effects of epigenetic modifying agents and little is known about the connections between chromatin structure and effector functions. HDAC inhibitors have been reported to interfere with IL-2 mediated activation, but effects on cytolytic activity are less clear (285, 286). Inhibition of histone methyltransferase function may have the potential to mediate similar functional alterations. Presumably,

NK differentiation requires significant alterations in chromatin structure. However, a paucity of NK-specific knockout targeting strategies and the relatively low number of NK

150

cells per animal make it exceedingly difficult to experimentally address questions of this nature.

PRDM1’s role as a tumor suppressor is well-established within lymphomas. In

DLBCL numerous mutations have been identified and deletion of 6q21 occurs with relatively high frequency. Recent in vitro and in vivo studies verify mechanistically that

abrogation of PRDM1 results in increased incidence of several hematopoietic

malignancies (10, 59). A role for PRDM1 as a tumor suppressor in NK malignancies is

suggested by the high frequency of a minimal deletion region on 6q21 that includes

PRDM1. Interestingly, lymphomas originating from the NK lineage exhibit a distinct

geographical distribution, with markedly higher incidences observed in Asia and South

America. A secondary genetic locus or environmental interaction may partially explain

the wide discrepancies in incidence of these highly aggressive and often fatal neoplasms.

Deletions and silencing of PRDM1 have been documented in these patients (62).

However, data linking disruption of PRDM1 to NK-specific functional defects is

currently lacking, due in no small part to the difficulty involved in obtaining primary

patient samples. Likewise, the contribution of microRNAs to NK lymphomagenesis

remains unexplored. It will be of interest to assess the expression of miR-155 (as well as

other potential PRDM1 targets) in this process.

Collectively, the data within this dissertation establish PRDM1 as a negative

regulator of NK effector function. PRDM1 is induced during cytokine-mediated activation and acts as a post-induction attenuator of effector cytokine production, but does not impede cytotoxic activity. Through modulating the expression of a wide variety of target genes PRDM1 shapes the phenotype of human NK cells (Figure 61). PRDM1

151

is a highly conserved transcription factor and orthologs have been identified in numerous metazoan phyla, indicating an ancient evolutionary history. Evolution of PRDM1’s functionality likely predates the emergence of the adaptive immune system, which occurred roughly 500 million years ago. Consistent with this notion, these studies position PRDM1 as a common regulator of both the adaptive and innate immune response.

Figure 61: Model of PRDM1 function in NK cells. Schematic demonstrating functionality of PRDM1 in response to cytokine stimulation and LAK culture.

152

REFERENCES

1. Keller, A. D., and T. Maniatis. 1991. Identification and characterization of a novel repressor of beta-interferon gene expression. Genes Dev 5:868-879.

2. Turner, C. A., Jr., D. H. Mack, and M. M. Davis. 1994. Blimp-1, a novel zinc finger-containing protein that can drive the maturation of B lymphocytes into immunoglobulin-secreting cells. Cell 77:297-306.

3. Huang, S. 1994. Blimp-1 is the murine homolog of the human transcriptional repressor PRDI-BF1. Cell 78:9.

4. Tunyaplin, C., M. A. Shapiro, and K. L. Calame. 2000. Characterization of the B lymphocyte-induced maturation protein-1 (Blimp-1) gene, mRNA isoforms and basal promoter. Nucl. Acids Res. 28:4846-4855.

5. Gyory, I., G. Fejer, N. Ghosh, E. Seto, and K. L. Wright. 2003. Identification of a functionally impaired positive regulatory domain I binding factor 1 transcription repressor in myeloma cell lines. J Immunol 170:3125-3133.

6. Schmidt, D., A. Nayak, J. E. Schumann, A. Schimpl, I. Berberich, and F. Berberich-Siebelt. 2008. Blimp-1[Delta]exon7: A naturally occurring Blimp-1 deletion mutant with auto-regulatory potential. Experimental Cell Research 314:3614.

7. Huang, S., G. Shao, and L. Liu. 1998. The PR Domain of the Rb-binding Zinc Finger Protein RIZ1 Is a Protein Binding Interface and Is Related to the SET Domain Functioning in Chromatin-mediated Gene Expression. J. Biol. Chem. 273:15933-15939.

8. Schneider, R., A. J. Bannister, and T. Kouzarides. 2002. Unsafe SETs: histone lysine methyltransferases and cancer. Trends Biochem Sci 27:396-402.

9. Wu, H., J. Min, V. V. Lunin, T. Antoshenko, L. Dombrovski, H. Zeng, A. Allali- Hassani, V. Campagna-Slater, M. Vedadi, C. H. Arrowsmith, A. N. Plotnikov, and M. Schapira. 2010. Structural biology of human H3K9 methyltransferases. PLoS One 5:e8570.

153

10. Mandelbaum, J., G. Bhagat, H. Tang, T. Mo, M. Brahmachary, Q. Shen, A. Chadburn, K. Rajewsky, A. Tarakhovsky, L. Pasqualucci, and R. Dalla-Favera. 2010. BLIMP1 Is a Tumor Suppressor Gene Frequently Disrupted in Activated B Cell-like Diffuse Large B Cell Lymphoma. Cancer Cell 18:568-579.

11. Freund, C., H. G. Schmalz, J. Sticht, and R. Kühne. 2008. Proline-Rich Sequence Recognition Domains (PRD): Ligands, Function and Inhibition. In Protein- Protein Interactions as New Drug Targets. E. Klussmann, and J. Scott, eds. Springer Berlin Heidelberg. 407-429.

12. Bikoff, E. K., M. A. Morgan, and E. J. Robertson. 2009. An expanding job description for Blimp-1/PRDM1. Curr Opin Genet Dev 19:379-385.

13. Kuo, T. C., and K. L. Calame. 2004. B Lymphocyte-Induced Maturation Protein (Blimp)-1, IFN Regulatory Factor (IRF)-1, and IRF-2 Can Bind to the Same Regulatory Sites. J Immunol 173:5556-5563.

14. Keller, A. D., and T. Maniatis. 1992. Only two of the five zinc fingers of the eukaryotic transcriptional repressor PRDI-BF1 are required for sequence-specific DNA binding. Mol Cell Biol 12:1940-1949.

15. Gyory, I., J. Wu, G. Fejer, E. Seto, and K. L. Wright. 2004. PRDI-BF1 recruits the histone H3 methyltransferase G9a in transcriptional silencing. Nat Immunol 5:299-308.

16. Yu, J., C. Angelin-Duclos, J. Greenwood, J. Liao, and K. Calame. 2000. Transcriptional repression by blimp-1 (PRDI-BF1) involves recruitment of histone deacetylase. Mol Cell Biol 20:2592-2603.

17. Ancelin, K., U. C. Lange, P. Hajkova, R. Schneider, A. J. Bannister, T. Kouzarides, and M. A. Surani. 2006. Blimp1 associates with Prmt5 and directs histone arginine methylation in mouse germ cells. Nat Cell Biol 8:623-630.

18. Eckert, D., K. Biermann, D. Nettersheim, A. J. Gillis, K. Steger, H. M. Jack, A. M. Muller, L. H. Looijenga, and H. Schorle. 2008. Expression of BLIMP1/PRMT5 and concurrent histone H2A/H4 arginine 3 dimethylation in fetal germ cells, CIS/IGCNU and germ cell tumors. BMC Dev Biol 8:106.

154

19. Su, S. T., H. Y. Ying, Y. K. Chiu, F. R. Lin, M. Y. Chen, and K. I. Lin. 2009. Involvement of Histone Demethylase LSD-1 in Blimp-1-Mediated Gene Repression during Plasma Cell Differentiation. Mol Cell Biol.

20. Ren, B., K. J. Chee, T. H. Kim, and T. Maniatis. 1999. PRDI-BF1/Blimp-1 repression is mediated by corepressors of the Groucho family of proteins. Genes Dev. 13:125-137.

21. Xie, M., G. Shao, I. M. Buyse, and S. Huang. 1997. Transcriptional repression mediated by the PR domain zinc finger gene RIZ. J Biol Chem 272:26360-26366.

22. Buyse, I. M., G. Shao, and S. Huang. 1995. The retinoblastoma protein binds to RIZ, a zinc-finger protein that shares an epitope with the adenovirus E1A protein. Proc Natl Acad Sci U S A 92:4467-4471.

23. Tokumaru, Y., S. Nomoto, C. Jeronimo, R. Henrique, S. Harden, B. Trink, and D. Sidransky. 2003. Biallelic inactivation of the RIZ1 gene in human gastric cancer. Oncogene 22:6954-6958.

24. Jiang, G. L., and S. Huang. 2000. The yin-yang of PR-domain family genes in tumorigenesis. Histol Histopathol 15:109-117.

25. Fumasoni, I., N. Meani, D. Rambaldi, G. Scafetta, M. Alcalay, and F. D. Ciccarelli. 2007. Family expansion and gene rearrangements contributed to the functional specialization of PRDM genes in vertebrates. BMC Evol Biol 7:187.

26. Seale, P., S. Kajimura, W. Yang, S. Chin, L. M. Rohas, M. Uldry, G. Tavernier, D. Langin, and B. M. Spiegelman. 2007. Transcriptional control of brown fat determination by PRDM16. Cell Metab 6:38-54.

27. Seale, P., B. Bjork, W. Yang, S. Kajimura, S. Chin, S. Kuang, A. Scime, S. Devarakonda, H. M. Conroe, H. Erdjument-Bromage, P. Tempst, M. A. Rudnicki, D. R. Beier, and B. M. Spiegelman. 2008. PRDM16 controls a brown fat/skeletal muscle switch. Nature 454:961-967.

28. Yamaji, M., Y. Seki, K. Kurimoto, Y. Yabuta, M. Yuasa, M. Shigeta, K. Yamanaka, Y. Ohinata, and M. Saitou. 2008. Critical function of Prdm14 for the establishment of the germ cell lineage in mice. Nat Genet 40:1016-1022.

155

29. Berg, I. L., R. Neumann, K. W. Lam, S. Sarbajna, L. Odenthal-Hesse, C. A. May, and A. J. Jeffreys. 2010. PRDM9 variation strongly influences recombination hotspot activity and meiotic instability in humans. Nat Genet.

30. Oliver, P. L., L. Goodstadt, J. J. Bayes, Z. Birtle, K. C. Roach, N. Phadnis, S. A. Beatson, G. Lunter, H. S. Malik, and C. P. Ponting. 2009. Accelerated evolution of the Prdm9 speciation gene across diverse metazoan taxa. PLoS Genet 5:e1000753.

31. Baudat, F., J. Buard, C. Grey, A. Fledel-Alon, C. Ober, M. Przeworski, G. Coop, and B. de Massy. 2010. PRDM9 is a major determinant of meiotic recombination hotspots in humans and mice. Science 327:836-840.

32. Myers, S., R. Bowden, A. Tumian, R. E. Bontrop, C. Freeman, T. S. MacFie, G. McVean, and P. Donnelly. 2010. Drive against hotspot motifs in primates implicates the PRDM9 gene in meiotic recombination. Science 327:876-879.

33. Parvanov, E. D., P. M. Petkov, and K. Paigen. 2010. Prdm9 controls activation of mammalian recombination hotspots. Science 327:835.

34. Shapiro-Shelef, M., K. I. Lin, L. J. McHeyzer-Williams, J. Liao, M. G. McHeyzer-Williams, and K. Calame. 2003. Blimp-1 is required for the formation of immunoglobulin secreting plasma cells and pre-plasma memory B cells. Immunity 19:607-620.

35. Chang, D. H., G. Cattoretti, and K. L. Calame. 2002. The dynamic expression pattern of B lymphocyte induced maturation protein-1 (Blimp-1) during mouse embryonic development. Mech Dev 117:305-309.

36. Roy, S., and T. Ng. 2004. Blimp-1 Specifies Neural Crest and Sensory Neuron Progenitors in the Zebrafish Embryo. Current Biology 14:1772.

37. Lee, B. C., and S. Roy. 2006. Blimp-1 is an essential component of the genetic program controlling development of the pectoral limb bud. Dev Biol 300:623-634.

38. Robertson, E. J., I. Charatsi, C. J. Joyner, C. H. Koonce, M. Morgan, A. Islam, C. Paterson, E. Lejsek, S. J. Arnold, A. Kallies, S. L. Nutt, and E. K. Bikoff. 2007. Blimp1 regulates development of the posterior forelimb, caudal pharyngeal arches, heart and sensory vibrissae in mice. Development 134:4335-4345.

156

39. Nishikawa, K., T. Nakashima, M. Hayashi, T. Fukunaga, S. Kato, T. Kodama, S. Takahashi, K. Calame, and H. Takayanagi. 2010. Blimp1-mediated repression of negative regulators is required for osteoclast differentiation. Proc Natl Acad Sci U S A 107:3117-3122.

40. Miyauchi, Y., K. Ninomiya, H. Miyamoto, A. Sakamoto, R. Iwasaki, H. Hoshi, K. Miyamoto, W. Hao, S. Yoshida, H. Morioka, K. Chiba, S. Kato, T. Tokuhisa, M. Saitou, Y. Toyama, T. Suda, and T. Miyamoto. 2010. The Blimp1-Bcl6 axis is critical to regulate osteoclast differentiation and bone homeostasis. J Exp Med 207:751-762.

41. Magnusdottir, E., S. Kalachikov, K. Mizukoshi, D. Savitsky, A. Ishida- Yamamoto, A. A. Panteleyev, and K. Calame. 2007. Epidermal terminal differentiation depends on B lymphocyte-induced maturation protein-1. Proc Natl Acad Sci U S A 104:14988-14993.

42. Lin, Y., K.-k. Wong, and K. Calame. 1997. Repression of c-myc Transcription by Blimp-1, an Inducer of Terminal B Cell Differentiation. Science 276:596-599.

43. Messika, E. J., P. S. Lu, Y.-J. Sung, T. Yao, J.-T. Chi, Y.-h. Chien, and M. M. Davis. 1998. Differential Effect of B Lymphocyte-induced Maturation Protein (Blimp-1) Expression on Cell Fate during B Cell Development. J. Exp. Med. 188:515-525.

44. Piskurich, J. F., K. I. Lin, Y. Lin, Y. Wang, J. P. Ting, and K. Calame. 2000. BLIMP-I mediates extinction of major histocompatibility class II transactivator expression in plasma cells. Nat Immunol 1:526-532.

45. Lin, K. I., C. Angelin-Duclos, T. C. Kuo, and K. Calame. 2002. Blimp-1- dependent repression of Pax-5 is required for differentiation of B cells to immunoglobulin M-secreting plasma cells. Mol Cell Biol 22:4771-4780.

46. Shaffer, A. L., K. I. Lin, T. C. Kuo, X. Yu, E. M. Hurt, A. Rosenwald, J. M. Giltnane, L. Yang, H. Zhao, K. Calame, and L. M. Staudt. 2002. Blimp-1 orchestrates plasma cell differentiation by extinguishing the mature B cell gene expression program. Immunity 17:51-62.

47. Sciammas, R., and M. M. Davis. 2004. Modular Nature of Blimp-1 in the Regulation of Gene Expression during B Cell Maturation. J Immunol 172:5427- 5440.

157

48. Shaffer, A. L., M. Shapiro-Shelef, N. N. Iwakoshi, A. H. Lee, S. B. Qian, H. Zhao, X. Yu, L. Yang, B. K. Tan, A. Rosenwald, E. M. Hurt, E. Petroulakis, N. Sonenberg, J. W. Yewdell, K. Calame, L. H. Glimcher, and L. M. Staudt. 2004. XBP1, downstream of Blimp-1, expands the secretory apparatus and other organelles, and increases protein synthesis in plasma cell differentiation. Immunity 21:81-93.

49. Klein, U., S. Casola, G. Cattoretti, Q. Shen, M. Lia, T. Mo, T. Ludwig, K. Rajewsky, and R. Dalla-Favera. 2006. Transcription factor IRF4 controls plasma cell differentiation and class-switch recombination. Nat Immunol 7:773-782.

50. Nera, K. P., P. Kohonen, E. Narvi, A. Peippo, L. Mustonen, P. Terho, K. Koskela, J. M. Buerstedde, and O. Lassila. 2006. Loss of Pax5 promotes plasma cell differentiation. Immunity 24:283-293.

51. Kallies, A., J. Hasbold, K. Fairfax, C. Pridans, D. Emslie, B. S. McKenzie, A. M. Lew, L. M. Corcoran, P. D. Hodgkin, D. M. Tarlinton, and S. L. Nutt. 2007. Initiation of Plasma-Cell Differentiation Is Independent of the Transcription Factor Blimp-1. Immunity 26:555.

52. Shapiro-Shelef, M., K.-I. Lin, D. Savitsky, J. Liao, and K. Calame. 2005. Blimp-1 is required for maintenance of long-lived plasma cells in the bone marrow. J. Exp. Med. 202:1471-1476.

53. Lin, F. R., H. K. Kuo, H. Y. Ying, F. H. Yang, and K. I. Lin. 2007. Induction of apoptosis in plasma cells by B lymphocyte-induced maturation protein-1 knockdown. Cancer Res 67:11914-11923.

54. Gaidano, G., R. Hauptschein, N. Parsa, K. Offit, P. Rao, G. Lenoir, D. Knowles, R. Chaganti, and R. Dalla-Favera. 1992. Deletions involving two distinct regions of 6q in B-cell non-Hodgkin lymphoma. Blood 80:1781-1787.

55. Thelander, E. F., K. Ichimura, M. Corcoran, G. Barbany, A. Nordgren, M. Heyman, M. Berglund, A. Mungall, R. Rosenquist, V. P. Collins, D. Grander, C. Larsson, and S. Lagercrantz. 2008. Characterization of 6q deletions in mature B cell lymphomas and childhood acute lymphoblastic leukemia. Leuk Lymphoma 49:477-487.

56. Pasqualucci, L., M. Compagno, J. Houldsworth, S. Monti, A. Grunn, S. V. Nandula, J. C. Aster, V. V. Murty, M. A. Shipp, and R. Dalla-Favera. 2006.

158

Inactivation of the PRDM1/BLIMP1 gene in diffuse large B cell lymphoma. J Exp Med 203:311-317.

57. Tam, W., M. Gomez, A. Chadburn, J. W. Lee, W. C. Chan, and D. M. Knowles. 2006. Mutational analysis of PRDM1 indicates a tumor-suppressor role in diffuse large B-cell lymphomas. Blood 107:4090-4100.

58. Tate, G., Y. Hirayama-Ohashi, K. Kishimoto, and T. Mitsuya. 2007. Novel BLIMP1/PRDM1 gene mutations in B-cell lymphoma. Cancer Genet Cytogenet 172:151-153.

59. Calado, D. P., B. Zhang, L. Srinivasan, Y. Sasaki, J. Seagal, C. Unitt, S. Rodig, J. Kutok, A. Tarakhovsky, M. Schmidt-Supprian, and K. Rajewsky. 2010. Constitutive canonical NF-kappaB activation cooperates with disruption of BLIMP1 in the pathogenesis of activated B cell-like diffuse large cell lymphoma. Cancer Cell 18:580-589.

60. Kwong, Y. L. 2005. Natural killer-cell malignancies: diagnosis and treatment. Leukemia 19:2186-2194.

61. Yoon, J., and Y. H. Ko. 2003. Deletion mapping of the long arm of chromosome 6 in peripheral T and NK cell lymphomas. Leuk Lymphoma 44:2077-2082.

62. Iqbal, J., C. Kucuk, R. J. Deleeuw, G. Srivastava, W. Tam, H. Geng, D. Klinkebiel, J. K. Christman, K. Patel, K. Cao, L. Shen, K. Dybkaer, I. F. Tsui, H. Ali, N. Shimizu, W. Y. Au, W. L. Lam, and W. C. Chan. 2009. Genomic analyses reveal global functional alterations that promote tumor growth and novel tumor suppressor genes in natural killer-cell malignancies. Leukemia 23:1139-1151.

63. Wang, X., K. Belguise, C. F. O'Neill, N. Sanchez-Morgan, M. Romagnoli, S. F. Eddy, N. D. Mineva, Z. Yu, C. Min, V. Trinkaus-Randall, D. Chalbos, and G. E. Sonenshein. 2009. RelB NF-kappaB represses estrogen receptor alpha expression via induction of the zinc finger protein Blimp1. Mol Cell Biol 29:3832-3844.

64. Martins, G. A., L. Cimmino, M. Shapiro-Shelef, M. Szabolcs, A. Herron, E. Magnusdottir, and K. Calame. 2006. Transcriptional repressor Blimp-1 regulates T cell homeostasis and function. Nat Immunol 7:457-465.

159

65. Kallies, A., E. D. Hawkins, G. T. Belz, D. Metcalf, M. Hommel, L. M. Corcoran, P. D. Hodgkin, and S. L. Nutt. 2006. Transcriptional repressor Blimp-1 is essential for T cell homeostasis and self-tolerance. Nat Immunol 7:466-474.

66. Cimmino, L., G. A. Martins, J. Liao, E. Magnusdottir, G. Grunig, R. K. Perez, and K. L. Calame. 2008. Blimp-1 attenuates Th1 differentiation by repression of ifng, tbx21, and bcl6 gene expression. J Immunol 181:2338-2347.

67. Martins, G. A., L. Cimmino, J. Liao, E. Magnusdottir, and K. Calame. 2008. Blimp-1 directly represses Il2 and the Il2 activator Fos, attenuating T cell proliferation and survival. J Exp Med 205:1959-1965.

68. Gong, D., and T. R. Malek. 2007. Cytokine-Dependent Blimp-1 Expression in Activated T Cells Inhibits IL-2 Production. J Immunol 178:242-252.

69. Johnston, R. J., A. C. Poholek, D. DiToro, I. Yusuf, D. Eto, B. Barnett, A. L. Dent, J. Craft, and S. Crotty. 2009. Bcl6 and Blimp-1 Are Reciprocal and Antagonistic Regulators of T Follicular Helper Cell Differentiation. Science 325:1006-1010.

70. Cretney, E., A. Xin, W. Shi, M. Minnich, F. Masson, M. Miasari, G. T. Belz, G. K. Smyth, M. Busslinger, S. L. Nutt, and A. Kallies. 2011. The transcription factors Blimp-1 and IRF4 jointly control the differentiation and function of effector regulatory T cells. Nat Immunol 12:304-311.

71. Kallies, A., A. Xin, G. T. Belz, and S. L. Nutt. 2009. Blimp-1 Transcription Factor Is Required for the Differentiation of Effector CD8(+) T Cells and Memory Responses. Immunity.

72. Rutishauser, R. L., G. A. Martins, S. Kalachikov, A. Chandele, I. A. Parish, E. Meffre, J. Jacob, K. Calame, and S. M. Kaech. 2009. Transcriptional Repressor Blimp-1 Promotes CD8(+) T Cell Terminal Differentiation and Represses the Acquisition of Central Memory T Cell Properties. Immunity.

73. Shin, H., S. D. Blackburn, A. M. Intlekofer, C. Kao, J. M. Angelosanto, S. L. Reiner, and E. J. Wherry. 2009. A Role for the Transcriptional Repressor Blimp-1 in CD8(+) T Cell Exhaustion during Chronic Viral Infection. Immunity.

74. Chang, D. H., C. Angelin-Duclos, and K. Calame. 2000. BLIMP-1: trigger for differentiation of myeloid lineage. Nat Immunol 1:169-176.

160

75. Smith, M. A., G. Wright, J. Wu, P. Tailor, K. Ozato, X. Chen, S. Wei, J. F. Piskurich, J. P. Ting, and K. L. Wright. 2011. Positive Regulatory Domain I (PRDM1) and IRF8/PU.1 Counter-regulate MHC Class II Transactivator (CIITA) Expression during Dendritic Cell Maturation. J Biol Chem 286:7893-7904.

76. Chan, Y. H., M. F. Chiang, Y. C. Tsai, S. T. Su, M. H. Chen, M. S. Hou, and K. I. Lin. 2009. Absence of the transcriptional repressor Blimp-1 in hematopoietic lineages reveals its role in dendritic cell homeostatic development and function. J Immunol 183:7039-7046.

77. Kiessling, R., E. Klein, and H. Wigzell. 1975. "Natural" killer cells in the mouse. I. Cytotoxic cells with specificity for mouse Moloney leukemia cells. Specificity and distribution according to genotype. European Journal of Immunology 5:112.

78. Herberman, R. B., M. E. Nunn, and D. H. Lavrin. 1975. Natural cytotoxic reactivity of mouse lymphoid cells against syngeneic acid allogeneic tumors. I. Distribution of reactivity and specificity. Int J Cancer 16:216-229.

79. Seaman, W. E., T. D. Gindhart, J. S. Greenspan, M. A. Blackman, and N. Talal. 1979. Natural killer cells, bone, and the bone marrow: studies in estrogen-treated mice and in congenitally osteopetrotic (mi/mi) mice. J Immunol 122:2541-2547.

80. Kumar, V., J. Ben-Ezra, M. Bennett, and G. Sonnenfeld. 1979. Natural killer cells in mice treated with 89strontium: normal target-binding cell numbers but inability to kill even after interferon administration. J Immunol 123:1832-1838.

81. Hackett, J., Jr., M. Bennett, and V. Kumar. 1985. Origin and differentiation of natural killer cells. I. Characteristics of a transplantable NK cell precursor. J Immunol 134:3731-3738.

82. Hackett, J., Jr., M. Bennett, G. C. Koo, and V. Kumar. 1986. Origin and differentiation of natural killer cells. III. Relationship between the precursors and effectors of natural killer and natural cytotoxic activity. Immunol Res 5:16-24.

83. Roth, C., C. Rothlin, S. Riou, D. H. Raulet, and G. Lemke. 2007. Stromal-cell regulation of natural killer cell differentiation. J Mol Med 85:1047-1056.

84. Colucci, F., M. A. Caligiuri, and J. P. Di Santo. 2003. What does it take to make a natural killer? Nat Rev Immunol 3:413-425.

161

85. Igarashi, H., S. C. Gregory, T. Yokota, N. Sakaguchi, and P. W. Kincade. 2002. Transcription from the RAG1 locus marks the earliest lymphocyte progenitors in bone marrow. Immunity 17:117-130.

86. Adolfsson, J., R. Mansson, N. Buza-Vidas, A. Hultquist, K. Liuba, C. T. Jensen, D. Bryder, L. Yang, O. J. Borge, L. A. Thoren, K. Anderson, E. Sitnicka, Y. Sasaki, M. Sigvardsson, and S. E. Jacobsen. 2005. Identification of Flt3+ lympho- myeloid stem cells lacking erythro-megakaryocytic potential a revised road map for adult blood lineage commitment. Cell 121:295-306.

87. Mrozek, E., P. Anderson, and M. A. Caligiuri. 1996. Role of interleukin-15 in the development of human CD56+ natural killer cells from CD34+ hematopoietic progenitor cells. Blood 87:2632-2640.

88. Williams, N. S., T. A. Moore, J. D. Schatzle, I. J. Puzanov, P. V. Sivakumar, A. Zlotnik, M. Bennett, and V. Kumar. 1997. Generation of lytic natural killer 1.1+, Ly-49- cells from multipotential murine bone marrow progenitors in a stroma-free culture: definition of cytokine requirements and developmental intermediates. J Exp Med 186:1609-1614.

89. Lodolce, J. P., D. L. Boone, S. Chai, R. E. Swain, T. Dassopoulos, S. Trettin, and A. Ma. 1998. IL-15 receptor maintains lymphoid homeostasis by supporting lymphocyte homing and proliferation. Immunity 9:669-676.

90. Kennedy, M. K., M. Glaccum, S. N. Brown, E. A. Butz, J. L. Viney, M. Embers, N. Matsuki, K. Charrier, L. Sedger, C. R. Willis, K. Brasel, P. J. Morrissey, K. Stocking, J. C. Schuh, S. Joyce, and J. J. Peschon. 2000. Reversible defects in natural killer and memory CD8 T cell lineages in interleukin 15-deficient mice. J Exp Med 191:771-780.

91. Suzuki, H., G. S. Duncan, H. Takimoto, and T. W. Mak. 1997. Abnormal development of intestinal intraepithelial lymphocytes and peripheral natural killer cells in mice lacking the IL-2 receptor beta chain. J Exp Med 185:499-505.

92. Rosmaraki, E. E., I. Douagi, C. Roth, F. Colucci, A. Cumano, and J. P. Di Santo. 2001. Identification of committed NK cell progenitors in adult murine bone marrow. Eur J Immunol 31:1900-1909.

93. Barton, K., N. Muthusamy, C. Fischer, C. N. Ting, T. L. Walunas, L. L. Lanier, and J. M. Leiden. 1998. The Ets-1 transcription factor is required for the development of natural killer cells in mice. Immunity 9:555-563. 162

94. Colucci, F., S. I. Samson, R. P. DeKoter, O. Lantz, H. Singh, and J. P. Di Santo. 2001. Differential requirement for the transcription factor PU.1 in the generation of natural killer cells versus B and T cells. Blood 97:2625-2632.

95. Ikawa, T., S. Fujimoto, H. Kawamoto, Y. Katsura, and Y. Yokota. 2001. Commitment to natural killer cells requires the helix-loop-helix inhibitor Id2. Proc Natl Acad Sci U S A 98:5164-5169.

96. Yokota, Y., A. Mansouri, S. Mori, S. Sugawara, S. Adachi, S. Nishikawa, and P. Gruss. 1999. Development of peripheral lymphoid organs and natural killer cells depends on the helix-loop-helix inhibitor Id2. Nature 397:702-706.

97. Freud, A. G., B. Becknell, S. Roychowdhury, H. C. Mao, A. K. Ferketich, G. J. Nuovo, T. L. Hughes, T. B. Marburger, J. Sung, R. A. Baiocchi, M. Guimond, and M. A. Caligiuri. 2005. A human CD34(+) subset resides in lymph nodes and differentiates into CD56bright natural killer cells. Immunity 22:295-304.

98. Romagnani, C., K. Juelke, M. Falco, B. Morandi, A. D'Agostino, R. Costa, G. Ratto, G. Forte, P. Carrega, G. Lui, R. Conte, T. Strowig, A. Moretta, C. Munz, A. Thiel, L. Moretta, and G. Ferlazzo. 2007. CD56brightCD16- Killer Ig-Like Receptor- NK Cells Display Longer Telomeres and Acquire Features of CD56dim NK Cells upon Activation. J Immunol 178:4947-4955.

99. Cooper, M. A., T. A. Fehniger, and M. A. Caligiuri. 2001. The biology of human natural killer-cell subsets. Trends Immunol 22:633-640.

100. Hanna, J., P. Bechtel, Y. Zhai, F. Youssef, K. McLachlan, and O. Mandelboim. 2004. Novel Insights on Human NK Cells' Immunological Modalities Revealed by Gene Expression Profiling. J Immunol 173:6547-6563.

101. Wendt, K., E. Wilk, S. Buyny, J. Buer, R. E. Schmidt, and R. Jacobs. 2006. Gene and protein characteristics reflect functional diversity of CD56dim and CD56bright NK cells. J Leukoc Biol:1529-1541.

102. Caligiuri, M. A., A. Zmuidzinas, T. J. Manley, H. Levine, K. A. Smith, and J. Ritz. 1990. Functional consequences of interleukin 2 receptor expression on resting human lymphocytes. Identification of a novel natural killer cell subset with high affinity receptors. J Exp Med 171:1509-1526.

163

103. Frey, M., N. B. Packianathan, T. A. Fehniger, M. E. Ross, W. C. Wang, C. C. Stewart, M. A. Caligiuri, and S. S. Evans. 1998. Differential expression and function of L-selectin on CD56bright and CD56dim natural killer cell subsets. J Immunol 161:400-408.

104. Koopman, L. A., H. D. Kopcow, B. Rybalov, J. E. Boyson, J. S. Orange, F. Schatz, R. Masch, C. J. Lockwood, A. D. Schachter, P. J. Park, and J. L. Strominger. 2003. Human Decidual Natural Killer Cells Are a Unique NK Cell Subset with Immunomodulatory Potential. J. Exp. Med. 198:1201-1212.

105. Beziat, V., B. Descours, C. Parizot, P. Debre, and V. Vieillard. 2010. NK cell terminal differentiation: correlated stepwise decrease of NKG2A and acquisition of KIRs. PLoS One 5:e11966.

106. Yu, J., H. C. Mao, M. Wei, T. Hughes, J. Zhang, I. K. Park, S. Liu, S. McClory, G. Marcucci, R. Trotta, and M. A. Caligiuri. 2010. CD94 surface density identifies a functional intermediary between the CD56bright and CD56dim human NK-cell subsets. Blood 115:274-281.

107. Hayakawa, Y., N. D. Huntington, S. L. Nutt, and M. J. Smyth. 2006. Functional subsets of mouse natural killer cells. Immunol Rev 214:47-55.

108. Hayakawa, Y., and M. J. Smyth. 2006. CD27 dissects mature NK cells into two subsets with distinct responsiveness and migratory capacity. J Immunol 176:1517- 1524.

109. Colonna, M., and J. Samaridis. 1995. Cloning of immunoglobulin-superfamily members associated with HLA-C and HLA-B recognition by human natural killer cells. Science 268:405-408.

110. Lazetic, S., C. Chang, J. Houchins, L. Lanier, and J. Phillips. 1996. Human natural killer cell receptors involved in MHC class I recognition are disulfide- linked heterodimers of CD94 and NKG2 subunits. The Journal of Immunology 157:4741-4745.

111. Wu, J., Y. Song, A. B. H. Bakker, S. Bauer, T. Spies, L. L. Lanier, and J. H. Phillips. 1999. An Activating Immunoreceptor Complex Formed by NKG2D and DAP10. Science 285:730-732.

164

112. Bauer, S., V. Groh, J. Wu, A. Steinle, J. H. Phillips, L. L. Lanier, and T. Spies. 1999. Activation of NK Cells and T Cells by NKG2D, a Receptor for Stress- Inducible MICA. Science 285:727-729.

113. Arnon, T. I., G. Markel, and O. Mandelboim. 2006. Tumor and viral recognition by natural killer cells receptors. Seminars in Cancer Biology 16:348-358.

114. Cooper, M. A., T. A. Fehniger, A. Fuchs, M. Colonna, and M. A. Caligiuri. 2004. NK cell and DC interactions. Trends in Immunology 25:47.

115. Trinchieri, G., M. Matsumoto-Kobayashi, S. C. Clark, J. Seehra, L. London, and B. Perussia. 1984. Response of resting human peripheral blood natural killer cells to interleukin 2. J Exp Med 160:1147-1169.

116. Henney, C. S., K. Kuribayashi, D. E. Kern, and S. Gillis. 1981. Interleukin-2 augments natural killer cell activity. Nature 291:335-338.

117. Dybkaer, K., J. Iqbal, G. Zhou, H. Geng, L. Xiao, A. Schmitz, F. d'Amore, and W. C. Chan. 2007. Genome wide transcriptional analysis of resting and IL2 activated human natural killer cells: gene expression signatures indicative of novel molecular signaling pathways. BMC Genomics 8:230.

118. Wang, K. S., J. Ritz, and D. A. Frank. 1999. IL-2 Induces STAT4 Activation in Primary NK Cells and NK Cell Lines, But Not in T Cells. J Immunol 162:299- 304.

119. Carson, W. E., T. A. Fehniger, S. Haldar, K. Eckhert, M. J. Lindemann, C. F. Lai, C. M. Croce, H. Baumann, and M. A. Caligiuri. 1997. A potential role for interleukin-15 in the regulation of human natural killer cell survival. J Clin Invest 99:937-943.

120. Cooper, M. A., J. E. Bush, T. A. Fehniger, J. B. VanDeusen, R. E. Waite, Y. Liu, H. L. Aguila, and M. A. Caligiuri. 2002. In vivo evidence for a dependence on interleukin 15 for survival of natural killer cells. Blood 100:3633-3638.

121. Schorle, H., T. Holtschke, T. Hunig, A. Schimpl, and I. Horak. 1991. Development and function of T cells in mice rendered interleukin-2 deficient by gene targeting. Nature 352:621-624.

165

122. Kundig, T. M., H. Schorle, M. F. Bachmann, H. Hengartner, R. M. Zinkernagel, and I. Horak. 1993. Immune responses in interleukin-2-deficient mice. Science 262:1059-1061.

123. Kobayashi, M., L. Fitz, M. Ryan, R. M. Hewick, S. C. Clark, S. Chan, R. Loudon, F. Sherman, B. Perussia, and G. Trinchieri. 1989. Identification and purification of natural killer cell stimulatory factor (NKSF), a cytokine with multiple biologic effects on human lymphocytes. J Exp Med 170:827-845.

124. Okamura, H., H. Tsutsui, T. Komatsu, M. Yutsudo, A. Hakura, T. Tanimoto, K. Torigoe, T. Okura, Y. Nukada, K. Hattori, K. Akita, M. Namba, F. Tanabe, K. Konishi, S. Fukuda, and M. Kurimoto. 1995. Cloning of a new cytokine that induces IFN-[gamma] production by T cells. Nature 378:88-91.

125. Robertson, M. J., R. J. Soiffer, S. F. Wolf, T. J. Manley, C. Donahue, D. Young, S. H. Herrmann, and J. Ritz. 1992. Response of human natural killer (NK) cells to NK cell stimulatory factor (NKSF): cytolytic activity and proliferation of NK cells are differentially regulated by NKSF. J Exp Med 175:779-788.

126. Chan, S. H., M. Kobayashi, D. Santoli, B. Perussia, and G. Trinchieri. 1992. Mechanisms of IFN-gamma induction by natural killer cell stimulatory factor (NKSF/IL-12). Role of transcription and mRNA stability in the synergistic interaction between NKSF and IL-2. J Immunol 148:92-98.

127. Mavropoulos, A., G. Sully, A. P. Cope, and A. R. Clark. 2005. Stabilization of IFN-gamma mRNA by MAPK p38 in IL-12- and IL-18-stimulated human NK cells. Blood 105:282-288.

128. Thierfelder, W. E., J. M. van Deursen, K. Yamamoto, R. A. Tripp, S. R. Sarawar, R. T. Carson, M. Y. Sangster, D. A. Vignali, P. C. Doherty, G. C. Grosveld, and J. N. Ihle. 1996. Requirement for Stat4 in interleukin-12-mediated responses of natural killer and T cells. Nature 382:171-174.

129. Robinson, D., K. Shibuya, A. Mui, F. Zonin, E. Murphy, T. Sana, S. B. Hartley, S. Menon, R. Kastelein, F. Bazan, and A. O'Garra. 1997. IGIF does not drive Th1 development but synergizes with IL-12 for interferon-gamma production and activates IRAK and NFkappaB. Immunity 7:571-581.

130. Lapaque, N., T. Walzer, S. Meresse, E. Vivier, and J. Trowsdale. 2009. Interactions between human NK cells and macrophages in response to Salmonella infection. J Immunol 182:4339-4348. 166

131. Ortaldo, J. R., R. Winkler-Pickett, J. Wigginton, M. Horner, E. W. Bere, A. T. Mason, N. Bhat, J. Cherry, M. Sanford, D. L. Hodge, and H. A. Young. 2006. Regulation of ITAM-positive receptors: role of IL-12 and IL-18. Blood 107:1468- 1475.

132. Lucas, M., W. Schachterle, K. Oberle, P. Aichele, and A. Diefenbach. 2007. Dendritic Cells Prime Natural Killer Cells by trans-Presenting Interleukin 15. Immunity 26:503-517.

133. Ortaldo, J. R., N. P. Lang, T. Timonen, and R. B. Herberman. 1981. Augmentation of human natural killer cell activity by interferon: conditions required for boosting and characteristics of the effector cells. J Interferon Res 1:253-262.

134. Kornberg, R. D. 1974. Chromatin structure: a repeating unit of histones and DNA. Science 184:868-871.

135. Kornberg, R. D., and J. O. Thomas. 1974. Chromatin structure; oligomers of the histones. Science 184:865-868.

136. Luger, K., A. W. Mader, R. K. Richmond, D. F. Sargent, and T. J. Richmond. 1997. Crystal structure of the nucleosome core particle at 2.8 A resolution. Nature 389:251-260.

137. Luger, K. 2003. Structure and dynamic behavior of nucleosomes. Curr Opin Genet Dev 13:127-135.

138. Allfrey, V. G., R. Faulkner, and A. E. Mirsky. 1964. Acetylation and Methylation of Histones and Their Possible Role in the Regulation of Rna Synthesis. Proc Natl Acad Sci U S A 51:786-794.

139. Jenuwein, T., and C. D. Allis. 2001. Translating the histone code. Science 293:1074-1080.

140. Hayakawa, T., and J. Nakayama. 2011. Physiological roles of class I HDAC complex and histone demethylase. J Biomed Biotechnol 2011:129383.

167

141. Jones, P. L., and Y. B. Shi. 2003. N-CoR-HDAC corepressor complexes: roles in transcriptional regulation by nuclear hormone receptors. Curr Top Microbiol Immunol 274:237-268.

142. Kuzmichev, A., T. Jenuwein, P. Tempst, and D. Reinberg. 2004. Different EZH2- containing complexes target methylation of histone H1 or nucleosomal histone H3. Mol Cell 14:183-193.

143. Ruthenburg, A. J., C. D. Allis, and J. Wysocka. 2007. Methylation of Lysine 4 on Histone H3: Intricacy of Writing and Reading a Single Epigenetic Mark. Molecular Cell 25:15-30.

144. Kouzarides, T. 2002. Histone methylation in transcriptional control. Current Opinion in Genetics & Development 12:198-209.

145. Crosio, C., G. M. Fimia, R. Loury, M. Kimura, Y. Okano, H. Zhou, S. Sen, C. D. Allis, and P. Sassone-Corsi. 2002. Mitotic Phosphorylation of Histone H3: Spatio- Temporal Regulation by Mammalian Aurora Kinases. Mol. Cell. Biol. 22:874- 885.

146. Fischle, W., Y. Wang, and C. David Allis. 2003. Binary switches and modification cassettes in histone biology and beyond. Nature 425:475-479.

147. Eissenberg, J. C., T. C. James, D. M. Foster-Hartnett, T. Hartnett, V. Ngan, and S. C. Elgin. 1990. Mutation in a heterochromatin-specific chromosomal protein is associated with suppr ession of position-effect variegation in Drosophila melanogaster. Proceedings of the National Academy of Sciences 87:9923-9927.

148. Nielsen, P. R., D. Nietlispach, H. R. Mott, J. Callaghan, A. Bannister, T. Kouzarides, A. G. Murzin, N. V. Murzina, and E. D. Laue. 2002. Structure of the HP1 chromodomain bound to histone H3 methylated at lysine 9. Nature 416:103- 107.

149. Jacobs, S. A., and S. Khorasanizadeh. 2002. Structure of HP1 chromodomain bound to a lysine 9-methylated histone H3 tail. Science 295:2080-2083.

150. Zhang, Y., and D. Reinberg. 2001. Transcription regulation by histone methylation: interplay between different covalent modifications of the core histone tails. Genes & Development 15:2343-2360.

168

151. Krouwels, I. M., K. Wiesmeijer, T. E. Abraham, C. Molenaar, N. P. Verwoerd, H. J. Tanke, and R. W. Dirks. 2005. A glue for heterochromatin maintenance: stable SUV39H1 binding to heterochromatin is reinforced by the SET domain. J Cell Biol 170:537-549.

152. Barski, A., S. Cuddapah, K. Cui, T. Y. Roh, D. E. Schones, Z. Wang, G. Wei, I. Chepelev, and K. Zhao. 2007. High-resolution profiling of histone methylations in the human genome. Cell 129:823-837.

153. Milner, C. M., and R. D. Campbell. 1993. The G9a gene in the human major histocompatibility complex encodes a novel protein containing ankyrin-like repeats. Biochem J 290 ( Pt 3):811-818.

154. Brown, S. E., R. D. Campbell, and C. M. Sanderson. 2001. Novel NG36/G9a gene products encoded within the human and mouse MHC class III regions. Mamm Genome 12:916-924.

155. Tachibana, M., K. Sugimoto, T. Fukushima, and Y. Shinkai. 2001. Set domain- containing protein, G9a, is a novel lysine-preferring mammalian histone methyltransferase with hyperactivity and specific selectivity to lysines 9 and 27 of histone H3. J Biol Chem 276:25309-25317.

156. Tachibana, M., K. Sugimoto, M. Nozaki, J. Ueda, T. Ohta, M. Ohki, M. Fukuda, N. Takeda, H. Niida, H. Kato, and Y. Shinkai. 2002. G9a histone methyltransferase plays a dominant role in euchromatic histone H3 lysine 9 methylation and is essential for early embryogenesis. Genes Dev 16:1779-1791.

157. Tachibana, M., J. Ueda, M. Fukuda, N. Takeda, T. Ohta, H. Iwanari, T. Sakihama, T. Kodama, T. Hamakubo, and Y. Shinkai. 2005. Histone methyltransferases G9a and GLP form heteromeric complexes and are both crucial for methylation of euchromatin at H3-K9. Genes Dev 19:815-826.

158. Shinkai, Y., and M. Tachibana. 2011. H3K9 methyltransferase G9a and the related molecule GLP. Genes Dev 25:781-788.

159. Ueda, J., M. Tachibana, T. Ikura, and Y. Shinkai. 2006. Zinc finger protein Wiz links G9a/GLP histone methyltransferases to the co-repressor molecule CtBP. J Biol Chem 281:20120-20128.

169

160. Esteve, P.-O., H. G. Chin, A. Smallwood, G. R. Feehery, O. Gangisetty, A. R. Karpf, M. F. Carey, and S. Pradhan. 2006. Direct interaction between DNMT1 and G9a coordinates DNA and histone methylation during replication. Genes Dev. 20:3089-3103.

161. Espada, J., E. Ballestar, M. F. Fraga, A. Villar-Garea, A. Juarranz, J. C. Stockert, K. D. Robertson, F. Fuks, and M. Esteller. 2004. Human DNA methyltransferase 1 is required for maintenance of the histone H3 modification pattern. J Biol Chem 279:37175-37184.

162. Tachibana, M., Y. Matsumura, M. Fukuda, H. Kimura, and Y. Shinkai. 2008. G9a/GLP complexes independently mediate H3K9 and DNA methylation to silence transcription. Embo J 27:2681-2690.

163. Kim, J. K., P. O. Esteve, S. E. Jacobsen, and S. Pradhan. 2009. UHRF1 binds G9a and participates in p21 transcriptional regulation in mammalian cells. Nucleic Acids Res 37:493-505.

164. Lehnertz, B., J. P. Northrop, F. Antignano, K. Burrows, S. Hadidi, S. C. Mullaly, F. M. V. Rossi, and C. Zaph. 2010. Activating and inhibitory functions for the histone lysine methyltransferase G9a in T helper cell differentiation and function. The Journal of Experimental Medicine 207:915-922.

165. Osipovich, O., R. Milley, A. Meade, M. Tachibana, Y. Shinkai, M. S. Krangel, and E. M. Oltz. 2004. Targeted inhibition of V(D)J recombination by a histone methyltransferase. Nat Immunol 5:309-316.

166. Thomas, L. R., H. Miyashita, R. M. Cobb, S. Pierce, M. Tachibana, E. Hobeika, M. Reth, Y. Shinkai, and E. M. Oltz. 2008. Functional analysis of histone methyltransferase g9a in B and T lymphocytes. J Immunol 181:485-493.

167. Maze, I., H. E. Covington, D. M. Dietz, Q. LaPlant, W. Renthal, S. J. Russo, M. Mechanic, E. Mouzon, R. L. Neve, S. J. Haggarty, Y. Ren, S. C. Sampath, Y. L. Hurd, P. Greengard, A. Tarakhovsky, A. Schaefer, and E. J. Nestler. 2010. Essential Role of the Histone Methyltransferase G9a in Cocaine-Induced Plasticity. Science 327:213-216.

168. Trievel, R. C. 2004. Structure and function of histone methyltransferases. Crit Rev Eukaryot Gene Expr 14:147-169.

170

169. Zhang, X., Z. Yang, S. I. Khan, J. R. Horton, H. Tamaru, E. U. Selker, and X. Cheng. 2003. Structural Basis for the Product Specificity of Histone Lysine Methyltransferases. Molecular Cell 12:177.

170. Zhang, X., H. Tamaru, S. I. Khan, J. R. Horton, L. J. Keefe, E. U. Selker, and X. Cheng. 2002. Structure of the Neurospora SET domain protein DIM-5, a histone H3 lysine methyltransferase. Cell 111:117-127.

171. Collins, R. E., M. Tachibana, H. Tamaru, K. M. Smith, D. Jia, X. Zhang, E. U. Selker, Y. Shinkai, and X. Cheng. 2005. In vitro and in vivo analyses of a Phe/Tyr switch controlling product specificity of histone lysine methyltransferases. J Biol Chem 280:5563-5570.

172. Collins, R. E., J. P. Northrop, J. R. Horton, D. Y. Lee, X. Zhang, M. R. Stallcup, and X. Cheng. 2008. The ankyrin repeats of G9a and GLP histone methyltransferases are mono- and dimethyllysine binding modules. Nat Struct Mol Biol 15:245-250.

173. Van Gelder, R. N., M. E. von Zastrow, A. Yool, W. C. Dement, J. D. Barchas, and J. H. Eberwine. 1990. Amplified RNA synthesized from limited qua ntities of heterogeneous cDNA. Proc Natl Acad Sci U S A 87:1663-1667.

174. Nilsson, M., J. Ljungberg, J. Richter, T. Kiefer, M. Magnusson, A. Lieber, B. Widegren, S. Karlsson, and X. Fan. 2004. Development of an adenoviral vector system with adenovirus serotype 35 tropism; efficient transient gene transfer into primary malignant hematopoietic cells. J Gene Med 6:631-641.

175. Ji, H., H. Jiang, W. Ma, D. S. Johnson, R. M. Myers, and W. H. Wong. 2008. An integrated software system for analyzing ChIP-chip and ChIP-seq data. Nat Biotech 26:1293-1300.

176. Johnson, W. E., W. Li, C. A. Meyer, R. Gottardo, J. S. Carroll, M. Brown, and X. S. Liu. 2006. Model-based analysis of tiling-arrays for ChIP-chip. Proceedings of the National Academy of Sciences 103:12457-12462.

177. Golubovskaya, V., A. Kaur, and W. Cance. 2004. Cloning and characterization of the promoter region of human focal adhesion kinase gene: nuclear factor kappa B and p53 binding sites. Biochim Biophys Acta 1678:111-125.

171

178. Glimcher, L. H., M. J. Townsend, B. M. Sullivan, and G. M. Lord. 2004. Recent developments in the transcriptional regulation of cytolytic effector cells. Nat Rev Immunol 4:900-911.

179. Rosenberger, C. M., A. E. Clark, P. M. Treuting, C. D. Johnson, and A. Aderem. 2008. ATF3 regulates MCMV infection in mice by modulating IFN-Î³ expression in natural killer cells. Proceedings of the National Academy of Sciences 105:2544- 2549.

180. Becknell, B., T. L. Hughes, A. G. Freud, B. W. Blaser, J. Yu, R. Trotta, H. C. Mao, M. L. Caligiuri de Jesus, M. Alghothani, D. M. Benson, Jr., A. Lehman, D. Jarjoura, D. Perrotti, M. D. Bates, and M. A. Caligiuri. 2006. The Hlx homeobox transcription factor negatively regulates interferon-{gamma} production in monokine-activated natural killer cells. Blood.

181. Martins, G., and K. Calame. 2008. Regulation and Functions of Blimp-1 in T and B Lymphocytes. Annual Review of Immunology 26:133-169.

182. Mittler, R., J. Foell, M. McCausland, S. Strahotin, L. Niu, A. Bapat, and L. Hewes. 2004. Anti-CD137 antibodies in the treatment of autoimmune disease and cancer. Immunologic Research 29:197-208.

183. Zingoni, A., T. Sornasse, B. G. Cocks, Y. Tanaka, A. Santoni, and L. L. Lanier. 2004. Cross-Talk between Activated Human NK Cells and CD4+ T Cells via OX40-OX40 Ligand Interactions. J Immunol 173:3716-3724.

184. Brilot, F., T. Strowig, S. M. Roberts, F. Arrey, and C. Münz. 2007. NK cell survival mediated through the regulatory synapse with human DCs requires IL- 15Rα. The Journal of Clinical Investigation 117:3316-3329.

185. Sayers, T. J., A. T. Mason, and J. R. Ortaldo. 1986. Regulation of human natural killer cell activity by interferon-gamma: lack of a role in interleukin 2-mediated augmentation. J Immunol 136:2176-2180.

186. Cruz-Munoz, M. E., Z. Dong, X. Shi, S. Zhang, and A. Veillette. 2009. Influence of CRACC, a SLAM family receptor coupled to the adaptor EAT-2, on natural killer cell function. Nat Immunol 10:297-305.

172

187. Echlin, D. R., H. J. Tae, N. Mitin, and E. J. Taparowsky. 2000. B-ATF functions as a negative regulator of AP-1 mediated transcription and blocks cellular transformation by Ras and Fos. Oncogene 19:1752-1763.

188. Dorsey, M. J., H. J. Tae, K. G. Sollenberger, N. T. Mascarenhas, L. M. Johansen, and E. J. Taparowsky. 1995. B-ATF: a novel human bZIP protein that associates with members of the AP-1 transcription factor family. Oncogene 11:2255-2265.

189. Okamura, T., K. Fujio, M. Shibuya, S. Sumitomo, H. Shoda, S. Sakaguchi, and K. Yamamoto. 2009. CD4+CD25−LAG3+ regulatory T cells controlled by the transcription factor Egr-2. Proceedings of the National Academy of Sciences 106:13974-13979.

190. Lauritsen, J.-P. H., S. Kurella, S.-Y. Lee, J. M. Lefebvre, M. Rhodes, J. Alberola- Ila, and D. L. Wiest. 2008. Egr2 Is Required for Bcl-2 Induction during Positive Selection. The Journal of Immunology 181:7778-7785.

191. Pearce, E. L., A. C. Mullen, G. A. Martins, C. M. Krawczyk, A. S. Hutchins, V. P. Zediak, M. Banica, C. B. DiCioccio, D. A. Gross, C.-a. Mao, H. Shen, N. Cereb, S. Y. Yang, T. Lindsten, J. Rossant, C. A. Hunter, and S. L. Reiner. 2003. Control of Effector CD8+ T Cell Function by the Transcription Factor Eomesodermin. Science 302:1041-1043.

192. Tayade, C., Y. Fang, G. P. Black, P. VA, A. Erlebacher, and B. A. Croy. 2005. Differential transcription of Eomes and T-bet during maturation of mouse uterine natural killer cells. Journal of Leukocyte Biology 78:1347-1355.

193. Trotta, R., R. Parihar, J. Yu, B. Becknell, J. Allard, 2nd, J. Wen, W. Ding, H. Mao, S. Tridandapani, W. E. Carson, and M. A. Caligiuri. 2005. Differential expression of SHIP1 in CD56bright and CD56dim NK cells provides a molecular basis for distinct functional responses to monokine costimulation. Blood 105:3011-3018.

194. Ettinger, R., G. P. Sims, A. M. Fairhurst, R. Robbins, Y. S. da Silva, R. Spolski, W. J. Leonard, and P. E. Lipsky. 2005. IL-21 Induces Differentiation of Human Naive and Memory B Cells into Antibody-Secreting Plasma Cells. J Immunol 175:7867-7879.

195. Desai, S., M. Maurin, M. A. Smith, S. C. Bolick, S. Dessureault, J. Tao, E. Sotomayor, and K. L. Wright. 2010. PRDM1 is required for mantle cell lymphoma response to bortezomib. Mol Cancer Res 8:907-918. 173

196. Anegon, I., M. C. Cuturi, G. Trinchieri, and B. Perussia. 1988. Interaction of Fc receptor (CD16) ligands induces transcription of interleukin 2 receptor (CD25) and lymphokine genes and expression of their products in human natural killer cells. J Exp Med 167:452-472.

197. Chehimi, J., S. E. Starr, I. Frank, M. Rengaraju, S. J. Jackson, C. Llanes, M. Kobayashi, B. Perussia, D. Young, E. Nickbarg, and et al. 1992. Natural killer (NK) cell stimulatory factor increases the cytotoxic activity of NK cells from both healthy donors and human immunodeficiency virus-infected patients. J Exp Med 175:789-796.

198. Huntington, N. D., C. A. J. Vosshenrich, and J. P. Di Santo. 2007. Developmental pathways that generate natural-killer-cell diversity in mice and humans. Nat Rev Immunol 7:703.

199. Hanna, J., T. Gonen-Gross, J. Fitchett, T. Rowe, M. Daniels, T. I. Arnon, R. Gazit, A. Joseph, K. W. Schjetne, A. Steinle, A. Porgador, D. Mevorach, D. Goldman-Wohl, S. Yagel, M. J. LaBarre, J. H. Buckner, and O. Mandelboim. 2004. Novel APC-like properties of human NK cells directly regulate T cell activation. J Clin Invest 114:1612-1623.

200. Schoenborn, J. R., M. O. Dorschner, M. Sekimata, D. M. Santer, M. Shnyreva, D. R. Fitzpatrick, J. A. Stamatoyannopoulos, and C. B. Wilson. 2007. Comprehensive epigenetic profiling identifies multiple distal regulatory elements directing transcription of the gene encoding interferon-gamma. Nat Immunol 8:732-742.

201. Taylor, J. M., K. Wicks, C. Vandiedonck, and J. C. Knight. 2008. Chromatin profiling across the human tumour necrosis factor gene locus reveals a complex, cell type-specific landscape with novel regulatory elements. Nucleic Acids Res 36:4845-4862.

202. Tsytsykova, A. V., R. Rajsbaum, J. V. Falvo, F. Ligeiro, S. R. Neely, and A. E. Goldfeld. 2007. Activation-dependent intrachromosomal interactions formed by the TNF gene promoter and two distal enhancers. Proc Natl Acad Sci U S A 104:16850-16855.

203. Chang, S., and T. M. Aune. 2005. Histone hyperacetylated domains across the Ifng gene region in natural killer cells and T cells. Proc Natl Acad Sci U S A 102:17095-17100.

174

204. Fehniger, T. A., M. A. Cooper, G. J. Nuovo, M. Cella, F. Facchetti, M. Colonna, and M. A. Caligiuri. 2003. CD56bright natural killer cells are present in human lymph nodes and are activated by T cell-derived IL-2: a potential new link between adaptive and innate immunity. Blood 101:3052-3057.

205. Chan, A., D. L. Hong, A. Atzberger, S. Kollnberger, A. D. Filer, C. D. Buckley, A. McMichael, T. Enver, and P. Bowness. 2007. CD56bright human NK cells differentiate into CD56dim cells: role of contact with peripheral fibroblasts. J Immunol 179:89-94.

206. Zhao, W. L., Y. Y. Liu, Q. L. Zhang, L. Wang, C. Leboeuf, Y. W. Zhang, J. Ma, J. F. Garcia, Y. P. Song, J. M. Li, Z. X. Shen, Z. Chen, A. Janin, and S. J. Chen. 2008. PRDM1 is involved in chemoresistance of T-cell lymphoma and down- regulated by the proteasome inhibitor. Blood 111:3867-3871.

207. Huang, Y., A. de Reynies, L. de Leval, B. Ghazi, N. Martin-Garcia, M. Travert, J. Bosq, J. Briere, B. Petit, E. Thomas, P. Coppo, T. Marafioti, J. F. Emile, M. H. Delfau-Larue, C. Schmitt, and P. Gaulard. 2010. Gene expression profiling identifies emerging oncogenic pathways operating in extranodal NK/T-cell lymphoma, nasal type. Blood 115:1226-1237.

208. Kallies, A., S. Carotta, N. D. Huntington, N. J. Bernard, D. M. Tarlinton, M. J. Smyth, and S. L. Nutt. 2011. A role for Blimp1 in the transcriptional network controlling natural killer cell maturation. Blood 117:1869-1879.

209. Sun, J. C., J. N. Beilke, and L. L. Lanier. 2009. Adaptive immune features of natural killer cells. Nature 457:557-561.

210. Farnham, P. J. 2009. Insights from genomic profiling of transcription factors. Nat Rev Genet 10:605-616.

211. Ji, H., and W. H. Wong. TileMap: create chromosomal map of tiling array hybridizations. Bioinformatics 21:3629-3636.

212. Vivier, E., A. J. da Silva, M. Ackerly, H. Levine, C. E. Rudd, and P. Anderson. 1993. Association of a 70-kDa tyrosine phosphoprotein with the CD16:ζ:γ complex expressed in human natural killer cells. European Journal of Immunology 23:1872-1876.

175

213. Hara, H., C. Ishihara, A. Takeuchi, L. Xue, S. W. Morris, J. M. Penninger, H. Yoshida, and T. Saito. 2008. Cell Type-Specific Regulation of ITAM-Mediated NF-κB Activation by the Adaptors, CARMA1 and CARD9. The Journal of Immunology 181:918-930.

214. Tanaka, N., K. Kaneko, H. Asao, H. Kasai, Y. Endo, T. Fujita, T. Takeshita, and K. Sugamura. 1999. Possible Involvement of a Novel STAM-associated Molecule “AMSH” in Intracellular Signal Transduction Mediated by Cytokines. Journal of Biological Chemistry 274:19129-19135.

215. Fortier, J. M., and J. Kornbluth. 2006. NK Lytic-Associated Molecule, Involved in NK Cytotoxic Function, Is an E3 Ligase. The Journal of Immunology 176:6454-6463.

216. Kozlowski, M., J. Schorey, T. Portis, V. Grigoriev, and J. Kornbluth. 1999. NK Lytic-Associated Molecule: A Novel Gene Selectively Expressed in Cells with Cytolytic Function. The Journal of Immunology 163:1775-1785.

217. Hoover, R. G., G. Gullickson, and J. Kornbluth. 2009. Impaired NK Cytolytic Activity and Enhanced Tumor Growth in NK Lytic-Associated Molecule- Deficient Mice. The Journal of Immunology 183:6913-6921.

218. Natkunam, Y., S. Zhao, D. Y. Mason, J. Chen, B. Taidi, M. Jones, A. S. Hammer, S. Hamilton Dutoit, I. S. Lossos, and R. Levy. 2007. The oncoprotein LMO2 is expressed in normal germinal-center B cells and in human B-cell lymphomas. Blood 109:1636-1642.

219. Doody, G. M., M. A. Care, N. J. Burgoyne, J. R. Bradford, M. Bota, C. Bonifer, D. R. Westhead, and R. M. Tooze. 2010. An extended set of PRDM1/BLIMP1 target genes links binding motif type to dynamic repression. Nucleic Acids Res:1- 15.

220. Lanier, L. L., B. Corliss, J. Wu, and J. H. Phillips. 1998. Association of DAP12 with Activating CD94/NKG2C NK Cell Receptors. Immunity 8:693-701.

221. Brooks, A. G., P. E. Posch, C. J. Scorzelli, F. Borrego, and J. E. Coligan. 1997. NKG2A Complexed with CD94 Defines a Novel Inhibitory Natural Killer Cell Receptor. The Journal of Experimental Medicine 185:795-800.

176

222. Huntington, N. D., H. Puthalakath, P. Gunn, E. Naik, E. M. Michalak, M. J. Smyth, H. Tabarias, M. A. Degli-Esposti, G. Dewson, S. N. Willis, N. Motoyama, D. C. S. Huang, S. L. Nutt, D. M. Tarlinton, and A. Strasser. 2007. Interleukin 15- mediated survival of natural killer cells is determined by interactions among Bim, Noxa and Mcl-1. Nat Immunol 8:856-863.

223. Morris, S. C., T. Orekhova, M. J. Meadows, S. M. Heidorn, J. Yang, and F. D. Finkelman. 2006. IL-4 Induces In Vivo Production of IFN-γ by NK and NKT Cells. The Journal of Immunology 176:5299-5305.

224. Bream, J. H., R. E. Curiel, C.-R. Yu, C. E. Egwuagu, M. J. Grusby, T. M. Aune, and H. A. Young. 2003. IL-4 synergistically enhances both IL-2 and IL- 12induced IFN-γ expression in murine NK cells. Blood 102:207-214.

225. Orange, J. S., N. Ramesh, E. Remold-O'Donnell, Y. Sasahara, L. Koopman, M. Byrne, F. A. Bonilla, F. S. Rosen, R. S. Geha, and J. L. Strominger. 2002. Wiskott–Aldrich syndrome protein is required for NK cell cytotoxicity and colocalizes with actin to NK cell-activating immunologic synapses. Proceedings of the National Academy of Sciences 99:11351-11356.

226. Behrens, T., J. Jagadeesh, P. Scherle, G. Kearns, J. Yewdell, and L. Staudt. 1994. Jaw1, A lymphoid-restricted membrane protein localized to the endoplasmic reticulum. The Journal of Immunology 153:682-690.

227. Snyder, H. L., I. Bačík, J. R. Bennink, G. Kearns, T. W. Behrens, T. Bächi, M. Orlowski, and J. W. Yewdell. 1997. Two Novel Routes of Transporter Associated with Antigen Processing (TAP)-independent Major Histocompatibility Complex Class I Antigen Processing. The Journal of Experimental Medicine 186:1087- 1098.

228. Li, Y., X. He, J. Schembri-King, S. Jakes, and J. Hayashi. 2000. Cloning and Characterization of Human Lnk, an Adaptor Protein with Pleckstrin Homology and Src Homology 2 Domains that Can Inhibit T Cell Activation. The Journal of Immunology 164:5199-5206.

229. Takaki, S., Y. Tezuka, K. Sauer, C. Kubo, S.-M. Kwon, E. Armstead, K. Nakao, M. Katsuki, R. M. Perlmutter, and K. Takatsu. 2003. Impaired Lymphopoiesis and Altered B Cell Subpopulations in Mice Overexpressing Lnk Adaptor Protein. The Journal of Immunology 170:703-710.

177

230. Pei, L., A. Castrillo, M. Chen, A. Hoffmann, and P. Tontonoz. 2005. Induction of NR4A orphan nuclear receptor expression in macrophages in response to inflammatory stimuli. J. Biol. Chem. 280:29256-29262.

231. Bohlander, S. K. 2005. ETV6: A versatile player in leukemogenesis. Seminars in Cancer Biology 15:162-174.

232. Singh, U. P., S. Singh, R. Singh, Y. Cong, D. D. Taub, and J. W. Lillard. 2008. CXCL10-Producing Mucosal CD4+ T Cells, NK Cells, and NKT Cells Are Associated with Chronic Colitis in IL-10−/− Mice, Which Can Be Abrogated by Anti-CXCL10 Antibody Inhibition. Journal of Interferon & Cytokine Research 28:31-43.

233. Robertson, M. J., K. J. Cochran, C. Cameron, J. M. Le, R. Tantravahi, and J. Ritz. 1996. Characterization of a cell line, NKL, derived from an aggressive human natural killer cell leukemia. Exp Hematol 24:406-415.

234. 2007. Identification and analysis of functional elements in 1% of the human genome by the ENCODE pilot project. Nature 447:799-816.

235. Gonsalves, S. E., A. M. Moses, Z. Razak, F. Robert, and J. T. Westwood. 2011. Whole-Genome Analysis Reveals That Active Heat Shock Factor Binding Sites Are Mostly Associated with Non-Heat Shock Genes in Drosophila melanogaster. PLoS ONE 6:e15934.

236. Kallies, A., S. Carotta, N. D. Huntington, N. J. Bernard, D. M. Tarlinton, M. J. Smyth, and S. L. Nutt. 2011. A role for Blimp1 in the transcriptional network controlling natural killer cell maturation. Blood 117:1869-1879.

237. Gismondi, A., S. Morrone, M. J. Humphries, M. Piccoli, L. Frati, and A. Santoni. 1991. Human natural killer cells express VLA-4 and VLA-5, which mediate their adhesion to fibronectin. J Immunol 146:384-392.

238. Kornberg, L., H. S. Earp, J. T. Parsons, M. Schaller, and R. L. Juliano. 1992. Cell adhesion or integrin clustering increases phosphorylation of a focal adhesion- associated tyrosine kinase. J Biol Chem 267:23439-23442.

239. Schaller, M. D., C. A. Borgman, B. S. Cobb, R. R. Vines, A. B. Reynolds, and J. T. Parsons. 1992. pp125FAK a structurally distinctive protein-tyrosine kinase associated with focal adhesions. Proc Natl Acad Sci U S A 89:5192-5196.

178

240. Parsons, J. T. 2003. Focal adhesion kinase: the first ten years. J Cell Sci 116:1409-1416.

241. Tilghman, R. W., J. K. Slack-Davis, N. Sergina, K. H. Martin, M. Iwanicki, E. D. Hershey, H. E. Beggs, L. F. Reichardt, and J. T. Parsons. 2005. Focal adhesion kinase is required for the spatial organization of the leading edge in migrating cells. Journal of Cell Science 118:2613-2623.

242. Huang, D., M. Khoe, M. Befekadu, S. Chung, Y. Takata, D. Ilic, and M. Bryer- Ash. 2007. Focal adhesion kinase mediates cell survival via NF-kappaB and ERK signaling pathways. Am J Physiol Cell Physiol 292:C1339-1352.

243. van Seventer, G. A., H. J. Salmen, S. F. Law, G. M. O'Neill, M. M. Mullen, A. M. Franz, S. B. Kanner, E. A. Golemis, and J. M. van Seventer. 2001. Focal adhesion kinase regulates beta1 integrin-dependent T cell migration through an HEF1 effector pathway. Eur J Immunol 31:1417-1427.

244. Iwata, S., Y. Ohashi, K. Kamiguchi, and C. Morimoto. 2000. Beta 1-integrin- mediated cell signaling in T lymphocytes. Journal of Dermatological Science 23:75-86.

245. Tse, K. W., M. Dang-Lawson, R. L. Lee, D. Vong, A. Bulic, L. Buckbinder, and M. R. Gold. 2009. B cell receptor-induced phosphorylation of Pyk2 and focal adhesion kinase involves integrins and the Rap GTPases and is required for B cell spreading. J Biol Chem 284:22865-22877.

246. Williams, D. A., M. Rios, C. Stephens, and V. P. Patel. 1991. Fibronectin and VLA-4 in haematopoietic stem cell-microenvironment interactions. Nature 352:438-441.

247. Salomon, D. R., C. F. Mojcik, A. C. Chang, S. Wadsworth, D. H. Adams, J. E. Coligan, and E. M. Shevach. 1994. Constitutive activation of integrin alpha 4 beta 1 defines a unique stage of human thymocyte development. The Journal of Experimental Medicine 179:1573-1584.

248. Rabinowich, H., W. C. Lin, M. Manciulea, R. B. Herberman, and T. L. Whiteside. 1995. Induction of protein tyrosine phosphorylation in human natural killer cells by triggering via alpha 4 beta 1 or alpha 5 beta 1 integrins. Blood 85:1858-1864.

179

249. Rabinowich, H., M. Manciulea, R. B. Herberman, and T. L. Whiteside. 1996. Beta1 integrin-mediated activation of focal adhesion kinase and its association with Fyn and Zap-70 in human NK cells. J Immunol 157:3860-3868.

250. Palmieri, G., A. Serra, R. De Maria, A. Gismondi, M. Milella, M. Piccoli, L. Frati, and A. Santoni. 1995. Cross-linking of alpha 4 beta 1 and alpha 5 beta 1 fibronectin receptors enhances natural killer cell cytotoxic activity. The Journal of Immunology 155:5314-5322.

251. Zhang, T., S. Liu, P. Yang, C. Han, J. Wang, J. Liu, Y. Han, Y. Yu, and X. Cao. 2009. Fibronectin maintains survival of mouse natural killer (NK) cells via CD11b/Src/beta-catenin pathway. Blood 114:4081-4088.

252. Gismondi, A., L. Bisogno, F. Mainiero, G. Palmieri, M. Piccoli, L. Frati, and A. Santoni. 1997. Proline-rich tyrosine kinase-2 activation by beta 1 integrin fibronectin receptor cross-linking and association with paxillin in human natural killer cells. J Immunol 159:4729-4736.

253. Ding, J., S. Huang, S. Wu, Y. Zhao, L. Liang, M. Yan, C. Ge, J. Yao, T. Chen, D. Wan, H. Wang, J. Gu, M. Yao, J. Li, H. Tu, and X. He. 2010. Gain of miR-151 on chromosome 8q24.3 facilitates tumour cell migration and spreading through downregulating RhoGDIA. Nat Cell Biol 12:390-399.

254. Chen, J.-S., X.-H. Huang, Q. Wang, X.-L. Chen, X.-H. Fu, H.-X. Tan, L.-J. Zhang, W. Li, and J. Bi. 2010. FAK is involved in invasion and metastasis of hepatocellular carcinoma. Clinical and Experimental Metastasis 27:71-82.

255. Itoh, S., T. Maeda, M. Shimada, S.-i. Aishima, K. Shirabe, S. Tanaka, and Y. Maehara. 2004. Role of Expression of Focal Adhesion Kinase in Progression of Hepatocellular Carcinoma. Clinical Cancer Research 10:2812-2817.

256. Fulci, V., T. Colombo, S. Chiaretti, M. Messina, F. Citarella, S. Tavolaro, A. Guarini, R. Foà, and G. Macino. 2009. Characterization of B- and T-lineage acute lymphoblastic leukemia by integrated analysis of MicroRNA and mRNA expression profiles. Genes, Chromosomes and Cancer 48:1069-1082.

257. Garcia-Manero, G., and P. Fenaux. 2011. Hypomethylating Agents and Other Novel Strategies in Myelodysplastic Syndromes. Journal of Clinical Oncology 29:516-523.

180

258. Duvic, M., R. Talpur, X. Ni, C. Zhang, P. Hazarika, C. Kelly, J. H. Chiao, J. F. Reilly, J. L. Ricker, V. M. Richon, and S. R. Frankel. 2007. Phase 2 trial of oral vorinostat (suberoylanilide hydroxamic acid, SAHA) for refractory cutaneous T- cell lymphoma (CTCL). Blood 109:31-39.

259. Marks, P. A. 2007. Discovery and development of SAHA as an anticancer agent. Oncogene 26:1351-1356.

260. Greiner, D., T. Bonaldi, R. Eskeland, E. Roemer, and A. Imhof. 2005. Identification of a specific inhibitor of the histone methyltransferase SU(VAR)3- 9. Nat Chem Biol 1:143-145.

261. Jones, P. A., and S. B. Baylin. 2007. The epigenomics of cancer. Cell 128:683- 692.

262. Watanabe, H., K. Soejima, H. Yasuda, I. Kawada, I. Nakachi, S. Yoda, K. Naoki, and A. Ishizaka. 2008. Deregulation of histone lysine methyltransferases contributes to oncogenic transformation of human bronchoepithelial cells. Cancer Cell International 8:15.

263. Chen, M.-W., K.-T. Hua, H.-J. Kao, C.-C. Chi, L.-H. Wei, G. Johansson, S.-G. Shiah, P.-S. Chen, Y.-M. Jeng, T.-Y. Cheng, T.-C. Lai, J.-S. Chang, Y.-H. Jan, M.-H. Chien, C.-J. Yang, M.-S. Huang, M. Hsiao, and M.-L. Kuo. 2010. H3K9 Histone Methyltransferase G9a Promotes Lung Cancer Invasion and Metastasis by Silencing the Cell Adhesion Molecule Ep-CAM. Cancer Research 70:7830- 7840.

264. Wozniak, R. J., W. T. Klimecki, S. S. Lau, Y. Feinstein, and B. W. Futscher. 2006. 5-Aza-2'-deoxycytidine-mediated reductions in G9A histone methyltransferase and histone H3 K9 di-methylation levels are linked to tumor suppressor gene reactivation. Oncogene.

265. Rea, S., F. Eisenhaber, D. O'Carroll, B. D. Strahl, Z. W. Sun, M. Schmid, S. Opravil, K. Mechtler, C. P. Ponting, C. D. Allis, and T. Jenuwein. 2000. Regulation of chromatin structure by site-specific histone H3 methyltransferases. Nature 406:593-599.

266. Yang, W. M., S. C. Tsai, Y. D. Wen, G. Fejer, and E. Seto. 2002. Functional domains of histone deacetylase-3. J Biol Chem 277:9447-9454.

181

267. Gazzar, M. E., B. K. Yoza, X. Chen, J. Hu, G. A. Hawkins, and C. E. McCall. 2008. G9a and HP1 Couple Histone and DNA Methylation to TNF{alpha} Transcription Silencing during Endotoxin Tolerance. J. Biol. Chem. 283:32198- 32208.

268. Chen, L., Z. Li, A. K. Zwolinska, M. A. Smith, B. Cross, J. Koomen, Z.-M. Yuan, T. Jenuwein, J.-C. Marine, K. L. Wright, and J. Chen. 2010. MDM2 recruitment of lysine methyltransferases regulates p53 transcriptional output. Embo J 29:2538-2552.

269. Iwasa, E., Y. Hamashima, S. Fujishiro, E. Higuchi, A. Ito, M. Yoshida, and M. Sodeoka. 2010. Total Synthesis of (+)-Chaetocin and its Analogues: Their Histone Methyltransferase G9a Inhibitory Activity. Journal of the American Chemical Society 132:4078-4079.

270. Kubicek, S., R. J. O'Sullivan, E. M. August, E. R. Hickey, Q. Zhang, M. L. Teodoro, S. Rea, K. Mechtler, J. A. Kowalski, C. A. Homon, T. A. Kelly, and T. Jenuwein. 2007. Reversal of H3K9me2 by a Small-Molecule Inhibitor for the G9a Histone Methyltransferase. Molecular Cell 25:473.

271. Frye, S. V. 2010. The art of the chemical probe. Nat Chem Biol 6:159-161.

272. Smith, M. A., M. Maurin, H. I. Cho, B. Becknell, A. G. Freud, J. Yu, S. Wei, J. Djeu, E. Celis, M. A. Caligiuri, and K. L. Wright. 2010. PRDM1/Blimp-1 controls effector cytokine production in human NK cells. J Immunol 185:6058- 6067.

273. O'Leary, J. G., M. Goodarzi, D. L. Drayton, and U. H. von Andrian. 2006. T cell- and B cell-independent adaptive immunity mediated by natural killer cells. Nat Immunol 7:507-516.

274. Paust, S., and U. H. von Andrian. 2011. Natural killer cell memory. Nat Immunol 131:500-508.

275. Savage, A. K., M. G. Constantinides, J. Han, D. Picard, E. Martin, B. Li, O. Lantz, and A. Bendelac. 2008. The Transcription Factor PLZF Directs the Effector Program of the NKT Cell Lineage. Immunity 29:391-403.

182

276. Wong, W. F., K. Kohu, T. Chiba, T. Sato, and M. Satake. 2011. Interplay of transcription factors in T-cell differentiation and function: the role of Runx. Immunology 132:157-164.

277. Perez, S. A., L. G. Mahaira, F. J. Demirtzoglou, P. A. Sotiropoulou, P. Ioannidis, E. G. Iliopoulou, A. D. Gritzapis, N. N. Sotiriadou, C. N. Baxevanis, and M. Papamichail. 2005. A potential role for hydrocortisone in the positive regulation of IL-15–activated NK-cell proliferation and survival. Blood 106:158-166.

278. Tan, Y. H., J. A. Armstrong, Y. H. Ke, and M. Ho. 1970. Regulation of Cellular Interferon Production: Enhancement by Antimetabolites. Proceedings of the National Academy of Sciences 67:464-471.

279. Sharova, L. V., A. A. Sharov, T. Nedorezov, Y. Piao, N. Shaik, and M. S. Ko. 2009. Database for mRNA half-life of 19 977 genes obtained by DNA microarray analysis of pluripotent and differentiating mouse embryonic stem cells. DNA Res 16:45-58.

280. Liu, Y. Y., C. Leboeuf, J. Y. Shi, J. M. Li, L. Wang, Y. Shen, J. F. Garcia, Z. X. Shen, Z. Chen, A. Janin, S. J. Chen, and W. L. Zhao. 2007. Rituximab plus CHOP (R-CHOP) overcomes PRDM1-associated resistance to chemotherapy in patients with diffuse large B-cell lymphoma. Blood 110:339-344.

281. Diermayr, S., H. Himmelreich, B. Durovic, A. Mathys-Schneeberger, U. Siegler, U. Langenkamp, J. Hofsteenge, A. Gratwohl, A. Tichelli, M. Paluszewska, W. Wiktor-Jedrzejczak, C. P. Kalberer, and A. Wodnar-Filipowicz. 2008. NKG2D ligand expression in AML increases in response to HDAC inhibitor valproic acid and contributes to allorecognition by NK-cell lines with single KIR-HLA class I specificities. Blood 111:1428-1436.

282. Kato, N., J. Tanaka, J. Sugita, T. Toubai, Y. Miura, M. Ibata, Y. Syono, S. Ota, T. Kondo, M. Asaka, and M. Imamura. 2007. Regulation of the expression of MHC class I-related chain A, B (MICA, MICB) via chromatin remodeling and its impact on the susceptibility of leukemic cells to the cytotoxicity of NKG2D- expressing cells. Leukemia 21:2103-2108.

283. Skov, S., M. T. Pedersen, L. Andresen, P. Thor Straten, A. Woetmann, and N. Ødum. 2005. Cancer Cells Become Susceptible to Natural Killer Cell Killing after Exposure to Histone Deacetylase Inhibitors Due to Glycogen Synthase Kinase-3– Dependent Expression of MHC Class I–Related Chain A and B. Cancer Research 65:11136-11145.

183

284. Armeanu, S., M. Bitzer, U. M. Lauer, S. Venturelli, A. Pathil, M. Krusch, S. Kaiser, J. Jobst, I. Smirnow, A. Wagner, A. Steinle, and H. R. Salih. 2005. Natural Killer Cell–Mediated Lysis of Hepatoma Cells via Specific Induction of NKG2D Ligands by the Histone Deacetylase Inhibitor Sodium Valproate. Cancer Research 65:6321-6329.

285. Schmudde, M., E. Friebe, J. Sonnemann, J. F. Beck, and B. M. Broker. 2010. Histone deacetylase inhibitors prevent activation of tumour-reactive NK cells and T cells but do not interfere with their cytolytic effector functions. Cancer Lett 295:173-181.

286. Ogbomo, H., M. Michaelis, J. Kreuter, H. W. Doerr, and J. Cinatl, Jr. 2007. Histone deacetylase inhibitors suppress natural killer cell cytolytic activity. FEBS Lett 581:1317-1322.

184

APPENDICES

185

Appendix I: Genes changed 4-fold in stimulated NK cells

Genes up-regulated with 24h IL-2 + IL-12 + IL-18

Un-annotated transcripts removed. Data filtered based on a “present” call and hybridization signal of >500 after cytokine stimulation (T24hr). Duplicate probe sets excluded.

• 541 genes up-regulated 3-fold or more • 305 genes up-regulated 4-fold or more

Fold Change Gene Symbol Gene description 2950.10 LTA lymphotoxin alpha (TNF superfamily, member 1) 2529.17 CSF2 colony stimulating factor 2 (granulocyte-macrophage) 918.23 IL26 interleukin 26 852.71 EBI3 Epstein-Barr virus induced gene 3 563.02 ZBTB32 zinc finger and BTB domain containing 32 436.34 GPR84 G protein-coupled receptor 84 399.46 IL2RA interleukin 2 receptor, alpha 351.10 ETV5 ets variant gene 5 (ets-related molecule) 298.32 MFSD2 major facilitator superfamily domain containing 2 289.36 TNFRSF9 tumor necrosis factor receptor superfamily, member 9 277.55 ITGA1 integrin, alpha 1 252.64 PTGER3 prostaglandin E receptor 3 (subtype EP3) 211.41 STEAP1 six transmembrane epithelial antigen of the prostate 1 203.36 ZNRF1 zinc and ring finger 1 175.35 SLCO4A1 solute carrier organic anion transporter family, member 4A1 170.62 ADM adrenomedullin 167.31 HK2 hexokinase 2 143.31 CPXM carboxypeptidase X (M14 family) 115.95 DPP4 dipeptidylpeptidase 4 (CD26) 85.42 IL1A interleukin 1, alpha 77.37 TNFSF4 tumor necrosis factor (ligand) superfamily, member 4 76.15 NETO2 neuropilin (NRP) and tolloid (TLL)-like 2 74.33 LAG3 lymphocyte-activation gene 3 64.95 P2RX5 purinergic receptor P2X, ligand-gated ion channel, 5 64.87 LGALS3 lectin, galactoside-binding, soluble, 3 (galectin 3) 63.99 TNF tumor necrosis factor (TNF superfamily, member 2) 62.85 RGS16 regulator of G-protein signalling 16 62.61 CCR5 chemokine (C-C motif) receptor 5 59.19 MYC v-myc myelocytomatosis viral oncogene homolog (avian) 57.60 EGR2 early growth response 2 (Krox-20 homolog, Drosophila) 57.20 SLC27A2 solute carrier family 27 (fatty acid transporter), member 2 55.64 ICOS inducible T-cell co-stimulator 55.10 FREQ frequenin homolog (Drosophila) 54.57 LIMK2 LIM domain kinase 2 186

52.00 SOD2 superoxide dismutase 2, mitochondrial 51.07 MYB v-myb myeloblastosis viral oncogene homolog (avian) 49.18 CHAC1 ChaC, cation transport regulator-like 1 (E. coli) 48.95 PHLDA1 pleckstrin homology-like domain, family A, member 1 47.73 IFNG interferon, gamma 47.00 PSAT1 phosphoserine aminotransferase 1 45.41 CXCL10 chemokine (C-X-C motif) ligand 10 43.26 AMID (AIF)-like mitochondrion-associated inducer of death 42.30 CCR1 chemokine (C-C motif) receptor 1 38.88 PDCD1L1 programmed cell death 1 ligand 1 38.43 CCND2 cyclin D2 37.64 CYP1B1 cytochrome P450, family 1, subfamily B, polypeptide 1 37.63 NDFIP2 Nedd4 family interacting protein 2 37.45 MTHFD1L methylenetetrahydrofolate dehydrogenase 1-like 35.50 IER3 immediate early response 3 33.01 BCAR3 breast cancer anti-estrogen resistance 3 32.25 SLAMF1 signaling lymphocytic activation molecule family member 1 30.91 DUSP4 dual specificity phosphatase 4 30.50 CDC6 CDC6 cell division cycle 6 homolog (S. cerevisiae) 29.08 HBEGF heparin-binding EGF-like growth factor 28.99 HAS2 hyaluronan synthase 2 28.82 CD274 CD274 antigen 27.44 CLIC4 chloride intracellular channel 4 26.66 HAPLN3 hyaluronan and proteoglycan link protein 3 25.29 CXCL16 chemokine (C-X-C motif) ligand 16 25.27 TNFSF10 tumor necrosis factor (ligand) superfamily, member 10 25.01 TSPAN33 tetraspanin 33 24.40 EPS8 epidermal growth factor receptor pathway substrate 8 24.30 ADAM19 ADAM metallopeptidase domain 19 (meltrin beta) 22.68 NFE2L3 nuclear factor (erythroid-derived 2)-like 3 21.85 CISH cytokine inducible SH2-containing protein 21.58 ICAM1 intercellular adhesion molecule 1 (CD54) 21.58 TNFRSF18 tumor necrosis factor receptor superfamily, member 18 20.44 DKFZP434B0335 DKFZP434B0335 protein 20.22 TNFRSF25 tumor necrosis factor receptor superfamily, member 25 19.72 PBEF1 pre-B-cell colony enhancing factor 1 19.08 BATF basic leucine zipper transcription factor, ATF-like 18.76 TRIP10 thyroid hormone receptor interactor 10 18.65 SLC7A1 solute carrier family 7), member 1 18.48 CD83 CD83 antigen 18.43 SLC31A1 solute carrier family 31 (copper transporters), member 1 18.33 ST8SIA4 ST8 alpha-N-acetyl-neuraminide alpha-2,8-sialyltransferase 4 18.15 CDKN1A cyclin-dependent kinase inhibitor 1A (p21, Cip1) 18.11 SLC1A5 solute carrier family 1, member 5 17.91 PPIF peptidylprolyl isomerase F (cyclophilin F) 17.89 CHEK1 CHK1 checkpoint homolog (S. pombe) 17.32 APOARGC apolipoprotein (a) related gene C 17.24 MAR3 membrane-associated ring finger (C3HC4) 3 16.75 GBP1 guanylate binding protein 1, interferon-inducible, 67kDa 187

16.73 STX11 syntaxin 11 16.61 SHMT2 serine hydroxymethyltransferase 2 (mitochondrial) 16.51 PAICS phosphoribosylaminoimidazole succinocarboxamide synthetase 16.36 NFKB2 nuclear factor of kappa light enhancer in B-cells 2 (p49/p100) 15.69 SNCA synuclein, alpha (non A4 component of amyloid precursor) 15.61 WARS tryptophanyl-tRNA synthetase 15.47 EBI2 Epstein-Barr virus induced gene 2 14.99 BCL2A1 BCL2-related protein A1 14.93 PBEF pre-B-cell colony-enhancing factor 14.92 CSDA cold shock domain protein A 14.39 UBE2T ubiquitin-conjugating enzyme E2T (putative) 14.30 TRPV1 transient receptor potential cation channel, subfamily V, 1 14.07 SAMSN1 SAM domain, SH3 domain and nuclear localisation signals, 1 13.98 KLRA1 killer cell lectin-like receptor subfamily A, member 1 13.95 MTHFD2 methylene tetrahydrofolate 13.73 E2F1 E2F transcription factor 1 13.37 CDK6 cyclin-dependent kinase 6 13.21 MT1E metallothionein 1E (functional) 13.15 SOCS1 suppressor of cytokine signaling 1 12.82 SEH1L SEH1-like (S. cerevisiae) 12.78 SART2 squamous cell carcinoma antigen recognized by T cells 2 12.55 MDFIC MyoD family inhibitor domain containing 12.50 CDT1 DNA replication factor 12.35 DUSP5 dual specificity phosphatase 5 12.31 UHRF1 ubiquitin-like, containing PHD and RING finger domains, 1 12.13 TRAF1 TNF receptor-associated factor 1 12.04 FRMD4B FERM domain containing 4B 11.96 PKM2 pyruvate kinase, muscle 11.58 NFKBIZ nuclear factor of kappa light enhancer in B-cells inhibitor, zeta 11.56 STK3 serine/threonine kinase 3 (STE20 homolog, yeast) 11.43 VEGF vascular endothelial growth factor 11.41 GPR171 G protein-coupled receptor 171 11.36 IRAK2 interleukin-1 receptor-associated kinase 2 11.20 LIG4 ligase IV, DNA, ATP-dependent 11.19 NME1 non-metastatic cells 1, protein (NM23A) expressed in 11.11 IL12RB2 interleukin 12 receptor, beta 2 11.09 SLAMF7 SLAM family member 7 11.05 CD48 CD48 antigen (B-cell membrane protein) 10.92 DTL denticleless homolog (Drosophila) 10.83 SLC43A3 solute carrier family 43, member 3 10.68 BYSL bystin-like 10.66 PIM2 pim-2 oncogene 10.64 SLC16A1 solute carrier family 1, member 1 10.58 CMAH cytidine monophosphate-N-acetylneuraminic acid hydroxylase 10.54 TANK TRAF family member-associated NFKB activator 10.29 RAB6IP1 RAB6 interacting protein 1 10.22 MYO1B myosin IB 10.21 RPP40 ribonuclease P 40kDa subunit 10.18 NFE2L1 nuclear factor (erythroid-derived 2)-like 1 188

10.15 CHAC2 ChaC, cation transport regulator-like 2 (E. coli) 10.03 FAS Fas (TNF receptor superfamily, member 6) 9.83 STAG3 stromal antigen 3 9.79 MIRN21 microRNA 21 9.61 RAB33A RAB33A, member RAS oncogene family 9.59 GPR34 G protein-coupled receptor 34 9.54 TWSG1 twisted gastrulation homolog 1 (Drosophila) 9.47 PIGV phosphatidylinositol glycan, class V 9.41 PTTG1 pituitary tumor-transforming 1 9.20 TMEM49 transmembrane protein 49 9.17 ADORA2A adenosine A2a receptor 8.97 ABTB2 ankyrin repeat and BTB (POZ) domain containing 2 8.96 GMPPB GDP-mannose pyrophosphorylase B 8.76 SHCBP1 SHC SH2-domain binding protein 1 8.70 IRF8 interferon regulatory factor 8 8.52 SLC35F2 solute carrier family 35, member F2 8.45 IL15RA interleukin 15 receptor, alpha 8.43 SDC4 syndecan 4 (amphiglycan, ryudocan) 8.43 NFKB1 nuclear factor of kappa light enhancer in B-cells 1 (p105) 8.35 JMJD3 jumonji domain containing 3 8.31 NUPL1 nucleoporin like 1 8.19 STIP1 stress-induced-phosphoprotein 1 (Hsp70/90-organizing protein) 8.18 SLC7A5 solute carrier family 7, member 5 8.15 CBX5 chromobox homolog 5 (HP1 alpha homolog, Drosophila) 8.07 TNFRSF4 tumor necrosis factor receptor superfamily, member 4 8.06 FNBP2 formin binding protein 2 8.06 TAX1BP3 Tax1 (human T-cell leukemia virus type I) binding protein 3 7.99 PHGDH phosphoglycerate dehydrogenase 7.97 FASLG Fas ligand (TNF superfamily, member 6) 7.92 FABP5 fatty acid binding protein 5 (psoriasis-associated) 7.87 SRPRB signal recognition particle receptor, B subunit 7.80 CSE1L CSE1 chromosome segregation 1-like (yeast) 7.75 PRDX3 peroxiredoxin 3 7.66 PIM1 pim-1 oncogene 7.59 BCOR BCL6 co-repressor 7.57 GADD45B growth arrest and DNA-damage-inducible, beta 7.44 TARS threonyl-tRNA synthetase 7.37 METTL1 methyltransferase like 1 7.32 POLR3D polymerase (RNA) III (DNA directed) polypeptide D, 44kDa 7.28 BCL2L1 BCL2-like 1 7.26 TRIB3 tribbles homolog 3 (Drosophila) 7.17 GPR174 G protein-coupled receptor 174 7.15 LACTB lactamase, beta 7.09 RAB10 RAB10, member RAS oncogene family 7.06 EXOSC3 exosome component 3 7.03 GART phosphoribosylaminoimidazole synthetase 7.02 SQLE squalene epoxidase 6.99 SKB1 SKB1 homolog (S. pombe) 6.98 TAP1 transporter 1, ATP-binding cassette, sub-family B (MDR/TAP) 189

6.98 FLOT1 flotillin 1 6.97 SYNCRIP synaptotagmin binding, cytoplasmic RNA interacting protein 6.96 HIVEP2 human immunodeficiency virus type I enhancer binding protein 2 6.89 SMS spermine synthase 6.88 LMAN1 lectin, mannose-binding, 1 6.85 SPHK1 sphingosine kinase 1 6.83 PPIL1 peptidylprolyl isomerase (cyclophilin)-like 1 6.83 NUDCD1 NudC domain containing 1 6.83 CDK4 cyclin-dependent kinase 4 6.77 SNAPC1 small nuclear RNA activating complex, polypeptide 1, 43kDa 6.72 SEC24D SEC24 related gene family, member D (S. cerevisiae) 6.69 HSPA1A heat shock 70kDa protein 1A 6.65 XCL1 chemokine (C motif) ligand 1 6.64 NANS N-acetylneuraminic acid synthase (sialic acid synthase) 6.62 GFPT1 glutamine-fructose-6-phosphate transaminase 1 6.61 MOSPD2 motile sperm domain containing 2 6.59 TYMS thymidylate synthetase 6.50 PLAGL1 pleiomorphic adenoma gene-like 1 6.49 SAR1B SAR1 gene homolog B (S. cerevisiae) 6.48 KLHL6 kelch-like 6 (Drosophila) 6.48 DLEU2 deleted in lymphocytic leukemia, 2 6.46 CCDC6 coiled-coil domain containing 6 6.45 ECGF1 endothelial cell growth factor 1 (platelet-derived) 6.36 GPR18 G protein-coupled receptor 18 6.32 FH fumarate hydratase 6.28 SLC39A14 solute carrier family 39 (zinc transporter), member 14 6.27 PGD phosphogluconate dehydrogenase 6.26 TFRC transferrin receptor (p90, CD71) 6.23 VASP vasodilator-stimulated phosphoprotein 6.23 ACVR1B activin A receptor, type IB 6.22 SFXN4 sideroflexin 4 6.19 AK2 adenylate kinase 2 6.18 KCNA3 potassium voltage-gated channel, shaker-related subfamily, 3 6.16 WDR12 WD repeat domain 12 6.05 MT1H metallothionein 1H 5.97 HM13 histocompatibility (minor) 13 5.95 APMCF1 APMCF1 protein 5.93 LARP4 La ribonucleoprotein domain family, member 4 5.90 SUV39H2 suppressor of variegation 3-9 homolog 2 (Drosophila) 5.90 STAT5A signal transducer and activator of transcription 5A 5.90 MT1G metallothionein 1G 5.89 PPAN peter pan homolog (Drosophila) 5.88 TUBB tubulin, beta 5.80 BET1 BET1 homolog (S. cerevisiae) 5.79 RBBP8 retinoblastoma binding protein 8 5.79 HAVCR2 hepatitis A virus cellular receptor 2 5.78 E2F6 E2F transcription factor 6 5.76 GADD45G growth arrest and DNA-damage-inducible, gamma 5.76 GOT1 glutamic-oxaloacetic transaminase 1, soluble 190

5.75 PFKFB3 6-phosphofructo-2-kinase/fructose-2,6-biphosphatase 3 5.72 FAM89A family with sequence similarity 89, member A 5.71 SFXN1 sideroflexin 1 5.71 TUBG1 tubulin, gamma 1 5.61 PDIA5 protein disulfide isomerase family A, member 5 5.57 LTB lymphotoxin beta (TNF superfamily, member 3) 5.52 SNX10 sorting nexin 10 5.50 WDR4 WD repeat domain 4 5.49 IBRDC3 IBR domain containing 3 5.44 MCM4 MCM4 minichromosome maintenance deficient 4 (S. cerevisiae) 5.37 PPP1R15A protein phosphatase 1, regulatory (inhibitor) subunit 15A 5.37 TM6SF1 transmembrane 6 superfamily member 1 5.35 TIMM10 translocase of inner mitochondrial membrane 10 homolog (yeast) 5.33 PNPT1 polyribonucleotide nucleotidyltransferase 1 5.31 CREM cAMP responsive element modulator 5.26 SERPINB8 serpin peptidase inhibitor, clade B (ovalbumin), member 8 5.24 DAPK1 death-associated protein kinase 1 5.24 WARS2 tryptophanyl tRNA synthetase 2 (mitochondrial) 5.24 IDH3A isocitrate dehydrogenase 3 (NAD+) alpha 5.18 ETS2 v-ets erythroblastosis virus E26 oncogene homolog 2 (avian) 5.17 EIF2S1 eukaryotic translation initiation factor 2, subunit 1 alpha, 35kDa 5.17 JARID1B Jumonji, AT rich interactive domain 1B (RBP2-like) 5.17 ETNK1 ethanolamine kinase 1 5.16 ACSL1 acyl-CoA synthetase long-chain family member 1 5.13 BCCIP BRCA2 and CDKN1A interacting protein 5.13 MCM6 MCM6 minichromosome maintenance deficient 6 (MIS5 homolog) 5.12 HTPAP HTPAP protein 5.09 LMNB2 lamin B2 5.04 KIF3B kinesin family member 3B 5.03 HSPD1 heat shock 60kDa protein 1 (chaperonin) 5.03 BIRC3 baculoviral IAP repeat-containing 3 5.02 NAB1 NGFI-A binding protein 1 (EGR1 binding protein 1) 5.02 IARS isoleucine-tRNA synthetase 5.02 AARS alanyl-tRNA synthetase 5.02 ABHD2 abhydrolase domain containing 2 4.97 NDUFAF1 NADH dehydrogenase 1 alpha subcomplex, assembly factor 1 4.95 TMEM64 transmembrane protein 64 4.95 PSMB5 proteasome (prosome, macropain) subunit, beta type, 5 4.95 CYB5 cytochrome b-5 4.91 BXDC1 brix domain containing 1 4.90 SOCS3 suppressor of cytokine signaling 3 4.90 MT1X metallothionein 1X 4.86 SPRED2 sprouty-related, EVH1 domain containing 2 4.86 CSTF2 cleavage stimulation factor, 3' pre-RNA, subunit 2, 64kDa 4.84 EED embryonic ectoderm development 4.81 FTH1 ferritin, heavy polypeptide 1 4.79 PELO pelota homolog (Drosophila) 4.79 IL4R interleukin 4 receptor 4.74 EIF2S2 eukaryotic translation initiation factor 2, subunit 2 beta, 38kDa 191

4.73 RALA v-ral simian leukemia viral oncogene homolog A (ras related) 4.72 CD40 CD40 antigen (TNF receptor superfamily member 5) 4.72 MCM2 minichromosome maintenance deficient 2, mitotin (S. cerevisiae) 4.69 CHCHD4 coiled-coIL-helix-coiled-coIL-helix domain containing 4 4.69 ARF4 ADP-ribosylation factor 4 4.68 HUWE1 HECT, UBA and WWE domain containing 1 4.62 CETN3 centrin, EF-hand protein, 3 (CDC31 homolog, yeast) 4.60 IVNS1ABP influenza virus NS1A binding protein 4.52 RQCD1 RCD1 required for cell differentiation1 homolog (S. pombe) 4.51 RRS1 RRS1 ribosome biogenesis regulator homolog (S. cerevisiae) 4.47 JTV1 JTV1 gene 4.46 TICAM2 toll-like receptor adaptor molecule 2 4.45 FARSLA phenylalanine-tRNA synthetase-like, alpha subunit 4.43 YIPF6 Yip1 domain family, member 6 4.43 LYCAT lysocardiolipin acyltransferase 4.42 POLE2 polymerase (DNA directed), epsilon 2 (p59 subunit) 4.41 TNIP2 TNFAIP3 interacting protein 2 4.39 DDX21 DEAD (Asp-Glu-Ala-Asp) box polypeptide 21 4.36 MT1F metallothionein 1F (functional) 4.35 TXN thioredoxin 4.31 GIPR gastric inhibitory polypeptide receptor 4.28 DKC1 dyskeratosis congenita 1, dyskerin 4.22 SEC11L3 SEC11-like 3 (S. cerevisiae) 4.21 MARS methionine-tRNA synthetase 4.17 MLLT4 mixed-lineage leukemia (trithorax homolog); translocated to, 4 4.04 TIPRL TIP41, TOR signalling pathway regulator-like (S. cerevisiae)

192

Genes down-regulated with 24h IL-2 + IL-12 + IL-18

Un-annotated transcripts removed. Data filtered based on a “present” call and hybridization signal of >500 before cytokine stimulation (T0hr). Duplicate probe sets excluded.

• 609 genes down-regulated 3-fold or more • 383 genes down-regulated 4-fold or more Fold Gene Change Symbol Gene description -391.76 ZBP1 Z-DNA binding protein 1 -362.98 HBA1 hemoglobin, alpha 1 -282.01 MGAM maltase-glucoamylase (alpha-glucosidase) -251.21 DUSP1 dual specificity phosphatase 1 -192.18 SNAI3 snail homolog 3 (Drosophila) -185.22 LDB2 LIM domain binding 2 -168.79 EDG8 endothelial differentiation, sphingolipid G-protein-coupled receptor, 8 -154.25 ENPP5 ectonucleotide pyrophosphatase/phosphodiesterase 5 (putative function) -122.01 DPEP2 dipeptidase 2 -106.26 HBA2 hemoglobin, alpha 2 -104.43 AREG amphiregulin (schwannoma-derived growth factor) -101.19 PLXND1 plexin D1 -95.36 SPON2 spondin 2, extracellular matrix protein -92.98 ADAM28 ADAM metallopeptidase domain 28 -87.98 KRTHB6 keratin, hair, basic, 6 (monilethrix) -77.54 MYOM2 myomesin (M-protein) 2, 165kDa -73.98 TLE1 transducin-like enhancer of split 1 (E(sp1) homolog, Drosophila) -66.44 CAPN12 calpain 12 -65.74 FZD4 frizzled homolog 4 (Drosophila) -65.62 TSC22D3 TSC22 domain family, member 3 -65.34 TKTL1 transketolase-like 1 -65.08 ZFP36L2 zinc finger protein 36, C3H type-like 2 -63.44 RAB37 RAB37, member RAS oncogene family -63.03 CD52 CD52 antigen (CAMPATH-1 antigen) -62.21 ENC1 ectodermal-neural cortex (with BTB-like domain) -62.20 SCML4 sex comb on midleg-like 4 (Drosophila) -61.34 FBXO32 F-box protein 32 -61.26 FAIM3 Fas apoptotic inhibitory molecule 3 -57.20 AHNAK AHNAK nucleoprotein (desmoyokin) -54.85 RASGRP2 RAS guanyl releasing protein 2 (calcium and DAG-regulated) -49.09 CXCR4 chemokine (C-X-C motif) receptor 4 -47.85 PER1 period homolog 1 (Drosophila) -43.91 PRSS23 protease, serine, 23 -40.62 SIGLECP3 sialic acid binding Ig-like lectin, pseudogene 3 -40.50 NR4A2 nuclear receptor subfamily 4, group A, member 2 -37.39 TSPAN32 tetraspanin 32 -37.04 SPTB spectrin, beta, erythrocytic (includes spherocytosis, clinical type I) 193

-37.04 HHEX hematopoietically expressed homeobox -36.77 TXNDC3 thioredoxin domain containing 3 (spermatozoa) -36.38 TMCC3 transmembrane and coiled-coil domain family 3 -32.32 ZFP36 zinc finger protein 36, C3H type, homolog (mouse) -30.76 SORL1 sortilin-related receptor, L(DLR class) A repeats-containing -27.99 SYNGR1 synaptogyrin 1 -27.87 ABLIM1 actin binding LIM protein 1 -27.00 ARHGAP12 Rho GTPase activating protein 12 -26.49 KLF2 Kruppel-like factor 2 (lung) -26.33 GAS7 growth arrest-specific 7 -26.24 CD160 CD160 antigen -25.97 MTAC2D1 membrane targeting (tandem) C2 domain containing 1 -25.60 NEIL1 nei endonuclease VIII-like 1 (E. coli) -24.97 RASA3 RAS p21 protein activator 3 -24.90 IL8RB interleukin 8 receptor, beta -24.74 SLC1A7 solute carrier family 1 (glutamate transporter), member 7 -24.62 MTSS1 metastasis suppressor 1 -24.11 TMEM71 transmembrane protein 71 -24.05 RNPC1 RNA-binding region (RNP1, RRM) containing 1 -23.98 PDCD4 programmed cell death 4 (neoplastic transformation inhibitor) -23.89 TMEM2 transmembrane protein 2 -23.41 SGSH N-sulfoglucosamine sulfohydrolase (sulfamidase) -22.61 FAM13A1 family with sequence similarity 13, member A1 -22.58 PDZK4 PDZ domain containing 4 -22.50 GCHFR GTP cyclohydrolase I feedback regulator -22.34 FAM62B Family with sequence similarity 62 (C2 domain containing) member B -21.41 SPOCK2 sparc/osteonectin, cwcv and kazal-like domains proteoglycan 2 -21.00 DLL1 delta-like 1 (Drosophila) -20.64 YPEL1 yippee-like 1 (Drosophila) -20.26 MMP23B matrix metallopeptidase 23B -20.11 C11orf32 sortilin-related receptor, L(DLR class) A repeats-containing * -19.86 MXI1 MAX interactor 1 -19.56 CENTG2 centaurin, gamma 2 -19.55 IRS2 insulin receptor substrate 2 -19.42 SIGIRR single immunoglobulin and toll-interleukin 1 receptor (TIR) domain -19.31 SBK1 SH3-binding domain kinase 1 -19.30 SSX2IP synovial sarcoma, X breakpoint 2 interacting protein -18.87 SNF1LK SNF1-like kinase -18.21 SHFM3P1 split hand/foot malformation (ectrodactyly) type 3 pseudogene 1 -17.94 ISG20 interferon stimulated exonuclease gene 20kDa -17.14 FRAT1 frequently rearranged in advanced T-cell lymphomas -17.02 FOS v-fos FBJ murine osteosarcoma viral oncogene homolog -16.86 MXD4 MAX dimerization protein 4 -16.80 KIR2DS5 killer cell immunoglobulin-like receptor, two domains, short tail, 5 -16.79 MLC1 megalencephalic leukoencephalopathy with subcortical cysts 1 -16.78 CLIC3 chloride intracellular channel 3 -16.68 KIR2DL1 killer cell immunoglobulin-like receptor, two domains, long tail, 1 -16.58 LRMP lymphoid-restricted membrane protein -16.48 CD300A CD300A antigen 194

-16.43 GARNL4 GTPase activating Rap/RanGAP domain-like 4 -16.33 SELPLG selectin P ligand -16.08 GPR56 G protein-coupled receptor 56 -15.70 AIM1 absent in melanoma 1 -15.33 KIR3DL3 killer cell immunoglobulin-like receptor, three domains, long tail, 3 -15.32 RNF166 ring finger protein 166 -15.20 AK5 adenylate kinase 5 -15.16 SYNE1 spectrin repeat containing, nuclear envelope 1 -14.96 ZBTB16 zinc finger and BTB domain containing 16 -14.90 MIR myosin regulatory light chain interacting protein -14.52 RAB27B RAB27B, member RAS oncogene family -14.49 CECR1 cat eye syndrome chromosome region, candidate 1 -14.35 PTGDS prostaglandin D2 synthase 21kDa (brain) -14.32 KLHL24 kelch-like 24 (Drosophila) -14.22 GZMM granzyme M (lymphocyte met-ase 1) -14.02 SLA2 Src-like-adaptor 2 -13.92 TMC8 transmembrane channel-like 8 -13.88 SPN sialophorin (gpL115, leukosialin, CD43) -13.86 KIR3DL1 killer cell immunoglobulin-like receptor, three domains, long tail, 1 -13.75 CXXC5 CXXC finger 5 -13.62 APBA2 amyloid beta (A4) precursor protein-binding, family A, member 2 -13.57 RGS2 regulator of G-protein signalling 2, 24kDa -13.52 RNF125 ring finger protein 125 -13.51 CHST12 carbohydrate (chondroitin 4) sulfotransferase 12 -13.44 KIR2DS4 killer cell immunoglobulin-like receptor, two domains, short tail, 4 -13.43 C3AR1 complement component 3a receptor 1 -13.33 STMN3 stathmin-like 3 -13.21 WHDC1L1 WAS protein homology region 2 domain containing 1-like 1 -13.08 AGPAT4 1-acylglycerol-3-phosphate O-acyltransferase 4 -13.07 YPEL2 yippee-like 2 (Drosophila) -13.05 BIN1 bridging integrator 1 -12.80 SFXN3 sideroflexin 3 -12.77 NR1D2 nuclear receptor subfamily 1, group D, member 2 -12.63 PDGFD platelet derived growth factor D -12.41 CNNM3 cyclin M3 -12.40 SOX4 SRY (sex determining region Y)-box 4 -12.39 ADCY7 adenylate cyclase 7 -12.08 SIGLEC7 sialic acid binding Ig-like lectin 7 -11.98 GSTM1 glutathione S-transferase M1 -11.97 TSEN54 tRNA splicing endonuclease 54 homolog (SEN54, S. cerevisiae) -11.80 PLEKHG3 pleckstrin homology domain containing, family G member 3 -11.71 BNC2 basonuclin 2 -11.71 SEPT6 septin 6 -11.70 AMY2B amylase, alpha 2B; pancreatic -11.65 FOSL2 FOS-like antigen 2 -11.65 PSCD1 pleckstrin homology, Sec7 and coiled-coil domains 1(cytohesin 1) -11.62 ATP8A1 ATPase, aminophospholipid transporter (APLT), type 8A, member 1 -11.49 CDKN2D cyclin-dependent kinase inhibitor 2D (p19, inhibits CDK4) -11.46 PTGDR prostaglandin D2 receptor (DP) 195

-11.37 TBC1D10C TBC1 domain family, member 10C -11.36 ZNF331 zinc finger protein 331 -11.34 PECAM1 platelet/endothelial cell adhesion molecule (CD31 antigen) -11.32 SH2D1B SH2 domain containing 1B -11.31 CD7 CD7 antigen (p41) -11.31 ATM ataxia telangiectasia mutated -11.26 NR4A3 nuclear receptor subfamily 4, group A, member 3 -11.20 CYBA cytochrome b-245, alpha polypeptide -11.19 PBXIP1 pre-B-cell leukemia transcription factor interacting protein 1 -11.16 ZAP70 zeta-chain (TCR) associated protein kinase 70kDa -11.11 MGAT4A mannosyl (alpha-1,3-) beta-1,4-N-acetylglucosaminyltransferase A -11.10 DHRS1 dehydrogenase/reductase (SDR family) member 1 -11.08 DAB2 disabled homolog 2, mitogen-responsive phosphoprotein (Drosophila) -11.00 ATXN7L1 ataxin 7-like 1 -11.00 BNIP3L BCL2/adenovirus E1B 19kDa interacting protein 3-like -10.88 UBE1L ubiquitin-activating enzyme E1-like -10.80 TSC22D1 TSC22 domain family, member 1 -10.80 AVPI1 arginine vasopressin-induced 1 -10.75 MYO1F myosin IF -10.74 LY9 lymphocyte antigen 9 -10.65 KIR3DL2 killer cell immunoglobulin-like receptor, three domains, long tail, 2 -10.47 GSN gelsolin (amyloidosis, Finnish type) -10.40 BCL11B B-cell CLL/lymphoma 11B (zinc finger protein) -10.20 RNF103 ring finger protein 103 -10.07 EVER2 epidermodysplasia verruciformis 2 -9.93 ERBB2 v-erb-b2 erythroblastic leukemia viral oncogene homolog 2, (avian) -9.89 SFRS5 splicing factor, arginine/serine-rich 5 -9.87 TSPAN2 tetraspanin 2 -9.80 COL6A2 collagen, type VI, alpha 2 -9.75 ABCB1 ATP-binding cassette, sub-family B (MDR/TAP), member 1 * -9.75 AUTS2 autism susceptibility candidate 2 -9.73 PTPN22 protein tyrosine phosphatase, non-receptor type 22 (lymphoid) -9.73 MPPE1 metallo phosphoesterase -9.66 DUSP2 dual specificity phosphatase 2 -9.63 MYLIP myosin regulatory light chain interacting protein -9.59 KLHL3 kelch-like 3 (Drosophila) -9.58 SNTB2 syntrophin, beta 2 (dystrophin-associated protein A1, basic component 2) -9.55 KIR2DL3 killer cell immunoglobulin-like receptor, two domains, long tail, 3 -9.50 CIDEB cell death-inducing DFFA-like effector b -9.46 BTBD15 BTB (POZ) domain containing 15 -9.44 LDLRAP1 low density lipoprotein receptor adaptor protein 1 -9.43 FAM8A1 family with sequence similarity 8, member A1 -9.41 LLGL2 lethal giant larvae homolog 2 (Drosophila) -9.34 PELI2 pellino homolog 2 (Drosophila) -9.11 OSBPL5 oxysterol binding protein-like 5 -9.08 IL11RA interleukin 11 receptor, alpha -9.08 PITPNM2 phosphatidylinositol transfer protein, membrane-associated 2 -9.07 MLLT3 mixed-lineage leukemia (trithorax homolog); translocated to, 3 -9.04 FBXL3A F-box and leucine-rich repeat protein 3A 196

-9.02 RAB11FIP4 RAB11 family interacting protein 4 (class II) -9.01 CASP8 caspase 8, apoptosis-related cysteine peptidase -8.94 FOXO1A forkhead box O1A (rhabdomyosarcoma) -8.91 KIR2DL4 killer cell ig-like receptor, two domains, long cytoplasmic tail, 4 -8.86 KLF12 Kruppel-like factor 12 -8.82 F2R coagulation factor II (thrombin) receptor -8.79 AKAP2 A kinase (PRKA) anchor protein 2 -8.72 AHSA2 AHA1, activator of heat shock 90kDa protein ATPase homolog 2 -8.69 KIR2DS1 killer cell ig-like receptor, two domains, short cytoplasmic tail, 1 -8.68 TMEM59 transmembrane protein 59 * -8.58 IFITM1 interferon induced transmembrane protein 1 (9-27) -8.42 RNASET2 ribonuclease T2 -8.40 KIR2DL5B killer cell ig-like receptor, two domains, long cytoplasmic tail, 5B -8.39 NMT2 N-myristoyltransferase 2 -8.34 ADHFE1 alcohol dehydrogenase, iron containing, 1 -8.34 GALNT12 UDP- GalNAc-:polypeptide N-acetylgalactosaminyltransferase 12 -8.31 ZNF238 zinc finger protein 238 -8.28 IFITM2 interferon induced transmembrane protein 2 (1-8D) -8.26 SLCO3A1 solute carrier organic anion transporter family, member 3A1 -8.24 NUAK2 NUAK family, SNF1-like kinase, 2 -8.14 CTSF cathepsin F -8.11 ZNF198 zinc finger protein 198 -8.05 SAMD3 sterile alpha motif domain containing 3 -8.05 PAPD5 PAP associated domain containing 5 -8.04 VPS26B vacuolar protein sorting 26 homolog B (yeast) -8.00 RCSD1 RCSD domain containing 1 -7.97 SEPT9 septin 9 -7.89 BZRAP1 benzodiazapine receptor (peripheral) associated protein 1 -7.89 PTPRN2 protein tyrosine phosphatase, receptor type, N polypeptide 2 -7.87 PLEKHA1 pleckstrin homology domain containing, family A member 1 -7.85 TRIB2 tribbles homolog 2 (Drosophila) -7.77 ABTB1 ankyrin repeat and BTB (POZ) domain containing 1 -7.74 SCAND1 SCAN domain containing 1 -7.72 ARHGAP8 Rho GTPase activating protein 8 -7.68 P2RY8 purinergic receptor P2Y, G-protein coupled, 8 -7.65 PSCD4 pleckstrin homology, Sec7 and coiled-coil domains 4 -7.63 YPEL3 yippee-like 3 (Drosophila) -7.62 WASPIP Wiskott-Aldrich syndrome protein interacting protein -7.61 FGL2 fibrinogen-like 2 -7.61 ITGAM integrin, alpha M (complement component receptor 3, alpha, CD11b) -7.58 ITGA4 integrin, alpha 4 (antigen CD49D, alpha 4 subunit of VLA-4 receptor) -7.56 SIPA1 signal-induced proliferation-associated gene 1 -7.56 SERTAD3 SERTA domain containing 3 -7.55 KLRG1 killer cell lectin-like receptor subfamily G, member 1 -7.52 KLRF1 killer cell lectin-like receptor subfamily F, member 1 -7.49 MIB2 mindbomb homolog 2 (Drosophila) -7.49 CEP78 centrosomal protein 78kDa -7.48 CACNA2D2 calcium channel, voltage-dependent, alpha 2/delta subunit 2 -7.42 ANXA1 annexin A1 197

-7.41 TRAF3IP3 TRAF3 interacting protein 3 -7.33 HSPC072 HSPC072 protein -7.31 OTUD1 OTU domain containing 1 -7.29 ASCL2 achaete-scute complex-like 2 (Drosophila) -7.29 ATHL1 ATH1, acid trehalase-like 1 (yeast) -7.15 IGF1R insulin-like growth factor 1 receptor -7.10 SEMA4C semaphorin 4C -7.08 RASGEF1A RasGEF domain family, member 1A -7.06 ZFP276 zinc finger protein 276 homolog (mouse) -7.02 CDK5R1 cyclin-dependent kinase 5, regulatory subunit 1 (p35) -7.01 BCR breakpoint cluster region -6.92 TP53INP1 tumor protein p53 inducible nuclear protein 1 -6.89 TGFB1 transforming growth factor, beta 1 (Camurati-Engelmann disease) -6.89 FLJ10244 Ral-A exchange factor RalGPS2 -6.87 NOD3 NOD3 protein -6.85 TYROBP TYRO protein tyrosine kinase binding protein -6.83 IL24 interleukin 24 -6.80 ARHGAP9 Rho GTPase activating protein 9 -6.77 MLL5 myeloid/lymphoid or mixed-lineage leukemia 5 (trithorax homolog) -6.76 PLEKHF1 pleckstrin homology domain containing, family F, member 1 -6.71 ELF2 E74-like factor 2 (ets domain transcription factor) -6.64 HLA-DPB1 major histocompatibility complex, class II, DP beta 1 -6.63 AXIN1 axin 1 -6.63 FYB FYN binding protein (FYB-120/130) -6.62 TAF9L TAF9-like RNA polymerase II, TBP-associated factor, 31kDa -6.60 GALT galactose-1-phosphate uridylyltransferase -6.60 WDFY1 WD repeat and FYVE domain containing 1 -6.59 AMT aminomethyltransferase (glycine cleavage system protein T) -6.58 DLG5 discs, large homolog 5 (Drosophila) -6.50 NRBP2 nuclear receptor binding protein 2 -6.46 SDCCAG33 serologically defined colon cancer antigen 33 -6.42 CENTB1 centaurin, beta 1 -6.42 CYBRD1 cytochrome b reductase 1 -6.42 PTPLAD2 protein tyrosine phosphatase-like A domain containing 2 -6.38 HBP1 HMG-box transcription factor 1 -6.37 TRIM74 tripartite motif-containing 50B or of tripartite motif-containing 74 * -6.33 SAMHD1 SAM domain and HD domain 1 -6.30 AK1 adenylate kinase 1 -6.30 LGR6 leucine-rich repeat-containing G protein-coupled receptor 6 -6.30 IFITM3 interferon induced transmembrane protein 3 (1-8U) -6.28 CALM1 calmodulin 1 (phosphorylase kinase, delta) -6.28 ADAMTS10 ADAM metallopeptidase with thrombospondin type 1 motif, 10 -6.27 FLOT2 flotillin 2 -6.26 PLCL2 phospholipase C-like 2 -6.26 EFHC1 EF-hand domain (C-terminal) containing 1 -6.23 CCL28 chemokine (C-C motif) ligand 28 -6.23 FCGR2B Fc fragment of IgG, low affinity IIb, receptor (CD32) -6.23 OSBPL7 oxysterol binding protein-like 7 -6.23 SMAD3 SMAD, mothers against DPP homolog 3 (Drosophila) 198

-6.19 SPTBN1 spectrin, beta, non-erythrocytic 1 -6.15 ACSS1 acyl-CoA synthetase short-chain family member 1 -6.15 PTPRE protein tyrosine phosphatase, receptor type, E -6.14 TIMP1 TIMP metallopeptidase inhibitor 1 -6.12 EVL Enah/Vasp-like -6.12 BTN3A1 butyrophilin, subfamily 3, member A1 -6.11 IFNGR1 interferon gamma receptor 1 -6.10 PLAC8 placenta-specific 8 -6.10 VAV3 vav 3 oncogene -6.06 IFIT2 interferon-induced protein with tetratricopeptide repeats 2 -6.04 BTG1 B-cell translocation gene 1, anti-proliferative -6.01 PPP2R5C protein phosphatase 2, regulatory subunit B (B56), gamma isoform -5.95 HIST1H1C histone 1, H1c -5.94 ADD3 adducin 3 (gamma) -5.94 LAT linker for activation of T cells -5.87 HSH2D hematopoietic SH2 domain containing -5.87 PLCB2 phospholipase C, beta 2 -5.82 SEC31L2 SEC31-like 2 (S. cerevisiae) -5.82 UNC84B unc-84 homolog B (C. elegans) -5.78 DTX3 deltex 3 homolog (Drosophila) -5.78 SLC4A4 solute carrier family 4, sodium bicarbonate cotransporter, member 4 -5.76 S100A4 S100 calcium binding protein A4 -5.76 TOB1 transducer of ERBB2, 1 -5.74 DGKQ diacylglycerol kinase, theta 110kDa -5.73 PLCD1 phospholipase C, delta 1 -5.65 PCL1 prenylcysteine lyase -5.64 FAM46A family with sequence similarity 46, member A -5.62 TTLL3 tubulin tyrosine ligase-like family, member 3 -5.56 DIP2 disco-interacting protein 2 (Drosophila) homolog -5.55 RKHD2 ring finger and KH domain containing 2 -5.55 PARP8 poly (ADP-ribose) polymerase family, member 8 -5.55 AGTPBP1 ATP/GTP binding protein 1 -5.54 CPT1A carnitine palmitoyltransferase 1A (liver) -5.52 LCK lymphocyte-specific protein tyrosine kinase -5.52 SERTAD2 SERTA domain containing 2 -5.50 IREB2 iron-responsive element binding protein 2 -5.47 LOC153222 adult retina protein -5.45 NAP1L5 nucleosome assembly protein 1-like 5 -5.45 CDC25B cell division cycle 25B -5.45 ZNF395 zinc finger protein 395 -5.45 LTB4R leukotriene B4 receptor -5.44 LGALS9 lectin, galactoside-binding, soluble, 9 (galectin 9) -5.44 CCNG2 cyclin G2 -5.43 LY6E lymphocyte antigen 6 complex, locus E -5.41 MLKL mixed lineage kinase domain-like -5.35 BCL10 B-cell CLL/lymphoma 10 -5.35 UBE2B ubiquitin-conjugating enzyme E2B (RAD6 homolog) * -5.34 MNAB membrane-associated nucleic acid binding protein -5.31 MAST3 microtubule associated serine/threonine kinase 3 199

-5.30 GNS glucosamine (N-acetyl)-6-sulfatase (Sanfilippo disease IIID) -5.26 RNF44 ring finger protein 44 -5.25 COL13A1 collagen, type XIII, alpha 1 -5.23 ANKRD9 ankyrin repeat domain 9 -5.22 PYHIN1 pyrin and HIN domain family, member 1 -5.20 CD47 CD47 antigen (Rh-related antigen, integrin-associated signal transducer) -5.19 ICAM3 intercellular adhesion molecule 3 -5.17 ALAD aminolevulinate, delta-, dehydratase -5.17 CARD11 caspase recruitment domain family, member 11 -5.16 SELL selectin L (lymphocyte adhesion molecule 1) -5.15 CAMK2G calcium/calmodulin-dependent protein kinase (CaM kinase) II gamma -5.15 KLF3 Kruppel-like factor 3 (basic) -5.14 DOK2 docking protein 2, 56kDa -5.10 CD8A CD8 antigen, alpha polypeptide (p32) -5.10 ADRB2 adrenergic, beta-2-, receptor, surface -5.07 HPCAL1 hippocalcin-like 1 -5.03 PYCARD PYD and CARD domain containing -5.01 HDAC4 histone deacetylase 4 -5.00 NUDT11 nudix (nucleoside diphosphate linked moiety X)-type motif 11 -4.99 KCNAB2 potassium voltage-gated channel, shaker-related subfamily, beta member 2 -4.96 SYTL1 synaptotagmin-like 1 -4.95 PRKACB protein kinase, cAMP-dependent, catalytic, beta -4.94 ITGB7 integrin, beta 7 -4.92 FRAT2 frequently rearranged in advanced T-cell lymphomas 2 -4.91 GFOD1 glucose-fructose oxidoreductase domain containing 1 -4.90 FAM13A1OS family with sequence similarity 13, member A1 opposite strand -4.89 INPP4A inositol polyphosphate-4-phosphatase, type I, 107kDa -4.88 CD6 CD6 antigen -4.82 RBL2 retinoblastoma-like 2 (p130) -4.81 LAPTM5 lysosomal associated multispanning membrane protein 5 -4.79 UNRIP unr-interacting protein -4.74 NALP1 NACHT, leucine rich repeat and PYD (pyrin domain) containing 1 -4.66 SPSB3 splA/ryanodine receptor domain and SOCS box containing 3 -4.63 PRKCBP1 protein kinase C binding protein 1 -4.61 PLA2G6 phospholipase A2, group VI (cytosolic, calcium-independent) -4.55 BOK BCL2-related ovarian killer -4.53 BRWD1 bromodomain and WD repeat domain containing 1 -4.46 MIDN midnolin -4.44 CABC1 chaperone, ABC1 activity of bc1 complex like (S. pombe) -4.40 TFDP2 transcription factor Dp-2 (E2F dimerization partner 2) -4.39 HSCARG HSCARG protein -4.38 GKAP1 G kinase anchoring protein 1 -4.33 DDIT4 DNA-damage-inducible transcript 4 -4.30 HSPBAP1 HSPB (heat shock 27kDa) associated protein 1 -4.29 EPC1 enhancer of polycomb homolog 1, (Drosophila) -4.27 ARRB1 arrestin, beta 1 -4.27 SSH2 slingshot homolog 2 (Drosophila) -4.26 ARL4C ADP-ribosylation factor-like 4C -4.26 MAPRE2 microtubule-associated protein, RP/EB family, member 2 200

-4.20 ZBTB1 zinc finger and BTB domain containing 1 -4.20 NDRG3 NDRG family member 3 -4.13 TNRC6B trinucleotide repeat containing 6B

201

Appendix II: Expression changes of ChIP-chip target genes

Genes on ChIP-chip list which increase with IL-2+IL-12+IL-18 stimulation (24h)

Must be called “present” after stimulation by Affymetrix algorithm Fold change greater than 1.3 in both donors Duplicate probe sets not removed

Gene ID Donor 1 Donor 2 206975_at LTA 4213.9 1686.3 206432_at HAS2 815.2 63.2 223383_at ZNRF1 39.2 367.5 231094_s_at MTHFD1L 38.7 26.7 230372_at HAS2 33.5 24.5 225520_at MTHFD1L 30.2 44.7 223382_s_at ZNRF1 28.6 33.5 225959_s_at ZNRF1 23.2 41.2 225960_at ZNRF1 14.0 19.5 225962_at ZNRF1 13.0 80.8 221331_x_at CTLA4 12.9 3.3 226056_at CDGAP 10.4 7.1 212561_at RAB6IP1 9.5 11.1 200759_x_at NFE2L1 8.2 12.2 214179_s_at NFE2L1 7.4 11.2 210538_s_at BIRC3 6.0 4.0 226057_at CDGAP 6.0 4.6 222981_s_at RAB10 6.0 8.2 236341_at CTLA4 5.7 5.4 216985_s_at STX3A 5.3 3.9 202748_at GBP2 5.1 3.2 221981_s_at WDR59 4.9 2.6 200758_s_at NFE2L1 4.9 6.3 208079_s_at STK6 (AURKA) 4.9 6.3 202557_at STCH 4.8 3.9 36564_at IBRDC3 4.7 6.3 213038_at IBRDC3 4.5 6.1 230499_at BIRC3 4.3 2.6 1553096_s_at BCL2L11 4.3 1.7 208536_s_at BCL2L11 4.1 1.3 216602_s_at FARSLA 4.1 4.8 204822_at TTK 4.1 4.8 202159_at FARSLA 4.0 4.3 1558143_a_at BCL2L11 3.9 1.7 242907_at GBP2 3.9 2.3 206536_s_at BIRC4 3.9 3.2 208985_s_at EIF3S1 3.7 4.3 209927_s_at C1orf77 3.7 3.6 52285_f_at CEP76 3.6 3.0 234362_s_at CTLA4 3.6 2.3 225831_at LUZP1 3.6 3.2 222343_at BCL2L11 3.6 1.7 1558173_a_at LUZP1 3.5 1.9 222509_s_at ZNF672 3.4 3.1 202

217364_x_at EIF3S1 3.3 2.7 222980_at RAB10 3.2 4.0 202558_s_at STCH 3.1 2.8 200050_at ZNF146 3.1 2.7 218810_at ZC3H12A 3.1 4.7 225267_at KPNA4 3.0 1.7 218979_at RMI1 3.0 3.5 212286_at ANKRD12 2.9 2.9 200935_at CALR 2.9 4.1 224775_at IWS1 2.9 2.0 218195_at C6orf211 2.8 2.5 223212_at ZDHHC16 2.8 4.0 212397_at RDX 2.8 3.1 224714_at MKI67IP 2.8 3.0 203372_s_at SOCS2 2.7 2.5 231794_at CTLA4 2.7 1.4 220153_at ENTPD7 2.7 3.0 201921_at GNG10 2.6 3.4 200039_s_at PSMB2 2.6 4.0 202941_at NDUFV2 2.5 3.3 200974_at ACTA2 2.5 3.1 228363_at BIRC4 2.5 1.9 221832_s_at KIAA0601 2.5 2.0 209653_at KPNA4 2.5 2.1 225859_at BIRC4 2.4 1.9 212250_at MTDH 2.4 3.0 1553224_at LUZP1 2.4 2.3 223397_s_at NIP7 2.4 2.5 224713_at MKI67IP 2.4 2.5 225682_s_at POLR3H 2.3 4.0 219311_at CEP76 2.3 3.1 204484_at PIK3C2B 2.3 1.3 203373_at SOCS2 2.3 2.3 1555486_a_at FLJ14213 2.3 4.6 202468_s_at CTNNAL1 2.2 4.6 209238_at STX3A 2.2 2.6 208264_s_at EIF3S1 2.2 2.5 205042_at GNE 2.2 2.2 202506_at SSFA2 2.2 1.6 223076_s_at NSUN2 2.1 1.8 226743_at SLFN11 2.1 4.2 212289_at ANKRD12 2.1 2.6 205114_s_at CCL3L1 2.1 2.3 209055_s_at CDC5L 2.1 2.1 202078_at COPS3 2.0 2.3 213117_at KLHL9 2.0 1.8 212398_at RDX 2.0 2.0 219383_at FLJ14213 2.0 2.2 1554433_a_at ZNF146 2.0 1.8 221831_at KIAA0601 2.0 1.8 202559_x_at C1orf77 2.0 2.0 209056_s_at CDC5L 2.0 1.9 238505_at ADPRH 2.0 1.8 201735_s_at CLCN3 2.0 1.9 203

236207_at SSFA2 1.9 1.7 220329_s_at C6orf96 1.9 1.9 202591_s_at SSBP1 1.9 2.9 203429_s_at C1orf9 1.9 1.7 205585_at ETV6 1.9 1.9 225268_at KPNA4 1.8 1.5 201771_at SCAMP3 1.8 2.8 244519_at ASXL1 1.8 1.8 223288_at USP38 1.8 1.6 204841_s_at EEA1 1.7 2.3 225858_s_at BIRC4 1.7 1.7 212251_at MTDH 1.7 1.8 213233_s_at KLHL9 1.7 2.3 202001_s_at NDUFA6 1.7 1.7 219575_s_at COG8 1.7 1.5 224768_at IWS1 1.7 1.8 224692_at PPP1R15B 1.6 1.7 1552792_at SOCS4 1.6 2.0 218505_at WDR59 1.6 1.6 205150_s_at KIAA0644 1.6 5.5 218099_at TEX2 1.6 1.7 205273_s_at PITRM1 1.6 1.8 218068_s_at ZNF672 1.6 2.0 214697_s_at ROD1 1.6 1.3 209432_s_at CREB3 1.6 1.9 235056_at ETV6 1.5 1.3 228162_at ESD 1.5 2.1 200945_s_at SEC31L1 1.5 1.6 212245_at MCFD2 1.5 1.3 200902_at SEP15 1.4 1.7 201039_s_at RAD23A 1.4 1.9 214838_at SFT2D2 1.4 1.5 225748_at C6orf93 1.4 1.8 212953_x_at CALR 1.4 2.3 201046_s_at RAD23A 1.4 1.8 223005_s_at C9orf5 1.3 1.6 219079_at CYB5R4 1.3 1.8

204

Genes on ChIP-chip list which decrease with IL-2+IL-12+IL-18 stimulation (24h)

Must be called “present” before stimulation by Affymetrix algorithm Fold change greater than 1.3 in both donors Duplicate probe sets not removed

Probeset Gene ID Donor 1 Donor 2 240718_at LRMP -42.8 -77.0 204867_at GCHFR -20.7 -24.3 204674_at LRMP -12.8 -20.4 35974_at LRMP -11.8 -25.5 214032_at ZAP70 -11.4 -10.9 1557261_at WHDC1L1 -9.4 -17.0 1555613_a_at ZAP70 -9.3 -8.9 213908_at WHDC1L1 -8.0 -14.6 208442_s_at ATM -7.7 -14.9 216979_at NR4A3 -7.4 -15.1 210858_x_at ATM -6.8 -11.4 212672_at ATM -6.8 -14.8 202587_s_at AK1 -6.8 -5.8 211021_s_at RGS14 -5.9 -41.1 1564093_at NEK1 -5.2 -6.9 202308_at SREBF1 -4.2 -2.5 1557257_at BCL10 -4.2 -6.6 207292_s_at MAPK7 -4.1 -2.3 243165_at BTF3 -3.7 -2.6 1554786_at C20orf32 -3.6 -14.9 235430_at C14orf43 -3.6 -3.3 221088_s_at PPP1R9A -3.5 -2.4 38290_at RGS14 -3.5 -3.8 1562442_a_at SSBP1 -3.3 -4.4 203825_at BRD3 -3.2 -2.4 212257_s_at SMARCA2 -3.2 -2.6 227515_at STAMBP -3.2 -2.4 234968_at C9orf55 -3.2 -11.1 1570352_at ATM -3.1 -10.3 1553387_at ATM -3.1 -2.5 225980_at C14orf43 -3.1 -4.0 215605_at NCOA2 -3.0 -5.7 236662_at CRAMP1L -3.0 -5.6 204280_at RGS14 -3.0 -3.3 203118_at PCSK7 -2.9 -1.7 214060_at SSBP1 -2.9 -3.3 209112_at CDKN1B -2.9 -3.7 206543_at SMARCA2 -2.8 -3.9 210365_at RUNX1 -2.7 -4.2 206544_x_at SMARCA2 -2.7 -2.3 222868_s_at IL18BP -2.7 -3.1 205

242349_at HECTD1 -2.7 -4.4 38241_at BTN3A3 -2.6 -3.6 232521_at PCSK7 -2.6 -2.8 204821_at BTN3A3 -2.6 -2.8 1558050_at EIF2B5 -2.6 -2.2 212875_s_at C21orf25 -2.5 -2.4 1554631_at ATM -2.5 -1.7 207564_x_at OGT -2.5 -2.4 200931_s_at VCL -2.5 -2.3 203312_x_at ARF6 -2.4 -1.7 232931_at ASCC3L1 -2.4 -2.0 204168_at MGST2 -2.4 -2.9 204675_at SRD5A1 -2.3 -2.8 207563_s_at OGT -2.3 -1.8 1554571_at APBB1IP -2.2 -2.6 217707_x_at SMARCA2 -2.2 -2.9 209984_at JMJD2C -2.2 -2.5 235539_at NUMA1 -2.2 -1.7 212756_s_at UBR2 -2.1 -1.9 211284_s_at GRN -2.1 -2.4 241027_at OPA1 -2.1 -5.0 212307_s_at OGT -2.1 -2.1 219994_at APBB1IP -2.1 -2.3 1559309_at MCFD2 -2.0 -1.8 219957_at RUFY2 -2.0 -3.6 201332_s_at STAT6 -2.0 -2.0 210959_s_at SRD5A1 -2.0 -2.0 200678_x_at GRN -2.0 -1.7 200747_s_at NUMA1 -2.0 -1.5 212032_s_at PTOV1 -2.0 -1.8 229787_s_at OGT -2.0 -2.7 208129_x_at RUNX1 -1.9 -1.4 206542_s_at SMARCA2 -1.9 -2.3 227648_at C3 -1.9 -1.8 35617_at MAPK7 -1.9 -2.0 204820_s_at BTN3A3 -1.9 -1.6 204494_s_at C15orf39 -1.9 -2.0 209285_s_at C3orf63 (RAP140) -1.9 -2.0 1557258_a_at BCL10 -1.8 -2.0 221729_at COL5A2 -1.8 -2.3 219114_at C3orf18 -1.8 -2.3 214251_s_at NUMA1 -1.8 -2.1 212760_at UBR2 -1.8 -1.9 205052_at AUH -1.8 -1.9 235745_at ERN1 -1.8 -1.8 212206_s_at H2AFV -1.7 -1.8 1553346_a_at TNRC6A -1.7 -2.2 52159_at HEMK1 -1.7 -1.5 209240_at OGT -1.7 -1.4 216041_x_at GRN -1.7 -1.4 218378_s_at PRKRIP1 -1.7 -2.0 219543_at MAWBP -1.7 -3.0 222108_at AMIGO2 -1.7 -1.9 224391_s_at SIAE -1.7 -4.3 206

239378_at PITRM1 -1.7 -3.5 231406_at TMEM142B -1.7 -1.5 241683_at HECTD1 -1.6 -1.8 224481_s_at HECTD1 -1.6 -1.7 201331_s_at STAT6 -1.6 -1.6 215087_at C15orf39 -1.6 -2.6 203285_s_at HS2ST1 -1.6 -1.4 209068_at HNRPDL -1.6 -1.5 218126_at FAM82C -1.6 -1.3 224705_s_at TNRC6A -1.6 -1.8 239212_at C6orf93 -1.6 -1.9 208990_s_at HNRPH3 -1.6 -1.6 226867_at C9orf55 -1.6 -2.4 213328_at NEK1 -1.5 -1.7 218620_s_at HEMK1 -1.5 -1.5 1553993_s_at MED25 -1.5 -1.8 1558111_at MBNL1 -1.5 -1.6 1555594_a_at MBNL1 -1.5 -1.6 218621_at HEMK1 -1.5 -1.6 214861_at JMJD2C -1.5 -2.4 223393_s_at ZNF537 -1.5 -1.5 212205_at H2AFV -1.5 -1.5 212454_x_at HNRPDL -1.5 -1.4 210110_x_at HNRPH3 -1.5 -1.9 225794_s_at C3 -1.5 -1.7 218811_at TMEM142B -1.4 -1.4 205732_s_at NCOA2 -1.4 -2.2 225172_at CRAMP1L -1.4 -1.9 223392_s_at ZNF537 -1.4 -1.4 221118_at PKD2L2 -1.4 -1.5 202932_at YES1 -1.4 -2.9 225795_at C3 -1.3 -1.4 233015_at MBNL1 -1.3 -1.5 201924_at AFF1 -1.3 -1.8 1556931_at CALR -1.3 -5.4 230185_at THAP9 -1.3 -2.2 205684_s_at C9orf55 -1.3 -2.3 1554678_s_at HNRPDL -1.3 -1.3

207

Appendix III: Genes changed 1.3-fold in NK cells with PRDM1 knockdown

Genes Increased with PRDM1 knockdown in Primary NK cells (48h)

Must be called “present” after k by Affymetrix algorithm Fold change greater than 1.3 in both donors Duplicate probe sets not removed

2+12 Up 2+12+18 Up

Donor 1 Donor 2 Donor 1 Donor 2

FLJ25791 4.72 1.78 C10orf97 8.64 1.31

HBA1 3.29 1.31 HIPK3 5.82 1.91

CXCL10 2.69 1.61 RAB8B 5.31 1.70

IL1R1 2.52 2.10 LRAP 4.96 2.04

DLL1 2.19 1.35 PSMA3 4.70 1.45

FAM26F 2.07 1.54 WHDC1L1 4.61 2.29

PTK2 2.03 2.12 RINZF 4.05 1.44

CD160 1.99 2.78 CXCL11 4.02 2.30

IPO11 1.93 1.65 SCOC 3.92 1.84

VNN2 1.87 1.53 SP1 3.91 1.42

DDX17 1.75 3.22 SGPP1 3.88 1.81

SLC16A7 1.73 1.50 MAF 3.88 1.37

DMRTA2 1.71 1.42 MLSTD2 3.85 1.92

HBA2 1.66 1.41 ZNF230 3.77 1.49

TNFRSF10B 1.64 1.88 NR4A3 3.70 1.32

CBX3 1.64 1.67 PDCD4 3.47 1.71

CLK4 1.57 1.35 FLI1 3.43 1.57

BCAR3 1.53 1.54 MBNL1 3.42 2.00

VEGF 1.51 1.49 MST4 3.33 1.33

JAK2 1.50 1.31 ITGA4 3.09 1.89

GART 1.50 1.38 NEDD1 3.08 1.43

ZAK 1.49 1.76 CAPRIN1 3.06 1.33

PHOSPHO2 1.46 1.47 B4GALT6 3.05 1.70

ZNF238 1.46 1.54 JAK1 3.03 1.41

FBXO8 1.45 1.37 MGC33214 2.98 1.59

FLJ37034 1.45 1.53 PSAT1 2.94 1.55

SOCS1 1.44 1.32 EIF4A2 2.92 1.35

TRFP 1.43 1.33 NCK1 2.89 1.57

IDH1 1.43 1.45 IGJ 2.86 1.84

TFAM 1.43 1.32 MCL1 2.79 1.44

ELF2 1.43 1.37 SPFH1 2.74 1.73

PARP15 1.42 1.45 CD58 2.73 1.35

NUAK2 1.41 1.54 LIG4 2.70 1.61

SEPT6 1.40 1.47 STAT6 2.68 1.76

TRIB2 1.40 1.31 STAT1 2.67 1.35

SIP1 1.40 1.54 PLCL2 2.64 1.53

POLR3G 1.39 1.36 PRKAB1 2.62 1.52

C3orf23 1.39 1.67 NCBP1 2.58 2.16

PGBD4 1.39 1.93 LYZ 2.57 1.79

EHBP1 1.39 2.08 SYPL1 2.54 1.42

AIM1 1.38 1.69 ARPP-19 2.53 1.36

208

FXR1 1.38 2.48 LACTB2 2.50 1.32

ALS2CR2 1.37 1.34 GFM2 2.50 2.52

DIAPH1 1.37 1.49 HEG1 2.45 1.93

TCF4 1.36 1.48 CAMK2D 2.41 1.54

ELK4 1.36 1.51 EFHC2 2.39 1.94

CCDC109B 1.35 1.68 STAM2 2.37 1.41

PLXNA4A 1.35 2.22 SKIL 2.35 1.46

GAS5 1.35 1.32 SH2B3 2.34 1.59

HADHA 1.34 1.33 CDKN2D 2.34 2.18

C16orf35 1.34 1.52 LOC51136 2.34 1.30

LRMP 1.32 1.35 TCF7L2 2.28 1.49

NEXN 1.32 1.34 CRSP3 2.27 1.37

MYC 1.32 1.40 TCF7 2.26 1.46

SYPL1 1.32 1.38 PPM1D 2.25 1.56

WHDC1L1 1.31 1.75 GK5 2.20 1.53

GPATC2 2.19 1.38 ZNRF3 2.19 2.17 MLLT10 2.18 1.38 WTAP 2.17 1.78 DDEF1 2.14 1.32 RKHD2 2.13 1.70 SH2D1A 2.13 1.37 LOC387763 2.12 2.10 NMD3 2.12 1.30 PCTK2 2.12 1.57 XCL1 2.11 1.31 BCL2L11 2.10 1.82 LPIN1 2.08 1.75

CYP1B1 2.08 1.35

B3GNT2 2.07 1.55

JAK2 2.05 1.57

MGC14595 2.05 1.31

ATF1 2.04 1.30

MAML2 2.02 1.55

SEPT6 2.02 1.57

CYBRD1 1.97 1.85

KIAA0828 1.97 1.42

RBM7 1.97 1.66

KLRK1 1.96 1.52

FH 1.95 2.21

ANXA1 1.95 1.46

ASXL2 1.94 1.52

CCDC126 1.94 1.65

SERINC1 1.93 1.37

PLEKHF2 1.91 1.36

SELL 1.91 1.63

CRIM1 1.89 1.33

NUMA1 1.89 1.56

MAP2K7 1.87 2.16

KLRD1 1.87 1.44

ADORA2B 1.84 1.49

SOCS1 1.84 1.40

PGM2 1.83 1.31

ZNF238 1.83 1.44

209

IL23R 1.82 2.62

KIAA2026 1.81 1.35

PTK2 1.80 1.46

PTPN4 1.78 1.31

FLJ23053 1.77 1.52

AIM1 1.76 1.69

PGGT1B 1.75 1.51

SETD2 1.75 1.58

HIVEP1 1.74 1.31

ETV6 1.73 1.40

TRA@ 1.72 1.44

ZDHHC2 1.71 1.40

PTPN1 1.70 1.45

LRMP 1.70 1.71

PSMD10 1.70 1.51

RAB6IP1 1.70 1.54

FLCN 1.68 1.35

DKFZp761P0423 1.68 1.40

KLHL6 1.68 1.37

KPNA1 1.68 1.31

KIAA1432 1.67 1.48

MBP 1.67 1.34

ARHGAP26 1.65 1.37

SURB7 1.64 1.62

IRF2BP2 1.63 1.32

MYADM 1.63 1.40

KLF2 1.60 2.29

C1orf174 1.60 1.39

SAV1 1.59 1.45

HCRP1 1.59 1.55

TAP1 1.58 1.40

DHX40 1.58 1.34

MOBKL2B 1.58 2.12

UST 1.58 1.88

EBI2 1.57 1.62

PEX12 1.57 1.38

CSF1 1.56 1.47

SPRY2 1.56 1.30

DLEU2 1.55 1.32

BCAR3 1.55 1.43

TMEM64 1.55 1.32

LOC352333 1.54 1.39

ZNF644 1.53 1.87

NIN 1.52 1.50

KIAA0974 1.50 1.66

ENPP3 1.49 1.85

HBA2 1.48 1.64

EPS8 1.47 1.47

FGFBP2 1.47 1.73

SCN10A 1.46 1.32

C13orf15 1.45 1.62

LPGAT1 1.44 1.45

GSK3B 1.44 1.31

GUSBP1 1.44 3.78

210

ZNF217 1.44 1.35

VTI1B 1.43 1.31

C12orf5 1.43 1.50

FBXO9 1.43 1.41

HTATIP2 1.43 1.60

DOCK9 1.43 1.37

GPR84 1.42 1.37

RSNL2 1.41 1.32

EIF5A2 1.40 1.54

PTEN 1.40 1.80

EXOSC6 1.40 1.35

CD160 1.40 1.57

NEXN 1.40 1.59

PHACTR2 1.39 1.37

SLC23A2 1.37 1.44

ARHGAP25 1.36 1.56

PELI2 1.35 1.51

APP 1.35 1.81

ARIH2 1.35 1.36

LOC391020 1.34 1.43

LOC150819 1.33 1.56

VNN2 1.33 1.32

SCML1 1.32 1.85

TLE1 1.31 1.64

KLRF1 1.31 1.68

211

Genes Decreased with PRDM1 knockdown in Primary NK cells (48h)

Must be called “present” before knockdown by Affymetrix algorithm Fold change greater than 1.3 in both donors Duplicate probe sets removed

2+12 Down 2+12+18 Down

Donor 1 Donor 2 Donor 1 Donor 2 DPF3 -7.79 -1.86 LOC348667 -3.33 -1.96

KBTBD4 -3.89 -1.57 IL6R -3.16 -1.37

MYCBP -3.74 -1.95 C8orf15 -3.14 -1.49

SNTB2 -3.72 -1.64 TncRNA -2.93 -1.45

GPR97 -2.34 -1.47 KLF12 -2.69 -1.38

SKIL -2.27 -1.43 ACACB -2.56 -1.56

VANGL1 -2.21 -1.85 ADM -2.46 -1.58

CLU -2.10 -2.13 ZMIZ2 -2.38 -1.41

SDCBP -1.88 -1.39 ITPR1 -2.37 -1.40

RTKN -1.84 -1.45 NAG8 -2.30 -1.54

PTGDS -1.83 -1.40 KIF21A -2.29 -1.67

NEIL1 -1.74 -1.63 RBM25 -2.12 -1.44

TMEM32 -1.71 -1.64 GPR107 -2.10 -1.96

INADL -1.70 -1.91 SCFD1 -2.10 -1.83

MYO1D -1.69 -1.49 CLCC1 -2.05 -1.47

ABCG1 -1.69 -1.37 FBXO31 -2.03 -1.33

NDFIP2 -1.66 -1.37 LOC286208 -2.01 -1.55

ACTN1 -1.64 -1.37 KIAA1618 -1.96 -2.26

OACT2 -1.63 -1.50 FCHO1 -1.95 -1.35

THRAP5 -1.62 -1.54 MGC23284 -1.94 -1.92

MED8 -1.61 -1.45 MCM10 -1.91 -1.56

RAB23 -1.61 -1.31 CDC25A -1.88 -1.35

C21orf91 -1.60 -1.39 GMPPA -1.84 -1.31

FAM46C -1.59 -1.93 GALNT6 -1.84 -1.31

FLJ10525 -1.55 -1.43 C15orf42 -1.83 -1.79

ARHGEF12 -1.55 -1.53 HDLBP -1.80 -1.36

PTGER3 -1.54 -1.39 PDIA4 -1.80 -1.36

ARPC1A -1.53 -1.54 CD7 -1.78 -1.32

GORASP1 -1.51 -1.32 BIRC5 -1.77 -1.31

PRDM1 -1.49 -1.81 FLJ22795 -1.75 -1.35

GABARAPL1 -1.47 -1.54 SMARCC1 -1.72 -1.37

KIR2DL3 -1.44 -1.53 MKI67 -1.71 -1.41

KIR2DL1 -1.44 -1.86 EWSR1 -1.71 -1.42

MLLT1 -1.42 -1.34 FOXM1 -1.68 -1.49

FLJ11184 -1.42 -1.93 CCND2 -1.67 -1.53

KIF21A -1.41 -1.31 EBP -1.66 -1.30

KIR2DS5 -1.41 -1.49 UNC84B -1.65 -1.37

DTYMK -1.41 -1.32 HELLS -1.64 -1.47

PAG1 -1.40 -1.59 NFKBIB -1.64 -1.35

KIR2DL5B -1.40 -1.73 FLJ36090 -1.62 -1.39

DLEU2 -1.39 -1.31 E2F2 -1.62 -1.51

LNX2 -1.39 -1.68 SDC4 -1.61 -1.39

ZNF537 -1.39 -1.59 ANXA11 -1.60 -1.30

TMEM33 -1.38 -1.40 CHPF -1.59 -1.32

PPP1R13B -1.38 -1.34 SIMP -1.59 -1.46

212

SAC3D1 -1.38 -1.39 ZBTB32 -1.59 -1.36

TIMP1 -1.36 -1.60 CDT1 -1.58 -1.52

MTMR4 -1.36 -1.70 MLLT1 -1.57 -1.45

ANXA5 -1.36 -1.40 ALCAM -1.55 -1.34

MAD1L1 -1.35 -1.30 SCARB1 -1.54 -1.73

KIR3DL2 -1.35 -1.61 PDE4DIP -1.54 -1.35

KIR3DL1 -1.35 -1.43 RRBP1 -1.53 -1.30

ATP6V0E2L -1.34 -1.37 ENO2 -1.52 -1.47

EHD4 -1.33 -1.44 S100A4 -1.51 -1.38

GALNACT-2 -1.31 -1.47 PCYT2 -1.50 -1.54

C6orf89 -1.31 -1.31 FLJ20254 -1.50 -1.35

TWSG1 -1.31 -1.34 AYTL2 -1.49 -1.33

MGC3020 -1.49 -1.32 SDF2L1 -1.49 -1.39 AURKB -1.48 -1.71 CDRT4 -1.48 -1.32 SF3B1 -1.47 -1.44 TMCC2 -1.47 -1.77 MAPK13 -1.47 -1.37 NALP2 -1.47 -1.31 DTYMK -1.47 -1.61 NCAPG -1.45 -1.30 PML -1.42 -1.32 NCAPD2 -1.41 -1.30 PDXK -1.39 -1.33

MFSD2 -1.38 -1.33

NT5DC2 -1.38 -1.33

SLC43A3 -1.37 -1.37

PTP4A3 -1.36 -1.36

NSDHL -1.36 -1.34

PCSK5 -1.36 -1.37

FLJ14627 -1.35 -1.36

PARP2 -1.33 -1.32

FAM57A -1.32 -1.49

HRBL -1.32 -1.39

KIAA1641 -1.31 -1.69

HYMAI -1.31 -1.64

213

ABOUT THE AUTHOR

Matthew Adams Smith was born in Boston, M.A., in 1978 and raised in Palmetto,

FL. After graduating as valedictorian from Palmetto High School in 1996, he attended

Florida State University in Tallahassee, FL. Matthew majored in biological sciences

graduating cum laude in 2000. While in Tallahassee he met Susan Blackwell to whom he

is now happily married. After graduation, Matthew worked for three years with the

Florida Department of Health, Bureau of Laboratories, pursuing a Master of Science in

Public Health at the University of South Florida’s College of Public Health under the

guidance of Dr. Lillian Stark. He then spent two years employed by the Centers for

Disease Control as a scientist with the BioWatch program before entering the doctoral

program in Biomedical Sciences at the University of South Florida’s College of

Medicine. While in the laboratory of Kenneth L. Wright, Ph.D., for the past 6 years,

Matthew has published two first-author papers in the Journal of Immunology and the

Journal of Biological Chemistry and has been a contributing author on three additional papers. He has presented his work at national and international meetings and received awards for local poster presentations. Upon graduation, he will continue to work at

Moffitt Cancer in the laboratory of Ed Seto, Ph.D.