Peripheral Tolerance Checkpoints Imposed by Ubiquitous Expression Limit Antigen-Specific Responses under Strongly Immunogenic Conditions This information is current as of September 30, 2021. Jeremy F. Brooks, Peter R. Murphy, James E. M. Barber, James W. Wells and Raymond J. Steptoe J Immunol published online 24 July 2020 http://www.jimmunol.org/content/early/2020/07/23/jimmun ol.2000377 Downloaded from

Supplementary http://www.jimmunol.org/content/suppl/2020/07/23/jimmunol.200037

Material 7.DCSupplemental http://www.jimmunol.org/

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication by guest on September 30, 2021

*average

Subscription Information about subscribing to The Journal of is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2020 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Published July 24, 2020, doi:10.4049/jimmunol.2000377 The Journal of Immunology

Peripheral Tolerance Checkpoints Imposed by Ubiquitous Antigen Expression Limit Antigen-Specific B Cell Responses under Strongly Immunogenic Conditions

Jeremy F. Brooks, Peter R. Murphy, James E. M. Barber, James W. Wells, and Raymond J. Steptoe

A series of layered peripheral checkpoints maintain self-reactive B cells in an unresponsive state. production occurs when these checkpoints are breached; however, when and how this occurs is largely unknown. In particular, how self-reactive B cells are restrained during bystander inflammation in otherwise healthy individuals is poorly understood. A weakness has been the unavailability of methods capable of dissecting physiologically relevant B cell responses without the use of an engineered

BCR. Resolving this will provide insights that decipher how this process goes awry during or could be exploited Downloaded from for therapy. In this study, we use a strong adjuvant to provide bystander innate and adaptive signals that promote B cell responsiveness in conjunction with newly developed B cell detection tools to study in detail the ways that peripheral tolerance mechanisms limit the expansion and function of self-reactive B cells activated under these conditions. We show that although self-reactive B cells are recruited into the germinal center, their development does not proceed, possibly because of rapid counterselection. Consequently, differentiation of plasma cells is blunted, and Ab responses are transient and devoid of affinity

maturation. We propose this approach, and these tools can be more widely applied to track Ag-specific B cell responses to more http://www.jimmunol.org/ disease-relevant Ags, without the need for BCR transgenic mice, in settings where tolerance pathways are compromised or have been genetically manipulated to drive stronger insights into the biology underlying B cell–mediated autoimmunity. The Journal of Immunology, 2020, 205: 000–000.

entral B cell tolerance mediated by and excluded from entry into follicles (4, 5), have a curtailed lifespan receptor editing in the bone marrow (BM) is incomplete, (4), and are deleted from the recirculating repertoire. Cell surface C and self-reactive B cells escape into the peripheral rep- and membrane-bound self-, which are BCR-accessible, ertoire. To counter this, many overlapping and complementary are potent inducers of B cell peripheral tolerance (6, 7), whereas peripheral tolerance mechanisms have evolved to limit the re- soluble proteins are less effective (8). Aside from these B cell– by guest on September 30, 2021 sponsiveness of potentially self-reactive B cells. Peripheral B cell intrinsic mechanisms, the presence or absence of cognate Ag- tolerance normally arises when B cells encounter cognate Ag in a specific help is a major controlling influence on the way that leads to “unproductive” BCR signaling, with chronic Ag productivity of B cell responses. Because strong T cell tolerance recognition being most potent and likely essential for this. Chronic is present for soluble self-proteins (6, 9), B cell responses and Ag stimulation leads to BCR desensitization and anergy (1, 2). autoantibody production to soluble self-proteins may be limited Anergic B cells are unable to effectively solicit T cell help (3), are primarily by loss of T cell help (9). Despite the many mechanisms that control self-reactive B cells, in autoimmune diseases, central and/or peripheral tolerance mech- University of Queensland Diamantina Institute, The University of Queensland, Wool- anisms fail. Transcriptional and molecular analyses have revealed loongabba, Queensland 4102, Australia many pathways that control B cell responsiveness (10), and genetic ORCIDs: 0000-0002-0702-9784 (P.R.M.); 0000-0002-1588-5660 (J.E.M.B.); 0000- 0002-9618-6940 (J.W.W.); 0000-0002-6513-6406 (R.J.S.). studies have defined where control pathways or mechanisms can Received for publication April 9, 2020. Accepted for publication July 6, 2020. fail in autoimmune diseases (reviewed in Ref. 11). Other studies identify “accessory signals” capable of breaking peripheral B cell J.F.B. was supported by a Research Training Program Scholarship and Children’s Hospital Foundation Top-Up award (50209, RPCPHD0072017); J.W.W. was sup- tolerance or bypassing the checkpoints that normally control ported by a Fellowship from Perpetual Trustees and R.J.S. was supported by a B cell responsiveness, including innate signaling (TLRs), T cell University of Queensland Vice Chancellor’s Senior Research Fellowship. collaboration (CD40L), and input (BAFF) (12). Some J.F.B. and R.J.S. designed and performed experiments, analyzed data, and wrote the of these mechanisms are of relevance to (13) manuscript. P.R.M. and J.E.M.B. performed experiments. J.W.W. provided intellec- tual insights and edited the manuscript. and provide valuable insights into how autoimmune disease is ini- Address correspondence and reprint requests to Dr. Raymond J. Steptoe, Univer- tiated or perpetuated. Importantly, understanding how the respon- sity of Queensland Diamantina Institute, Translational Research Institute, Level 6, siveness of self-reactive B cells is controlled will provide important The University of Queensland, 37 Kent Street, Woolloongabba, QLD 4102, insights into how B cell tolerance can be engineered for therapeutic Australia. E-mail address: [email protected] goals. This would be particularly pertinent for approaches, such as The online version of this article contains supplemental material. gene therapy, in which Ags of interest to which tolerance is desired, Abbreviations used in this article: act.OVA, actin.mOVA; BM, bone marrow; Bmem, ; CSR, class switch recombination; dLN, draining lymph node; FDC, are expressed chronically or ubiquitously, enabling normal B cell follicular dendritic cell; GC, germinal center; HEL, hen egg lysozyme; OVA/CFA, tolerance mechanisms to be exploited. OVA emulsified in CFA; OVA/IFA, OVA emulsified in IFA; PB, plasmablast; PC, The limited availability of tools to accurately monitor naive ; Tg, transgenic. endogenous self-reactive B cells in non-BCR–engineered models Copyright Ó 2020 by The American Association of Immunologists, Inc. 0022-1767/20/$37.50 has been a challenge to progress in the understanding of B cell

www.jimmunol.org/cgi/doi/10.4049/jimmunol.2000377 2 B CELL TOLERANCE UNDER IMMUNOGENIC CONDITIONS tolerance. Most insights have been gained in models in which the cold PBS/FCS (2.5% v/v). BM was repeatedly passaged through a BCR repertoire is genetically manipulated, typically to increase 21-gauge needle to prepare a single-cell suspension, and erythrocytes were the clonal frequency, or for adoptive transfer, to enable Ag- lysed using ammonium chloride buffer. Ab and tetramer staining was performed as previously described (17), and all staining steps were per- specific B cells to be detected. However, this has the conse- formed on ice. For intracellular tetramer staining, cell suspensions were quence of potentially altering tolerance outcomes (14). We aim to stained for CD138–AF700, CD19–BV785, and FITC-conjugated lineage understand the robustness of peripheral tolerance and B cell mix. In some experiments, cells were also stained with Sca-1 PerCP-Cy5.5 checkpoints in a setting devoid of BCR manipulation and I (BioLegend). Cells were then fixed (Fixation Buffer, 420801; BioLegend) and permeabilized (Fix/Perm Buffer, 55473; BD Biosciences, San Jose, which a self-antigen is expressed ubiquitously at the cell surface. CA) and stained with tetramer (OVA-PE tetramer only) as described for A key goal is to understand whether the knowledge gained using surface tetramer staining (17). BCR transgenic (Tg) settings also stands for a setting in which The absolute cell number was enumerated using Flow-Count Fluoro- there is no genetic manipulation of BCR. We employ newly de- spheres (Beckman Coulter, Brea, CA) as previously described (18). Data veloped tools to dissect the fate and function of self-reactive were collected on an LSR Fortessa X-20 (BD Biosciences) and analyzed using FACSDiva Version 8 (BD Biosciences). B cells after administration of Ag with a potent adjuvant. In de- tailed analyses, we found, despite the provision of strong adju- ELISA vant signals, self-reactive B cell responses were transient, affinity For OVA-specific IgG1, IgG2b, IgG2c or IgM titers, plasma was serially maturation was restrained, and long-lived plasma cell (PC) and diluted (4-fold dilutions) and added to ELISA plates (MaxiSorp; Nunc, memory B cell (Bmem) formation was curtailed, thus demon- Waltham, MA) precoated with 10 mg/ml OVA (Grade V, A5503; Sigma- strating the robustness of peripheral B cell checkpoints under Aldrich). OVA-specific Ab was detected using biotinylated anti-mouse Abs (anti-IgG1 [RMG1-1] 0.02 mg/ml and anti-IgG2b [RMG2b-1] 0.05 mg/ml; these conditions in a non-autoimmune background. The system BioLegend; anti-IgG2c 0.2 mg/ml; Southern Biotech, Birmingham, AL; Downloaded from explored in this study has revealed meaningful insights into control anti-IgM 0.2 mg/ml; Thermo Fisher Scientific) followed by streptavidin- of B cell responsiveness under “normal” non-BCR–engineered conjugated HRP (1:2000 dilution, P030701-2; Dako, Carpinteria, CA) and conditions and would form the basis of a useful system in which visualized with TMB (421101; BioLegend). Absorbance was measured at to test the influence of genetic and environmental factors on the 450 nm and corrected for absorbance at 540 nm using a Multiskan Go spectrophotometer (Thermo Fisher Scientific). outcome of peripheral B cell tolerance.

Affinity ELISA http://www.jimmunol.org/ Materials and Methods The affinity ELISA was adapted from a previously published method (19). The absolute OVA-specific IgG1 concentration was determined for each Mice sample by ELISA, as described above, using OVA-14 (Sigma-Aldrich) Actin.mOVA (act.OVA) mice ubiquitously expressing membrane-bound as a standard and equalized across all samples. Samples were blocked 3 24 3 212 OVA under the control of a b-actin promoter (15) were maintained at with increasing concentrations of OVA (1 10 –1 10 M), left 3 24 the Translational Research Institute Biological Resources Facility (Brisbane, unblocked, or blocked with an irrelevant Ag (HEL, 1 10 M). Samples Australia) by heterozygous backcrossing to C57BL/6J/Arc (Animal were then transferred to ELISA plates precoated with OVA (1 mg/ml). The Resources Centre, Perth, Australia). Male and female, age-matched assay then proceeded as described above for OVA-specific IgG1 ELISA. (typically within 2–4 wk) Tg and non-Tg littermates 6–16 wk of age Statistical analysis were used. Studies were approved by The University of Queensland by guest on September 30, 2021 Ethics Committee. Comparison of means used a Student t test or ANOVAwith Tukey post hoc analysis for multiple comparisons for normally distributed data and a Immunizations Kruskal–Wallis and Mann–Whitney U test for nonnormally distributed OVA (Grade V, A5503; Sigma-Aldrich, St. Louis, MO) 2 mg/ml in PBS was data. Normality was determined using the Shapiro–Wilk test (GraphPad emulsified in CFA (CFA, F5881, or Difco, 23810; Sigma-Aldrich) or IFA Prism V8; GraphPad Software, San Diego, CA). For Ab titers, statistical (F5506; Sigma-Aldrich) using a Mini-BeadBeater (Biospec Products, testing was performed on log-transformed data. Correlations were per- Bartlesville, OK), and 100 ml was injected s.c. at the tail base. In some formed using a semilog linear regression (y-log axis, x-linear axis) or log- experiments, 100 mg of OVA was administered i.p. with 50 mg of anti- log regression (GraphPad Prism V8; GraphPad Software). CD40 (FGK-45; Bio X Cell, Lebanon, NH) on day 0, with additional anti-CD40 (i.p.) at day 2, 4, 6, and 8. In some experiments, 50 mgof Results OVAwasmixedwith20mg of Quil-A (Sigma-Aldrich) in PBS and in- jected s.c. at the tail base. For adjuvant-free conditions, OVA was either Provision of highly immunogenic stimuli partially overrides injected alone or depleted of LPS as described previously (16), and 100 mg peripheral B cell tolerance checkpoints was injected s.c. at the tail base. Where sham-immunized controls were Our aim was to understand how peripheral checkpoints might included, PBS and adjuvant were prepared and injected as described for OVA immunizations. control the response of B cells to immunogenic stimuli in mice that are rendered “tolerant” through constitutive expression of a model Flow cytometry self-antigen. We first tested the outcome of immunization of non- OVA-labeled tetramer preparation has been described (17). OVA and hen Tg controls and OVA-expressing act.OVA mice with OVA and a egg lysozyme (HEL) were biotinylated in-house and tetramerized with range of adjuvants that might carry diverse or different accessory streptavidin–PE (BioLegend, San Diego, CA or Prozyme, Hayward, CA), signals (Fig. 1A). For this we used OVA or OVA that had been streptavidin–allophycocyanin (BioLegend) or streptavidin–BV510 (BD Biosciences, San Jose, CA) for 2 h on ice. OVA tetramers were used at treated to remove LPS, each without any adjuvant, OVA emulsi- 14.6–21.2 nM (depending on batch) for staining. HEL tetramers were fied in IFA (OVA/IFA) or CFA (OVA/CFA). We also immunized used at 6.6–7.3 nM. IgM–PECy7 (RMM-1), IgD–BUV395 (11-26c.2a), with OVA in conjunction with serial injections of agonistic anti- CD4–FITC (GK1.5), CD8–FITC (53-6.7), CD11c–FITC (N418), B220– CD40 mAb or OVA with Quil-A. In non-Tg controls there was a allophycocyanin (RA3-B62), CD19–allophycocyanin (6D5), or CD19– graded response to immunization 14 d after immunization. OVA BV785 (6D5) and GL7–Pacific Blue (GL7) were purchased from BioLegend. CD138–AF700 (281-2), GL7–BV421 (GL7), IgG1–biotin (A85-1), and without adjuvant induced modest OVA-specific IgG1 levels, and, IgG2a,2b-biotin (R2-40) were purchased from BD Biosciences. NK1.1 notably, OVA-specific IgG1 production was abrogated by LPS (PK136), Gr1 (RB6-8C5), and CD11b (M1-70) were purchased from Bio depletion. OVA/CFA and OVA/IFA, however, led to the production of X Cell or grown in-house and labeled with FITC in-house. substantial levels OVA-specific IgG1 in non-Tg controls, (Fig. 1B). Draining lymph nodes (inguinal) from each immunized mouse were collected, pooled and pressed through a 70 mm strainer (Corning, Corning, Other immunization regimes led to intermediate levels of OVA- NY) to create single-cell suspensions. For analysis of BM, femurs and specific IgG1 production, similar to OVA without adjuvant. In con- tibiae from immunized mice were harvested, and BM was flushed with trast to non-Tg controls, in act.OVA mice, only OVA/CFA reliably The Journal of Immunology 3 led to production of detectable levels of OVA-specific IgG1, and Affinity maturation but not class switch recombination of the overall level was very low (Fig. 1B; ∼0.2 mg/ml). Therefore, a OVA-specific GC B cells is impaired in act.OVA mice strong adjuvant was required to elicit OVA-specific Ab production When self-reactive B cells are inappropriately recruited into a GC, in OVA-expressing mice, and even then, minimal amounts of affinity counterselection regulates the GC–PC axis to prevent the OVA-specific Ab were produced. Interestingly, in act.OVA mice, ultimate production of self-reactive Abs (20). Thus far, the data both OVA and LPS-free OVA without adjuvant were incapable of suggested that a modest but consistent number of OVA-specific eliciting OVA-specific IgG1, suggesting that OVA-specific B cells B cells were recruited into the GC reaction in act.OVA mice in act.OVA were largely unresponsive, and this manifests in this (Fig. 2). Therefore, we sought to identify whether the typical out- study as an insensitivity to the adjuvant effects of endotoxin rel- comes of a GC reaction, such as generation of -switched, ative to non-Tg mice. affinity-matured, and long-lived Ab responses, occurred. Analysis When Ab titers for IgM, IgG1, IgG2b, and IgG2c were plotted of IgG+ (IgG1, IgG2b), class-switched B cells revealed that their for individual mice immunized with OVA/CFA, non-Tg mice total number was relatively similar for OVA/CFA–immunized non- produced anti-OVA Abs of all isotypes (Fig. 1C). Notably act.OVA Tg and act.OVA mice (Supplemental Fig. 2B). However, the fre- mice produced significantly lower amounts of OVA-specific Abs quency of OVA-specific cells within the switched B cell population across all isotypes tested (Fig. 1C). Therefore, the strong adjuvant (Fig. 3A) was remarkably similar to that for OVA-specific cells CFA provides immunostimulatory signals that partially overcome within the GC B cell pool (Fig. 2E). When the total number was peripheral B cell checkpoints, potentially through diverse innate enumerated, OVA-specific IgG-switched B cells were around 5-fold signals or from bystander T cell responses activated by the my- (5.03-fold overall) more abundant in OVA/CFA–immunized non-Tg cobacterial Mycobacterium tuberculosis components of CFA, Downloaded from mice than act.OVA counterparts (Fig. 3B), and this was relatively which are absent from the similar but M. tuberculosis–free IFA. similar to the relative abundance in the OVA-specific GC com- To gain more insight into the nature of the Ab response induced partment (Fig. 2F, 4.27-fold overall). These data strongly indicate by OVA/CFA immunization, we monitored OVA-specific IgG1 in that, although fewer OVA-specific B cells appear to be recruited into the circulation. In non-Tg controls, OVA-specific IgG1 titers rose GC reactions in act.OVA mice, once recruited, they proceed through rapidly (detectable by day 7), reached a maximum 14 d after class switch recombination (CSR) with equivalent efficiency to non- immunization, and then remained stable at this level until at least http://www.jimmunol.org/ day 70 after immunization (Fig. 1D). By contrast, in act.OVA Tg mice. We further explored this by probing for OVA-specific mice, OVA-specific IgG1 accumulation was slower (not detectable B cells that may have failed to undergo CSR in act.OVA mice. at day 7), was much lower overall, and accumulation was tran- Looking within the bulk unswitched OVA-specific B cell pop- sient, peaking 14 d after immunization and diminishing to near ulations using the multitetramer approach (17), it was apparent that undetectable levels by day 35 after immunization (Fig. 1D). there was no major enrichment of unswitched OVA-specific B cells in act.OVA mice relative to non-Tg controls (Supplemental Fig. 2C, OVA-specific B cell activation and germinal center 2D), and consistent with this, in a single analysis, the majority of differentiation is limited when OVA is widely expressed OVA-specific GC B cells in OVA/CFA–immunized non-Tg and

To understand in more detail which specific components of re- act.OVA mice had undergone switching (Supplemental Fig. 2E). by guest on September 30, 2021 + sponsiveness to OVA were modulated by widespread OVA ex- Because we did not directly assess OVA-specific IgM B cells using + pression we analyzed B cell responses at the cellular level (Fig. 2A, GC markers but rather used just IgG switched GC B cells, we are Supplemental Fig. 1). After immunization, the overall B cell unable to comment on the minor fraction of unswitched OVA-specific population was expanded in both non-Tg and act.OVA mice to a GC B cells in wild-type and act.OVA mice. similar extent relative to naive controls (Supplemental Fig. 2A), To understand whether affinity maturation was modulated, we and germinal center (GC) B cell differentiation was present compared the relative affinity of circulating OVA-specific IgG1 in sham- (PBS/CFA) and OVA/CFA–immunized non-Tg and from non-Tg controls and act.OVA mice after OVA/CFA immu- act.OVA mice, although the scale may have been larger in nization using an ELISA-based competitive blocking assay. OVA/CFA–immunized non-Tg mice (Fig. 2B–D). Comparing the IC50, the affinity of OVA-specific IgG1 present in Given the responsiveness observed in both sham- and OVA/CFA– the circulation 14 d after immunization was reduced by approxi- immunized act.OVA mice, which may be directed at M. tuberculosis mately two orders of magnitude in act.OVA mice relative to non-Tg Ags in CFA, we next used tetramers to dissect the OVA-specific controls (Fig. 3C). This was most evident when comparing samples components of the B cell response to immunization. Based on Ab blocked with a molar concentration of OVA around the IC50 for 27 28 production (Fig. 1), we speculated that OVA-specific GC B cell OVA/CFA–immunized non-Tg controls (1 3 10 –1 3 10 M; differentiation may have occurred in act.OVA mice, despite Fig. 3D). We next probed affinity directly at the cellular level using widespread expression of OVA. In non-Tg mice, a large propor- an assay we developed, in which preincubation with titrated con- tion (∼25%) of GC B cells were OVA-specific cells, whereas they centrations of soluble OVA is used to block OVA tetramer binding to formed a significantly smaller fraction (∼5%) in act.OVA mice reveal differences in surface BCR affinity (17). Tetramer binding to (Fig. 2E) such that OVA-specific GC B cells were ∼4.3-fold more B cells from draining lymph nodes (dLNs) of OVA/CFA–immunized numerous in non-Tg mice than in act.OVA mice after OVA/CFA non-Tg controls was inhibited by ∼100-fold lower concentrations of immunization (Fig. 2F). Overall, failure of OVA-specific GC soluble OVA than for B cells from OVA/CFA–immunized act.OVA B cells to develop in sham-immunized mice and their reduced mice, indicating a ∼100-fold greater BCR affinity for the cells from 2 proportion among total GC B cells in act.OVA mice supports the non-Tg mice. Notably, 1 3 10 4 M soluble OVA effectively blocked premise that a large component of the GC response in both non-Tg tetramer binding to the affinity-matured OVA-specific BCR in OVA/ and act.OVA mice is the result of adjuvant-associated signals (or CFA–immunized non-Tg mice but only partially blocked binding to Ags). Nevertheless, the presence of OVA-specific B cells partici- the presumptively nonmatured BCR in act.OVA mice (Fig. 3E). 2 pating in ongoing GC responses in act.OVA mice confirms a Using 1 3 10 4 M soluble OVA block, there was less inhibition of partial breach in peripheral tolerance in the presence of immu- tetramer binding for B cells from OVA/CFA–immunized act.OVA nogenic adjuvant, possibly as a consequence of bystander T cell mice than for sham- or OVA/CFA–immunized non-Tg mice (Fig. 3F), help provided through adjuvant-associated signals. indicating the presence of low-affinity BCR on OVA-specific B cells 4 B CELL TOLERANCE UNDER IMMUNOGENIC CONDITIONS Downloaded from

FIGURE 1. OVA/CFA partially overrides peripheral B cell checkpoints. (A) Schematic of experimental design. Non-Tg and act.OVA mice were sham immunized (PBS/

CFA) or immunized with OVA/CFA, OVA/IFA, OVA/ http://www.jimmunol.org/ antiCD40, OVA/Quil-A, or OVA without adjuvant, and then OVA-specific IgG1 ELISAs were performed on plasma collected 14 d later (B and C) or at the timepoints indicated (D). (B) Titration curves for non-Tg and act.OVA mice immunized as indicated. (C) OVA-specific IgG1, IgG2b, IgG2c, and IgM titers for OVA/CFA–immunized and sham-immunized mice. (D) Time course of plasma OVA-specific IgG1 titers determined at the indicated timepoints. Data show mean 6 SD with number of rep- licates indicated (B), individual values pooled from two to by guest on September 30, 2021 six experiments with median and interquartile range shown (C), or mean 6 SD (n = 3–10). Mann–Whitney U test (C) or ANOVA with Tukey posttest was used (D) for log-transformed data. Per group: (B) n = 5–15 mice, (C) n = 4–10 mice, and (D) n = 3–10 mice.

in OVA/CFA–immunized act.OVA mice. The very low frequency OVA-specific B cells had contributed to GC reactions (e.g., of OVA-specific B cells in act.OVA mice precludes analysis of Fig. 2), affinity was not increased across the OVA-specific B cell their relative BCR affinity. This data suggested that, although population when OVA was ubiquitously expressed. The Journal of Immunology 5

FIGURE 2. OVA-specific GC responses are blunted in act.OVA mice. (A) Schematic of experimental de- sign. (B) Representative FACS plots showing gating for GC B cells within lin2 CD19+ B cells in pooled (in- guinal) dLN (top) and gating for OVA tetramer binding (bottom). (C) Total GC B cell frequency as a propor- tion of all CD19+ B cells. (D) Total GC B cell number per dLN. (E) OVA-specific GC B cells as a proportion of all GC B cells. (F) Total number of OVA-specific GC B cells per dLN. Data show representative FACS Downloaded from plots (B) or values for individual mice with mean 6 SD (C–F). Data are pooled from three to five experiments. ANOVA with Tukey posttest was used (C–F). Per group: (C–F) n = 4–15 mice. http://www.jimmunol.org/ by guest on September 30, 2021

As might be predicted, the OVA-specific Ab output was related Globally, the proportion (Supplemental Fig. 3B, 3C) and total to the frequency of OVA-specific B cells within the total GC B cell number (Fig. 4A–C) of PBs and PCs were similar in act.OVA and pool in act.OVA and non-Tg mice (Fig. 3G) and to the total GC non-Tg mice after OVA/CFA immunization. PBs, which retain size (Fig. 3H), suggesting that CFA-derived stimuli contribute surface Ig expression, could be directly surveyed by OVA tetramer substantially to OVA-specific Ab production in act.OVA mice but, (Fig. 4D), and as expected, OVA-specific PBs were almost com- on a per-cell basis, GC B cells in act.OVA mice appeared ham- pletely absent from the PB pool of OVA/CFA–immunized act.OVA pered in their capacity to produce OVA-specific IgG1. We propose mice relative to non-Tg counterparts (Fig. 4E, 4F). To identify OVA- that the GC checkpoint leads to counterselection of OVA-specific specific PCs, which do not retain surface Ig (Supplemental Fig. 3D), B cells that might have initially gained entry to the GC reaction as we used intracellular OVA tetramer staining. Validating this, OVA- a result of accessory signals from CFA. Therefore, not only is the specific PCs were present among lineage2veCD19varCD138+ cells in production of OVA-specific Abs curtailed, affinity maturation is the dLN following immunization (Fig. 4G) but not in the irrelevant substantially limited. Additionally, there was also no difference in CD1382velineage+ population (Fig. 4G). Specificity of detection was surface IgG levels (Supplemental Fig. 2F), suggesting that the high, as blocking with OVA prior to tetramer staining abolished reduction of OVA-specific cells in the GC in act.OVA mice was tetramer binding to PCs (Fig. 4G). When we applied this analysis not due to unsuccessful competition for Ag consequent to atten- method to OVA/CFA–immunized act.OVA mice, we found virtu- uated receptor expression but rather because of low BCR affinity ally no OVA-specific PCs within the PC pool in dLN (Fig. 4H, 4I) for OVA. and, significantly, in BM (Fig. 4J), the expected niche for long- lived PCs. In contrast, OVA-specific PCs were readily detected in Transient Ab production reflects failed PC differentiation in OVA/CFA–immunized non-Tg controls (Fig. 4I, 4J). Together, act.OVA mice these data demonstrate that cooperation between GC and PC Our data thus far indicate that OVA-specific B cell activation and differentiation checkpoints limit the lifespan and affinity of progression through GC differentiation and affinity maturation is OVA-specific Ab production. substantially blunted, but OVA-specific GC and IgG-switched B cells do in fact accumulate, albeit at significantly reduced lev- OVA-specific Bmem fail to develop in OVA/CFA–immunized els, after OVA/CFA immunization of act.OVA mice. Additionally, act.OVA mice the Ab kinetics in act.OVA mice is consistent with a reduced output Reactivation of Bmem mediates rapid Ab production upon Ag re- of OVA-specific PCs from GC reactions. Therefore, to gain insight encounter. Based on our data thus far, we proposed that differen- into whether the low levels of OVA-specific IgG that accumulates tiation of OVA-specific Bmem would be impaired in act.OVA mice. might reflect reduced OVA-specific GC activity in act.OVA mice, To determine whether OVA-specific Bmem arose as a consequence we next examined plasmablast (PB) and PC formation. of OVA/CFA immunization, we monitored the kinetics of Bmem 6 B CELL TOLERANCE UNDER IMMUNOGENIC CONDITIONS

formation (Supplemental Fig. 4A–E) and then analyzed dLN in detail 35 d postimmunization. We chose this timepoint because, among IgG-switched B cells in OVA/CFA–immunized non-Tg and act.OVA mice, the Bmem pool remained relatively stable from this timepoint onward, but the abundance of GC B cells had waned from their peak level at day 14 (Supplemental Fig. 4C). Addi- tionally, among IgG-switched B cells, the Bmem compartment was dominant over the GC B cell compartment at day 35 (Fig. 5A, Supplemental Fig. 4C–E). Although the overall Bmem compart- ments were relatively similar between non-Tg controls and act.OVA mice (Supplemental Fig. 4C, 4D), OVA-specific Bmem were remarkably rare in dLN of OVA/CFA–immunized act.OVA mice but not non-Tg controls (Fig. 5B–D). Although there had been consistent but blunted development of GC B cells in the dLN of OVA/CFA–immunized act.OVA mice 14 d after immunization (Fig. 2), at 35 d after immunization, OVA-specific GC B cells were virtually absent from dLN (Fig. 5E–G) and considerably less abundant relative to non-Tg counterparts than at day 14

(72-fold less abundant at day 35 compared with 4.3-fold less at Downloaded from day 14; Figs. 2, 5G). This suggests that the OVA-specific GC reaction in act.OVA mice diminishes much more quickly than in non-Tg controls, possibly as a result of limited OVA-specific B cell survival as a consequence of limited T cell help or through counterselection. Consistent with this, there is virtually no circulating OVA-specific

IgG1 detectable at this timepoint (Fig. 1D, Supplemental Fig. 4F), http://www.jimmunol.org/ indicating PB and PC differentiation has ceased in act.OVA mice. Together, these data confirm that when endogenous self-antigen is widely expressed a GC checkpoint controls Ag-specific GC B cell survival and differentiation fate and efficiently terminates OVA-specific Bmem and PC differentiation that arise as a con- sequence of bystander activation.

Discussion Many mechanisms combine to control development of unwanted, by guest on September 30, 2021 self-reactive B cell responses and subsequent production of pathogenic . That naive self-reactive B cells are relatively abundant in healthy individuals (21, 22) indicates the effectiveness of these mechanisms (12). Characterizing the mechanisms that maintain tolerance in the presence of activating FIGURE 3. OVA-specific CSR and affinity maturation are reduced in stimuli may give insights into the genesis of Ab-mediated auto- act.OVA mice. Non-Tg or act.OVA mice were sham immunized (PBS/ immune disease and, conversely, how tolerance approaches might CFA) or immunized with OVA/CFA and analyzed 14 d later. (A and B) be best deployed to achieve therapeutic goals. Studies of B cell + C D Accumulation of OVA-specific IgG -switched B cells in dLN. ( and ) tolerance and its failure are typically performed in conditions in Relative affinity of OVA-specific IgG1 Abs in plasma collected 14 d after which the BCR repertoire is genetically engineered to increase the immunization was compared using an ELISA-based assay. (E) Blockade of frequency of the Ag-specific B cells to be studied, but this can tetramer binding by titrated concentrations of soluble OVA for B cells from dLN taken from sham- or OVA/CFA–immunized non-Tg mice shown as alter tolerance outcomes. To understand the effectiveness of pe- proportion (percentage) of OVA tetramer+ cells relative to unblocked ripheral B cell control mechanisms, we studied self-reactive B cell samples. (F) Extent of tetramer binding inhibition by 1 3 1024 MOVA. responses in a non–BCR-Tg repertoire, using adjuvants that pro- (G and H) Correlation of OVA-specific IgG1 titer and GC B cell number. vide bystander help through innate and T cell input. We found that Data show values for individual mice with mean 6 SD (A and B), ab- only a highly potent adjuvant elicited self-antigen–specific Ab sorbance (percentage of maximum) versus concentration of OVA block production, and even under these conditions, Ab production was (C), percentage reduction in absorbance relative to unblocked samples at transient. We also found that, although CSR occurred, affinity the indicated concentrations of OVA for individual mice pooled from three maturation was blocked, and a post-GC checkpoint appeared to 6 D + experiments with mean SD ( ), percentage reduction in OVA tetramer contribute to limiting Ag-specific Ab production. cells for blocked relative to unblocked samples showing mean 6 SD (n =4 Based on studies of BCR-Tg mice, self-reactive B cells arriving per analysis point) (E), or individual mice showing mean 6 SD (F) and IgG1 titer versus OVA-specific (G) or total (H) GC B cell frequency for in the periphery have effectively avoided deletion, and/or receptor individual mice. Data are representative (C) or pooled from two to five editing has insufficiently quenched the capacity to bind self. This experiments. Kruskal–Wallis (A and B), ANOVA with Tukey posttest (F), represents only a minor fraction of the peripheral B cell pool, but t test (D and E), log/log (G), or log/linear regression (H) were used. Per residual self-reactivity must be contained by additional check- group: (A) n = 4–14 mice, (D) n = 5–6 mice or 4–8 mice, (F) n = 5–11 points. The stringency of these additional checkpoints has been mice, (G) and (H), n = 8–9 mice. tested using environmental perturbations (microbiota, infection) or genetic polymorphisms (autoimmune susceptibility genes). Even in non-autoimmune–prone genetic backgrounds, some aspects of deletion, developmental arrest, and anergy may be overcome by The Journal of Immunology 7 Downloaded from http://www.jimmunol.org/

FIGURE 5. OVA-specific Bmem fail to develop in act.OVA mice after immunization with immunogenic adjuvant. Non-Tg and act.OVA mice were immunized with OVA/CFA and lymph nodes were analyzed at the indicated timepoints. (A) Representative FACS plots showing GC (GL7+CD382) and Bmem (CD38+GL72), gated as per Supplemental Fig. 1. (B) Representative FACS plots showing OVA tetramer staining of Bmem at C

day 35, gated as per Supplemental Fig. 1C. ( ) OVA-specific Bmem as a by guest on September 30, 2021 proportion of all Bmem in dLN at day 35. (D) Total OVA-specific Bmem in dLN at day 35. (E) Representative FACS plots showing tetramer staining of GC B cells at day 35, gated as per Supplemental Fig. 1. (F) OVA-specific GC B cells as a proportion of all GC B cells at day 35. (G) Total OVA-specific GC B cells in dLN at day 35. Data show representative FACS plots (A, B,and E)ormean6 SD or individual mice with mean 6 SD (C, D, F,andG). Data are pooled from two experiments with two to three mice per group. Unpaired t test was used (C, D, F,andG). Per group: (C–G) n = five mice.

augmenting antiapoptotic pathways (23) or by provision of T cell help (24) and innate stimuli (25). Although these have been in- formative, the typical readouts for these experiments are techni- cally limited and include the tracking of BCR-Tg cells and/or indirect probing for the frequency of self-reactive BCRs by FIGURE 4. OVA-specific PC responses are blunted in act.OVA mice. cloning and screening them as rAbs by ELISA (22). Whether Non-Tg and act.OVA mice were sham immunized (PBS/CFA) or immu- these readouts directly reflect ongoing self-reactivity by a physi- nized with OVA/CFA, and dLN (pooled inguinal lymph node) and BM ological repertoire remains unclear. In this study, we provide two were analyzed 14 d later. (A) Representative gating for PB and PC based important advances. First, we use sensitive and specific tetramer 2 var on CD19 and CD138 expression within lin /CD19 cells as shown in labeling to track the activation and progression of rare but en- Supplemental Fig. 1. (B) Total PB number in dLN. (C) Total number of PCs in dogenous B cells specific for a model self-antigen through the GC D dLN. ( ) Representative surface OVA tetramer staining of PBs in dLN with reaction. Second, we couple analyses of Ab production and af- no OVA blocking. (E) Proportion and total number of OVA-specific PBs finity to Ag-specific PC differentiation to characterize the fate of within all PBs in dLN. (F) Total number of OVA-specific PBs in dLN (G) Representative intracellular OVA tetramer staining of the indicated pop- GC-derived self-reactive B cells. Overall, our data are consistent ulations with and without OVA block (gated on CD19+ cells). (H) Intracellular with the well-appreciated role of tight regulation of GC entry and OVA tetramer staining of PCs in dLN from non-Tg and act.OVA mice with no ongoing participation but highlight a novel checkpoint post-GC OVA blocking. (I and J) Frequency of OVA-specific PCs in dLN and BM. that limits long-lived Ab production. Data show representative FACS plots (A, D, G,andH) or individual mice with Chronic sensing of self-antigen in the periphery enforces an mean 6 SD (B, C, E, F, I,andJ). Data are pooled from two to three experi- anergy transcriptional program (10, 26) and the downregulation of ments. ANOVAwith Tukey posttest (B, C,andE), a Kruskal–Wallis test (F), or surface IgM (22). The two key goals of this are 1) dampening a Mann–Whitney U test was used (I and J). Per group: (B) n =4–11mice,(C) signal transduction via the BCR upon self-ligation and 2) reducing E F I J n = 4–7 mice, ( and ) n = 4–11 mice, and ( and ) n = 4–6 mice. the probability of encountering T cell help. Because GC entry is 8 B CELL TOLERANCE UNDER IMMUNOGENIC CONDITIONS governed by T cell–B cell interaction, anergic B cells are ana- actually augment negative selection in the GC when T cell help is tomically alienated from follicles (4) and fail to receive necessary limiting because the effective concentration of local OVA Ag has stimulation through surface coreceptors that enhance BCR sig- been increased, and soluble Ag can drive apoptosis in the GC (37). naling (3, 5, 24). One unresolved aspect is to what extent in- The rapid but transient kinetics and low affinity observed for flammation and noncognate T cell help can drive promiscuous anti-OVA Abs resemble that described for autoantibodies from recruitment of autoreactive B cells into a GC reaction. It is well extrafollicular reactions that are relatively T cell–independent, but documented that viral infection can lead to collateral autoanti- for which innate signals (e.g., microbe-derived LPS) provide bodies (27) and, based on analyses of , that support (38). Indeed, innate signals, particularly those mediated this can proceed through a canonical GC (28). Additionally, Abs through TLR7 and TLR9 are sufficient to promote extrafollicular that are generated against pathogens are frequently polyreactive B cell responses and autoantibody production (39–41) as well as and cross-react with endogenous self-antigens (21). An oft-picked T cell–independent CSR (42). Additionally, extrafollicular B cell example is that Abs neutralizing HIV are derived from germline activation and PB differentiation may be a key site of autoanti- self-reactive precursors (28), indicating that self-reactive B cell body production, at least in rheumatoid arthritis (43, 44) and recruitment into a GC is a rare but biologically meaningful systemic lupus erythematosus (45), driven by self-antigens that are event. In a recent study (29), infection with g-herpesvirus elicited complexed with local TLR ligands such as DNA or RNA (40, 41). affinity-matured autoantibodies that did not have germline autor- In this study, we propose that various M. tuberculosis–associated eactivity, leading to the inference that self-binding had been ac- TLR signals (predominantly TLR2/4/9) serve a similar function quired and positively selected inside the GC. Our findings in this and promote activation of peripheral anergic OVA-specific B cells, study using CFA agree in that total GC B cell frequency correlates transiently breaking their state of functional unresponsiveness. Downloaded from with OVA-specific Ab output. We add further support to this by Interestingly, although extrafollicular somatic hypermutation (43) showing directly that self-reactive B cells can occupy the GC can be driven by bacteria-associated (38, 44) or other TLR signals, following CFA immunization, albeit in a truncated manner, and it was not apparent in this study. Although the majority of studies that this is related to the magnitude of Ab production. characterizing extrafollicular reactions are based on soluble self- Survival and progression in the GC are driven in a Darwinian antigen (43), the same self-antigen, when membrane-bound as

fashion by competition for limiting T cell help. Many, if not all, employed in this study, is more tolerogenic (6, 9) and may provide http://www.jimmunol.org/ possible fates for GC B cells (cell cycle, deletion, memory and PC more stringent control. In any case, both GC and extrafollicular differentiation) require T cell input (30). This is, in part, because checkpoints operated to limit the production of self-reactive Abs GC B cell signaling is intrinsically rewired to depend on T cell in the face of signals that might promote B cell activation. help (30). Because the OVA-specific T cell compartment is tol- The discrete stimuli that license self-reactive B cells are largely erant (15), we suggest that bystander signals from M. tuberculosis unknown in a polyclonal setting. These have instead been deduced in CFA promote early self-reactive GC B cell responses at the time using BCR-Tg models, which are useful tools to track defined of immunization that were terminated through either a failure to populations of Ag-specific cells; however, qualitative and quan- compete in ongoing GC or through negative selection or both. titative variations in tolerance outcomes have been reported when Although T cell tolerance is robust in act.OVA mice (31), we the BCR repertoire is genetically engineered. Rather, in this study, by guest on September 30, 2021 cannot rule out the slim possibility that CFA also permitted a partial we chose to engineer the Ag repertoire, with the goal that this breach in OVA-specific T cell tolerance. It is likely that providing an would minimize changes associated with BCR engineering. En- artificial source of cognate T cell help, either by adoptive transfer of gineering of the Ag repertoire has been explored extensively in primed OVA-specific CD4+ T cells or by covalently linking OVA to a studies of T cell tolerance, and it is apparent that this approach is foreign Ag, as others have done (32), self-reactive GC B cell responses not without caveats (46). Of course, ideally, studies would use may be either boosted or even rescued from deletion. Perhaps more unmanipulated BCR repertoires and bona fide autoantigens. useful, genetically enhancing positive signaling strength by modulat- Limiting this is that the frequency of autoantigen-specific B cells ing calcium flux (33) or removing negative regulation by inactivating is extremely low, and generating controls devoid of autoantigen inhibitory phosphatases (1) would reveal B cell–intrinsic contributions expression can be challenging because of the physiological role in addition to bystander T cell help, as shown in this study, that would of many autoantigens. We believe that development of sensitive be sufficient to drive bona fide autoimmunity. This would be partic- tetramer-based approaches (17) is a step forward, but some auto- ularly insightful in cases that autoimmune susceptibility genes syn- antigens may not be amenable to this, although adapted tetramer ergize with infection to initiate autoantibody production (34). approaches for DNA and RNA, for example, have recently become Because OVA-specific GC B cells were not present at later available (47, 48). An additional consideration for models is that timepoints and had not differentiated to Bmem or long-lived PCs, responses to autoantigens involve contributions from the tissue we conclude that they were shunted toward deletion within the GC. microenvironment (immune contexture, structural architecture), Ag This is most likely because of the way self-antigen is encountered expression pattern (membrane-bound or soluble, cell-type restricted), by B cells within the GC of act.OVA mice. For example, immune and genetic susceptibility, and these may not be recapitulated in all complexes or opsonized Ag acquired by follicular dendritic cells model systems. Notwithstanding such considerations, the findings in (FDC) is instrumental in driving affinity maturation during a GC this study provide important validation for several B cell controls reaction by providing a source of Ag that B cells use to elicit described elsewhere in other polyclonal and BCR-Tg settings. T cell help (35). In contrast, expression of a self-antigen by FDC The findings in this study provide some insight into the control of prevents cognate affinity maturation and potentially drives dele- Ab production against self-antigens. They implicate a mechanism tion of cognate GC B cells (36). Because OVA is ubiquitously by which switched (low-affinity) self-reactive Abs may arise in a expressed in act.OVA mice, we propose transgenic expression of polyclonal repertoire when coinfection occurs in the presence of lo- OVA within the GC, perhaps in FDC, was at least one component cally released self-antigens. Our findings show that some mechanisms critical for curtailing the OVA-specific GC response. Because few defined in BCR-Tg models hold in this study in this setting that OVA-specific B cells become activated in act.OVA mice, we also reflects a normal physiological polyclonal repertoire and refine our speculate that Ag availability following immunization is higher, as understanding of the control of B cell responses. We propose that uptake by OVA-binding B cells would be much less. This may our studies, performed in this work in a non-autoimmune–prone The Journal of Immunology 9 setting, along with tools we have developed (17) form the basis of 21. Wardemann, H., S. Yurasov, A. Schaefer, J. W. Young, E. Meffre, and M. C. Nussenzweig. 2003. Predominant autoantibody production by early a valuable system for studying genetic susceptibility alleles or the human B cell precursors. Science 301: 1374–1377. role of individual molecules in peripheral control of B cell re- 22. Watanabe, A., K. Y. Su, M. Kuraoka, G. Yang, A. E. Reynolds, A. G. Schmidt, sponses. Our data, which imply a critical role for T cell help, S. C. Harrison, B. F. Haynes, E. W. St Clair, and G. Kelsoe. 2019. Self-tolerance curtails the B cell repertoire to microbial . JCI Insight 4: e122551. indicate that therapeutic approaches that directly modulate B cells 23. Oliver, P. M., T. Vass, J. Kappler, and P. Marrack. 2006. Loss of the proapoptotic and T cells and that counter genetic susceptibilities, such as he- protein, Bim, breaks B cell anergy. J. Exp. Med. 203: 731–741. matopoietic stem cell–based gene therapy (49) are likely to be 24. Rathmell, J. C., S. Fournier, B. C. Weintraub, J. P. Allison, and C. C. Goodnow. 1998. Repression of B7.2 on self-reactive B cells is essential to prevent proliferation highly effective at modulating unwanted Ab production. and allow Fas-mediated deletion by CD4(+) T cells. J. Exp. Med. 188: 651–659. 25. Hannum, L. G., D. Ni, A. M. Haberman, M. G. Weigert, and M. J. Shlomchik. Acknowledgments 1996. A disease-related rheumatoid factor autoantibody is not tolerized in a normal mouse: implications for the origins of autoantibodies in autoimmune This research was carried out at the Translational Research Institute. The disease. J. Exp. Med. 184: 1269–1278. Translational Research Institute is supported by a grant from the Australian 26.Sabouri,Z.,S.Perotti,E.Spierings,P.Humburg,M.Yabas,H.Bergmann, Government. We acknowledge the Translational Research Institute core K. Horikawa, C. Roots, S. Lambe, C. Young, et al. 2016. IgD attenuates the IgM- facilities, in particular the Flow Cytometry Core Facility for outstanding induced anergy response in transitional and mature B cells. Nat. Commun. 7: 13381. 27. Getts, D. R., E. M. L. Chastain, R. L. Terry, and S. D. Miller. 2013. Virus in- technical support. fection, antiviral , and autoimmunity. Immunol. Rev. 255: 197–209. 28. Schroeder, K. M., A. Agazio, and R. M. Torres. 2017. Immunological tolerance as Disclosures a barrier to protective HIV . Curr. Opin. Immunol. 47: 26–34. 29. Sakakibara, S., T. Yasui, H. Jinzai, K. O’donnell, C.-Y. Tsai, T. Minamitani, The authors have no financial conflicts of interest. K. Takeda, G. T. Belz, D. M. Tarlinton, and H. Kikutani. 2020. Self-reactive and polyreactive B cells are generated and selected in the germinal center during g-herpesvirus infection. Int. Immunol. 32: 27–38. Downloaded from References 30. Shlomchik, M. J., W. Luo, and F. Weisel. 2019. Linking signaling and selection in the germinal center. Immunol. Rev. 288: 49–63. 1. Gauld, S. B., R. J. Benschop, K. T. Merrell, and J. C. Cambier. 2005. Mainte- 31. Coleman, M. A., J. A. Bridge, S. W. Lane, C. M. Dixon, G. R. Hill, J. W. Wells, nance of B cell anergy requires constant antigen receptor occupancy and sig- R. Thomas, and R. J. Steptoe. 2013. Tolerance induction with gene-modified stem naling. Nat. Immunol. 6: 1160–1167. cells and immune-preserving conditioning in primed mice: restricting antigen to 2. Goodnow, C. C., R. Brink, and E. Adams. 1991. Breakdown of self-tolerance in anergic B . Nature 352: 532–536. differentiated antigen-presenting cells permits efficacy. Blood 121: 1049–1058. 3. Cooke, M. P., A. W. Heath, K. M. Shokat, Y. Zeng, F. D. Finkelman, 32. Turner, J. S., M. Marthi, Z. L. Benet, and I. Grigorova. 2017. Transiently P. S. Linsley, M. Howard, and C. C. Goodnow. 1994. Immunoglobulin signal antigen-primed B cells return to naive-like state in absence of T-cell help. Nat. http://www.jimmunol.org/ transduction guides the specificity of B cell-T cell interactions and is blocked in Commun. 8: 15072. tolerant self-reactive B cells. J. Exp. Med. 179: 425–438. 33. Healy, J. I., R. E. Dolmetsch, L. A. Timmerman, J. G. Cyster, M. L. Thomas, 4. Cyster, J. G., S. B. Hartley, and C. C. Goodnow. 1994. Competition for follicular G. R. Crabtree, R. S. Lewis, and C. C. Goodnow. 1997. Different nuclear signals niches excludes self-reactive cells from the recirculating B-cell repertoire. Na- are activated by the B cell receptor during positive versus negative signaling. ture 371: 389–395. Immunity 6: 419–428. 5. Cyster, J. G., and C. C. Goodnow. 1995. Antigen-induced exclusion from fol- 34. Kivity, S., N. Agmon-Levin, M. Blank, and Y. Shoenfeld. 2009. Infections and licles and anergy are separate and complementary processes that influence pe- autoimmunity--friends or foes? Trends Immunol. 30: 409–414. ripheral B cell fate. Immunity 3: 691–701. 35. Heesters, B. A., R. C. Myers, and M. C. Carroll. 2014. Follicular dendritic cells: 6. Hartley, S. B., J. Crosbie, R. Brink, A. B. Kantor, A. Basten, and C. C. Goodnow. dynamic antigen libraries. Nat. Rev. Immunol. 14: 495–504. 1991. Elimination from peripheral lymphoid tissues of self-reactive 36. Chan, T. D., K. Wood, J. R. Hermes, D. Butt, C. J. Jolly, A. Basten, and R. Brink. B lymphocytes recognizing membrane-bound antigens. Nature 353: 765–769. 2012. Elimination of germinal-center-derived self-reactive B cells is governed by 7. Yau, I. W., M. H. Cato, J. Jellusova, T. Hurtado de Mendoza, R. Brink, and the location and concentration of self-antigen. Immunity 37: 893–904. by guest on September 30, 2021 R. C. Rickert. 2013. Censoring of self-reactive B cells by follicular dendritic 37. Han, S., B. Zheng, J. Dal Porto, and G. Kelsoe. 1995. In situ studies of the cell-displayed self-antigen. J. Immunol. 191: 1082–1090. primary immune response to (4-hydroxy-3-nitrophenyl)acetyl. IV. Affinity- 8. Adelstein, S., H. Pritchard-Briscoe, T. A. Anderson, J. Crosbie, G. Gammon, dependent, antigen-driven B cell apoptosis in germinal centers as a mecha- R. H. Loblay, A. Basten, and C. C. Goodnow. 1991. Induction of self-tolerance in nism for maintaining self-tolerance. J. Exp. Med. 182: 1635–1644. T cells but not B cells of transgenic mice expressing little self antigen. Science 38. Di Niro, R., S. J. Lee, J. A. Vander Heiden, R. A. Elsner, N. Trivedi, 251: 1223–1225. J. M. Bannock, N. T. Gupta, S. H. Kleinstein, F. Vigneault, T. J. Gilbert, et al. 9. Taylor, J. J., R. J. Martinez, P. J. Titcombe, L. O. Barsness, S. R. Thomas, 2015. Salmonella infection drives promiscuous B cell activation followed by N. Zhang, S. D. Katzman, M. K. Jenkins, and D. L. Mueller. 2012. Deletion and extrafollicular affinity maturation. Immunity 43: 120–131. anergy of polyclonal B cells specific for ubiquitous membrane-bound self- 39. Green, N. M., and A. Marshak-Rothstein. 2011. Toll-like receptor driven B cell antigen. J. Exp. Med. 209: 2065–2077. activation in the induction of systemic autoimmunity. Semin. Immunol. 23: 106–112. 10. Glynne, R., S. Akkaraju, J. I. Healy, J. Rayner, C. C. Goodnow, and D. H. Mack. 2000. 40. Leadbetter, E. A., I. R. Rifkin, A. M. Hohlbaum, B. C. Beaudette, M. J. Shlomchik, How self-tolerance and the immunosuppressive drug FK506 prevent B-cell mito- and A. Marshak-Rothstein. 2002. Chromatin-IgG complexes activate B cells by genesis. [Published erratum appears in 2000 Nature 407: 413.] Nature 403: 672–676. dual engagement of IgM and toll-like receptors. Nature 416: 603–607. 11. Marson, A., W. J. Housley, and D. A. Hafler. 2015. Genetic basis of autoim- 41. Lau, C. M., C. Broughton, A. S. Tabor, S. Akira, R. A. Flavell, M. J. Mamula, munity. J. Clin. Invest. 125: 2234–2241. S. R. Christensen, M. J. Shlomchik, G. A. Viglianti, I. R. Rifkin, and A. Marshak- 12. Shlomchik, M. J. 2008. Sites and stages of autoreactive B cell activation and Rothstein. 2005. RNA-associated autoantigens activate B cells by combined B cell regulation. Immunity 28: 18–28. antigen receptor/Toll-like receptor 7 engagement. J. Exp. Med. 202: 1171–1177. 13. Steri, M., V. Orru`, M. L. Idda, M. Pitzalis, M. Pala, I. Zara, C. Sidore, V. Faa`, 42. He, B., X. Qiao, and A. Cerutti. 2004. CpG DNA induces IgG class switch DNA M. Floris, M. Deiana, et al. 2017. Overexpression of the BAFF and recombination by activating human B cells through an innate pathway that re- autoimmunity risk. N. Engl. J. Med. 376: 1615–1626. quires TLR9 and cooperates with IL-10. J. Immunol. 173: 4479–4491. 14. Cambier, J. C., S. B. Gauld, K. T. Merrell, and B. J. Vilen. 2007. B-cell anergy: 43. William, J., C. Euler, S. Christensen, and M. J. Shlomchik. 2002. Evolution of from transgenic models to naturally occurring anergic B cells? Nat. Rev. Immunol. autoantibody responses via somatic hypermutation outside of germinal centers. 7: 633–643. Science 297: 2066–2070. 15. Ehst, B. D., E. Ingulli, and M. K. Jenkins. 2003. Development of a novel 44. William, J., C. Euler, and M. J. Shlomchik. 2005. Short-lived plasmablasts transgenic mouse for the study of interactions between CD4 and CD8 T cells dominate the early spontaneous rheumatoid factor response: differentiation during graft rejection. Am. J. Transplant. 3: 1355–1362. pathways, hypermutating cell types, and affinity maturation outside the germinal 16. Al-Kouba, J., A. N. Wilkinson, M. R. Starkey, R. Rudraraju, R. B. Werder, center. J. Immunol. 174: 6879–6887. X. Liu, S. C. Law, J. C. Horvat, J. F. Brooks, G. R. Hill, et al. 2017. - 45. Suurmond, J., and B. Diamond. 2015. Autoantibodies in systemic autoimmune encoding bone marrow transfer inactivates allergic T cell responses, alleviating diseases: specificity and pathogenicity. J. Clin. Invest. 125: 2194–2202. airway inflammation. JCI Insight 2: e85742. 46. Miyagawa, F., J. Gutermuth, H. Zhang, and S. I. Katz. 2010. The use of mouse 17. Brooks, J. F., X. Liu, J. M. Davies, J. W. Wells, and R. J. Steptoe. 2018. models to better understand mechanisms of autoimmunity and tolerance. Tetramer-based identification of naı¨ve antigen-specific B cells within a poly- J. Autoimmun. 35: 192–198. clonal repertoire. Eur. J. Immunol. 48: 1251–1254. 47. Suurmond, J., Y. Atisha-Fregoso, E. Marasco, A. N. Barlev, N. Ahmed, S. A. Calderon, 18. Steptoe, R. J., S. Stankovic, S. Lopaticki, L. K. Jones, L. C. Harrison, and M. Y. Wong, M. C. Mackay, C. Aranow, and B. Diamond. 2019. Loss of an IgG plasma G. Morahan. 2004. Persistence of recipient lymphocytes in NOD mice after ir- cell checkpoint in patients with lupus. J. Clin. Immunol. 143: 1586–1597. radiation and bone marrow transplantation. J. Autoimmun. 22: 131–138. 48. Suurmond, J., Y. Atisha-Fregoso, A. N. Barlev, S. A. Calderon, M. C. Mackay, 19. Macdonald, R. A., C. S. Hosking, and C. L. Jones. 1988. The measurement of C. Aranow, and B. Diamond. 2019. Patterns of ANA+ B cells for SLE patient relative affinity by ELISA using thiocyanate elution. J. Immunol. stratification. JCI Insight 4: e127885. Methods 106: 191–194. 49. Brooks, J. F., J. M. Davies, J. W. Wells, and R. J. Steptoe. 2018. Re-educating 20. Brink, R., and T. G. Phan. 2018. Self-reactive B cells in the germinal center immunity in respiratory : the potential for hematopoietic stem cell- reaction. Annu. Rev. Immunol. 36: 339–357. mediated gene therapy. J. Mol. Med. (Berl.) 96: 21–30.