Zinc Finger Protein Zfp335 Is Required for the Formation of the Naïve T Cell Compartment.” the Co-Authors on This Publication Were Shuang

Brenda Yuyuan Han

! ii! ACKNOWLEDGEMENTS

My time in graduate school has been a transformative period in my life, during which I have learned and grown tremendously both as a scientist and as a person, and I would like to take this opportunity to express my heartfelt gratitude to those who have helped me along the way.

First and foremost, I would like to thank my Ph.D supervisor Dr. Jason Cyster for his exceptional guidance and mentorship throughout the five years I have spent in his lab. I would be the first to admit that I started graduate school with only the haziest comprehension of how the scientific process really worked, but Jason has since taught me many invaluable lessons in being a good scientist, such as how to engage with the literature, design informative experiments, analyze and interpret data with care and rigor. From Jason, I learned the value of performing experiments iteratively, allocating time and resources efficiently, and maintaining focus and discipline in my efforts to keep moving my project forward. He has been patient, encouraging and supportive as a mentor, and for that I will always be grateful.

I would also like to thank Mark Ansel and Art Weiss, who have been so generous with their time as members of my qualifying exam committee and thesis committee. I have benefitted greatly from their insightful and constructive feedback on my thesis work both during and outside of committee meetings, and I will continue to seek their counsel on career and life beyond graduate school.

Various Cyster lab members, both past and present, have been an integral part of my graduate school experience and I feel fortunate to have had the opportunity to work alongside so many talented and energetic individuals. I am grateful to Lisa Kelly and Betsy Gray for sharing their senior grad student wisdom and for their kind words of encouragement back when I was still a clueless novice. In particular, I owe a huge debt of gratitude to Betsy: no matter how busy

! iii! she was, she somehow always managed to find time to help and give me advice when I needed it.

Betsy’s compassionate and generous spirit, disciplined work habits, organizational skills and conscientiousness were qualities that I greatly admired, and to this day I still think of her as my role model in lab. Tim Schmidt’s irrepressible good nature, upbeat attitude and overall bold approach to science and life made him an awesome lab sibling, and I shall always cherish the memory of how we worked together to organize a wildly successful Immunology Happy Hour that decisively raised the bar for all subsequent events. I can always rely on fellow graduate students Lauren Rodda and Eric Dang for enjoyable, refreshingly uncynical conversations, and I like to think that between the three of us we have for ourselves a Cyster lab grad student support circle. Lauren has an inquiring mind and a penchant for asking questions that force me to rethink my assumptions. Eric combines a towering scientific intellect with a wicked sense of humor and a cheerful willingness to help, and I very much appreciate having him as a bay mate, co- conspirator and (occasional) singing partner.

One of the best things about being in the Cyster lab is that with so many talented postdocs around, there was always someone I could go to if I needed guidance. Kazuhiro Suzuki, my rotation mentor, was a patient teacher and careful scientist who showed me the gold standard in lab technique. Oliver Bannard added a distinctly British flavor to the lab humor dynamics; he possesses a scientific acuity and passion for grappling with complexity that I truly appreciate and admire. I have the greatest respect for Tangsheng Yi, who not only was an awe-inspiring juggernaut of productivity but also an eternally cheerful dispenser of practical wisdom about work and life. Andrea Reboldi, my benchmate of five years who joined the lab at around the same time as I did, knows everything about everything and is a prolific purveyor of quality scientific advice and gossip. Francisco “Paco” Ramírez-Valle, our resident dermatologist, joined

! iv! our bay a couple years later and has been a model of medical professionalism as well as a continuing boon to lab morale ever since. Jagan Muppidi is our go-to guy for everything from

Plat-E cells to iOS upgrades and these days he seems to be the lab’s token grown-up. Michael

Barnes is an integral (if idiosyncratic) part of the lab’s social fabric and has been a steady source of project-specific feedback and bubble wrap-popping noises over the years.

I would like to give special thanks to my co-author Shuang Wu: her biochemistry expertise and hard work were indispensable to the work presented in this thesis, and she has been a dependable teammate throughout the course of this project. Many thanks also to Robert

Horton, with whom I worked closely on bioinformatics analysis. His enthusiasm and programming skills were instrumental to moving the project forward, and he was always happy to instruct me on the intricacies of R and Unix.

I also owe many thanks to the people who keep the wheels of the lab turning (more-or- less) smoothly. Claire Chan, our amazing administrative assistant, takes care of everything from ordering reagents to arranging for equipment repair to printing FedEx shipping labels; were it not for her effectiveness at getting things done, I am certain that the lab would fall apart within days.

Ying Xu, our molecular biology specialist, taught me how to clone my first construct and perform my first real-time PCR analysis, and I feel fortunate to be able to tap on her vast store of experience. Jinping An’s help in the mouse room has allowed me to keep both my mouse colony as well as my time spent in the bowels of LARC under control. I am also thankful for our wonderful lab assistant Bernarda Lopez, who performs most of the mundane, time-consuming lab chores so the rest of us can focus on research and never have to worry about running out of clean glassware or buffers.

! v! I would like to express my sincere thanks to Lisa Magargal, Monique Piazza, Nathan

Jew, Demian Sainz and the rest of the Biomedical Sciences office staff for their excellent administrative support throughout the course of graduate school. I also wish to thank my family, whose unconditional love has endured throughout all the long years I have spent away from home and on distant shores. Lastly, mere words alone cannot convey my gratitude for Chuan-

Sheng Foo, who has not only been a bedrock of emotional support and constant encouragement, but also is the best scientific collaborator one could hope for. Many of my best ideas were developed during our weekend discussion sessions and without him, this dissertation would literally have not been possible. Thank you for everything, CS.

! vi! CONTRIBUTIONS TO PRESENTED WORK

All work in this dissertation was done under the direct supervision and guidance of Dr.

Jason G. Cyster. Chapter 2 was published in eLife as “Zinc finger protein Zfp335 is required for the formation of the naïve T cell compartment.” The co-authors on this publication were Shuang

Wu, Chuan-Sheng Foo, Robert M. Horton, Craig N. Jenne, Susan R. Watson, Belinda Whittle,

Chris C. Goodnow and Jason G. Cyster. Dr. Jason Cyster and I collaboratively conceptualized and designed the research presented in this paper. The original ENU mutagenesis screen was designed and carried out by C.N. Jenne and S.R. Watson at the Australian National University under the supervision of J.G. Cyster and C.C. Goodnow. C.N. Jenne performed genetic mapping and B. Whittle coordinated exome sequencing which identified the Zfp335bloto mutation. S. Wu and I collaboratively designed the EMSA experiments in Fig. 7; all EMSA experiments in this paper were performed by S. Wu. Ying Xu designed allele-specific PCR primers for genotyping of Zfp335bloto mutant mice, cloned Zfp335 retroviral constructs, and prepared cDNA from cells I had sorted for quantitative RT-PCR analysis and Affymetrix array hybridization. R.M. Horton provided bioinformatics support for earlier analyses of ChIP-seq data leading up to the final version presented in this manuscript, which I developed in collaboration with C-S. Foo. C-S. Foo performed the motif analyses in Fig. 7C and 7D and also provided helpful discussion regarding the manuscript. I performed all other experiments, interpreted the data and generated the figures.

I wrote and revised the manuscript in consultation with Dr. Jason Cyster.

! vii! Zinc finger protein Zfp335 is required for the formation of the naïve

T cell compartment

Brenda Yuyuan Han

ABSTRACT

The generation of naïve T lymphocytes is critical for immune function yet the mechanisms governing their maturation remain incompletely understood. Through ENU mutagenesis screening, we identified a mouse mutant, bloto, characterized by a deficiency in naïve T cells. The causative genetic lesion was identified as a hypomorphic mutation in the zinc finger protein Zfp335. Zfp335bloto/bloto mice had a paucity of naïve T lymphocytes due to a cell- intrinsic defect in mature thymocytes, as well as in peripheral T cells that have recently undergone thymic egress. The T cell defect in Zfp335bloto/bloto mice could not be explained by altered thymic selection, proliferation or Bcl2-dependent survival. ChIP-seq analysis revealed that Zfp335 binds primarily to promoter regions, and we identified a set of target genes in thymocytes that are enriched in categories related to protein metabolism, mitochondrial function, and transcriptional regulation. Our analysis also showed that Zfp335bloto occupancy at a subset of sites was significantly decreased, and that this reduced binding correlated with deregulated target gene expression. Restoring the expression of one target, Ankle2, partially rescued T cell maturation. Finally, we identified a new DNA recognition motif and demonstrated that it was bound by Zfp335 in vitro. We suggest that Zfp335 binds directly to DNA in a sequence-specific manner to control gene transcription and in doing so, functions as an essential regulator of late- stage intrathymic and post-thymic T cell maturation.

! viii! TABLE OF CONTENTS

CHAPTER ONE: Introduction ………………………………………………………………… 1

T lymphocytes in immunity ……………………………………………………………... 2 Thymic selection ……………………………………………………………………….... 3 SP thymocyte maturation and emigration ……………………………………………….. 4 Homeostatic maintenance of naïve T cells …………………………………………….... 6 Recent thymic emigrants (RTEs) ………………………………………………………... 7 Regulation of T cell maturation in the thymus and the periphery ………………………. 9 Zinc finger protein 335 (Zfp335) ………………………………………………………. 11

CHAPTER TWO: Zinc finger protein Zfp335 is required for the formation of the naïve T cell compartment …………………………………………………………………………………… 13

Abstract ………………………………………………………………………………… 14 Introduction …………………………………………………………………………….. 15 Results ………………………………………………………………………………….. 17 Discussion ……………………………………………………………………………… 58 Materials and Methods …………………………………………………………………. 64

CHAPTER THREE: Conclusions and Discussion …………………………………………... 81

Zfp335 is required for the development of mature naïve T cells ………………………. 82 Cellular mechanisms by which Zfp335 regulates T cell maturation …………………... 82 Deciphering the function of Zfp335 through its interaction partners ………………….. 88 How does the bloto mutation affect Zfp335 function? ………………………………… 93 Elevated constitutive type I interferon signature in blt/blt mice ……………………….. 96 Relevance of Zfp335 to human immunological disease ……………………………….. 99

REFERENCES ………………………………………………………………………………. 100

! ix! LIST OF TABLES

CHAPTER TWO:

Supplementary Table 1: Summary statistics for ChIP-seq read alignment, mapping and peak calling …………………………………………………………………………….. 78

Supplementary Table 2: Functional annotation of Zfp335 target genes, with major enriched categories represented ………………………………………………………... 79

Supplementary Table 3: Sequences of oligonucleotides used for EMSA …………… 80

Supplementary Table 4: Sequences of primers used for RT-qPCR ………………….. 80

Supplementary Table 5: Sequences of primers used for ChIP-qPCR ………………... 80

! x! LIST OF FIGURES

CHAPTER ONE:

Figure 1: Maturation of SP thymocytes and recent thymic emigrants ...………………. 12

CHAPTER TWO:

Figure 1: Identification of an ENU mouse mutant with a cell-intrinsic deficiency in peripheral T cells ……………………………………………………………………….. 19

Figure 1 – Supplement 1: Similar relative decrease in blt/blt T cells in mixed chimeras vs. intact mice, indicating the lack of a competitive or rescue effect by WT cells ...……………………………………………………………………… 22

Figure 1 – Supplement 2: Mice heterozygous for the bloto mutation do not exhibit a T cell phenotype ...……………………………………………………. 22

Figure 1 – Supplement 3: blt/blt mice exhibit a selective defect in αβ T cells …………………………………………………………………………………... 23

Figure 2: Identification of causative missense mutation within a C2H2 zinc finger of Zfp335 ………………………………………………………………………………….. 25

Figure 2 – Supplement 1: Linkage mapping of bloto mutation to a 2.72 Mb region on chromosome 2 containing Zfp335 ……...... …………………. 27

Figure 2 – Supplement 2: Protein sequence analysis and structural modeling of mutated C2H2 zinc finger in Zfp335 …………………………………………... 28

Figure 3: Zfp335R1092W-induced T cell dysregulation affects mainly mature SP thymocytes and recent thymic emigrants ………………………………………………. 32

Figure 3 – Supplement 1: Impaired late-stage SP thymocyte development and early post-thymic peripheral T cell maturation in blt/blt mice ………………… 34

Figure 4: The T cell maturation defect in blt/blt mice is not caused by altered thymic selection or Bcl2-dependent survival …………………………………………………... 37

Figure 4 – Supplement 1: blt/blt mice exhibit intact positive and negative selection in the thymus …………………………………………………………. 39

! xi! Figure 4 – Supplement 2: Normal expression of IL-7 receptor and Bcl2 family members ………………………………………………………………………... 40

Figure 4 – Supplement 3: blt/blt naïve T cells proliferate normally in response to TCR stimulation in vitro and show no significant reduction in cycling of mature SP thymocytes ………………………………………………………………….. 41

Figure 5: Genome-wide analysis of Zfp335 binding sites in wild-type thymocytes based on ChIP-seq using an antibody against the C-terminus of Zfp335 …………………….. 44

Figure 6: Decreased Zfp335 binding in blt/blt thymocytes is detected for a subset of target genes …………………………………………………………………………….. 46

Figure 6 – Supplement 1: Analysis of correlation between changes in Zfp335 binding and gene expression in blt/blt thymocytes …………………………….. 48

Figure 7: Identification of a novel DNA motif bound by Zfp335 …………………….. 51

Figure 7 – Supplement 1: EMSA analysis of previously published Zfp335 motif …………………………………………………………………………………... 53

Figure 8: Ankle2 is a functional target gene of Zfp335 and its dysregulation by R1092W Zfp335 contributes to the maturation defect in blt/blt T cells …………………… 56

Figure 8 – Supplement 1: Ectopic expression of other Zfp335 target genes is not sufficient to reverse the T cell maturation defect ………………………………. 57 ! Supplementary Figure 1: Hierarchical clustering analysis of 108 genes differentially expressed in blt/blt vs. WT mature CD4SP thymocytes and CD4+ RTEs ……………... 77

CHAPTER THREE:

Figure 9: Effect of Zfp335R1092W on T cell maturation ………………………………... 83

Figure 10: Proposed models for how the bloto (R1092W) mutation affects Zfp335 binding to select target sites……..……………………………………………………… 95

! xii! CHAPTER ONE

Introduction

! 1! T lymphocytes in immunity

The adaptive immune system has evolved to defend the organism against broad range of pathogens, including bacteria, viruses, fungi and parasites. The effectiveness of the immune system depends on its ability to recognize and effectively neutralize pathogens, which are highly complex and diverse. At the same time, it must be capable of discriminating foreign agents from

“self”, so as to limit pathological immune responses against the host organism’s own tissues. T lymphocytes are central players in adaptive immunity – together with B lymphocytes, they provide diversity and specificity of antigen recognition and mediate appropriate immune responses to different pathogens.

T lymphocytes express T cell receptors (TCRs) that recognize specific antigens bound to major histocompatibility complex (MHC) molecules on other cells. T cells expressing the CD4 co-receptor (CD4+ T cells) specifically recognize antigens bound to class II MHC; CD8+ T cells bind to peptides presented by class I MHC. Under suitable conditions, TCR binding to cognate antigen/MHC triggers T cells to undergo activation, proliferation and differentiation into various types of effector cells to mediate an appropriate immune response; for example, through the secretion of cytokines that influence the activity of other immune cells, or by directly eliminating infected cells through cytotoxic mechanisms.

The generation of mature, self-tolerant T cells that express functional TCRs and are able to respond appropriately to antigenic stimuli occurs through a highly regulated developmental process that takes place primarily in the thymus and involves multiples stages of thymic selection, followed by post-selection maturation and emigration into the periphery.

! 2! Thymic selection

T cell development occurs in ordered stages that are classically defined by the temporally coordinated expression of the co-receptors CD4 and CD8. Lymphoid progenitors arising from hematopoietic stem cells in the bone marrow migrate to the thymus and under the influence of the thymic microenvironment, they commence differentiation as CD4−CD8− double negative

(DN) thymocytes. At this stage, the Tcrb locus undergoes V(D)J recombination, and productive rearrangement of the TCR β-chain results in its expression on the cell surface, in association with

CD3 and the invariant pre-TCRα chain (pTα). Successful assembly of this pre-TCR complex triggers proliferative expansion and progression to the CD4+CD8+ double positive (DP) stage, a process called β-selection. At this point, the cell is committed to the αβ T cell lineage (Starr et al., 2003).

At the DP stage, cells stop dividing and begin rearranging their Tcra locus. Productive

TCR α-chain rearrangement enables expression of a heterodimeric αβ TCR on the cell surface, which is then used to survey self-peptide MHC complexes presented by specialized cells in the thymic cortex, also known as cortical thymic epithelial cells (cTECs) (Klein et al., 2014). DP thymocytes expressing TCRs that form permissive interactions with self-peptide MHC on cTECs receive signals that allow them to survive. Positive selection is shortly followed by upregulation of TCR and CD69 on the cell surface. If the first attempt fails, DP thymocytes are able to continue testing multiple different TCR α-chains by processively rearranging their TCRα locus

(Starr et al., 2003). Continued failure to express a positively selected receptor results in DP cells undergoing apoptosis in a process referred to as death by neglect.

Positively selected DP thymocytes transition to the single-positive (SP) stage, during which cells expressing TCRs specific for class II MHC downregulate CD8 to become CD4 SP,

! 3! while those bearing MHC class I-restricted TCRs become CD8 SP. SP thymocytes migrate to the thymic medulla, where they interact with medullary thymic epithelial cells (mTECs) expressing peripheral tissue-restricted self-antigens under the control of the autoimmune regulator Aire, as well as with other antigen-presenting cells such as dendritic cells (DCs). At this stage, thymocytes that bind self-peptide/MHC presented by mTECs and DCs with high affinity are signaled to die by apoptosis, in a process known as negative selection or clonal deletion

(Hogquist et al., 2005). Negative selection prevents the maturation and export of T cells that present a risk of autoimmunity in the periphery. In addition, some self-reactive cells are selected to differentiate into regulatory T cells (Tregs), which help maintain self-tolerance (Stritesky et al., 2012).

SP thymocyte maturation and emigration

While residing in the medulla, SP thymocytes go through a series of maturational changes before they can be exported to the periphery as functionally competent T cells. SP thymocytes span a range of developmental states that may be distinguished by surface phenotype and functional differences. To systematically characterize SP thymocyte maturation, the field has widely adopted a two-stage scheme that divides the population into “semi-mature” and “mature”

SP subsets. The semi-mature phenotype (CD62LloCD24hi Qa2loCD69hi) arises directly after positive selection and is associated with susceptibility to apoptosis upon TCR stimulation. Semi- mature SP thymocytes are therefore thought to be the subset most sensitive to negative selection in the medulla (Sprent and Kishimoto, 2002). As SP thymocyte differentiation progresses,

CD62L and Qa2 are upregulated on the cell surface while CD24 and CD69 are downregulated; mature SP thymocytes are therefore identified as CD62LhiCD24loQa2hiCD69lo (Weinreich and

! 4! Hogquist, 2008). In contrast to their semi-mature precursors, mature SP thymocytes respond to

TCR stimulation not with apoptosis but proliferation and cytokine secretion. Only mature SP thymocytes upregulate sphingosine-1-phosphate receptor (S1PR1), which is required for S1P- mediated emigration to the periphery (Matloubian et al., 2004; Weinreich and Hogquist, 2008); this is thought to occur mainly at the cortico-medullary junction through the thymic perivascular space (PVS) (Zachariah and Cyster, 2010). From time of entry into the SP compartment to thymic exit, the entire process of SP thymocyte maturation has been estimated to take 4–5 days

(McCaughtry et al., 2007).

The phenotypic and functional shifts occurring during SP thymocyte maturation are accompanied by numerous changes in gene expression (Teng et al., 2011; Mingueneau et al.,

2013). Certain genes related to immune function are developmentally regulated as cells mature; these genes include MHCs, cytokine receptors (e.g. Il2rb, Il6ra), transcription factors (e.g. Klf2,

Foxo1) (Kerdiles et al., 2009; Carlson et al., 2006; Gubbels Bupp et al., 2009) and signaling proteins (e.g. Socs3, Pik3cd) that regulate immune responses. There are also changes in the gene expression of chemokine receptors and adhesion molecules involved in cortex-to-medulla migration (e.g. Ccr9, Plxnd1) (Choi et al., 2008; Campbell et al., 1999), thymic egress (e.g.

S1pr1) (Matloubian et al., 2004) and trafficking to secondary lymphoid organs (e.g. Sell)

(Arbonés et al., 1994).

Compared to earlier stages of T cell development, the mechanisms of regulation for the

SP maturation program are less well-understood. The most well-studied requirement is for transcription factor KLF2, which has been shown to be required for thymic emigration by controlling S1PR1 expression, as well as for naïve T cell homing to peripheral lymphoid organs through the upregulation of CD62L, CCR7 and β7-integrin (Carlson et al., 2006). Foxo1 is

! 5! thought to serve a similar function in thymic egress and T cell trafficking through its influence on the expression of S1PR1 and CD62L, although how it exerts this effect is less clear, and may occur via direct regulation or indirectly through KLF2 (Gubbels Bupp et al., 2009; Kerdiles et al., 2009; Fabre et al., 2008). Other potential regulators implicated in T cell maturation include the transcriptional repressor Nkap (Hsu et al., 2011) and the chromatin remodeling complex

NURF (Landry et al., 2011); however, the molecular mechanisms underpinning their roles in this developmental context have not been elucidated. A study of Egr1-deficient mice suggests that the quality of signals received during positive selection may affect not only the generation of SP thymocytes but also their ability to survive at later stages of maturation and even after leaving the thymus (Schnell and Kersh, 2005; Schnell et al., 2006).

Homeostatic maintenance of naïve T cells

After completing maturation in the thymic medulla, SP thymocytes are exported to the periphery, where they join a reservoir of T cells continuously recirculating through the body, surveying for MHC-presented antigen in secondary lymphoid organs. Before encountering antigen, T cells typically have a naïve phenotype characterized by high surface levels of the lymph node-homing cell adhesion molecule CD62L and low expression of the activation marker

CD44 (CD62LhiCD44lo).

Naïve T cells are quiescent and long-lived, being able to survive in interphase for extended periods of time – up to months in mice and years in humans (Sprent and Surh, 2011).

Prolonged viability of naïve T cells is dependent on continuous, weak interactions with self- peptide MHC complexes in the periphery, which provide T cells with tonic TCR signals that promote survival while being below the threshold necessary for overt activation. Additional

! 6! extrinsic survival cues are provided by the cytokine IL-7, which is produced by stromal cells, most prominently in the lymph nodes and bone marrow. IL-7 has been shown to be essential for naïve T cell maintenance in vivo: adoptive transfer into IL-7-deficient hosts or conditional ablation of the IL-7 receptor (IL-7R) results in naïve T cells dying within 1–2 weeks (Surh and

Sprent, 2008). IL-7R stimulation promotes T cell survival in part by inducing expression of anti- apoptotic proteins such as Bcl-2 and Mcl-1 (Takada and Jameson, 2009). IL-7 has also been shown to be important in maintaining normal metabolic activity and cell size through activation of the phosphoinositide 3-kinase (PI3K)/Akt and mammalian target of rapamycin (mTOR) pathways (Rathmell et al., 2001).

Recent thymic emigrants (RTEs)

The traditional view that all T cell maturation is completed prior to thymic egress, resulting in a uniformly mature peripheral naïve T cell population, has been substantially revised over the past decade as accumulating evidence suggests that newly exported T cells, also referred to as recent thymic emigrants (RTEs), constitute a distinctive subpopulation that undergoes further maturation in the periphery before acquiring the full set of surface phenotype and functional characteristics typical of fully mature naïve T cells.

In the past, the study of RTE biology had been hampered by the lack of unique surface markers that could be used to distinguish them from older naïve T cells. To identify RTEs, a standard technique was to inject fluorescein isothiocyanate (FITC) directly into the thymus and recover FITC-labeled T cells in the periphery 18–24 h later (Fink, 2012). While this approach was useful for the analysis of the youngest RTEs, the introduction of surgical stress and the potential for labeling mature T cells that recirculate back to the thymus presented caveats. The

! 7! study of RTEs has been facilitated in recent years by the development of Rag1-GFP knock-in

(Kuwata et al., 1999) and Rag2p-GFP transgenic reporter mice (Boursalian et al., 2004). Under the control of the Rag1 or Rag2 promoter, GFP expression is turned on during TCR rearrangement but extinguished shortly after. The residual GFP signal declines gradually with time, such that RTEs can be identified as GFP+ T cells in the periphery, with the youngest cells showing the highest GFP intensity (Boursalian et al., 2004).

Phenotypic analyses of RTEs revealed gradual upregulation of surface markers such as

Qa2, IL-7Rα, CD28 and CD45RB as cells become more mature, while other markers, notably

CD24 and CD3/TCR, are downregulated over time (Fink and Hendricks, 2011). In addition, recent studies have identified functional differences suggesting that RTEs are less immunocompetent than mature naïve T cells. For instance, compared to their mature counterparts, in vitro-stimulated CD4+ RTEs exhibit reduced proliferation and produce less IL-2,

IFN-γ and IL-4 (Boursalian et al., 2004; Hendricks and Fink, 2011). CD8+ RTEs have been shown to be less efficient at producing IL-2 and IFN-γ and generating KLRG1loIL-7Rαhi memory precursor effector cells (MPECs) in response to bacterial or viral infection (Makaroff et al., 2009).

Differences in homeostatic properties between RTEs and mature naïve T cells have also been described. Houston et al. reported that in lymphoreplete environments, RTEs displayed a competitive disadvantage relative to mature T cells, presumbly due to reduced survival.

However, under lymphopenic conditions, RTEs proliferated more rapidly than mature naïve cells; this property has been speculated to help RTEs rapidly populate a depleted peripheral T cell pool, such as in neonates or lymphopenic adults (Houston et al., 2011).

! 8! Regulation of T cell maturation in the thymus and the periphery

As reviewed above, much is known about the earlier stages of T cell development and selection in the thymus, as well as the factors influencing homeostasis of the bulk naïve T cell population. In contrast, the processes and molecular mechanisms involved in the generation of mature, fully functional naïve T cells from the SP thymocyte stage onward are far less understood.

The majority of published reports specifically focused on the biology of SP thymocytes and RTEs adhere to the traditional model that these constitute two distinct populations that are clearly demarcated by differences in anatomical location (thymus vs. periphery), and as a consequence, tend to study them separately. However, accumulating evidence from more recent studies suggest that the boundaries between mature SP thymocytes and RTEs may be more blurred than previously appreciated, and that the maturation processes in both compartments may form part of the same developmental progression and be regulated by the same mechanisms. In support of this concept, various gene deficiencies in mice have been reported to cause defects seemingly selective for both SP thymocytes and RTEs. For example, a study by Hsu et al. found that deletion of the transcriptional repressor Nkap at the DP stage left positive selection intact but strongly impaired SP thymocyte and RTE maturation. Compared to wild-type controls, Nkap- deficient mice had diminished numbers of Qa2hiCD24lo mature SP thymocytes and a severe deficiency in peripheral naïve T cells. Of the few naïve T cells remaining, most exhibited an immature Rag1-GFPhi phenotype, indicating that the maturation block began in SP thymocytes and continued as they transitioned into the periphery, eliminating the majority of RTEs within a few days after they exit the thymus (Hsu et al., 2011).

! 9! Multiple genes involved in NF-κB signaling have been implicated in regulating T cell maturation at various stages. For instance, ablation of RelB in hematopoietic cells has been shown to result in impaired accumulation of mature SP thymocytes (Guerin et al., 2002), and deletion of NEMO causes a developmental block at the semi-mature SP stage (Schmidt-Supprian et al., 2003). A recent study revealed that IKK2 was transiently required in new T cells for normal IL-7Rα induction and cell survival, but dispensable in fully mature naïve T cells (Silva et al., 2014). The authors also reported that IKK2 deficiency had no obvious impact on overall thymocyte development; however, a more detailed characterization of SP maturation was not shown. Nevertheless, it is interesting to note that CD24lo CD8 SP but not CD4 SP thymocytes showed decreased IL-7Rα expression in the absence of IKK2; this correlated with a greater fold- reduction in CD8+ naïve T cells compared to CD4+ T cells, which suggests that RTE survival is – at least to some degree – under the influence of events initiated in the thymus, particularly at the

SP thymocyte stage.

Finally, recent reports indicate that autophagy may be critical for normal T cell maturation (Pua et al., 2007; 2009; He et al., 2012). Mice with a T cell-specific deletion of the key autophagy gene Atg7 exhibited small decreases in CD4 and CD8 SP thymocytes that were followed by a more dramatic reduction in naïve T cell numbers (Pua et al., 2009). These results were suggestive of a block in the developmental transition from SP thymocytes to mature T cells, once again lending support to the idea that although the stages of development represented by SP thymocytes, RTEs and mature naïve T cells are temporally distinct, they may share common features of regulation within the context of an overall maturation program. However, the identity and function of the regulators controlling T cell maturation remain largely unknown.

! 10! Zinc finger protein 335 (Zfp335)

Zfp335, also known as NIF-1 (NRC-interacting factor), is a C2H2 zinc finger protein that was originally cloned and identified in a yeast two-hybrid screen as a novel interaction partner for nuclear receptor coactivator (NRC) (Mahajan et al., 2002). In initial reports, Zfp335 was characterized as an enhancer of nuclear hormone receptor activity (Mahajan et al., 2002); this was proposed to occur via interaction with NRC, although evidence for this association remained weak and unsupported by subsequent studies (Garapaty et al., 2009). Instead, mass spectrometry analysis of Zfp335-associated proteins identified several potential new interaction partners, among which were histone methyltransferase complex components such as ASH2L, RBBP5 and

WDR5 (Garapaty et al., 2009). The same study showed that purified Zfp335 complexes contained H3-specific methyltransferase activity, but did not define the enzymatic component responsible or establish the identity of the methylated H3 residue.

Little was known about the function of Zfp335 in vivo until the work described in this thesis and parallel studies by Christopher Walsh’s group that identified a mutation in the human ortholog ZNF335 causing severe microcephaly in a consanguineous pedigree (Yang et al., 2012).

Selective depletion of Zfp335 in mouse embryonic brain tissue revealed an essential role in neurogenesis and neuronal differentiation, and it was proposed that Zfp335 acts as a transcriptional regulator controlling the expression of genes important for neuronal development, potentially by recruiting H3K4 methyltransferase complexes to its target promoters (Yang et al.,

2012).

In Chapter 2, we identify a new role for Zfp335 as a regulator of T cell maturation. We show that mice harboring a hypomorphic point mutation in Zfp335 have a deficiency in naïve T cells due to a cell-intrinsic defect in SP thymocytes and RTEs. We provide evidence that Zfp335

! 11! binds predominantly to gene promoters in thymocytes; we also identify a novel consensus motif and demonstrate that it is bound to Zfp335 in vitro. Furthermore, we show that the mutation in

Zfp335 is associated with decreased binding to a subset of sites and dysregulated expression of some target genes, and that restoration of one such target, Ankle2, is able to partially correct the

T cell maturation defect.

Thymus Periphery T cell maturation

post-selection thymic maturation extrathymic maturation negative selection CD62Llo CD62Lhi IL-7Rα CD24hi CD24lo Qa2 TCRhi Qa2lo Qa2hi CD45RB CD69hi CD69hi CD69lo Ly6C thymic CD24 egress CD3/TCR post semi mat RTE MN -DP cortex to SP KLF2 SP S1P medulla Foxo1 low high migration S1PR1

TCR-induced apoptosis

positive selection positive pre- DP IL-2, IFNγ secretion Competence for TCR-induced proliferation

NEMO (Schmidt-Supprian et al., 2003) RelB (Guerin et al., 2002) Bptf (Landry et al., 2011) Nkap (Hsu et al., 2011) Atg7 (Pua et al., 2009)

IKK2 (Silva et al., 2014)

Figure 1. Maturation of SP thymocytes and recent thymic emigrants.

Schematic illustrating major surface phenotype and functional changes as T cells progress through late-stage thymic development and maturation in the periphery. The lower section shows genes encoding intracellular proteins that have been implicated in T cell maturation, as discussed in the text; bold lines define the affected developmental windows. DP, CD4 and CD8 double positive; pre-DP; pre-positive selection DP; post-DP, post-positive selection DP; SP, CD4 or CD8 single positive; semi-SP, semi-mature SP; mat-SP, mature SP; RTE, recent thymic emigrant; MN, mature naïve.

! 12! CHAPTER 2

Zinc finger protein Zfp335 is required for the formation of the naïve

T cell compartment

This chapter was originally published as: Han B.Y., Wu S., Foo C-S., Horton R.M., Jenne C.N., Watson S.R., Whittle B., Goodnow C.C., Cyster J.G. Zinc finger protein Zfp335 is required for the formation of the naïve T cell compartment. Elife. 2014 Oct 24;3. doi: 10.7554/eLife.03549.

! 13! Abstract

The generation of naïve T lymphocytes is critical for immune function yet the mechanisms governing their maturation remain incompletely understood. We have identified a mouse mutant, bloto, that harbors a hypomorphic mutation in the zinc finger protein Zfp335. Zfp335bloto/bloto mice exhibit a naïve T cell deficiency due to an intrinsic developmental defect that begins to manifest in the thymus and continues into the periphery, affecting T cells that have recently

bloto undergone thymic egress. The effects of Zfp335 are multigenic and cannot be attributed to altered thymic selection, proliferation or Bcl2-dependent survival. Zfp335 binds to promoter regions via a consensus motif, and its target genes are enriched in categories related to protein metabolism, mitochondrial function and transcriptional regulation. Restoring the expression of one target, Ankle2, partially rescues T cell maturation. These findings identify Zfp335 as a transcription factor and essential regulator of late-stage intrathymic and post-thymic T cell maturation.

! 14! Introduction

In order to mount effective adaptive responses against a diverse range of pathogens and antigens, the immune system has to generate sufficient numbers of mature peripheral T cells that express functional T cell receptors (TCRs). T cell development is a complex and highly regulated process that involves multiple stages of selection and maturation, both within the thymus and after thymic export. In the thymus, productive rearrangement of the TCR β-chain in

CD4−CD8− double negative (DN) thymocytes drives progression to the CD4+CD8+ double positive (DP) stage (Starr et al., 2003). After rearrangement of the TCR α-chain, DP thymocytes express mature TCRs which are then used to survey self-peptide/MHC complexes presented by specialized epithelial cells in the thymic cortex (cTECs) (Klein et al., 2014). A small percentage of DP thymocytes receive positively selecting TCR signals which promote their survival, in part through upregulation of IL-7Rα (Sinclair et al., 2011). Positively selected thymocytes become committed to either the CD4 or CD8 single-positive (SP) lineage and migrate to the thymic medulla, where they undergo further negative selection mediated by interactions with antigen- presenting cells such as dendritic cells (DCs) or AIRE-dependent medullary thymic epithelial cells (mTECs) (Hogquist et al., 2005; Klein et al., 2014), during which thymocytes expressing self-reactive TCRs either undergo apoptosis or are diverted to alternative fates, such as becoming regulatory T cells (Tregs) (Stritesky et al., 2012).

As SP thymocytes undergo maturation, expression of the surface marker CD24 is decreased while CD62L expression is upregulated. As such, SP thymocytes may be further divided into two phenotypically distinct populations, often referred to as semi-mature

(CD62LloCD24hi) and mature (CD62LhiCD24lo). These phenotypic changes are associated with an important functional difference: semi-mature SP thymocytes are susceptible to apoptosis upon

! 15! TCR stimulation, whereas mature SP thymocytes are not and respond instead by proliferating

(Sprent and Kishimoto, 2002; Weinreich and Hogquist, 2008). In addition, only mature SP thymocytes upregulate sphingosine-1-phosphate receptor (S1PR1), which is required for egress from the thymus (Matloubian et al., 2004; Weinreich and Hogquist, 2008). The process of SP thymocyte maturation, from entry into the SP compartment to thymic egress, has been estimated to take 4−5 days (McCaughtry et al., 2007). New T cells, also referred to as recent thymic emigrants (RTEs), undergo a phase of post-thymic phenotypic and functional maturation before they are incorporated into the long-lived peripheral naïve T cell pool (Fink, 2012).

The transition from the semi-mature to mature stage of SP thymocyte development is marked by numerous changes in gene expression. Some of these changes, including the upregulation of S1PR1 and CD62L are mediated by the transcription factor KLF2 (Carlson et al.,

2006). In the periphery, the process of post-thymic maturation is also associated with transcriptional changes, though on a smaller scale (Mingueneau et al., 2013). For instance, an increase in IL-7Rα expression during this period has been shown to promote RTE survival (Silva et al., 2014). Proper regulation of the transcriptional program underlying late-stage SP thymocyte and post-thymic T cell maturation is thus critically important for the establishment of a normal naïve T cell compartment. Multiple genes involved in NF-κB signaling have been reported to be required for the development of mature T cells (Guerin et al., 2002; Schmidt-Supprian et al.,

2003; Zhang and He, 2005; Sato et al., 2005; Wan et al., 2006; Liu et al., 2006; Silva et al.,

2014), primarily through mechanisms related to TCR signaling and protection from apoptosis. In addition, roles for the transcriptional repressor Nkap (Hsu et al., 2011) and chromatin remodeling factor Bptf (Landry et al., 2011) have been identified in recent years. However, the transcriptional regulators controlling these stages of T cell maturation remain largely unknown.

! 16! The C2H2 zinc finger family constitutes the largest class of transcription factors in mammalian genomes, and many key transcriptional regulators in immune cell development, such as Ikaros and Plzf, contain multiple C2H2 zinc fingers (Brayer and Segal, 2008). The C2H2 zinc finger fold is classically recognized to be a DNA-binding domain (Iuchi, 2005; Wolfe et al.,

2000), although it may also participate in interactions with RNA (Brown, 2005) or other proteins

(Brayer and Segal, 2008).

In this study, we identify a C2H2 zinc finger protein, Zfp335, as an essential regulator of

T cell maturation. Zfp335, also known as NIF-1 (Mahajan et al., 2002; Garapaty et al., 2009), is ubiquitously expressed and is essential for early development, with homozygous deletion resulting in embryonic lethality at E7.5 (Yang et al., 2012). Here, we report that an ENU-induced mutant allele of Zfp335 results in defective accumulation of naïve T cells, largely as a consequence of impaired maturation in SP thymocytes and RTEs. We show that this maturation defect is independent of thymic selection or effects on proliferation, but is associated with reduced viability. We identify a set of Zfp335 target genes in thymocytes and present evidence that decreased Zfp335 occupancy at a subset of these targets alters gene expression in mutant thymocytes. Taken together, our findings provide evidence that Zfp335 functions as a transcription factor and key regulator of a transcriptional program required for T cell maturation.

Results

The ENU mouse mutant bloto has a deficiency in peripheral T cells

As part of an N-ethyl-N-nitrosourea (ENU) mutagenesis screen (Nelms and Goodnow,

2001) for lymphocyte phenotypes, we identified a variant C57BL/6 mouse pedigree with decreased frequencies of CD4+ and CD8+ T cells in peripheral blood (Figure 1A), which we

! 17! named bloto (blood T cells low; allele henceforth designated blt). This trait was fully penetrant and occurred at a frequency consistent with recessive inheritance. Homozygotes were viable, fertile and displayed no gross external abnormalities.

Further characterization of the bloto mutant revealed a strong reduction in overall T cell frequencies in secondary lymphoid organs, especially in the CD62LhiCD44lo naïve T cell population (Figure 1B). Analysis of T cell development in the thymus revealed no significant decrease in frequencies or numbers of CD4−CD8− DN or CD4+CD8+ DP thymocytes of blt/blt mice relative to heterozygous controls (Figure 1C, D). However, blt/blt mice had slightly lower

SP thymocyte frequencies, and subgating on semi-mature (CD62LloCD24hi) and mature

(CD62LhiCD24lo) SP thymocytes showed significant underrepresentation of the mature subset

(Figure 1C), with an approximately 2-fold decrease in numbers of both CD4 and CD8 mature

SP thymocytes (Figure 1D). In comparison, CD4+ and CD8+ naïve T cell numbers in the spleen were reduced about 5- to 8-fold (Figure 1E), suggesting both a thymic and peripheral component to the bloto T cell developmental defect. In mixed bone marrow chimeras, lower percentages of blt/blt as compared to wild-type cells were observed in the SP thymocyte and naïve T cell populations, demonstrating that the T cell phenotype is cell-intrinsic and recapitulating the progressive developmental defect seen in intact mice (Figure 1F). We observed a similar relative decrease in blt/blt T cells in mixed chimeras compared to intact mice

(Figure 1 – figure supplement 1A), which indicates the lack of a competitive or rescue effect by wild-type cells and further demonstrates the cell-intrinsic nature of the bloto T cell defect.

This phenotype is a fully recessive trait with no evidence for haploinsufficiency or a dominant negative effect, since heterozygous mice exhibit no decrease in naïve T cells compared to wild- type controls (Figure 1 – figure supplement 2A), and blt/+ T cells do not decline relative to

! 18! !!

Figure 1. Identification of an ENU mouse mutant with a cell-intrinsic deficiency in peripheral T cells.

(A) Frequency of CD4+ and CD8+ T cells in peripheral blood of 8-week-old heterozygous (blt/+) or homozygous (blt/blt) mice as detected by flow cytometry. Numbers in quadrants indicate mean frequencies ± s.d. (n = 3 mice per genotype). (B) Frequency of total splenic CD4+ and CD8+ T cells (left); percentage of CD4+ T cells with a naïve (CD62LhiCD44lo) or effector (CD62LloCD44hi) phenotype (right). (C) Frequency of major thymocyte subsets (left); proportion of semi-mature (CD62LloCD24hi) and mature (CD62LhiCD24lo) subsets within the CD4SP thymocyte population (right).

(figure continued on next page)

! 19!

Figure 1 (continued). Identification of an ENU mouse mutant with a cell-intrinsic deficiency in peripheral T cells.

(D) Absolute number of DN, DP, CD4SP and CD8SP thymocytes in blt/+ vs. blt/blt mice. Semi- mature and mature CD4SP thymocytes were gated as in (C). CD8SP thymocytes were gated as follows: semi-mature (TCRβhiCD62LloCD24int), mature (TCRβhiCD62LhiCD24lo). Mature CD4SP and CD8SP thymocytes are reduced in numbers by approximately 1.8- and 2.3-fold, respectively. (E) Quantification of CD4+ and CD8+ naïve T cells in the spleen, gated as in (B), showing a 4.6-fold and 7.8-fold decrease in CD4+ and CD8+ naïve T cells, respectively. (F) Ratio of blt/blt (CD45.2+) vs. wild-type (CD45.1+CD45.2+) cells for splenic NK cells (NK1.1+TCRβ−), DP, semi-mature (semi) and mature (mat) SP thymocytes and naïve splenic T cells from lethally irradiated WT CD45.1+ hosts reconstituted with a 1:1 mix of blt/blt and WT bone marrow cells.

Data in (B) and (C) are representative of seven to eight independent experiments with matched blt/+ and blt/blt littermates and are summarized in (D) and (E). Mice were analyzed at 8 to 10 weeks of age (A−E) or 8 to 12 weeks post-reconstitution (F). Each symbol represents an individual mouse; small horizontal lines indicate the mean; n.s, not significant; * P < 0.05 and ** P < 0.01 (two-tailed Mann-Whitney test).

! 20! wild-type cells even in a competitive mixed chimeric setting (Figure 1 – figure supplement

2B).

Despite the strong defect in naïve T cells, we noted little difference in numbers of T cells with an effector/memory phenotype (CD62LloCD44hi) (Figure 1 – figure supplement 3A). This is likely due to homeostatic expansion of surviving cells in T cell-deficient blt/blt mice, as this increase in memory relative to naïve T cells was not observed in mixed chimeras in which the effects of T cell lymphopenia were alleviated by the presence of wild-type cells (data not shown). Non-conventional αβT cell lineages, such as Foxp3+ regulatory T cells and iNKTs, were also affected (Figure 1 – figure supplement 3B), but not to a greater degree than conventional

CD4+ and CD8+ T cells. However, there were no deficiencies in other major lymphocyte lineages such as NK cells (Figure 1F; Figure 1 – figure supplement 3C), γδT cells, and B cells (Figure

1 – figure supplement 3C), suggesting that the bloto mutation has a selective effect on αβT cells.

Identification of a missense mutation in Zfp335

To identify the causative genetic lesion, the peripheral blood T cell deficiency was used to map the mutation in an F2 intercross to a genomic interval between 163.16 to 165.88 Mb on chromosome 2 (Figure 2 – figure supplement 1A). Whole-exome sequencing of DNA from an affected mouse identified a single novel single-nucleotide variant within the interval of interest: a

C to T missense mutation in exon 21 of Zfp335 (Figure 2A). This results in the replacement of a positively-charged Arg residue at position 1092 (henceforth referred to as R1092) by Trp, a bulky non-polar amino acid.

! 21!

Figure 1 – Supplement 1. Similar relative decrease in blt/blt T cells in mixed chimeras vs. intact mice, indicating the lack of a competitive or rescue effect by WT cells.

(A) Ratio of blt/blt to WT mature SP thymocytes and naïve T cells normalized to the ratio in DP thymocytes from mixed chimeras (open symbols), compared to the ratio of the same subsets between matched pairs of intact mice (filled symbols), based on data reported in Figure 1. (mean ± s.d., n = 9)

Figure 1 – Supplement 2. Mice heterozygous for the bloto mutation do not exhibit a T cell phenotype.

(A) Number of CD4+ and CD8+ naïve T cells from spleens of WT (n = 6) and blt/+ (n = 4) mice. (B) Ratio of blt/+ (CD45.2+) vs. WT (CD45.1+CD45.2+) cells in splenic NK cells and indicated thymocyte and T cell populations from irradiation chimeras reconstituted with a mix of blt/+ and WT bone marrow as in Figure 1F.

! 22!

Figure 1 – Supplement 3. blt/blt mice exhibit a selective defect in αβ T cells.

(A) Number of effector/memory phenotype (CD44hiCD62Llo) CD4+ and CD8+ T cells in the spleen. (B) Number of Foxp3+ CD4SP thymocytes (left), splenic Foxp3+CD4+ Tregs (center) and splenic iNKTs identified by positive CD1d tetramer staining (right). (C) Number of splenic B cells (CD19+; left), NK cells (NK1.1+TCRβ+; center) and γδ T cells (TCRγδ+; right). n.s, not significant; * P < 0.05, ** P < 0.01, *** P < 0.001 (two-tailed Mann-Whitney test).

! 23! Zfp335 is a 1337-amino acid protein containing 13 predicted C2H2 zinc finger domains

(Figure 2B). Its role as a transcriptional regulator in neurogenesis and neuronal differentiation has recently been described (Yang et al., 2012), but any immunological function has thus far been unknown. The R1092W mutation falls within the 12th zinc finger (ZF12) near the C- terminus (Figure 2B), at a position that is highly conserved across vertebrate evolution (Figure

2C). Homology modeling places R1092 in the ZF12 α-helix at position +6 (Figure 2 – figure supplement 2A, 2B), one of the canonical positions mediating DNA base recognition by C2H2 zinc fingers (Wolfe et al., 2000). The presence of a TNEKP linker between ZF12 and ZF13

(Figure 2 – figure supplement 2A) and its similarity to the conserved TGEKP linker, a key structural feature of DNA-binding C2H2 zinc fingers (Wolfe et al., 2000), further hint at the possibility that R1092 may play a direct role in DNA binding by Zfp335.

Zfp335 transcript levels were not decreased in blt/blt thymocytes and T cells; in fact, a slight increase was observed relative to blt/+ controls, particularly in the most mature subsets

(Figure 2D). Western blotting analysis of thymocytes showed no reduction in the amount of

Zfp335 protein (Figure 2E), indicating that the R1092W mutation had no adverse effect on protein expression or stability. Confocal imaging of sorted mature SP thymocytes showed that both wild-type and mutant Zfp335 localized to the nucleus, forming punctate foci within regions of euchromatin (Figure 2F). No detectable differences in subnuclear distribution were observed.

These data suggest that the bloto mutation is hypomorphic rather than null, as it results in normal levels of stable protein that can localize appropriately to the nucleus but has impaired function due to the selective disruption of ZF12.

An in vivo gene complementation test was carried out by retroviral transduction of wild- type Zfp335 into blt/blt bone marrow for hematopoietic reconstitution of irradiated hosts. The T

! 24!

Figure 2. Identification of causative missense mutation within a C2H2 zinc finger of Zfp335.

(A) Sequence trace analysis of the mutated codon in homozygous (blt/blt) compared to heterogygous (blt/+) mice, showing an Arg-to-Trp substitution at position 1092. (B) Linear schematic of the thirteen C2H2 zinc finger (ZF) domains (shaded boxes) in Zfp335. Asterisk indicates the R1092W bloto mutation in ZF12 (black box). Diagram drawn to approximate scale. (C) Multiple sequence alignment of predicted Zfp335 orthologs from dog (Canis lupus), pig (Sus scrofa), human (Homo sapiens), mouse (Mus musculus), chicken (Gallus gallus) and zebrafish (Danio rerio). Asterisk indicates Arg residue affected by bloto mutation. Amino acids are colored according to their physicochemical properties. (D) Quantitative RT-PCR analysis of Zfp335 mRNA from indicated FACS-purified thymocyte subsets and naïve T cells (n = 3−4 mice, mean ± s.d. for biological replicates); ISP, immature CD8+ thymocytes identified by lack of TCRβ expression; results are presented relative to expression of Hprt. (E) Western blot for Zfp335 protein in thymocytes from wild-type (+/+) and homozygous mutant (b/b) mice, with actin as loading control. (F) Immunofluorescence analysis of Zfp335 nuclear localization in mature CD4SP thymocytes; nucleus counterstained with DAPI. (right) Secondary antibody-only negative staining control. Scale bar: 2µm.

(figure continued on next page)

! 25!

Figure 2 (continued). Identification of causative missense mutation within a C2H2 zinc finger of Zfp335.

(G) Frequency of CD4+ and CD8+ T cells differentiating from blt/blt hematopoietic stem cells transduced (Thy1.1+) with either wild-type Zfp335 (Zfp335WT) or control MSCV-IRES-Thy1.1 vector, compared to non-transduced (Thy1.1−) cells from the same mouse, 8 to 10 weeks after reconstitution of irradiated hosts. (H) Transduced (Thy1.1+) cells as a percentage of indicated thymocyte and T cell subsets from irradiation chimeras that had received bone marrow retrovirally transduced with WT Zfp335, bloto Zfp335 or control vector. Data points are connected by a separate line for individual mice. Data are representative of three independent experiments.

! 26!

Figure 2 – Supplement 1. Linkage mapping of bloto mutation to a 2.72 Mb region on chromosome 2 containing Zfp335.

(A) Black bars represent B6 homozygosity and gray bars represent B6/CBA heterozygosity as determined by SNP analysis. Data from two affected and three unaffected F2 progeny shown.

! 27!

Figure 2 – Supplement 2. Protein sequence analysis and structural modeling of mutated C2H2 zinc finger in Zfp335.

(A) Amino acid sequence of ZF12 and ZF13 (1073−1127 a.a). Positions −1, +2, +3 and +6, which are thought to mediate base recognition in DNA-binding C2H2 zinc fingers, are indicated for ZF12. R1092 (red); canonical C2H2 linker (green box). (B) Structural model of three zinc fingers near the C-terminus of Zfp335 (1044−1126 a.a). The mutated Arg residue is highlighted in red; gray spheres represent zinc ions coordinated within the zinc finger fold.

! 28! cell development block was strongly reversed in blt/blt cells transduced with wild-type Zfp335

R1092W but not control vector (Figure 2G, H), hence establishing that Zfp335 was the causative

R1092W mutation. Overexpression of Zfp335 yielded an intermediate rescue effect (Figure 2H), suggesting that supraphysiological protein expression may partially compensate for impaired function caused by a hypomorphic mutation. Interestingly, we observed that transduction frequencies for Zfp335WT in DP thymocytes (Figure 2H) or non-T lymphocytes (data not shown) were typically low (<10%) compared to transduction frequencies achieved with

R1092W Zfp335 or other genes, suggesting that overexpression of Zfp335 may have an inhibitory effect on early hematopoiesis, leading to poorer reconstitution of transduced progenitors.

blt/blt mice have defects in SP thymocyte maturation and homeostasis of recent thymic emigrants

To further characterize the block in intrathymic development, we examined the turnover of thymocyte populations by continuous in vivo bromodeoxyuridine (BrdU) labeling, where the percentage of BrdU+ cells indicates population turnover. After 4 days of BrdU administration, when comparing blt/blt mice to heterozygous controls, the mature SP population showed a decrease in turnover, whereas thymocyte subsets from earlier stages of development labeled with similar kinetics (Figure 3A; Figure 3 – figure supplement 1A). Genome-wide transcriptome analysis of sorted mature CD4SP thymocytes revealed a gene expression profile consistent with impaired SP thymocyte maturation; by showing, for instance, decreased expression of genes known to be upregulated during maturation (Teng et al., 2011) (Figure 3B). By comparing staining intensities of various surface markers associated with SP maturation (e.g. CD24,

CD62L) in pre-gated mature SP subsets, we also observed a trend towards a less mature surface

! 29! phenotype (data not shown). Taken together, these data suggest that blt/blt mice have decreased efficiency of entry into the mature SP compartment as well as progression through the final maturation stages within the mature SP thymocyte population.

As described earlier, blt/blt mice have a more severe defect in the accumulation of naïve

T cells compared to that of mature SP thymocytes, suggesting that Zfp335R1092W-induced dysregulation extends to events in the periphery following thymic export. To assess the impact of

R1092W Zfp335 on naïve T cell homeostasis, we adoptively transferred blt/blt peripheral T cells together with wild-type controls into congenic lymphoreplete hosts and assessed their relative maintenance over 7 days. Surprisingly, blt/blt T cells persisted just as well as control wild-type T

R1092W cells (Figure 3 – figure supplement 1B), suggesting that Zfp335 does not significantly impair the survival of the bulk naïve T cell population. This led us to hypothesize that the bloto defect may be largely confined to new T cells that have recently left the thymus, otherwise known as recent thymic emigrants (RTEs). In adult mice, these cells comprise a relatively small percentage of total naïve T cells and may therefore not be significantly represented in measurements involving the bulk naïve T cell population. We initially assessed the ability of SP thymocytes to survive once introduced into the periphery, in essence behaving as surrogate

RTEs. Significantly, blt/blt cells decayed more rapidly than co-transferred controls, hinting that

RTE survival may be impaired in blt/blt mice (Figure 3 – figure supplement 1B).

In order to study the RTE population in blt/blt mice in greater detail, we crossed the blt/blt mutant to the Rag1-GFP reporter line (Kuwata et al., 1999). In these reporter mice, GFP signal intensity is inversely proportional to time spent in the periphery (Boursalian et al., 2004), allowing for the identification of RTEs as GFP+ cells within the naïve T cell population. We found that the GFPhi subset was significantly overrepresented in blt/blt naïve T cells and was

! 30! skewed towards cells with the highest GFP signal intensities that have most recently exited the thymus, consistent with a partial block in post-thymic naïve T cell maturation (Figure 3 – figure supplement 1C). To directly assess RTE maintenance in vivo, a test population of either Rag1-

GFP blt/+ or Rag1-GFP blt/blt peripheral T cells was mixed with control non-fluorescent T cells at a known ratio and injected i.v. into congenic lymphoreplete hosts. Consistent with previous experiments (Figure 3 – figure supplement 1B), there was no significant decline in relative numbers of total blt/blt naïve T cells recovered from the spleen 1, 3 and 5 days post-transfer

(Figure 3C). However, the blt/blt Rag1-GFP+ population declined more rapidly than blt/+ Rag1-

GFP+ controls, particularly within the first day of adoptive transfer (Figure 3C). Interestingly, in contrast to the spleen, we noted a small decrease in the maintenance of total blt/blt naïve T cells, relative to blt/+ cells, that were recovered from peripheral lymph nodes. Similarly, the lymph nodes exhibited a larger decline in blt/blt Rag1-GFP+ relative to blt/+ Rag1-GFP+ T cells (Figure

3D). These data suggest that some of the apparent decline in blt/blt RTEs (Figure 3D) may be due to less efficient short-term accumulation in lymph nodes. Nonetheless, the fact that we observe in both spleen and lymph nodes a higher rate of decay in the Rag1-GFP+ population as compared to the overall effect on the bulk naïve T cell pool strongly suggests that blt/blt RTEs survive less well than control RTEs.

In terms of absolute numbers, GFPhi RTEs in blt/blt mice were reduced by a magnitude largely matching that of GFP− mature naïve cells (Figure 3 – figure supplement 1D), suggesting that most of the drop-off in cell numbers may have occurred at the earliest stages of RTE maturation. To test this hypothesis, we treated mice with FTY720, a potent inhibitor of thymic egress (Matloubian et al., 2004). This strategy ensured that new T cells, which would have otherwise been exported into the periphery, were trapped in the thymus where their accumulation

! 31!

Figure 3. Zfp335R1092W-induced T cell dysregulation affects mainly mature SP thymocytes and recent thymic emigrants.

(A) Percentage of BrdU+ cells in indicated thymocyte populations from blt/blt mice relative to blt/+ controls after 4 days of continuous BrdU labeling (mean ± s.d., n = 4). (B) Gene set enrichment analysis (GSEA) analysis of gene expression data from blt/blt vs. WT mature CD4SP thymocytes showing significant negative correlation with genes known to be upregulated during SP thymocyte maturation (MSigDB gene set: GSE30083). (C and D) Input-normalized fraction of total (left panel) or GFP+ (right panel) Rag1-GFP blt/+ and Rag1-GFP blt/blt cells within the total CD45.2+ CD4+ naïve donor population from recipient spleens (C) and peripheral lymph nodes (D) at indicated time points after co-transfer with control CD45.2+ blt/+ cells.

(figure continued on next page)

! 32!

Figure 3 (continued). Zfp335R1092W-induced T cell dysregulation affects mainly mature SP thymocytes and recent thymic emigrants.

(E) Flow cytometry analysis of CD4SP cells in the thymus of blt/blt mice and blt/+ controls after 4 days of FTY720 or saline treatment. Percentage of cells in mature SP (CD62LhiCD24lo) gate shown. Results are quantified (right) for CD4SP and CD8SP thymocytes. *** P < 0.001 (two- tailed Mann-Whitney test), data pooled from four independent experiments.

! 33!

Figure 3 – Supplement 1. Impaired late-stage SP thymocyte development and early post- thymic peripheral T cell maturation in blt/blt mice.

(A) Turnover of DN, DP, semi-mature and mature CD4SP thymocytes assessed after 2−4 days of continuous in vivo BrdU labeling (mean ± s.d., n = 4). (B) Input-normalized fraction of donor CD45.2+ CD4+ naïve T cells (closed symbols) or CD4SP thymocytes (open symbols) that were blt/blt, recovered from spleen at indicated time points after transfer of blt/blt and control peripheral lymphocytes or thymocytes to lymphoreplete CD45.1+ hosts (mean ± s.d., n = 2–4). (C) (left) GFP signal in CD62LhiCD44lo CD4+ and CD8+ T cells from spleen of Rag1-GFP transgenic blt/+ vs. blt/blt mice. (right) Percentage of GFPhi naïve T cells, gated as shown in histograms (mean ± s.d., n = 6−8). (D) Percentage of total naïve T cells, GFPhi RTEs and GFPlo mature naïve (MN) T cells from spleens of blt/blt mice relative to matched blt/+ littermate controls. Data obtained from mice analyzed between 7 to 10 weeks of age (mean ± s.d., n = 10).

! 34! could be measured. After 4 days of thymic egress blockade, blt/blt mice showed greatly impaired accumulation of SP cells with a mature CD62LhiCD24lo phenotype (Figure 3E), suggesting that most of the losses in RTEs take place within a short period after they enter the periphery. These data also indicate that the cell loss is due to intrinsic defects in the maturing T cells and is not dependent on their location in a particular lymphoid compartment.

Intact thymic selection in blt/blt mice

In mice with a polyclonal TCR repertoire, we observed normal frequencies of positively selected DP thymocytes with a CD69hiTCRβint phenotype (Figure 4 – figure supplement 1A), with no differences in CD5 surface expression on post-selection thymocytes (Figure 4 – figure supplement 1B), suggesting that positive selection is not strongly impaired by the bloto mutation. Because compensatory TCR rearrangements may mask a potential positive selection defect, we examined thymic development in blt/blt mice expressing the class II MHC-restricted

OTII TCR transgene, in which impaired positive selection would be expected to cause a dramatic reduction in CD4SP thymocytes. However, this was not observed in OTII blt/blt mice, which had only slightly lower CD4SP frequencies compared to controls and a fold reduction in CD4SP numbers (Figure 4A) similar to that seen in polyclonal blt/blt mice (Figure 1D).

Using the OTII/RIP-mOVA model of AIRE-dependent clonal deletion (Anderson et al.,

2005), we found no evidence that blt/blt mice had altered negative selection (Figure 4B).

Consistent with this conclusion, mRNA expression of Nur77, a key proapoptotic regulator induced during negative selection (Baldwin and Hogquist, 2007), was not elevated in blt/blt semi-mature SP thymocytes (Figure 4 – figure supplement 1D). Hence, our data indicate that altered thymic selection does not contribute to the T cell deficiency in blt/blt mice.

! 35! The bloto T cell deficiency is not due to defects in Bcl2-dependent survival, IL-7Rα expression or proliferation

The findings in Figure 3C, D and Figure 3 – figure supplement 1B suggested decreased survival may at least in part explain why blt/blt T lymphocytes fail to accumulate normally.

Consistent with this notion, blt/blt mature SP thymocytes showed a greater loss of viability in vitro over time compared to blt/+ controls, while a lesser effect was seen for blt/blt semi-mature

SP cells (Figure 4C). Annexin V staining and measurement of active caspase 3 in freshly isolated thymocytes did not reveal differences between blt/blt and blt/+ mice (data not shown), likely because cells in the earliest stages of apoptosis are efficiently cleared in vivo (Surh and

Sprent, 1994). To examine whether Bcl2-regulated apoptotic pathways were involved in promoting the loss of blt/blt T cells, we crossed blt/blt mice to a transgenic line expressing human BCL2 under control of the Lck promoter (Sentman et al., 1991). Overexpession of BCL2 failed to rescue the defect in peripheral naïve blt/blt T cells (Figure 4D), indicating that it is not due to reduced Bcl2-dependent survival. In addition, no differences in the expression of Bcl2 family pro- and anti-apoptotic genes were observed in blt/blt mature SP thymocytes (Figure 4 – figure supplement 2A).

IL-7 is a critical regulator of naïve T cell homeostasis (Surh and Sprent, 2008); in particular, IL-7Rα expression is induced in new T cells and is required for their survival and integration into the peripheral pool (Silva et al., 2014). However, no significant differences in IL-

7Rα expression were detected in blt/blt thymocytes and T cells, either at the transcript level

(Figure 4 – figure supplement 2B) or by surface receptor staining (Figure 4 – figure supplement 2C). Furthermore, survival of blt/blt thymocytes and naïve T cells in the presence of

! 36!

Figure 4. The T cell maturation defect in blt/blt mice is not caused by altered thymic selection or Bcl2-dependent survival.

(A) Flow cytometry analysis of major thymocyte populations from OTII TCR transgenic blt/+ and blt/blt mice, gated on total live thymocytes (left); quantification of OTII TCR-expressing Vα2+ CD4SP thymocytes from OTII blt/+ (n = 14) and OTII blt/blt (n = 14) mice (right). (B) Frequency of thymocyte subsets in lethally irradiated WT B6 (top left) or RIP-mOVATg (bottom

! 37! left) recipients reconstituted with T cell-depleted bone marrow from OTII blt/+ and OTII blt/blt mice; quantification of Vα2+ CD4SP thymocytes from indicated RIP-mOVATg chimeric mice (right). (C) In vitro viability of sorted CD45.1− blt/+ vs. blt/blt semi-mature (left) and mature (right) CD4SP thymocytes co-cultured with CD45.1+ WT CD4SP thymocytes. Live cells were pre-gated as Annexin V− DAPI− and percentage of CD45.1− cells was normalized to input. (D) Frequency of CD4+ and CD8+ T cells in the spleen of blt/+ vs. blt/blt mice expressing a human BCL2 transgene (BCL2Tg) under the control of the proximal Lck promoter (left); quantification of naïve T cells from BCL2Tg blt/+ (n = 10) and BCL2Tg blt/blt (n = 10) mice (right).

Data representative of eight (A), three (B), two (C) and six (D) independent experiments; n.s, not significant, *** P < 0.001 (two-tailed Mann-Whitney test).

! 38!

Figure 4 – Supplement 1. blt/blt mice exhibit intact positive and negative selection in the thymus.

(A) Surface expression of CD69 and TCRβ on DP thymocytes from blt/+ and blt/blt mice with a polyclonal TCR repertoire. Positively selected DP thymocytes are CD69hiTCRβint. (B) CD5 surface expression on CD69hiTCRβint and CD69hiTCRβint post-positive selection populations, gated on total live thymocytes. (C) Frequency of thymocyte subsets in OTI blt/+ and OTI blt/blt mice (mean ± s.d., n = 4). (D) Quantitative RT-PCR analysis of Nr4a1 (Nur77) mRNA in FACS- purified CD4 and CD8 semi-mature SP thymocytes (mean ± s.d., n = 3).

! 39!

Figure 4 – Supplement 2. Normal expression of IL-7 receptor and Bcl2 family members.

(A) Normalized expression levels of indicated Bcl2 family genes from Affymetrix array analysis of mRNA from sorted WT and blt/blt mature CD4SP thymocytes (mean ± s.d., n = 3). (B) Quantitative RT-PCR analysis of Il7r transcript in FACS-purified mature SP thymocytes and naïve T cells (mean ± s.d., n = 3). (C) Surface expression of IL-7Rα on WT (CD45.1+CD45.2+) and blt/blt (CD45.2+) mature SP thymocytes and naïve T cells from mixed chimeras. Data are representative of eight mice.

! 40!

Figure 4 – Supplement 3. blt/blt naïve T cells proliferate normally in response to TCR stimulation in vitro and show no significant reduction in cycling of mature SP thymocytes.

(A) BrdU labeling of DP and mature CD4SP thymocytes from blt/+ (n = 5) and blt/blt (n = 4) mice after a 4 h pulse (mean ± s.d., two-tailed Mann-Whitney test). (B) Percentage of indicated thymocyte subsets in S/G2/M phases of the cell cycle, as defined by >2n DNA content. Data pooled from three independent experiments (mean ± s.d., n = 6−7). (C) Analysis of CFSE dilution by congenically marked WT and blt/blt CD4+ naïve T cells in mixed culture, after 3 days of stimulation with plate-bound αCD3 and αCD28. Data are representative of two independent experiments.

! 41! IL-7 in vitro was comparable to that of co-cultured wild-type cells (data not shown), suggesting that the T cell defect is not due to loss of IL-7R function.

R1092W Lastly, Zfp335 had no significant effect on the fraction of thymocytes in cell cycle, as shown by short-term BrdU labeling (Figure 4 – figure supplement 3A) and DNA content analysis (Figure 4 – figure supplement 3B). Similarly, blt/blt naïve T cells were able to expand at a normal rate following TCR stimulation in vitro (Figure 4 – figure supplement 3C), demonstrating that blt/blt T cells are not defective in their ability to undergo proliferation.

Zfp335 binds to active gene promoters in thymocytes

Zfp335 has recently been shown to bind a variety of gene promoters in mouse embryonic brain (Yang et al., 2012), but its genome-wide binding characteristics and targets relevant to its function in T cell development remain to be defined. In addition, because our structure-function predictions (Figure 2 – figure supplement 2) led us to hypothesize a DNA-binding role for the mutated Arg, we wished to know if the R1092W mutation disrupted the ability of Zfp335 to bind to its targets in vivo.

To address these questions, we performed ChIP-seq analysis of Zfp335 binding sites in total thymocytes isolated from wild-type and blt/blt mice. ChIP-seq data sets were generated using two separate polyclonal antibodies and are referred to as “ChIP-C” and “ChIP-N” respectively. Using data obtained from wild-type thymocytes, peak calling with a q-value threshold of < 0.05 identified 157 Zfp335-binding regions in the vicinity of 177 genes. Genome browser inspection of ChIP-seq signal tracks confirmed that these peaks, although fairly limited in number, represented regions of significantly enriched binding intensity. Genome-wide,

Zfp335 peaks were strongly enriched in gene promoters (Figure 5A) and located upstream of

! 42! transcriptional start sites (TSS) (Figure 5B). Zfp335-bound regions were associated with high levels of H3K4me3, a hallmark of active gene promoters, and low levels of the enhancer- associated modification H3K27ac and the repressive chromatin mark H3K27me3 (Figure 5C), consistent with Zfp335 functioning primarily as a regulator of promoter-dependent gene transcription. Zfp335 target genes were enriched for functional categories representing a diverse range of biological processes, including protein synthesis and metabolism, mitochondrial function, cell cycle regulation, RNA processing and transcriptional regulation (Figure 5D;

Supplementary Table 2).

Decreased Zfp335 binding at a subset of target genes in blt/blt thymocytes

With the same q-value cutoff (< 0.05) that yielded a total of 157 binding events for wild- type thymocytes, we detected 141 peaks in blt/blt thymocytes (Supplementary Table 1). By visual inspection of ChIP-seq data on a genome browser, we determined that of the 28 peaks detected in wild-type but not blt/blt thymocytes, 22 showed a convincing loss of binding while the rest were false positives due to noise at low-confidence peaks. Of the nine peaks that were

R1092W WT called for Zfp335 but not Zfp335 , four were not true binding events, but rather signal artifacts arising from repeat regions, while the remaining five were low-confidence peaks. This strongly suggests that the bloto mutation does not lead to the gain of novel binding sites, which is consistent with our hypothesis that it is a loss-of-function hypomorph.

Interestingly, we did not observe a global decrease in Zfp335 binding intensities across all target sites. Reduced binding was detected for a subset of target genes: Ankle2, Nme6 and

Mrps5 are shown as representative examples (Figure 6A). However, at other target sites, such as

Rbbp5, Polr2e and Pes1, Zfp335 binding did not appear to be significantly impaired (Figure 6B).

! 43!

Figure 5. Genome-wide analysis of Zfp335 binding sites in wild-type thymocytes based on ChIP-seq using an antibody against the C-terminus of Zfp335.

(A) Genomic feature annotation of Zfp335 peaks reveals strong enrichment in promoter regions (≤1 kb upstream of TSS) and 5’ UTRs relative to genomic background. (B) Average profile of peak center distances from nearest RefSeq TSS for 141 Zfp335 peaks located within ±1.5 kb of a TSS, showing a positional preference for binding upstream of the TSS. (C) Average density of H3K4me3 (blue), H3K27ac (green) and H3K27me3 (purple) marks for a region from −2 kb to +2 kb relative to Zfp335 peak summits, based on ENCODE histone modification ChIP-seq data for murine whole thymus. (D) Gene ontology analysis of genes associated with Zfp335 binding sites using GREAT. Top enriched annotation terms in the MSigDB pathway ontology are shown.

! 44! Only nine Zfp335-bound regions showed decreases in ChIP-seq peak intensities in blt/blt thymocytes that were robust enough to be detected across both sets of ChIP-C and ChIP-N replicates (Figure 6C). We performed ChIP-qPCR to validate a selection of target genes that we had identified as differentially bound by Zfp335R1092W, versus targets that showed normal binding, and found the ChIP-qPCR data to be in agreement with the ChIP-seq-based assessment

(Figure 6D). Moreover, ChIP-qPCR analysis of sorted CD4SP thymocytes yielded similar results to the analysis with total thymocytes (Figure 6 – figure supplement 1A).

To understand how reduced Zfp335 binding at a subset of direct targets in blt/blt thymocytes could be biologically significant, we integrated our ChIP-seq data with gene expression profiles that we had obtained for blt/blt and wild-type mature CD4SP thymocytes. Of

R1092W the ten target genes listed in Figure 6C as having significantly reduced Zfp335 binding, only three (Ankle2, Nme6, Cnpy2) were downregulated in expression by more than 2-fold

(Figure 6E). Since our ChIP-seq results were derived from total thymocytes, of which only a small percentage are mature CD4SP, one caveat was that we might have missed gene expression changes in the larger population, so we sorted DP thymocytes (>90% of total) and tested them by

RT-qPCR. Relative to blt/+ controls, blt/blt DP thymocytes had no detectable differences in the expression of Mrps5 and Rabggtb, even though these genes exhibit strongly reduced Zfp335 binding in blt/blt thymocytes (Figure 6 – figure supplement 1B). This is not an unexpected finding given that transcription factors are known to bind sites which they do not functionally regulate (Smale, 2014). Nonetheless, we did observe deregulated expression of many Zfp335 target genes in our CD4SP array data, including many which did not exhibit reduced Zfp335 occupancy according to our stringent criteria (Figure 6 – figure supplement 1C). For example,

R1092W although Wdr47 failed to meet these criteria because it showed decreased Zfp335 binding

! 45!

! 46! Figure 6. Decreased Zfp335 binding in blt/blt thymocytes is detected for a subset of target genes.

(A) Signal tracks showing Zfp335 occupancy at three target genes (Ankle2, Nme6, Mrps5) for which significantly decreased binding in blt/blt relative to WT thymocytes is observed using both the C-terminus-specific antibody (ChIP-C) and the N-terminus-specific antibody (ChIP-N). Vertical axis, fragment pileup per million reads (normalized to library sequencing depth). Input, sequencing of input genomic DNA (background control). (B) Signal tracks showing Zfp335 occupancy at three target genes (Rbbp5, Polr2e, Pes1) for which no reduction in binding is detected in blt/blt thymocytes. (C) Identification of nine putative differentially bound target sites and their ten associated genes from the intersection of differential peaks (bloto < WT) called for both ChIP-C and ChIP-N datasets. We consider Aimp1 and Tbck to be associated with a single peak as they share a bidirectional promoter. (D) ChIP-qPCR analysis of Zfp335 binding at selected targets to validate ChIP-seq-based assessment of differential binding in blt/blt thymocytes. ChIP enrichment was calculated as percent input; results are presented as the fold- change in ChIP enrichment for blt/blt vs. WT (mean ± s.d., n = 3 for three independent experiments). (E) Relative Zfp335 binding and gene expression changes for target genes associated with the ChIP-C set of differentially bound regions: horizontal axis, expression fold- change (log2) values from microarray analysis of blt/blt vs WT mature CD4SP thymocytes; vertical axis, score reflecting likelihood that Zfp335 binding is significantly enriched in WT relative to blt/blt thymocytes. Red circles, target genes identified as differentially bound in both ChIP-C and ChIP-N datasets (Figure 6C); black circles, target genes associated with reduced Zfp335 binding in the ChIP-C but not ChIP-N dataset. Wdr47 is highlighted (filled black circle) as a target gene that was identified as differentially bound only in the ChIP-C dataset (Figure 6 – figure supplement 1D, E) but showed significantly downregulated expression in blt/blt thymocytes (Figure 6 – figure supplement 1F).

! 47!

! 48! Figure 6 – Supplement 1. Analysis of correlation between changes in Zfp335 binding and gene expression in blt/blt thymocytes.

(A) ChIP-qPCR analysis of Zfp335 binding at selected targets in sorted CD4SP thymocytes. ChIP enrichment was calculated as percent input; results are presented as the fold-change in ChIP enrichment for blt/blt vs. WT for two biological replicates. (B) Quantitative RT-PCR analysis of Mrps5 and Rabggtb mRNA in sorted DP thymocytes (mean ± s.d., n = 3). (C) GSEA plot comparing expression values of Zfp335 target genes (TSS within ±1 kb of a Zfp335 peak) in blt/blt vs. WT mature CD4SP thymocytes (left). Heatmaps depicting relative expression of the top 50 downregulated (right) and top 11 upregulated (bottom left) target genes; Wdr47 is marked with an asterisk. (D) ChIP-seq signal tracks showing Zfp335 occupancy at Wdr47, a target gene for which reduced Zfp335 binding is detected in the ChIP-C but not ChIP-N data set. (E) ChIP- qPCR confirmation of reduced Zfp335 binding to the Wdr47 promoter. Antibody used was specific for C-terminal epitope of Zfp335. ChIP enrichment represented as % input; matched data points for blt/+ (filled symbols) vs. blt/blt (open symbols) thymocytes from three independent ChIP experiments shown. (F) Quantitative RT-PCR analysis of Wdr47 mRNA in sorted DP thymocytes (mean ± s.d., n = 2−3).

! 49! in the ChIP-C but not ChIP-N dataset (Figure 6 – figure supplement 1D, E), its expression was nonetheless significantly downregulated in blt/blt thymocytes (Figure 6E, Figure 6 – figure supplement 1C, F). With the exception of a handful of strongly downregulated targets, gene expression changes were usually mild. Most deregulated target genes exhibited decreased expression, though a few were modestly upregulated (Figure 6 – figure supplement 1C).

Interestingly, this small group of upregulated targets include Zfp335 itself, which we confirmed by RT-qPCR (Figure 2D), suggesting that Zfp335 participates in an autoregulatory negative feedback loop. It is probably reasonable to assume that many of these differentially expressed

R1092W genes do in fact have reduced Zfp335 binding at their promoters, which we failed to detect owing to limitations in ChIP-seq sensitivity and/or the inability of our approach to identify differential binding events that are not evident at the whole population level because they occur exclusively in mature CD4SP thymocytes.

Identification of a novel consensus motif for Zfp335

As a member of the C2H2 zinc finger protein family, it is highly likely that Zfp335 interacts with its genomic targets through direct binding to a specific DNA sequence motif.

Using a set of high-confidence peaks from our Zfp335WT ChIP-seq data, we performed de novo motif analysis and identified a novel 22 bp bipartite motif consisting of two conserved elements separated by a variable spacer (Figure 7A). The positional distribution of this putative consensus motif was unimodal and located near the centers of Zfp335 peaks (Figure 7B), consistent with the hypothesis that it is the direct DNA-binding motif. Motif sites found within Zfp335 peaks showed a distinct DNase I genomic footprint (Figure 7C) and strong sequence conservation

! 50!

Figure 7. Identification of a novel DNA motif bound by Zfp335.

(A) Sequence motif identified by de novo motif search of WT Zfp335 ChIP-seq peaks. (B) Density histogram showing localization of motif relative to Zfp335 peaks. (C) DNase I genomic footprinting analysis of motif sites in Zfp335 ChIP-seq peaks (blue) compared with motif sites in all regions ±2kb of TSS (orange), using ENCODE DGF data for whole thymus. The 22 bp motif is marked on both sides by dashed lines. (D) Sequence conservation (phyloP) profiles around Zfp335 motif sites within ChIP-seq peaks (blue) versus sites in all regions ±2kb of TSS (orange).

(figure continued on next page)

! 51!

Figure 7 (continued). Identification of a novel DNA motif bound by Zfp335.

(E) Sequences of oligonucleotide probes used in (F) and (G). Z1 probe sequence was derived from Zfp335 binding site at the Zfp335 promoter and contains the primary consensus motif (capitalized, bold letters). For probes M1-M3, the first half (red), second half (green) or both parts of the consensus motif are mutated as shown. (F) Gel shift assay demonstrating sequence- specific binding of Zfp335 protein to labeled Z1 probe. Nuclear extracts from 293T cells transfected with control (−) or FLAG-Zfp335 (+) expression vectors were used. Signal from Zfp335-specific complexes (black arrowhead) is eliminated with an excess of unlabeled Z1 competitor oligo. Relative amounts of total Zfp335 protein verified by Western blot (bottom panel). Data are representative of three independent experiments. (G) Zfp335 binding to labeled Z1 probe in the presence of competition from unlabeled mutant oligos Z1-M1, Z1-M2 and Z1- M3. Signal intensity inversely correlates with ability of mutant probe to bind Zfp335: M1 is least able to bind, followed by M2, then M3. Black arrowhead, Zfp335 complex. Data are representative of three independent experiments.

! 52!

Figure 7 – Supplement 1. EMSA analysis of previously published Zfp335 motif.

(A) Gel shift assay showing Zfp335 complex formation with labeled Z1 probe but not with Pdap1 (Pd) probe containing the motif reported in a previous Zfp335 study (Yang et al., 2012). The standard Ikaros gel shift probe (Molnár and Georgopoulos, 1994; Cobb et al., 2000), IKbs4 (Ik), was used as negative control. (B) Zfp335 binding to labeled Z1 probe is competed away in the presence of excess unlabeled Z1 oligo, whereas Pdap1 (Pd) and Ikaros (Ik) oligos have no effect. Data from (A) and (B) are representative of two independent experiments.

! 53! across evolution (Figure 7D) compared with motif sites outside peaks, further suggesting that it is a functional DNA-binding motif for Zfp335.

To determine if Zfp335 was able to bind to this DNA sequence in vitro, we performed gel shift assays with labeled oligonucleotide probe containing the predicted consensus motif (Figure

7E) and 293T cell nuclear extracts ectopically expressing Zfp335 protein. We found that Zfp335 formed a gel shift complex with the labeled probe (Figure 7F). This complex increased in abundance proportional to the amount of Zfp335 protein and was eliminated by competition with excess unlabeled probe, demonstrating that it is sequence-specific and Zfp335-dependent

(Figure 7F). Competition experiments with unlabeled probes containing mutations of conserved nucleotides (Figure 7E) showed that Zfp335 binding was abolished when both DNA elements of the bipartite motif were mutated, whereas mutations in either the first or second element had an intermediate effect, indicating that both parts of the consensus motif are required for full Zfp335 binding (Figure 7G). We noted a stronger effect on competition upon mutation of the first element compared to the second, suggesting that it makes a greater contribution to Zfp335 binding (Figure 7G). We also performed gel shift assays with an oligonucleotide containing the previously reported Zfp335 recognition motif (Yang et al., 2012), but failed to detect binding above the negative control (Ikaros binding site) (Figure 7 – figure supplement 1). Re-analysis of the embryonic brain ChIP-seq data published by Yang et al. according to the parameters we applied to our dataset (see Methods) identified the motif shown in Figure 7A. These observations suggest that the discrepancy between our and the previously reported motif arises from differences in motif finding strategy. Importantly, the identification of a common motif in these distinct datasets strongly suggests that this sequence element is recognized by Zfp335 in a diversity of cell types.

! 54! R1092W Ankle2 dysregulation by Zfp335 contributes to the T cell maturation defect

To understand which of the direct targets of Zfp335 were functionally relevant to the T cell maturation defect in blt/blt mice, we focused our efforts on a set of genes for which we had clear evidence of reduced Zfp335 occupancy and mRNA expression, the most prominent of which was Ankle2 (ankyrin repeat and LEM domain-containing protein 2). Zfp335 binding to the Ankle2 promoter was effectively abolished in blt/blt thymocytes (Figure 6A, D); this was accompanied by decreased transcript expression across multiple stages of T cell development

(Figure 8A) and consequently, a virtual absence of Ankle2 protein (Figure 8B). Exogenous expression of Zfp335 significantly increased Ankle2 mRNA levels in blt/blt T cells (Figure 8C), providing evidence that Zfp335 is both necessary and sufficient for Ankle2 expression.

We were able to partially reverse the T cell maturation defect in the periphery by exogenously expressing Ankle2 in Rag1-GFP blt/blt cells (Figure 8D, E), as shown by the overall increase in representation of Ankle2-transduced (Thy1.1 reporter+) cells within the more mature Rag1-GFPlo naïve T cell subset compared to the less mature Rag1-GFPhi subset.

However, overexpression of Ankle2 had a weaker effect compared to that achieved by Zfp335

(Figure 8E), consistent with the idea that Ankle2, though important, is but one of several downstream targets that are required for Zfp335-dependent T cell maturation. In addition to

Ankle2, we tested five other Zfp335 target genes for their ability to rescue the T cell maturation defect in blt/blt bone marrow chimeras, but did not detect a significant effect for any of these genes (Figure 8 – figure supplement 1).

! 55!

Figure 8. Ankle2 is a functional target gene of Zfp335 and its dysregulation by R1092W Zfp335 contributes to the maturation defect in blt/blt T cells.

(A) Quantitative RT-PCR analysis of Ankle2 transcript levels in indicated thymocyte and naïve T cell populations sorted from blt/+ and blt/blt mice (mean ± s.d., n = 3−4). (B) Western blot for Ankle2 protein in wild-type (+/+) and mutant (b/b) thymocytes, with actin as loading control.

! 56! (C) Rag1-GFP blt/blt bone marrow was retrovirally transduced with either WT Zfp335 or control vector and used to reconstitute irradiated hosts. CD4+ RTEs from these chimeras were sorted into transduced (Thy1.1+) and non-transduced (Thy1.1−) populations and analyzed for Ankle2 expression by RT-qPCR (mean ± s.d., n = 3). (D) Gating strategy for spleen CD4+ naïve T cells, subdivided into GFPhi (less mature) and GFPlo (more mature) populations. Red line, Rag1-GFP+ naïve T cells; blue dashed line, Rag1-GFP+ mature SP thymocytes; grey fill, Rag1-GFP− CD45.1+ host naïve T cells (background control). GFPlo and GFPhi T cells were then gated on Thy1.1 reporter expression as indicated. Flow cytometry plots shown are for chimeras reconstituted with Ankle2- or empty vector-transduced Rag1-GFP blt/blt bone marrow. (E) Change in the percentage of cells transduced with Ankle2 (n = 6), WT Zfp335 (n = 5) or control vector (n = 10) during naïve T cell maturation (Δ%Thy1.1+ = %Thy1.1+ GFPlo − %Thy1.1+ GFPhi). A higher Δ%Thy1.1+ indicates enrichment of the reporter+ cells in the more mature GFPlo population compared to the less mature GFPhi population. Data are represented as Tukey box plots; * P < 0.05, ** P < 0.01 (one-tailed Mann-Whitney test). Chimeras were analyzed 8 weeks post-reconstitution.

Figure 8 – Supplement 1. Ectopic expression of other Zfp335 target genes is not sufficient to reverse the T cell maturation defect.

(A) blt/blt bone marrow was retrovirally transduced with Ankle2, Cnpy2, Nme6, Sep15, Tbck, Wdr47 or control vector, and used to reconstitute irradiated chimeras. The normalized ratios of %Thy1.1+ CD4+ naïve T cells relative to %Thy1.1+ DP thymocytes are shown. Overexpression of Ankle2 yielded a ratio greater than that for vector control, suggesting it is able to drive T cell maturation to some degree. Other constructs tested either showed weak or insignificant effects (Nme6, Wdr47, Sep15, Tbck), or may be inhibitory for maturation (Cnpy2). Each symbol represents one chimeric mouse.; ** P < 0.01 (one-tailed Mann-Whitney test). ! ! ! ! ! !

! 57! Discussion

Here, we identify a novel role for Zfp335 as an essential regulator of T cell maturation.

By analyzing mice with a hypomorphic missense mutation in a C2H2 zinc finger of Zfp335, we reveal a selective defect in the accumulation of naïve T cells resulting from a maturation block in

SP thymocytes and RTEs. In line with another recent study (Yang et al., 2012), we have shown that Zfp335 regulates transcription by binding to promoters of target genes. We have identified a set of direct targets in thymocytes and provide evidence that Zfp335 occupancy at a small subset of target sites was significantly decreased in mutant T cells. In addition, we identified a new

DNA recognition motif that it is bound by Zfp335. Taken together, our findings suggest that

Zfp335 acts as a novel transcription factor that regulates the expression of multiple genes required for late stage naïve T cell maturation.

We provide evidence that Zfp335 regulates T cell maturation in part by promoting

Ankle2 transcription. A study in C. elegans and HeLa cells showed a role for Ankle2 in nuclear envelope reassembly by promoting dephosphorylation of BAF during mitotic exit (Asencio et al.,

2012). As we found no significant defects in proliferation (Figure 4 – figure supplement 3) and did not detect nuclear envelope abnormalities in blt/blt mature SP thymocytes by immunofluorescence microscopy (data not shown), it is likely that the requirement for Ankle2 is dependent on some other as-yet-unknown function of this protein. To our knowledge, this is the first time an in vivo role for Ankle2 has been reported, and a more detailed understanding of its function and mechanism of action in T cells will be an important subject for future studies.

The T cell deficiency in blt/blt mice does not appear to be a consequence of defects in thymic selection, survival, proliferation or IL-7Rα expression, but is associated with reduced viability of mature SP thymocytes and recent thymic emigrants. Our finding that BCL2

! 58! overexpression failed to rescue the relative deficiency in blt/blt T cells, together with the unaltered expression of Bcl2 family genes (Figure 4 – figure supplement 2A) or other well- defined pro-apoptotic genes such as members of the death receptor family (data not shown), suggests Zfp335 has an indirect and likely multigenic pro-survival influence. Although our data do not provide support for altered thymic selection being an explanation for the reduced T cell numbers in blt/blt mice, we cannot exclude the possibility that the selection of some T cell specificities is affected by the Zfp335 mutation and future deep sequencing studies will be needed to fully address this issue. The intact proliferation of blt/blt T cells contrasts with the defective proliferation of human lymphoblastic cells and neuronal stem cells carrying a H1111R mutation in ZFP335 (ZNF335) (Yang et al., 2012). We suspect that this difference is a consequence of the almost complete loss of protein caused by the H1111R mutation, compared to the more subtle influence of the R1092W mutation studied here.

Naïve T cell deficiencies have been observed in several mouse lines with deletions in genes related to NF-κB signaling, such as RelB (Guerin et al., 2002), NEMO (Schmidt-Supprian et al., 2003), c-FLIP (Zhang and He, 2005), TAK1 (Sato et al., 2005; Liu et al., 2006; Wan et al.,

2006) and IKK2 (Schmidt-Supprian et al., 2003; Silva et al., 2014). However, the expression of these genes was not altered in blt/blt T lymphocytes and we did not detect enrichment for NF-κB targets in the gene expression analysis (data not shown). Two other genes known to be required for T cell maturation are the transcriptional repressor Nkap (Hsu et al., 2011) and Bptf, a component of the ISWI-containing chromatin remodeling complex NURF (Landry et al., 2011).

Similar to what we observed in blt/blt mice, the maturation block in Nkap- and Bptf- deficient T cells was not caused by altered thymic selection and could not be rescued by Bcl2 overexpression. However, neither Nkap nor Bptf showed significant expression changes in blt/blt

! 59! T lymphocytes. Moreover, we did not find a strong correlation between the expression profiles of blt/blt and Bptf-deficient SP thymocytes (unpubl. obs.). Nonetheless, this does not exclude the possibility of a partial overlap between genes regulated by Bptf and Zfp335 – one potential mechanism could be that Bptf-dependent nucleosome repositioning is required for efficient

Zfp335 binding to some of its target sites.

A small number of studies have shown that autophagy plays a role in mature T cell survival (Pua et al., 2007; He et al., 2012; Parekh et al., 2013), possibly by ensuring the clearance of excess mitochondria to minimize the harmful impact of reactive oxygen species (ROS) as newly mature cells transition from the thymus to the periphery (Pua et al., 2009). Although we observed genes involved in oxidative phosphorylation enriched amongst Zfp335 targets, our gene expression data did not reveal a clear autophagy signature, and initial experiments found no significant differences in mitochondrial content or ROS levels in blt/blt mature SP thymocytes

(data not shown). However, we cannot rule out the possibility that other autophagy-dependent processes may be affected by Zfp335 deficiency.

From our microarray analyses, we found that most gene expression changes in blt/blt T lymphocytes were subtle. In our mature CD4SP thymocyte dataset, of the 1504 genes passing a

P-value threshold of < 0.05, only 16 genes, or 1% of the total, had a 2-fold or greater change in expression, while the vast majority (93%) showed a less than 1.5-fold difference between blt/blt and WT. This may not be a surprising result, given that we were only able to detect strongly

R1092W diminished Zfp335 occupancy at a limited subset of target sites. However, due to the highly conservative criteria used and other technical limitations discussed earlier, this is most likely an underestimate of the true degree of differential binding, which may affect a broader range of target genes and account for many of the more subtle changes in gene expression.

! 60! Consistent with this notion, we observed that 33 direct Zfp335 targets, representing ~20% of the total, were differentially expressed in mature CD4SP thymocytes (P < 0.05), and of these genes, only four (Ankle2, Cnpy2, Nme6, Tbck) met our differential binding criteria. Furthermore, several known or putative transcriptional regulators are among genes directly targeted by Zfp335

(Supplementary Table 2), or differentially expressed in blt/blt thymocytes and RTEs

(Supplementary Figure 1), suggesting that Zfp335 operates within the context of a broader gene regulatory network. It is likely that the T cell maturation defect arises from the cumulative effects of relatively small gene expression changes within the Zfp335 regulatory network, possibly resulting in mild perturbations of multiple pathways whose functional consequences may not be individually significant, but in combination may negatively impact overall T cell fitness.

It is interesting to note that survival of mature naïve T cells appears to be intact, despite

RT-qPCR evidence indicating that a number of Zfp335 target genes found to be downregulated in blt/blt mature SP thymocytes and RTEs (Ankle2, Cnpy2, Sep15 and Wdr47) are similarly downregulated in the total naïve T cell population (Figure 8A and data not shown). If we extend this observation to the genome-wide level to assume that Zfp335R1092W-induced transcriptional dysregulation in mature naïve T cells approximates the situation in RTEs, one way to explain why mature T cells are less affected could be that the deregulated genes and their associated pathways are most critical during late-thymic to early post-thymic maturation, after which they become dispensable. In support of this concept, it has been reported that IKK2 is required transiently in RTEs but not in mature T cells for normal IL-7Rα upregulation and homeostasis

(Silva et al., 2014), although we have no evidence that this pathway is affected in blt/blt T lymphocytes. Alternatively, stochastic variations in gene expression or upregulation of

! 61! compensatory mechanisms may have resulted in comparatively “fitter” cells being preferentially selected into the mature naïve T cell pool.

Zfp335 is broadly expressed in hematopoietic cells (Heng et al., 2008) and in a variety of non-lymphoid tissues such as brain, kidney, heart and lung (Wu et al., 2009). Given its widespread expression and the early embryonic lethality resulting from its complete ablation

(Yang et al., 2012), it is highly likely that Zfp335 serves wider developmental roles. The poor contribution of Zfp335-overexpressing cells to hematopoietic lineages in BM chimeric mice also hints at a wider role. It may therefore seem remarkable that a mutation in such a ubiquitously expressed gene could produce a defect that appears to be selective for this particular stage of T cell development. It is possible that loss of Zfp335 binding to the target sites described in our study may be unique to T lymphocytes, resulting in T cell-specific dysregulation; however, this hypothesis is not favored by preliminary evidence showing that target genes such as Ankle2 were also downregulated in B cells (data not shown). As discussed earlier, our data that only a small number of target genes are significantly deregulated suggests that other aspects of the Zfp335- dependent transcriptional program may remain sufficiently intact to allow most developmental processes, except for T cell maturation, to proceed relatively normally. For future studies, it will be important to test whether a null allele has additional consequences for immune regulation beyond the selective T cell maturation effects revealed by the bloto mutation.

Genome-wide, we estimate that there are more than 2000 promoter regions containing

Zfp335 motif sites. Our study detected approximately 150 ChIP-seq peaks in thymocytes, consistent with the generally accepted principle that additional molecular requirements beyond

DNA sequence preference determine the stability of transcription factor binding to a given site.

At this early stage, it is not known what these requirements are for Zfp335, but they are likely to

! 62! involve local chromatin context and interactions with co-binding partners which remain to be defined. It is also unclear whether Zfp335 displays cell type-specific binding patterns, which may allow Zfp335 to fulfill various developmental roles by regulating different gene expression programs. Finally, although it has been proposed that Zfp335 activates target genes by recruiting

H3K4 methyltransferases (Yang et al., 2012), it is not clear if this is the sole mechanism, or that

Zfp335 functions exclusively as a transcriptional activator. Consistent with the view that Zfp335 drives transcriptional activation, we observed that most target genes with deregulated expression in blt/blt thymocytes tend to be downregulated. However, we have noted exceptions in which some target genes – including Zfp335 itself – were modestly upregulated, suggesting the possibility that Zfp335 may also act as a repressor. Further investigations will be needed to identify the interaction partners with which Zfp335 cooperates to control gene expression.

To conclude, our findings regarding Zfp335R1092W add to other recent work (Hsu et al.,

2011; Silva et al., 2014; Landry et al., 2011) to highlight unique gene expression requirements in late stage thymocytes and recent thymic emigrants for formation of a normal sized naïve T cell compartment. In this regard, the function of Zfp335 is similar to its action in the brain where it is needed for formation of a normal sized forebrain structure (Yang et al., 2012). In future work, it will be important to build from these findings to understand how this broad set of gene expression changes are integrated to promote formation of a normal sized compartment of cells.

It will also be important to understand whether Zfp335 acts constitutively during late stages of T cell development, or whether its function is regulated by external inputs and thus serves as a checkpoint that can influence the size and properties of the naïve T cell compartment.

! 63! Materials and Methods

Mice

bloto The Zfp335 strain was established through ethylnitrosourea (ENU)-mediated mutagenesis of

C57BL/6 (B6) mice at the Australian National University using methods previously described

(Randall et al., 2009). Putative mutants were identified as having blood CD4+ and CD8+ T cell frequencies more than one standard deviation below the mean. CD45.1+ congenic mice were from the National Cancer Institute (01B96; B6-LY5.2/Cr). CD45.1+CD45.2+ mice were generated by crossing B6 and Boy/J (Jackson Laboratory, 002014; B6.SJL-PtprcaPepcb/BoyJ) mice. OTII TCR transgenic mice (Tg(TcraTcrb)426-6Cbn) were from an internal colony. Rag1-

GFP transgenic mice were provided by N. Sakaguchi (Kumamoto University, Kumamoto, Japan)

(Kuwata et al., 1999). Lck-BCL2 transgenic mice were generated by S. Korsmeyer (Dana-Farber

Cancer Institute, Boston, MA) (Sentman et al., 1991) and provided by A. Winoto (University of

California Berkeley, Berkeley, CA). RIP-mOVA transgenic mice (Tg(Ins2-OVA)59Wehi) were provided by S. Sanjabi (University of California San Francisco). Mice were housed in specific pathogen-free conditions and all experiments were done according to the Institutional Animal

Care and Use Committee guidelines of the University of California San Francisco.

Genetic mapping and sequencing of the bloto mutation

Affected bloto mice were crossed onto the CBA/J background to generate heterozygous F1 mice.

F1 mice were intercrossed to yield F2 progeny homozygous for the bloto mutation and carrying a mix of C57BL/6 and CBA/J single nucleotide polymorphisms (SNPs). SNP mapping using an

Amplifluor assay (EMD Millipore, Billerica, MA) with Platinum Taq (Life Technologies,

Carlsbad, CA) was carried out on genomic DNA isolated from affected and unaffected mice.

! 64! Exome enrichment was performed using the SeqCap EZ Mouse Exome kit (Roche Nimblegen,

Basel, Switzerland), followed by 75 bp paired-end sequencing on the Illumina Genome Analyzer

IIx platform (Illumina, San Diego, CA). Computational analysis to detect novel single-nucleotide variants was done as previously described (Andrews et al., 2012). The affected exon was PCR- amplified from genomic DNA and Sanger sequencing (TACGen, Richmond, CA) was carried out to confirm the mutation.

Genotyping

bloto Zfp335 mice were genotyped by allele-specific PCR using the following primers: WT-F: 5’-

AGAACAAGAAGGATCTGAGGC-3’; bloto-F: 5’-AAGAACAAGAAGGATCTGAGGT-3’; common-R: 5’-GGCTCGGGCTGTAGAAGT-3’. WT and bloto allele-specific primers were run in separate reactions with GoTaq Hot Start polymerase (Promega, Madison, WI).

Constructs

Full-length Zfp335 was cloned from cDNA into an MSCV retroviral vector containing a Thy1.1

(CD90.1) reporter downstream of an internal ribosome entry site (IRES). The bloto mutation

(c.3274C > T) was introduced by site-directed mutagenesis. For use in transfections, WT and bloto Zfp335 inserts were subcloned into a pcDNA3.1 vector (Life Technologies) with a FLAG epitope tag at the N-terminus. Ankle2, Cnpy2, Nme6, Sep15, Tbck and Wdr47 were PCR amplified from mouse cDNA and cloned into the MSCV-IRES-Thy1.1 vector. All constructs were verified by sequencing. Reference sequences used in this study are as follows – protein:

Zfp335 (NP_950192.2); mRNA: Zfp335 (NM_199027.2), Ankle2 (NM_001253814.1), Cnpy2

(NM_019953.1), Nme6 (NM_018757.1), Sep15 (NM_053102.2), Tbck (NM_001163455.1),

! 65! Wdr47 (NM_181400.3).

Flow cytometry

Cells were isolated from thymus, spleen and lymph nodes by mechanical disaggregation through a 40-µm nylon sieve and stained as described (Schmidt et al., 2013). Antibodies were as follows: anti-CD4 (GK1.5, RM4-5), CD8 (53-6.7) (Biolegend, San Diego, CA; Tonbo Biosciences, San

Diego, CA); CD62L (MEL-14), CD44 (IM7), CD69 (H1.2F3), CD24 (M1/69), CD45.1 (A20),

CD45.2 (104), TCRβ (H57-597), CD19 (6D5), TCRγδ (GL3), Thy1.1 (OX-7), CD25 (PC61)

(Biolegend); NK1.1 (PK136), Vα2 (B20.1), CD5 (53-7.3) (BD Biosciences, San Jose, CA); IL-

7Rα (A7R34), Foxp3 (FJK-16s), BrdU (BU20A) (eBioscience); mCD1d/PBS-57 tetramer (NIH

Tetramer Core Facility). 4',6-diamidino-2-phenylindole (DAPI) was used for dead cell exclusion.

For all intracellular staining, cells were stained for surface antigens before fixation. Foxp3 was stained using the Foxp3 Staining Buffer Set (eBioscience). BrdU staining was performed as per manufacturer’s guidelines (BD Biosciences). For cell cycle analysis by DNA content, cells were fixed with BD Cytofix/Cytoperm Buffer and stained with 5 µM DAPI in Perm/Wash buffer (BD

Biosciences). Annexin V staining was performed using the Annexin V-PE Apoptosis Detection kit (BD Biosciences) according to manufacturer’s instructions. Samples were acquired on an

LSRII cytometer (BD Biosciences) and analyzed with FlowJo software (TreeStar, Ashland, OR).

Cell sorting

For microarray analysis, hematopoietic chimeras were generated with a mix of CD45.1+CD45.2+

Rag1-GFP WT and CD45.2+ Rag1-GFP blt/blt bone marrow. Thymocytes were isolated in

MACS buffer (PBS + 2% FBS, 2 mM EDTA), incubated with anti-CD8 microbeads (Miltenyi

! 66! Biotec, Bergisch Gladbach, Germany) and MACS-depleted to enrich for CD4SP thymocytes, after which cells were stained with anti-CD45.1, CD4, CD8, CD69 and CD62L. Mature CD4SP thymocytes (GFPhiCD4+CD8−CD62LhiCD69lo) were sorted into WT (CD45.1+) and blt/blt

(CD45.1−) subsets. For RTE sorting, spleen and lymph node cells were pooled and erythrocytes were lysed in a solution of Tris-buffered NH4Cl. Cells were labeled with anti-CD45.1, CD4,

CD8, CD62L and CD44. RTEs were sorted as CD4+CD62LhiCD44lo Rag1-GFPhi CD45.1+ (WT) and CD45.1− (blt/blt) subsets. Dead cells were excluded with DAPI. Samples were sorted on a

FACSAria with ≥98% purity.

Immunofluorescence

Mature CD4SP thymocytes were sorted and allowed to adhere to poly-L-lysine coated glass slides (P0425; Sigma-Aldrich, St. Louis, MO). Cells were then fixed with 4% PFA in PBS and permeabilized with 0.1% Triton X-100 for 10 min each on ice, followed by blocking (5% normal goat serum, 2% BSA in PBS) and staining (2% goat serum, 0.1% BSA, 0.1% Tween-20 in PBS).

For Zfp335 detection, slides were incubated at RT with primary antibody (A300-797A, A300-

798A; Bethyl Laboratories, Montgomery, TX) for at least 3 h, followed by biotin goat anti-rabbit

(BD Biosciences) and lastly SA-Cy3 (Jackson Immunoresearch, West Grove, PA). Slides were counterstained with 1 µM DAPI and mounted with Fluoromount-G (Southern Biotech,

Birmingham, AL). Confocal imaging was performed on a Leica SP5 inverted microscope with a

63× oil immersion objective. Images were processed with Leica LAS software.

Bone marrow chimeras

Mixed bone marrow chimeras were generated by intravenously transferring 3×106 to 5×106 cells

! 67! from the following mixes into lethally irradiated (2×450 rads, 3 h apart) CD45.1+ congenic mice: blt/blt CD45.2+: WT CD45.1+CD45.2+; blt/+ CD45.2+: WT CD45.1+CD45.2+, at a 50:50 ratio.

For OTII/ RIP-mOVA experiments, bone marrow from OTII blt/+ or OTII blt/blt mice was incubated with biotin-anti-CD3, CD4 and CD8 followed by anti-biotin microbeads (Miltenyi

Biotec) and T cells were depleted by MACS prior to transfer into lethally irradiated WT B6 or

RIP-mOVA recipients. Retroviral transduction of bone marrow from blt/blt or Rag1-GFP blt/blt mice was performed as described (Green et al., 2011) using the Platinum-E packaging cell line.

Chimeric mice were analyzed at least 8 weeks after reconstitution.

Adoptive transfers

For peripheral T cell transfers, spleen and lymph node cells were harvested from WT (control) or blt/blt mice and RBC-lysed. For thymocyte transfers, bulk thymocytes were prepared from blt/+

(control) or blt/blt mice. To distinguish between the two populations, either control or blt/blt cells were labeled with 1 µM CFSE (Life Technologies) for 10 min at 37°C before they were mixed and intravenously injected into lymphoreplete Boy/J (CD45.1+) mice. Recipients were analyzed at days 1 and 7 post-transfer and the percentage of blt/blt naïve T cells in the CD45.2+ donor population was determined. For Rag1-GFP T cell transfers, spleen and lymph node cells were harvested from Rag1-GFP blt/+ or Rag1-GFP blt/blt mice and RBC-lysed. Non-transgenic

CD45.2+ blt/+ cells were used as a mixing control. Either control or Rag1-GFP cells were labeled with 10 µM CellTraceTM Violet (Life Technologies) for 20 min at 37°C before mixing and intravenous transfer into Boy/J recipients. At days 1, 3 and 7 post-transfer, Rag1-GFP blt/+ or Rag1-GFP blt/blt naïve T cells were assessed as a proportion of the total CD45.2+ donor population.

! 68!

BrdU and FTY720 treatment

For BrdU experiments, mice received two intraperitoneal (i.p.) injections of 1 mg BrdU spaced 2 h apart and were euthanized 4 h after the first dose, or were fed 1 mg/ml BrdU in drinking water for longer-term labeling. To block thymic egress and induce mature SP thymocyte accumulation, mice were injected i.p. twice with FTY720 (Cayman Chemical, Ann Arbor, MI) dissolved in saline at a dose of 1 mg/kg body weight on days 0 and 2, and sacrificed on day 4.

T cell stimulation and culture

FACS-purified CD4+ naïve T cells were labeled with 5 µM CFSE for 10 min at 37°C, quenched with fetal bovine serum (FBS) and washed in 10% FBS. Labeled cells were stimulated with plate-bound anti-mouse CD3 (clone 2C11; 1 µg/ml) and anti-mouse CD28 (clone 37.51; 1

µg/ml) for 3 days. For in vitro cell viability assays, semi-mature and mature CD4SP thymocytes were sorted from chimeras reconstituted with a mix of either blt/+ and WT or blt/blt and WT

BM. Sorted cells were cultured in a 96-well plate at a density of 6–10×104 cells per well. All cells were cultured at 37°C in 5% CO2, using RPMI media supplemented with 10% FBS, L- glutamine, β-mercaptoethanol, penicillin and streptomycin.

Western blotting

Thymocytes were lysed in RIPA buffer containing protease inhibitor cocktail (EMD Millipore).

Samples were resolved on NuPAGE Bis-Tris gels (Life Technologies) and transferred to

Immobilon-FL membranes (EMD Millipore). Primary antibodies used: anti-Zfp335 (A300-

797A, A300-798A; Bethyl Laboratories), anti-actin (A2066, Sigma-Aldrich) and anti-FLAG M2

! 69! (F1804, Sigma-Aldrich). Ankle2 was detected using rabbit antiserum raised against human

ANKLE2, provided by I. Mattaj (EMBL, Heidelberg, Germany) (Asencio et al., 2012).

Secondary antibodies used: goat anti-mouse IRDye 800CW, donkey anti-rabbit IRDye 700DX

(Rockland Immunochemicals, Gilbertsville, PA). Blots were scanned using the Odyssey Infrared

Imaging System (LI-COR Biosciences, Lincoln, NE).

Electrophoretic mobility shift assay (EMSA)

HEK293T cells were transfected with FLAG-tagged WT or bloto Zfp335 (cloned as described) and nuclear extracts prepared using a modified Dignam protocol. Relative protein amounts were determined by Western blot using anti-Zfp335 (A300-797A). Gel mobility shift assays were performed as described (Lo et al., 1991). Briefly, nuclear extracts were mixed with 10 mM

HEPES pH 7.9, 0.1 mM EDTA, 10% glycerol, 0.5 mM DTT, 0.1 mM PMSF, 0.08 mg/ml BSA,

0.04 mg/ml poly dI/dC, 0.4 mM ZnCl2 and 10 mM biotin-labeled probe and incubated at room temperature for 20 min. For experiments using competitor oligos, 20- (Figure 7F) or 50-fold

(Figure 7G) molar excess of unlabeled probe was pre-incubated with nuclear extracts on ice for

30 min before adding biotinylated probe. Samples were loaded on a 4% polyacrylamide gel with

3.5% glycerol in 0.25X TBE and run at constant voltage in 0.25X TBE running buffer, then transferred to Biodyne B nylon membranes (Pall Corporation, Port Washington, NY) in 0.5X

TBE at 380 mA for 1 h. Signal was detected with LightShift Chemiluminescent Nucleic Acid

Detection Module kit (Thermo Scientific, Waltham, MA). All probes were derived from synthetic double-stranded 5’-biotinylated oligonucleotides, listed in Supplementary Table 3

(Integrated DNA Technologies, San Jose, CA).

! 70! Quantitative RT-PCR and microarray analysis

Total RNA was isolated using the RNeasy Micro Kit (Qiagen, Venlo, The Netherlands); cDNA was synthesized using MMLV reverse transcriptase and random primers (Life Technologies) according to manufacturer’s instructions. Real-time PCR was carried out using a StepOnePlus real-time PCR system (Applied Biosystems, Foster City, CA) with SYBR Green PCR Master

Mix (Applied Biosystems) and the appropriate primer pairs (Supplementary Table 4; Integrated

DNA Technologies). Relative mRNA abundance of target genes was determined by subtracting the threshold cycle for the internal reference (Hprt1) from that of the target. Primer pairs were tested for linear amplification over two orders of magnitude.

For microarray analysis, cDNA was prepared using the Ovation Pico WTA System V2 and labeled using the Encore Biotin Module (NuGEN, San Carlos, CA). Labeled cDNA libraries were hybridized to Affymetrix Mouse 1.0 ST arrays (Affymetrix, Santa Clara, CA) and scanned with the Affymetrix GeneChip® Scanner 3000 7G System. Raw data were normalized by RMA and probesets mapped to unique Entrez Gene IDs using a custom Brainarray CDF. The limma R package was used for analyzing differential gene expression. GENE-E software (Broad Institute) was used for heat map generation and hierarchical clustering.

ChIP-qPCR

Total thymocytes (20×106) or CD4SP thymocytes (4–10×106) were harvested from WT and blt/blt mice, cross-linked in 1% formaldehyde for 10 min at room temperature, quenched with

125 mM glycine and washed twice in ice-cold PBS. Cells were lysed in buffer containing 20 mM

Tris-HCl (pH 8.0), 85 mM KCl, 0.5% Nonidet-P40, followed by nuclear lysis buffer (50 mM

! 71! Tris-HCl pH 8.0, 10 mM EDTA, 1% SDS). Chromatin was sonicated with a Bioruptor

(Diagenode, Liège, Belgium), cleared by centrifugation and diluted in buffer containing 20 mM

Tris-HCl (pH 8.0), 1.1 mM EDTA, 140 mM NaCl, 0.01% SDS and 1.1% Triton X-100. Diluted chromatin was incubated overnight at 4°C with antibody bound to Protein A Dynabeads (Life

Technologies). All buffers up to this point were supplemented with protease inhibitors. Beads were washed in low salt buffer (10 mM Tris-HCl pH 8.0, 150 mM NaCl, 1 mM EDTA, 1%

Triton X-100), high salt buffer (20 mM Tris-HCl pH 8.0, 500 mM NaCl, 2 mM EDTA, 1%

Triton X-100, 0.1% SDS), LiCl buffer (10 mM Tris-HCl pH 8.0, 250 mM LiCl, 1 mM EDTA,

0.5% sodium deoxycholate, 0.5% Nonidet-P40) and TE buffer. Protein/DNA complexes were eluted (50 mM Tris-HCl pH 8.0, 10 mM EDTA, 1% SDS) at 65°C for 30 min with shaking on a

Thermomixer (Eppendorf). Eluted complexes were reverse-crosslinked overnight at 65°C and then treated with RNase A and proteinase K. ChIP DNA was purified using the QIAquick PCR purification kit (Qiagen) and quantitative PCR was performed using SYBR Green PCR Master

Mix (Applied Biosystems) and primer pairs listed in Supplementary Table 5. For Zfp335 ChIP, a polyclonal antibody specific for a C-terminal epitope (A300-798A; Bethyl Laboratories) was used.

ChIP-seq

ChIP was performed with total thymocytes (60×106 cells) as described above. Two antibodies were used: one raised against a C-terminal epitope (A300-798A; Bethyl Laboratories) and the other against an N-terminal epitope (A300-797A; Bethyl Laboratories) of ZNF335. 7−10 ng

ChIP DNA and 10 ng input DNA were used for library preparation, performed according to

Illumina’s TruSeq protocol with some modifications. DNA clean-up, removal of adapter dimers

! 72! and size selection were done using Agencourt AMPure XP beads (Beckman Coulter, Brea, CA).

Libraries were checked for quality using the High Sensitivity DNA Bioanalyzer kit (Agilent

Technologies, Santa Clara, CA), quantified with the Qubit dsDNA HS Assay kit (Life

Technologies), and sequenced as 50 bp single-end reads on the Illumina HiSeq 2000 platform.

ChIP-seq data analysis

Illumina adapter sequences were removed using the cutadapt tool. Trimmed reads were aligned to the mm9 reference genome using bwa, allowing for a maximum of two mismatches. Reads aligned with a MAPQ score of less than 20 were filtered out using samtools. Basic peak calling was performed with MACS2 (parameters: -g mm --bw=300 –q 0.05). Peaks were annotated with their nearest RefSeq TSS using HOMER. Genomic feature annotation summary statistics were generated using the Galaxy/Cistrome CEAS module (version 1.0.0), and the full set of target genes (n = 177) defined as having a TSS within 1 kb of high-confidence peaks were identified using the BETA-minus module (version 1.0.0). To call differential binding events, samtools rmdup was first used to remove duplicate reads, after which regions of differential enrichment were identified using the MACS2 callpeak and bdgdiff modules. Normalized pileup tracks were generated from nonredundant reads using MACS2 callpeak --SPMR (fragment pileup per million reads) and converted to the bigWig format for visualization on the UCSC Genome Browser. The

C-terminal antibody (A300-798A) was found to give the best enrichment and so we based our analyses of general Zfp335 binding properties on the ChIP-C WT data set, unless otherwise stated. H3K4me3, H3K27ac and H3K27me3 ChIP-seq data for mouse thymus were downloaded from ENCODE/LICR (Consortium, 2012), and aggregate density profiles were computed using bwtool.

! 73!

Gene ontology and pathway analysis

Gene ontology enrichment analysis of Zfp335 binding regions was performed using the Genomic

Regions Enrichment of Annotations Tool (GREAT). Each ChIP-seq peak was associated with the two nearest genes within 10 kb.

Motif analysis

400 bp sequences centered on peak summits were extracted from the 119 top-scoring WT

Zfp335 peak regions, repeat-masked and used for de novo motif discovery with MEME

(parameters: min. width = 6, max. width = 30, zero or one instances of a given motif per sequence). The top-scoring motif obtained with MEME was replicated using HOMER’s motif finding function. HOMER was used to detect Zfp335 motif occurrences and locations within defined genomic regions. HOMER was used to generate a histogram of motif density (bin size =

50 bp) for a region from −500 bp to +500 bp of Zfp335 ChIP-seq peaks. DNase I-seq signal and phyloP conservation scores for motifs occurring within Zfp335 peaks or ±2 kb of RefSeq TSS were aggregated using bwtool. The DNase I digital genomic footprinting (DGF) signal track for mouse thymus was obtained from ENCODE/UW (Consortium, 2012); sequence conservation tracks were downloaded from UCSC Genome Browser (Raney et al., 2014).

Bioinformatics

Software tools used:

• R 3.0.2

• affy R package (Gautier et al., 2004)

! 74! • limma R package (Smyth)

• Brainarray custom CDF (mogene10st_Mm_ENTREZG_17.1.0) (Dai et al., 2005)

• cutadapt 1.4.1 (Martin, 2011)

• bwa 0.7.7 (Li and Durbin, 2009)

• samtools 0.1.18 (Li et al., 2009)

• bedtools 2.17.0 (Quinlan and Hall, 2010)

• bwtool (Pohl and Beato, 2014)

• MACS 2.0.10 (Zhang et al., 2008)

• MEME 4.9.1 (Bailey et al., 2009)

• HOMER 4.5 (Heinz et al., 2010)

• GREAT 2.0.2 (McLean et al., 2010)

• GSEA 2.0.14 (Subramanian et al., 2005)

• Galaxy/Cistrome (Liu et al., 2011)

• UCSC Genome Browser (http://genome.ucsc.edu) (Kent et al., 2002; 2010; Rosenbloom

et al., 2012)

Protein structural modeling

A homology model was generated by SWISS-MODEL utilizing using the known structure of a designed DNA-binding zinc finger protein (Protein Data Bank ID: 1MEYC) as template. Figures were created using MacPyMOL (version 1.5.0.5).

Statistical analysis

Data were analyzed with Prism 5 (GraphPad Software). The two-tailed non-parametric Mann-

! 75! Whitney test was used for comparison of two unpaired groups for all datasets unless otherwise indicated.

Data availability

Microarray and ChIP-seq datasets generated in this study were deposited to NCBI’s Gene

Expression Omnibus under SuperSeries GSE58293.

Acknowledgements

We thank A. Winoto (University of California Berkeley) and S. Sanjabi (University of California

San Francisco) for mice, J. Landry (Virginia Commonwealth University) for sharing Bptf−/− expression data, H. Schjerven, G. Yeh and S. Smale (University of California Los Angeles) for

EMSA protocols and advice, P. Hartley and E. Dang for ChIP-seq protocols, and M. Thomson and A. Marson (University of California San Francisco) for bioinformatics discussion. We acknowledge the ENCODE Consortium for histone ChIP-seq and DNase I-seq datasets, and the

NIH Tetramer Core Facility (contract HHSN272201300006C)!for provision of the mCD1d/PBS-

57 tetramer. Special thanks to M. Ansel for critical reading of the manuscript, Y. Xu and J. An for technical assistance, and members of the Cyster lab for helpful discussion.

! 76!

Supplementary Figure 1: Hierarchical clustering analysis of 108 genes differentially expressed (P < 0.05, fold-change > 1.2) in blt/blt vs. WT mature CD4SP thymocytes and CD4+ RTEs.

! 77! Number of Sample Peaks reads

Total Unambiguously q < Alignable to mm9 Non-redundant sequenced mapped 0.05 Input_WT 25,390,396 20,593,485 13,022,918 11,815,042 NA Input_bloto 26,493,787 22,701,346 14,648,439 13,413,356 NA ChIP-N_WT 26,787,869 22,273,230 14,465,720 11,887,477 104 ChIP-N_bloto 23,942,405 20,318,391 13,809,035 10,577,178 131 ChIP-C_WT 23,774,983 18,110,353 11,400,834 9,196,919 157 ChIP-C_bloto 19,939,618 18,360,947 13,199,717 5,978,773 141

Percentage of Sample total Unambiguously Alignable to mm9 Non-redundant mapped Input_WT 81.1 51.3 46.5

Input_bloto 85.7 55.3 50.6

ChIP-N_WT 83.1 54.0 44.4

ChIP-N_bloto 84.9 57.7 44.2

ChIP-C_WT 76.2 48.0 38.7

ChIP-C_bloto 92.1 66.2 30.0

Supplementary Table 1. Summary statistics for ChIP-seq read alignment, mapping and peak calling.

! 78! Transcriptional Mitochondrion Cell cycle regulation A530054K11Rik Atp5l Anapc1

AA987161' Bckdha Anapc7

Gtf3c4 Ccdc58 Ankle2

Polr2e Cox7a2l Bre

Polr3f Cs Cdc73

Rbbp5 Fam162a Cep76

Suds3 Fars2 Gadd45gip1

Taf9 Gadd45gip1 Hectd3

Zfp335 Hadha Psmg2

Zfp383 Hadhb Rad17

Zfp738 Hspa9 Smc2

Iscu Ube2i

Ribosome/ translation Lyrm4

Aimp1 Mrpl11

Eif2s3x Mrpl38 RNA processing

Eif4enif1 Mrpl9 Nsun2

Eif4h Mrps5 Rexo1

Exosc5 Mrps7 2510012J08Rik

Fars2 Ndufa10 Eftud2

Farsb Ndufa4 Exosc5

Mrpl11 Nme6 Fars2

Mrpl38 Suclg1 Pcbp1

Mrpl9 Supv3l1 Sf3b3

Mrps5 Tomm20 Sart3

Mrps7 Trap1 Zranb2

Mrto4 Txnrd2

Nsun2 Vdac2

Pes1 Xpnpep3

Rpl21

Rpl35a

Rpl7

Rps10

Rps12

Rps23

Rps24

Rps27a

Vars

Supplementary Table 2. Functional annotation of Zfp335 target genes, with major enriched categories represented.

! 79! Name Motif Sequence (5' − 3') Z1 Zfp335 GTGTCTCGGGGCTGTCCCCGGCGCTGCCTGAGCTCCAGG Z1-M1 Zfp335 GTGTCTCGGGACGTTACCCGGCGCGATTGACGCTCCAGG Z1-M2 Zfp335 GTGTCTCGGGACGTTACCCGGCGCTGCCTGAGCTCCAGG Z1-M3 Zfp335 GTGTCTCGGGGCTGTCCCCGGCGCGATTGACGCTCCAGG Pdap1 (Pd) Zfp335 (Yang et. al.) GGGCCACCCCTACAGGAAGTGAGAGGAAAAGGGCC IKbs4 (Ik) Ikaros TGACAGGGAATACACATTCCCAAAAGC

Supplementary Table 3. Sequences of oligonucleotides used for EMSA.

Gene Forward primer (5' − 3') Reverse primer (5' − 3') Ankle2 TAATGGGTCACTACTATGTGCCA GGGCAGTATTTGAGGCTTCAG Il7r CAGCAAGGGGTGAAAGCAAC GGTCGTAGTTTTCTCTGTGGGA Mrps5 CAGGGGCCAGATCATTGGTG GTTCTGGACGACCCCACTTT Nr4a1 TGATGTTCCCGCCTTTGC CAATGCGATTCTGCAGCTCTT Rabggtb CGCCTGACACCCTGTTGC GCCGCTCATTCTCAGGTATTCA Wdr47 TTCAGCCTCTAGAATGTATGGAGA AATGCTGGGGCTCATCTTCT Zfp335 CGCCTAGAGACCTCAGACCT GATCCTGCAACTCTGGAGGG

Supplementary Table 4. Sequences of primers used for RT-qPCR.

Gene Forward primer (5' − 3') Reverse primer (5' − 3') Aimp1/Tbck GCGCTTCCCACTACTACACA GCTTCATGTGAGTGTCCCGA Ankle2 CGCGTTCTAGCCTGATCGTT CGTGGTTAGACCAGGCGG Cnpy2 TTTGCAAGTCCTAGCCCACC CTTACAGTGAGGACCGAGCG Mrps5 CCCTGCATTGTGTAGCTTGC TAGACTCGACAGACCCGAGG Nme6 AGAGAAGCCCAGGTTGTTCG ACACCTCCTACTCGTCTCCC Pes1 ATAACGAGGTCGCCAAAAGC CAGCTGAGGGCTGTCTGAAA Polr2e TTAGTCCGCAACTGAACGCT CTACCGGCTGTGGAAGATCC Rbbp5 CCAAGGAACACACGGCATTG GGGACCACGTAAGTGAGCAA Wdr47 CATAGCGTCACCCGAAGGTT GGGCGGGCAGAATCTAAGAC Zfp335 CGGGCAGCATGTGACTTCTA AGAGAGATCTGCGGGGTTTC

Supplementary Table 5. Sequences of primers used for ChIP-qPCR.

! 80! CHAPTER 3

Conclusions and Discussion

! 81! Zfp335 is required for the development of mature naïve T cells

Compared to other stages of T cell development, the maturation process that occurs when post-selection SP thymocytes transition into the periphery to become fully competent naïve T cells remains poorly understood. Here, we have identified Zfp335 as an essential transcriptional regulator of T cell maturation (Figure 9). This work also represents the first report of an immune function for Zfp335, whose in vivo significance – with the exception of one recent study describing its role in brain development – has been largely obscure. We demonstrated that mice with a hypomorphic mutation in Zfp335 (blt/blt) had a maturation block in SP thymocytes and

RTEs, resulting in the defective accumulation of naïve T cells in the periphery. We showed that

Zfp335 binds predominantly to promoter regions targeting approximately 200 genes in thymocytes, and provide evidence for significantly decreased Zfp335 binding at subset of targets in blt/blt cells. We provide evidence that Zfp335 binds to its target sites via recognition of an unusually large, bipartite DNA sequence motif. We propose that the disruption of Zfp335 binding to some of its target sites and the resulting impact on gene regulation accounts for the observed developmental block in naïve T cell maturation. In support of this model, we presented evidence that severely decreased Ankle2 expression concomitant with loss of Zfp335 binding at the Ankle2 promoter contributed to the T cell maturation defect in blt/blt mice.

Cellular mechanisms by which Zfp335 regulates T cell maturation

Our findings have provided new insight into the molecular function of Zfp335; however, major unresolved questions remain as to what cellular mechanisms are responsible for the role

Zfp335 plays in T cell biology, and how they may account for the T cell maturation defect in blt/blt mice. We showed that the T cell defect could not be explained by altered thymic selection

! 82! semi-SP mat-SP RTE MN

Zfp335WT

Zfp335R1092W

similar levels of Zfp335 expression

defect in blt/blt mice

Impaired Zfp335R1092W binding to target genes

altered gene expression other transcriptional (e.g. Ankle2) regulators

perturbations in various cellular processes/ pathways (e.g. autophagy, ribosomal function, metabolism, etc.)

decreased T cell fitness

Figure 9. Effect of Zfp335R1092W on T cell maturation.

(top half) Schematic illustrating developmental window most affected by maturation block in blt/blt mice. (bottom half) Proposed model for how the R1092W mutation affects Zfp335 function in maintaining T cell fitness during this developmental transition. (semi-SP, semi- mature SP; mat-SP, mature SP; RTE, recent thymic emigrant; MN, mature naïve)

! 83! (Figure 4A, B; Figure 4 – figure supplement 1), proliferation (Figure 4 – figure supplement

3), or IL-7Rα expression (Figure 4 – figure supplement 2B, C), but was associated with reduced viability of mature SP thymocytes (Figure 4C). However, overexpression of BCL2 failed to reverse the T cell defect (Figure 4D), suggesting that the effects of Zfp335 on SP thymocyte and naïve T cell survival are not solely dependent on classical apoptotic pathways, and therefore future efforts to elucidate these functions would necessitate a closer examination of other cellular processes that may be relevant to this developmental transition.

A convenient starting point for generating hypotheses to address these questions would be to consider the transcriptional changes associated with pre- and post-thymic T cell maturation.

In a comprehensive study of αβ T cell differentiation based on transcriptome profiling data generated by the Immunological Genome Project (ImmGen) consortium, it was reported that a number of genes were significantly upregulated (>2-fold) for both CD4+ and CD8+ lineages during the transition from mature SP thymocytes to peripheral naïve T cells, but these transcriptional changes was judged to be limited in scale compared to those occurring at earlier stages of thymocyte differentiation (Mingueneau et al., 2013). Nonetheless, there is new evidence to suggest that one of the genes most strongly induced during post-thymic maturation,

Cd55, may be functionally important for RTE maintenance. CD55, also known as decay- accelerating factor (DAF), protects cells from complement-mediated lysis by blocking the activation of complement on the cell surface. A recent study found that Nkap-deficient peripheral

T cells had a defect in CD55 upregulation which correlated with increased complement deposition on the cell surface (Hsu et al., 2014), leading the authors to propose that the severe T cell maturation block seen in Nkapfl/fl CD4-Cre mice was due to the elimination of immature

RTEs by complement.

! 84! In addition, Nkap-deficient T cells exhibited decreased sialylation of surface glycoproteins, which was linked to increased IgM binding and complement activation. Hsu et al. argued that this is consistent with ImmGen transcriptome profiles showing that expression of the

α2,8 sialyltransferase St8sia6 is strongly upregulated (~6-fold) in mature CD4SP compared to semi-mature CD4SP thymocytes. More generally, the example of St8sia6 lends conceptual support to the idea that transcriptional changes during semi-mature to mature SP thymocyte differentiation contribute to the T cell maturation process together with events occurring post- thymic emigration. Consistent with that notion, a substantial number of genes are concordantly up- or downregulated both pre-egress (mature vs. semi-mature SP) and post-egress (naïve T cells vs. mature SP) (data from www.immgen.org). Interestingly, a number of Zfp335 direct targets are among the set of genes differentially regulated during post-egress maturation: Rabggtb (up

~2-fold in CD4+), Tsn, Atp5l and Sec61b (down ~2-fold in CD8+); it may be of interest to investigate the effect of Zfp335 on the expression of these genes and their potential contributions to mature T cell formation.

We explored the possibility that complement-mediated clearance may explain the peripheral T cell deficiency in blt/blt mice, similar to what had been reported for Nkapfl/fl CD4-

Cre mice (Hsu et al., 2014). However, our preliminary findings indicated no significant decrease in surface CD55 staining on blt/blt RTEs and mature naïve T cells, or any increase in binding of the lectin PNA, suggesting that sialic acid incorporation at the cell surface was not defective

(data not shown). In a mixed chimeric setting, there were no significant differences in C3 deposition on blt/blt naïve T cells compared to wild-type cells in the same animal (data not shown). Moreover, we note that thymic accumulation of blt/blt CD62LhiCD24lo SP cells following prolonged blockade of egress was greatly reduced (Figure 3E), even though they had

! 85! not entered the circulation where they would have been more exposed to the effects of complement. Thus, so far, our data do not support the hypothesis that defective protection from complement plays a major role in the loss of blt/blt T cells.

Of the genes that were found to be downregulated after thymic egress (Mingueneau et al.,

2013), many were related to the cell cycle; this is consistent with studies showing that mature SP thymocytes can undergo some limited cell division (Pénit and Vasseur, 1997; Le Campion et al.,

2000; McCaughtry et al., 2007), although this is only evident in a small fraction of the population at any given point in time and it is not clear what its functional significance may be, since proliferation is not a necessary precondition for thymic egress (Pénit and Vasseur, 1997).

Interestingly, our microarray profiling data revealed that many of the cell cycle-associated genes that were downregulated after thymic egress (according to ImmGen datasets), for example

Ccnb2 and Mki67, showed significant upregulation in blt/blt mature CD4SP thymocytes (fold- change = 2.4x for Ccnb2, 1.8x for Mki67; unpublished data).

This observation may seem paradoxical given that its apparent prediction – that blt/blt cells undergo increased proliferation – is clearly not supported by our experimental data. In fact, our assessment of proliferation by two independent approaches, namely BrdU pulse labeling and

DNA content analysis, indicated a trend towards slightly reduced frequencies of blt/blt mature

SP thymocytes in cell cycle (Figure 4 – figure supplement 3A, B). These data, in conjunction with previous findings of severe proliferation defects in Zfp335-deficient human lymphoblastic cells and neuronal stem cells (Yang et al., 2012), suggest that we cannot dismiss the possibility that Zfp335R1092W may have a modest cell cycle effect. However, based on the consensus in the field that the vast majority of SP thymocytes and naïve T cells are quiescent (Surh and Sprent,

2008), together with our own observations that Rag1-GFPhi RTEs do not exhibit a significant

! 86! degree of proliferation (data not shown), we argue that this small effect is unlikely to contribute significantly to the overall T cell defect in blt/blt mice. As for the elevated expression of cell cycle genes observed in blt/blt mature SP thymocytes, we interpret this to be consistent with a molecular signature of impaired maturation.

Recent work from various groups has shown that autophagy plays an essential role in T cell homeostasis (Pua et al., 2007; 2009; Parekh et al., 2013; Afzal et al., 2015). In these studies, mice deficient in various components of the autophagy machinery exhibited phenotypes that were in some aspects similar to that seen in blt/blt mice – normal DN and DP thymocyte numbers, small but reproducible decreases in SP thymocytes, and a more dramatic reduction in peripheral naïve T cells. However, Atg5- and Atg7-deficient T cells also showed enhanced apoptosis and impaired TCR-induced proliferation (Pua et al., 2007; 2009), which was not seen in blt/blt T cells, suggesting that any autophagy defects – if they do exist – are probably much less severe than in the Atg5 and Atg7 knockout models. In addition, expression profiling of blt/blt mature CD4SP thymocytes and RTEs did not reveal significant alterations in major canonical autophagy-related genes, although Gene Set Enrichment Analysis (GSEA) did suggest an enrichment for lysosomal genes downregulated in blt/blt cells (data not shown), which raises the speculative possibility that defects in lysosome function may affect autophagosome formation and contribute to impaired T cell maturation.

It has been reported that mitochondrial content is developmentally regulated during T cell maturation – highest in DP thymocytes, followed by SP thymocytes, and lowest in peripheral T cells (Pua et al., 2009). Autophagy was required for mitochondrial clearance during the transition from the thymus to the periphery, and the failure of Atg7−/− T cells to clear excess mitochondria was thought to be the cause of increased cell death. The authors propose that this may result from

! 87! the overproduction of toxic reactive oxygen species (ROS) as T cells transition into the periphery, since peripheral blood has a much higher oxygen tension than the thymus; they also suggest that an imbalance in pro- and anti-apoptotic mitochondrial proteins may contribute to cell death. However, preliminary experiments failed to detect aberrant mitochondrial content or

ROS levels in blt/blt mature SP thymocytes (data not shown), although we cannot rule out the possibility that there may be defects in other autophagy-dependent processes. For instance, autophagy plays a critical role in maintaining cellular health by preventing the accumulation of damaged organelles and protein aggregates. In addition, naïve T cells are known to rely mostly on mitochondrial ATP generation, and it has been hypothesized that autophagy may help support mitochondrial respiration in T cells by generating substrates for fatty acid oxidation through lipid degradation and providing mitochondrial quality control (Puleston et al., 2014). Whether autophagy contributes to the block in blt/blt T cell maturation will be interesting to investigate in future studies, perhaps through experimental approaches that modulate autophagy in vivo – for instance, by treating mice with the mTOR-independent autophagy-inducing compound spermidine (Puleston et al., 2014).

Deciphering the function of Zfp335 through its interaction partners

Previous work on the biochemical purification and analysis of Zfp335-containing protein complexes from human cell lines and mouse embryonic brain tissue (Garapaty et al., 2009; Yang et al., 2012) yielded a large number of candidate interaction partners, of which some have formed the basis for current speculation as to the molecular function(s) of Zfp335. One of these interaction partners is Ki-67, a protein that is highly expressed during the G1/S/G2/M phases of the cell cycle, and as such is widely used as a marker to assess proliferation (Scholzen and

! 88! Gerdes, 2000). Ki-67 is predominantly localized in the nucleolus during interphase (Kill, 1996;

Verheijen et al., 1989), but relocates to the surface of condensed chromosomes at the onset of mitosis (Scholzen and Gerdes, 2000). Despite its well-established connection to cell proliferation, for a long time the functional significance of Ki-67 remained unclear, although it has been proposed to aid in stabilizing the mitotic spindle (Vanneste et al., 2009) or in regulating chromatin structure (Scholzen et al., 2002). More recently, it has been reported that Ki-67 is required for assembly of the perichromosomal layer. Depletion of Ki-67 resulted in the failure of nucleolar proteins to associate with the chromosome periphery during mitosis, a process that the authors suggest is important for efficient nucleolar assembly during mitotic exit and successful completion of subsequent cell divisions (Booth et al., 2014).

Ki-67 was first found to interact with Zfp335 in a proteomics analysis using HeLa S3 cells (Garapaty et al., 2009). This finding was later corroborated by Yang et al., who also showed that the R1111H mutation in ZNF335 led to decreased Ki-67 binding, with the implication that this may contribute to the growth defects seen in ZNF335R1111H lymphoblastic cell lines, although no further evidence was presented that explicitly tested this hypothesis (Yang et al.,

2012). It should be noted, however, that Yang et al. did not provide details on the experimental system used for obtaining these data (e.g. cell line, overexpressed vs. endogenous ZNF335 and

Ki-67), and it is unclear whether the decreased pulldown of Ki-67 by ZNF335R1111H was due to the loss of a direct protein-protein interaction, or if ZNF335/ Ki-67 co-immunoprecipitation occurred through non-specific, “sticky” binding to genomic DNA. In the latter case, the apparent loss of Ki-67 interaction with ZNF335R1111H may have been due to an inability of the mutant protein to bind DNA.

! 89! Based on the Garapaty et al. data, we decided to explore the possibility that the bloto mutation may affect a hypothetical interaction between Zfp335 and Ki-67. In initial experiments, co-immunoprecipitation assays using nuclear extracts from WEHI-231 B cells retrovirally transduced with either FLAG-Zfp335WT or FLAG-Zfp335R1092W suggested that both mutant and wild-type proteins were able to pulldown endogenously expressed Ki-67 (data not shown), although technical challenges with the sensitivity of endogenous Ki-67 detection and in normalizing Zfp335WT and Zfp335R1092W expression levels made it difficult to draw conclusions about quantitative differences in Ki-67 binding. Confocal imaging of total thymocytes showed that although there was partial overlap, Zfp335 and Ki-67 had distinct staining patterns within the nucleus – Zfp335 clustered in punctate foci, whereas Ki-67 was more broadly distributed

(data not shown), which may not be surprising in view of what we now know about the role of

Ki-67 as a nuclear/ nucleolar organizer (Booth et al., 2014), as opposed to that of Zfp335 as a promoter-binding transcription factor.

By intracellular flow cytometry, we did not detect any changes in the frequencies of Ki-

67+ cells within DN, DP or semi-mature SP thymocyte populations from blt/blt mice compared to blt/+ controls. However, for CD4 and CD8 mature SP thymocytes, there was a significant decrease (~1.8-fold) in the Ki-67+ population. Interestingly, the Ki-67+ population in blt/blt mature SP thymocytes was strongly enriched for cells with the highest Ki-67 staining intensity, whereas the same gate in blt/+ controls had a distribution that was more weighted towards Ki-

67int cells (data not shown). This enrichment for Ki-67hi cells is consistent with our gene expression data indicating higher Mki67 mRNA levels in blt/blt mature SP thymocytes, despite an overall decrease in the Ki-67+ subset. As a possible explanation for this decrease, one might posit that the R1092W mutation affects Zfp335 binding to Ki-67 and that disruption of this

! 90! interaction destabilizes Ki-67. However, such a model would seem inconsistent with data indicating that Ki-67+ blt/blt mature SP thymocytes actually express on average more Ki-67.

A more plausible way to reconcile these findings would be to interpret our Ki-67 staining data in the context of SP thymocyte maturation. In this model, some mature SP thymocytes enter the cell cycle transiently, as discussed earlier in this chapter and shown by various studies (Pénit and Vasseur, 1997; Le Campion et al., 2000), and in doing so they become Ki-67hi. We assume that these cells exit the cell cycle rapidly and return to their default quiescent state, with concurrent downregulation of Ki-67. Therefore, cells within the Ki-67+ gate progress from Ki-

67hi (proliferating) to Ki-67int (post-cell cycle exit) before becoming Ki-67−. This implies that Ki-

67int cells have spent more time in the medulla and are more “mature” relative to the Ki-67hi subset; this notion is supported by preliminary evidence indicating slightly lower Rag1-GFP and

CD24 expression on Ki-67int vs. Ki-67hi mature SP thymocytes (data not shown). Hence, we speculate that the depletion of Ki-67int cells and overall decrease in Ki-67+ frequencies is most likely a side-effect of the maturation defect in blt/blt SP thymocytes, and not the direct outcome of altered Zfp335/Ki-67 interaction.

In their 2009 study, Garapaty et al. identified RBBP5, ASH2L and WDR5 as Zfp335- associated proteins. All three proteins are core components of histone methyltransferase complexes, which led the authors to postulate the association of Zfp335 with histone methyltransferase activity. This was supported by in vitro assays that demonstrated H3-specific methyltransferase activity on purified histone octamers; however, this appeared not to be H3K4- specific. Furthermore, Western blot analysis of Zfp335 complexes did not detect H3K4 methyltransferases such as SET1, MLL1, MLL3, MLL4 (MLL2, SET1A and SET1B were not assessed), although the authors suggested that this may have been due to the stringent conditions

! 91! used during protein purification (Garapaty et al., 2009). Hence, they were unable to determine the exact nature of the histone methyltransferase activity and identity of the enzymatic component(s) within Zfp335-associated complexes.

The 2012 study of Yang et al. replicated the RBBP5/ ASH2L/ WDR5 interactions originally reported by Garapaty et al.; in addition, they found new evidence for co- immunoprecipitation of MLL1, an H3K4 methyltransferase. This led the authors to propose that

Zfp335 associates with proteins of the H3K4 methyltransferase complex, thereby recruiting it to target promoters to activate gene expression (Yang et al., 2012). However, they did not address the inconsistency between their model and Garapaty et al.’s prior finding that Zfp335 complexes lacked H3K4-specific methyltransferase activity.

We attempted to reproduce some of the published interactions and determine if they were affected by the bloto mutation. However, in our hands, we were unable to detect co-IP of RBBP5 or WDR5 with Zfp335 (data not shown), suggesting that these interactions may not be robust to variability in experimental conditions. Based on the limited data available at this point, we suggest that recruitment of H3K4 methyltransferases is likely not the only mechanism by which

Zfp335 regulates transcription. In fact, the IP/MS data sets published by Yang et al. contain many other potentially interesting hits (e.g. CHD4, RBBP4, SETD2) that suggest possible involvement of other histone modifying or chromatin remodeling complexes. For example,

CHD4 and RBBP4 are both components of the NuRD (nucleosome remodeling and deacetylation) complex (Clapier and Cairns, 2009), and RBBP4 also forms part of the NURF complex, which interestingly has been implicated in late-stage thymocyte development (Landry et al., 2011). Likewise, Zfp335 may not function solely as a transcriptional activator – we have also noted that at least some target genes, most notably Zfp335 itself, appear to be modestly

! 92! upregulated in blt/blt thymocytes (Figure 6 – figure supplement 1), suggesting that Zfp335 may also act as a repressor depending on the context. In future studies, identification and characterization of interaction partners and other co-binding transcription factors may help elucidate the mechanisms by which Zfp335 regulates gene expression.

How does the bloto mutation affect Zfp335 function?

In Chapter Two, we present various lines of evidence indicating that the bloto (R1092W) mutation results in a hypomorphic allele of Zfp335 that does not have a deleterious effect on steady-state transcript levels (Figure 2D), protein stability (Figure 2E) or nuclear localization

(Figure 2F), but is able to partially rescue the bloto T cell defect when overexpressed (Figure

2H). Consistent with our structural prediction that Arg1092 is a DNA-binding residue (Figure 2

– figure supplement 2), we showed that the R1092W mutation diminished the ability of Zfp335 to bind to some of its target sites (Figure 6). Furthermore, we have thus far been unable to detect clear differences in protein-protein interactions for Zfp335R1092W vs. Zfp335WT, either through semi-quantitative IP/MS approaches (in collaboration with Nevan Krogan’s laboratory) or co-IP analysis of selected candidates (data not shown). On the whole, these subtle effects on protein function are consistent with the comparatively mild phenotype observed in blt/blt mice, which appear to be essentially healthy apart from the selective T cell maturation defect described in our work. In contrast, null alleles of Zfp335 are associated with far more severe phenotypes: from the early embryonic lethality resulting from global deletion to the extensive neurodevelopmental abnormalities and perinatal mortality caused by brain-specific ablation (Yang et al., 2012). Thus,

Zfp335R1092W retains sufficient wild-type activity to accomplish essential functions in early

! 93! development, neurogenesis and most likely other aspects of development that are yet to be studied, but not enough to drive normal T cell maturation.

We believe that the hypomorphic properties of the Zfp335R1092W allele can be primarily attributed to the fact that it does not lead to a uniform loss of Zfp335 occupancy across all binding sites in the genome. Consistent with this, we found that Zfp335R1092W bound the bipartite consensus motif similarly to wild-type Zfp335 in gel shift assays (data not shown). Our ChIP-seq data indicated that the effect of the bloto mutation on Zfp335 occupancy varied greatly between different target sites, with some exhibiting a dramatic reduction in binding, others that seemed completely unaffected, and the remainder showing a continuum of effects. It is still unclear how the bloto mutation exerts this selective effect on Zfp335 binding. To address this question, we propose two hypothetical models, summarized in Figure 10: 1) The mutation may disrupt the ability of Zfp335 to interact with some other protein(s), perhaps a co-binding transcription factor or co-activator, that is required for stabilizing Zfp335 engagement at a subset of target sites or within certain local chromatin environments, but not under other conditions. 2) Being a large protein with multiple zinc finger clusters, Zfp335 may have secondary DNA binding specificities that modulate interaction with the core consensus motif, analogous to what has been reported for

CTCF, another multi-domain C2H2 zinc finger protein (Nakahashi et al., 2013). We are currently investigating these possibilities.

! 94! 1) Assisted binding through protein-protein interactions

Zfp335 X Zfp335 X

2) Modulation by secondary DNA binding motifs

Zfp335 R1092 Zfp335 consensus motif secondary motif Factor X binding site

Figure 10. Proposed models for how the bloto (R1092W) mutation affects Zfp335 binding to select target sites.

Diagram depicting hypothetical models of Zfp335 binding to sites where Zfp335 occupancy is affected by the R1092W mutation: 1) Other proteins (e.g. Factor X) help stabilize Zfp335 binding to its motif, which may be a low- quality match for the consensus sequence (left); or, in the absence of its motif, Zfp335 is recruited to the site by interactions with Factor X (right). R1092 is critical for these protein- protein interactions, and mutating it results in loss of Zfp335 binding. 2) Zfp335 has additional DNA binding specificities, and R1092 is involved in recognizing secondary motifs which influence the engagement of Zfp335 with its consensus motif.

! 95! Elevated constitutive type I interferon signature in blt/blt mice

In addition to the T cell maturation phenotype described in our published work, we have also found evidence for increased steady state levels of type I interferon (IFN) in blt/blt mice, in the absence of acute infection. Transcriptome profiling of mature CD4SP thymocytes revealed a strong type I IFN gene expression signature in blt/blt cells relative to blt/+ controls, with significantly increased expression of type I IFN-stimulated genes (ISGs) such as Oas1a, Mx1,

Ly6a and Usp18, among many others (data not shown). This phenotype was cell-extrinsic: in mixed BM chimeric mice, where they were exposed to the same cytokine milieu, wild-type SP thymocytes upregulated ISGs to the same extent as blt/blt cells (data not shown). Preliminary findings indicate that ISG upregulation in blt/blt cells is dependent on type I IFN signaling: compared to blt/+ controls, blt/blt mature SP thymocytes consistently showed higher surface expression of the type I IFN-induced molecules Ly6A and Qa2, but this difference was eliminated on an IFNAR1-deficient background (data not shown). We believe that our data are consistent with the model that blt/blt mice have higher levels of constitutively secreted type I

IFNs in the local tissue microenvironments and/or sytemic circulation, although definitive proof would require direct assessment of type I IFN expression, which would be technically challenging because baseline IFN expression levels are very low and close to the threshold of detection of most assays (Gough et al., 2012).

We have found that this phenotype is recapitulated in lethally irradiated wild-type mice reconstituted with blt/blt bone marrow but not in reciprocal chimeras (wild-type BM ! blt/blt hosts), suggesting that the sources of elevated IFN in blt/blt mice are hematopoietic in origin

(data not shown). However, these sources have not yet been identified. Initial experiments provide evidence for interferon-stimulated Ly6A upregulation in blt/blt SP thymocytes and naïve

! 96! T cells, suggesting that cellular sources of elevated IFNs are present in the thymus as well as the periphery. Given that type I IFN can be secreted by a diverse range of cell types and can regulate its own production through a positive feedback loop (Ivashkiv and Donlin, 2013), it is likely that multiple distinct sources are involved. It is also possible that the overall IFN increase is an aggregate of relatively small contributions from individual populations, making it potentially challenging to isolate and identify the specific cell types responsible.

In addition to their well-studied role in anti-viral defense, type I IFNs are pleiotropic cytokines with a broad range of immunomodulatory functions. Several studies have shown that constitutive type I IFN is important for maintaining homeostasis of multiple immune lineages.

For example, Ifnar1−/− mice have decreased splenic NK cells (Swann et al., 2007) and increased frequencies of Gr-1+ and CD11b+ myeloid cells in peripheral blood (Hwang et al., 1995).

Constitutive type I IFN signaling is also thought to play a role in maintaining the hematopoietic stem cell (HSC) niche (Essers et al., 2009). Thus, in our studies of blt/blt mice, we have found that it is important to distinguish between immunological changes arising from Zfp335-regulated cell-autonomous mechanisms versus those that are secondary to the phenomenon of elevated tonic type I IFN signaling. For this reason, we have made use of mixed hematopoietic chimeras to validate phenotypes originally discovered in intact mice. The T cell maturation phenotype described in our published work was similarly evident in mixed chimeras (Figure 1F; Figure 1 – figure supplement 1A), indicating that increased type I IFN signaling in blt/blt mice was not sufficient to cause this defect. Conversely, we have observed increased frequencies of

CD11b+CD115+ monocytes in blt/blt mice but not in mixed chimeras (data not shown), which suggests this may be a secondary phenotype driven by elevated IFN, given the known role of type I IFNs in monocyte differentiation (Lee et al., 2009).

! 97! The mechanisms for how Zfp335 influences basal type I IFN expression are at this point completely unknown and open to speculation. Here, we discuss a few speculative models that could be explored in future studies:

It is now thought that many KRAB domain-containing tandem C2H2 zinc finger proteins

(KZNFs) may have evolved to preserve genomic integrity by repressing various classes of endogenous retroelements (Jacobs et al., 2014; Thomas and Schneider, 2011). It is tempting to speculate that Zfp335 may serve a similar function, such that defective retroelement repression by Zfp335R1092W leads to the accumulation of retrotransposon-derived RNA and DNA in the cytoplasm, which activates cytosolic nucleic acid sensing pathways and triggers type I IFN production. These pathways may also become activated as a consequence of defects in autophagy. Recent studies have shown that autophagosomes mediate the degradation of cytoplasmic nucleic acids arising from a variety of sources – phagocytosed material, damaged and discarded chromosomal DNA (Lan et al., 2014), or retrotransposon intermediates (Guo et al.,

2014). Mitochondrial DNA has also been shown to activate cytosolic DNA sensing pathways

(Rongvaux et al., 2014; West et al., 2000), hence defective autophagic clearance of damaged mitochondria may provide another link between autophagy and type I IFN induction. Moreover, this model would be consistent with our hypothesis, as discussed earlier in this chapter, that perturbations in autophagy may contribute to maturational defects in blt/blt T cells.

Lastly, it is possible that Zfp335 may regulate constitutive type I IFNs at the transcriptional level. Although we did not detect Zfp335 binding to type I IFN genes in thymocytes, this may not necessarily hold true for other cell types. Zfp335 may also have an indirect role as part of a wider gene regulatory network. For instance, it would be interesting to investigate the possibility of cross-talk between Zfp335 and the transcription factors AP-1 and

! 98! NF-κB, which have been reported to be essential for constitutive type I IFN production (Gough et al., 2012).

Relevance of Zfp335 to human immunological disease

Zfp335 is broadly expressed in various immune cell lineages, including B cells, NK cells and myeloid cells (www.immgen.org). Given our published work demonstrating the significance of Zfp335 in T cell development and preliminary data suggesting an additional role in type I interferon regulation, it would be important to investigate possible connections to diseases of the human immune system – immune deficiencies, autoimmune and inflammatory disorders, or hematological malignancies. To date, no studies have directly implicated ZNF335 in immunological diseases. This may not be so surprising: Zfp335 has been shown to be essential for early embryonic development in mice, and a severely hypomorphic mutant ZNF335 allele in human patients causes extreme microcephaly and death before the age of one (Yang et al., 2012), so it is conceivable that many loss-of-function mutations in ZNF335 might have such serious consequences that patients do not survive long enough for immune abnormalities to be clinically recognized. However, as future studies better define the functional domains of ZNF335, as well as the cis-regulatory elements that control its expression in various cell types, new insights are likely to emerge on how ZNF335 may contribute to immune disease.

! 99! References

Afzal, S., Z. Hao, M. Itsumi, Y. Abouelkheer, D. Brenner, Y. Gao, A. Wakeham, C. Hong, W.Y. Li, J. Sylvester, S.O. Gilani, A. Brüstle, J. Haight, A.J. You-Ten, G.H.Y. Lin, S. Inoue, and T.W. Mak. 2015. Autophagy-independent functions of UVRAG are essential for peripheral naive T-cell homeostasis. Proc Natl Acad Sci U S A. doi:10.1073/pnas.1423588112.

Anderson, M.S., E.S. Venanzi, Z. Chen, S.P. Berzins, C. Benoist, and D. Mathis. 2005. The Cellular Mechanism of Aire Control of T Cell Tolerance. Immunity. 23:227–239. doi:10.1016/j.immuni.2005.07.005.

Andrews, T.D., B. Whittle, M.A. Field, B. Balakishnan, Y. Zhang, Y. Shao, V. Cho, M. Kirk, M. Singh, Y. Xia, J. Hager, S. Winslade, G. Sjollema, B. Beutler, A. Enders, and C.C. Goodnow. 2012. Massively parallel sequencing of the mouse exome to accurately identify rare, induced mutations: an immediate source for thousands of new mouse models. Open Biol. 2:120061. doi:10.1098/rsob.120061.

Arbonés, M.L., D.C. Ord, K. Ley, H. Ratech, C. Maynard-Curry, G. Otten, D.J. Capon, and T.F. Tedder. 1994. Lymphocyte homing and leukocyte rolling and migration are impaired in L- selectin-deficient mice. Immunity. 1:247–260.

Asencio, C., I.F. Davidson, R. Santarella-Mellwig, T.B.N. Ly-Hartig, M. Mall, M.R. Wallenfang, I.W. Mattaj, and M. Gorjánácz. 2012. Coordination of kinase and phosphatase activities by Lem4 enables nuclear envelope reassembly during mitosis. Cell. 150:122–135. doi:10.1016/j.cell.2012.04.043.

Bailey, T.L., M. Boden, F.A. Buske, M. Frith, C.E. Grant, L. Clementi, J. Ren, W.W. Li, and W.S. Noble. 2009. MEME SUITE: tools for motif discovery and searching. Nucleic Acids Research. 37:W202–W208. doi:10.1093/nar/gkp335.

Baldwin, T.A., and K.A. Hogquist. 2007. Transcriptional analysis of clonal deletion in vivo. J. Immunol. 179:837–844.

Booth, D.G., M. Takagi, L. Sanchez-Pulido, E. Petfalski, G. Vargiu, K. Samejima, N. Imamoto, C.P. Ponting, D. Tollervey, W.C. Earnshaw, P. Vagnarelli, and A.A. Hyman. 2014. Ki-67 is a PP1-interacting protein that organises the mitotic chromosome periphery. eLife. 3. doi:10.7554/eLife.01641.

Boursalian, T.E., J. Golob, D.M. Soper, C.J. Cooper, and P.J. Fink. 2004. Continued maturation of thymic emigrants in the periphery. Nat Immunol. 5:418–425. doi:10.1038/ni1049.

Brayer, K.J., and D.J. Segal. 2008. Keep Your Fingers Off My DNA: Protein–Protein Interactions Mediated by C2H2 Zinc Finger Domains. Cell Biochem Biophys. 50:111–131. doi:10.1007/s12013-008-9008-5.

Brown, R.S. 2005. Zinc finger proteins: getting a grip on RNA. Current Opinion in Structural Biology. 15:94–98. doi:10.1016/j.sbi.2005.01.006.

! 100! Campbell, J.J., J. Pan, and E.C. Butcher. 1999. Cutting edge: developmental switches in chemokine responses during T cell maturation. J. Immunol. 163:2353–2357.

Carlson, C.M., B.T. Endrizzi, J. Wu, X. Ding, M.A. Weinreich, E.R. Walsh, M.A. Wani, J.B. Lingrel, K.A. Hogquist, and S.C. Jameson. 2006. Kruppel-like factor 2 regulates thymocyte and T-cell migration. Nature. 442:299–302. doi:10.1038/nature04882.

Choi, Y.I., J.S. Duke-Cohan, W.B. Ahmed, M.A. Handley, F. Mann, J.A. Epstein, L.K. Clayton, and E.L. Reinherz. 2008. PlexinD1 Glycoprotein Controls Migration of Positively Selected Thymocytes into the Medulla. Immunity. 29:888–898. doi:10.1016/j.immuni.2008.10.008.

Clapier, C.R., and B.R. Cairns. 2009. The biology of chromatin remodeling complexes. Annu. Rev. Biochem. 78:273–304. doi:10.1146/annurev.biochem.77.062706.153223.

Cobb, B.S., S. Morales-Alcelay, G. Kleiger, K.E. Brown, A.G. Fisher, and S.T. Smale. 2000. Targeting of Ikaros to pericentromeric heterochromatin by direct DNA binding. Genes & Development. 14:2146–2160.

Consortium, T.E.P. 2012. An integrated encyclopedia of DNA elements in the human genome. Nature. 489:57–74. doi:doi:10.1038/nature11247.

Dai, M., P. Wang, A.D. Boyd, G. Kostov, B. Athey, E.G. Jones, W.E. Bunney, R.M. Myers, T.P. Speed, H. Akil, S.J. Watson, and F. Meng. 2005. Evolving gene/transcript definitions significantly alter the interpretation of GeneChip data. Nucleic Acids Research. 33:e175. doi:10.1093/nar/gni179.

Essers, M.A.G., S. Offner, W.E. Blanco-Bose, Z. Waibler, U. Kalinke, M.A. Duchosal, and A. Trumpp. 2009. IFNα activates dormant haematopoietic stem cells in vivo. Nature. 458:904– 908. doi:10.1038/nature07815.

Fabre, S., F. Carrette, J. Chen, V. Lang, M. Semichon, C. Denoyelle, V. Lazar, N. Cagnard, A. Dubart-Kupperschmitt, M. Mangeney, D.A. Fruman, and G. Bismuth. 2008. FOXO1 regulates L-Selectin and a network of human T cell homing molecules downstream of phosphatidylinositol 3-kinase. The Journal of Immunology. 181:2980–2989.

Fink, P.J. 2012. The Biology of Recent Thymic Emigrants. Annu. Rev. Immunol. 31:121102143727003. doi:10.1146/annurev-immunol-032712-100010.

Fink, P.J., and D.W. Hendricks. 2011. Post-thymic maturation: young T cells assert their individuality. Nat Rev Immunol. 11:544–549. doi:10.1038/nri3028.

Garapaty, S., C.-F. Xu, P. Trojer, M.A. Mahajan, T.A. Neubert, and H.H. Samuels. 2009. Identification and characterization of a novel nuclear protein complex involved in nuclear hormone receptor-mediated gene regulation. J. Biol. Chem. 284:7542–7552. doi:10.1074/jbc.M805872200.

Gautier, L., L. Cope, B.M. Bolstad, and R.A. Irizarry. 2004. affy--analysis of Affymetrix GeneChip data at the probe level. Bioinformatics. 20:307–315.

! 101! doi:10.1093/bioinformatics/btg405.

Gough, D.J., N.L. Messina, C.J.P. Clarke, R.W. Johnstone, and D.E. Levy. 2012. Constitutive Type I Interferon Modulates Homeostatic Balance through Tonic Signaling. Immunity. 36:166–174. doi:10.1016/j.immuni.2012.01.011.

Green, J.A., K. Suzuki, B. Cho, L.D. Willison, D. Palmer, C.D.C. Allen, T.H. Schmidt, Y. Xu, R.L. Proia, S.R. Coughlin, and J.G. Cyster. 2011. The sphingosine 1-phosphate receptor S1P2 maintains the homeostasis of germinal center B cells and promotes niche confinement. Nat Immunol. 12:672–680. doi:10.1038/ni.2047.

Gubbels Bupp, M.R., B. Edwards, C. Guo, D. Wei, G. Chen, B. Wong, E. Masteller, and S.L. Peng. 2009. T cells require Foxo1to populate the peripheral lymphoid organs. Eur. J. Immunol. 39:2991–2999. doi:10.1002/eji.200939427.

Guerin, S., M.-L. Baron, R. Valero, M. Herrant, P. Auberger, and P. Naquet. 2002. RelB reduces thymocyte apoptosis and regulates terminal thymocyte maturation. Eur. J. Immunol. 32:1–9. doi:10.1002/1521-4141(200201)32:1<1::AID-IMMU1>3.0.CO;2-S.

Guo, H., M. Chitiprolu, D. Gagnon, L. Meng, C. Perez-Iratxeta, D. Lagace, and D. Gibbings. 2014. Autophagy supports genomic stability by degrading retrotransposon RNA. Nat Comms. 5:5276–. doi:10.1038/ncomms6276.

He, Y.-W., M.-X. He, I. Mcleod, and W. Jia. 2012. Macroautophagy in T Lymphocyte Development and Function. Frontiers in Immunology. 3. doi:10.3389/fimmu.2012.00022.

Heinz, S., C. Benner, N. Spann, E. Bertolino, Y.C. Lin, P. Laslo, J.X. Cheng, C. Murre, H. Singh, and C.K. Glass. 2010. Simple Combinations of Lineage-Determining Transcription Factors Prime cis-Regulatory Elements Required for Macrophage and B Cell Identities. Molecular Cell. 38:576–589. doi:10.1016/j.molcel.2010.05.004.

Hendricks, D.W., and P.J. Fink. 2011. Recent thymic emigrants are biased against the T-helper type 1 and toward the T-helper type 2 effector lineage. Blood. 117:1239–1249. doi:10.1182/blood-2010-07-299263.

Heng, T.S.P., M.W. Painter, K. Elpek, V. Lukacs-Kornek, N. Mauermann, S.J. Turley, D. Koller, F.S. Kim, A.J. Wagers, N. Asinovski, S. Davis, M. Fassett, M. Feuerer, D.H.D. Gray, S. Haxhinasto, J.A. Hill, G. Hyatt, C. Laplace, K. Leatherbee, D. Mathis, C. Benoist, R. Jianu, D.H. Laidlaw, J.A. Best, J. Knell, A.W. Goldrath, J. Jarjoura, J.C. Sun, Y. Zhu, L.L. Lanier, A. Ergun, Z. Li, J.J. Collins, S.A. Shinton, R.R. Hardy, R. Friedline, K. Sylvia, and J. Kang. 2008. The Immunological Genome Project: networks of gene expression in immune cells. Nat Immunol. 9:1091–1094. doi:10.1038/ni1008-1091.

Hogquist, K.A., T.A. Baldwin, and S.C. Jameson. 2005. Central tolerance: learning self-control in the thymus. Nat Rev Immunol. 5:772–782. doi:10.1038/nri1707.

Houston, E.G., L.E. Higdon, and P.J. Fink. 2011. Recent thymic emigrants are preferentially incorporated only into the depleted T-cell pool. Proc Natl Acad Sci U S A. 108:5366–5371.

! 102! doi:10.1073/pnas.1015286108.

Hsu, F.-C., A.G. Pajerowski, M. Nelson-Holte, R. Sundsbak, and V.S. Shapiro. 2011. NKAP is required for T cell maturation and acquisition of functional competency. Journal of Experimental Medicine. 208:1291–1304. doi:10.1084/jem.20101874.

Hsu, F.-C., M.J. Shapiro, M.W. Chen, D.C. McWilliams, L.M. Seaburg, S.N. Tangen, and V.S. Shapiro. 2014. Immature recent thymic emigrants are eliminated by complement. The Journal of Immunology. 193:6005–6015. doi:10.4049/jimmunol.1401871.

Hwang, S.Y., P.J. Hertzog, K.A. Holland, S.H. Sumarsono, M.J. Tymms, J.A. Hamilton, G. Whitty, I. Bertoncello, and I. Kola. 1995. A null mutation in the gene encoding a type I interferon receptor component eliminates antiproliferative and antiviral responses to interferons alpha and beta and alters macrophage responses. Proc Natl Acad Sci U S A. 92:11284–11288.

Iuchi, S. 2005. C2H2 Zinc Fingers As DNA Binding Domains. In link.springer.com. Springer US, Boston, MA. 7–13.

Ivashkiv, L.B., and L.T. Donlin. 2013. Regulation of type I interferon responses. Nat Rev Immunol. 14:36–49. doi:10.1038/nri3581.

Jacobs, F.M.J., D. Greenberg, N. Nguyen, M. Haeussler, A.D. Ewing, S. Katzman, B. Paten, S.R. Salama, and D. Haussler. 2014. An evolutionary arms race between KRAB zinc-finger genes ZNF91/93 and SVA/L1 retrotransposons. Nature. 516:242–245. doi:10.1038/nature13760.

Kent, W.J., A.S. Zweig, G. Barber, A.S. Hinrichs, and D. Karolchik. 2010. BigWig and BigBed: enabling browsing of large distributed datasets. Bioinformatics. 26:2204–2207. doi:10.1093/bioinformatics/btq351.

Kent, W.J., C.W. Sugnet, T.S. Furey, K.M. Roskin, T.H. Pringle, A.M. Zahler, and A.D. Haussler. 2002. The Human Genome Browser at UCSC. Genome Research. 12:996–1006. doi:10.1101/gr.229102.

Kerdiles, Y.M., D.R. Beisner, R. Tinoco, A.S. Dejean, D.H. Castrillon, R.A. DePinho, and S.M. Hedrick. 2009. Foxo1 links homing and survival of naive T cells by regulating L-selectin, CCR7 and interleukin 7 receptor. Nat Immunol. 10:176–184. doi:10.1038/ni.1689.

Kill, I.R. 1996. Localisation of the Ki-67 antigen within the nucleolus. Evidence for a fibrillarin- deficient region of the dense fibrillar component. J. Cell. Sci. 109:1253–1263.

Klein, L., B. Kyewski, P.M. Allen, and K.A. Hogquist. 2014. Positive and negative selection of the T cell repertoire: what thymocytes see (and don't see). Nat Rev Immunol. 14:377–391. doi:10.1038/nri3667.

Kuwata, N., H. Igarashi, T. Ohmura, S. Aizawa, and N. Sakaguchi. 1999. Cutting edge: absence of expression of RAG1 in peritoneal B-1 cells detected by knocking into RAG1 locus with green fluorescent protein gene. J. Immunol. 163:6355–6359.

! 103! Lan, Y.Y., D. Londoño, R. Bouley, M.S. Rooney, and N. Hacohen. 2014. Dnase2a Deficiency Uncovers Lysosomal Clearance of Damaged Nuclear DNA via Autophagy. Cell Reports. 9:180–192. doi:10.1016/j.celrep.2014.08.074.

Landry, J.W., S. Banerjee, B. Taylor, P.D. Aplan, A. Singer, and C. Wu. 2011. Chromatin remodeling complex NURF regulates thymocyte maturation. Genes & Development. 25:275–286. doi:10.1101/gad.2007311.

Le Campion, A., F. Vasseur, and C. Pénit. 2000. Regulation and kinetics of premigrant thymocyte expansion. Eur. J. Immunol. 30:738–746.

Lee, P.Y., Y. Li, Y. Kumagai, Y. Xu, J.S. Weinstein, E.S. Kellner, D.C. Nacionales, E.J. Butfiloski, N. van Rooijen, S. Akira, E.S. Sobel, M. Satoh, and W.H. Reeves. 2009. Type I Interferon Modulates Monocyte Recruitment and Maturation in Chronic Inflammation. The American Journal of Pathology. 175:2023–2033. doi:10.2353/ajpath.2009.090328.

Li, H., and R. Durbin. 2009. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics. 25:1754–1760. doi:10.1093/bioinformatics/btp324.

Li, H., B. Handsaker, A. Wysoker, T. Fennell, J. Ruan, N. Homer, G. Marth, G. Abecasis, R. Durbin, 1000 Genome Project Data Processing Subgroup. 2009. The Sequence Alignment/Map format and SAMtools. Bioinformatics. 25:2078–2079. doi:10.1093/bioinformatics/btp352.

Liu, H.H., M. Xie, M.D. Schneider, and Z.J. Chen. 2006. Essential role of TAK1 in thymocyte development and activation. Proc Natl Acad Sci U S A. 103:11677–11682.

Liu, T., J.A. Ortiz, L. Taing, C.A. Meyer, B. Lee, Y. Zhang, H. Shin, S.S. Wong, J. Ma, Y. Lei, U.J. Pape, M. Poidinger, Y. Chen, K. Yeung, M. Brown, Y. Turpaz, and X.S. Liu. 2011. Cistrome: an integrative platform for transcriptional regulation studies. Genome Biol. 12:R83. doi:10.1101/gr.115519.110.

Lo, K., N.R. Landau, and S.T. Smale. 1991. LyF-1, a transcriptional regulator that interacts with a novel class of promoters for lymphocyte-specific genes. Molecular and Cellular Biology. 11:5229–5243.

Mahajan, M.A., A. Murray, and H.H. Samuels. 2002. NRC-interacting factor 1 is a novel cotransducer that interacts with and regulates the activity of the nuclear hormone receptor coactivator NRC. Molecular and Cellular Biology. 22:6883–6894. doi:10.1128/MCB.22.19.6883–6894.2002.

Makaroff, L.E., D.W. Hendricks, R.E. Niec, and P.J. Fink. 2009. Postthymic maturation influences the CD8 T cell response to antigen. Proc Natl Acad Sci U S A. 106:4799–4804. doi:10.1073/pnas.0812354106.

Martin, M. 2011. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet j. 17:10. doi:10.14806/ej.17.1.200.

! 104! Matloubian, M., C.G. Lo, G. Cinamon, M.J. Lesneski, Y. Xu, V. Brinkmann, M.L. Allende, R.L. Proia, and J.G. Cyster. 2004. Lymphocyte egress from thymus and peripheral lymphoid organs is dependent on S1P receptor 1. Nature. 427:355–360. doi:10.1038/nature02284.

McCaughtry, T.M., M.S. Wilken, and K.A. Hogquist. 2007. Thymic emigration revisited. Journal of Experimental Medicine. 204:2513–2520. doi:10.1084/jem.20070601.

McLean, C.Y., D. Bristor, M. Hiller, S.L. Clarke, B.T. Schaar, C.B. Lowe, A.M. Wenger, and G. Bejerano. 2010. GREAT improves functional interpretation of cis-regulatory regions. Nat Biotechnol. 28:495–501. doi:10.1038/nbt.1630.

Mingueneau, M., T. Kreslavsky, D. Gray, T. Heng, R. Cruse, J. Ericson, S. Bendall, M.H. Spitzer, G.P. Nolan, K. Kobayashi, H. von Boehmer, D. Mathis, C. Benoist, A.J. Best, J. Knell, A. Goldrath, V. Jojic, D. Koller, T. Shay, A. Regev, N. Cohen, P. Brennan, M. Brenner, F. Kim, T.N. Rao, A. Wagers, T. Heng, J. Ericson, K. Rothamel, A. Ortiz-Lopez, D. Mathis, C. Benoist, N.A. Bezman, J.C. Sun, G. Min-Oo, C.C. Kim, L.L. Lanier, J. Miller, B. Brown, M. Merad, E.L. Gautier, C. Jakubzick, G.J. Randolph, P. Monach, D.A. Blair, M.L. Dustin, S.A. Shinton, R.R. Hardy, D. Laidlaw, J. Collins, R. Gazit, D.J. Rossi, N. Malhotra, K. Sylvia, J. Kang, T. Kreslavsky, A. Fletcher, K. Elpek, A. Bellemare-Pelletier, D. Malhotra, and S. Turley. 2013. The transcriptional landscape of αβ T cell differentiation. Nat Immunol. 14:619–632. doi:10.1038/ni.2590.

Molnár, A., and K. Georgopoulos. 1994. The Ikaros gene encodes a family of functionally diverse zinc finger DNA-binding proteins. Molecular and Cellular Biology. 14:8292–8303.

Nakahashi, H., K.-R.K. Kwon, W. Resch, L. Vian, M. Dose, D. Stavreva, O. Hakim, N. Pruett, S. Nelson, A. Yamane, J. Qian, W. Dubois, S. Welsh, R.D. Phair, B.F. Pugh, V. Lobanenkov, G.L. Hager, and R. Casellas. 2013. A Genome-wide Map of CTCF Multivalency Redefines the CTCF Code. Cell Reports. 3:1678–1689. doi:10.1016/j.celrep.2013.04.024.

Nelms, K.A., and C.C. Goodnow. 2001. Genome-wide ENU mutagenesis to reveal immune regulators. Immunity. 15:409–418.

Parekh, V.V., L. Wu, K.L. Boyd, J.A. Williams, J.A. Gaddy, D. Olivares-Villagomez, T.L. Cover, W.X. Zong, J. Zhang, and L. Van Kaer. 2013. Impaired Autophagy, Defective T Cell Homeostasis, and a Wasting Syndrome in Mice with a T Cell-Specific Deletion of Vps34. The Journal of Immunology. 190:5086–5101. doi:10.4049/jimmunol.1202071.

Pénit, C., and F. Vasseur. 1997. Expansion of mature thymocyte subsets before emigration to the periphery. J. Immunol. 159:4848–4856.

Pohl, A., and M. Beato. 2014. bwtool: a tool for bigWig files. Bioinformatics. 30:1618–1619. doi:10.1093/bioinformatics/btu056.

Pua, H.H., I. Dzhagalov, M. Chuck, N. Mizushima, and Y.W. He. 2007. A critical role for the autophagy gene Atg5 in T cell survival and proliferation. Journal of Experimental Medicine. 204:25–31. doi:10.1084/jem.20061303.

! 105! Pua, H.H., J. Guo, M. Komatsu, and Y.W. He. 2009. Autophagy Is Essential for Mitochondrial Clearance in Mature T Lymphocytes. The Journal of Immunology. 182:4046–4055. doi:10.4049/jimmunol.0801143.

Puleston, D.J., H. Zhang, T.J. Powell, E. Lipina, S. Sims, I. Panse, A.S. Watson, V. Cerundolo, A.R. Townsend, P. Klenerman, A.K. Simon, and B. Levine. 2014. Autophagy is a critical regulator of memory CD8+ T cell formation. eLife. 3:e03706. doi:10.7554/eLife.03706.

Quinlan, A.R., and I.M. Hall. 2010. BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics. 26:841–842. doi:10.1093/bioinformatics/btq033.

Randall, K.L., T. Lambe, A. Johnson, B. Treanor, E. Kucharska, H. Domaschenz, B. Whittle, L.E. Tze, A. Enders, T.L. Crockford, T. Bouriez-Jones, D. Alston, J.G. Cyster, M.J. Lenardo, F. Mackay, E.K. Deenick, S.G. Tangye, T.D. Chan, T. Camidge, R. Brink, C.G. Vinuesa, F.D. Batista, R.J. Cornall, and C.C. Goodnow. 2009. Dock8 mutations cripple B cell immunological synapses, germinal centers and long-lived antibody production. Nat Rev Immunol. 10:1283–1291. doi:doi:10.1038/ni.1820.

Raney, B.J., T.R. Dreszer, G.P. Barber, H. Clawson, P.A. Fujita, T. Wang, N. Nguyen, B. Paten, A.S. Zweig, D. Karolchik, and W.J. Kent. 2014. Track data hubs enable visualization of user-defined genome-wide annotations on the UCSC Genome Browser. Bioinformatics. 30:1003–1005. doi:10.1093/bioinformatics/btt637.

Rathmell, J.C., E.A. Farkash, W. Gao, and C.B. Thompson. 2001. IL-7 enhances the survival and maintains the size of naive T cells. J. Immunol. 167:6869–6876.

Rongvaux, A., R. Jackson, C.C.D. Harman, T. Li, A.P. West, M.R. de Zoete, Y. Wu, B. Yordy, S.A. Lakhani, C.-Y. Kuan, T. Taniguchi, G.S. Shadel, Z.J. Chen, A. Iwasaki, and R.A. Flavell. 2014. Apoptotic Caspases Prevent the Induction of Type I Interferons by Mitochondrial DNA. Cell. 159:1563–1577. doi:10.1016/j.cell.2014.11.037.

Rosenbloom, K.R., C.A. Sloan, V.S. Malladi, T.R. Dreszer, K. Learned, V.M. Kirkup, M.C. Wong, M. Maddren, R. Fang, S.G. Heitner, B.T. Lee, G.P. Barber, R.A. Harte, M. Diekhans, J.C. Long, S.P. Wilder, A.S. Zweig, D. Karolchik, R.M. Kuhn, D. Haussler, and W.J. Kent. 2012. ENCODE Data in the UCSC Genome Browser: year 5 update. Nucleic Acids Research. 41:D56–D63. doi:10.1093/nar/gks1172.

Sato, S., H. Sanjo, K. Takeda, J. Ninomiya-Tsuji, M. Yamamoto, T. Kawai, K. Matsumoto, O. Takeuchi, and S. Akira. 2005. Essential function for the kinase TAK1 in innate and adaptive immune responses. Nat Immunol. 6:1087–1095. doi:10.1038/ni1255.

Schmidt, T.H., O. Bannard, E.E. Gray, and J.G. Cyster. 2013. CXCR4 promotes B cell egress from Peyer's patches. Journal of Experimental Medicine. 210:1099–1107. doi:10.1084/jem.20122574.

Schmidt-Supprian, M., G. Courtois, J. Tian, A.J. Coyle, A. Israël, K. Rajewsky, and M. Pasparakis. 2003. Mature T Cells Depend on Signaling through the IKK Complex. Immunity. 19:377–389. doi:10.1016/S1074-7613(03)00237-1.

! 106! Schnell, F.J., A.L. Zoller, S.R. Patel, I.R. Williams, and G.J. Kersh. 2006. Early growth response gene 1 provides negative feedback to inhibit entry of progenitor cells into the thymus. J. Immunol. 176:4740–4747.

Schnell, F.J., and G.J. Kersh. 2005. Control of recent thymic emigrant survival by positive selection signals and early growth response gene 1. J. Immunol. 175:2270–2277.

Scholzen, T., and J. Gerdes. 2000. The Ki-67 protein: from the known and the unknown. J. Cell. Physiol. 182:311–322. doi:10.1002/(SICI)1097-4652(200003)182:3<311::AID- JCP1>3.0.CO;2-9.

Scholzen, T., E. Endl, C. Wohlenberg, S. van der Sar, I.G. Cowell, J. Gerdes, and P.B. Singh. 2002. The Ki-67 protein interacts with members of the heterochromatin protein 1 (HP1) family: a potential role in the regulation of higher-order chromatin structure. J. Pathol. 196:135–144. doi:10.1002/path.1016.

Sentman, C.L., J.R. Shutter, D. Hockenbery, O. Kanagawa, and S.J. Korsmeyer. 1991. bcl-2 inhibits multiple forms of apoptosis but not negative selection in thymocytes. Cell. 67:879– 888. doi:10.1016/0092-8674(91)90361-2.

Silva, A., G. Cornish, S.C. Ley, and B. Seddon. 2014. NF-kB signaling mediates homeostatic maturation of new T cells. Proc Natl Acad Sci U S A. doi:10.1073/pnas.1319397111.

Sinclair, C., M. Saini, I. Schim van der Loeff, S. Sakaguchi, and B. Seddon. 2011. The Long- Term Survival Potential of Mature T Lymphocytes Is Programmed During Development in the Thymus. Sci Signal. 4:ra77–ra77. doi:10.1126/scisignal.2002246.

Smale, S.T. 2014. Transcriptional regulation in the immune system: a status report. Trends in Immunology. doi:10.1016/j.it.2014.03.003.

Smyth, G.K. Linear Models and Empirical Bayes Methods for Assessing Differential Expression in Microarray Experiments. Statistical Applications in Genetics and Molecular Biology. 3. doi:10.2202/1544-6115.1027.

Sprent, J., and C.D. Surh. 2011. Normal T cell homeostasis: the conversion of naive cells into memory-phenotype cells. Nat Immunol. 131:478–484. doi:10.1038/ni.2018.

Sprent, J., and H. Kishimoto. 2002. The thymus and negative selection. Immunol. Rev. 185:126– 135. doi:10.1034/j.1600-065X.2002.18512.x.

Starr, T.K., S.C. Jameson, and K.A. Hogquist. 2003. Positive and Negative Selection of T cells. Annu. Rev. Immunol. 21:139–176. doi:10.1146/annurev.immunol.21.120601.141107.

Stritesky, G.L., S.C. Jameson, and K.A. Hogquist. 2012. Selection of self-reactive T cells in the thymus. Annu. Rev. Immunol. 30:95–114. doi:10.1146/annurev-immunol-020711-075035.

Subramanian, A., P. Tamayo, V.K. Mootha, S. Mukherjee, B.L. Ebert, M.A. Gillette, A. Paulovich, S.L. Pomeroy, T.R. Golub, E.S. Lander, and J.P. Mesirov. 2005. Gene set

! 107! enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A. 102:15545–15550. doi:10.1073/pnas.0506580102.

Surh, C.D., and J. Sprent. 1994. T-cell apoptosis detected in situ during positive and negative selection in the thymus. Nature. 372:100–103. doi:10.1038/372100a0.

Surh, C.D., and J. Sprent. 2008. Homeostasis of Naive and Memory T Cells. Immunity. 29:848– 862. doi:10.1016/j.immuni.2008.11.002.

Swann, J.B., Y. Hayakawa, N. Zerafa, K.C.F. Sheehan, B. Scott, R.D. Schreiber, P. Hertzog, and M.J. Smyth. 2007. Type I IFN Contributes to NK Cell Homeostasis, Activation, and Antitumor Function. The Journal of Immunology. 178:7540–7549. doi:10.4049/jimmunol.178.12.7540.

Takada, K., and S.C. Jameson. 2009. Naive T cell homeostasis: from awareness of space to a sense of place. Nat Rev Immunol. 9:823–832. doi:10.1038/nri2657.

Teng, F., Y. Zhou, R. Jin, Y. Chen, X. Pei, Y. Liu, J. Dong, W. Wang, X. Pang, X. Qian, W.-F. Chen, Y. Zhang, and Q. Ge. 2011. The molecular signature underlying the thymic migration and maturation of TCRαβ+ CD4+ CD8 thymocytes. PLoS ONE. 6:e25567. doi:10.1371/journal.pone.0025567.

Thomas, J.H., and S. Schneider. 2011. Coevolution of retroelements and tandem zinc finger genes. Genome Research. 21:1800–1812. doi:10.1101/gr.121749.111.

Vanneste, D., M. Takagi, N. Imamoto, and I. Vernos. 2009. The role of Hklp2 in the stabilization and maintenance of spindle bipolarity. Curr. Biol. 19:1712–1717. doi:10.1016/j.cub.2009.09.019.

Verheijen, R., H.J. Kuijpers, R.O. Schlingemann, A.L. Boehmer, R. van Driel, G.J. Brakenhoff, and F.C. Ramaekers. 1989. Ki-67 detects a nuclear matrix-associated proliferation-related antigen. I. Intracellular localization during interphase. J. Cell. Sci. 92 ( Pt 1):123–130.

Wan, Y.Y., H. Chi, M. Xie, M.D. Schneider, and R.A. Flavell. 2006. The kinase TAK1 integrates antigen and cytokine receptor signaling for T cell development, survival and function. Nat Immunol. 7:851–858. doi:10.1038/ni1355.

Weinreich, M.A., and K.A. Hogquist. 2008. Thymic Emigration: When and How T Cells Leave Home. The Journal of Immunology. 181:2265–2270. doi:10.4049/jimmunol.181.4.2265.

West, A.P., W. Khoury-Hanold, M. Staron, M.C. Tal, C.M. Pineda, S.M. Lang, M. Bestwick, B.A. Duguay, N. Raimundo, D.A. MacDuff, S.M. Kaech, J.R. Smiley, R.E. Means, A. Iwasaki, and G.S. Shadel. 2000. Mitochondrial DNA stress primes the antiviral innate immune response. Nature. –. doi:doi:10.1038/nature14156.

Wolfe, S.A., L. Nekludova, and C.O. Pabo. 2000. DNA recognition by Cys2His2 zinc finger proteins. Annu Rev Biophys Biomol Struct. 29:183–212. doi:10.1146/annurev.biophys.29.1.183.

! 108! Wu, C., C. Orozco, J. Boyer, M. Leglise, J. Goodale, S. Batalov, C.L. Hodge, J. Haase, J. Janes, J.W. Huss, and A.I. Su. 2009. BioGPS: an extensible and customizable portal for querying and organizing gene annotation resources. Genome Biol. 10:R130. doi:10.1186/gb-2009-10- 11-r130.

Yang, Y.J., A.E. Baltus, R.S. Mathew, E.A. Murphy, G.D. Evrony, D.M. Gonzalez, E.P. Wang, C.A. Marshall-Walker, B.J. Barry, J. Murn, A. Tatarakis, M.A. Mahajan, H.H. Samuels, Y. Shi, J.A. Golden, M. Mahajnah, R. Shenhav, and C.A. Walsh. 2012. Microcephaly gene links trithorax and REST/NRSF to control neural stem cell proliferation and differentiation. Cell. 151:1097–1112. doi:10.1016/j.cell.2012.10.043.

Zachariah, M.A., and J.G. Cyster. 2010. Neural Crest-Derived Pericytes Promote Egress of Mature Thymocytes at the Corticomedullary Junction. Science. 328:1129–1135. doi:10.1126/science.1188222.

Zhang, N., and Y.-W. He. 2005. An essential role for c-FLIP in the efficient development of mature T lymphocytes. J. Exp. Med. 202:395–404. doi:10.1084/jem.20050117.

Zhang, Y., T. Liu, C.A. Meyer, J. Eeckhoute, D.S. Johnson, B.E. Bernstein, C. Nussbaum, R.M. Myers, M. Brown, W. Li, and X.S. Liu. 2008. Model-based Analysis of ChIP-Seq (MACS). Genome Biol. 9:R137. doi:10.1186/gb-2008-9-9-r137.

! 109!