by
THOMAS ALEXANDER PACKARD B.S., University of Colorado Denver, 2008
A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment
of the requirements for the degree of Doctor of Philosophy
Immunology Program 2015
ii
This thesis for the Doctor of Philosophy degree by Thomas Alexander Packard
has been approved for the Immunology Program
by
Raul Torres, Chair Eric Clambey Rachel Friedman
Peter Gottlieb John Kappler Gregory Owens John Cambier, Advisor
iii Packard, Thomas Alexander (Ph.D., Immunology)
The Anti-Insulin B Cell Receptor in Autoimmune Diabetes
Thesis directed by Distinguished Professor and Chair John C. Cambier
ABSTRACT
B cells are required during the pathogenesis of autoimmune diabetes and this functionality is independent of antibody secretion, but dependent on MHC expression and B cell receptor (BCR) specificity. Together, this suggests that autoreactive B cells drive diabetes by presenting antigen to effector T cells. An insulin-binding BCR is sufficient for diabetes development in NOD, but monoclonal transgenic (125Tg) have reduced disease compared to polyclonal transgenic (VH125) animals. This prompted us to investigate the insulin-binding affinities of BCRs derived from the VH125 animal, and consequential B cell functions in tolerance and autoimmune processes.
We hypothesized that BCR affinity for insulin impacts the level and type of tolerance, and that intrinsic genetic susceptibility of the B cell affects these thresholds. Further, we posited that the BCR affinity alters the role of the insulin-specific B cell in diabetes: assigning the clone to a fate of rigid tolerance in the case of too much avidity, or complete ignorance if too low. B cells bearing BCR of optimal affinity are the most able to participate in pathogenesis, through acquiring insulin, activation, and antigen presentation.
Within we use a novel method of combined transgenic and retrogenic BCR to demonstrate that affinity of the BCR is critical to bind insulin and induce signaling. We found that the 125 BCR exhibits a much higher avidity than predicted by studies of soluble Ig, explaining much of its observed functionality. We show that BCR affinity for insulin and genetic background affect the tolerization of B cells exposed to the autoantigen. Insulin-specific B cells can activate insulin-specific T cells, and tolerized B
iv cells are decreased in this capability. Finally, we show that in vivo acquisition of insulin requires sufficient BCR affinity and permissive host/tissue environment.
We propose the model that BCR affinity, pancreas environment, and B cell tolerance genes converge in the NOD animal to allow for acquisition of insulin and autoimmunity.
The form and content of this abstract are approved. I recommend its publication. Approved: John C. Cambier
v
ACKNOWLEDGMENTS
I would like to acknowledge John Cambier, who has been my mentor over the past 7 years: to his training I owe not just my knowledge of science, but how to think as a scientist. My thesis advisory committee, chaired by Raul Torres, and including members Eric Clambey, Rachel Friedman, Peter Gottlieb, John Kappler, Gregory Owens, and past member Dirk Homann, has guided me throughout my graduate career. They have provided essential input in the direction of the project, constant support, and collaboration.
I would like to thank my wife, Sarah, for her love and encouragement. She has been my friend, source of strength, and recent mother of our child—and Evelyn Annette deserves acknowledgement for being a perfect baby, and sleeping as much as a father could hope! Thank you to my parents for everything they have done: my father for his strong support of our family, and my mother for teaching us and inspiring me to become a scientist. To the rest of my family, siblings, grandparents and in-laws: thank you for everything you have done to help me achieve this capstone of my education.
I would like to recognize everyone who has helped me on this path. Those who are noted within the text have been particularly helpful specifically to this project, and I am grateful. Thank you to the past and present members of the Cambier Laboratory, and I would like to especially thank Mia Smith and Andrew Getahun.
However, the people mentioned here only represent a small number of the people who have provided me with education, discussion, and inspiration. To those not mentioned, you know who you are, thank you.
TABLE OF CONTENTS
CHAPTER
I. INTRODUCTION ... 1
II. MATERIALS AND METHODS ... 29
III. RESULTS AND DISCUSSION ... 47
IV. CONCLUSIONS AND FUTURE DIRECTIONS ... 91
APPENDIX A. INSULIN BINDING B CELLS IN HUMANS ... 117
vii
LIST OF FIGURES
FIGURE
1. BCR heavy chain transgenes convey susceptibility or resistance to diabetic
disease in NOD mice. ... 4
2. Endogenous insulin is bound by 125Tg B cells. ... 5
3. Incidence of diabetes in VH125Tg NOD mice is similar in the presence [filled circles] or absence [empty circles] of endogenous B cells & T1DM develops in NOD mice that harbor anti-insulin 125Tg. ... 7
4. Divergent outcomes for diabetes in NOD mice are associated with different Ig HC Tgs ... 8
5. Pairing VH125 to multiple VL genes confers insulin-binding capacity with varying ligand affinity. ... 10
6. B cell receptor signaling and regulation. ... 13
7. Example of SPR demonstrating high and low affinity insulin-binding of model Igs. ... 47
8. SPR can detect Protein G-adsorbed A12 Ig binding to soluble insulin. ... 50
9. A12 and EW6 Igs are not polyreactive. ... 51
10. Construction and validation of TR-B cells that bind insulin. ... 53
11. VH125-family insulin-binding Igs recognize a shared epitope. ... 55
12. mAb123 can detect receptor occupancy in vitro following incubation with unlabeled insulin ... 56
13. 125 TR-B cells bind nanomolar biotin-insulin and flux Ca2+ in response to streptavidin crosslinking ... 57
14. 125 has a lower off-rate and increased half life than A12. ... 58
15. Alignment of CDR3s from model Ig VL genes. ... 60
16. CDR3 tyrosine is required for A12 binding to insulin ... 62
17. YF TR-B cells do not mobilize calcium in response to insulin. ... 62
18. Culture with insulin drives affinity-dependent downregulation of autoreactive BCR. ... 64
20. Insulin-binding affinity affects downregulation of surface IgM. ... 66 21. B6 low affinity TR-B cells have decreased IgM following insulin exposure, as
compared to their NOD counterparts. ... 67 22. Differential calcium responses following tolerance based on affinity and
genetic background ... 69 23. NOD TR-B cells express slightly higher surface IgM at baseline and upon
treatment with insulin. ... 70 24. Immature TR-B cells exposed to Ag do not significantly upregulate activation
markers. ... 71 25. TCR hybridoma responses to insulin protein and mimitope peptides. ... 75 26. TR-B cells present antigen to insulin-specific T cell clone 12-4.1 and this is
reduced following in vitro tolerance. ... 76 27. Potential mechanisms for decreased APC function by TR-B cells following
insulin pretreatment. ... 79 28. TR-B cells can be detected 7 days post-adoptive transfer. ... 82 29. Splenic TR-B cells exhibit a mature-follicular phenotype one week
post-transfer. ... 83 30. 2-photon micrographs of TR-B cells in the PLN. ... 85 31. High affinity TR-B cells trend towards enrichment in the PLN. ... 85 32. High affinity A12 TR-B cells have detectable receptor occupancy in the
spleen and PLN of NOD mice, while EW6 are likely clonally ignorant. ... 86 33. TR-B cells maintain insulin-binding capability following adoptive transfer. ... 87 34. BCR affinity, host and tissue environment contribute to the acquisition of
autoantigen. ... 88 35. Attempts to visualize insulin specific T/B intereactions in vivo were
unsuccessful ... 89 A.1. IBC labeling and enrichment from human PBLs ... 120 A.2. Representative cytograms of phenotypic analyses of PBLs enriched for
IBCs ... 121 A.3. New-onset and pre-T1D patients have decreased BND frequency among
A.4. BND cells have increased number of positive-charged residues and increased
length of heavy chain CDR3, along with increased Jh6 usage ... 122 A.5. BCR of BND cells is of higher affinity for insulin and more polyreactive ... 122
x
LIST OF ABBREVIATIONS
Ab, antibody Ag, antigen
ANOVA, analysis of variance APC, Ag-presenting cell β-ME, 2-mercaptoethanol BCR, B cell antigen-receptor
BND, human naïve IgD+IgM− B cells
BSA, bovine serum albumin
CDR, complementarity determining region
CFSE, carboxyfluorescein succinimidyl ester
CTL, cytotoxic T lymphocyte
DMEM, Dulbecco's modified Eagle's medium
DNA, deoxyribonucleic acid
EC50, 50% effective concentration
ELISA, enzyme-linked immunosorbent assay
Fab, Ag-binding fragment
FACS, fluorescence-activated cell sorter FBS, fetal bovine serum
g, gram h, hour
HLA, human histocompatibility leukocyte Ag
HRP, horseradish peroxidase IFN, interferon
Ig, immunoglobulin IgH, Ig heavy chain IL, interleukin
IMDM, Iscove's modified Dulbecco's medium
ITAM, immunoreceptor tyrosine-based activation motif
ITIM, immunoreceptor tyrosine-based inhibitory motif i.v. intravenous Ka, association constant Kd, dissociation constant KD, affinity constant Koff, off-rate kD, kilodalton LB, lysogeny broth LPS, lipopolysaccharide mAb, monoclonal Ab
MACS, magnetic-activated cell sorting
mg, milligram
MHC, major histocompatibility complex min, minute
mL, milliliter
mRNA, messenger RNA μg, microgram
μL, microliter
m.w., molecular weight
n, number in study or group
ND, not detected
NOD, nonobese diabetic NS, not significant OD, optical density OVA, ovalbumin
p, probability
PAGE, polyacrylamide gel electrophoresis
PBL, peripheral blood lymphocyte PBMC, peripheral blood mononuclear cell
PBS, phosphate-buffered saline PCR, polymerase chain reaction PE, phycoerythrin
PerCP, peridinin chlorophyll protein PG, prostaglandin
PI3K, phosphatidylinositol 3-kinase r, recombinant
R, receptor
RAG, recombination-activating gene RBC, red blood cell
RCF, relative centrifugal force RNA, ribonucleic acid
rpm, revolutions per minute RT-PCR, reverse transcriptase polymerase chain reaction s, second
s.c., subcutaneous SD, standard deviation SDS, sodium dodecyl sulfate SEM, standard error of the mean
SHIP, src homology 2-containing inositol 5' phosphatase
SHP, Src homology region 2 domain-containing phosphatase-1
t1/2, half-life, half-time
xi
TBST, TBS with Tween 20 TCR, T cell antigen-receptor Tg, transgenic
TLR, Toll-like receptor TNF, tumor necrosis factor TR, transgenic/retrogenic V region, variable region of Ig
V(D)J or VDJ, variable diversity joining VL, variable region of light chain VH, variable region of heavy chain VH125, gene bearing variable region of the 125 mAb
VH281, germline revertant of VH125 CDR2
CHAPTER I INTRODUCTION
The B cell antigen-receptor (BCR) is central to the development and functionality of the B lymphocyte. The work described herein seeks to explore the role of BCR in determining the functionality of B cells in the context of autoimmune diabetes. We and others have found that B cells—and more specifically, B cells of certain specificity—are essential for the normal development of disease in the non-obese diabetic (NOD) model (Serreze et al. 1996; Hulbert et al. 2001). We set out to interrogate the role of affinity of insulin-specific BCR in antigen acquisition, tolerance mechanisms, and participation in disease.
B cells in type 1 diabetes
Type 1 diabetes (T1D) is an autoimmune disease characterized by immune-mediated destruction of the insulin-producing beta cells of the pancreas. Across the globe T1D is increasing in incidence; and currently there exist limited therapeutic interventions, no prevention strategies, and no cure (Gale 2002). The most common therapy for treating T1D is insulin-replacement, and patients who fail to respond favorably to these regimens are candidates for islet or whole pancreas transplantation (Robertson et al. 2006). However, neither of these therapies addresses the underlying autoimmune etiology driving disease.
Recently, a clinical study demonstrated limited, but significant, efficacy of B cell depletion therapy in new-onset T1D patients (Pescovitz et al. 2009). This study showed significant preservation of islet cell function one year following B cell depletion. These findings and others, prompted us to further investigate the role for B cells in autoimmune diabetes. To improve on the non-specific depletion of B cells, and related
immunodeficiency, we are interested in characterizing and targeting the most relevant populations of autoantigen-specific B cells.
While our laboratory continues to study T1D in human patients, and recently published a characterization of insulin-specific B cells during disease development (Smith et al. 2014) (some of this work is described in Appendix A), the bulk of the studies herein utilize the NOD mouse model of autoimmune diabetes. Makino, et al., originally described the NOD mouse in 1980 (Makino et al. 1980); and despite inherent limitations and caveats characteristic of such studies, it represents the best current mouse model of human T1D.
Since 1980, much work has been done to determine the immune mechanisms underlying disease in NOD animals. It is well accepted that CD4+ T cells are the ultimate effector cells responsible for beta cell destruction, and a variety of islet antigen-specific CD4+ T cells can transfer disease to NOD.RAG-/-.SCID animals (Burton et al. 2008; Babad et al. 2010). Despite the critical role for CD4+ T cells in mediating pathology, the etiology of disease development in unmanipulated NOD is much more complex.
Behind the activation of CD4+ T cells lies a complex network of other critical contributors. It has been shown that B cells are necessary for diabetes development (Serreze et al. 1996; Noorchashm et al. 1997; Akashi et al. 1997), and depletion of B cells can ameliorate disease (Xiu et al. 2008; Fiorina et al. 2008; Hu et al. 2007; Zekavat et al. 2008; Mariño et al. 2011; Forsgren et al. 1991). Furthermore, specificity of the BCR drives this pathogenicity (Hulbert et al. 2001; Silveira et al. 2002). However, despite the overwhelming prevalence of autoantibodies to islet antigens in human diabetics, and their usefulness as biomarkers (Orban et al. 2009), studies in NOD demonstrate that membrane-bound immunoglobulin (mIg) is critical to drive disease, and secreted immunoglobulin (Ig) is dispensable (Wong et al. 2004). Thus, anti-islet antibody titres
may simply function as surrogates, reflecting B cell activation in disease, and not be directly active in pathology. Together, our current model proposes that B cells function through critical antigen presentation, licensing capacity, or a combination of both; however, much work remains in directly testing this functionality.
Insulin-binding B cells play an important role in autoimmune diabetes
Antibodies to insulin, though not perfectly correlated with disease, arise early and often precede any clinical symptoms (Orban et al. 2009). Additionally, in retrospective analyses of T1D patients from the B cell-depletion clinical trial found that insulin autoantibody titers at entry to the study inversely correlated with preservation of insulin production (Yu et al. 2011). Thus, B cells specific for insulin are of interest in T1D. As mentioned above, NOD mice with fixed transgenic BCR of innocuous specificity are protected from disease (Hulbert et al. 2001; Silveira et al. 2002) (Figures 1 & 4). By contrast, mice with a BCR specific to insulin do develop disease (Hulbert et al. 2001).
In the VH125 transgenic, endogenous light chains are rearranged to pair with VH125, giving rise to a heterogeneous population of insulin-binding B cells, representing ~1% of circulating mature B cells in the VH125.NOD. In our colony, we observe complete penetrance of disease among female NOD mice bearing the VH125 heavy chain transgene (Figure 1). This contrasts with the VH281.NOD animals, which express BCR incapable of binding insulin, and are protected from diabetes. Our NOD transgenic colonies display similar diabetes incidence to that published by the Thomas laboratory (Hulbert et al. 2001). These data support a role for insulin-binding B cells as drivers of disease in NOD.
Figure 1. BCR heavy chain transgenes convey susceptibility or resistance to diabetic
disease in NOD mice. Female mice housed in specific pathogen-free conditions at National Jewish Health Biological Resource Center (Rochelle Hinman). Diabetes was defined as two consecutive elevated blood glucose readings (>200 mg/dl), VH125.NOD n=22, NOD n=18, VH281.NOD n=21.
Studies of the 125Tg monoclonal anti-insulin B cell
The laboratory of Dr. James Thomas at Vanderbilt University created the 125Tg animal (Rojas et al. 2001; Hulbert et al. 2001). The transgenic BCR was cloned from an anti-insulin hybridoma originally generated by immunizing BALB/c animals with insulin (Schroer et al. 1983). This monoclonal antibody (mAb) 125, along with mAb123 and others, were characterized in their ability to bind insulin. The Thomas laboratory has done much work studying multiple iterations of transgenic animals derived from the mAb125 hybridoma, introduced below.
They first described the 125Tg in the context of C57BL/6 animals, and demonstrated that these B cells have characteristics similar to anergy, and detectable receptor occupancy (Figure 2) (Rojas et al. 2001). This result was surprising due to the fact that the affinity of mAb125 for insulin was determined to be 3x10-8 M for human insulin, and had been reported to bind murine insulin at 5x10-6-10-7 M (Rojas et al. 2001;
Schroer et al. 1983; Acevedo-Suárez et al. 2005). Circulating concentrations of insulin are hundreds to thousands of times lower, thus the predicted status of the 125Tg B cell would be clonal ignorance. In fact, estimates vary as to the concentration of circulating insulin between species and strains; but likely resting concentrations are much less than 100 pM, with high glucose challenge driving release of perhaps 1 nM in the circulation closest to secretion (Andrikopoulos et al. 2005; Wang et al. 2013). These levels would be compatible with affinity for the insulin receptor, which binds insulin with 5.7×10−11 M and 6.3×10−9 M affinities, for the first and second binding, respectively (Hoyne et al. 2000). Thus, we hypothesized that these sub-nanomolar concentrations of circulating insulin in vivo—paired with competition with insulin receptor binding—would be insufficient to activate or anergize all but the highest avidity BCR-bearing cells.
Figure 2. Endogenous insulin is bound by 125Tg B cells. Biotinylated anti-insulin
mAb123 and avidin FITC (mAb123 FITC) were used in flow cytometry to identify B cells (PE-B220, left axis) that bind and display insulin on their surface. A (left), No binding in spleens of B6 mice; B (right), most 125Tg B cells are bound by mAb123, indicating the display insulin on their surface. The binding in B is totally inhibited by soluble insulin (50 μg/ml).1
The results of the Rojas, Hulbert & Thomas study demonstrating an anergic phenotype of 125Tg B cells in vivo were unexpected. The prediction would be that a B
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
1 This figure and figure legend reproduced here with permission from JI. Rojas M,
Hulbert C, Thomas JW. Anergy and not Clonal Ignorance Determines the Fate of B Cells that Recognize a Physiological Autoantigen. J Immunol. 2001, 166:3194-3200. doi: 10.4049/jimmunol.166.5.3194. Copyright 2001, The American Association of
cell bearing a receptor with 10-7 M affinity would be <0.1% receptor occupied at the highest possible concentrations of insulin. However, they found that virtually all of the 125Tg B cells were receptor occupied (Figure 2) (Rojas et al. 2001).
Furthermore, insulin should exist in a monomeric, soluble form and thus incapable of activating BCR signaling known to be required for induction of anergy. To the contrary, they report that circulating insulin occupies a sufficient level of 125Tg BCR to drive an “anergic” status of the transgenic B cells, which were unresponsive to immunization with exogenous insulin. They proposed this disparity was possibly explained by an induced fit binding model related to a “molten” quality of hormonal insulin (Rojas et al. 2001).
Concurrent with characterization of 125Tg.C57BL/6, Hulbert, et al. demonstrated that the 125Tg was sufficient to support development of diabetes in NOD, with a reported disease penetrance of ~%40 in female 125Tg animals (Figure 3) (Hulbert et al. 2001). Later, they showed that this diabetes occurred despite intact “anergy”, as defined by an inability to respond to immunization with insulin (Acevedo-Suárez et al. 2005). The 125Tg B cells were capable of upregulating costimulatory molecules (e.g. CD86), and were poised to present antigen to diabetogenic T cells.
When light chain usage is unrestricted, VH125.NOD animals develop disease with accelerated onset and increased prevalence as compared to wild type (WT) NOD; and in contrast, mice with the VH281 transgene—which is the germline revertant of VH125, incapable of binding insulin—exhibited nearly complete protection (Figures 1 & 4) (Thomas and Hulbert 1996; Hulbert et al. 2001). The disparity in disease prevalence between VH125.NOD and 125Tg.NOD prompted us to further investigate the role of affinity of the BCR for insulin in the context of NOD.
Figure 3. Incidence of diabetes in VH125Tg NOD mice is similar in the presence [filled
circles] or absence [empty circles] of endogenous B cells (left). VH125Tg NOD mice were intercrossed and then backcrossed with NOD.μMT to produce cohorts of female mice without endogenous B cells (μMT-/-, n = 10) or without (sic) endogenous B cells (μMT-/+, n = 15).2 T1DM develops in NOD mice that harbor anti-insulin 125Tg (right). Diabetes incidence curves compare disease development in 125Tg NOD mice ([open circle]; n = 20) with that in nontransgenic littermate controls ([filled circle]; n = 40). All mice were female, and diabetes was based on two consecutive blood sugar measurements >200 mg/dl.3
As the rate and penetrance of diabetes can vary between NOD colonies, we were interested in the relative diathesis or protection for disease in the heavy chain Tg animals housed in our SPF facility (Leiter 1993). We observe similar, though more polarized, outcomes in our colonies of transgenic animals (Figures 1, 4).
The Thomas laboratory extended their studies to demonstrate that insulin-binding B cells of a variety of affinities participate in disease (Woodward and Thomas 2005), that 125Tg B cells are effective antigen-presenting cells (APCs) to insulin-specific T cells
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2
This figure and figure legend reproduced here with permission from JI. Hulbert C, Riseili B, Rojas M, Thomas JW. Cutting Edge: B Cell Specificity Contributes to the Outcome of Diabetes in Nonobese Diabetic Mice. J Immunol. 2001, 167:5535-5538. doi: 10.4049/jimmunol.167.10.5535. Copyright 2001, The American Association of Immunologists, Inc.
3
This figure and figure legend reproduced here with permission from JI. Uncoupling of Anergy from Developmental Arrest in Anti-Insulin B Cells Supports the Development of Autoimmune Diabetes. Acevedo-Suárez CA, Hulbert C, Woodward EJ, Thomas JW. J Immunol. 2005, 174:827-833. doi: 10.4049/jimmunol.174.2.827. Copyright 2005, The American Association of Immunologists, Inc.
(Kendall et al. 2013), and that depletion of B cells bound to insulin delays or prevented diabetes (Henry et al. 2012).
Figure 4. Divergent outcomes for diabetes in NOD mice are associated with different Ig
HC Tgs. Incidence of diabetes in cohorts of female mice carrying VH125HC Tg ([diamond], n = 22), VH281HC Tg ([square], n = 19), and nontransgenic littermates ([triangle], n = 40). VH125 contains mutations in CDRH2 that favor insulin binding while VH281 is unmutated. Mice were considered diabetic if two consecutive blood sugars were >200 mg/dl.4
Together, these studies demonstrate that BCR specificity is a critical component of autoimmune diabetes development. The 125Tg model demonstrated that insulin-binding B cells are sufficient to satisfy this specificity requirement; however, when comparing the penetrance of disease between transgenic cohorts, ~40% for 125Tg and >90% for VH125, we proposed a more in-depth evaluation of the functional role of BCR affinity for insulin. In the heavy chain-only VH125, multiple endogenous light chain rearrangements confer insulin specificity (Woodward and Thomas 2005), and these unique clones have a diversity of binding affinities for their ligand. We hypothesize that these other insulin-binding B cells are contributing to the rapidity and penetrance of disease in VH125 as compared to 125Tg, i.e. B cells bearing BCR of higher or lower affinity than 125Tg may explain the increased pathogenesis.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
4
This figure and figure legend reproduced here with permission from JI. Hulbert C, Riseili B, Rojas M, Thomas JW. Cutting Edge: B Cell Specificity Contributes to the Outcome of Diabetes in Nonobese Diabetic Mice. J Immunol. 2001, 167:5535-5538. doi: 10.4049/jimmunol.167.10.5535. Copyright 2001, The American Association of Immunologists, Inc.
Insulitis in NOD pathology
Despite the lack of clinical diabetes, and lifelong normoglycemia observed among VH281.NOD animals (Figure 1), they develop a smoldering insulitis (Hulbert et al. 2001). Insulitis is characterized by endothelial swelling, accumulation of macrophages and dendritic cells, and lymphocytic infiltration (Jansen et al. 1994). Low-grade insulitis can be detected as early as 3-weeks of age in female NOD animals, and normally progresses to beta cell destruction and clinical hyperglycemia by about 20-weeks of age. However, insulitis has been reported to be reduced or absent in mice deficient of B cells (Noorchashm et al. 1997; Akashi et al. 1997). Though somewhat reduced in comparison to VH125, 90% of the VH281.NOD mice exhibit significant insulitis (infiltration > 10%) (Hulbert et al. 2001). Thus, early infiltration events may be BCR-independent, but destruction of beta cells to the extent necessary for diabetes is BCR-dependent.
Model VH125-family insulin-binding Igs
We used the VH125 heavy chain transgenic model as a platform with which to pair light chains that would confer varying, known, affinities for insulin (Figure 5). The light chain VLEW6 was generously provided to us by the Thomas laboratory, and we generated the VLA12 light chain, both cloned from hybridomas generated from immunized VH125 animals. By expressing the light chain 125 with VH125, we can reconstitute an analog to the mAb125 immunoglobulin, analogous to the 125Tg BCR. Together, these combinations provide the ability to expand on studies of 125Tg, to include immunoglobulin molecules of expected higher and lower affinities for insulin.
Figure 5. Pairing VH125 to multiple VL genes confers insulin-binding capacity with
varying ligand affinity. VH281 does not bind insulin when paired with any known light chain, and YF (a mutant of A12) disrupts insulin binding. Affinities for insulin binding determined by surface plasmon resonance using recombinant human immunoglobulin (Frank Conrad, unpublished). † The affinity of mAb125 previously reported as 3x10-8 M (Schroer et al. 1983), and our preparation of recombinant immunoglobulin pairing VH125 and VL125 was determined to be 1.6x10-8 M.
Transgenic/Retrogenic BCR approach
Due to our interest in determining the functions of insulin-binding BCR rather than soluble Ig, and in hopes of developing a platform for rapid analysis of myriad BCR specificities, we utilized a retrogenic approach. By expressing BCR components using retroviral transduction, we could theoretically avoid the time and expense involved in creating transgenic BCR animals. The method for retrogenic expression of Ag-receptor was first described for the TCR (Holst et al. 2006), and has been extensively deployed in the study of T cell specificities since, including multiple pancreas-specific TCRs (Burton et al. 2008).
However, to date, only a single, very recent study has been published pertaining to retrogenic BCR (Freitag et al. 2014). They found it possible to express the MD4 BCR retrogenically in an IL-7 dependent pro-B cell line, but could not achieve chimerism with retrogenic stem cells. The authors posited that in light of their difficulties and the lack of any published retrogenic BCR, “achieving BCR-expression in vivo with retrogenic
technology is highly challenging if not impossible.” (Freitag et al. 2014) Unfortunately, this publication was released long after the inception of this project, and thus we attempted the highly challenging or impossible.
We chose pMIG as our vector, which was recommended for use in the creation of TCR retrogenics (Holst et al. 2006). The pMSCV backbone of the virus is less silenced in stem cells than alternative vectors, and the eGFP after an internal ribosomal entry site (IRES) allows for detection of transduced cells. We proposed to package the retrovirus with pCL-Eco using Phoenix Eco cell lines, due to success in TCR transductions (Naviaux et al. 1996; Swift et al. 2001; Holst et al. 2006).
LT-HSC isolation and expansion
Retrogenic TCR protocols, and the recent BCR attempt, utilize 5-FU-treated BM, cultured in a mixture of IL-3, IL-6, and SCF for less than a week before being used to reconstitute an animal (Holst et al. 2006; Freitag et al. 2014). An alternative approach would be to isolate and expand the hematopoietic stem cells (HSCs), most importantly, the long-term population of HSCs (LT-HSCs). LT-HSCs are self-renewing and multi-potent progenitor cells that give rise to all hematopoietic lineage cells upon reconstitution of lethally irradiated animals (Seita and Weissman 2010). Because they are generally not rapidly proliferating, they are thought to be difficult cells to infect with retroviral vectors. However, recent work has demonstrated methods to induce rapid and large expansion LT-HSCs in vitro (Varnum-Finney et al. 2011). We proposed that we might be able to take advantage of the mitotic activity induced during these LT-HSC expansion cultures to more readily infect this progenitor population, as ecotropic retrovirus does not readily infect non-mitotic cells (Naviaux et al. 1996).
IL-7 in vitro differentiation of B cells
Most of the studies described herein employed B cells derived from IL-7 cultured bone marrow. IL-7 signaling is essential to the development of B cells in vivo (von Freeden-Jeffry et al. 1995). IL-7 induces CLP expression of Early B cell Factor, an essential regulator of B cell development (Kikuchi et al. 2008; Hagman et al. 1991). Immature B cells can be readily generated in large quantity and purity by simply culturing bone marrow cells in vitro cultures supplemented with IL-7. As the cells are actively proliferating in this culture, they are favorable targets for retroviral transduction (Rowland, DePersis, et al. 2010).
BCR signaling overview
The BCR is comprised of mIg non-covalently associated with the heterodimeric Igα (CD79a) and Igβ (CD79b) signaling molecules. The mIg component of the BCR conveys the “specificity” of a given B cell clone. During normal B cell development, the heavy and light chain genes encoding mIg are stochastically rearranged, resulting in an incredibly diverse repertoire of reactivity. Canonical recognition of antigen (Ag) by BCR is mediated via binding of the complementarity-determining region (CDR) of the mIg to cognate Ag. Binding of Ag sufficient to induce crosslinking of BCR facilitates trans-phosphorylation of the immunoreceptor tyrosine-based activation motif (ITAM) containing CD79a and CD79b via Src-family tyrosine kinases (SFK) and Syk.
Aggregation of the BCR by Ag-binding drives multiple activating and inhibitory signaling pathways (Figure 6), discussed at more length in this review (Packard and Cambier 2013). The net result of successful activation signals drive clonal proliferation, as well as upregulation of costimulatory molecules, critical for priming effective activation in responding T cells.
Figure 6. B cell receptor signaling and regulation. (Upper panel) B cell activation
Upon ligation of the BCR (1), ITAMs become phosphorylated via activity of SFKs (such as Lyn) and Syk. Syk and SFKs then phosphorylate signalosome components (2). The signalosome is associated with CD79a non-ITAM phosphotyrosine residues via binding of the adaptor protein Blnk (not illustrated). Activated Btk phosphorylates PLCγ2, which in turn cleaves the phosphoinositide PI(4,5)P2, releasing IP3 into the cytosol and forming
DAG (3). IP3 binds IP3R in the endoplasmic reticulum, releasing Ca2+ into the cytoplasm.
The decrease in endoplasmic reticulum [Ca2+] activates STIM1, which binds ORAI in the plasma membrane, forming the CRAC channel and allowing for the influx of extracellular Ca2+ ions (4). RasGRP and protein kinase C (PKC) are activated by binding DAG, and feed into the MEK/MAP kinase (5) and NFκB activation pathways, respectively. CD19 plays an important role in amplifying the BCR signal via processive activation of Lyn, and activation of PI3K (6). Along with the recruitment of PH domain-containing signalosome components, the accumulation of PI(3,4,5)P3 drives activation of Akt (7). (Lower panel) B
cell deactivation: Lyn phosphorylates immunoreceptor tyrosine-based inhibition motifs (ITIMs) in CD22 and FCγRIIb. These ITIMs activate SHP1 and SHIP1, which function to inhibit BCR signaling. The protein phosphatase SHP1 has many substrates, including CD79, Syk, Grb2, and Vav, as well as others not shown. Additionally, ITIMs and mono-phosphorylated ITAMs can activate the lipid phosphatase SHIP1. SHIP1 hydrolyzes the phosphate at position 5 of PI(3,4,5)P3, while PTEN removes that at position 3. This
decrease in PI(3,4,5)P3 concentration results in the disassociation of many PH
domain-containing molecules, inhibiting signalosome assembly and downstream signaling. Illustrated structures not to scale, references in [Packard and Cambier 2013].5
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
5 This figure and figure legend reproduced here with permission and under the Creative
Commons Attribution-Non Commercial License. Packard TA, Cambier JC. B lymphocyte antigen receptor signaling: initiation, amplification, and regulation. F1000 Prime Reports. 2013, 5:40. doi: 10.12703/P5-40
BCR proximal signaling via SFK and Ca2+-release drive clathrin-dependent, receptor-mediated endocytosis of the BCR-Ag complex (Stoddart et al. 2002). Additionally, pSyk recruits c-Cbl and/or Itch, which ubiquitinate CD79, marking these nascent endosomes for preferential trafficking to late endosomes (Zhang et al. 2007; Katkere et al. 2012). This “sorting” of endosomes, combined with the overall increase in cellular activation following BCR-mediated recognition of Ag, likely synergize to confer upon B cells a unique advantage among professional APCs in presentation of cognate antigen. Indeed, B cells have been described as having a 1,000 to 10,00-fold increase in the ability to present their cognate Ag to T cells as compared to non-specific B cells or other professional APCs (Kakiuchi et al. 1983; Lanzavecchia 1985; Lanzavecchia 1990; Pierce et al. 1988).
BCR signaling induction by soluble, monovalent Ag
Of particular importance for induction of signaling by soluble Ags such as insulin, signal initiation requires that this interaction be minimally bivalent in order to cross-link BCR (Minguet et al. 2010; Woodruff et al. 1967). Insulin exists in multimeric forms, such as hexamers, and is further complexed in crystalline insulin in beta cell secretory granules (Ciszak and Smith 1994). These complexes are dependent on Zn2+, present at high concentration in the beta cell, and presumably these rapidly dissolve to monomeric, active hormone, following secretion into the circulation. An excellent review on the role of zinc in the pancreas was recently published by Li, in which he discusses Zn2+ release as a function of exocytosis of insulin by the pancreas (Li 2013). Thus, it is possible that in proximity to the pancreas insulin may exist in higher valency then the spleen.
Due to its monovalency it is unlikely that insulin induces BCR crosslinking, and thus signaling, in the periphery. However, some controversy exists in the field regarding
aggregation dependence of BCR signaling, with reports varying as to the ability of monovalent, soluble Ag to activate detectable signaling. Much of this controversy may be due to the nature of model antigens utilized in the studies. Most notable is hen-egg lysozyme (HEL), which has been implicated as capable of stimulating BCR signaling as a soluble monomer (Kim et al. 2006; Mukherjee et al. 2013). However, multiple non-HEL soluble antigens have shown to be incapable of monovalent signaling, examples include: fAb’ monomers, recombinant monovalent peptides, and unconjugated haptens, to name a few (Woodruff et al. 1967; Minguet et al. 2010; Gauld et al. 2005). We have found in our laboratory that HEL in solution quickly forms multimers detectable by column chromatography (Dr. Linda Akerlund, unpublished), and thus may not truly be monovalent. Furthermore, as HEL is a highly charged lysozyme, it may bind to sugars on cell surfaces, or bind via electrostatic interactions (pI: 11.35, Sigma).
Monovalent Ags are capable of activating BCR when bound to lipid bilayers (Tolar et al. 2009; Taylor et al. 2012). Thus, it is theoretically possible that a BCR could recognize insulin bound to insulin receptor, and engage signaling. For example, the exposed face of a molecule of insulin bound to insulin receptor on a neighboring cell— such as a monocyte (which express high levels of surface insulin receptor)—could function to present antigen to an insulin-specific B cell. Furthermore, a low affinity, insulin-binding BCR could bind in cis to hormone-bound insulin receptor in a cooperative manner. However, in early studies epitope mapping the anti-insulin hybridomas, mAb125 and mAb123 were demonstrated to be incapable of binding insulin bound to insulin receptor (Taylor et al. 1984). Of the 16 anti-insulin hybridomas tested in that study, a single Ab was capable of binding insulin in the context of receptor, and with reduced affinity. From this small survey, it is our opinion that this phenomenon of insulin receptor-mediated activation of insulin-specific B cells in vivo, though possible, is not unlikely; and
is not involved in the VH125 family binding to insulin, and does not account for the anergic phenotype of 125Tg B cells.
Antigen presentation may not require active signaling, as the BCR may quietly gather antigen and internalize spontaneously; though still serving concentrate the cognate antigen in endosomes for processing and presentation. The Thomas laboratory has demonstrated that this likely can occur in insulin-specific B cells (Kendall et al. 2013). However, this method arguably would not confer the same efficacy as BCR-activated antigen presentation, absent BCR ubiquitination driving the preferential association of BCR-Ag endosomes with lysosomes and other antigen processing and presentation machinery. Additionally, Ag internalized passively by BCR would also not drive activation of the B cell, nor upregulation of costimulation molecules.
Polyreactive B cells
Some proportion of insulin-binding B cells in mice and humans are polyreactive (Casali and Notkins 1989; Chen et al. 1995; Wardemann et al. 2003). Insulin has an isoelectric point of 5.3 (Merck), bearing a negative charge at physiological pH. Thus, binding is correlated to positive charged residues in and around Ig CDRs (Thomas and Hulbert 1996; Wardemann et al. 2003). The classification of “polyreactive” varies; though our laboratory defines it as binding to insulin, chromatin, and LPS, as measured by ELISA (Smith et al. 2014).
Polyreactive B cells are thought to play an important house-keeping role: either through direct binding of apoptotic cells or microbes, or by secreting natural Ig, which in turn can assist in clearance of debris or pathogens (Casali and Notkins 1989; Baumgarth 2011). Though detectable in all B cell compartments, they are reduced through consecutive checkpoints in B cell development, and enriched in compartments
traditionally associated with autoreactive cells, such as marginal zone, B-1, and BND in
humans (Kanayama et al. 2005; Baumgarth 2011; Duty et al. 2009).
B cell tolerance
Central Tolerance. B cell tolerance is divided in two main categories, central and peripheral. The definition of central tolerance is limited by location, i.e. bone marrow. During development in the bone marrow, the B cell somatically rearranges V, D, and J segments to form the heavy chain of the BCR Ig (IgH) at the pre-pro B stage. This IgH is tested for functionality (e.g. in frame rearrangement, no early stop codons), as well as self-reactivity by pairing with a germline encoded surrogate light chain (Igϕ). IgH and Igϕ together constituting the pre-BCR, is transported to the cell surface and upon functional expression of the pre-BCR the cell advances to the pro-B stage. The pre-BCR provides signals to inhibit further rearrangement at the heavy chain loci, and to begin V/J rearrangement to form a light chain. Following light chain V/J rearrangement, the newly formed heavy + light chain are transported to the cell surface, as the cell undergoes a second rearrangement checkpoint, assaying for functionality and self-reactivity. If the BCR is functionally expressed, it will provide “tonic” signal—serving as positive selection—and if it fails to receive activating BCR signal during this formative time, it will escape negative selection, progressing to the transitional stage and exiting the bone marrow. An excellent review of B cells development was recently published by Clark, et al. which covers these topics in more detail (Clark et al. 2013).
During development, if the nascent BCR recognizes a ligand in the bone marrow as Ag, the immature B cell will attempt to rearrange a new light chain to abrogate this reactivity (Gay et al. 1993; Tiegs et al. 1993). This is accomplished by re-induction of recombination-activating gene (RAG). This receptor editing is the most common type of
central tolerance in B cells (Halverson et al. 2004). If the immature B cell is capable of rearranging a functional and less-autoreactive BCR, it will progress to the periphery. Cells that are incapable of neutralizing this reactivity, or rearrange a non-functional receptor die by apoptosis, termed “clonal deletion” (Nemazee and Bürki 1989).
Peripheral Tolerance. The transitional cells that emerge from the bone marrow have cleared central tolerance, and are now subject to peripheral tolerance. It is the opinion of the author that peripheral tolerance is a vague mélange without clear boundaries between mechanisms. The first—which is not a programmed state of tolerance, and therefore perhaps not tolerance—is clonal ignorance. This is a state in which a BCR is specific for a self-Ag but either is excluded from encountering the Ag, or lacks BCR signal induction upon binding Ag, as can be the case with soluble, monovalent self-Ags. The net activation result is equivalent: ignorant B cells are phenotypically naïve. However, BCRs that are bound to soluble, monovalent Ag are receptor-occupied, and thus, because of steric blockade may have diminished signaling upon encountering multivalent form of the same epitope.
B cells bearing autoreactive BCR that bind self and signal in the periphery are tolerized by either “anergy” or split tolerance. As mentioned above, split tolerance refers to the inconcordant recognition of linked Ag by B and T cells. B cells that fail to receive sufficient T cell help following BCR signal go on to die by a form of activation-induced cell death (AICD). The cognate recognition of physically linked antigens (i.e. linked recognition) by both B and T cells is central to the theory of split tolerance, in which a consensus of T and B specificity increases the theoretical tolerance of the system (Zinkernagel et al. 1991). This requirement for the engagement of disparate tolerance programs exponentially decreases the likelihood of breaks in tolerance, as a B and T cell must both recognize and be non-tolerant to a linked set of Ags.
In many ways, anergy is quite similar to—or a form of—split tolerance, in that anergic cells are destined to die, but this can be somewhat overcome by T cell help (Eris et al. 1994). However, anergy is best defined by two requirements: BCR signaling must be chronically engaged with Ag, and the anergic B cell is hyporesponsive to subsequent activation. For a more detailed review of these and other tenets of anergy, and an overview of anergic mouse models, see Cambier, et al. (Cambier et al. 2007).
Evidence of chronic BCR engagement and signaling is found in the elevated and altered tyrosine phosphorylation profile, and elevated basal intracellular Ca2+ level in anergic B cells. This BCR signal is required for the maintenance of anergy, as decreased receptor occupancy in vivo, or disruption of cross-linking using monovalent hapten ex vivo, restores activation capacity (Gauld et al. 2005).
Our laboratory and others have previously shown that this chronic BCR signal causes activation of negative regulatory molecules, such as SH2 domain-containing inositol phosphatase-1 (SHIP-1) and Src-homology phosphatase-1 (SHP-1, PTPN6), a review of which can be found here (Yarkoni et al. 2010). SHIP-1 is an inositol phosphatase that cleaves phosphoinositol 3,4,5 tri-phosphate (PI(3,4,5,)P3) at the 5’
position resulting in the generation of PI(3,4)P2. Along with other lipid phosphatases
such as PTEN, this drives depletion of PI(3,4,5)P3, which is critical for organization of
activating BCR signaling components. On the protein phosphatase side, SHP-1 has many substrates, dephosphorylating multiple components of the BCR signal cascade, and critical to maintenance of tolerance in B cells (Pao et al. 2007).
The exact definition of anergic hyporesponsiveness is somewhat nebulous and controversial, but in general is characterized in vivo as a failure to mount an antibody response to vaccination. Phenotypically, anergic cells are described as having
decreased surface IgM, despite unchanged IgD, and reduced Ca2+ response to αIgM or Ag-induced signaling (Goodnow et al. 1988; Merrell et al. 2006).
B cells enriched for autoreactivity that share many of the characteristics of anergy with murine counterparts have been described in humans in multiple studies (Duty et al. 2009; Quách et al. 2011). One such circulating population was defined by the Wilson group by expression of cell surface markers CD19+, CD27-, IgD+, IgM-, dubbed “BND”
(Duty et al. 2009). We found that patients with new-onset T1D had drastic decreases in this compartment coincident with onset of diabetes (Smith et al. 2014). We showed that those BND cells that bound insulin were enriched in avidity to insulin and polyreactivity in
healthy humans, and decreased numbers of circulating BNDs were found among both
insulin-binding and non-insulin binding populations in new-onset T1D patients.
To summarize, anergy is initiated and maintained by chronically occupied, constitutively signaling BCR; this serves to downregulate surface IgM and engage a negative feedback network of protein and lipid phosphatases, which blunt subsequent restimulation and maintain unresponsiveness to homeostatic host conditions. This is in contrast to clonal ignorance, in which cells are naïve to Ag, due to lack of exposure, or binding to signaling-incompetent Ag.
B cell tolerance in insulin-specific B cells
In the context of previous studies using 125Tg, the term anergy was used to describe in vitro downregulation of surface IgM, impaired Ca2+ response, and reduced proliferation, but enhanced APC ability (Rojas et al. 2001; Acevedo-Suárez et al. 2005; Henry et al. 2009; Kendall et al. 2013). Additionally, in vivo 125Tg B cells are described as anergic due to their inability to mount an antibody response in response to insulin immunization. This finding is interesting in potentially informing the differences between
split tolerance and anergy: namely, wild-type animals mount detectable responses to immunization with adjuvanted insulin, whereas 125Tg are impaired (Rojas et al. 2001). Both of these have normal T cell repertoires, suggesting that the silencing is intrinsic to the B cell, rather than due to a deficit of T cell help. However, as mentioned above, circulating 125Tg cells have significant receptor occupancy in vivo and this may be sterically blocking the recognition of the immunogen. In addition to steric hindrance, or competition for recognition, a recent study using non-competing BCR-ligands found that monovalent engagement of the Ag-binding domain could inhibit subsequent BCR signaling through competitive and non-competitive crosslinking (Avalos et al. 2014).
B-T interactions following B cell activation
Following BCR activation in vivo, the B cell migrates to the border of the T cell zone, where it seeks CD4+ T cell help (Crotty 2015). This classically defined interaction has been extensively studied and serves to satisfy the requirements for activation of B and T lymphocytes. The BCR activation serves as “Signal 1” for the B cell, which internalizes the Ag and any physically linked components, processes and presents peptides derived from this Ag complex in the context of MHC class II.
As mentioned above, CD4+ T cells are critical effector cells in diabetes pathology. These lymphocytes express the co-receptor CD4, which binds to and restricts their reactivity to major histocompatibility complex (MHC) class II. They are conventionally activated by professional APCs—like B cells—which constitutively express MHC class II. The CD4+ T cell can engage the complex of MHC class II + peptide presented on the B cell, via the T cell antigen-receptor (TCR), and this TCR activation serves as Signal 1 for the T cells. Both B and T cells express activation
molecules upon Ag receptor stimulation, and the interactions of these costimulatory molecules provide “Signal 2” to the collaborating lymphocyte.
In addition to priming CD4+ T cells, B cells have been shown to be important in cross-presentation to CD8+ T cells in the context of diabetes (Mariño et al. 2012). In this study the authors demonstrate that expansion of CD8+ T cells specific for islet-specific glucose-6-phosphatase catalytic subunit-related protein (IGRP) is dependent on B cells. Further they show that this B cells activation of IGRP-specific CD8+ T cells is dependent on expression of MHC class I expression by B cells, as well as BCR specificity. The overall conclusion of this study is that autoAg-binding B cells are necessary and capable cross-presenters of pancreatic Ag in the context of diabetes.
In an excellent study of Ag presentation by B cells, Yan, et al. showed that B cells were important in the induction of autoimmunity (Yan et al. 2006). Yet again this effect was dependent on the BCR specificity, using a linked hapten-autoAg system. Together, a recurrent theme is found throughout the literature: autoAg-specific B cells are potent drivers of autoimmunity by presenting cognate Ag to T cells, further, that this priming is necessary and non-redundant to that performed by other professional APCs.
T cell recognition of insulin in autoimmune diabetes
Foundational studies of T cell recognition of insulin in the context of diabetes have been carried out by the laboratories of Drs. John Kappler, Kathryn Haskins, and Emil Unanue. Early studies by Unanue, et al., proposed that one method by which autoreactive T cells bypass central tolerance mechanisms is via recognition of non-traditional self-peptides in MHC class II (Pu et al. 2002; Mohan et al. 2010). Namely, canonical antigen presentation in class II occurs via endocytosis of cell-associated or protein Ag (acquisition), lysosomal degradation (processing), and presentation; and the
T cells recognizing the resultant epitopes are called “Type A”. However, they propose that an alternative, extracellular event can cause MHC class II-associated peptide to be displaced by new, different peptides—bypassing traditional acquisition and processing steps—calling T cells that recognize resultant epitopes “Type B”.
Dr. Kappler’s laboratory described requirements for register-binding that characterize diabetogenic TCR epitopes. They showed that all tested insulin-specific TCRs recognized B9-23 peptide bound to I-Ag7 in the third register (Stadinski et al. 2010). Expanding on characterization of TCR epitopes on insulin, they showed that insulin-specific T cells seemed to subset based on their sensitivity to a p8 position residue of the peptide. They used mimotopes of insulin, in which p6 and p9 were optimized for binding to I-Ag7, locking them in the third register. Comparing responses to either the WT glutamic acid at p8 (P8E) to a glycine (P8G), they found that some TCRs preferentially bound the P8G, and termed these Type 1; and some P8E, calling them Type 2 (Crawford et al. 2011).
To coalesce the above designations: Type 1 and Type B, along with Type 2 and Type A may have significant overlap, respectively. This is due to the observation that some of the TCR hybridomas classified as Type 1 preferentially recognize the P8G mimitope, and therefore are Type B (Crawford et al. 2011). Conversely, those that preferentially respond to P8E peptides (Type 2) had been defined by others as Type A. It is possible that differential processing and modified peptides in the pancreatic tissue are truncated, fused with other peptides, or otherwise modified such that the p8 glutamic acid is not present, allowing for activation of Type 1/B.
As this work focuses on B cells that bear BCR specificity for insulin, we would argue that they should most significantly have advantages in antigen presentation via the canonical, receptor-mediated endocytosis route of acquisition and subsequent
processing and presentation. This would likely specify a Type A/2 T cell response, as opposed to the passive MHC class II peptide displacement proposed by Unanue’s Type B model (Pu et al. 2002; Mohan et al. 2010). However, this does not rule out the possibility that disparate BCR engagement in the pancreas by intact insulin could activate the B cell, while in turn MHC class II peptides are being displaced by exogenous Type B/1 epitope peptides. This type of “unlinked” recognition—or bystander activation— does not seem likely in light of requirements for BCR autoantigen specificity stated above. In other words, if activated B cells are sufficient to present unlinked-Ag to T cells, how are VH281.NOD mice so protected from diabetes? It is likely that we have yet to uncover the link between BCR specificity as it relates to presentation to T cell types 1/B and 2/A. An interesting collaboration between Drs. Thomas & Unanue is currently ongoing; and in a poster presentation they showed that 125Tg cells were capable of activating Type 2 TCR hybridomas when incubated with intact insulin plus stimulating factors (Wan et al. 2015). This would belie the premise of Type B-recognition of tissue-derived passive peptides; instead elevating the insulin-specific B cell to a central role in presenting altered peptides in the context of autoimmunity.
Antigen presentation in vivo
As we are interested in the ability of insulin-specific B cells to present Ag to insulin-specific T cells, we proposed to visualize this interaction in situ using multiphoton microscopy. Specifically we used a 2-photon microscope, which allows for detection of fluorescence quite deep in tissue. This is due to the fact that the laser used to excite the fluorophore produces higher wavelength photons (lower energy) than the fluorophore, and thus the fluorophore is only excited upon effectively simultaneous absorption of energy from two photons. Due to this physical requirement, varying the frequency of the
laser pulse allows for a tight plane (Z) in which these 2-photon events may occur, thereby reducing much of the out-of-focus light produced in traditional light microscopy techniques.
By employing this technique, we could explant PLNs from animals that had received GFP+ TR-B cells and dye-labeled insulin-specific T cells and visualize their putative interactions dynamically. Previous studies have utilized multiphoton microscopy to visualize Ag presentation by B cells (Carrasco and Batista 2007; Phan et al. 2007). Multiphoton imaging allows cellular contacts to be measured in living tissue, and stable APC/T cell-contacts are associated with effective priming of CD4+ T cells (Mempel et al. 2004). We hypothesized that TR-B cells with high affinity for insulin might form more stable contacts with T cells, and attempted to visualize this synapse.
Aims of the thesis
Summary. In light of the presumed important role that B cells play in the development of autoimmune diabetes, we proposed to investigate the function of BCR affinity insulin, an established autoantigen in disease. We decided to use a novel approach in combining retrogenic and transgenic BCR light and heavy chains, respectively; in the hopes that this could serve as a platform for the rapid examination of new BCR clones, to assess the function of B cells bearing Ag- receptors with a variety of affinities for autoantigen.
Through expression of these immunoglobulin molecules as BCR, we proposed to evaluate the role of affinity for insulin in affecting B cell functionality. We hypothesize that affinity drives the mechanism of tolerance in the context of non-susceptible genetic background; furthermore, that BCR affinity for insulin determines B cell participation in disease in susceptible animals. Informed by the 125Tg model, we expected the VH125 +
VLA12 (A12) BCR to be receptor-occupied, due to its higher affinity. The ability to bind circulating insulin may drive anergy, and the genetic background of the B cell may affect this tolerized status. In contrast, the VH125 + VLEW6 (EW6) BCR may be clonally ignorant due to its micromolar affinity for a picomolar concentration ligand.
Extending our expectations to roles in disease, we predicted that A12—if insufficiently tolerized—would most effectively acquire antigen for activation and presentation. Furthermore, we reckoned that if EW6 was clonally ignorant, cells bearing this BCR would be unable to acquire antigen or become activated for effective presentation to diabetogenic CD4+ T cells.
Aim 1: Does BCR affinity for insulin affect mechanisms of B cell tolerance? We hypothesize that BCR affinity for insulin will play a critical role in selecting the mode of tolerance employed. Though central and peripheral tolerance mechanisms may be engaged, we propose to focus on the peripheral, due to the fact that circulating insulin-binding cells are common in mouse and man. We predict that low affinity insulin-insulin-binding B cells will be rendered tolerant by clonal ignorance in the periphery due to the low and pulsatile concentration of insulin in circulation. The high affinity cells, by contrast, may bind sufficient insulin to be rendered anergic, as seen with 125Tg B cells. This Aim will be interrogated in vitro by culturing TR-B cells in the presence of insulin and assessing their phenotype and functionality, and in vivo following adoptive transfer including detection of receptor occupancy. In addition to effects of BCR affinity on tolerance, the genetic background might also affect these parameters. Specifically, we hypothesize that TR-B cells derived from animals of non-diabetes susceptible background (i.e. B6) will have different modes and thresholds of tolerance as compared to those derived from the diabetes susceptible NOD background. Furthermore, host factors, such as the insulitis
found in the NOD pancreas microenvironment, may also play a role in the level of Ag exposure and therefore tolerance program engagement.
Aim 2: Does BCR affinity for insulin affect the B cells’ potentially pathogenic functions in diabetes? Ag presentation to autoreactive T cells is the primary method by which B cells are thought to be pathogenic in autoimmune diabetes. We hypothesize that the affinity of the BCR for insulin plays a critical function in subsequent presentation and priming of insulin-specific T cells. Examples of BCR-mediated functions that we propose to be affinity-dependent include acquisition, activation, and anergy. First, in order to present insulin, the B cell must acquire it for processing and loading into MHC. We predict that affinity of the anti-insulin BCR affects the B cells ability to acquire this autoAg in vivo. Second, insulin may be acquired by binding BCR in a non-signal transducing manner due its monovalent nature, thus altering activation of the B cell. We posit that high affinity B cells that are activated by insulin will have enhanced presentation capability, due to upregulation of costimulatory molecules. Third, if B cells are rendered tolerant by chronic or early exposure to insulin, this may reduce their subsequent ability to present Ag to insulin-specific T cells, or prime the T cells towards tolerance.
CHAPTER II
MATERIALS AND METHODS
The studies described herein were conducted in the laboratory of Dr. John Cambier at National Jewish Health and University of Colorado Denver. Except for as noted, all experiments were performed by Thomas A. Packard (TAP).
Animal models
Animals were bred, housed, and cared for in the specific pathogen-free facility at National Jewish Health Biological Resource Facility. All experimental use was subject to, and in accordance with, prior approval by the National Jewish Institutional Animal Care and Use Committee. NOD and C57BL/6 animals were obtained from The Jackson Laboratory (Bar Harbor, ME). Dr. James Thomas (Vanderbilt) generously provided VH125.NOD, VH281.NOD, VH125.B6, and VH281.B6 animals. The VH125.NOD were crossed to RAG2-/-.NOD (Jackson), and VH125.B6 crossed to RAG1-/-.B6 (Jackson). VH281.B6 animals were crossed to H2g7.B6. Animals were between age 8 and 12 weeks of age at the time of experiments.
Flow cytometric analysis
Many of the experiments carried out within use flow cytometric analysis or fluorescence-activated cell sorting (FACS). Flow cytometry was performed by TAP at the Flow Cytometry Core at National Jewish Health. FACS was performed by Shirley Sobus on the MoFlo XDP (Beckman Coulter) or Josh Loomis on the iCyt Synergy (Sony Biotechnology). The most commonly used flow cytometer was the LSR II (BD Biosciences), followed by the LSR Fortessa (BD Biosciences) and the CyAn ADP (Beckman Coulter).
Analyses of FCS data were performed using FlowJo software, versions 7 through 10 (Tree Star).
Graphing and statistical analyses
Data visualization and statistical analyses were performed using Prism version 5 (GraphPad). Calculation of EC50 uses non-linear regression dose-response analysis: log
(agonist) vs. normalized response (variable slope). Analyses of 2D off-rate uses dissociation – one phase exponential decay model to calculate rate constants, and 2-way ANOVA and Akaike Information Criterion-corrected (AICc) to compare the differences. Unpaired t-tests used to compare populations in column formats. Linearity regression calculated p values for non-zero slope. Throughout, asterisks used to denote p-values by indicated statistical test: * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001.
Recombinant Ig production
VL genes were originally cloned by Frank Conrad, via PCR, into an expression vector containing the human kappa constant region as previously described (Tiller et al. 2008). Additionally, VH125 was cloned into an expression vector containing the human IgG1 constant region, as previously described (Tiller et al. 2008). Human kidney endothelial (HEK) 293 cells were maintained in DMEM + L-glutamine + Pen/Strep and 10% FBS which had been pre-treated with Protein G to remove bovine Ig. These heavy and light chain plasmids were co-transfected into human endothelial HEK 293 cells at 1:1 molar ratio, using Effectene Transfection Reagent (Qiagen) according to manufacturer’s protocol. Five days post-transfection, the supernatant was removed and centrifuged at 2000 RCF and 0.22 μM filtered. The clarified supernatant was incubated with Protein G-sepharose beads (GE Life Sciences) overnight at 4°C. Protein-G beads
were precipitated with 200 RCF centrifugation, and loaded into a chromatography column (Bio-Rad). Bound Ig was eluted with a 100mM pH 1.9 glycine buffer. Eluted Ig was buffer-exchanged via 50 kD centrifugal filter unit (Amicon) into PBS. Purified Ig was analyzed by PAGE and Coomassie staining, and concentration determined by Nano Drop using preprogrammed settings for IgG molecular weight and extinction coefficient.
Following the original preparation by Frank Conrad, TAP cloned the VL + human kappa ORFs, and VH125 + human IgG1 into an expression vector obtained from the NIAID Vaccine Research Center, which had contained a broadly-neutralizing HIV Ab, VRC01 (Wu et al. 2010). This VRC vector had been used to produce recombinant Ig in high yield by the Cancer Center Protein Production core in their 293F system; and we proposed to utilize it for expression of our model recombinant Igs. After confirming successful replacement of the VRC01-Ab with Igs of interest, large quantities of plasmid were purified by Maxiprep (Qiagen) per the manufacturer’s protocol. The Cancer Center Protein Production core co-transfected a total of 0.5 mg of heavy and light chain plasmids in 1:1 molar ratio in 0.5 L of 293F grown in FreeStyle 293 Expression Medium (Life Technologies) using 293Fectin (Life Technologies). Five days post-transfection they returned the clarified supernatant to us for isolation of Ig.
As Ig was being isolated from a large volume of supernatant, 2.5 mL of Protein G-sepharose bead volume (GE Life Sciences) was loaded in a 15 mL chromatography column (Bio-Rad), and washed with PBS. 293F supernatant was flowed over the Protein G column at a rate of ~2 mL per minute at ambient pressure. Following Ig adsorption, the column was washed with >10x bead volume with PBS and eluted with 10x bead volume pH 1.9 100 mM glycine buffer. Eluted Ig was dialyzed in PBS overnight using a 75kD dialysis cassette (Pierce Thermo-Fisher). A12 Ig dialyzed in this manner precipitated in the pH 7.4 PBS following dialysis. Solubility testing found that A12 IgG1 could not be
