ROLES OF THE RHO KINASES IN B CELL DIFFERENTIATION AND LYMPHOMAGENESIS

A Dissertation Presented to the Faculty of Weill Cornell Graduate School of Medical Sciences in Partial Fulfillment of the Requirements of the Degree of Doctor of Philosophy

by Edd C. Ricker May 2018

© 2018 Edd C. Ricker

ROLES OF THE RHO KINASES IN B CELL DIFFERENTIATION AND LYMPHOMAGENESIS Edd C. Ricker, Ph.D. Cornell University 2018

ABC-DLBCLs are aggressive B cell malignancies characterized by deregulations in the molecular networks controlling plasma cell (PC) differentiation. The survival of ABC-DLBCLs is known to require the transcription factor IRF4. The mechanisms controlling IRF4 activity in ABC- DLBCLs are not fully understood. ROCK1 and ROCK2 are two serine-threonine kinases that serve as major effector proteins for RhoA, which was recently found to be mutated in several lymphomas. Here we show that IRF4 is constitutively phosphorylated in ABC-DLBCLs. IRF4 phosphorylation is mediated by ROCK2, which is constitutively activated in ABC-DLBCLs, but not in other DLBCLs. ROCK2-mediated IRF4 phosphorylation can be induced by signals that promote PC differentiation and modulates the ability of IRF4 to regulate the expression of a subset of PC genes. Inhibition of ROCK2 in ABC-DLBCLs alters their transcriptional profile not only by controlling IRF4 activity but also by regulating c-MYC protein levels. In addition to ROCK2, ROCK1 also regulates key survival pathways in ABC-DLBCLs and pan-ROCK inhibition decreases the survival of ABC-DLBCLs, but not that of GCB-DLBCLs. We also identified critical roles for ROCK2 in physiological B cell differentiation and showed that lack of ROCK2 in B cells leads to impaired germinal center formation and humoral responses. Together, these findings reveal an important role for ROCK2 in modulating physiological B cell responses upon antigen challenge, delineate the pathophysiological implications of ROCK activation in ABC-DLBCL and other B cell malignancies, and propose that ROCK inhibition could represent a novel therapeutic approach for the treatment of diseases characterized by dysfunctional B cell responses.

BIOGRAPHICAL SKETCH

Edd C. Ricker is from Pepperell, Massachusetts, where he graduated as the Class of 2008 Salutatorian from North Middlesex Regional High School. Edd attended the University of Massachusetts in Amherst, Massachusetts, where he earned a Bachelor of Science in Biochemistry and Molecular Biology with a Minor in Microbiology in 2012, graduating summa cum laude and with Honors with Greatest Distinction from the Commonwealth Honors College. As an undergraduate, he participated in an HHMI Summer Research Internship and the Junior Fellowship in the Life Sciences program, conducting research in the laboratory of Dr. Patricia Wadsworth. In 2012, Edd matriculated into the Immunology and Microbial Pathogenesis PhD program at Weill Cornell Medical College, and joined the laboratory of Alessandra Pernis, MD, in 2013. During his time in the laboratory, Edd was the recipient of an NIH F31 Ruth L. Kirschstein National Service Award, attended and presented his work at international research conferences, served as a teaching assistant for the Immunology and Microbial Pathogenesis graduate program, and remained active in several community organizations at Weill Cornell Medicine, the Hospital for Special Surgery, and in New York City.

iii

This dissertation is dedicated to my mother, Margaret Ricker, father, Edd Ricker, brother, Dylan Ricker, and wife, Jocelyn Ricker.

iv ACKNOWLEDGEMENTS

There are many people that contributed to my success as a graduate student and to the completion of this dissertation project. I would like to thank the Immunology and Microbial Pathogenesis Program at Weill Cornell Medicine and the Hospital for Special Surgery Research Institute for administrative and scientific support, as well as for providing an enriching and vigorous scientific and academic community.

Notably, I would like to thank Leandro Cerchietti, Jayanta Chaudhuri, and Selina Chen-Kiang for their invaluable insights, support, and mentorship as active members of my thesis committee. Specifically, for the projects outlined in this dissertation, I would like to thank Sanjay Gupta, Michela Manni, Rossella Marullo, Akanksha Verma, and Chao Ye for assistance with experiments; Leandro Cerchietti, Olivier Elemento, Tania Pannellini, and Wayne Tam for assistance with interpreting experiments; Franck Barrat, Carl Blobel, Lionel Ivashkiv, Theresa Lu, Inez Rogatsky, and Jane Salmon for providing reagents and critical intellectual input. I would also like to thank the laboratories of Franck Barrat, Leandro Cerchietti, Lionel Ivashkiv, and Theresa Lu for helpful discussions and reagents. This work was supported by an NIH F31 Ruth L. Kirschstein National Service Award.

I would like to thank Dr. Alessandra Pernis, my thesis advisor, for her guidance and mentorship throughout my time as a graduate student. Alessandra worked closely with me during my graduate training to ensure my success and growth as an independent scientist. She continually challenged me to think more

v critically and rigorously, while providing a wealth of professional and personal support. Alessandra also enriched my graduate training by promoting a diverse laboratory environment where I was able to learn new technical and scientific skills from experienced postdoctoral fellows and research scientists. In addition to Alessandra, I would also like to thank all members of the Pernis lab, past and present, for creating an enjoyable and stimulating work environment.

Finally, I would like to thank all my family and friends for their continued support and encouragement over the past six years. Specifically, I would like to thank my mother, Margaret Ricker, and father, Edd Ricker, for demonstrating the qualities of persistence, integrity, and perspective, characteristics that I have attempted to preserve throughout my graduate studies. I would also like to thank my wife, Jocelyn Ricker, for her unwavering support, despite late nights and weekends in the lab. Her continued encouragement and humor has kept me balanced and inspired me to persist through the inevitable trials and tribulations of a PhD program.

vi TABLE OF CONTENTS

Biographical Sketch 0iii Dedication 0iv Acknowledgements 00v Table of Contents 0vii List of Figures 0viii Chapter I: General Introduction 001 Chapter II: Roles of the Rho Kinases in ABC-DLBCLs 026 Results 030 Discussion 061 Methods 066 Chapter III: Roles of ROCK2 in Primary B Cell Responses 077 Results 078 Discussion 95 Methods 99 Chapter IV: Conclusions and Future Perspectives 104 References 107

vii LIST OF FIGURES

Figure 1. Germinal Center B Cell Responses Are Regulated by an 4 Antagonistic Network of Transcription Factors Figure 2. IRF4 is a Multifaceted Regulator of Lymphocyte Biology 8 Figure 3. RhoA-GTPases are Molecular Switches that Activate the 12 Rho Kinases Figure 4. B-NHLs are Classified Based on Genetic Correspondence 21 with Physiological Stages in B Cell Differentiation Figure 5. IRF4 is Constitutively Phosphorylated in ABC-DLBCLs 31 Figure 6. STAT3 Phosphorylation is Not Decreased by Y-27632 in 33 B-NHLs

Figure 7. IRF4 is Phosphorylated Upon Stimulation with aCD40 34 and IL-21 Figure 8. ROCK2 Activity is Constitutive in ABC-DLBCLs and 37

Induced Upon Stimulation with aCD40 and IL-21 Figure 9. ROCK2 Phosphorylates IRF4 Upon Stimulation with 40

aCD40 and IL-21 Figure 10. ROCK2 Regulates the Expression of a Subset of IRF4 41

Target Genes Figure 11. ROCK2 Controls the Binding of IRF4 to Several Regulatory 43 Regions Figure 12. ROCK2 Phosphorylates IRF4 in ABC-DLBCLs 45 Figure 13. ROCK2 Regulates a Transcriptional Program in ABC-DLBCLs 48 Figure 14. ROCK2 Promotes an IRF4-Regulated Transcriptional 49 Program in ABC-DLBCLs

viii Figure 15. ROCK2 Promotes MYC Protein Levels in ABC-DLBCLs 51 Figure 16. ROCK1 and ROCK2 Regulate Specific Transcriptional 55 Programs in ABC-DLBCLs Figure 17. Pan-ROCK Inhibition Induces Lethality in ABC-DLBCLs 57 Figure 18. ROCK is Activated in Primary DLBCLs 60 Figure 19. ROCK2 Activity in Increased Upon B Cell Differentiation 79

Figure 20. B Cell Development is Normal in Cg1-Cre.ROCK2 Mice 81 Figure 21. Cg1-Cre.ROCK2 Mice Have Decreased GC Responses 84 following TD Immunization

Figure 22. Cg1-Cre.ROCK2 Mice Have Impaired GC Responses 87 following TD Immunization

Figure 23. Cg1-Cre.ROCK2 Mice Exhibit Defective Humoral 90 Responses following TD Immunization

Figure 24. TFH Differentiation is Normal in Cg1-Cre.ROCK2 Mice 92 following TD Immunization Figure 25. CD4-Cre.ROCK2 Mice Exhibit Normal Responses to TD 93 Immunization Figure 26. Model Showing the Roles of ROCK2 in Physiological and 106 Pathophysiological B Cell Differentiation

ix CHAPTER I GENERAL INTRODUCTION

Effective humoral responses depend on the ability of lymphocytes to respond to rapid and dynamic cues. Indeed, B cell fate decisions can be controlled by several signals, in particular those mediated by the B cell receptor (BCR), Toll- like receptors (TLRs), and T cell-dependent help. Proper B cell responses require the integration of these signals to alter transient processes, such as cell migration and proliferation, and to induce longer-lasting changes, including chromatin remodeling and alterations in the gene expression program. Precise regulation of these processes is not only important for efficient pathogen clearance, but is also essential to prevent the emergence of several diseases, including autoimmunity and B cell malignancies. Delineating the regulatory mechanisms that control B cell fate decisions in response to dynamic and rapid signals will, thus, be crucial to expand our understanding of immune regulation and more effectively target disorders marked by immune dysregulation.

Physiological B Cell Responses

Although some humoral responses can be initiated in the absence of T cell help, highly efficient immune responses necessary for the elimination of most pathogens rely on T-B interactions that culminate in the production of high- affinity antibody secreting plasma cells (PCs)1, 2. These interactions promote B cell activation through the engagement of CD40 by CD40L and through the presence of other T cell-derived signals, such as the production of cytokines like IL-213, 4, 5. These signals play critical roles in supporting B cell activation by

1 driving the expression of transcriptional programs that serve as central regulators of divergent B cell differentiation pathways, including the formation of germinal center B cells, plasma cells, and memory B cells1, 2, 6.

Germinal Center Responses Germinal centers (GCs) are transient anatomical structures that function as the major sites of clonal expansion and affinity maturation for antigen-specific B cells in response to T cell-dependent activation1, 7. The development of a functional GC after antigen exposure can be divided into three distinct stages: early initiation (days 0-2), late initiation (days 3-4), and establishment (days 5- 6)1. The early initiation phase of the GC response begins immediately following antigen encounter, whereby antigen-activated B and T cells migrate to the interfollicular border and engage in sustained interactions, resulting in further B

8, 9, 10, 11 cell activation and differentiation of follicular T helper (TFH) cells . During the late initiation phase, TFH cells migrate into the primary follicle and activated B cells face an important cell fate decision; some activated B cells terminally differentiate into short-lived extrafollicular plasmablasts/plasma cells (PB/PCs), while others migrate back into the follicle and begin to proliferate1, 10, 12, 13. During this proliferative stage, activated B cells displace resident follicular B cells and form the early GC structure. The rapid proliferation of activated B cells results in the continued expansion of the early GC, until the mature GC is established.

The mature GC is histologically characterized by the appearance of two polarized microenvironments: the GC dark zone (DZ) and GC light zone (LZ)7, 14, 15. This polarity carries functional relevance, as it allows for the compartmentalization of specific GC processes. The GC DZ is primarily

2 composed of rapidly dividing GC B cells known as centroblasts and is believed to be the site of BCR repertoire diversification due to increased somatic hypermutation of the immunoglobulin variable region (IgV)7, 16, 17. The GC LZ is composed of a sparser population of GC B cells, known as centrocytes, and is

1, 7, 15 also rich in TFH cell and follicular dendritic cell (FDC) populations . T-B interactions in the LZ between TFH cells and centrocytes are critical for affinity maturation to occur15. Although not completely understood, current models propose that B cell clones with high-affinity BCRs are able to outcompete lower- affinity clones for T cell help and receive sufficient signals to exit the GC and terminally differentiation into PB/PCs7, 15, 18, 19, 20. B cell clones with low-affinity BCRs, however, can be stimulated to either reenter the DZ for further rounds of hypermutation in a process known as cyclic reentry, or can be induced to undergo apoptosis1, 15. Although a complete understanding of the signals and dynamics driving these divergent fates remains unknown, recent studies have suggested a level of temporal control to cyclic reentry and affinity maturation15, 19.

Given the complex and dynamic nature of GC responses, regulation of the GC B cell program is coordinated by an antagonistic network of transcription factors1

(Figure 1); the intricacies of which remain incompletely understood. One of the major hubs in this network is BCL6, a transcriptional repressor that is essential for both the formation and maintenance of the GC11, 21. During initiation of the GC, BCL6 is required for the repositioning of activated B cells towards the center of the follicles and may fine-tune T-B interactions by enhancing the integrin- mediated conjugation of T and B cells11, 21. In the established GC, BCL6 acts to enforce a transcriptional network that promotes affinity maturation by increasing

3

Figure 1. Germinal Center B Cell Responses Are Regulated by an Antagonistic Network of Transcription Factors. Diagram outlining key regulators of the GC B cell response. During GC initiation, IRF4 is required for the induction of BCL6, a transcriptional repressor that acts as the central regulator of the GC B cell program. MEF2C, MYC, and NF-kB are also expressed during the early stages of GC initiation. In the established GC, GC B cells express multiple transcription factors that maintain BCL6 expression, including IRF8 and MEF2B. GC functions are also maintained through the expression of AICDA, which promotes CSR and SHM, and FoxO1, which contributes to the centroblast phenotype of dark zone GC B cells. In addition to promoting the GC B cell phenotype, BCL6 also acts to repress genes involved in PB/PC differentiation, including PRDM1, the gene encoding BLIMP1. BLIMP1 is the master transcription factor of the PB/PC phenotype and acts to both promote the expression of PC-related genes and repress GC genes, including BCL6. After IRF4 silencing following GC initiation, IRF4 is re-expressed and activated in exiting GC B cells and acts as a molecular switch to silence BCL6 expression and induce BLIMP1 expression, thus terminating the GC program and inducing PB differentiation.

4 the cellular threshold for the DNA damage response and by repressing the expression of microRNAs that serve as negative regulators of AID expression22, 23, 24. BCL6 also maintains the GC B cell phenotype by directly repressing genes that promote terminal differentiation, such as Prdm125, 26. Importantly, BCL6 is

27, 28, 29, 30 also required for the TFH cell phenotype , highlighting the ability of BCL6 to regulate GC dynamics through complementary effects on multiple cell types.

In addition to BCL6, other transcription factors also play important roles in promoting the GC B cell program1. Many of these transcription factors are temporally regulated during the GC response and only appear at distinct stages of activation. For instance, MYC is expressed at the onset of the GC response in activated B cells and is believed to provide an early proliferative burst to help establish the GC B cell program before its expression is silenced31, 32, 33. Interestingly, MYC is re-expressed in a subset of centrocytes at later stages in the established GC, where it has been suggested to promote cyclic reentry and to maintain the GC response31, 32. Other transcription factors known to play important roles in the GC program include regulators of BCL6 expression, such as MEF2B34, and, as discussed in greater detail below, members of the Interferon Regulatory Factor (IRF) family of transcription factors1, 35.

Upon terminal differentiation to PB/PCs, GC B cells undergo a transcriptional remodeling marked by the repression of Bcl6 and the induction of Prdm11, 2. Prdm1 is the gene encoding BLIMP1 and serves as the master regulator of the PB/PC phenotype2. BLIMP1 acts on a diverse range of transcriptional targets to promote metabolic and organelle reprogramming in order to facilitate differentiation into a high antibody-secreting state and directly represses the GC

5 program through the inhibition of Bcl636, 37, 38. Regulation of this transition, from a GC B cell to a PB/PC, is achieved by the activation of a distinct member of the IRF family, IRF42, 39.

Role of IRF4 in B Cell Activation and Differentiation. The Interferon Regulatory Factor (IRF) family of transcription factors is composed of nine members, IRF1-9, which have been well studied for their roles in promoting both innate and adaptive immune responses39, 40. Proteins of the IRF family share structural similarities, whereby they contain an N-terminal DNA-binding domain (DBD) with five highly conserved tryptophan-rich repeats (W)40. The DBD recognizes motifs containing 5’-AAnnGAAA-3’39, 40. The IRFs also contain an IRF-associated domain (IAD) of either Type I or Type II, which mediates interactions between the IRFs and other binding partners as well as a C-terminal region, which can serve an autoinhibitory function39, 40.

Among the IRF family, IRF4 has emerged as an essential regulator of T-B collaborations during immune responses39, 40. Indeed, IRF4 expression is induced downstream of signals that activate T and B cells, such as antigen receptor stimulation, CD40 ligation, and cytokines, including IL-4 and IL-2139, 41.

In T cells, IRF4 plays major roles in the differentiation and effector functions of

42, 43, 44, 45 several TH subsets, including TH2, TH9, TFH, and TH17 cells . Importantly, IRF4 is absolutely required for the production of IL-21, a major signal that drives B cell activation45, 46, 47. IRF4 has also been shown to be critical

48 for the effector function of regulatory T (Treg) cells , providing additional support for IRF4 as a critical modulator of adaptive immune responses (Figure 2).

6 In addition to driving the production of B cell activating factors by T cells, fundamental cell-intrinsic roles for IRF4 have also been described in the development, differentiation, and activation of B lymphocytes39. During development, IRF4 plays important, yet redundant roles in the pro-B to pre-B cell transition by promoting rearrangement of the Ig light chain locus49, 50. In mature B cells, IRF4 regulates class switch recombination (CSR) through its ability to control the expression of activation-induced cytosine deaminase (AID) and has been shown to be absolutely required for the exit of B cells from the GC and for the formation of PB/PCs51, 52. IRF4 plays dual roles in promoting this transition by directly suppressing BCL6 expression and inducing BLIMP1 expression by targeting several regulatory regions in the BCL6 and PRDM1 loci, respectively39, 53, 54. Interestingly, a role for IRF4 in the initiation of the GC program has also been described, whereby IRF4 induces BCL6 expression by directly binding to a site 25kB upstream from the BCL6 transcription start site 55, 56. This site differs from the site targeted to mediate BCL6 repression during GC exit 54, suggesting multifaceted roles for IRF4 during B cell activation.

The ability of IRF4 to execute distinct transcriptional programs in a context- dependent manner has been partially explained by a dose-dependent model of

IRF4 regulation52, 55, 57. In this model, genomic targeting of IRF4 is believed to be greatly dependent on IRF4 expression levels. This model was first proposed based on observations in activated B cells whereby low levels of IRF4 expression was shown to promote class switch recombination (CSR) by inducing AICDA expression, while high levels of IRF4 induced PB/PC differentiation through the direct targeting of the PRDM1 locus52, 55, 57. Since this initial description, dose-dependent effects of IRF4 in T cell activation have also

7

Figure 2. IRF4 is a Multifaceted Regulator of Lymphocyte Biology. (A) Schematic diagram of the domain structure of human Interferon Regulatory Factor 4 (IRF4). IRF4 has an N-terminal DNA binding domain (DBD) with five tryptophan (W)-rich repeats. The C-terminal domain of IRF4 contains an IRF- association domain (IAD), which mediates interactions between IRF4 and various binding partners. The C-terminal region also contains auto-inhibitory functions. (B) Schematic diagram showing representative roles for IRF4 in various stages of lymphocyte development and differentiation. In T cells, IRF4 promotes the differentiation of TH2, TH9, TFH, and TH17 cells. IRF4 also + promotes the effector functions of Treg cells and cytotoxic CD8 lymphocytes. In B cells, IRF4 promotes the rearrangement of the Ig light chain locus during VD(J) recombination and functions in mature B cells to promote processes such as CSR, GC initiation, and PB/PC formation through the dynamic regulations of AICDA, BCL6, and PRDM1.

8 been described58. Mechanistically, the dose-dependent activity of IRF4 may be caused by differential complex formation between IRF4 and its interacting partners. ChIP-seq studies have demonstrated that at high expression levels, IRF4 predominantly targets IFN-stimulated response elements (ISREs) in the genome due to the enhanced formation of IRF4 homodimers55. ISREs can presumably also be targeted by IRF4 heterodimers, whereby IRF4 interacts with other IRF family members, including IRF8. Contrasting the targeting of ISREs at high expression levels, at lower levels of expression, IRF4 is likely to complex with ETS-family members, such as SpiB and PU.1, and AP-1 family members, such as BATF, to target Ets-IRF composite elements (EICEs) and AP1-IRF composite elements (AICEs), respectively59, 60.

In addition to the concentration-dependent control of IRF4 activity, other modes of IRF4 regulation have also been described. One such example includes negative regulation of IRF4 activity by the SWEF family of proteins61. The SWEF proteins, SWAP70 and DEF6, are a pair of unique Rac-GEFs, which, among other functions, can bind and sequester IRF4 in the nucleus of lymphocytes61. In activated B cells, sequestration of IRF4 by the SWEF proteins suppresses the ability of IRF4 to promote AICDA and PRDM1 expression, restricting CSR and PB/PC differentiation, respectively61. Additionally, ROCK2, a serine- threonine kinase, is able to directly regulate IRF4 activity in T lymphocytes.

ROCK2 phosphorylates IRF4 on two serine residues (S446, S447) within the a- helical region of the protein located C-terminal to the IAD62. Structural studies indicate that phosphorylation of these residues promotes conformational rearrangements that modulate the ability of the IAD to mediate interactions with cofactors63. In T cells, this phosphorylation has functional consequences, as

9 blocking ROCK activity attenuates the binding of IRF4 to the IL-21 and IL-17 promoters and prevents the optimal production of these effector cytokines62. A major aim of this dissertation is to determine whether ROCK2 also regulates IRF4 function in B cells.

The RhoA-ROCK pathway Rho-GTPases, such as RhoA, are emerging as important regulators of lymphocyte biology owing to their ability to be rapidly activated downstream of a broad range of biochemical and biomechanical signals64, 65, 66. As a GTPase, RhoA acts as a molecular switch that cycles between active (GTP-bound) and inactive (GDP-bound) states through the interaction with guanine nucleotide exchange factors (GEFs, which promote the exchange of GDP for GTP), GTPase activating proteins (GAPs, which enhance the intrinsic GTPase activity of RhoA), and guanosine nucleotide dissociation inhibitors (GDIs, which bind and sequester inactivated RhoA in the cytoplasm)66 (Figure 3A). GEF activation, and subsequently that of RhoA, occurs in response to dynamic stimuli, including chemokines, growth factors, cell-matrix interactions, and mechanical signals65. Interestingly, at least 24 GEFs have been reported to activate RhoA67. This redundancy in RhoA activation allows for multiple upstream regulators to converge onto RhoA and likely compartmentalizes GEFs with specific RhoA substrates, thus facilitating the transmission of distinct downstream effector functions68.

Following activation, RhoA interacts with several effector molecules, including the Rho Kinases (ROCKs), two highly homologous serine-threonine kinases (ROCK1 and ROCK2) that regulate a variety of biological processes that allows

10 for the coordination of a tissue response to stress and injury65, 69, 70, 71. The RhoA-ROCK pathway has been implicated in the control of cytoskeletal reorganization and migration, proliferation, survival, and gene expression65, 69, 70, 71. Despite the fundamental reliance of T and B cells on these processes, the precise involvement of the ROCKs in lymphocyte biology is yet to be elucidated64. This dissertation examines how ROCK1 and ROCK2, both collectively and independently, regulate B cell responses under physiological and pathophysiological settings.

Rho Kinase Structure and Regulation ROCK1 and ROCK2 are encoded by two separate genes, yet share ~65% identity in their protein structure. The ROCKs contain an N-terminal catalytic kinase domain, which is followed by a coiled-coil region containing the Rho- binding domain (RBD), and a C-terminal autoinhibitory domain (Figure 3B). The catalytic kinase domains of ROCK1 and ROCK2 are highly homologous, exhibiting ~92% identity65, 69, 70, 71. This high degree of identity is reflected in the ability of ROCK1 and ROCK2 to target similar substrates in vitro and has led to the prevalent view that the ROCKs exclusively play compensatory and redundant roles in vivo69, 70, 71. However, recent reports utilizing newly developed isoform-specific inhibitors and genetic models62, 64, 72, as well as data presented in this dissertation, have aimed to challenge this conclusion and to delineate isoform specific roles of ROCK1 and ROCK2 in distinct biological contexts.

Due to the ubiquitous expression pattern of the ROCKs, regulation of these proteins mainly occurs at the level of kinase activation. At basal state, the

11

Figure 3. RhoA-GTPases are Molecular Switches that Activate the Rho Kinases. (A) Schematic diagram showing the RhoA activation cycle. Inactive GDP-bound RhoA is sequestered in the cytoplasm through interactions with GDP-dissociation inhibitors (GDIs). Activation of RhoA is facilitated by guanine- nucleotide exchange factors (GEFs) which promote the exchange of GDP for GTP. Activated RhoA-GTP binds with and activates several effector molecules, including the Rho Kinases (ROCKs). RhoA-GTP becomes inactivated through its intrinsic ability to hydrolyze GTP, a process promoted by GTPase-activating proteins (GAPs). (B) Schematic diagram of the domain structure of the two human ROCK-family members: ROCK1 and ROCK2. ROCK1 and ROCK2 contain N-terminal kinase domains (KD), which shares 92% identity. The Rho- binding domains (RBD) are located in the C-terminal end of the coiled-coil regions (CC). The C-termini contain pleckstrin homology domains (PH).

12 ROCKs exist in an inactive “closed” confirmation whereby the C-terminal autoinhibitory domain binds with and inhibits the activity of the N-terminal kinase domain. Binding of activated RhoA to the RBD disrupts this association, however, and leads to kinase activation. Although this is the classical method of ROCK activation, Rho-independent mechanisms of activation involving cleavage of the C-terminal autoinhibitory domain by Caspase-3 and Granzyme B have also been described69, 70, 71.

Given the high degree of identity between the kinase domains of ROCK1 and ROCK2, studies examining ROCK biology have historically been dependent on pan-kinase inhibitors, such as Y-27632. Y-27632 is an ATP-competitive ROCK inhibitor belonging to the 4-aminopyridine series73, 74, 75. Although able to inhibit both ROCK1 and ROCK2 in preclinical models, clinical development of this drug has been limited due to concerns of potential off-target effects and limited potency (Ki ~140nM)76. The limited potency of Y-27632 can especially be observed in human studies, whereby higher doses of Y-27632 are required to see functional effects compared to murine cells77. Despite these concerns for the use of Y-27632, other pan-ROCK inhibitors, including Fasudil, have proven to be well tolerated in healthy individuals and exhibit an excellent safety profile75,

78, 79, suggesting that the ROCK pathway could be a promising therapeutic target for diseases driven by hyperactive ROCK signaling. Recently, SurfaceLogix developed a new inhibitor with 100-fold selectivity for ROCK2 compared to ROCK174, 75, 76. This compound, recently renamed KD025 (Ki ~41nM), has been shown to effectively inhibit ROCK2 in both human and murine systems74, 75, 76.

Functions of Rho Kinases in Non-Hematopoietic Cells

13 Upon activation, the ROCKs target a wide range of substrates that allow them to regulate diverse biological processes and serve as coordinators of a tissue response to stress and injury69, 70, 71. Perhaps the best-studied roles of the ROCKs are in the regulation of cytoskeletal dynamics. The ROCKs play a central role in promoting actin-myosin contractility, intermediate filament assembly, and microtubule dynamics69, 70, 71. The ROCKs also contribute to the activation of the ERM (ezrin, radizin, moesin) complex of proteins, leading to the cross-linking of integral membrane proteins to the actin cytoskeleton69, 70. The RhoA-ROCK pathway has also been shown to be involved in organelle positioning and intracellular vesicular transport69, 70, 71. Cumulatively, through the regulation of these cytoskeletal dynamics, the ROCKs are able to control key cellular processes such as cell migration, adhesion, membrane dynamics, and the establishment of cellular polarity68, 69, 70, 71.

As a major effector of the cellular stress response, it is not surprising that the ROCKs have also been shown to play important roles in cell proliferation and survival. Indeed, activated ROCK accumulates at the cleavage furrow during cytokinesis to promote daughter cell formation, and can regulate early stages of cell cycle progression by directly targeting cyclins and cyclin-dependent kinases

(CDKs), including CyclinE and CDK269, 70, 71. The functional outcomes of these phosphorylation events appear to be highly cell type- and context-dependent, as the ROCKs have been suggested to play both pro- and anti-proliferative roles. The roles of the ROCKs in cell survival also appear to be context- dependent, as numerous cancer lines have been shown to rely on ROCK activity for survival, while ROCK activation during apoptosis has been suggested to be required for the progression of cell death by inducing

14 membrane blebbing and DNA fragmentation69, 70, 71. The ROCKs may also play a complex role in the regulation of autophagy, with starvation-induced autophagy being dependent on ROCK1, and ROCK2 playing an antagonistic role in the autophagic response80, 81, 82. Insight into these often-contrasting findings will require a mechanistic examination of both the upstream regulators leading to ROCK activation as well as their downstream effector targets in these settings.

In addition to regulating cytoskeletal reorganization and cell survival, characterizations of the ROCKs as regulators of gene expression reprogramming have begun to emerge. ROCK-dependent regulation of gene expression occurs through both direct and indirect mechanisms. Interestingly, the ROCK-dependent regulation of transcription may, in some cases, be closely linked with its regulation of cytoskeletal dynamics. For example, the ROCKs are able to induce a profibrotic transcriptional program via a mechanism that involves severing of the actin cytoskeleton to induce the nuclear shuttling of the MRTF-A/B and YAP/TAZ transcription factors83. The ROCKs can also directly target transcriptional regulators and active ROCK has been detected in the nucleus of several cell types70, 71. Supporting a direct role for the ROCKs in the regulation of gene expression, ROCK2 can phosphorylate and enhance the acetyltransferase activity of p30084.

The RhoA-ROCK Pathway in Immune Responses While the RhoA-ROCK pathway has been extensively investigated in nonhematopoietic cells, its role in the immune system is just beginning to be examined64. The best-studied involvement of the RhoA-ROCK pathway in

15 lymphocyte biology has focused on its effects on cytoskeletal reorganization. The utilization of a RhoA activity biosensor has demonstrated that active RhoA can be detected in migrating T cells at the leading edge in lamellipodia and filopodia68. Interestingly, activated RhoA was also detected in the uropod of T cells, suggesting that the ROCKs can play multiple roles in controlling cellular migration70. Indeed, the ROCKs can promote uropod contractility, modulate integrin-mediated T cell adhesion, and have been shown to be required for transendothelial migration (TEM), especially through endothelial cells with low permeability68, 85. Involvement of the RhoA-ROCK pathway in additional T cell cytoskeletal processes, such as the regulation of lipid raft dynamics, has also been suggested86, 87.

In addition to cytoskeletal effects, the RhoA-ROCK pathway has also been implicated in the regulation of T helper (TH) cell differentiation. RhoA-deficient T

88 cells exhibit impaired TH2, but not TH1 differentiation in vitro . This effect may be dependent on ROCK1, as heterozygous ROCK1-deficient mice exhibited decreased expression of TH2 cytokines in bronchoalveolar lavage (BAL) fluid from a murine model of allergic inflammation89. In contrast, ROCK2 is selectively activated downstream of TH17-skewing conditions, but not under neutral, TH1-

62 or TH2- conditions . Activated ROCK2 phosphorylates IRF4 and promotes the

62 expression of IL-17 and IL-21, key effector cytokines of the TH17 lineage . In line with these findings, naïve T cells from heterozygous ROCK2-deficient mice exhibit impaired TH17 differentiation, marked by decreases in IL-17 and IL-21 production, as well as attenuated expression of RORgt, the master regulator of

62 the TH17 lineage . These findings have also been corroborated in human T cells, whereby an induction in ROCK activity is promoted downstream of TH17-

16 skewing signals77, 90. Addition of the selective ROCK2 inhibitor, KD025, is also able to dampen TH17 differentiation in human T cells by decreasing the activation of STAT3, suggesting that ROCK2 may target multiple pathways to coordinate cell differentiation72. The suppression of STAT3 signaling corresponded with an increase in STAT5 phosphorylation, suggesting that

72 ROCK2 activation may skew the TH17-Treg balance . Through the disruption of the STAT3-STAT5 balance, ROCK2 has also been suggested to regulate the

91 differentiation of follicular T helper (TFH) cells . Although additional studies are needed to more precisely delineate the roles of the ROCKs in T cell activation, these initial investigations suggest that ROCK1 and ROCK2 may promote the differentiation of distinct TH subsets.

Despite the limited studies examining the RhoA-ROCK pathway in T lymphocytes, the importance of this pathway in B cell biology has been even less explored. RhoA was shown to play a non-redundant role in peripheral B cell development through dual effects in promoting the expression of and signaling through the BAFF receptor92. However, whether these are ROCK- mediated effects or whether the ROCKs play roles in B cell development remains unknown. Studies utilizing various in vitro systems have shown that

ROCK1 activity is required for antigen extraction from immune synapses by the BCR and that activated RhoA interferes with the ability of TLR ligands to enhance BCR signaling by restricting BCR mobility93, 94. These studies, along with others, whereby an involvement of RhoA was described in the regulation of BCR-induced proliferation95, suggests that the RhoA-ROCK pathway plays a complex role in regulating B cell responses downstream of dynamic stimuli. The RhoA-ROCK pathway has also been shown to regulate the migration of B cells,

17 as ROCK inhibition can alter the adhesion and migration of both normal and malignant B cells96, 97. Recent studies have, furthermore, uncovered a role for one of the Rho-GEFs, ARHGEF1, in the retention of B cells within the GC98. Understanding the contexts in which the ROCKs control pro- and anti-migratory effects will be of great interest for the proper therapeutic targeting of this pathway.

Pathophysiological B Cell Responses IRF4 and Immune Dysregulation Autoimmunity. The central role of IRF4 in B cell biology has been further corroborated by the finding that deregulation in IRF4 activity plays a fundamental role in the pathogenesis of several diseases, including autoimmunity and B cell malignancies. Systemic autoimmune disorders, such as Systemic Lupus Erythematosus (SLE), are often characterized by autoantibody production and multi-organ involvement as a result of aberrant activation of T and B cell subsets99. Indeed, dysregulation of several IRF4- regulated TH effector subsets, including TH17 and TFH cells, have been shown to contribute to autoimmune pathogenesis in both murine models and in patients100, 101. Furthermore, GC and PC abnormalities are also associated with autoimmunity102, 103. Roles for IRF4 in mediating the pathogenesis of systemic autoimmunity are supported by murine studies in autoimmune B6.lpr mice, in which deletion of IRF4 offers protection from the development of a lupus-like syndrome104. Additionally, as discussed in more detail below, dysregulated post-translational modifications that alter the transcriptional activity of IRF4 have also been observed in various murine models of systemic autoimmunity62, 105.

18 Lymphomagenesis. In addition to autoimmune pathogenesis, IRF4 dysregulation is also a common feature of several B cell malignancies, including Activated B cell (ABC-) diffuse large B cell lymphoma (DLBCL) and Multiple Myeloma (MM)53, 106. Relevant to the discussion of physiological GC responses above, most B cell lymphomas originate from dysregulated GC responses and aberrancies in the modulators of GC responses are often observed in B cell lymphomas107, 108 (Figure 4). Despite their similar origins, B cell non-Hodgkin lymphomas (B-NHLs) comprise a range of genetically and phenotypically distinct malignancies that can be further subcategorized via the cell of origin (COO) classification system, whereby malignancies are classified based on the correspondence of their gene expression profiles with physiological stages in the GC B cell response107, 108. For instance, Burkitt lymphoma (BL) are believed to derive from centroblasts and exhibit translocations that result in the ectopic expression of MYC109, 110, 111. BLs are also characterized by aberrant activation of the PI3K signaling pathway112. As discussed below, inhibitory mutations in the RhoA pathway that disrupt GC B cell migration and confinement are also observed in a subset of BL cases98.

Diffuse large B cell lymphomas (DLBCLs) are the most common B-NHL, comprising approximately 40% of newly diagnosed B-NHL cases. DLBCLs are furthermore divided into three distinct subsets: GC B cell-like (GCB)-DLBCLs, which derive from centrocytes; Activated B cell (ABC)-DLBCLs, which genetically resemble post-GC cells in the early stages of PB/PC differentiation; and primary mediastinal large B cell lymphoma (PMBCLs), which are believed to derive from a thymic post-GC B cell109, 113, 114. GCB-DLBCLs often exhibit abnormalities in epigenetic regulators that promote the expression of a GC B

19 cell-like transcriptome115, 116, 117, 118. For instance, gain-of-function mutations in EZH2, a methyltransferase that supports the BCL6-dependent transcriptional program by repressing key proliferation and terminal differentiation regulators, including CDKN1A, PRDM1 and IRF4, is observed in ~21% of GCB-DLBCLs cases115, 119, 120.

Contrasting the GC-like phenotype of BLs and GCB-DLBCLs, ABC-DLBCLs resemble early plasmablast-like cells and are characterized by chronic signaling downstream of the BCR, culminating in the constitutive activation of NF-kB, and a disruption in the transcription factor network controlling terminal differentiation107. NF-kB activation is critical for the survival of ABC-DLBCLs and many genetic alterations in ABC-DLBCLs converge on the activation of this transcription factor family. Indeed, ~20% of ABC-DLBCLs cases carry activating mutations in CD79A/B, signaling components of the BCR complex, that culminate in NF-kB activation121. Mutations in downstream regulators of NF-kB activation are also commonly observed in ABC-DLBCLs including activating mutations in CARD11 (observed in ~10% cases), a scaffolding protein that promotes NF-kB activation downstream of various signals including BCR and CD40 stimulation, and inactivating mutations in TNFAIP3 (the gene encoding

A20; observed in ~30% cases), a negative regulator of NF-kB activity122, 123. In addition to driving chronic BCR signaling, common mutations in ABC-DLBCLs can also result in the activation of several other pro-survival pathways, including TLR, STAT3, and IFN signaling108, 124.

In addition to chronic activation of pro-survival and proliferative pathways, ABC- DLBCLs also exhibit mutations that prevent their terminal differentiation107. One

20

Figure 4. B-NHLs are Classified Based on Genetic Correspondence with Physiological Stages in B Cell Differentiation. Schematic and table showing incidences, prognoses, and common aberrancies in Burkitt’s lymphomas (BLs), Germinal Center B cell-like (GCB-) diffuse large B cell lymphomas (DLBCLs), and Activated B cell (ABC-) DLBCLs. BLs appear to derive from centroblasts and exhibit translocations that result in the ectopic expression of MYC. GCB- DLBCLs appear to derive from centrocytes and contain deregulations in several epigenetic regulators known to promote the GC B cell program. BLs and GCB- DLBCLs have also been reported to contain inactivating mutations in the RhoA pathway. Patients with ABC-DLBCL exhibit a lower survival rate and poorer prognosis compared to those with GCB-DLBCL and BL. ABC-DLBCLs genetically resemble exiting GC B cells, and often contain deregulations in the pathways that promote PB differentiation, including IRF4. ABC-DLBCLs have a dependence on IRF4 activity for their survival.

21 common mechanism by which this is achieved is through a blockade in BLIMP1 upregulation in these cells125, 126, 127. Indeed, bi-allelic inactivation of PRDM1 (the gene encoding for BLIMP1) is observed in ~30% of ABC-DLBCL cases125. Alternatively, another mechanism by which terminal differentiation can be blocked is through the failure of these cells to fully suppress BCL6 expression25, 54. Persistent BCL6 expression is often a result of chromosomal translocations by which regulatory regions of the BCL6 locus are altered. Among the regulatory regions commonly affected by translocation are the IRF4 binding sites that promote BCL6 repression54. Interestingly, given the post-GC nature of these cells, ABC-DLBCLs express high levels of activated IRF4106. IRF4 contributes both to the survival of ABC-DLBCLs by repressing autocrine Type I IFN signaling, and IRF4 activity is also required to maintain the PB-like features of these cells106. Illustrating the importance of IRF4 activity in ABC-DLBCL pathogenesis, silencing of IRF4 selectively induced lethality in the ABC-DLBCL subset, but not in GCB-DLBCLs106. Furthermore, gain-of-function mutations in the IRF4 interacting partner, SPIB, are seen in ~25% of cases and silencing of SPIB also results in lethality of the ABC-DLBCL subset128, 129. Importantly, the molecular phenotypes of B-NHLs seem to have clinical significance, as ABC- DLBCLs have a poorer prognosis and require a more aggressive treatment course compared to GCB-DLBCLs and BL.

The RhoA-ROCK Pathway and Immune Dysregulation Autoimmunity. While deregulations in the RhoA-ROCK pathway have been well described in cardiovascular, renal, and neurological disorders73, 130, 131, their impact on the pathogenesis of immune-mediated diseases is just beginning to be appreciated132. Consistent with the ability of ROCK to phosphorylate IRF4

22 62 and regulate TH17 differentiation , aberrant activation of this pathway has been observed in murine models of rheumatoid arthritis (RA) and SLE. Indeed, T cells from a spontaneous mouse model of RA exhibited increased activation of ROCK2 and dysregulated production of IL-17 and IL-21, which was shown to be dependent on both RhoA and ROCK262. Enhanced ROCK activation has also been observed in inducible models of RA, including in synovial tissue from mice with collagen-induced arthritis133. Notably, in vivo administration of the pan-ROCK inhibitor, Fasudil, or the ROCK2-selective inhibitor, KD025, was able to decrease IL-17 and IL-21 production, autoantibody titers, and disease severity in both spontaneous and induced models of RA62, 72.

Aberrant activation of ROCK2 has also been observed in T cells from several spontaneous murine models of SLE, including MRL/lpr mice62. The administration of Fasudil to these mice resulted in an impressive decrease in effector cytokine and autoantibody production and in proteinuria levels62. Furthermore, NZB/W F1 mice, a distinct spontaneous model of SLE, showed decreased PC formation and protection from nephritis upon treatment with the pan-ROCK inhibitor105. Human studies also support the notion that dysregulated ROCK activation might contribute to SLE pathogenesis, as enhanced phosphorylation of ROCK substrates, like the ERM proteins, have been observed in T cells from SLE patients134. Furthermore, approximately 60% of SLE patients display elevated levels of ROCK activity in their PBMCs compared to healthy controls77, 90. Production of IL-17 and IL-21 by SLE T cells is also inhibited following treatment with a pan-ROCK inhibitor, or a selective ROCK2 inhibitor, further supporting a role for the RhoA-ROCK pathway, and ROCK2, in particular, in SLE pathogenesis72, 90. More recently, studies utilizing ROCK2

23 selective inhibitors have also shown beneficial roles for ROCK inhibition in patients with psoriasis and graft-versus-host disease (GVHD)135, 136, suggesting that dysregulation in the RhoA-ROCK pathway may represent a common pathogenic mechanism in multiple autoimmune disorders.

Lymphomagenesis. Dysregulation in components of the RhoA pathway have recently been described in a number of hematopoietic malignancies64. Approximately 60-70% of angioimmunoblastic T cell lymphoma cases, a rare peripheral T cell lymphoma that phenotypically resembles TFH cells, have been found to express an inactivating mutation in RhoA (encoding p.Gly17Val)137, 138, 139, 140. Molecular studies have revealed the Gly17Val RhoA mutant is unable to bind GTP and is believed to act as a dominant-negative by sequestering activated GEFs141. Similar mutations in the GTP-binding domain of RhoA have also been observed in cutaneous T cell lymphoma142. Interestingly, both inactivating mutations in RhoA and mutations that lead to constitutive RhoA activation, have been identified in adult T cell leukemia/lymphoma143, 144. Potentially inactivating mutations in RhoA and in upstream regulators of the RhoA pathway, including S1PR2, GNA13, ARHGEF1, and P2RY8, have also been reported in BL and in 30% of GCB-DLBCL98, 145, 146. RhoA mutations in BL are commonly found within the GEF-binding domain and inhibit the ability of RhoA to bind and be activated by GEFs and may contribute to the transformation and peripheral dissemination of centrocytes146. Whether these alterations in RhoA activation are reflected in aberrant activation of its downstream effectors like the ROCKs is presently unknown. In addition, although the migration of several B cell malignancies, including classic Hodgkin Lymphoma (cHL), chronic lymphocytic leukemia (CLL), and multiple myeloma

24 (MM), can be inhibited by treatment with a pan-ROCK inhibitor96, 97, 147, 148, the precise contributions of ROCK1 and ROCK2 to lymphomagenesis is unknown.

Central Hypothesis In this thesis, I aimed to identify and delineate the functions of ROCK1 and ROCK2 in physiological and pathophysiological B cell differentiation. Given the ability of ROCK2 to phosphorylate and regulate the activity of IRF4 in T cells62, and the critical dependence on IRF4 of physiological humoral responses and of the survival and pathogenesis of ABC-DLBCL106, I hypothesized that ROCK2 played a central role in normal and pathogenic B cell responses, likely through both IRF4-dependent and IRF4-independent mechanisms. Furthermore, I hypothesized that modulations of ROCK activity could offer therapeutic potential both by fine-tuning the effectiveness of physiological humoral responses as well as by disrupting pathophysiological signaling in DLBCLs. In the following chapters, I detail our findings on the roles of ROCK1 and ROCK2 in controlling the biology of ABC-DLBCLs, identify novel roles for B cell-specific ROCK2 in promoting T cell-dependent humoral responses, and propose that ROCK inhibition could represent a novel therapeutic approach for the treatment of diseases characterized by dysfunctional B cell responses.

25 CHAPTER II ROLES OF THE RHO KINASES IN ABC-DLBCLS

B cell non-Hodgkin lymphomas (B-NHLs) encompass a wide-range of distinct malignancies, which include Burkitt’s lymphomas (BLs) and diffuse large B cell lymphomas (DLBCLs)107, 108. While BLs are derived from GC dark zone B cells109, 111, DLBCLs can be subdivided into molecular subtypes corresponding to B cells arrested at different stages of GC transit based on their transcriptional profiles107, 108, 109. The two major DLBCL subtypes are GC B cell-like (GCB-) DLBCLs, whose transcriptome resembles that of light zone B cells, and Activated B cell-like (ABC-) DLBCLs, which instead express markers that normally accompany plasmablast commitment107, 108, 113. These distinctions have important clinical implications since DLBCL patients with an ABC profile exhibit worse survival than patients with a GCB profile107, 108. Patient prognosis, however, is also impacted by the presence of additional abnormalities, such as the expression of MYC and BCL2, which can result from either translocations, as in Double-hit lymphomas (DHLs), or from over-expression, as in Double- expressor lymphomas (DELs)149, 150, 151.

One of the key features distinguishing GCB- from ABC-DLBCLs is the expression of the transcription factor IRF4, which is expressed by ABC- DLBCLs, but not GCB-DLBCLs106. IRF4 is an essential regulator of the development and differentiation of multiple immune subsets, including T and B cells39, 40. Within the T cell compartment, IRF4 is induced downstream of TCR signaling and is involved in mediating the differentiation of several TH subsets, such as TH17 and TFH cells and the production of key cytokines like IL-21, a

26 major regulator of humoral responses43, 45, 46, 47. In B cells, IRF4 is upregulated upon B cell activation and regulates multiple stages of B cell differentiation, including GC B cell formation, class switch recombination, the exit of properly selected GC B cells, and their terminal differentiation into plasmablasts (PBs)/plasma cells (PCs)39, 51, 52. The multifaceted actions of IRF4 rely on its ability to interact with multiple partners including ETS-, AP1-, and other IRF- family members and target distinct regulator elements such as EICEs (ETS-IRF composite elements), AICEs (AP1-IRF composite elements), and ISREs (IRF- stimulated response elements) in a context-dependent manner52, 55, 57, 59, 60. Notably, deregulation of IRF4 expression and/or activity promotes oncogenesis in several B cell malignancies, including ABC-DLBCLs and Multiple Myeloma (MM)53, 106, 152. The diverse tumorigenic effects of IRF4 have again been associated with context-dependent functions of IRF4, whereby its interaction with the ETS-family member SPIB is critical for the survival of ABC-DLBCLs, while a positive auto-regulatory loop between IRF4 and c-MYC fuels myeloma cell survival106, 129, 152.

Given the essential and complex roles of IRF4 in controlling immune responses, regulatory mechanisms must exist to ensure the proper execution of distinct

IRF4-mediated transcriptional programs under different settings. Previous work identified the serine-threonine kinase ROCK2 as a regulator of IRF4 activity in

62 TH17 cells . ROCK2-dependent phosphorylation of IRF4 occurs on two serine residues (S446 and S447), located within the C-terminus of the protein62. Structural studies have indicated that the C-terminal portion of the IRFs contains an autoinhibitory domain that modulates the ability of the IRF association domain (IAD) to interact with other cofactors63. ROCK2-dependent

27 phosphorylation of the S446/S447 residues within IRF4 has functional implications in T cells since it augments the binding of IRF4 to the IL-17 and IL- 21 promoters and leads to the production of high levels of these effector cytokines62. Whether ROCK2 can phosphorylate and regulate IRF4 activity in other cell compartments, like B cells, however, has yet to be explored.

ROCK2 and its only other family member, ROCK1, are highly homologous serine-threonine kinases that serve as major effectors of the Rho subfamily of small GTPases, which includes RhoA69, 70, 71. Like other small GTPases, RhoA cycles between an active (GTP-bound) and an inactive (GDP-bound) state, a process controlled by Rho-GEFs, Rho-GAPs, and Rho-GDIs66. Upon RhoA binding to the RhoA binding domain (RBD), the ROCKs undergo a conformational change that disrupts the association of the autoinhibitory C- terminal region with the N-terminal kinase domain resulting in kinase activation69, 70, 71. The ROCKs are widely expressed and control a diverse range of biological processes such as cytoskeletal dynamics, gene expression, and cell survival, enabling them to act as critical coordinators of a tissue response to stress and injury69, 70, 71. The most extensively characterized role of the RhoA- ROCK pathway in B cells centers on the regulation of cytoskeletal reorganization and encompasses effects on immunological synapse dynamics and B cell migration64. Since ROCK1 and ROCK2 exhibit a high degree of identity in their kinase domains, they can phosphorylate similar substrates in vitro and both kinases are inhibited by commonly available pan-ROCK inhibitors, such as Y-27632 and Fasudil69, 70, 71. Few selective ROCK2 inhibitors, like KD025, which possesses 100-fold selectivity for ROCK2 versus ROCK1,

28 have, however, also been developed74, 75, 76. The precise scope of the isoform- specific roles of ROCK1 and ROCK2 in immune cells is mostly unknown.

Aberrant activation of the RhoA-ROCK pathway has been implicated in the pathogenesis of multiple diseases including cancer73, 130, 131, 153. Genetic alterations affecting the RhoA-GTPase pathway have recently been implicated in lymphomagenesis64. Indeed, approximately 60-70% of Angioimmunoblastic T cell lymphomas (AITLs) have been found to express an inactivating mutation in RhoA (Gly17Val) 137, 138, 139, 140. The RhoA Gly17Val mutant does not bind GTP and is believed to act in a dominant-negative manner by sequestering activated GEFs141. Mutations in components of the RhoA pathway have also been identified in BLs and in ~30% of GCB-DLBCLs98, 145, 146. RhoA mutations in BLs are commonly found within the GEF-binding domain and inhibit the ability of RhoA to bind to and becomes activated by GEFs146. Mice lacking Ga13, an upstream regulator of RhoA, or ARHGEF1, which mediates the activation of

RhoA in response to Ga13 and other receptors, develop B cell-derived lymphomas characterized by the dissemination of GC B cells from the lymph nodes into the periphery, suggesting that this pathway plays a tumor suppressive role in the development of GCB lymphomas98. As additional investigations of the RhoA pathway in lymphomas are being undertaken, however, an unexpected degree of complexity is emerging. Indeed, both loss- and gain-of function RhoA mutations have been identified in adult T cell lymphoma/leukemia142, 143, 144. While RhoA and its upstream regulators have been implicated in lymphomagenesis, the impact of key downstream RhoA effectors, like the ROCKs, in promoting oncogenesis has not been investigated.

29 In view of the central role of IRF4 in ABC-DLBCLs and the ability of ROCK2 to phosphorylate IRF4, in this study we directly assessed the regulation and role of ROCK2 in ABC-DLBCLs. We found that ROCK2 is constitutively activated and phosphorylates IRF4 in ABC-DLBCLs and that this pathway can be induced upon stimulation with key T-cell dependent signals such as those provided by CD40 engagement and IL-21. Disruption of this pathway through the silencing of ROCK2 altered the transcriptional profile of ABC-DLBCLs through both IRF4- dependent and IRF4-independent mechanisms. Furthermore, treatment with a pan-ROCK inhibitor selectively induced lethality in ABC-DLBCLs, suggesting that this pathway could be targeted for the treatment of patients suffering from this disease.

Results IRF4 is Constitutively Phosphorylated in ABC-DLBCLs IRF4 has previously been shown to play a key role in the function and survival of ABC-DLBCLs106. It is, however, unknown whether post-translational modifications can modulate IRF4 function in these cells. Given the ability of ROCK2 to phosphorylate IRF4 in T cells62, we thus investigated whether IRF4 might be aberrantly phosphorylated in ABC-DLBCLs by using an antibody that detects the phosphorylation of IRF4 at S446/S447, the previously identified ROCK2 target sites62. Probing nuclear extracts from a panel of different B cell malignancies with this antibody demonstrated that IRF4 is phosphorylated in several ABC-DLBCL cell lines (Figure 5A). In contrast, BL cell lines, which do not depend on IRF4 for survival106, exhibited no or minimal IRF4 phosphorylation compared to ABC-DLBCLs, despite expressing similar levels

30

Figure 5. IRF4 is Constitutively Phosphorylated in ABC-DLBCLs. (A) Representative western blot analysis (left) and pooled quantification (right) of phosphorylated IRF4 (S446/S447), total IRF4, and housekeeping protein LaminB, from nuclear extracts of BL (Ramos, BL-41, BL-2), GCB-DLBCL (BJAB, DB, HT), and ABC-DLBCL lines (HBL-1, OCI-LY3, RIVA, SU-DHL-2, U2932). Quantification is calculated as the densitometry ratio between phosphorylated IRF4 to total IRF4 and normalized to U2932 levels per experiment. Western blots representative of at least 3 independent experiments per cell line. Error bars show SEM, P-value by ordinary one-way ANOVA. (B) Representative western blot analysis (left) and pooled quantification (right) of indicated proteins from nuclear extracts of ABC-DLBCL lines either left untreated or cultured in the presence of 90µM Y-27632 for 3-6hr. Quantification is calculated as in (A) and normalized per cell line to the untreated condition. Western blots representative of at least 2 independent experiments per cell line. Error bars show SEM, P-value by unpaired two-tailed t-test. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

31 of total IRF4 protein (Figure 5A). Consistent with previous reports, IRF4 was not expressed in GCB-DLBCL cell lines106, 113 (Figure 5A). To assess whether IRF4 phosphorylation in ABC-DLBCLs was dependent on ROCK activity, we cultured these cells in the presence of Y-27632, a pan-ROCK inhibitor. In line with our previous findings that the IRF4 S446/S447 residues are phosphorylated by ROCK2 in T cells62, ROCK inhibition decreased IRF4 phosphorylation in ABC- DLBCL cell lines (Figure 5B). As reported, ABC-DLBCL cell lines displayed a wide-range of constitutively phosphorylated STAT3154 (Figure 6A). STAT3 phosphorylation did not correlate with the levels of phosphorylated IRF4 and, in contrast to IRF4 phosphorylation, was not diminished by Y-27632 treatment (Figure 6B). Thus, IRF4 is constitutively phosphorylated in ABC-DLBCLs in a ROCK-dependent manner.

Given that ABC-DLBCLs are believed to correspond to B cells arrested in a PB-

107, 108 like state , we next investigated whether TFH-derived signals that normally drive the differentiation of GC B cells into PB/PCs, such as aCD40 and IL-2145, 46, 47, could induce the phosphorylation of IRF4. To address this possibility, we cultured Ramos cells, a human BL cell line previously employed as a model system to study the signals driving GC exit54, 155, 156, in the presence of aCD40 and IL-21. While stimulation with either aCD40 or IL-21 alone led to a modest induction of IRF4 phosphorylation, the combination of these two signals resulted in a robust increase in IRF4 phosphorylation levels (Figure 7A). IRF4 phosphorylation downstream of aCD40 and IL-21 was dependent on ROCK activity, as treatment with Y-27632 prevented this phosphorylation event (Figure 7B). Since IRF4 was constitutively phosphorylated in lymphomas exhibiting PB- like features and could be induced downstream of signals that promote PB/PC

32

Figure 6. STAT3 Phosphorylation Is Not Decreased by Y-27632 in B-NHLs. (A) Representative western blot analysis (left) and pooled quantification (right) of phosphorylated STAT3 (Y705), total STAT3, and housekeeping protein HDAC1, from nuclear extracts of BL (Ramos, BL-41, BL-2), GCB-DLBCL (BJAB, DB, HT), and ABC-DLBCL lines (HBL-1, OCI-LY3, RIVA, SU-DHL-2, U2932). Quantification is calculated as the densitometry ratio between phosphorylated STAT3 to total STAT3 and normalized to Ramos levels per experiment. (B) Representative western blot analysis (left) and pooled quantification (right) of indicated proteins from nuclear extracts of BL or ABC- DLBCL lines either left untreated or cultured in the presence of 90µM Y-27632 for 3-6hr. Quantification is calculated as in (A) and normalized per cell line to the untreated condition. Western blots representative of 2 independent experiments per cell line. Error bars show SEM.

33

Figure 7. IRF4 is Phosphorylated Upon Stimulation with aCD40 and IL-21. (A) Representative western blot analysis (left) and pooled quantification (right) of phosphorylated IRF4 (S446/S447), total IRF4, and housekeeping protein LaminB, from nuclear extracts of Ramos cells either unstimulated or following 6hr stimulation with 1µg/mL aCD40 and/or 100ng/mL IL-21. Quantification is calculated as the densitometry ratio between phosphorylated IRF4 to total IRF4 and normalized to the aCD40+IL-21 condition per experiment. Western blots representative of 3 independent experiments. Error bars show SEM, P-value by unpaired two-tailed t test. (B) Representative western blot analysis (left) and pooled quantification (right) of indicated proteins from nuclear extracts of Ramos cells either left untreated or pre-treated with 60-90µM Y-27632 for 2hr before stimulation with aCD40 and IL-21 for 6hr, as in (A). Quantification is calculated as in (A). Western blots representative of 2-4 independent experiments. Error bars show SEM, P-value by unpaired two-tailed t-test. (C) Representative western blot analysis of indicated proteins from lysates of sorted plasmablasts (PBs; Blimp1-YFP+ CD138+) or follicular B cells (FoBs; Blimp1-YFP- CD138- B220+ CD23+) from Blimp1-YFP reporter mice at d7 post-immunization with 100µg NP-CGG. Western blot representative of 3 independent experiments. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

34 differentiation, we next assessed whether IRF4 phosphorylation could be observed in PB/PCs that differentiate under physiological conditions. To this end, we immunized Blimp1-YFP mice157, in which Blimp1-expressing cells can be identified via the expression of yfp, and sorted either follicular B cells (B220+ Blimp1- CD23+) or PB/PCs (Blimp1+ CD138+) from the spleens of these mice. Phosphorylation of IRF4 could be detected in Blimp1+ PB/PCs but not in follicular B cells (Figure 7C), suggesting that IRF4 phosphorylation occurs physiologically during the terminal differentiation of B cells into PB/PCs.

ROCK2 is Constitutively Activated in ABC-DLBCLs Since the phosphorylation of IRF4 detected in ABC-DLBCLs, but not other lymphomas, was dependent on ROCK activity, we next investigated whether ROCK activation was also differentially regulated in distinct subtypes of B cell lymphomas. In order to specifically assess ROCK1- and ROCK2-specific activation, we performed in vitro kinase assays (IVKAs), in which immunoprecipitated ROCK1 or ROCK2 were assessed for their ability to phosphorylate recombinant MYPT1 (rMYPT1), a well-known ROCK substrate62, 158. Nuclear extracts were utilized for these assays since this is the main compartment where IRF4 phosphorylation is detected. While ROCK1 was highly activated in all B-NHL cell lines examined (Figure 8A), high levels of ROCK2 activity were only observed in ABC-DLBCL cell lines (Figure 8B), suggesting that ROCK2 is selectively activated in ABC-DLBCLs. Given that we had observed that IRF4 phosphorylation can be regulated by stimulation of B cells with aCD40 and IL-21, we next evaluated the ability of these signals to upregulate ROCK2 activity. While high levels of ROCK1 activation could be observed irrespective of the stimulatory conditions (Figure 8C), exposure of

35 Ramos cells to either aCD40 or IL-21 alone induced the activation of ROCK2, an effect that was further augmented in the concomitant presence of both aCD40 and IL-21 (Figure 8D). Collectively, these data suggest that while ROCK1 is constitutively activated in several B-NHLs, ROCK2 is selectively activated in ABC-DLBCLs. Furthermore, ROCK2 activation can be induced upon stimulation of B cells with signals, like aCD40 and IL-21, which normally promote B cell terminal differentiation.

ROCK activation is primarily induced upon binding of activated RhoA, which, in turn, can be regulated by GEFs69, 70, 71. While at least 24 GEFs have been reported to activate RhoA, ARHGEF1 is known to be an important GEF regulating RhoA activity in B cells98, 159, 160, 161. Furthermore, we previously observed an association between ARHGEF1 activation and ROCK2 activity in CD4+ T cells158. To assess the activation state of ARHGEF1 in B-NHL lines, we employed a RhoA-G17A pull-down assay (Figure 8E). The RhoA-G17A mutant can stably bind to active Rho-GEFs, facilitating the detection of specific GEFs that are activated under different stimulatory conditions. Nuclear extracts from B-NHL cell lines were incubated with recombinant nucleotide-free RhoA-G17A, precipitates were subjected to SDS-PAGE, and then immunoblotted with an anti-ARHGEF1 antibody. ARHGEF1 was highly activated in several ABC- DLBCL cell lines (Figure 8E). Furthermore, while only low levels of ARHGEF1 activation could be observed in Ramos cells in the absence of stimulation, ARHGEF1 activation could be induced upon stimulation with IL-21 (Figure 8E). Thus, increased ARHGEF1 activity is likely to contribute to the activation of ROCK2 in ABC-DLBCLs.

36

Figure 8. ROCK2 Activity is Constitutive in ABC-DLBCLs and Induced Upon Stimulation with aCD40 and IL-21. (A-B) ROCK1 and ROCK2 kinase activity was examined by incubating immunoprecipitated (A) ROCK1 or (B) ROCK2 from nuclear extracts of BL, GCB-DLBCL, or ABC-DLBCL lines with purified recombinant MYPT1 (rMYPT1) as a substrate. Phosphorylated rMYPT1 (pMYPT1) was detected using an antibody against pMYPT1. Total ROCK1 or ROCK2 input levels for each sample are shown in the lower panels. (C-D) ROCK1 and ROCK2 kinase activity was examined as in (A-B) from nuclear extracts of Ramos cells either unstimulated or following 6hr stimulation with 1µg/mL aCD40 and/or 100ng/mL IL-21. Western blots representative of 2-3 independent experiments (left) and pooled quantifications are calculated as the densitometry ratio between phosphorylated MYPT1 to total ROCK1 or ROCK2 input and normalized to (A-B) OCI-LY3 levels or to (C-D) the aCD40+IL-21 condition per experiment (right). (E) RhoA-G17A-conjugated agarose beads were used to pull-down active ARHGEF1 from whole cell extracts of BL, GCB- DLBCL, or ABC-DLBCL lines, and from whole cell extracts of Ramos cells either unstimulated or following 6hr stimulation with 1µg/mL aCD40 and/or 100ng/mL IL-21. Western blot representative of 3 independent experiments (left) and quantification is calculated as the densitometry ratio between pulled-down ARHGEF1 to input ARHGEF1 and is normalized to the aCD40+IL-21 condition per experiment (right). Error bars show SEM, P-value by unpaired two-tailed t- test. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

37

38 ROCK2 Phosphorylates IRF4 and Regulates IRF4 Activity in Ramos B Cells The strong association between the activation state of ROCK2 and IRF4 phosphorylation detected in Ramos B cells upon stimulation or constitutively in ABC-DLBCLs led us to hypothesize that IRF4 phosphorylation in these settings was primarily mediated by ROCK2. To start addressing this possibility, we first silenced ROCK1 or ROCK2 in Ramos cells with shRNA constructs targeting either of the two ROCKs (Figure 9A-B). While silencing of ROCK1 had only a small effect on IRF4 phosphorylation, silencing of ROCK2 markedly decreased the phosphorylation of IRF4 upon stimulation with aCD40 and IL-21 (Figure 9C). Thus, ROCK2 is the major ROCK isoform mediating IRF4 phosphorylation downstream of CD40 and IL-21.

In addition to promoting the phosphorylation of IRF4, CD40 stimulation induces Ramos cells to acquire a gene expression signature resembling that of B cells exiting the GC54, 155, 156. This transition is characterized by the IRF4-dependent repression of BCL6 expression and the upregulation of other targets including PRDM1, a central hub in the PB/PC transcriptional network54. To begin assessing whether IRF4 phosphorylation might affect this transition, we stimulated Ramos cells silenced for either ROCK1 (ROCK1 knockdown) or

ROCK2 (ROCK2 knockdown) and monitored the expression of known IRF4 targets. The upregulation of several IRF4 target genes, including PRDM1, ELL2, IRF4, CIITA, and IL10 was diminished in stimulated Ramos ROCK2 knockdown cells but not in Ramos ROCK1 knockdown cells (Figure 10A). The effects of the ROCK2 knockdown on PRDM1 and IRF4 mRNA expression furthermore corresponded to a marked reduction in protein levels following stimulation (Figure 10C-D). The upregulation of few targets, like ZEB2, was diminished not

39

Figure 9. ROCK2 Phosphorylates IRF4 Upon Stimulation with aCD40 and IL-21. (A) Representative western blot analysis (left) and pooled quantification (right) of ROCK1, ROCK2, and housekeeping protein b-Tubulin, from lysates of Ramos cells after stable lentiviral infection with shRNA constructs targeting either ROCK1 (ROCK1 KD; orange) or ROCK2 (ROCK2 KD; blue) or with a scrambled shRNA control construct (Scr; black). Western blot representative of 3 independent infections. Quantification is calculated as the densitometry ratio between each ROCK protein to b-Tubulin and normalized to the scrambled shRNA control per experiment. Error bars show SEM, P-value by ordinary one- way ANOVA. (B) Representative RT-qPCR analysis of ROCK1 and ROCK2 expression relative to 28S. Data representative of 2 independent infections. Error bars show SD, P-value by ordinary one-way ANOVA. (C) Representative western blot analysis (left) and pooled quantification (right) of phosphorylated IRF4, total IRF4, and HDAC1 (as loading control), from nuclear extracts of stable Ramos ROCK KD lines left either unstimulated or following 6hr stimulation with 1µg/mL aCD40 and/or 100ng/mL IL-21. Quantification is calculated as the densitometry ratio between phosphorylated IRF4 to total IRF4 and normalized to the aCD40+IL-21 condition in the scrambled shRNA control lines per experiment. Western blots representative of 3 independent experiments. Error bars show SEM, P-value by unpaired two-tailed t test. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

40

Figure 10. ROCK2 Regulates the Expression of a Subset of IRF4 Target Genes. Stable Ramos ROCK KD lines were left untreated or stimulated for 6hr with 1µg/mL aCD40 and/or 100ng/mL IL-21. (A-B) Pooled RT-qPCR analysis of indicated PRDM1, ELL2, IRF4, CIITA, IL10, and ZEB2. Data pooled from 4 independent experiments. Error bars show SEM, P-value by one-way ANOVA. (C) Representative western blot analysis of BLIMP1 and housekeeping protein HDAC1 from nuclear extracts of stable Ramos ROCK KD lines. Western blot representative of 2 independent experiments. (D) IC FACS histograms of IRF4. (E) Pooled RT-qPCR analysis of BCL6, PAX5, and XBP1. Data pooled from 4 independent experiments. Error bars show SEM, P-value by one-way ANOVA. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

41 only in ROCK2 knockdown cells, but also in ROCK1 knockdown cells (Figure 10B). Notably, ROCK2 silencing did not exert global effects on other IRF4 targets, such as BCL6, or other key regulators of B cell differentiation, like PAX5 and XBP1 (Figure 10E). Taken together, these data suggest that, upon stimulation of B cells with signals that promote GC B cell exit and PB/PC differentiation, ROCK2 regulates the expression of a specific subset of IRF4 target genes.

To more directly assess whether ROCK2 silencing impacted the ability of IRF4 to target specific regulatory regions, we performed chromatin immunoprecipitation (ChIP) assays followed by qPCR using primers encompassing regulatory sites known to be bound by IRF454, 106, 152. Silencing of ROCK2 resulted in a significant decrease in the binding of IRF4 to regulatory regions within the PRDM1 locus, including the promoter region and a site within intron 4 (Figure 11A). Binding of IRF4 to a previously identified site in the promoter region of ELL2 and in the IL10 enhancer was similarly decreased following ROCK2 silencing (Figure 11B). In line with the inability of ROCK2 silencing to affect BCL6 transcripts, binding of IRF4 to the BCL6 promoter was not significantly altered in the ROCK2 knockdown lines (Figure 11C). To further assess the effects of phosphorylation on the DNA binding activity of IRF4, we transfected 293T cells, which contain baseline ROCK activity, with either a wild- type IRF4 construct (IRF4 WT) or a construct encoding a mutant of IRF4 that cannot undergo S446/S447 phosphorylation (IRF4 AA) and performed oligonucleotide precipitation assays (ONPs) using oligonucleotides containing the IRF4 binding sites in the IL10 enhancer or the ELL2 promoter (Figure 11D). Binding of the IRF4 AA mutant to both sites was lower than the binding observed

42

Figure 11. ROCK2 Controls the Binding of IRF4 to Several Regulatory Regions. (A-C) Representative IRF4 ChIP-qPCR analysis of IRF4 binding to regulatory regions in the (A) PRDM1, (B) ELL2, IL10, and (C) BCL6 loci from stable Ramos ROCK2 KD lines left untreated or following stimulation for 6hr with 1µg/mL aCD40 and/or 100ng/mL IL-21. Data representative of 2 independent experiments. Error bars show SD, P-value by unpaired two-tailed t test. (D) Oligonucleotide precipitation assay (ONP) of nuclear extracts from 293T cells transfected with WT or phosphomutant (AA) IRF4, assessed with a biotinylated oligonucleotide from the IL10 enhancer region (top) or the ELL2 promoter region (bottom), followed by immunoblot analysis of precipitated proteins or input with anti-IRF4. Data representative of two independent experiments (left). Quantification (right) calculated as the densitometry ratio between precipitated IRF4 to input IRF4. (E) 293T cells were co-transfected with MYC-tagged IRF4 WT or MYC-tagged IRF4 AA and either FLAG-tagged IRF4 WT or FLAG-tagged IRF4 AA. Immunoprecipitations were performed using an anti-FLAG antibody and analyzed by immunoblotting using anti-MYC and anti- FLAG antibodies. Western blots representative of 3 independent experiments. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

43 with IRF4 WT. To assess whether IRF4 phosphorylation by ROCK2 could alter its ability to homodimerize, we furthermore transfected 293T cells with various combinations of differentially tagged IRF4 WT and IRF4 AA constructs followed by co-immunoprecipitations. IRF4 homodimers could be observed when IRF4 WT-MYC and IRF4 WT-FLAG proteins were co-expressed or when IRF4 WT- FLAG was expressed with IRF4 AA-MYC, but not when IRF4 AA-MYC and IRF4 AA-FLAG were co-expressed (Figure 11E). Taken together, these data suggest that the ROCK2-dependent phosphorylation of IRF4 promotes IRF4 homodimerization and its binding to selected regulatory regions.

ROCK2 Phosphorylates IRF4 and Promotes an IRF4-Regulated Transcriptional Program in ABC-DLBCLs The ability of ROCK2 to regulate the activity of IRF4 in response to signals driving PB/PC differentiation raised the possibility that its constitutive activation in ABC-DLBCLs could contribute to their phenotype. To more directly evaluate the role of ROCK2 in ABC-DLBCLs, we utilized shRNA constructs to silence ROCK2 in U2932 cells, generating U2932 ROCK2 knockdown cells (Figure 12A-B). U2932 ROCK1 knockdown cells were also generated to directly compare the roles of the two ROCK isoforms in these cells (Figure 12A-B). In line with the results obtained in stimulated Ramos cells, the constitutive phosphorylation of IRF4 in U2932 cells was decreased upon silencing of ROCK2, but not of ROCK1 (Figure 12C), suggesting that ROCK2 is the major ROCK isoform responsible for the phosphorylation of IRF4 in ABC-DLBCLs.

44

Figure 12. ROCK2 Phosphorylates IRF4 in ABC-DLBCLs. (A) Representative western blot analysis (left) and pooled quantification (right) of ROCK1, ROCK2, and housekeeping protein b-Tubulin, from lysates of U2932 cells after stable lentiviral infection with shRNA constructs targeting either ROCK1 (ROCK1 KD; orange) or ROCK2 (ROCK2 KD; blue) or with a scrambled shRNA control construct (Scr; black). Western blot representative of 4 independent infections. Quantification is calculated as the densitometry ratio between each ROCK protein to b-Tubulin and normalized to the scrambled shRNA control per experiment. Error bars show SEM, P-value by ordinary one- way ANOVA. (B) Representative RT-qPCR analysis of ROCK1 and ROCK2 expression relative to 28S. Data representative of 2 independent infections. Error bars show SD, P-value by ordinary one-way ANOVA. (C) Representative western blot analysis (left) and pooled quantification (right) of phosphorylated IRF4 and total IRF4 from nuclear extracts of stable U2932 ROCK KD cells. Quantification is calculated as the densitometry ratio between phosphorylated IRF4 to total IRF4 and normalized to the scrambled shRNA control levels per experiment. Western blots representative of 5 independent experiments. Error bars show SEM, P-value by unpaired two-tailed t test. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

45 In an effort to more broadly examine the functional relevance of constitutive ROCK2 activation in ABC-DLBCLs, we next performed RNA-seq analysis on U2932 ROCK1 knockdown and ROCK2 knockdown cells and compared the findings to those in scrambled shRNA controls. Principal component analysis revealed that ROCK1 and ROCK2 both regulated a unique transcriptional profile in ABC-DLBCLs (Figure 13A). We identified 1391 genes that were differentially expressed (false discovery rate (FDR), q <0.05) in U2932 ROCK2 knockdown cells compared to scrambled shRNA control cells. Over-representation pathway analysis of the differentially expressed genes following ROCK2 silencing revealed that ROCK2 induced genes encoded proteins involved in the regulation of gene expression (such as TAF1C, TFAP2A, and ELL), in DNA repair pathways (such as XRCC3, BRCA1, RAD9A, RAD52, BRCA2, and MDC1), and several cell cycle regulators (including CDT1, CDK11A, E2F2, and E2F1) (Figure 13B-E). In contrast, ROCK2 repressed the expression of genes encoding proteins involved in BCR signaling (such as LYN, CD79A, BLNK, and FYN) and in co-stimulatory signaling (including PDCD1, CD86, and HLA- family members) (Figure 13D and 13F-G). Since terminal differentiation requires B cells to transition from a highly activated state to a plasmacytic phenotype, these findings suggest that ROCK2 may support the PB-like features of ABC-DLBCLs by promoting the expression of pathways that allow for the maintenance of proliferation and survival, while repressing components of the B cell activation program, such as BCR and co-stimulatory signaling.

To identify transcription factors and epigenetic regulators that could potentially mediate these observed transcriptional effects following ROCK2 silencing, we performed an upstream regulator analysis on the ROCK2-regulated geneset

46 using the ENCODE ChIP-seq database. This analysis revealed several transcription factors known to play key roles in the pathogenesis of ABC- DLBCLs, including IRF4 (Figure 14A). Indeed, this analysis revealed 190 IRF4- regulated genes that were also differentially expressed following ROCK2 silencing. To further assess whether ROCK2 significantly regulated the expression of known IRF4 targets in ABC-DLBCLs, the ROCK2-regulated geneset in U2932 was compared with a previously published dataset of IRF4 targets in in ABC-DLBCLs106 (Figure 14B). We found a significant enrichment of IRF4 repressed targets in U2932 ROCK2 knockdown cells compared to scrambled shRNA controls (Figure 14B-C). IRF4-induced genes, however, did not show a significant enrichment between the U2932 ROCK2 knockdown cells and scrambled shRNA controls (Figure 14B). Among the IRF4-repressed targets, ROCK2 repressed the expression of several genes including IL10, CIITA, BHLHE40, and ZEB2 (Figure 14C-D). Interestingly, some of these ROCK2-repressed IRF4 targets, such as IL10, ZEB2, and CIITA, were induced by ROCK2 in Ramos following aCD40 and IL-21 stimulation (Figure 10A-B). These findings offer the suggestion that phosphorylation of IRF4 can have distinct functional consequences at various stages of B cell differentiation, likely due to the dynamic expression and activity of IRF4 interacting partners, such as

ETS-family members. Taken together, these data show that ROCK2 modulates the expression of IRF4-repressed target genes in ABC-DLBCLs.

ROCK2 Promotes MYC Protein Levels in ABC-DLBCLs In addition to IRF4, MYC was also identified as an upstream regulator of the ROCK2-regulated geneset in U2932 (Figure 14A). Interestingly, there was a substantial overlap between the genes regulated by MYC and IRF4, as 119 out

47

Figure 13. ROCK2 Regulates a Transcriptional Program in ABC-DLBCLs. (A) Principal component analysis of RNA-seq expression data from all biological replicates of stable U2932 ROCK KD cells (n = 3). (B) Plots showing the - log10(p-value) values for the top over-represented pathways from the genes (top) induced or (bottom) repressed by ROCK2 in U2932. (C-G) Heat map depictions of differentially expressed genes in the U2932 ROCK2 KD cells compared to the scrambled shRNA control cells composing the (C) gene expression, (D) DNA repair, (E) cell cycle, (F) BCR signaling, and (G) costimulation by the CD28 family pathways from the over-representation pathway analysis in (B).

48

Figure 14. ROCK2 Promotes an IRF4-Regulated Transcriptional Program in ABC-DLBCLs. (A) Plot showing the top enriched upstream regulators of the ROCK2-regulated geneset in U2932 using EnrichR analysis. Dotted line indicates significance cutoff at p =0.05. (B) GSEA plot showing the significant enrichment of previously defined (left) IRF4-repressed and (right) IRF4-induced genesets from ABC-DLBCLs in U2932 ROCK2 knockdown cells. (C) Heat map depiction of target genes identified in (B) that are contributing to the enrichment of the IRF4-repressed targets in ABC-DLBCL pathway in the U2932 ROCK2 KD cells. (D) Representative RT-qPCR analysis of IL10, CIITA, BHLHE40, ZEB2 relative to 28S in U2932 ROCK1 knockdown (orange), ROCK2 knockdown (blue), or scrambled shRNA control cells (black). Data representative of 3 independent experiments. Error bars show SD, P-value by ordinary one-way ANOVA. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

49 of the 190 differentially expressed IRF4 targets were also MYC targets (Figure 15A). To start assessing the mechanisms by which ROCK2 might regulate MYC, we cultured ABC-DLBCL cells in the presence of either Y-27632, a pan- ROCK inhibitor, or KD025, a selective ROCK2 inhibitor. No significant changes in MYC transcript levels were observed upon ROCK2 inhibition (Figure 15B). Given that MYC is well-known to be under complex post-transcriptional regulation164, an evaluation of MYC protein levels was also conducted (Figure 15C). Interestingly, in the presence of the selective ROCK2 inhibitor, there was a decrease in the protein levels of MYC in all ABC-DLBCL cell lines examined, while no inhibitory effects were observed in HT, a GCB-DLBCL cell line (Figure 15C). We also did not observe decreases in MYC protein levels in the ABC- DLBCL cell lines upon treatment with Y-27632 (Figure 15C). Given that activation of Rho-family GTPases have been preciously shown to lead to phosphorylation of MYC and promote MYC protein stability165, 166, we evaluated whether pretreatment of ABC-DLBCLs with MG-132, a 26S proteasome inhibitor, could alter the effects of ROCK2 inhibition on MYC protein levels (Figure 15D-E). Pretreatment with MG-132 partially rescued the decreased CMYC protein levels observed upon addition of the ROCK2 selective inhibitor, KD025, but had no effect on transcript levels (Figure 15D-F). No differences in

IRF4 protein levels were instead observed with these perturbations (Figure 15D- E). These data suggest that ROCK2 supports a MYC-regulated transcriptional program in ABC-DLBCLs by promoting the stability of MYC protein.

50

Figure 15. ROCK2 Promotes MYC Protein Levels in ABC-DLBCLs. (A) Venn diagram showing the overlap of IRF4 and MYC targets from ENCODE upstream regulator analysis on the U2932 ROCK2-regulated geneset in Figure 14A. (B) Pooled RT-qPCR analysis of MYC relative to 28S from ABC-DLBCL and GCB- DLBCL lines cultured for 6hr either alone, with 90µM Y-27632, or with 1-5µM KD025. Data pooled from 3 independent experiments. Error bars show SEM, P- value by ordinary one-way ANOVA. (C) Representative western blot analysis of MYC and housekeeping protein b-tubulin, from lysates of DLBCL lines treated as in (B). (D-F) HBL-1 (ABC-DLBCL line) was pre-treated for 1hr with 5µM MG- 132, followed by treatment with either 90µM Y-27632 or 1µM KD025 for 3-6hr. (D) Representative western blot analysis and (E) pooled quantification of MYC, IRF4, and housekeeping protein b-tubulin. Western blot representative of (D) and quantifications pooled from (E) 2 independent experiments. Error bars show SEM. (F) Representative RT-qPCR analysis of MYC relative to 28S. Error bars show SD.

51

52 ROCK1 and ROCK2 Regulate Specific Transcriptional Programs in ABC- DLBCLs Most previous studies examining the functions of the ROCKs have primarily relied on the use of pan-ROCK inhibitors and have thus not directly addressed the specific biological roles of ROCK1 and ROCK2. To start delineating the individual contributions of ROCK1 and ROCK2 to the phenotype of ABC- DLBCLs, we performed RNA-seq on the U2932 ROCK1 knockdown cells and compared the differentially expressed genes with those identified following ROCK2 silencing. We identified 740 genes that were significantly (false discovery rate (FDR), q <0.05) upregulated and 950 genes significantly downregulated in U2932 ROCK1 knockdown cells compared to scrambled shRNA controls. While several of the ROCK1 target genes were also differentially expressed following ROCK2 silencing, there was a significant fraction of genes uniquely regulated by ROCK1 (Figure 16A-B). We performed over-representation pathway analysis on these genesets using ConsensusPathDB (CPDB) and compared the differentially expressed genes to the ENCODE ChIP-seq database to identify functional programs and upstream regulators that were specifically regulated by either ROCK1 or ROCK2 in ABC- DLBCLs. ROCK1 induced the expression of several genes involved in the regulation of cholesterol and sterol biosynthesis, including IDI1, SQLE, SC5D, and SREBF2 (Figure 16B). Upstream regulator analysis highlighted several ETS-family members as potential mediators of the ROCK1-induced geneset, including ELF1, ELK4, and GABPA. GABPA was previously reported to target various cell metabolic and biosynthetic genes and was shown to play roles in mature B cell function167, 168. The ROCK1-repressed geneset showed an enrichment of genes involved in the Proteoglycans in cancer pathway (Figure

53 16B), which included several genes encoding integrins and other cytoskeletal proteins (ITGB3, ITGA2, FLNB, MSN), suggesting that ROCK1 might contribute to cytoskeletal reorganization by regulating the transcription of several genes. ROCK1 also repressed the expression of several Rho-GAPs, including ARHGAP12, ARHGAP24, and TAGAP, suggesting that, in ABC-DLBCLs, ROCK1 may reinforce its own activation through the transcriptional repression of RhoA inhibitors.

In addition to the genes uniquely regulated by ROCK1 or ROCK2 in ABC- DLBCLs, we also observed a substantial overlap of genes whose expression was regulated by either ROCK1 or ROCK2 (Figure 16A-B). Over-representation pathway analysis of these genes revealed that both ROCK1 and ROCK2 promoted the expression of genes involved in DNA repair, including BRCA1, MDC1, NELFA, and CHEK1 (Figure 16A). Although ROCK2 was also able to promote the expression of other DNA repair genes, these findings suggest that both kinases may be required for the maximal expression and function of DNA repair pathways in ABC-DLBCLs. Upstream regulator analysis of these ROCK1 and ROCK2 co-induced genes revealed several transcriptional regulators, including MYC, suggesting that ROCK1 may also contribute to MYC activity in

ABC-DLBCLs (Figure 16A). Amongst the genes repressed by ROCK1 or ROCK2, we identified several genes involved in metabolic processes, especially glycogen storage (GYG1, PGM1) (Figure 16B). By repressing the expression of enzymes involved in glycogen storage, the ROCKs may promote the fitness of ABC-DLBCLs by allowing more glucose to be oxidized and utilized catabolically.

54

Figure 16. ROCK1 and ROCK2 Regulate Specific Transcriptional Programs in ABC-DLBCLs. Venn diagrams showing the overlap of ROCK1- and ROCK2- (A) induced and (B) repressed genes from the U2932 ROCK KD RNA-seq (middle). ConsensusPathDB analysis and upstream regulator analysis of overlapping and unique ROCK1- and/or ROCK2- regulated target genes are also shown.

55 Taken together, these data reveal that ROCK1 and ROCK2 regulate transcriptional programs that contain both unique and overlapping features that together, support the phenotype of ABC-DLBCLs.

ROCK Inhibition Induces Lethality in ABC-DLBCLs Given the ability of ROCK1 and ROCK2 to promote a transcriptional program related to DNA repair and metabolic fitness, we next asked whether ROCK signaling contributed to the survival of DLBCLs. To address this, we cultured ABC-DLBCL, GCB-DLBCL, and BL cells in the presence of increasing doses of Y-27632 for 4 days and monitored cell viability with MTS assays (Figure 17A). The ABC-DLBCL cells exhibited significant decreases in viability following treatment with Y-27632 (Figure 17A). The viability of GCB-DLBCL and BL cells, however, was not affected by the pan-ROCK inhibitor (Figure 17A). In addition to the decreased viability in response to Y-27632, we also observed an accumulation of non-viable sub-G0 cells and an induction in caspase-3 activity in the ABC-DLBCL cells at earlier timepoints after Y-27632 treatment (Figure 17B-C). In contrast to the induced lethality upon pan-ROCK inhibition, treatment with KD025, the selective ROCK2 inhibitor, only minimally effected the survival of ABC-DLBCL cells, suggesting that both ROCK isoforms contribute to the survival of ABC-DLBCLs (Figure 17D). To further assess the potential of ROCK inhibition as a therapeutic approach for ABC-DLBCLs, we generated xenograft models in which an ABC-DLBCL cell line (U2932) or a GCB-DLBCL cell line (HT) was engrafted into immunodeficient NSG mice. Once the tumors were implanted and started growing, mice were treated with Y-27632 at a dose in which we could detect decreases in IRF4 phosphorylation (Figure 17D). Pan- ROCK inhibition resulted in reduced tumor growth in the ABC-DLBCL xenograft,

56

Figure 17. Pan-ROCK Inhibition Induces Lethality in ABC-DLBCLs. (A) Viability (MTS proliferation assay) analysis of BL, GCB-DLBCL, and ABC- DLBCL cells following 4d treatment with 0-90µM Y-27632. Data pooled from 2- 4 independent experiments per cell line. Error bars show SEM, P-value by unpaired two-tailed t test. (B) Representative IC FACS histogram of propidium iodide (PI) incorporation showing sub-G0 cell populations in DLBCL cells either left untreated or following 48hr treatment with 90µM Y-27632. Data representative of 3 independent experiments per cell line. (C) Representative IC FACS histograms of cleaved caspase-3 in DLBCL cells either left untreated or following 24hr treatment with 90µM Y-27632. Data representative of 3 independent experiments per cell line. (D) Representative IC FACS histogram of propidium iodide (PI) incorporation showing sub-G0 cell populations in DLBCL cells either left untreated for following 48hr treatment with 1µM KD025. Data representative of 2 independent experiments. (D) Western blot analysis (left) and quantification (right) of phosphorylated IRF4, total IRF4, and housekeeping protein HDAC1 from nuclear extracts of U2932 ABC-DLBCL lines established as a subcutaneous tumor in immunodeficient NSG mice and treated with 20-40mg/kg Y-27632 for 24hr by intraperitoneal injection. Error bars show SEM, P-value by unpaired two-tailed t test. (E-F) U2932 ABC-DLBCL (left) and HT GCB-DLBCL (right) cells were established as a subcutaneous tumor in immunodeficient NSG mice and treated daily for 10-15 days with PBS (vehicle), or Y-27632 (40mg/kg) by intraperitoneal injection. (E) Tumor progression was monitored as a function of tumor volume. (F) Weight of tumor implanted female (left) and male (right) mice following treatment with either PBS or Y-27632. Pooled data from 8-10 mice per treatment condition per cell line. Error bars show SEM, P-value by unpaired two-tailed test. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

57

58 but showed no effect in the GCB-DLBCL xenograft, suggesting that the selective effects of ROCK inhibition on ABC-DLBCL lethality were also observed in an in vivo setting (Figure 17E). Y-27632 administration did not result in any noticeable toxicities and mouse weight remained normal between the vehicle and Y-27632-treated groups, irrespective of sex (Figure 17F). Thus, ROCK activity is required for the survival of ABC-DLBCLs and inhibition of ROCK activity can induce lethality in ABC-DLBCLs.

ROCK Activation in Primary ABC-DLBCLs To assess whether hyperactivation of the RhoA-ROCK pathway was also a feature of primary ABC-DLBCLs, we monitored the phosphorylation of the ERM (pERM) proteins, classical ROCK substrates69, in primary DLBCL tissue by immunohistochemistry. We first validated the use of pERM as a surrogate marker for ROCK activity in DLBCL cell lines by intracellular FACS (Figure 18A- B). DLBCLs exhibited high levels of ERM phosphorylation that was significantly decreased following treatment with Y-27632 (Figure 18A-B). Using pERM as a readout for ROCK activity, we stained a panel of primary DLBCL tumors and categorized pERM expression levels as highly positive, weakly positive, or negative (Figure 18C-D). Out of the 109 DLBCL cases examined, 64% exhibited some level of pERM positivity (Figure 18C-D), suggesting that ROCK is activated in a subset of primary DLBCL cases. Interestingly, in addition to the positively stained ABC-DLBCLs, we also observed ERM phosphorylation in a fraction of GCB-DLBCL tumors, suggesting that ROCK may also be activated in a subset of primary GCB-DLBCLs, although this could be mediated by ROCK1. We furthermore assessed whether ROCK2 activation was a feature of

59

Figure 18. ROCK is Activated in Primary DLBCL. (A-B) (A) Representative IC FACS histograms and (B) pooled quantifications of phosphorylated ERM expression in DLBCL lines either left untreated or treated with 90µM Y-27632 for 3-6hr. Data representative of and quantification pooled from 3 independent experiments. Error bars show SEM, P-value by unpaired two-tailed t test. (C-E) IHC of phosphorylated ERM on primary DLBCL TMAs consisting of GCB- and non-GCB (ABC-) DLBCLs. Plot showing the fraction of highly positive, weakly positive, and negative pERM levels in the (D) total DLBCL cases or (E) subset as GCB-DLBCLs versus non-GCB (ABC-) DLBCLs. (F) GSEA plot showing the significant enrichment of the ROCK2-induced geneset from U2932 in primary ABC-DLBCL cases. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

60 primary ABC-DLBCL by comparing our identified ROCK2-dependent transcriptional profile in ABC-DLBCLs with the transcriptional profile of primary DLBCL cases sequenced at Weill Cornell Medical College. We found that there was a significant enrichment of ROCK2-induced targets in primary ABC- DLBCLs compared to GCB-DLBCLs (Figure 18F), suggesting that ROCK2 activity promotes a gene expression profile characteristic of primary ABC- DLBCLs. Taken together, these data show that ROCK is activated in primary ABC-DLBCL, may contribute to the transcriptional profile of these cells, and that ROCK signaling could potentially be therapeutically targeted for the treatment of ABC-DLBCLs.

Discussion ABC-DLBCLs are particularly aggressive B cell malignancies characterized by chronic BCR signaling and deregulations in the molecular networks controlling plasma cell differentiation107, 108. Here, we have demonstrated that ROCK1 is activated in both GCB-DLBCLs and ABC-DLBCLs, while constitutive activation of ROCK2 is primarily observed only in ABC-DLBCLs. We furthermore show that ROCK2, together with ROCK1, promotes the expression of unique transcriptional programs that support the phenotype of ABC-DLBCLs. At a mechanistic level, ROCK2 regulated the activity of key transcription factors involved in the pathogenesis of ABC-DLBCLs through the direct phosphorylation of IRF4 and through the promotion of MYC protein levels. We furthermore showed that inhibition of ROCK signaling selectively induced lethality in ABC-DLBCLs, but not in GCB-DLBCLs or BLs. These studies thus elucidate previously unknown roles for ROCK2 in the pathophysiological

61 regulation of ABC-DLBCLs and suggest that ROCK inhibition has therapeutic potential for the treatment of B cell malignancies.

IRF4 is a critical regulator of the PB/PC phenotype and deregulations in IRF4 activity have been associated with a range of pathologies, including ABC- DLBCLs39, 53, 106. While previous studies have highlighted the importance of IRF4 expression to the survival and pathogenesis of ABC-DLBCLs106, our study identifies a novel mode of IRF4 regulation in ABC-DLBCLs through its direct phosphorylation by ROCK2. IRF4 phosphorylation affects the ability of IRF4 to bind to several, but not all regulatory regions, thus allowing ROCK2 to regulate a subset of IRF4 functions without controlling global IRF4 activity. Models of IRF4 activity in B cells have evolved around a dose-dependence hypothesis, whereby lower levels of IRF4 expression allow for the interaction of IRF4 with ETS-family members, while increased concentrations of IRF4 promote homodimerization52, 55, 57. These complexes exhibit unique functional consequences, as they direct IRF4 to distinct sites in the genome and promote the expression of specific transcriptional programs. Our findings place ROCK2 as a potential regulator of these interactions, as we showed that IRF4 phosphorylation promotes homodimerization. We also found that IRF4 phosphorylation was induced downstream of signals known to promote physiological terminal B cell differentiation, aCD40 and IL-21, and detected significant levels of phosphorylated IRF4 in primary murine PBs following immunization. These findings suggested that the phosphorylation of IRF4 is not unique to ABC-DLBCLs, but, rather, may represent a physiological process during terminal B cell differentiation that becomes deregulated in ABC-DLBCLs due to the PB-like nature of this lymphoma subtype.

62

Consistent with the high levels of IRF4 phosphorylation in ABC-DLBCLs, we also identified ROCK2 as being constitutively activated in ABC-DLBCLs. In contrast to the heightened activation of ROCK2 in ABC-DLBCLs, we observed low levels of ROCK2 activity in representative GCB-DLBCL and BL cell lines. Given that inactivating mutations in upstream regulators of the RhoA pathway, including ARHGEF1, have been identified in primary GCB-DLBCLs and BLs98, 145, 146, our findings present an interesting model whereby fine-tuning the activity of the RhoA-ROCK pathway may be essential to prevent pathologies; aberrant activation of the RhoA-ROCK pathway may promote ABC-DLBCLs, while inactivation of this pathway may contribute to the pathogenesis of GCB-DLBCLs or BLs. Similar to what was observed with IRF4 phosphorylation, ROCK2 activation was also induced downstream of signals that promote terminal B cell differentiation. We found that ROCK2 activity promoted the expression of several genes downstream of these signals, including PRDM1, IL10, and IRF4. Interestingly, since IRF4 can bind to its own promoter region53, ROCK2 may promote IRF4 autoregulation, suggesting that in addition to direct phosphorylation, ROCK2 may reinforce an IRF4-dependent program by fine- tuning IRF4 expression levels.

ROCK2 also acted in IRF4-independent manners to promote biological processes in ABC-DLBCLs. One such mechanism was through the regulation of MYC protein levels. MYC expression in DLBCLs increase lymphoma aggressiveness and is associated with a poorer clinical prognosis149, 150, 151. Indeed, double-hit lymphomas (DHLs) and double-expressing lymphomas (DELs), subtypes of DLBCLs that are characterized by aberrant expression of

63 MYC and BCL2 either due to gene rearrangements (as in the case of DHLs) or via other mechanisms, have been shown to be particularly aggressive and are associated with increased risk of relapse149, 150, 151, 169, 170. Although MYC is expressed at high levels in many DLBCLs, most lack amplifications or translocations of the MYC locus23, 108, suggesting that other mechanisms are required to achieve MYC over-expression in ABC-DLBCLs. Our findings suggest that ROCK2 activation contributes to increased MYC protein levels in ABC-DLBCLs by blocking its proteasome-dependent degradation. Bioinformatics analysis also identified RELA as an upstream regulator of the

ROCK2-regulated geneset in ABC-DLBCL. RELA is a canonical NF-kB subunit that is activated downstream of aCD40 signaling in B cells and is required for GC-derived PB/PC formation following immunization163. In addition to its physiological roles in PB/PC formation, nuclear RELA has been detected across a variety of B cell malignancies and NF-kB activation is required for the survival of several lymphomas, including ABC-DLBCLs107, 108. The ROCKs have previously been shown to regulate RELA activity in several cell types and regulation of RELA may contribute to the ROCK2-dependent program in ABC- DLBCLs171, 172, 173.

The transcriptional profiles of ABC-DLBCLs following ROCK1 or ROCK2 knockdown revealed that the ROCKs regulate unique pathways that converge to promote biological features of ABC-DLBCL. ROCK2 regulates the activities of IRF4 and MYC, and promotes DNA repair pathways, pathways critical for the survival of ABC-DLBCLs. ROCK1, however, may preferentially target metabolic programs by inducing cholesterol biosynthetic pathways. Although cytoskeletal transcriptional programs were also found to be regulated by ROCK1 and

64 ROCK2, as previously described69, we found that the ROCKs may also act to reinforce their own activation in ABC-DLBCL by repressing the expression of several Rho-GAPs. In contrast to the inducible nature of ROCK2 activation, ROCK1 activity was not induced downstream of signals known to promote terminal B cell differentiation and was found at high levels independently of lymphoma subtype. Although the signals driving constitutive ROCK1 activation in B cell lymphomas remain unclear, it is possible that signaling downstream of the BCR and/or mechanosensing pathways may regulate ROCK1 activity.

The selective lethality of ABC-DLBCLs following pan-ROCK inhibition suggests that ROCK inhibition could be a potential therapeutic strategy for the treatment of ABC-DLBCLs. Interestingly, we found that the transcriptional program affected following ROCK2 silencing in ABC-DLBCL cell lines was also significantly enriched in primary ABC-DLBCL cases compared to GCB-DLBCL cases, suggesting that ROCK inhibition could have beneficial effects in primary ABC-DLBCL patients. In addition to the transcriptional effects that we observed, survival downstream of ROCK signaling in ABC-DLBCLs could also be promoted by effects on cytoskeletal dynamics. Indeed, ERM phosphorylation has been suggested to promote DLBCL survival due to its ability to promote cell surface BCR organization and subsequent signaling174. In addition to Y-27632- induced lethality, our findings also yield insights into the potential use of ROCK2 selective inhibitors in combination with other chemotherapeutic agents. RNA- seq analysis revealed that ROCK2 represses antigen processing and presentation in ABC-DLBCLs. Loss of MHC-II expression correlates with poor clinical outcome in patients with DLBCLs following CHOP or R-CHOP therapies, possibly due to decreased tumor-infiltrating T cells and thus, a loss in immune

65 surveillance175. Our findings suggest that selective ROCK2 inhibition could be used in combination with current therapeutics to enhance efficacy.

Our findings outline novel roles for ROCK2 in the regulation of IRF4 and MYC activity in ABC-DLBCLs. In addition to the implications in lymphomagenesis, these findings may carry broader significance across multiple disease settings. Indeed, several of the dysregulated pathways seen in ABC-DLBCLs have also been observed in autoimmune patients including hyperactive NF-kB and TLR signaling, as well as deregulations in IRF activity. Additionally, ROCK hyperactivation has been observed in several murine models and in human patients with autoimmunity. Dysregulation in ROCK2 activity may thus contribute to the pathophysiology of several diseases marked by hyperactive B cell responses.

Methods Cell Culture

o Cell lines were grown in media to log phase at 37 C, 5% CO2 as follows: Burkitt’s lymphoma lines (BLs; Ramos, BL-41, BL-2), GC B cell-like diffuse large B cell lymphoma lines (GCB-DLBCLs; BJAB, DB, HT), and Activated B cell like

DLBCL lines (ABC-DLBCLs; OCI-LY3, U2932, RIVA, HBL-1, SU-DHL-2) in Iscove’s DMEM (Corning) + 10% FBS (Atlanta Biologics) + Pen/Strep (Corning).

For aCD40 and/or IL-21 stimulations, Ramos cells were plated at 1 x106 cells/mL and stimulated for various timepoints with 1µg/mL aCD40 (BioLegend) and/or 100ng/mL IL-21 (Peprotech). For Y-27632 (EMD Millipore) treatment experiments, cells were plated at 0.5-1.0 x106 cells/mL in the indicated media for 3-24hr before collection. For MG-132 experiments, cells were pre-treated

66 with 5µM MG-132 for 1hr prior to treatment with Y-27632 or KD025. HEK-293T cells were grown in DMEM (Corning) + 10% FBS (Atlanta Biologics) + Pen/Strep (Corning). 293T cells were transfected with various IRF4 constructs using either standard calcium phosphate procedures or using the TransIT-293 Transfection Reagent (Mirus) according to manufacturer’s protocol.

Constructs, Drugs, and Recombinant Proteins Expression plasmids for MYC- and FLAG-tagged human IRF4 and its phoshphodead mutants were generated as previously described62. The compound Y-27632 was obtained from EMD Millipore. Human recombinant IL-

21 was obtained from Peprotech. Human purified aCD40 was obtained from BioLegend. NP-conjugated chicken gamma-globulin (CGG) was obtained from Biosearch Technologies. MG-132 was obtained from Sigma.

MTS Proliferation Assays Cells were plated in triplicate at a density of 5,000-10,000 cells per well in 96- well plates. Cell viability after indicated treatments was assayed by adding 3- (4,5-dimethylthiazol-2-yl)-5-(3-carbpoxymethoxyphenol)-2-(4-sulphophenyl)- 2H tetrazolium and an electron coupling reagent (Promega), incubated for 3hr and measure by the amount of 490nm absorbance using a 96-well plate reader. The background was subtracted using media only controls.

Lentiviral Infection and Generation of Stable Cell Lines Lentiviral infections were performed using a 3rd generation packaging system. MISSION shRNA constructs targeting ROCK1 and ROCK2 were cloned into pLKO.1 vectors containing puromycin resistance cassettes and either tagRFP

67 (for ROCK1 shRNA) or tGFP (for ROCK2 and scrambled shRNAs) cassettes and were obtained from Sigma. Lentiviral particles were produced by transient transfection of 293T with 10µg of vector DNA along with the 5µg of the packaging construct psPAX2 (#12260; Addgene) and 5µg of the VSV-G envelope construct pMD2.G (#12259; Addgene) using standard calcium phosphate precipitation procedures. Virus-containing supernatants were collected at 48-72hr post-transfection and passed through a 0.45mm filter. For infection of cell lines, 1.0 x105 cells were resuspended in 6mL virus-containing supernatant with 10ng/mL polybrene, incubated at 37oC for 30min, and then centrifuged at 2,200 rpm for 2.5hr at 32oC, as previously described176. Virus- containing media was washed out from the cells 6hr after centrifugation and cells were cultured in IDMEM. Infected cells were sorted 24-72hr after infection on the basis of GFP- (for ROCK2 KD and scrambled shRNA controls) or RFP- (for ROCK1 KD) expression on a FACSAriaII. Sorted cells were further selected and maintained in IDMEM containing 2µg/mL puromycin. The sequences for the shRNAs were as follows: Scr: non-target shRNA ROCK1: TRCN0000195202 5’-CCG GCG ATC GTC TCT AGG ATG ATA TCT CGA GAT ATC ATC CTA GAG ACG ATC GTT TTT TG-3’

ROCK2: TRCN0000194836 5’-CCG GCC TTG ATG TCT GTC TAT TAT TCT CGA GAA TAA TAG ACA GAC ATC AAG GTT TTT TG-3’

RNA Extraction and RT-qPCR Total RNA was isolated from cells using the RNeasy Plus Mini kit (Qiagen). cDNAs were prepared using the iScript cDNA synthesis kit (BioRad). Real-time PCR was performed using the iTaq Universal SYBR Green Supermix (BioRad).

68 Gene expression was calculated using the delta-delta-Ct method and normalized to 28S. The following custom human primers were used: 28S: For 5’-GGA CAA GCC GGT GAC CTA C-3’ Rev 5’-CAT GAA CTT GCG AAG GAA AAC AT-3’ BCL6: For 5’- CGC AAC TCT GAA GAG CCA CCT GCG-3’ Rev 5’-TTT GTG ACG GAA ATG CAG GTT A-3’ BHLHE40: For 5’-AGC AGT GGT TCT TGA ACT TAC C-3’ Rev 5’-ACA AGC TGC GAA GAC TTC AGG-3’ CIITA: For 5’-CTA CCT GGA GCC TCT TAA CAG CGA T-3’ Rev 5’-TGG AGA AAG GCA TTG GAA TCT GG-3’ CMYC: For 5’-TTC GGG TAG TGG AAA ACC AG-3’ Rev 5’-CAG CAG CTC GAA TTT CTT CC-3’ ELL2: For 5’-CAT CAC CGT ACT GCA TGT GAA-3’ Rev 5’-ACT GGA TTG AAG GTC GAA AAG G-3’ IL10: For 5’-TCA AGG CGC ATG TGA ACT CC-3’ Rev 5’- GAT GTC AAA CTC ACT CAT GGC T-3’ IRF4: For 5’-GCT GAT CGA CCA GAT CGA CAG-3’ Rev 5’-CGG TTG TAG TCC TGC TTG C-3’ PAX5: For 5’-ACT TGC TCA TCA AGG TGT CAG-3’

Rev 5’-TCC TCC AAT TAC CCC AGG CTT-3’ PRDM1: For 5’-ACC AAG GAA TCT GCT TTT CAA GTA TG-3’ Rev 5’-CAT CAC TCC AAT AAC CTC TTC ACT GT-3’ ROCK1: For 5’-GGT GGT CGG TTG GGG TAT TTT-3’ Rev 5’-CGC CCT AAC CTC ACT TCC C-3’ ROCK2: For 5’-TCA GAG GTC TAC AGA TGA AGG C-3’ Rev 5’-CCA GGG GCT ATT GGC AAA GG-3’

69 XBP1: For 5’-CCT GGT TGC TGA AGA GGA GG-3’ Rev 5’-CCA TGG GGA GAT GTT CTG GAG-3’ ZEB2: For 5’- CAA CAG GCG CAA ACA AGC C-3’ Rev 5’-GGT TGG CAA TAC CGT CAT CC-3’

RNA-seq Analysis Total RNA was isolated using RNeasy Plus Mini kits (Qiagen). Preparation of a Nextera library was used to prepare Illumina-compatible sequencing libraries. Quality of all RNA and library preparations was evaluated with BioAnalyzer 2100 (Agilent). Libraries were paired-end sequenced by the Weill Cornell Epigenomics Core using HiSeq2500 at a depth of ~30-50 million fragments per sample. Sequencing performance was evaluated using FASTQC. 50-bp paired reads were mapped to a human reference genome (hg19 from UCSC) using STAR (Version 2.4.2) aligner. Aligned reads were quantified against the reference annotation to obtain FPKM (fragments per kilobase per million) and raw counts using Cufflinks (Version 2.2.1) and HTSeq, respectively. In order to classify the samples based on gene expression profiles, unsupervised clustering, namely Hierarchical data clustering and principal component analysis were performed. Both methods were performed on the log2 transformed FPKM expression values in R statistical software. Differential expression to compare expression profiles of respective groups was performed on normalized raw counts using the limma package in R. Genes with false discovery rate (FDR), q <0.05 were considered to be significantly differentially expressed.

70 Pathway analysis using GSEA (gene set enrichment analysis) software177 from the Broad Institute was used to identify functions of differentially expressed genes. Genes were ranked by the t-statistic value obtained from comparisons and the pre-ranked version of the tool was used to identify significantly enriched biological pathways. Pathways enriched with FDR q <0.20 were considered to be significant. Additionally, gene-name based pathway analysis was carried out using the online webtool ConcensusPathDB (CPDB)178.

Cell Extracts, Western Blotting, and Immunoprecipitations Whole cell extracts were prepared using lysis buffer (1% NP40, 10% glycerol, 50mM Tris pH 8.0, 150mM NaCl, 5mM NaF, 5mM EDTA, 1mM NaPP, 1mM Na3VO4, PhosSTOP (Sigma), cOmplete Protease Inhibitor (Sigma)). Nuclear and cytoplasmic extracts were prepared using the NE-PER Nuclear and Cytoplasmic Extraction Reagent kit (ThermoScientific). The purity of nuclear and cytoplasmic fractions was verified by probing with antibodies against

LaminB1 (D4Q4Z; Cell Signaling) or HDAC1 (#2062; Cell Signaling), and b- tubulin (D66; Sigma). Antibodies against BLIMP1 (6D3), CMYC (9E10), IRF4 (M-17), IRF8 (C-19), ROCK1 (H-85), and ROCK2 (H-85) were obtained from Santa Cruz Biotechnology. Antibodies against pSTAT3 (Y705, #9131) were obstained from Cell Signaling. Antibodies against STAT3 (84) were obtained from BD Biosciences. Rabbit polyclonal antibody specific for pIRF4 (recognizing phosphorylated S446/S447 residues) was generated by 21st Century Biochemicals Inc., as previously described62. For protein-protein interaction studies, cell extracts were immunoprecipitated with an anti-FLAG agarose affinity matrix (Sigma) according to manufacturer’s instructions. The immunoprecipitates were resolved by 10% SDS-PAGE and analyzed by

71 immunoblot with anti-MYC (Santa Cruz) and anti-FLAG (Sigma) antibodies. Western blot images were prepared using Adobe Photoshop and quantifications were calculated using ImageJ software.

Flow Cytometry Monoclonal antibodies to human HLA-DR (L243) were obtained from BioLegend. Surface staining was done on ice for 30min after blocking in 5% BSA in PBS. For intracellular staining, cells were fixed and permeabilized with the Foxp3 Staining Buffer Set (eBioscience) at 4oC for 40min, according to the manufacturer’s instructions. The cells were blocked with 5% FBS in PBS for 15min at room temperature and stained antibodies against IRF4 (200x; 3E4; eBioscience) or activated caspase-3 (C92-605; BD) for 30min on ice. For assessment of the dead sub-G0 population, cells were fixed using the Foxp3

Staining Buffer Set (eBioscience) at 4oC for 40min, and incubated with 50µg/mL propidium iodide (PI) in the presence of 100µg/mL RNaseA (Sigma). For pERM staining, cells were fixed and permeabilized with Cytofix/ Cytoperm buffer (BD) and washed twice with Perm/Wash buffer (BD). The cells were stained with an antibody against pERM (250x; 48G2; Cell Signaling) for 40min at room temperature. Primary antibody was detected with either anti-rabbit Alexa-fluor

488 (Invitrogen) or anti-rabbit R-phycoerythrin (Invitrogen). All flow cytometry data was acquired on a FACS Canto (BD) and analyzed with FlowJo (TreeStar) software.

Chromatin Immunoprecipitation (ChIP) Assays Stable Ramos ROCK KD lines were stimulated for 6hr and 20x106 cells were harvested and cross-linked with formaldehyde per condition. Chromatin extracts

72 were prepared using the truChIP Chromatin Shearing Reagent Kit (Covaris) according to manufacturer’s instructions. 100µg of the DNA-protein complexes was used for immunoprecipitions with anti-IRF4 (M-17; Santa Cruz) or normal goat Ig control (Santa Cruz) antibodies. After cross-linking was reversed and proteins were digested, the DNA was purified from the immunoprecipitates as well as from input extracts and then was analyzed by qPCR using the following primers: BCL6 Promoter: For 5’-TTTTTCTCGTGGTGCCTAAT-3’ Rev 5’-TGGGCTCCTCCTCTGTGACG-3’ ELL2 Promoter: For 5’-GTGCCTGGCACTAGTGGGAGCTCAAC-3’ Rev 5’-AAGGACACCGAGACACAGATTCACATA G-3’ IL10 Enhancer: For 5’-CCTCGGCAGGAGCGAGCGAACATGA-3’ Rev 5’-CTTGTGCTGACCTGGTGGATTTGGAAAG-3’ PRDM1 Intron 4: For 5’-CTGTTGCTGAGTGGGAGAGT-3’ Rev 5’-CTCCCTAAGATGCTCTACAGGTG-3’ PRDM1 Promoter: For 5’-GGACAGAGGCTGAGTTTGAAGA-3’ Rev 5’-CGCCATCAGCACCAGAATC-3’

Oligonucleotide Precipitation (ONP) Assays

ONP assays were conducted as previously described62. In brief, nuclear extracts were precleared with streptavidin-agarose beads and then incubated with biotinylated double-stranded oligonucleotides corresponding to the amplified IRF4-bound ChIP regions. Proteins bound to the biotin-labeled DNA were collected by streptavidin-agarose beads, separated by 10% SDS-PAGE and analyzed by immunobot with anti-IRF4 (D9P5H; Cell Signaling) antibodies. Sequences of oligonculeotides were as follows:

73 ELL2 Promoter: 5’-TAA AGT GCC TGG CAC TAG TGG GAG CTC AAC AGA TGT CAA TTC TTT TCC CCA CTT TTA GCT GAA TAA CTA TGT GAA TCT GTG TCT CGG TGT CCT T-3’ IL10 Enhancer: 5’-CCT TCT GTT CCC CTC GGC AGG AGC GAG CGA ACA TGA GTG GTT CTG ATA CAT ATT TTC CTT CCT TTC CAA ATC CAC CAG GTC AGC ACA AG-3’

In Vitro Kinase Assays ROCK kinase assays were performed as previously described62, 158. In brief, ROCK1 or ROCK2 was immunoprecipitated from nuclear extracts of cell lines using an anti-ROCK1 (C-19) or anti-ROCK2 (C-20) antibody (Santa Cruz Biotechnology). The immunoprecipitated ROCK1 or ROCK2 was incubated with purified recombinant MYPT1 substrate in kinase buffer containing ATP according to the manufacturer’s instructions (Cell Biolabs). Levels of phosphorylated recombinant MYPT1 was detected by immunoblotting using an anti-MYPT1 antibody.

Active ARHGEF1 Pull-down RhoA-G17A-conjugated agarose beads (STA-431; Cell Biolabs) were used to pull-down active ARHGEF1 from whole cell extracts prepared from cell lines, following the manufacturer’s instructions. Precipitated active ARHGEF1 was detected by immunoblotting using an anti-ARHGEF1 antibody (H-165; Santa Cruz).

Mice, Immunizations, and Cell Sorting

74 Blimp1-YFP reporter mice were obtained from S. Kaech (Yale University, New Haven, CT) and have been described previously157, 179. For immunization experiments, 8-10 week old mice were immunized with 100µg of NP30-40-CGG (Biosearch Technologies) precipitated in alum. Single-cell suspensions were pooled from spleens of immunized mice and were enriched for B cells and plasmablasts by positive selection using biotinylated B220 (RA3-6B2; BD) and CD138 (281-2; BD) antibodies and streptavidin microbeads (Miltenyi Biotech). Plasmablasts (Blimp1-YFP+ CD138+) and follicular B cells (B220+ Blimp1-YFP- CD23+) were sorted on a FACSAriaII. All the mice used in the experiments were housed in a specific pathogen-free animal facility at Weill Cornell Medical College. Experiments were performed according to the protocols approved by the Institutional Animal Care and Use Committee of Weill Cornell Medicine.

Xenograft Models The xenograft tumor models of human ABC-DLBCL and GCB-DLBCL were established by subcutaneous (s.c.) injection of 10-20x106 U2932 or HT cells into the flank of male and female NOD/SCID gamma (NSG) mice. The tumor growth was monitored by measuring size in two orthogonal dimensions. Tumor volume was calculated by using the formula ½ (log dimension) x (short dimension)2.

Once the average tumor volume reached 80mm3, therapy was started. The tumor-bearing mice were divided into 2 groups, with comparable tumor volume among groups. The mice were treated intraperitoneally everyday with vehicle alone (PBS) or Y-27632 (40mg/kg). At d12-15 after initiation of therapy, the mice were euthanized and the tumors were collected. Experiments were performed according to the protocols approved by the Institutional Animal Care and Use Committee of Weill Cornell Medicine.

75

Tissue Microarray (TMA) Immunohistochemistry Immunohistochemistry (IHC) was applied to a tissue microarray (TMA) encompassing 109 DLBCL tissues using a rabbit monoclonal antibody against pERM (16,000x; 48G2; Cell Signaling). Slides were evaluated for the percentage and degree of positive tumor cells.

76 CHAPTER III ROLE OF ROCK2 IN PRIMARY B CELL RESPONSES

Although extensively investigated in the non-hematopoietic compartment, the contribution of the RhoA-ROCK pathway to the development and differentiation of immune subsets is yet to be fully elucidated64. Initial insights have been obtained primarily through perturbations in RhoA activity with less attention being paid to the roles of specific RhoA effector proteins, including the ROCKs64. A non-redundant role for RhoA in B cell development has been described whereby RhoA deletion resulted in a marked reduction of mature peripheral B cell populations (including transitional, follicular, and marginal zone B cells) in the spleen92. The effect of RhoA deletion on B cell development was partially attributed to defective survival in response to BAFF92. RhoA has also been shown to modulate B cell cytoskeletal dynamics, as studies have shown that activated RhoA can inhibit intracellular BCR trafficking in response to TLR ligands93. The in vitro migration of normal and malignant B cells is also regulated by the RhoA pathway and recent studies have uncovered a role for one of the Rho-GEFs, ARHGEF1, in the retention of B cells within the GC96, 97, 98. The extent by which the ROCKs mediate many of the RhoA-dependent effects on B cell cytoskeletal dynamics, however, remains unknown.

In the previous chapter, we examined the contribution of ROCK1 and ROCK2 to the pathogenesis of ABC-DLBCLs and found that the two ROCKs can promote B cell lymphomagenesis by distinct as well as complementary mechanisms. In this chapter, I have investigated the hypothesis that the ROCKs play major roles, not only in B cell pathophysiology, but also in physiological

77 humoral responses. In particular, I focused my studies on the role of ROCK2 given our initial evidence that the activity of ROCK2, unlike that of ROCK1, can be regulated by B cell differentiation signals.

Results ROCK2 Activity is Increased Upon B Cell Differentiation Since ROCK2 activity could be induced in Ramos cells upon stimulation with aCD40 and IL-21 (Figure 8D), we first explored whether these signals could also control the activation of ROCK2 in primary B cells. For these studies, we purified CD23+ follicular B cells from C57BL/6 mice and stimulated them with aIgM, aCD40, and IL-21 for 3 days180, 181. In this culture system, the expression of key regulators of the GC program mimics what is observed during the early stages of GC entry with upregulation of Bcl6 and Aicda and downregulation of Irf4 (Figure 19A) 180, 181. In vitro kinase assays (IVKAs) were then performed to assess ROCK1- and ROCK2-specific activity in response to these signals. ROCK1 was highly activated in purified B cells at baseline and ROCK1 activity was unaffected by subsequent stimulation with combinations of aIgM, aCD40, and IL-21 (Figure 19B). In contrast, ROCK2 activity was low at baseline (Figure 19C). ROCK2 activation, could, however, be observed upon stimulation with aCD40 (Figure 19C). Thus, unlike ROCK1, ROCK2 activity is induced upon stimulation of primary B cells.

The in vitro studies were complemented by an in vivo assessment of ROCK activation in B cells following immunization of Blimp1-YFP reporter mice with a T cell-dependent (TD) antigen, NP-CGG (Figure 19D). ERM phosphorylation (pERM) was used as a surrogate marker for ROCK activity since treatment with

78

Figure 19. ROCK2 Activity is Increased Upon B Cell Differentiation. (A-C) CD23+ cells were purified from C57BL/6 mice and cultured with combinations of 5µg/mL aIgM, 5µg/mL aCD40, and/or 50ng/mL IL-21 for 3 days. (A) RT-qPCR analysis of Bcl6, Prdm1, Irf4, and Aicda expression from cultured B cells. qPCR data representative of at least 3 independent experiments. Error bars show SD, P-value by one-way ANOVA. (C-D) ROCK1 and ROCK2 kinase activity was examined by incubating immunoprecipitated (C) ROCK1 or (D) ROCK2 from extracts of cultured B cells with purified recombinant MYPT1 (rMPYT1) as substrate. Phosphorylated rMPYT1 (pMYPT1) was detected using an antibody against pMYPT1. Total ROCK1 or ROCK2 input levels for each sample are shown in the lower panels. Western blots representative of 3 independent experiments (left) and pooled quantifications are calculated as the densitometry ratio between phosphorylated rMYPT1 to total ROCK1 or ROCK2 input and normalized to the aIgM + aCD40 condition per experiment (left). Error bars show SEM, P-value by unpaired two-tailed t test. (D) Representative FACS hisogram (left) and pooled quantification (right) of pERM expression gated on total B cells (B220+), germinal center B cells (GCBs; B220+ Fas+ GL7+), and plasmablasts/plasma cells (PB/PCs; CD138+ YFP+) from male and female ROCK2f/f.WT.Blimp1-YFP reporter mice immunized with 100µg NP-CGG i.p. for 7 days. Quantification shows mean fluorescent intensity of pERM expression in the indicated populations from 3 independent experiments (n = 9). Error bars show SEM, P-value by ordinary one-way ANOVA. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

79 Y-27632, a pan-ROCK inhibitor, was able to significantly decrease pERM levels in B cell lines (Figure 18A-B). Consistent with the high levels of IRF4 phosphorylation in PB/PCs (Figure 7C), PB/PCs exhibited the highest levels of pERM following immunization (Figure 19D). Interestingly, GC B cells also exhibited significantly higher levels of pERM than did total B cells (Figure 19D), suggesting that, upon immunization, ROCK activation can be observed in both GC B cells and PB/PCs.

Absence of ROCK2 Leads to Impaired GC Responses The ability of ROCK2 to be activated in B cells in response to T-cell dependent signals such as CD40 engagement raised the possibility that ROCK2 could be required during the response of B cells to T-dependent antigens. To directly evaluate this possibility, we crossed ROCK2f/f mice with mice expressing the

Cre recombinase downstream from the Cg1 promoter (Cg1-Cre). The Cg1- Cre.ROCK2f/f mice allow for the selective deletion of Rock2 in GC B cells, which occurs at 3-5 days post immunization with a TD antigen51, 182. ROCK2f/f (WT) mice served as a control for these experiments. Deletion of Rock2 in Cg1- Cre.ROCK2f/f mice was first confirmed by in vitro cultures whereby purified B cells were induced to class switch to IgG1 upon stimulation with either LPS and

IL-4 stimulation or aCD40 and IL-21 (Figure 20A-B). Expression of ROCK1 was comparable in WT and Cg1-Cre.ROCK2f/f B cell cultures, indicating that ROCK1 expression is not upregulated to compensate for the lack of ROCK2 in these mice (Figure 20A-B). Development of bone marrow and splenic B cell populations was normal in Cg1-Cre.ROCK2f/f mice compared to WT mice (Figure 20C-F). Baseline levels of serum immunoglobulins were also similar

80

Figure 20. B Cell Development is Normal in Cg1-Cre.ROCK2 Mice. (A-B) RT- qPCR analysis of Rock1 and Rock2 from purified CD43- purified B cells from control ROCK2f/f (WT) or Cg1-Cre.ROCK2f/f (cKO) mice stimulated in vitro for 3d with either (A) 25µg/mL LPS + 20ng/mL IL-4 or (B) 5µg/mL aIgM, 5mg/mL aCD40, and 50ng/mL IL-21. Data representative of 2 independent experiments. Eror bars show SD, P-value by unpaired two-tailed t test. (C-D) Representative FACS plots and pooled quantifications of frequencies (top) and cell numbers (bottom) of Pro-B cells (Pro-B; B220+ CD43+ IgM-), Pre-B cells (Pre-B; B220+ CD43- IgM-), and Immature B cells (Imm B; B220+ CD43- IgM+) from the bone marrow and of Transitional 1 B cells (T1; B220+ CD23- CD21- IgM+), Transitional 2 B cells (B220+ CD23+ CD21hi IgM+), follicular B cells (FoB; B220+ CD23+ CD21lo IgMmid), and marginal zone B cells (MZB; B220+ CD23- CD21hi IgM+) from the spleens of 6-10wk old control ROCK2f/f (WT; closed circle) or Cg1- Cre.ROCK2f/f (cKO; open circle) mice. (E-F) Representative FACS plots and pooled quantifications of plasma cells (PB/PCs; B220lo CD138+) from the bone marrow (left) and spleen (right) of WT and cKO mice, as in (C-D). Quantification shows frequencies and cell numbers of respective populations from at least 3 independent experiments (n >= 5 per genotype). Error bars show SEM, P-value by unpaired two-tailed t test. (G) Pooled ELISA data of IgM, IgG1, IgG2c, IgG3, IgA, and IgE protein levels in serum from 6-10wk old WT or cKO mice. Data pooled from at least 3 independent experiments (n >= 8 per genotype). Error bars show SEM; P-value by unpaired two-tailed t test. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

81

82 between WT and Cg1-Cre.ROCK2f/f mice except for slight reductions in IgG2c levels (Figure 20G).

To examine the role of B cell ROCK2 during an immune response, we next immunized WT and Cg1-Cre.ROCK2f/f mice with the TD-antigen, NP-CGG. As compared to WT mice, the frequencies and numbers of GC B cells (B220+ GL7+ Fas+) at day 7 and day 10 after immunization were greatly reduced in the spleens of Cg1-Cre.ROCK2f/f mice (Figure 21A-B). We also assessed antigen- specific B cell responses after NP-CGG immunization by monitoring the binding of a fluorescently labeled NP protein to various B cell subpopulations, as previously described5. Although not detected at day 4 after immunization, a significant population of WT B cells bound NP by day 7 (Figure 21C-D). The frequency and total number of NP-specific B cells in the spleens of immunized

Cg1-Cre.ROCK2f/f mice were greatly reduced compared to WT mice, an effect that was also seen at day 10 after immunization (Figure 21C-D). In accordance with the decreases in total NP-specific B cells, there was also a drastic reduction in NP-specific GC B cells (NP+ CD38lo) at days 7 and 10 post-immunization in

Cg1-Cre.ROCK2f/f mice compared to WT mice (Figure 21C and E). In addition to the NP-specific GC B cells, we also identified a previously described population of emerging memory B cells (NP+ CD38hi) in WT mice at day 75 (Figure 21C). The frequency and total numbers of NP-specific emerging memory B cells were also lower in Cg1-Cre.ROCK2f/f mice compared to WT mice at days 7 and 10 post-immunization (Figure 21F). This, however, may be a consequence of the decrease in total NP-specific B cells, since the frequency of CD38hi B cells among NP-specific B cells was comparable between the Cg1-

83

Figure 21. Cg1-Cre.ROCK2 Mice Have Decreased GC Responses following TD Immunization. Male and female control ROCK2f/f (WT; closed circle) and Cg1-Cre.ROCK2f/f (cKO; open circle) mice were immunized with 100µg NP- CGG i.p. for 4-10 days. (A-B) Representative FACS plots and pooled quantifications of total GC B cells (B220+ Fas+ GL7+) from the spleens of indicated mice. (C-G) Representative FACS plots and pooled quantifications of (D) NP-specific B cells (B220+IgM-IgD-Gr-1-IgG1+NP+), (E) NP-specific GC B cells (B220+IgM-IgD-Gr-1-IgG1+NP+CD38lo), and (F-G) NP-specific memory B cells (B220+IgM-IgD-Gr-1-IgG1+NP+CD38hi) from the spleens of indicated mice. Quantifications pooled from at least 2-3 independent experiments (n>=5 per genotype). Error bars show SEM, P-value by unpaired two-tailed t test. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

84 Cre.ROCK2f/f mice and WT mice (Figure 21G). Thus, B cell specific ROCK2 is required for optimal GC responses.

Absence of ROCK2 Results in Impaired GC Dynamics and Functions Formation of a mature GC by day 7 after immunization is characterized by the establishment of GC polarity1, 7. Indeed, mature GCs exhibit both a centroblast- rich dark zone and a centrocyte-rich light zone and exhibit a centroblast to centrocyte ratio of ~2:118. To examine whether the failure of Cg1-Cre.ROCK2f/f mice to possess a GC B cell population at day 7 after immunization was accompanied by a disruption in GC polarity, we examined the frequencies of centroblast and centrocyte populations following immunization. Centrocytes and centroblasts can be distinguished by surface expression of the activation markers CD86 or CD83 and by the chemokine receptor CXCR4, respectively18. Consistent with the reduced numbers of GC B cells following immunization in

Cg1-Cre.ROCK2f/f mice, we observed decreased numbers of both CXCR4hiCD86lo centroblasts (CBs) and CXCR4loCD86hi centrocytes (CCs) in

Cg1-Cre.ROCK2f/f mice compared to WT mice at day 7 after immunization (Figure 22A-B). Despite decreases in the numbers of both GC subpopulations, the effect on the centroblast population was more pronounced and Cg1- Cre.ROCK2f/f mice showed reductions in the frequency of centroblasts within the GC (Figure 22C). This decrease in centroblast frequency corresponded with an increase in the frequency of centrocytes among GC B cells (Figure 22C), resulting in the inability of the GC B cells in Cg1-Cre.ROCK2f/f mice to reach a centroblast to centrocyte ratio of ~2:1. FoxO1 was recently identified as a regulator of the centroblast transcriptional program in vivo183, 184. No differences in total FoxO1 expression were detected by IC FACS in GC B cells from Cg1-

85 Cre.ROCK2f/f mice as compared to WT mice after immunization (Figure 22D). To assess whether the failure to properly expand the centroblast subpopulation in the Cg1-Cre.ROCK2f/f mice was due to defects in proliferation, we monitored the expression of Ki-67 in GC B cells following immunization (Figure 22E). We observed reductions in the expression of Ki-67 within GC B cells from Cg1- Cre.ROCK2ff mice as compared to WT mice 7 days after immunization (Figure 22E). Consistent with decreased GC B cell proliferation, we also observed decreases in the frequency of switched B cells within the GCs of Cg1- Cre.ROCK2f/f mice at days 7 and 10 after immunization (Figure 22F-G). These findings suggest that ROCK2 promotes the proper expansion of GC B cell subpopulations after immunization.

Given the decreased formation of GC B cells in immunized Cg1-Cre.ROCK2f/f mice, we next assessed whether ROCK2 regulates the expression of Bcl6, the master regulator of the GC B cell program. Bcl6 protein levels were significantly reduced in GC B cells from Cg1-Cre.ROCK2f/f mice compared to WT mice at day 7 after immunization (Figure 22H-I). Interestingly, Bcl6 protein levels at day

4 after immunization were normal in Cg1-Cre.ROCK2f/f mice (Figure 22H-I), suggesting that ROCK2 may play a role in the maintenance of Bcl6 expression.

Taken together, these data highlight important roles for ROCK2 in mediating GC B cell responses.

B Cell-specific ROCK2 is Required for Optimal Humoral Responses After TD Immunization

In addition to the disrupted GC B cell development in immunized Cg1- Cre.ROCK2f/f mice, we also observed decreased frequencies and numbers of

86

Figure 22. Cg1-Cre.ROCK2 Mice Have Impaired GC Responses following TD Immunization. Male and female control ROCK2f/f (WT; closed circle) and Cg1-Cre.ROCK2f/f mice (cKO; open circle) mice were immunized with 100µg NP-CGG for 4-10 days. (A) Representative FACS plots of centroblasts (CBs; CXCR4hi CD86lo) and centrocytes (CCs; CXCR4lo CD86hi) gated on GC B cells (B220+ Fas+ GL7+) from the indicated mice at day 7 post-immunization. (B) Pooled quantifications of CB and CC cell numbers from indicated mice at day 7 post-immunization. (C) Pooled quantifications showing frequencies of CBs (left) and CCs (right) of GC B cells from indicated mice at day 7 post-immunization. (D) Representative FACS histogram (left) and quantification (right) showing FoxO1 expression in GC B cells (B220+ Fas+ GL7+) from the indicated mice 7 days post-immunization. (E) Representative FACS plot (left) and pooled quantification (right) of Ki-67 expression in GC B cells (B220+ Fas+ GL7+) from indicated mice at day 7 post-immunization. (F-G) Representative FACS plots and pooled quantification of IgG1 and IgM expression gated on GC B cells (B220+ GL7+ Fas+) from spleens of mice 7-10 days after immunization. (H-I) Representative FACS histograms and pooled quantifications of BCL6 expression in WT (blue) or cKO (red) mice 4-7 days after immunization. Data representative of at least 2 (day 4; n>=6 per genotype), 3 (day 7; n>=8 per genotype), or 2 independent experiments (day 10; n>=4 per genotype). Error bars show SEM, P-value by unpaired two-tailed t test. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

87

88 PB/PCs in the spleens of Cg1-Cre.ROCK2f/f mice compared to WT mice at day 7 (Figure 23A-B). This finding was accompanied by decreased serum IgG1 levels at later time points after immunization, while serum levels of IgM were normal in the immunized Cg1-Cre.ROCK2f/f mice (Figure 23C). We also assessed the levels of antigen-specific IgG1 antibodies at day 10 after immunization by detecting NP30-reactive IgG1 (whereby “30” indicates 30 moieties of NP conjugated to BSA and represents detection of both low and high affinity antibodies) and observed a great reduction in NP-specific IgG1 antibodies in Cg1-Cre.ROCK2f/f mice following immunization (Figure 23D). As a measure of affinity maturation, we assessed serum titers of NP7-reactive IgG1

(whereby “7” indicates 7 moieties of NP conjugated to BSA) and calculated the ratio of NP7-reactive IgG1 to NP30-reactive IgG1 (Figure 23D-E). We found that the levels of high-affinity antigen-specific IgG1 were lower in Cg1-Cre.ROCK2f/f mice compared to WT mice (Figure 23D) and furthermore, that the ratio of

f/f NP7/30-reactive IgG1 was also significantly reduced in the Cg1-Cre.ROCK2 mice (Figure 23E), suggesting that ROCK2 is required for normal affinity maturation. Together, these findings show that humoral immune responses to

TD antigens are impaired in Cg1-Cre.ROCK2f/f mice.

TFH Differentiation is Normal in Mice with B Cell-specific ROCK2 Deletion To assess whether the impaired GC responses in Cg1-Cre.ROCK2f/f mice were only due to alterations in the B cell compartment or were also accompanied by alterations in the development of follicular T helper (TFH) cells, we evaluated the spleens of immunized mice for the presence of TFH cells. Normal frequencies and numbers of TFH cells (as marked by co-expression of CXCR5 and PD1 on CD4+ T cells) were observed in Cg1-Cre.ROCK2f/f mice as compared to WT

89

Figure 23. Cg1-Cre.ROCK2 Mice Exhibit Defective Humoral Responses Following TD Immunization. Male and female control ROCK2f/f (WT; closed circles) and Cg1-Cre.ROCK2f/f (cKO; open circles) mice were immunized with 100µg NP-CGG i.p. for 7-10 days. (A-B) Representative FACS plots and pooled quantifications of plasmablasts (PBs; B220lo CD138+) from the spleens of indicated mice. Quantification shows frequencies and cell numbers of PBs from 3 independent experiments (n>=10 per genotype). Error bars show SEM, P-value by unpaired two-tailed t test. (C) Pooled ELISA data of IgM and IgG1 protein levels in serum from WT and cKO mice at day 10 post-immunization. Data pooled from at least 2 independent experiments (n>=4 per genotype). Error bars SEM, P-value by unpaired two-tailed t test. (D-E) Representative ELISAs of NP-specific IgG1 in serum collected from immunized WT and cKO mice at day 10, presented as OD450 readings of serial dilutions for NP7 (molar ratio of NP to BSA, 7), NP30 (molar ratio of NP to BSA, 30), and (E) the NP7 to NP30 ratio. Data representative of 2 independent experiments. Error bars SD, P-value by unpaired two-tailed t test. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

90 mice (Figure 24A-B). Recently, a novel subset of GC-localized T cells known as

T follicular regulatory (TFR) cells was also identified. TFR cells are defined as

+ + FoxP3-expressing TFH-like cells (CXCR5 PD1 ) and were shown to negatively regulate GC B cell responses through both direct and indirect mechanisms. It has been proposed that the ratio of TFR to TFH cells is an important factor in humoral immunity and offers correlative value in predicting the magnitude of

185, 186 antibody responses . No changes in the frequencies of TFR cells in Cg1- Cre.ROCK2f/f mice as compared to WT mice could be observed following immunization (Figure 24C). Thus, TFH cell differentiation occurs normally in Cg1- Cre.ROCK2f/f mice after TD immunization and the impaired GC responses are not due to aberrant expansion of TFR cells.

Given that B cell-specific ROCK2 appears to be dispensable for TFH cell differentiation following immunization with a TD antigen, we next asked whether T cell-specific ROCK2 play a role in mediating TD humoral responses. For these studies, we crossed the ROCK2f/f mice with mice expressing the Cre recombinase downstream of the CD4 promoter (CD4-Cre). We immunized CD4-Cre.ROCK2f/f mice or control ROCK2f/f (WT) mice with NP-CGG and assessed the ability of these mice to form TFH cells. No differences in TFH cell frequencies or numbers between CD4-Cre.ROCK2f/f mice and WT mice at day 7 after immunization were detected (Figure 25A-B). Total and antigen-specific GC B cell responses were also normal in CD4-Cre.ROCK2f/f mice compared to WT mice after immunization (Figure 25C-G). Furthermore, CD4-Cre.ROCK2f/f mice exhibited normal levels of total and NP-specific immunoglobulins at 28 days after immunization and affinity maturation, as monitored by the ratio of

NP7-reactive IgG1 to NP30-reactive IgG1, was normal in these mice (Figure

91

f/f Figure 24. TFH Differentiation is Normal in Cg1-Cre.ROCK2 Mice Following TD Immunization. Male and female control ROCK2f/f (WT; closed circles) and Cg1-Cre.ROCK2f/f (cKO; open circles) mice were immunized with 100µg NP-CGG i.p. for 7-10 days. (A-B) Representative FACS plots and pooled + + + quantifications of follicular T helper cells (TFH; PD1 CXCR5 ) gated on CD4 cells from the spleens of indicated mice. (C) Quantifications of regulatory + + + + follicular T cells (TFR; FoxP3 ) gated on PD1 CXCR5 CD4 cells from the spleens of indicated mice at 7-10 days post-immunization. Quantifications pooled from at least 2 independent experiments (n>=5 per genotype). Error bars show SEM, P-value by unpaired two-tailed t test. Data representative of single experiment (n>=4 per genotype). * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

92

Figure 25. CD4-Cre.ROCK2f/f Mice Exhibit Normal Responses to TD Immunization. Male and female control ROCK2f/f (WT; closed circle) and CD4- Cre.ROCK2f/f mice (CD4-cKO; open circle) mice were immunized with 100µg NP-CGG for 7 or 28 days. (A-B) Representative FACS plots (left) and pooled + + + quantifications (right) of (A) TFH cell populations (CXCR5 PD1 gated on CD4 cells) and (B) GC B cell populations (GL7+Fas+ gated on B220+ cells) from spleens of the indicated mice at day 7 post-immunization. (C-D) Representative FACS plots and quantifications of NP-specific B cells (NP+IgG1+ gated on B220+IgM-IgD-Gr1- cells) and NP-specific GC B cells (NP+IgG1+CD38lo) from spleens of indicated mice at day 7 post-immunization. (E) ELISA analysis of total levels of IgM and IgG1 from the serum of indicated mice 28 days after immunization. (F) ELISA analysis of NP-specific IgG1 in serum collected from immunized WT and CD4-cKO mice, presented as OD450 readings of serial dilutions for NP30 (molar ratio of NP to BSA, 30) (left), NP7 (molar ratio of NP to BSA, 7) (center), and the NP7 to NP30 ratio (right). Error bars show SEM, P- value by unpaired two-tailed t test. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

93

94 25H-J). These data thus suggest that T cell specific ROCK2 is not required for efficient humoral responses upon immunization with a TD antigen.

Discussion Effective humoral responses require B cells to rapidly respond to diverse and dynamic stimuli that shape B cell fate decisions. The precise mechanisms driving B cell differentiation in response to these signals, however, remain incompletely understood. Here, we demonstrated a requirement for ROCK2 in B cell differentiation following immunization with a protein antigen. We found that B cell-specific ROCK2 deletion profoundly affected primary humoral responses following TD immunization, reducing splenic PB formation and GC persistence and function. These studies thus uncovered previously unknown roles for ROCK2 in mediating B cell differentiation and suggest that targeting ROCK2 activity may have broad applications in several fields including vaccine development and autoimmunity.

Although low-affinity B cell responses can occur in the absence of T cell help, high-affinity humoral responses require a dynamic interplay between T and B cells1, 2. T-B interactions begin immediately after antigen encounter, promote the initiation of GCs and continue throughout the GC response, resulting in the production of high-affinity antibody-secreting PCs1, 2. These interactions promote B cell activation and differentiation through the engagement of CD40 by CD40L and through the presence of other T cell-derived signals, including IL-213, 4, 5. We found that ROCK2 activation was induced in primary B cells downstream of these T cell-derived signals and detected high phosphorylation levels of the ERM proteins, classical ROCK substrates, in GC B cells and in

95 PB/PCs following immunization. Interestingly, we did not observe an induction in ROCK1 activity downstream of T cell-derived signals, suggesting that, of the two ROCK isoforms, ROCK2 is a key regulator of B cell differentiation in the setting of T-B collaborations.

The requirement of productive T-B interactions for high-affinity antigen-specific immune responses has been illustrated through multiple studies whereby disruption of CD40/CD40L signaling through either pharmacological or genetic approaches prevented the formation of GCs and PB/PCs upon TD immunization187, 188. Additionally, initiation of CD40 blockade following GC formation also resulted in the premature termination of the GC response189. Consistent with ROCK2 being a potential mediator of T-B interactions, we found that mice lacking ROCK2 in their B cell compartment also failed to maintain GC B cell responses following TD immunization. The decreased proliferation of GC B cells in the absence of ROCK2 likely contributed to this reduction in GC B cells. During the maturation of the GC reaction, GC B cell subpopulations emerge and polarity is established within the GC1, 7. The mature GC consists of ~ 2:1 ratio of centroblasts (GC B cells residing in the dark zone) to centrocytes (those residing in the light zone) 1, 7, 18. These polarized regions of the GC may act to segregate GC processes, whereby somatic hypermutation and proliferation occur in the dark zone, while selection occurs in the light zone1, 7. Models of cyclic re-entry postulate that productive T-B interactions in the light zone lead to the interzonal migration of centrocytes back into the dark zone for further rounds of proliferation and hypermutation1, 7. ROCK2-deficient mice exhibited a diminished centroblast population, raising the possibility that ROCK2 is important in cyclic re-entry and/or in the establishment of GC polarity.

96

The GC B cell program is regulated by a diverse network of transcription factors. Notably, BCL6 is absolutely required for the formation and maintenance of GC B cells and has been attributed as the master regulator of the GC B cell program1, 7. We found that GC B cells exhibited decreased levels of BCL6 protein in the absence of ROCK2. Expression of BCL6 in the GC is maintained by several transcription factors, including IRF81, 7. IRF8 shares a high degree of similarity with IRF440 and, given the ability of ROCK2 to phosphorylate and regulate IRF4 activity, it is possible that ROCK2 may also target IRF8 and regulate its ability to promote BCL6 expression within the GC. In addition to promoting Bcl6 expression, IRF8 also facilitates antigen specific T-B interactions through several mechanisms, including the upregulation of MHC-II surface expression, and promotes affinity maturation190, raising the possibility that IRF8 may mediate ROCK2-dependent processes in GC B cells. An additional role for ROCK2 in the maintenance of the GC B cell response may arise from its ability to regulate MYC. In the previous chapter, we outlined a role for ROCK2 in promoting the stability of MYC protein in ABC-DLBCLs. Critical roles for MYC in GC B cell responses have been described, whereby B cell- specific deletion of MYC results in the collapse of the established GC. In the context of GC B cells, ROCK2 may promote MYC protein expression downstream of T cell-derived signals, resulting in re-entry into the dark zone and continued maintenance of the GC.

In line with the impaired GC responses observed in the absence of ROCK2, the ROCK2-deficient mice also exhibited reduced frequencies of splenic PB/PCs and had lower serum titers of total and antigen-specific antibodies. This

97 decrease in PB/PC formation was observed at early time points after immunization and likely consisted of both GC-derived and extrafollicular PB/PC responses. The potential for ROCK2 to regulate the formation of GCs and the formation of extrafollicular PB/PCs following immunization suggests dynamic roles for ROCK2 in mediating B cell differentiation downstream of T cell-derived signals.

Recent studies using the selective ROCK2 inhibitor, KD025, have suggested that ROCK2 plays important roles in the differentiation of TFH cells in the context of SLE and graft-versus-host disease91, 136. Since the transcriptional networks regulating TFH and GC B cell programs share similarities, including a dependency on BCL6 and IRF41, 7, 191, we asked whether T cell-specific ROCK2 might play complementary roles with B cell-specific ROCK2 in promoting physiological GC responses. In contrast to the KD025 studies performed in autoimmune settings91, 136, we found that mice with T cell-specific deletion of

ROCK2 exhibited normal TFH differentiation and humoral responses following TD immunization. Given the ability of ROCK1 and ROCK2 to target similar substrates in vitro69, 70, 71, ROCK1 activity might compensate for the loss of ROCK2 in T cells under physiological conditions. In autoimmune settings, however, ROCK1 might not be able to compensate for the augmented levels of ROCK2 activity132. Additionally, ROCK2 might play a more prominent role in the differentiation of pathogenic TFH cells, rather than in the differentiation of physiological TFH cells following transient immune challenge. Indeed, pathogenic TFH cells found in systemic autoimmune models often have unique

101 features compared to physiological TFH cells, including the production of IFNg .

98 Our findings identify novel roles for ROCK2 in B cell differentiation following immunization with a TD antigen. In addition to implications in physiological immune responses, these findings may carry broader implications. Indeed, as the prevalence of systemic autoimmunity in the western world continues to grow, novel therapeutics that can dampen the activation and differentiation of autoreactive B cells are greatly needed. Hyperactivation of ROCK has been observed in several murine models and in primary patients with autoimmunity132, suggesting that targeting this pathway may have clinical benefit. These findings illustrate that modulation of ROCK2 activity may thus have broad implications in vaccine biology as well as in autoimmunity.

Methods Mice and Immunizations

C57BL/6 mice, Cg1-Cre, and CD4-Cre were obtained from Jackson Laboratory. Blimp1-YFP reporter mice were obtained from S. Kaech (Yale University, New Haven, CT) and have been described previously157, 179. ROCK2f/f mice were obtained from James K. Liao (University of Chicago, Chicago, IL) and have been described previously192. For immunization experiments, 8-10 week old mice were immunized with 100µg NP-CGG (Biosearch Technologies) precipitated in alum. All the mice used in the experiments were housed in a specific pathogen- free animal facility at Weill Cornell Medical College. Experiments were performed according to the protocols approved by the Institutional Animal Care and Use Committee of Weill Cornell Medicine.

Drugs and Recombinant Proteins

99 Murine recombinant IL-4 and IL-21 were obtained from Peprotech. Murine purfieid aCD40 was obtained from BioLegend. Murine aIgM was obtained from Jackson ImmunoResearch. LPS was obtained from Sigma. NP-conjugated chicken gamma globulin (CGG) was obtained from Biosearch Technologies.

B Cell Cultures Single-cell suspensions from pooled spleens were enriched for follicular B cells with biotinylated anti-CD23 (BD Bioscience) and streptavidin microbeads (Miltenyi Biotec) following the manufacturer’s instructions. CD23+ B cells were cultured in RPMI 1640 medium (Corning) supplemented with 10% FBS (Atlanta Biologicals), 100 U/mL penicillin (Corning), 100 mg/mL streptomycin (Corning), 1X Non-Essential Amino Acids (Corning), 2mM L-glutamine (Corning), 25mM

HEPES (pH 7.2-7.6) and 50µM b-mercaptoethanol and were stimulated with 5

µg/mL F(ab’)2 anti-mouse IgM (Jackson ImmunoResearch), 5µg/mL Ultra-LEAF purified anti-mouse CD40 (BioLegend), and/or 50 ng/mL IL-21 (Peprotech). For proliferation assays, CD23+ B cells were labeled with 2.5µM Cell Trace Violet (Invitrogen) for 2min at 25oC prior to stimulation. For IgG1-skewing cultures, B cells are enriched by negative selection with CD43-conjugated microbeads (Miltenyi Biotec) following the manufacturer’s instructions, and stimulated with

25 µg/mL LPS (Sigma) and 20ng/mL IL-4 (Peprotech) for 3 days.

Flow Cytometry Monoclonal antibodies to the following mouse proteins were obtained from BioLegend and used for multi-parameter flow cytometry: B220 (400x; RA3- 6B2), CD21 (200x; 7E9), CD23 (200x; B3B4), CD38 (600x; 90), CD4 (400x; RM4-5), CD86 (200x; GL-1), Gr-1 (400x; RB6-8C5), IgG1 (1000x; RMG1-1),

100 MHC-II I-Ab (1000x; AF6-120.1) and PD-1 (200x; 29F.1A12). Antibodies to murine BCL6 (50x; K112-91), CD138 (600x; 281-2), CXCR5 (200x; 2G8), Fas (200x), GL7 (600x), and IgD (500x; 11-26c.2a) were obtained from BD. Antibodies to murine CD43 (200x; R2/60), CXCR4 (200x; 2B11), FoxP3 (100x; FJK-16s), IgM (1500x; II/4E), and Ki-67 (200x; SolA15) were obtained from eBioscience. Antibodies to murine FoxO1 (250x; C29H4) were obtained from Cell Signaling Technologies. Antibodies to murine Ephrin-B1 (250x; BAF473) were obtained from R&D Systems. PE-conjugated NP (200x) was obtained from Biosearch Technologies. For CXCR4 and CXCR5 staining, cells were incubated in the dark at room temperature for 25min. For detection of nuclear proteins, cells were fixed after surface staining at 4oC with the Foxp3 Staining Buffer Set (eBioscience) according to manufacturer’s instruction. Data was acquired on FACS Canto (Becton Dickinson) and analyze with FlowJo (TreeStar) software.

For pERM staining, splenocytes were fixed and permeabilized after surface staining with Cytofix/ Cytoperm buffer (BD) and washed twice with Perm/Wash Buffer (BD). The cells were stained with an antibody against pERM (250x; 48G2; Cell Signaling) for 40min at room temperature. Primary antibody was detected with either anti-rabbit Alexa-fluor 488 (Invitrogen) or anti-rabbit R-phycoerythrin

(Invitrogen).

RNA Extraction and RT-qPCR Total RNA was isolated from cells using the RNeasy Plus Mini kit (Qiagen). cDNAs were prepared using the iScript cDNA synthesis kit (BioRad). Real-Time PCR was performed using the iTaq Universal SYBR Green Supermix (BioRad). Gene expression was calculated using the delta-delta-Ct method and

101 normalized to Ppia (Cyclophilin A). Primers for Bcl6, Prdm1, Rock1, and Rock2 were obtained from Qiagen. The following custom primers were used: Aicda: For 5’-GCC ACC TTC GCA ACA AGT CTC-3’ Rev 5’-CCG GGC ACA GTC ATA GCA C-3’ Irf4: For 5’-TCC TCT GGA TGG CTC CAG ATG G-3’ Rev 5’-CAC CAA AGC ACA GAG TCA CCT G-3’ Ppia: For 5’-TTG CCA TTC CTG GAC CAA A-3’ Rev 5’-ATG GCA CTG GCG GCA GGT CC-3’

Cell Extracts and Western Blotting Whole cell extracts were prepared using lysis buffer (1% NP40, 10% glycerol, 50mM Tris pH 8.0, 150mM NaCl, 5mM NaF, 5mM EDTA, 1mM NaPP, 1mM

Na3VO4, PhosSTOP (Sigma), cOmplete Protease Inhibitor (Sigma)). Antibodies against ROCK1 (H-85) and ROCK2 (H-85) were obtained from Santa Cruz Biotechnology.

In vitro Kinase Assays ROCK kinase assays were performed as previously described62, 158. In brief, ROCK1 or ROCK2 was immunoprecipitated from whole cell extracts of cultured murine B cells using an anti-ROCK1 (C-19) or anti-ROCK2 (C-20) antibody (Santa Cruz Biotechnology). The immunoprecipitated ROCK1 or ROCK2 was incubated with purified recombinant MYPT1 substrate in kinase buffer containing ATP according to the manufacturer’s instructions (Cell Biolabs). Levels of phosphorylated recombinant MYPT1 were detected by immunoblotting using an anti-pMYPT1 (T696) antibody.

102 ELISA Total serum immunoglobulins were detected by ELISA (Southern Biotech) according to manufacturer’s protocol. In brief, 96-well plates were coated with

10µg/mL goat anti-mouse Ig, blocked with 1% BSA in PBS, and incubated with various dilutions of serum. Total immunoglobulin levels were detected with isotype-specific AP-conjugated antibodies and developed with PNPP substrate. For detection of NP-specific antibodies, 96-well plates were coated with either

o NP7-BSA or NP30-BSA (Biosearch Technologies) overnight at 4 C, followed by blocking with 5% BSA. Serum samples were loaded into plates with eight serial dilutions (from 1:800 to 1:13,107,200) and incubated for 2hr at 25oC. NP titers were detected with goat anti-mouse IgG1-HRP and visualized with TMB substrate (BioLegend).

103 CHAPTER IV CONCLUSIONS AND FUTURE DIRECTIONS

Our findings outline novel roles for the Rho Kinases in physiological and pathophysiological B cell differentiation. We find that ROCK2 activity is induced downstream of T cell-derived signals that normally promote B cell activation and differentiation, including aCD40 and IL-21, and that ROCK2 plays important roles in modulating B cell differentiation. In primary models, effects of ROCK2 deficiency are observed early following TD immunization, as mice with B cell- specific deletion of Rock2 fail to form and maintain optimal GC responses. These defective GC responses correspond with decreased Bcl6 protein expression in the residual GC B cells and are characterized by defective class switching and a skewed centroblast to centrocyte ratio. Additionally, our lymphoma studies showed that downstream of T cell-derived signals, ROCK2 also promotes the phosphorylation of IRF4, enhancing the ability of IRF4 to regulate distinct genes involved in terminal B cell differentiation, including PRDM1 and ELL2. These data highlight important roles for ROCK2 at multiple stages in B cell differentiation downstream of T-B interactions (Figure 26).

Although normally regulated by T cell-derived signals, we found that ROCK2 activation is constitutive in ABC-DLBCLs and suggest that deregulated ROCK signaling plays roles in shaping dynamic biological processes in ABC-DLBCLs (Figure 26). We found that a subset of primary ABC-DLBCL tissues exhibited elevated levels of phosphorylated ROCK substrates and that our identified ROCK2-regulated gene signature was enriched in primary ABC-DLBCL patients. We furthermore showed that ROCK2 contributed to the maintenance

104 of the ABC-DLBCL phenotype through multiple mechanisms, including the direct phosphorylation of IRF4 and through the promotion of MYC protein levels. We also found that ROCK1 activation was constitutive in ABC-DLBCLs and that each ROCK isoform had a unique contribution to the transcriptional profile of ABC-DLBCLs. Interestingly, inhibition of both ROCK kinases selectively resulted in lethality of the ABC-DLBCLs, but not other GC-derived malignancies, suggesting that ROCK inhibition could have therapeutic potential in ABC- DLBCL patients.

Future studies related to this dissertation project include: 1) further detailing the molecular mechanisms by which ROCK2 is required for GC formation and maintenance, 2) determining the functional implications of ROCK1 activity in physiological B cell differentiation, and 3) exploring whether ROCK1 and ROCK2 play pathogenic roles in B cells in other disease settings, such as systemic autoimmunity. These follow-up studies will be essential to further elucidate the roles of the Rho Kinases in B cell biology and will be required to build a more comprehensive understanding of the potential that therapeutic ROCK inhibition might carry in the clinic. Collectively, our findings delineate the pathophysiological implications of ROCK activation in ABC-DBCLs and other B cell malignancies, reveal an important role for ROCK2 in modulating physiological B cell responses upon antigen challenge, and propose that ROCK inhibition could represent a novel therapeutic approach for the treatment of diseases characterized by dysfunctional B cell responses.

105

Figure 26. Model Showing the Roles of ROCK2 in Physiological and Pathophysiological B Cell Differentiation. ROCK2 is activated downstream of T cell-derived signals, including aCD40 and IL-21, in B cells, possibly through the activation of ARHGEF1 (left). Upon activation, ROCK2 can phosphorylate IRF4 and enhance its ability to regulate several IRF4 target genes involved in terminal B cell differentiation, including PRDM1, ELL2, and IL10. At earlier stages in B cell differentiation, following TD immunization, ROCK2 is required for the promotion of GC B cell responses. ROCK2-deficient GC B cells exhibit decreased protein levels of Bcl6, show impaired class switching, and have a skewed centroblast (CB) to centrocyte (CC) ratio. In ABC-DLBCLs, ROCK2 is constitutively activated (right) and acts to phosphorylate IRF4 and promote MYC protein stability. ROCK1 is also constitutively activated in ABC-DLBCLs and promotes the expression of genes involved in cholesterol biosynthesis. Together, ROCK1 and ROCK2 regulate a gene signature that converge on DNA repair and promote the survival of ABC-DLBCL.

106 REFERENCES

1. De Silva, N.S. & Klein, U. Dynamics of B cells in germinal centres. Nat Rev Immunol 15, 137-148 (2015).

2. Nutt, S.L., Hodgkin, P.D., Tarlinton, D.M. & Corcoran, L.M. The generation of antibody-secreting plasma cells. Nat Rev Immunol 15, 160- 171 (2015).

3. McHeyzer-Williams, L.J., Pelletier, N., Mark, L., Fazilleau, N. & McHeyzer-Williams, M.G. Follicular helper T cells as cognate regulators of B cell immunity. Curr Opin Immunol 21, 266-273 (2009).

4. Linterman, M.A. et al. IL-21 acts directly on B cells to regulate Bcl-6 expression and germinal center responses. J Exp Med 207, 353-363 (2010).

5. Zotos, D. et al. IL-21 regulates germinal center B cell differentiation and proliferation through a B cell-intrinsic mechanism. J Exp Med 207, 365- 378 (2010).

6. Kurosaki, T., Kometani, K. & Ise, W. Memory B cells. Nat Rev Immunol 15, 149-159 (2015).

7. Victora, G.D. & Nussenzweig, M.C. Germinal centers. Annu Rev Immunol 30, 429-457 (2012).

8. Okada, T. et al. Antigen-engaged B cells undergo chemotaxis toward the T zone and form motile conjugates with helper T cells. PLoS Biol 3, e150 (2005).

9. Qi, H., Cannons, J.L., Klauschen, F., Schwartzberg, P.L. & Germain, R.N. SAP-controlled T-B cell interactions underlie germinal centre formation. Nature 455, 764-769 (2008).

10. Kerfoot, S.M. et al. Germinal center B cell and T follicular helper cell development initiates in the interfollicular zone. Immunity 34, 947-960 (2011).

11. Kitano, M. et al. Bcl6 protein expression shapes pre-germinal center B cell dynamics and follicular helper T cell heterogeneity. Immunity 34, 961-972 (2011).

107 12. Gatto, D., Paus, D., Basten, A., Mackay, C.R. & Brink, R. Guidance of B cells by the orphan G protein-coupled receptor EBI2 shapes humoral immune responses. Immunity 31, 259-269 (2009).

13. Pereira, J.P., Kelly, L.M., Xu, Y. & Cyster, J.G. EBI2 mediates B cell segregation between the outer and centre follicle. Nature 460, 1122-1126 (2009).

14. MacLennan, I.C. Germinal centers. Annu Rev Immunol 12, 117-139 (1994).

15. Mesin, L., Ersching, J. & Victora, G.D. Germinal Center B Cell Dynamics. Immunity 45, 471-482 (2016).

16. Schwickert, T.A. et al. In vivo imaging of germinal centres reveals a dynamic open structure. Nature 446, 83-87 (2007).

17. Hauser, A.E. et al. Definition of germinal-center B cell migration in vivo reveals predominant intrazonal circulation patterns. Immunity 26, 655- 667 (2007).

18. Victora, G.D. et al. Germinal center dynamics revealed by multiphoton microscopy with a photoactivatable fluorescent reporter. Cell 143, 592- 605 (2010).

19. Gitlin, A.D., Shulman, Z. & Nussenzweig, M.C. Clonal selection in the germinal centre by regulated proliferation and hypermutation. Nature 509, 637-640 (2014).

20. Liu, D. et al. T-B-cell entanglement and ICOSL-driven feed-forward regulation of germinal centre reaction. Nature 517, 214-218 (2015).

21. Huang, C. et al. The BCL6 RD2 domain governs commitment of activated B cells to form germinal centers. Cell Rep 8, 1497-1508 (2014).

22. Basso, K. et al. Integrated biochemical and computational approach identifies BCL6 direct target genes controlling multiple pathways in normal germinal center B cells. Blood 115, 975-984 (2010).

23. Ci, W. et al. The BCL6 transcriptional program features repression of multiple oncogenes in primary B cells and is deregulated in DLBCL. Blood 113, 5536-5548 (2009).

108 24. Huang, C., Geng, H., Boss, I., Wang, L. & Melnick, A. Cooperative transcriptional repression by BCL6 and BACH2 in germinal center B-cell differentiation. Blood 123, 1012-1020 (2014).

25. Tunyaplin, C. et al. Direct repression of prdm1 by Bcl-6 inhibits plasmacytic differentiation. J Immunol 173, 1158-1165 (2004).

26. Shaffer, A.L. et al. BCL-6 represses genes that function in lymphocyte differentiation, inflammation, and cell cycle control. Immunity 13, 199-212 (2000).

27. Hatzi, K. et al. BCL6 orchestrates Tfh cell differentiation via multiple distinct mechanisms. J Exp Med 212, 539-553 (2015).

28. Johnston, R.J. et al. Bcl6 and Blimp-1 are reciprocal and antagonistic regulators of T follicular helper cell differentiation. Science 325, 1006- 1010 (2009).

29. Nurieva, R.I. et al. Bcl6 mediates the development of T follicular helper cells. Science 325, 1001-1005 (2009).

30. Yu, D. et al. The transcriptional repressor Bcl-6 directs T follicular helper cell lineage commitment. Immunity 31, 457-468 (2009).

31. Dominguez-Sola, D. et al. The proto-oncogene MYC is required for selection in the germinal center and cyclic reentry. Nat Immunol 13, 1083-1091 (2012).

32. Calado, D.P. et al. The cell-cycle regulator c-Myc is essential for the formation and maintenance of germinal centers. Nat Immunol 13, 1092- 1100 (2012).

33. Luo, W., Weisel, F. & Shlomchik, M.J. B Cell Receptor and CD40 Signaling Are Rewired for Synergistic Induction of the c-Myc Transcription Factor in Germinal Center B Cells. Immunity 48, 313-326 e315 (2018).

34. Ying, C.Y. et al. MEF2B mutations lead to deregulated expression of the oncogene BCL6 in diffuse large B cell lymphoma. Nat Immunol 14, 1084- 1092 (2013).

35. Lee, C.H. et al. Regulation of the germinal center gene program by interferon (IFN) regulatory factor 8/IFN consensus sequence-binding protein. J Exp Med 203, 63-72 (2006).

109 36. Shaffer, A.L. et al. Blimp-1 orchestrates plasma cell differentiation by extinguishing the mature B cell gene expression program. Immunity 17, 51-62 (2002).

37. Minnich, M. et al. Multifunctional role of the transcription factor Blimp-1 in coordinating plasma cell differentiation. Nat Immunol 17, 331-343 (2016).

38. Tellier, J. et al. Blimp-1 controls plasma cell function through the regulation of immunoglobulin secretion and the unfolded protein response. Nat Immunol 17, 323-330 (2016).

39. De Silva, N.S., Simonetti, G., Heise, N. & Klein, U. The diverse roles of IRF4 in late germinal center B-cell differentiation. Immunol Rev 247, 73- 92 (2012).

40. Rogatsky, I., Chandrasekaran, U., Manni, M., Yi, W. & Pernis, A.B. Epigenetics and the IRFs: a complex interplay in the control of immunity and autoimmunity. Autoimmunity 47, 242-255 (2014).

41. Kwon, H. et al. Analysis of interleukin-21-induced Prdm1 gene regulation reveals functional cooperation of STAT3 and IRF4 transcription factors. Immunity 31, 941-952 (2009).

42. Staudt, V. et al. Interferon-regulatory factor 4 is essential for the developmental program of T helper 9 cells. Immunity 33, 192-202 (2010).

43. Bollig, N. et al. Transcription factor IRF4 determines germinal center formation through follicular T-helper cell differentiation. Proc Natl Acad Sci U S A 109, 8664-8669 (2012).

44. Lohoff, M. et al. Dysregulated T helper cell differentiation in the absence of interferon regulatory factor 4. Proc Natl Acad Sci U S A 99, 11808- 11812 (2002).

45. Brustle, A. et al. The development of inflammatory T(H)-17 cells requires interferon-regulatory factor 4. Nat Immunol 8, 958-966 (2007).

46. Chen, Q. et al. IRF-4-binding protein inhibits interleukin-17 and interleukin-21 production by controlling the activity of IRF-4 transcription factor. Immunity 29, 899-911 (2008).

47. Huber, M. et al. IRF4 is essential for IL-21-mediated induction, amplification, and stabilization of the Th17 phenotype. Proc Natl Acad Sci U S A 105, 20846-20851 (2008).

110

48. Zheng, Y. et al. Regulatory T-cell suppressor program co-opts transcription factor IRF4 to control T(H)2 responses. Nature 458, 351- 356 (2009).

49. Lu, R., Medina, K.L., Lancki, D.W. & Singh, H. IRF-4,8 orchestrate the pre-B-to-B transition in lymphocyte development. Genes Dev 17, 1703- 1708 (2003).

50. Ma, S., Pathak, S., Trinh, L. & Lu, R. Interferon regulatory factors 4 and 8 induce the expression of Ikaros and Aiolos to down-regulate pre-B-cell receptor and promote cell-cycle withdrawal in pre-B-cell development. Blood 111, 1396-1403 (2008).

51. Klein, U. et al. Transcription factor IRF4 controls plasma cell differentiation and class-switch recombination. Nat Immunol 7, 773-782 (2006).

52. Sciammas, R. et al. Graded expression of interferon regulatory factor-4 coordinates isotype switching with plasma cell differentiation. Immunity 25, 225-236 (2006).

53. Shaffer, A.L., Emre, N.C., Romesser, P.B. & Staudt, L.M. IRF4: Immunity. Malignancy! Therapy? Clin Cancer Res 15, 2954-2961 (2009).

54. Saito, M. et al. A signaling pathway mediating downregulation of BCL6 in germinal center B cells is blocked by BCL6 gene alterations in B cell lymphoma. Cancer Cell 12, 280-292 (2007).

55. Ochiai, K. et al. Transcriptional regulation of germinal center B and plasma cell fates by dynamical control of IRF4. Immunity 38, 918-929 (2013).

56. Willis, S.N. et al. Transcription factor IRF4 regulates germinal center cell formation through a B cell-intrinsic mechanism. J Immunol 192, 3200- 3206 (2014).

57. Sciammas, R. et al. An incoherent regulatory network architecture that orchestrates B cell diversification in response to antigen signaling. Mol Syst Biol 7, 495 (2011).

58. Krishnamoorthy, V. et al. The IRF4 Gene Regulatory Module Functions as a Read-Write Integrator to Dynamically Coordinate T Helper Cell Fate. Immunity 47, 481-497 e487 (2017).

111 59. Glasmacher, E. et al. A genomic regulatory element that directs assembly and function of immune-specific AP-1-IRF complexes. Science 338, 975-980 (2012).

60. Li, P. et al. BATF-JUN is critical for IRF4-mediated transcription in T cells. Nature 490, 543-546 (2012).

61. Biswas, P.S. et al. Dual regulation of IRF4 function in T and B cells is required for the coordination of T-B cell interactions and the prevention of autoimmunity. J Exp Med 209, 581-596 (2012).

62. Biswas, P.S. et al. Phosphorylation of IRF4 by ROCK2 regulates IL-17 and IL-21 production and the development of autoimmunity in mice. J Clin Invest 120, 3280-3295 (2010).

63. Remesh, S.G., Santosh, V. & Escalante, C.R. Structural Studies of IRF4 Reveal a Flexible Autoinhibitory Region and a Compact Linker Domain. J Biol Chem 290, 27779-27790 (2015).

64. Ricker, E., Chowdhury, L., Yi, W. & Pernis, A.B. The RhoA-ROCK pathway in the regulation of T and B cell responses. F1000Res 5 (2016).

65. Thumkeo, D., Watanabe, S. & Narumiya, S. Physiological roles of Rho and Rho effectors in mammals. Eur J Cell Biol 92, 303-315 (2013).

66. Jaffe, A.B. & Hall, A. Rho GTPases: biochemistry and biology. Annu Rev Cell Dev Biol 21, 247-269 (2005).

67. Tybulewicz, V.L. & Henderson, R.B. Rho family GTPases and their regulators in lymphocytes. Nat Rev Immunol 9, 630-644 (2009).

68. Heasman, S.J. & Ridley, A.J. Multiple roles for RhoA during T cell transendothelial migration. Small GTPases 1, 174-179 (2010).

69. Amano, M., Nakayama, M. & Kaibuchi, K. Rho-kinase/ROCK: A key regulator of the cytoskeleton and cell polarity. Cytoskeleton (Hoboken) 67, 545-554 (2010).

70. Schofield, A.V. & Bernard, O. Rho-associated coiled-coil kinase (ROCK) signaling and disease. Crit Rev Biochem Mol Biol 48, 301-316 (2013).

71. Julian, L. & Olson, M.F. Rho-associated coiled-coil containing kinases (ROCK): structure, regulation, and functions. Small GTPases 5, e29846 (2014).

112 72. Zanin-Zhorov, A. et al. Selective oral ROCK2 inhibitor down-regulates IL- 21 and IL-17 secretion in human T cells via STAT3-dependent mechanism. Proc Natl Acad Sci U S A 111, 16814-16819 (2014).

73. Mueller, B.K., Mack, H. & Teusch, N. Rho kinase, a promising drug target for neurological disorders. Nat Rev Drug Discov 4, 387-398 (2005).

74. Pan, P. et al. Advances in the development of Rho-associated protein kinase (ROCK) inhibitors. Drug Discov Today 18, 1323-1333 (2013).

75. Defert, O. & Boland, S. Rho kinase inhibitors: a patent review (2014 - 2016). Expert Opin Ther Pat 27, 507-515 (2017).

76. Hahmann, C. & Schroeter, T. Rho-kinase inhibitors as therapeutics: from pan inhibition to isoform selectivity. Cell Mol Life Sci 67, 171-177 (2010).

77. Isgro, J. et al. Enhanced rho-associated protein kinase activation in patients with systemic lupus erythematosus. Arthritis Rheum 65, 1592- 1602 (2013).

78. Shi, J. & Wei, L. Rho kinases in cardiovascular physiology and pathophysiology: the effect of fasudil. J Cardiovasc Pharmacol 62, 341- 354 (2013).

79. Chen, M. et al. Fasudil and its analogs: a new powerful weapon in the long war against central nervous system disorders? Expert Opin Investig Drugs 22, 537-550 (2013).

80. Gurkar, A.U. et al. Identification of ROCK1 kinase as a critical regulator of Beclin1-mediated autophagy during metabolic stress. Nat Commun 4, 2189 (2013).

81. Koch, J.C. et al. ROCK2 is a major regulator of axonal degeneration, neuronal death and axonal regeneration in the CNS. Cell Death Dis 5, e1225 (2014).

82. Mleczak, A., Millar, S., Tooze, S.A., Olson, M.F. & Chan, E.Y. Regulation of autophagosome formation by Rho kinase. Cell Signal 25, 1-11 (2013).

83. Knipe, R.S., Tager, A.M. & Liao, J.K. The Rho kinases: critical mediators of multiple profibrotic processes and rational targets for new therapies for pulmonary fibrosis. Pharmacol Rev 67, 103-117 (2015).

84. Tanaka, T. et al. Nuclear Rho kinase, ROCK2, targets p300 acetyltransferase. J Biol Chem 281, 15320-15329 (2006).

113

85. Soriano, S.F. et al. In vivo analysis of uropod function during physiological T cell trafficking. J Immunol 187, 2356-2364 (2011).

86. Lou, Z., Billadeau, D.D., Savoy, D.N., Schoon, R.A. & Leibson, P.J. A role for a RhoA/ROCK/LIM-kinase pathway in the regulation of cytotoxic lymphocytes. J Immunol 167, 5749-5757 (2001).

87. Tharaux, P.L. et al. Rho kinase promotes alloimmune responses by regulating the proliferation and structure of T cells. J Immunol 171, 96- 105 (2003).

88. Yang, J.Q. et al. RhoA orchestrates glycolysis for TH2 cell differentiation and allergic airway inflammation. J Allergy Clin Immunol 137, 231-245 e234 (2016).

89. Zhu, M. et al. Role of Rho kinase isoforms in murine allergic airway responses. Eur Respir J 38, 841-850 (2011).

90. Rozo, C. et al. Targeting the RhoA-ROCK pathway to reverse T-cell dysfunction in SLE. Ann Rheum Dis 76, 740-747 (2017).

91. Weiss, J.M. et al. ROCK2 signaling is required to induce a subset of T follicular helper cells through opposing effects on STATs in autoimmune settings. Sci Signal 9, ra73 (2016).

92. Zhang, S., Zhou, X., Lang, R.A. & Guo, F. RhoA of the Rho family small GTPases is essential for B lymphocyte development. PLoS One 7, e33773 (2012).

93. Freeman, S.A. et al. Toll-like receptor ligands sensitize B-cell receptor signalling by reducing actin-dependent spatial confinement of the receptor. Nat Commun 6, 6168 (2015).

94. Natkanski, E. et al. B cells use mechanical energy to discriminate antigen affinities. Science 340, 1587-1590 (2013).

95. Saci, A. & Carpenter, C.L. RhoA GTPase regulates B cell receptor signaling. Mol Cell 17, 205-214 (2005).

96. Mele, S., Devereux, S. & Ridley, A.J. Rho and Rap guanosine triphosphatase signaling in B cells and chronic lymphocytic leukemia. Leuk Lymphoma 55, 1993-2001 (2014).

114 97. Cuesta-Mateos, C. et al. Analysis of migratory and prosurvival pathways induced by the homeostatic chemokines CCL19 and CCL21 in B-cell chronic lymphocytic leukemia. Exp Hematol 38, 756-764, 764 e751-754 (2010).

98. Muppidi, J.R. et al. Loss of signalling via Galpha13 in germinal centre B- cell-derived lymphoma. Nature 516, 254-258 (2014).

99. Liu, Z. & Davidson, A. Taming lupus-a new understanding of pathogenesis is leading to clinical advances. Nat Med 18, 871-882 (2012).

100. Crispin, J.C. & Tsokos, G.C. Interleukin-17-producing T cells in lupus. Curr Opin Rheumatol 22, 499-503 (2010).

101. Craft, J.E. Follicular helper T cells in immunity and systemic autoimmunity. Nat Rev Rheumatol 8, 337-347 (2012).

102. Woods, M., Zou, Y.R. & Davidson, A. Defects in Germinal Center Selection in SLE. Front Immunol 6, 425 (2015).

103. Bird, A.K., Meednu, N. & Anolik, J.H. New insights into B cell biology in systemic lupus erythematosus and Sjogren's syndrome. Curr Opin Rheumatol 27, 461-467 (2015).

104. Lech, M. et al. IRF4 deficiency abrogates lupus nephritis despite enhancing systemic cytokine production. J Am Soc Nephrol 22, 1443- 1452 (2011).

105. Stirzaker, R.A. et al. Administration of fasudil, a ROCK inhibitor, attenuates disease in lupus-prone NZB/W F1 female mice. Lupus 21, 656-661 (2012).

106. Yang, Y. et al. Exploiting synthetic lethality for the therapy of ABC diffuse large B cell lymphoma. Cancer Cell 21, 723-737 (2012).

107. Basso, K. & Dalla-Favera, R. Germinal centres and B cell lymphomagenesis. Nat Rev Immunol 15, 172-184 (2015).

108. Shaffer, A.L., 3rd, Young, R.M. & Staudt, L.M. Pathogenesis of human B cell lymphomas. Annu Rev Immunol 30, 565-610 (2012).

109. Victora, G.D. et al. Identification of human germinal center light and dark zone cells and their relationship to human B-cell lymphomas. Blood 120, 2240-2248 (2012).

115

110. Dave, S.S. et al. Molecular diagnosis of Burkitt's lymphoma. N Engl J Med 354, 2431-2442 (2006).

111. Schmitz, R., Ceribelli, M., Pittaluga, S., Wright, G. & Staudt, L.M. Oncogenic mechanisms in Burkitt lymphoma. Cold Spring Harb Perspect Med 4 (2014).

112. Schmitz, R. et al. Burkitt lymphoma pathogenesis and therapeutic targets from structural and functional genomics. Nature 490, 116-120 (2012).

113. Alizadeh, A.A. et al. Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature 403, 503-511 (2000).

114. Rosenwald, A. et al. Molecular diagnosis of primary mediastinal B cell lymphoma identifies a clinically favorable subgroup of diffuse large B cell lymphoma related to Hodgkin lymphoma. J Exp Med 198, 851-862 (2003).

115. Morin, R.D. et al. Somatic mutations altering EZH2 (Tyr641) in follicular and diffuse large B-cell lymphomas of germinal-center origin. Nat Genet 42, 181-185 (2010).

116. Lohr, J.G. et al. Discovery and prioritization of somatic mutations in diffuse large B-cell lymphoma (DLBCL) by whole-exome sequencing. Proc Natl Acad Sci U S A 109, 3879-3884 (2012).

117. Morin, R.D. et al. Frequent mutation of histone-modifying genes in non- Hodgkin lymphoma. Nature 476, 298-303 (2011).

118. Pasqualucci, L. et al. Analysis of the coding genome of diffuse large B- cell lymphoma. Nat Genet 43, 830-837 (2011).

119. Beguelin, W. et al. EZH2 is required for germinal center formation and somatic EZH2 mutations promote lymphoid transformation. Cancer Cell 23, 677-692 (2013).

120. Caganova, M. et al. Germinal center dysregulation by histone methyltransferase EZH2 promotes lymphomagenesis. J Clin Invest 123, 5009-5022 (2013).

121. Davis, R.E. et al. Chronic active B-cell-receptor signalling in diffuse large B-cell lymphoma. Nature 463, 88-92 (2010).

116 122. Lenz, G. et al. Oncogenic CARD11 mutations in human diffuse large B cell lymphoma. Science 319, 1676-1679 (2008).

123. Compagno, M. et al. Mutations of multiple genes cause deregulation of NF-kappaB in diffuse large B-cell lymphoma. Nature 459, 717-721 (2009).

124. Ngo, V.N. et al. Oncogenically active MYD88 mutations in human lymphoma. Nature 470, 115-119 (2011).

125. Mandelbaum, J. et al. BLIMP1 is a tumor suppressor gene frequently disrupted in activated B cell-like diffuse large B cell lymphoma. Cancer Cell 18, 568-579 (2010).

126. Pasqualucci, L. et al. Inactivation of the PRDM1/BLIMP1 gene in diffuse large B cell lymphoma. J Exp Med 203, 311-317 (2006).

127. Tam, W. et al. Mutational analysis of PRDM1 indicates a tumor- suppressor role in diffuse large B-cell lymphomas. Blood 107, 4090-4100 (2006).

128. Lenz, G. et al. Aberrant immunoglobulin class switch recombination and switch translocations in activated B cell-like diffuse large B cell lymphoma. J Exp Med 204, 633-643 (2007).

129. Lenz, G. et al. Molecular subtypes of diffuse large B-cell lymphoma arise by distinct genetic pathways. Proc Natl Acad Sci U S A 105, 13520-13525 (2008).

130. Zhou, Q., Gensch, C. & Liao, J.K. Rho-associated coiled-coil-forming kinases (ROCKs): potential targets for the treatment of atherosclerosis and vascular disease. Trends Pharmacol Sci 32, 167-173 (2011).

131. Komers, R. Rho kinase inhibition in diabetic kidney disease. Br J Clin Pharmacol 76, 551-559 (2013).

132. Pernis, A.B., Ricker, E., Weng, C.H., Rozo, C. & Yi, W. Rho Kinases in Autoimmune Diseases. Annu Rev Med 67, 355-374 (2016).

133. He, Y. et al. Antiinflammatory effect of Rho kinase blockade via inhibition of NF-kappaB activation in rheumatoid arthritis. Arthritis Rheum 58, 3366-3376 (2008).

117 134. Li, Y. et al. Phosphorylated ERM is responsible for increased T cell polarization, adhesion, and migration in patients with systemic lupus erythematosus. J Immunol 178, 1938-1947 (2007).

135. Zanin-Zhorov, A. et al. Cutting Edge: Selective Oral ROCK2 Inhibitor Reduces Clinical Scores in Patients with Psoriasis Vulgaris and Normalizes Skin Pathology via Concurrent Regulation of IL-17 and IL-10. J Immunol 198, 3809-3814 (2017).

136. Flynn, R. et al. Targeted Rho-associated kinase 2 inhibition suppresses murine and human chronic GVHD through a Stat3-dependent mechanism. Blood 127, 2144-2154 (2016).

137. Cools, J. RHOA mutations in peripheral T cell lymphoma. Nat Genet 46, 320-321 (2014).

138. Yoo, H.Y. et al. A recurrent inactivating mutation in RHOA GTPase in angioimmunoblastic T cell lymphoma. Nat Genet 46, 371-375 (2014).

139. Sakata-Yanagimoto, M. et al. Somatic RHOA mutation in angioimmunoblastic T cell lymphoma. Nat Genet 46, 171-175 (2014).

140. Palomero, T. et al. Recurrent mutations in epigenetic regulators, RHOA and FYN kinase in peripheral T cell lymphomas. Nat Genet 46, 166-170 (2014).

141. Chiba, S., Enami, T., Ogawa, S. & Sakata-Yanagimoto, M. G17V RHOA: Genetic evidence of GTP-unbound RHOA playing a role in tumorigenesis in T cells. Small GTPases 6, 100-103 (2015).

142. Choi, J. et al. Genomic landscape of cutaneous T cell lymphoma. Nat Genet 47, 1011-1019 (2015).

143. Nagata, Y. et al. Variegated RHOA mutations in adult T-cell leukemia/lymphoma. Blood 127, 596-604 (2016).

144. Kataoka, K. & Ogawa, S. Variegated RHOA mutations in human cancers. Exp Hematol 44, 1123-1129 (2016).

145. Rohde, M. et al. Recurrent RHOA mutations in pediatric Burkitt lymphoma treated according to the NHL-BFM protocols. Genes Chromosomes Cancer 53, 911-916 (2014).

146. O'Hayre, M. et al. Inactivating mutations in GNA13 and RHOA in Burkitt's lymphoma and diffuse large B-cell lymphoma: a tumor suppressor

118 function for the Galpha13/RhoA axis in B cells. Oncogene 35, 3771-3780 (2016).

147. Azab, A.K. et al. RhoA and Rac1 GTPases play major and differential roles in stromal cell-derived factor-1-induced cell adhesion and chemotaxis in multiple myeloma. Blood 114, 619-629 (2009).

148. Linke, F. et al. WNT5A: a motility-promoting factor in Hodgkin lymphoma. Oncogene 36, 13-23 (2017).

149. Nguyen, L., Papenhausen, P. & Shao, H. The Role of c-MYC in B-Cell Lymphomas: Diagnostic and Molecular Aspects. Genes (Basel) 8 (2017).

150. Johnson, N.A. et al. Concurrent expression of MYC and BCL2 in diffuse large B-cell lymphoma treated with rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone. J Clin Oncol 30, 3452-3459 (2012).

151. Green, T.M. et al. Immunohistochemical double-hit score is a strong predictor of outcome in patients with diffuse large B-cell lymphoma treated with rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone. J Clin Oncol 30, 3460-3467 (2012).

152. Shaffer, A.L. et al. IRF4 addiction in multiple myeloma. Nature 454, 226- 231 (2008).

153. Wei, L., Surma, M., Shi, S., Lambert-Cheatham, N. & Shi, J. Novel Insights into the Roles of Rho Kinase in Cancer. Arch Immunol Ther Exp (Warsz) 64, 259-278 (2016).

154. Lam, L.T. et al. Cooperative signaling through the signal transducer and activator of transcription 3 and nuclear factor-{kappa}B pathways in subtypes of diffuse large B-cell lymphoma. Blood 111, 3701-3713 (2008).

155. Basso, K. et al. Tracking CD40 signaling during germinal center development. Blood 104, 4088-4096 (2004).

156. Niu, H., Cattoretti, G. & Dalla-Favera, R. BCL6 controls the expression of the B7-1/CD80 costimulatory receptor in germinal center B cells. J Exp Med 198, 211-221 (2003).

157. Rutishauser, R.L. et al. Transcriptional repressor Blimp-1 promotes CD8(+) T cell terminal differentiation and represses the acquisition of central memory T cell properties. Immunity 31, 296-308 (2009).

119 158. Weng, C.H., Gupta, S., Geraghty, P., Foronjy, R. & Pernis, A.B. Cigarette smoke inhibits ROCK2 activation in T cells and modulates IL-22 production. Mol Immunol 71, 115-122 (2016).

159. Girkontaite, I. et al. Lsc is required for marginal zone B cells, regulation of lymphocyte motility and immune responses. Nat Immunol 2, 855-862 (2001).

160. Rubtsov, A. et al. Lsc regulates marginal-zone B cell migration and adhesion and is required for the IgM T-dependent antibody response. Immunity 23, 527-538 (2005).

161. Green, J.A. et al. The sphingosine 1-phosphate receptor S1P(2) maintains the homeostasis of germinal center B cells and promotes niche confinement. Nat Immunol 12, 672-680 (2011).

162. Jackson, S.W. et al. B cell IFN-gamma receptor signaling promotes autoimmune germinal centers via cell-intrinsic induction of BCL-6. J Exp Med 213, 733-750 (2016).

163. Heise, N. et al. Germinal center B cell maintenance and differentiation are controlled by distinct NF-kappaB transcription factor subunits. J Exp Med 211, 2103-2118 (2014).

164. Farrell, A.S. & Sears, R.C. MYC degradation. Cold Spring Harb Perspect Med 4 (2014).

165. Bustelo, X.R. A transcriptional cross-talk between RhoA and c-Myc inhibits the RhoA/Rock-dependent cytoskeleton. Small GTPases 1, 69- 74 (2010).

166. Sears, R. et al. Multiple Ras-dependent phosphorylation pathways regulate Myc protein stability. Genes Dev 14, 2501-2514 (2000).

167. Perdomo-Sabogal, A. et al. Human Lineage-Specific Transcriptional Regulation through GA-Binding Protein Transcription Factor Alpha (GABPa). Mol Biol Evol 33, 1231-1244 (2016).

168. Xue, H.H. et al. The transcription factor GABP is a critical regulator of B lymphocyte development. Immunity 26, 421-431 (2007).

169. Savage, K.J. et al. MYC gene rearrangements are associated with a poor prognosis in diffuse large B-cell lymphoma patients treated with R-CHOP chemotherapy. Blood 114, 3533-3537 (2009).

120 170. Hu, S. et al. MYC/BCL2 protein coexpression contributes to the inferior survival of activated B-cell subtype of diffuse large B-cell lymphoma and demonstrates high-risk gene expression signatures: a report from The International DLBCL Rituximab-CHOP Consortium Program. Blood 121, 4021-4031; quiz 4250 (2013).

171. Matoba, K. et al. Rho-kinase regulation of TNF-alpha-induced nuclear translocation of NF-kappaB RelA/p65 and M-CSF expression via p38 MAPK in mesangial cells. Am J Physiol Renal Physiol 307, F571-580 (2014).

172. Shimada, H. & Rajagopalan, L.E. Rho kinase-2 activation in human endothelial cells drives lysophosphatidic acid-mediated expression of cell adhesion molecules via NF-kappaB p65. J Biol Chem 285, 12536- 12542 (2010).

173. Meyer-Schwesinger, C. et al. Rho kinase inhibition attenuates LPS- induced renal failure in mice in part by attenuation of NF-kappaB p65 signaling. Am J Physiol Renal Physiol 296, F1088-1099 (2009).

174. Pore, D. et al. Identification of Ezrin-Radixin-Moesin proteins as novel regulators of pathogenic B-cell receptor signaling and tumor growth in diffuse large B-cell lymphoma. Leukemia 29, 1857-1867 (2015).

175. Rimsza, L.M. et al. Loss of MHC class II gene and protein expression in diffuse large B-cell lymphoma is related to decreased tumor immunosurveillance and poor patient survival regardless of other prognostic factors: a follow-up study from the Leukemia and Lymphoma Molecular Profiling Project. Blood 103, 4251-4258 (2004).

176. Aubrey, B.J. et al. An inducible lentiviral guide RNA platform enables the identification of tumor-essential genes and tumor-promoting mutations in vivo. Cell Rep 10, 1422-1432 (2015).

177. Mootha, V.K. et al. PGC-1alpha-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes. Nat Genet 34, 267-273 (2003).

178. Kamburov, A., Wierling, C., Lehrach, H. & Herwig, R. ConsensusPathDB--a database for integrating human functional interaction networks. Nucleic Acids Res 37, D623-628 (2009).

179. Parish, I.A. et al. Chronic viral infection promotes sustained Th1-derived immunoregulatory IL-10 via BLIMP-1. J Clin Invest 124, 3455-3468 (2014).

121

180. Goenka, R. et al. Cutting edge: dendritic cell-restricted antigen presentation initiates the follicular helper T cell program but cannot complete ultimate effector differentiation. J Immunol 187, 1091-1095 (2011).

181. Goenka, R. et al. Local BLyS production by T follicular cells mediates retention of high affinity B cells during affinity maturation. J Exp Med 211, 45-56 (2014).

182. Casola, S. et al. Tracking germinal center B cells expressing germ-line immunoglobulin gamma1 transcripts by conditional gene targeting. Proc Natl Acad Sci U S A 103, 7396-7401 (2006).

183. Dominguez-Sola, D. et al. The FOXO1 Transcription Factor Instructs the Germinal Center Dark Zone Program. Immunity 43, 1064-1074 (2015).

184. Sander, S. et al. PI3 Kinase and FOXO1 Transcription Factor Activity Differentially Control B Cells in the Germinal Center Light and Dark Zones. Immunity 43, 1075-1086 (2015).

185. Sage, P.T., Alvarez, D., Godec, J., von Andrian, U.H. & Sharpe, A.H. Circulating T follicular regulatory and helper cells have memory-like properties. J Clin Invest 124, 5191-5204 (2014).

186. Sage, P.T., Francisco, L.M., Carman, C.V. & Sharpe, A.H. The receptor PD-1 controls follicular regulatory T cells in the lymph nodes and blood. Nat Immunol 14, 152-161 (2013).

187. Xu, J. et al. Mice deficient for the CD40 ligand. Immunity 1, 423-431 (1994).

188. Foy, T.M. et al. In vivo CD40-gp39 interactions are essential for thymus- dependent humoral immunity. II. Prolonged suppression of the humoral immune response by an antibody to the ligand for CD40, gp39. J Exp Med 178, 1567-1575 (1993).

189. Takahashi, Y., Dutta, P.R., Cerasoli, D.M. & Kelsoe, G. In situ studies of the primary immune response to (4-hydroxy-3-nitrophenyl)acetyl. V. Affinity maturation develops in two stages of clonal selection. J Exp Med 187, 885-895 (1998).

190. Xu, H. et al. Regulation of bifurcating B cell trajectories by mutual antagonism between transcription factors IRF4 and IRF8. Nat Immunol 16, 1274-1281 (2015).

122

191. Crotty, S. T follicular helper cell differentiation, function, and roles in disease. Immunity 41, 529-542 (2014).

192. Okamoto, R. et al. FHL2 prevents cardiac hypertrophy in mice with cardiac-specific deletion of ROCK2. FASEB J 27, 1439-1449 (2013).

123