1. Immunology and Inflammation
Download icon

IgM and IgD B cell receptors differentially respond to endogenous antigens and control B cell fate

  1. Mark Noviski
  2. James L Mueller
  3. Anne Satterthwaite
  4. Lee Ann Garrett-Sinha
  5. Frank Brombacher
  6. Julie Zikherman  Is a corresponding author
  1. University of California San Francisco, United States
  2. UT Southwestern Medical Center, United States
  3. University at Buffalo, The State University of New York, United States
  4. International Center for Genetic Engineering and Biotechnology (ICGEB), South Africa
  5. University of Cape Town & Medical Research Council (SAMRC), South Africa
Research Article
  • Cited 4
  • Views 3,517
  • Annotations
Cite this article as: eLife 2018;7:e35074 doi: 10.7554/eLife.35074

Abstract

Naive B cells co-express two BCR isotypes, IgM and IgD, with identical antigen-binding domains but distinct constant regions. IgM but not IgD is downregulated on autoreactive B cells. Because these isotypes are presumed to be redundant, it is unknown how this could impose tolerance. We introduced the Nur77-eGFP reporter of BCR signaling into mice that express each BCR isotype alone. Despite signaling strongly in vitro, IgD is less sensitive than IgM to endogenous antigen in vivo and developmental fate decisions are skewed accordingly. IgD-only Lyn−/− B cells cannot generate autoantibodies and short-lived plasma cells (SLPCs) in vivo, a fate thought to be driven by intense BCR signaling induced by endogenous antigens. Similarly, IgD-only B cells generate normal germinal center, but impaired IgG1+ SLPC responses to T-dependent immunization. We propose a role for IgD in maintaining the quiescence of autoreactive B cells and restricting their differentiation into autoantibody secreting cells.

https://doi.org/10.7554/eLife.35074.001

eLife digest

To defend an organism against invaders such as viruses and bacteria, cells of the immune system need to recognize and respond to foreign microbes. However, these immune cells must also avoid attacking ‘self’ – for example, the healthy tissues of the body – as this could lead to autoimmune disease.

B cells are a type of immune cell that is essential in order to produce antibodies, protective proteins that can identify and defend against a broad range of germs: in addition, certain antibodies can also recognize ‘self’. When a B cell first develops, it places its antibody on its surface and uses this protein as a receptor (termed ‘B cell receptor’) to sense its surroundings.

Prior to mounting an immune response, B cells carry two closely related versions of the B cell receptor on their surface: IgM and IgD. Both IgM and IgD perform many of the same roles and can largely substitute for one another. However, B cells that recognize ‘self’ decrease their levels of IgM but keep high amounts of IgD on their surface. It is unclear why this is the case, but one possibility is that IgM and IgD may see ‘self’ differently.

To investigate this, Noviski et al. used mice with B cells that only carry either IgM or IgD, and tracked how these cells reacted to molecules from ‘self’ and foreign origins. IgM-only B cells reacted more strongly to ‘self’ molecules than IgD-only cells, which suggests that IgM is more sensitive to ‘self’ than IgD. In fact, in mice which are at risk for an autoimmune disease similar to lupus, deleting the IgM receptor prevented antibodies against ‘self’ from being produced. Therefore, reducing the amount of IgM receptors may be a way to keep B cells that are reactive against ‘self’ from inappropriately attacking the host. Meanwhile, the IgD receptor still allows B cells to mount protective antibody responses against foreign microbes.

Future studies are necessary to determine whether these differences between IgM and IgD also exist in humans. If this is the case, blocking the IgM receptor could become a new kind of treatment for certain autoimmune diseases.

https://doi.org/10.7554/eLife.35074.002

Introduction

The pre-immune mature naïve B cell compartment must balance the need for a diverse antibody repertoire with the risk of autoantibody-mediated disease. Early in development, B cells randomly rearrange their immunoglobulin genes through VDJ recombination and encounter a series of tolerance checkpoints that serve to remove autoreactive B cell receptors (BCRs) from the repertoire. Strongly autoreactive immature B cells rearrange additional light chains to ‘edit’ their autoreactivity, and they ultimately undergo deletion if this is unsuccessful (Shlomchik, 2008). Yet, despite stringent counter-selection of autoreactivity, the mature follicular (Fo) B cell compartment retains cells reactive towards endogenous antigens (Wardemann et al., 2003; Zikherman et al., 2012). How these cells are restrained from mounting autoimmune responses is not fully understood.

We previously described a BAC Tg reporter mouse (Nur77-eGFP) in which GFP expression is under the control of the regulatory region of Nr4a1, an immediate early gene rapidly induced by antigen receptor signaling (Mittelstadt and DeFranco, 1993; Winoto and Littman, 2002). We showed that Nur77-eGFP expression in naïve B cells is proportional to strength of antigenic stimulation and consequently to autoreactivity (Zikherman et al., 2012). The most prominent characteristic of GFPhi reporter B cells is decreased surface IgM BCR relative to GFPlo cells. Goodnow and colleagues first suggested that IgM downregulation may mark autoreactive B cells in the normal mature repertoire shortly after they reported selective downregulation of IgM in the IgHEL (hen egg lysozyme) BCR Tg/soluble HEL Tg model system of B cell autoreactivity (Goodnow et al., 1988; Goodnow et al., 1989). Indeed, multiple studies in mice and humans have identified naturally occurring IgD+IgMlo cells as autoreactive and ‘anergic’ or functionally unresponsive (Duty et al., 2009; Kirchenbaum et al., 2014; Quách et al., 2011; Zikherman et al., 2012). These results corroborate observations with several BCR transgenic model systems (Cambier et al., 2007). However, whether or how IgM downregulation might constrain autoreactivity remains unclear because naturally-occurring autoreactive B cells maintain high expression of the IgD BCR isotype (Zikherman et al., 2012).

IgM and IgD are splice isoforms of a common precursor heavy chain mRNA (Moore et al., 1981). While they differ in their Fc domains, both BCR isotypes contain the same antigen-binding domain, as well as identical 3-amino acid cytoplasmic tails, and they pair with Igα/β in order to initiate the canonical BCR signaling cascade (Blum et al., 1993; Radaev et al., 2010). However, the expression pattern of the isotypes differs; IgM expression begins as soon as heavy and light chains recombine early in B cell development, and persists until class switch recombination occurs following B cell activation (Chen and Cerutti, 2010). IgD is uniquely co-expressed with IgM during a narrow developmental window on late transitional and mature naïve Fo B cells as a result of alternate splicing regulated by the zinc-finger protein ZFP318 (Enders et al., 2014; Pioli et al., 2014). This suggests that IgD may play a critical role specifically in mature naïve B cells. Initial characterization of IgM- and IgD-deficient mice revealed only mild phenotypes and substantial redundancy; each isotype could mediate B cell development, initiate antibody responses to T-dependent and -independent immunization, and induce normal levels of steady-state serum IgG (Lutz et al., 1998; Nitschke et al., 1993; Roes and Rajewsky, 1993). This is consistent with prior studies in BCR transgenic model systems demonstrating that each isotype alone can mediate B cell development, deletion, and activation (Brink et al., 1992).

IgM and IgD differ structurally; IgD has a much longer, flexible hinge region linking its Fab and Fc regions than IgM does (Chen and Cerutti, 2010). Recently, Ubelhart, Jumaa, and colleagues demonstrated that a short and inflexible hinge confers upon IgM the unique capacity to signal in response to monovalent antigens, while both isotypes can respond to multivalent antigens (Übelhart et al., 2015). However, the Goodnow lab subsequently showed that IgHEL Tg splenic B cells expressing either the IgD or IgM BCR exclusively could mobilize calcium in response to soluble HEL antigen in vitro, and each isotype can mediate a common gene expression program characteristic of anergy in vivo (Sabouri et al., 2016). Therefore, it remains unclear whether IgM and IgD BCRs expressed in an unrestricted B cell repertoire differentially sense bona fide endogenous antigens in vivo, particularly as the identity and nature of such antigens is largely unknown and not restricted to soluble monovalent antigens. For example, 5–10% of circulating naïve B cells in healthy humans express the well-characterized unmutated and autoreactive human heavy chain V-segment IGHV4-34, and these cells exhibit anergy, selective IgM downregulation, and recognize cell-surface antigens on erythrocytes and B cells in vivo (Quách et al., 2011; Reed et al., 2016).

We and others previously hypothesized that downregulation of IgM on Fo B cells might serve to restrain their response to endogenous antigen. Conversely, high expression of IgD on these cells might play a ‘tolerogenic’ role (Zikherman et al., 2012). To determine how IgM and IgD differentially regulate B cell responses to endogenous antigens, we generated Nur77-eGFP reporter mice deficient for either IgM or IgD (Lutz et al., 1998; Nitschke et al., 1993). Using this reporter of antigen-dependent signaling, we show that IgD is less sensitive than IgM to bona fide endogenous antigens in vivo despite higher surface expression and robust responsiveness to receptor ligation in vitro. Indeed, marginal zone (MZ) and B1a cells that express IgD but lack IgM induce less Nur77-eGFP than WT, and this is not attributable to repertoire differences. To further support these observations, we examine a series of cell fate decisions for which in vivo BCR signaling requirements have been previously defined. We show that IgD drives a pattern of development consistent with reduced endogenous antigen recognition, favoring MZ B cell fate and disfavoring B1a cell fate. Similarly, IgD alone is less efficient than IgM at driving SLPC expansion in response to endogenous antigens, a process that requires robust BCR signaling. However, IgD is sufficient for germinal center (GC) B cell differentiation. Our data suggest that reduced endogenous antigen sensing (i.e. signal transduction in response to antigen binding) by the IgD BCR isotype shunts autoreactive IgDhi IgMlo Fo B cells away from differentiation into SLPCs. We propose that predominant IgD expression maintains the quiescence of autoreactive B cells in response to chronic endogenous antigen stimulation, and limits autoantibody secretion in the context of rapid immune responses.

Results

Endogenous antigen is both necessary and sufficient for expression of Nur77-eGFP reporter in B cells in vivo

We previously showed that antigen recognition was both necessary and sufficient for Nur77-eGFP reporter expression by B cells in vivo (Figure 1A)(Zikherman et al., 2012). Thus, reporter expression reflects endogenous antigen recognition in vivo. Conversely, under steady-state conditions in vivo, the Nur77-eGFP reporter does not reflect signaling through other receptors expressed in B cells; indeed, loss of either CD40, or TLR3, 7, and 9 signaling has no effect on reporter expression in B cells in vivo (Figure 1—figure supplement 1A). Moreover, neither BAFF, nor IL-4, nor CXCR4-dependent signaling can regulate reporter expression in vitro (Figure 1—figure supplement 1B–C) (Zikherman et al., 2012). To further confirm that Nur77 expression in naïve B cells under steady state conditions is not regulated by microbial stimulation of pattern recognition receptors (e.g. TLRs 1, 2, 4, 6, and the TLR-4-like molecule RP105), we studied mice raised under either germ-free conditions or conventional specific-pathogen-free conditions. We observed no induction of endogenous Nur77 protein in splenic B cells or endogenous Nr4a1 transcript in splenocytes in the presence of commensal flora (Figure 1—figure supplement 1D,E). Moreover, MyD88-deficient and MyD88-sufficient splenocytes and peritoneal B1a cells express comparable amounts of endogenous Nr4a1 transcript and protein respectively under steady state conditions (Figure 1—figure supplement 1F,G). Taken together, these data demonstrate that Nur77-eGFP expression in B cells under steady-state conditions in vivo is a specific readout of antigen-dependent signaling through the BCR (Table 1). We therefore sought to take advantage of the Nur77-eGFP reporter in order to probe the responsiveness of a diverse BCR repertoire of mature B cells expressing either IgM or IgD alone to the vast range of endogenous antigens they may encounter in vivo.

Figure 1 with 3 supplements see all
IgD expression enables a dynamic range of IgM responsiveness.

(A) GFP expression in mature Fo splenic B cells (CD19+CD23+CD93-) from Nur77-eGFP BAC Tg reporter mice with either a wild-type BCR repertoire (left), or harboring IgHEL Tg specific for the cognate antigen HEL (hen egg lysozyme) in the absence (middle), or presence (right) of endogenous cognate antigen driven by soluble HEL Tg. WT Fo B cells lacking GFP reporter are included for reference (gray shaded histograms). (B) Surface IgM and IgD expression in splenic Fo B cells from WT, IgM−/−, and IgD−/− mice expressing the Nur77-eGFP reporter. (C) Splenocytes from WT, IgM−/−, and IgD−/− mice were loaded with Indo-1 and stimulated with 2.5 μg/mL of F(ab’)2 anti-IgM or 1:400 anti-IgD. Fo B cells with the highest 20% and lowest 20% Nur77-eGFP expression are compared. (D) Representative histograms and quantification of surface Igκ expression in IgM−/− and IgD−/− Fo B cells normalized to WT. (E) Representative histograms and quantification of Nur77-eGFP expression in IgM−/− and IgD−/− Fo B cells normalized to WT. (F) Median surface Igκ expression of WT, IgM−/−, and IgD−/− Fo B cells was calculated for 200 bins of equal width across the Nur77-eGFP spectrum. For (A), (B) and (F), data are representative of at least n = 4 independent experiments. For (C), n = 3 independent experiments for anti-IgM and n = 2 independent experiments for anti-IgD. For (D) and (E), n = 7 and n = 4, respectively, WT, IgM−/−, and IgD−/− mice. Welch’s t test was used to calculate p values, and mean +SEM is displayed. ***p<0.001.

https://doi.org/10.7554/eLife.35074.003
Table 1
Nur77-eGFP is a specific reporter of antigen-dependent signaling in vivo.
https://doi.org/10.7554/eLife.35074.010
ConclusionStimulus/perturbationPathwayReadoutCell typeReferences
Does not modulate Nur77 at steady state in vivoCD40L-/-CD40Nur77-eGFPB cellsManuscript Figure 1—figure supplement 1A
TLR7-/-TLR7Nur77-eGFPB cellsManuscript Figure 1—figure supplement 1A
Un93b13d/3dTLR3/7/9Nur77-eGFPB cellsManuscript Figure 1—figure supplement 1A
MyD88fl/fl MB1-CreMyD88Endog. Nur77 protein/transcriptPerC B1a cells/spleenManuscript Figure 1—figure supplement 1F,G
Germ-free miceMyD88/TRIFEndog. Nur77 protein/transcriptSplenic B cells/spleenManuscript Figure 1—figure supplement 1D,E
Const. act. STAT5Jak/StatNur77-eGFPThymocytesMoran et al. JEM 2011, Figure 8.
Does not induce Nur77 in vitroBAFFBAFFRNur77-eGFPB cellsZikherman et al. (2012): Figure S1G
IL-4Jak/StatNur77-eGFPB cellsManuscript Figure 1—figure supplement 1B
IL-2, IL-15Jak/StatNur77-eGFPCD8 (IL-2, 15), CD4 (IL-2)Au-Yeung et al. JI 2017, Figures S1A, 3C, 4A
CXCL12/
SDF-1
CXCR4Nur77-eGFPB cellsManuscript Figure 1—figure supplement 1C
Induces Nur77 in vitro but does not require IgM or IgD specificallyLPSTLR4, Rp150Nur77-eGFPB cellsZikherman et al. (2012): Figure S1G; Manuscript Figure 1—figure supplement 3A
CpGTLR9Nur77-eGFPB cellsZikherman et al. (2012), Figure S1G; Manuscript Figure 1—figure supplement 3A
Pam3CSK4TLR1/2Nur77-eGFPB cellsManuscript Figure 1—figure supplement 3A
Anti-IgκBCRNur77-eGFPB cellsManuscript Figure 2F
Modulates pathway at steady state in vivoIgHEL TgAntigen/BCRNur77-eGFPB cellsZikherman et al. (2012), Figure 3B,C; Manuscript Figure 1A
IgHEL BCR Tg/sHEL AgAntigen/BCRNur77-eGFPB cellsZikherman et al. (2012), Figure 3B,C; Manuscript Figure 1A
Lyn-/-BCR via ITIMsNur77-eGFPB cellsManuscript Figure 5—figure supplement 3A
CD45 allelic seriesBCR via SFKsNur77-eGFPB cellsZikherman et al. (2012), Figure 3A,B

Dual expression of IgM and IgD BCRs is required to establish a broad dynamic range of BCR responsiveness across the repertoire

Surface IgM, but not IgD, expression is inversely correlated with endogenous antigen recognition and GFP expression across a diverse repertoire of mature naïve follicular (Fo) B cells, such that B cells reactive to endogenous antigens express high levels of IgD and low levels of IgM on their surface (Figure 1B) (Zikherman et al., 2012). The variation in surface IgM expression across the B cell repertoire is profound, spanning a 100-fold range. However, because IgD is expressed at high and invariant levels in all WT naïve Fo B cells, total surface BCR, unlike IgM, has little dynamic range across the WT repertoire (Figure 1—figure supplement 2A).

To explore how IgM and IgD differentially regulate B cells reactive to endogenous antigens, we generated Nur77-eGFP mice deficient for either IgM or IgD (Lutz et al., 1998; Nitschke et al., 1993). In IgM−/− mice, all B cells express IgD alone prior to class switch recombination (CSR), and in IgD−/− mice, all B cells express IgM alone prior to CSR. We observed that IgM is still downregulated on IgD−/− GFPhi Fo B cells, but the dynamic range of IgM expression is highly restricted relative to the broad range observed on WT cells (Figure 1B).

We previously showed that IgM down-modulation on GFPhi reporter B cells largely accounts for impaired signaling through IgM (Figure 1C) (Zikherman et al., 2012). However, the narrow dynamic range of IgM on IgD-deficient cells renders reporter B cells unable to fine-tune signaling through the IgM BCR across the GFP repertoire (Figure 1C). Furthermore, IgD-mediated signaling is not altered across the GFP repertoire in the presence or absence of IgM (Figure 1C). As a result, reduced responsiveness of GFPhi B cells is most profound in B cells that express the IgM isotype in the presence of IgD. This suggests that IgD expression may be necessary to establish and/or maintain a broad and functionally dynamic range of surface IgM expression across the B cell repertoire. This is consistent with a previously proposed survival function for the IgD BCR (Roes and Rajewsky, 1993; Sabouri et al., 2016).

IgM−/− follicular B cells express more surface BCR than IgD−/− follicular B cells, but similar Nur77-eGFP

Deletion of either the IgM or the IgD BCR isotype results in compensatory upregulation of the remaining isotype. This may be due in part to a change in competition for pairing with Igα/β heterodimers, which is essential for trafficking of IgM to the cell surface (Hombach et al., 1990; Sabouri et al., 2016). IgM−/− B cells express about twice as much surface BCR as IgD−/− B cells, as measured by surface anti-light chain staining (Figure 1D and Figure 1—figure supplement 2B–E). Nevertheless, distribution of Nur77-eGFP expression across the naïve B cell repertoire is nearly identical in IgM−/− and IgD−/− Fo B cells (Figure 1E). Further, at every level of Nur77-eGFP, cells that express IgD alone require more surface BCR to drive an equivalent amount of GFP compared to cells that express IgM alone (Figure 1F). This suggests that on a per-receptor basis, IgD BCR may be less efficient at inducing Nur77-eGFP in response to endogenous antigens. Importantly, this is not attributable to differential dependence upon IgM or IgD expression for signaling via CXCR4 or downstream of canonical TLR ligands (Figure 1—figure supplement 1C, Figure 1—figure supplement 3A).

IgD BCR crosslinking induces robust signaling in vitro

Because IgD drives less Nur77-eGFP per receptor than IgM in vivo, we wanted to determine whether there were defects in signal transduction downstream of IgD. To mimic antigen binding to the membrane-distal end of the BCR and allow for direct comparison between the isotypes, we stimulated IgD-only (IgM−/−) and IgM-only (IgD−/−) cells with anti-Igκ and compared downstream signaling. BCR ligation in IgD-only B cells induced equivalent amounts of Erk phosphorylation relative to IgM-only B cells (Figure 2A), and there were no gross differences in pErk kinetics (Figure 2B). Additionally, IgD-only cells induced more robust intracellular calcium increase and S6 phosphorylation than IgM-only cells (Figure 2C and Figure 2—figure supplement 1A). This was not due to differential effects of the Fc portion of the stimulatory antibody on IgM-only and IgD-only B cells as anti-Igκ-F(ab’)2 and anti-Igκ stimulation produced identical signaling (Figure 2—figure supplement 1B).

Figure 2 with 1 supplement see all
IgD signals strongly in vitro but weakly in vivo.

(A) Median intracellular pErk in splenic CD23+ B cells stimulated with anti-Igκ for 15 min. (B) Erk phosphorylation kinetics in splenic CD23+ B cells stimulated with 15 μg/mL anti-Igκ. (C) Splenocytes from IgM−/− and IgD−/− mice were loaded with Indo-1 and stimulated with 2.5 or 5 μg/mL anti-Igκ. B220+CD23+CD93- Fo B cells are compared. (D) CD86 induction in CD23+ IgM−/− and IgD−/− splenocytes stimulated with anti-Igκ for 18 hr. (E) Summary data for CD86 MFI in (D). (F) Nur77-eGFP induction in cells from (D). (G) Nur77-eGFP in cells from (D) incubated with medium alone (0 μg/mL anti-Igκ). (H) Representative histograms and summary data for MHC-II induction in unstimulated cells from (D). (I) Basal calcium in unstimulated IgM−/− and IgD−/− Fo B cells was calculated by normalizing the geometric mean of [Indo-1(violet)/Indo-1(blue)] to WT B cells in the same experiment. For (A), signaling in cells from n = 2 IgM−/− and IgD−/− mice is displayed, and results for 1.5, 3, and 15 μg/mL of anti-Igκ were replicated in n = 3 independent experiments. Data in (B) was compiled from n = 3 independent experiments with n = 3 mice of each genotype in each experiment. Data in (C) are representative of n = 4 independent experiments for 5 μg/mL and n = 2 independent experiments for 2.5 μg/mL. For (D-H), values were calculated for splenocytes from n = 3 mice of each genotype. For (I), basal calcium ratios from n = 6 independent experiments are compiled. Welch’s t test was used to calculate p values, and mean ±SEM is displayed. *p<0.05, **p<0.01, ***p<0.001.

https://doi.org/10.7554/eLife.35074.011

IgD BCRs sense endogenous antigens less efficiently than IgM BCRs

Due in part to increased surface receptor expression, IgD-only B cells induce comparable Erk phosphorylation and enhanced calcium mobilization relative to IgM-only B cells stimulated with anti-Igκ. As a result, we observe significantly more CD86 and Nur77-eGFP induction in IgD-only B cells after 18 hr of anti-Igκ stimulation in vitro (Figure 2D–F). However, IgD-only cells induced less Nur77-eGFP and MHC-II upregulation when cultured in the absence of exogenous stimulus (Figure 2G and H). This discrepancy suggests that despite increased receptor expression and efficient coupling to downstream signaling machinery, residual endogenous antigens occupying the IgD BCR are less efficient at inducing signaling ex vivo. Consistent with this hypothesis, basal calcium analyzed immediately ex vivo is depressed in cells that express only IgD (Figure 2I). These data suggest that while signal transduction downstream of the IgD BCR is robust in vitro, IgD responds less efficiently than IgM to the relevant antigens it encounters in vivo.

Reduced endogenous antigen sensing by IgD BCR in innate-like B cells

While IgD-only and IgM-only Fo B cells express comparable levels of Nur77-eGFP (Figure 1E), we suspected that competitive pressures for survival might constrain the acceptable range of BCR signaling among mature Fo B cells. We further speculated that compensatory mechanisms such as altered surface receptor and altered BCR repertoire might fine tune how much signaling Fo B cells experience in vivo in order to lie within this range. In contrast to Fo B cells, IgD-only marginal zone (MZ) and B1a cells expressed significantly less Nur77-eGFP than WT cells despite higher levels of surface BCR (Figure 3A, Figure 1—figure supplement 2B–E). To determine whether this difference was due to isotype or altered BCR repertoire, we probed GFP expression in B1a cells specific for phosphatidylcholine (PtC), an endogenous antigen exposed on the surface of dying cells that is thought to select developing B cells into the B1a compartment (Baumgarth, 2011). We found a large difference in GFP even among B1a cells with a common specificity, and this occurs in spite of high surface PtC binding in IgD-only B1a cells (Figure 3B and C).

Figure 3 with 1 supplement see all
Reduced in vivo antigen sensing by IgD in innate-like B cells.

(A) Nur77-eGFP in peritoneal B1a (CD19+CD5+CD23-) and splenic MZ (B220+CD21hiCD23lo) B cells from WT and IgM−/− mice. (B) Nur77-eGFP in PtC-binding peritoneal B1a cells from WT, IgM−/−, and IgD−/− mice. (C) Nur77-eGFP and PtC MFIs were calculated for total B1a and PtC-binding B1a cells. The ratio of the IgM−/− (IgD-only) MFI to the IgD−/− (IgM-only) MFI is displayed with a 95% confidence interval. (D) Representative histograms of Nur77-eGFP in IgM−/− and IgD−/− splenic MZ B cells from mice without (left) and with (right) a BAFF overexpression transgene. (E) Quantification of Nur77-eGFP in MZ B cells from (D). For (A) and (E), n = 3 mice of each genotype were analyzed. For (B), histograms are representative of n = 4 mice of each genotype. For (C), ratios are pooled for n = 4 mice of each genotype from three independent experiments. For (D), histograms are representative of n = 3 mice of each genotype. Welch’s t test (A and E) was used to calculate p values, and mean +SEM is displayed (except in C). *p<0.05.

https://doi.org/10.7554/eLife.35074.015

We took an independent and complementary approach to address the same question in MZ B cells; we crossed IgM−/− and IgD−/− reporter mice to a genetic background that drives over-expression of the B cell survival factor BAFF (Gavin et al., 2005). BAFF overexpression ‘unrestricts’ the BCR repertoire by removing competition for survival, thereby reducing both positive and negative selection pressures and permitting cells with either insufficient or excessive BCR signaling to persist in the repertoire, resulting in a massively expanded MZ compartment (Stadanlick and Cancro, 2008). Importantly, BAFF does not directly regulate reporter GFP expression, and therefore GFP reflects endogenous antigen stimulation on this genetic background (Zikherman et al., 2012). In BAFF Tg mice, IgD-only MZ cells expressed much less GFP than those expressing IgM alone (Figure 3D and E). This was not explained by differences in surface receptor levels between these populations as they are comparable on this genetic background (Figure 3—figure supplement 1A). Indeed, IgD-only cells with low GFP and IgM-only cells with high GFP appear to be preferentially rescued in mice with excess BAFF, suggesting that distinct BCR repertoires, far from accounting for GFP differences between IgM-only and IgD-only MZ B cells, actually obscure these differences. Conversely, either ‘fixing’ or unrestricting the BCR repertoire unmasks impaired GFP upregulation by IgD relative to IgM.

As a result of allelic exclusion of the heavy chain locus during B cell development, IgM+/− heterozygous mice develop two genetically distinct sets of B cells in which half express only IgD and the other half express both IgM and IgD. We exploited this property to confirm that the Nur77-eGFP difference in the MZ and B1a compartments is cell intrinsic and not a consequence of the presence or absence of serum IgM (Figure 3—figure supplement 1B).

Since IgM−/− and IgD−/− mice were generated on the Balb/c and 129 genetic backgrounds respectively, their germline VDJ loci are different from that of C57BL/6 mice to which they have been back-crossed. Throughout our study, we have validated our findings in either Balb/c-B6 F1 mice or IgHa/b heterozygous mice, which have two sets B cells with identical IgM and IgD isotype expression but distinct germline VDJ loci. We took this approach to confirm that differences in MZ GFP were due to BCR isotype and not VDJ locus (Figure 3—figure supplement 1C).

Cell-intrinsic skewing of B cell development by the IgM and IgD BCRs

Generation of the B1a compartment requires endogenous antigen recognition and strong BCR signaling, while the opposite is true for MZ B cells (Figure 4—figure supplement 1A) (Cariappa and Pillai, 2002; Casola et al., 2004). Indeed, B1a cells are modestly reduced in IgM-deficient mice, while MZ B cells are modestly increased (Figure 4—figure supplement 1B–C). Because serum IgM deficiency leads to increased B1a numbers, we assessed B1a and MZ B cell development in a competitive setting to isolate the cell-intrinsic effects of IgM and IgD BCRs (Boes et al., 1998). Analogous to the IgM+/− mice described above, IgM+/− IgD−/+ heterozygous mice develop two genetically distinct sets of B cells in which half express only IgD and the other half express only IgM (Figure 4A). Although IgD-only B cells can partially populate the B1a compartment in the absence of competition (Lutz et al., 1998; Übelhart et al., 2015), IgD-only cells are virtually excluded from this compartment in a competitive setting (Figure 4B). Indeed, this parallels the loss of PtC-binding IgD-only peritoneal B cells observed in competition with wild type cells (Übelhart et al., 2015). In contrast, IgD-only B cells preferentially populate the MZ B cell compartment in competition with IgM-only B cells (Figure 4B). These data are consistent with a signal strength model of B1a/MZ B cell development (Figure 4—figure supplement 1A), and this suggests that in vivo signaling, antigen recognition, or both are reduced in IgD-only B cells. These differences in signaling could in turn modulate the generation and/or survival of IgD-only B1a and MZ B cells.

Figure 4 with 1 supplement see all
Cell-intrinsic skewing of B cell development by IgM and IgD BCRs.

(A) Allelic exclusion leads to a 1:1 mixture of IgM-only and IgD-only B cells in IgM+/− IgD−/+ mice. (B) Proportion of peritoneal B1a (CD19+CD5+CD23-) and splenic MZ (B220+CD21hiCD23lo) B cells originating from each Ig locus in IgM+/− IgD−/+ mice. (C) Relative competition between IgM+ and IgD+ B cells in IgM+/− IgD−/+ mice was calculated for bone marrow, splenic, and peritoneal B cell compartments. Results include data from (B) for reference. Immature (CD23-CD93+); T2-like (CD23+CD93+); mature recirculating (CD23+CD93-); T1 (CD93+CD23-); T2/3 (CD93+CD23+); Fo (CD93-CD23+); MZ (CD21hiCD23lo); B1a (CD5+CD23-); B1b (CD5-CD23-); B2 (CD5-CD23+). (D) Relative competition between WT (IgMb+) and IgM-null (IgDa+) B cells in IgM+/− mice was determined as in (C). (E) Relative competition between WT (IgMb+) and IgD-null (IgMa+) B cells in IgD+/− mice was determined for splenic and peritoneal compartments as described in (C). (F) Competition in peritoneal B1a and splenic MZ compartments in IgM+/− mice with or without a BAFF overexpression transgene. Results include data from (D) for reference. For (B) and (C), n = 3–5 mice were analyzed. For (D), n = 3–8 mice were analyzed. For (E), n = 5 mice were analyzed. For (F), n = 4–5 mice of each genotype were analyzed. Welch’s t test was used to calculate p values, and mean ±SEM is displayed. **p<0.01, ***p<0.001.

https://doi.org/10.7554/eLife.35074.019

We sought to determine whether skewed development of B cell lineages was driven primarily by properties of IgD, IgM, or both by comparing development of B cell populations in IgM+/− IgD−/+, IgM+/−, and IgD+/− mice (Figure 4C–4E). We conclude that skewing of MZ and B1a fates is primarily attributable to lack of IgM expression on IgD-only cells rather than lack of IgD expression on IgM-only cells because we observe little skewing in IgD+/− mice (Figure 4E). Further, the competitive advantage of IgD-only cells in the marginal zone niche is reduced by BAFF over-expression, consistent with loosening of competitive selection pressures (Figure 4F).

IgM-only B cells exhibit a disadvantage in the mature Fo compartment relative to IgD-only B cells (Figure 4C). We find that absence of IgD rather than excess IgM expression accounts for some of this disadvantage as WT and IgD-only cells compete equally well in this compartment (Figure 4D). This may reflect impaired positive selection or survival in absence of IgD. Indeed, similar to our observations, a disadvantage for Fo B cells lacking IgD was observed in IgD+/− mice previously (Roes and Rajewsky, 1993). As with differences in Nur77-eGFP expression, we confirmed that the developmental fates of IgM-only and IgD-only cells were due to BCR isotype and not VDJ locus (Figure 4—figure supplement 1D).

Either IgD or IgM BCR is sufficient to mediate polyclonal B cell activation and germinal center differentiation in Lyn−/− mice

Mice deficient for either IgM or IgD can mount T-independent and T-dependent immune responses to model antigens (Lutz et al., 1998; Nitschke et al., 1993; Roes and Rajewsky, 1993). However, endogenous antigens may have unique properties that are absent in model antigen systems. To test whether each BCR isotype was competent to mediate autoimmune responses to endogenous antigens in lupus-prone mice, we generated IgM+/− and IgD+/− mice on the Lyn−/− background (Chan et al., 1997). The Src family kinase Lyn is essential to mediate ITIM-dependent inhibitory signals in B cells and myeloid cells (Scapini et al., 2009; Xu et al., 2005). Lyn−/− mice consequently develop a spontaneous lupus-like disease characterized by anti-DNA antibodies and nephritis on the C57BL/6 genetic background (Chan et al., 1997; Hibbs et al., 1995; Nishizumi et al., 1995). It has been shown that both B-cell-specific MyD88 expression and T cells are essential for IgG2a/c anti-dsDNA autoantibody production in Lyn−/− mice, and conditional deletion of Lyn in B cells is sufficient for autoimmunity (Hua et al., 2014; Lamagna et al., 2014).

Secreted natural IgM is thought to play important homeostatic functions that repress autoimmunity, particularly clearance of dead cell debris (Boes et al., 2000; Manson et al., 2005). By performing our analysis in mice with wild type and IgM-only or IgD-only B cells in a common milieu (IgM+/− Lyn−/− and IgD+/− Lyn−/− mice), we could control for this and isolate the B cell-intrinsic effects of the IgM and IgD BCRs. Importantly, precursor Fo B cells are present at roughly equal ratios from WT and IgM-only or IgD-only loci in these mice, and this is unaffected by VDJ background (Figure 5—figure supplement 1A–C). Moreover, Lyn−/− B cells expressing either IgM or IgD alone exhibit enhanced BCR signaling in vitro relative to Lyn+/+ B cells, suggesting that neither isotype is uniquely coupled to ITIM-containing receptors (Figure 5—figure supplement 2A–B).

Polyclonal B cell activation in Lyn−/− mice precedes and is genetically separable from autoantibody production; it is driven by B cell-intrinsic loss of Lyn, is thought to result from enhanced BCR signaling, and is independent of MyD88 expression (Hua et al., 2014; Lamagna et al., 2014). Indeed, Lyn−/− reporter B cells have elevated Nur77-eGFP expression consistent with enhanced BCR signal transduction in vivo (Figure 5—figure supplement 3A). We had expected that activation of IgD-only Lyn−/− B cells might be impaired due to reduced sensing of endogenous antigens, but neither CD86 nor CD69 upregulation were significantly different on WT and IgD-only B cells in the absence of Lyn (Figure 5A). This implies that both BCRs can provide sufficient signals to drive polyclonal activation by endogenous antigens.

Figure 5 with 3 supplements see all
IgD can drive polyclonal activation and germinal center entry, but not anti-dsDNA IgG2a production, in Lyn−/− mice.

(A) Surface CD69 and CD86 expression on CD23+ splenic B cells from each Ig locus in IgM+/− mice on Lyn+/+ and Lyn−/− backgrounds. (B) Percentage of unswitched germinal center (CD19+ Fashi GL-7hi IgM/IgD+) B cells from each Ig locus in IgM+/− and IgD+/− mice on the Lyn−/− background. (C) Anti-dsDNA IgG2a titers from each Ig locus in IgM+/− Lyn−/− mice were calculated by ELISA using pooled IgHa/b autoimmune serum with high-titer autoantibodies from each locus as a reference (titer set at 1000). Each color represents a single mouse tracked over time. (D) Anti-dsDNA IgG2a titers in IgD+/− Lyn−/− mice were calculated as in (C). (E) Paired anti-dsDNA titers from each locus in individual IgM+/− Lyn−/− mice from (C) with additional mice from 24 weeks to 12 months. (F) Paired anti-dsDNA titers from each locus in IgD+/− Lyn−/− mice from (D). (G) Ratio of anti-dsDNA IgG2a[a] to IgG2a[b] from all samples in (E) and (F) with an anti-dsDNA IgG2a titer >250 from either locus; cutoff defined by titers in young WT mice. For (A), n = 4 IgM+/− Lyn−/− mice are compared to a reference IgM+/− Lyn+/+ mouse. Qualitatively similar results were obtained in two independent experiments. For (B) n = 5–6 mice of each genotype were analyzed. For (C) and (D), n = 9 and n = 12 mice were tracked. For (G), n = 36 IgM+/− Lyn−/− and n = 39 IgD+/− Lyn−/− anti-dsDNA IgG2a+ samples were compared. Welch’s t test was used to calculate p values, and mean ±SEM is displayed. **p<0.01.

https://doi.org/10.7554/eLife.35074.024

Although GC fate is disfavored relative to SLPC fate in Lyn−/− mice, GCs do arise spontaneously with time (Hibbs et al., 1995; Hua et al., 2014). We found that both IgD-only and IgM-only cells in Lyn-deficient mice made comparable contributions to the GC compartment in competition with wild type B cells, independent of VDJ background (Figure 5B and Figure 5—figure supplement 3B–C). We propose that endogenous antigen sensing by IgD, while normally dampened, is sufficient to drive both polyclonal activation and GC differentiation on the Lyn−/− background. This further implies that antigen capture by IgD and presentation on MHC-II is robust enough to recruit adequate T cell help to support these cell fates.

IgM BCR is required to drive dsDNA antibody production in Lyn−/− mice

IgM- and IgD-deficient B cells express BCR heavy chain allotype [a] (IgHa), and their secreted antibodies can be differentiated from WT B6 antibodies (IgHb) even after isotype switching (e.g. IgG2a[a] vs. IgG2a[b] – also referred to as IgG2c). Allotype-specific antibodies have been previously used in the context of lupus mouse models to track autoantibodies generated by B cells of distinct genetic origin (Mills et al., 2015; Pisitkun et al., 2006). We took an analogous approach to assess IgG2a/c anti-dsDNA autoantibodies emanating from each genetic locus in IgM+/−, IgD+/−, and IgHa/b control mice deficient for Lyn. In order to compensate for reduced penetrance of autoantibody production in IgM+/− Lyn−/− mice, we collected serum from this genotype at time points beyond 24 weeks to increase the number of samples in which tolerance had been broken. While B cells expressing IgM (of either IgH locus) generated IgG2a/c anti-dsDNA, IgD-only B cells were relatively protected from generating anti-dsDNA antibodies, even at late time points (Figure 5C–G, Figure 5—figure supplement 3D–E). Importantly, this is attributable to BCR isotype and not VDJ locus (Figure 5—figure supplement 3D–E).

We noted that anti-dsDNA antibody titers fluctuated substantially over time in Lyn−/− mice, in some cases dropping markedly over a four-week period (Figure 5C and D). Anti-dsDNA IgG autoantibodies in patients with systemic lupus erythematosus (SLE) similarly fluctuate over time – in contrast to other anti-nuclear specificities - and correlate with disease flares (Liu et al., 2011). Furthermore, unmutated germline BCRs from naïve B cells are recruited directly into the circulating plasmablast compartment during SLE flares (Tipton et al., 2015). Taken together, this suggests that anti-dsDNA antibodies in mice and humans are secreted at least in part by SLPCs emerging from an extra-follicular immune response, rather than long-lived plasma cells (LLPCs) that originate in GCs. Our data suggests that, in the absence of Lyn, the IgD BCR is sufficient to mediate B cell activation and GC entry, but the IgM BCR is essential to produce dsDNA-specific SLPCs.

IgM BCR is required for expansion of unswitched plasma cells in Lyn−/− mice

In addition to stochastic generation of autoantibodies over time, Lyn−/− mice exhibit a massive expansion of the splenic IgM plasma cell compartment that corresponds to a roughly 10-fold increase in serum IgM (Hibbs et al., 1995). While natural IgM is thought to be secreted by B1a-derived plasma cells (Baumgarth, 2011), elevated serum IgM in Lyn−/− mice arises from the Fo B2 compartment since Lyn−/− mice lack MZ B cells, and bone marrow chimeras lacking the B1a compartment reconstitute the hyper-IgM secretion phenotype (Luo et al., 2014). Moreover, while BCR signaling is prominently enhanced in Lyn−/− B2 B cells, it is markedly dampened in Lyn−/− B1a cells (Figure 6—figure supplement 1A) (Skrzypczynska et al., 2016).

IgM plasma cell expansion is temporally and genetically separable from autoimmunity in Lyn−/− mice; this phenotype develops in mice as young as 7–10 weeks of age with 100% penetrance, is B-cell intrinsic, and is not dependent upon MyD88 expression (Hua et al., 2014; Infantino et al., 2014; Lamagna et al., 2014). A pathway involving Btk, Ets1, and Blimp-1 is thought to drive this phenotype. Ets1 inhibits plasma cell differentiation in part by antagonizing the key transcriptional regulator of plasma cell fate, Blimp-1 (John et al., 2008). Mice deficient for Ets1 display IgM plasma cell expansion, Ets1 levels are reduced in Lyn−/− mice, and restoration of Ets1 expression normalizes IgM plasma cell numbers (Luo et al., 2014). Furthermore, reducing levels of Btk, a critical kinase that mediates BCR signaling, is sufficient to suppress IgM plasma cell expansion in Lyn−/− mice, and Ets1 levels are concomitantly restored to normal levels in Btklo Lyn−/− mice (Gutierrez et al., 2010; Luo et al., 2014; Mayeux et al., 2015). Therefore, IgM plasma cell expansion in Lyn-/- mice is driven by exaggerated Btk-dependent BCR stimulation of B2 B cells by endogenous antigens (Figure 6—figure supplement 2A). Indeed, enhanced BCR signaling has been shown to favor SLPC fate over GC differentiation in Lyn-sufficient mice as well (Chan et al., 2009; Nutt et al., 2015; Paus et al., 2006). We therefore wondered whether impaired endogenous antigen sensing by IgD influenced Ets1 expression and unswitched PC expansion in Lyn−/− mice.

As previously shown, Lyn−/− B cells downregulate Ets1 protein expression (Figure 6A) (Luo et al., 2014). However, Lyn-deficient B cells expressing only IgD did not do so, consistent with a reduced ability of IgD to transmit antigen-dependent signals in vivo (Figure 6A). As previously reported, we found that the frequency of plasma cells in the spleens of Lyn−/− mice was increased fourfold relative to Lyn+/+ mice, but the bone marrow plasma cell compartment size was relatively unaffected (Figure 6B–C) (Infantino et al., 2014). The increase in splenic plasma cells was attributable entirely to unswitched (IgM+) plasma cells (Figure 6C). However, unswitched plasma cell expansion was completely absent in IgM−/− Lyn−/− spleens (Figure 6C). This was reflected in steady-state serum antibody levels; while Lyn−/− mice have elevated serum IgM relative to WT mice, IgM−/− Lyn−/− mice did not have elevated serum IgD relative to IgM−/− mice (Figure 6D–E).

Figure 6 with 2 supplements see all
Cell-intrinsic IgM expression is required for unswitched plasma cell expansion in Lyn−/− mice.

(A) Representative blot and quantification of Ets1 and GAPDH protein in purified splenic B cells from WT, Lyn−/−, and IgM−/− Lyn−/− mice. (B) Composition of the CD138+ plasma cell compartments in the spleen and bone marrow of WT, Lyn−/−, and IgM−/− Lyn−/− mice was determined by intracellular staining of IgM, IgD, and IgA. (C) Percentages in (B) multiplied by the fraction of live cells positive for CD138 in each tissue. Unswitched cells are positive for either IgM or IgD. Statistics correspond to unswitched plasma cell percentages; differences in IgA+ cells were not significant. (D) Serum IgM in 16-week-old mice was quantified for B6.IgHa (WT) and Lyn−/− mice by ELISA. A sample from an IgM−/− mouse is shown for reference. (E) Serum IgD in 16-week-old mice was quantified for IgM−/− and IgM−/− Lyn−/− mice by ELISA. A sample from a WT mouse is shown for reference. (F) Gating scheme for quantifying the unswitched splenic plasma cell composition of IgM+/− Lyn−/− mice. (G) Percentage of unswitched splenic plasma cells (CD138+B220loIgM/IgD+) from each locus in IgM+/− and IgD+/− mice on the Lyn−/− background. For (B) and (C), figures are representative of n = 4–5 mice of each genotype. For (D), values from n = 3 WT and n = 4 Lyn−/− mice are averaged. For (E), values from n = 3 IgM−/− and n = 4 IgM−/− Lyn−/− mice are averaged. For (F) and (G), n = 4–5 mice of each genotype were used. Welch’s t test was used to calculate p values, and mean ±SEM is displayed. *p<0.05, **p<0.01, ***p<0.001.

https://doi.org/10.7554/eLife.35074.029

To confirm that resistance to unswitched plasma cell expansion was a cell-intrinsic feature of IgD-only Lyn−/− B cells, we assessed this phenotype in IgM+/− Lyn−/− mice. Similar to the non-competitive setting, IgD-only Lyn−/− cells did not contribute to an expansion in the unswitched plasma cell compartment (Figure 6F-G). This effect was due to isotype and not VDJ allele usage because both IgHa and IgHb B cells in IgHa/b Lyn-/- mice generated unswitched plasma cells in a competitive setting (Figure 5—figure supplement 3C).

IgM BCR is required for efficient generation of short-lived IgG1+ plasma cells but is dispensable for GC B cell fate

IgD-only Lyn−/− Fo B cells are completely prevented from generating an expanded unswitched SLPC compartment in response to chronic endogenous antigen stimulation, but they are competent to enter the GC. To determine how this defect relates to responses towards exogenous model antigens, we sought to determine whether IgD-only Lyn+/+ B cells exhibit skewing of cell fate in the context of a T-dependent response to a model antigen. To generate a robust polyclonal B cell response in which T cell help should not be limiting, we used the classic immunogen sheep RBCs. During this immune response, Fo B cells can either enter the germinal center or rapidly differentiate into IgG1+SLPCs (Chan et al., 2009; Paus et al., 2006). We found that wild type, IgM−/−, and IgD−/− mice all produced extremely robust GC responses 5 days after SRBC injection, but B cells expressing IgD alone were unable to produce IgG1+ SLPCs at this time point (Figure 7A–D).

Figure 7 with 2 supplements see all
IgD-only cells have intact germinal center responses but impaired IgG1+ SLPC responses.

(A) Splenic (CD19+) B cells from WT, IgM−/−, and IgD−/− mice unimmunized or 5 days after i.p. immunization with 200 μL of 10% SRBCs. (B) Quantification of germinal center (Fashi GL-7hi) cells in (A). (C) Splenocytes from mice in (A). (D) Quantification of CD138+ IgG1+ plasma cells in (C). (E) WT (IgMb+) and IgM-null (IgDa+) germinal center B cells as a percentage of live splenocytes in unimmunized and IgM+/− mice 5 days after i.p. immunization with 200 μL of 10% SRBCs. (F) WT (IgG1b+) and IgM-null (IgG1a+) switched plasma cells (CD138 +IgG1+) as a percentage of live splenocytes in IgM+/− mice unimmunized or 5 days after i.p. immunization with 200 μL of 10% SRBCs. (G) Fraction of unswitched NP-specific germinal center cells (CD19+ Fashi GL-7hi IgM/IgD+) from the IgHa locus in the spleens of Balb/c-B6 F1 and IgM+/− mice 7–8 days after i.p. immunization with 100 μg NP-RSA. (H) Fraction of IgG1+CD138+ plasma cells from the IgHa locus in Balb/c-B6 F1 and IgM+/− mice 7–8 days after i.p. immunization with 100 μg NP-RSA. (I) NP-specific IgG1a and IgG1b titers at OD = 0.2 were calculated for the mice in (G–H) by ELISA. The IgG1a to IgG1b titer ratio was calculated for each mouse, and all ratios were normalized such that the average IgG1a/IgG1b ratio in Balb/c-B6 F1 samples = 1.0. For (A-D), statistics from n = 4 unimmunized mice of each genotype and n = 3 WT, n = 6 IgM−/−, and n = 7 IgD−/− immunized mice were pooled. For (E-F), n = 5 unimmunized and n = 5 immunized mice are shown. For (G-I), n = 5 Balb/c-B6 F1 mice and n = 3 IgM+/− mice are shown. One-way ANOVA with Tukey’s multiple comparisons test (B and D), a paired t test (E-F), and Welch’s t test (G-I) were used to calculate p values, and mean +SEM is displayed. *p<0.05, **p<0.01, ***p<0.001.

https://doi.org/10.7554/eLife.35074.035

Since serum IgM is absent in IgM−/− mice and is known to enhance PC generation (Boes et al., 1998), we wanted to assess the cell-intrinsic effect of IgD and IgM BCRs on the immune response to SRBCs. To do so, we immunized IgM+/− mice in which half of Fo B cells express IgD alone and half express both IgM and IgD. We could track the relative contribution of these two cell populations to the GC response because the bulk of GC B cells have not yet isotype-switched and express surface IgM or IgD. Both cell types made a robust contribution to the germinal center response (Figure 7E). Next we tracked the generation of IgG1+ PCs by detecting allotype [a] or allotype [b] IgG1. In contrast to the GC response, IgD-only B cells were significantly disfavored in this compartment, although the defect was less severe than that observed in IgM−/− mice, which lack serum IgM (Figure 7F). To control for differences in the VDJ locus, we performed this experiment using IgHa/b control mice and observed no significant differences between allotypes, implying that BCR isotypes rather than VDJ locus accounted for our observations (Figure 7—figure supplement 1A–B).

In contrast to impaired IgG1 SLPC responses by IgD-only B cells, we observed that IgD-only B cells were able to generate unswitched plasma cells in response to SRBC immunization in both non-competitive and competitive settings (Figure 7—figure supplement 1C–D). Unswitched PC numbers correlate well with MZ B cell numbers in these mice (Figure 4—figure supplements 1B and Figure 4D). Moreover, MZ B cells efficiently generate short-lived unswitched plasma cells even in response to TD-antigens (Phan et al., 2005; Song and Cerny, 2003). We therefore suspect that IgG1 PCs emanate from the Fo B cell compartment, while the unswitched PC response may originate in the MZ B cell compartment. Normal SLPC generation by IgM-deficient MZ B cells in response to immunization would be consistent with previously reported normal T-independent II responses in IgM-deficient mice (Lutz et al., 1998). However, we cannot exclude a contribution of follicular-derived PCs that have failed to class switch to IgG1.

We sought to confirm that the IgG1+ PC defect was generalizable to other T-dependent immunogens. The B cell response to NP hapten is stereotyped and makes predominant use of the VH186.2 heavy chain (Loh et al., 1983). We therefore immunized both IgM+/− and IgHa/b mice with NP hapten conjugated to rabbit serum albumin (NP-RSA) in order to control for differences in VDJ locus. Although allotype [a] makes a less robust response to NP than allotype [b], we normalized responses of IgD-only cells to wild type allotype [a] cells and found that GC responses at early time points were intact, but IgG1+ PCs and NP-specific IgG1 titers were significantly reduced 7–8 days after immunization (Figure 7G–I). These data suggest that IgD-only follicular B cells are competent to enter the GC, but they have defects in adopting the SLPC fate (or class-switching) in response to T-dependent immunization.

Discussion

We and others previously hypothesized that a major tolerance strategy employed by autoreactive Fo B cells is selective downregulation of IgM. However, since all Fo B cells express a high and invariant amount of surface IgD, it was unclear whether and how loss of IgM expression could affect B cell responses to antigenic stimulation. It was recently reported that IgM, but not IgD, is uniquely responsive to monomeric antigens due to a short and inflexible linker region coupling the Fab and Fc portions of IgM (Übelhart et al., 2015). This could account for selective quiescence of mildly autoreactive B cells to monovalent endogenous antigens. However, the identity and structural characteristics of endogenous antigens are not well understood, and include cell-surface antigens on the surface of red cells and B cells (Quách et al., 2011; Reed et al., 2016). Moreover, recent work from Goodnow and colleagues showed that HEL-specific IgD BCR, when expressed on primary mouse splenocytes, can indeed mobilize calcium in vitro and can mediate gene expression changes in vivo in response to monovalent cognate antigen (Sabouri et al., 2016).

How do our observations help move beyond and reconcile the seemingly contradictory published work on antigen sensing by IgM and IgD? In vitro signaling studies of HEL-specific BCRs by the Jumaa and Goodnow labs differ in several important ways; the former work assesses calcium signaling in pre-B cells that lack Slp-65 expression, while the latter studies splenic B cells with a considerably more mature phenotype. In addition, differences in reagents or strength of stimulus may also contribute to opposing conclusions. Our data suggests that the reduced sensitivity of IgD to bona fide endogenous antigen may lie somewhere between these two extremes; we find that IgD can sense endogenous antigens, just less efficiently than IgM. Consistent with this conclusion, both HEL-specific IgM and IgD BCRs can mediate B cell anergy and drive a common gene expression program in response to chronic high affinity soluble antigen exposure in vivo (Brink et al., 1992; Sabouri et al., 2016).

Here we show that B cells expressing IgD BCR alone inefficiently sense endogenous antigens, and exhibit impaired in vivo cell fate decisions that are dependent upon BCR signal strength. This is epitomized by dramatic skewing of IgD-only B cells away from the B1a and towards the MZ fate in a competitive setting. Importantly, signal transduction mediated by the IgD BCR in response to anti-Igκ stimulation in vitro is intact for all assayed parameters, suggesting that antigen sensing and signal initiation rather than signal transduction downstream of IgD is impaired. We propose that this impaired sensing is conferred by structural properties of IgD, either through direct binding of antigen and signal transduction through the BCR or through differential pairing with co-receptors that directly modulate the BCR signaling pathway.

Characterization of bona fide endogenous antigens that drive Nur77-eGFP expression in the polyclonal repertoire is necessary in order to further dissect the biophysical basis for reduced antigen sensing by IgD in vivo. These antigens may not be exclusively restricted to those that are germline-encoded or generated by host enzymes; rather they may include non-inflammatory antigens taken up through the gastrointestinal tract and recognized by specific BCRs. At least some relevant endogenous antigens may be membrane-bound (Quách et al., 2011; Reed et al., 2016), and recent studies have illustrated the importance of membrane spreading, contraction, and stiffness discrimination in BCR signaling (Fleire et al., 2006; Shaheen et al., 2017). We propose that relevant endogenous antigens are presented in contexts that do not efficiently trigger signaling through the flexible structure of IgD. This might explain the discrepancy between weak responsiveness of IgD to endogenous antigens in vivo and strong signaling in response to crosslinking antibodies in vitro.

An alternative mechanism is that differential clustering of co-receptors with IgD and IgM might influence how BCR signals are integrated in vivo. It has long been proposed that there are not only quantitative, but also qualitative differences between IgM and IgD signaling. Early studies of IgM and IgD signaling in cell lines suggested that kinetics but not quality of signaling triggered by each isotype differed (Kim and Reth, 1995). However, we have been unable to identify a difference in kinetics of BCR signaling in primary IgM- or IgD-deficient B cells in vitro. More recently, studies using TIRF, dSTORM, and PLA (proximity ligation assay) showed that IgD is more densely clustered on the cell surface than IgM, and these distinct IgM- and IgD-containing ‘islands’ are differentially associated with co-receptors such as CD19 in resting and activated B cells (Kläsener et al., 2014; Maity et al., 2015; Mattila et al., 2013). Since function of the ITIM-containing inhibitory co-receptor CD22 depends on its ability to efficiently access the BCR, differences in cluster density of IgM and IgD might render them differentially sensitive to such inhibitory tone (Gasparrini et al., 2016). It is possible that differential association with such co-receptors contributes to differences in the function of the IgM and IgD BCRs in vivo by modulating BCR signal strength, or by selectively perturbing specific downstream signaling events.

In addition to B cell co-receptors that have long been known to directly modulate canonical BCR signaling, a growing list of immunoreceptors expressed on B cells have more recently been shown to require expression of the BCR for optimal function; Becker et al. have shown that expression of the IgD BCR is critical for CXCR4-dependent signaling (Becker et al., 2017). This may influence the biology of B cells expressing only IgM; indeed, CXCR4-deficient B cells exhibit reduced plasma cell migration from spleen to bone marrow (but intact splenic SLPC responses) (Nie et al., 2004). Importantly, defective CXCR4 signaling in IgM-only B cells could not account for reduced Nur77-eGFP expression in IgD-only B cells because reporter expression is insensitive to CXCR4 signaling (Figure 1—figure supplement 1C). Nor would it account for skewed cell fate decisions by IgD-only B cells in competition with wild type B cells, both of which retain intact CXCR4 signaling. B cell responses to TLR4 ligands also require expression of the BCR, but importantly, canonical TLR ligands do not exhibit selective dependence on either the IgM or the IgD isotypes (Figure 1—figure supplement 3A). Finally, recent work establishes a role for the Fc-receptor for IgM in modulating B cell responses, and could therefore play a role downstream of serum IgM that is absent in IgM−/− mice (Nguyen et al., 2017a; Nguyen et al., 2017b). We control for this effect by confirming that all phenotypes in this study are cell-intrinsic and independent of secreted IgM.

Accumulating evidence suggests that strong BCR signals favor SLPC over GC fate (Chan et al., 2009; Nutt et al., 2015; Paus et al., 2006). This is thought to be transcriptionally mediated at least in part by loss of Ets1 expression and induction of Irf4 in a BCR signal-strength dependent manner (Nutt et al., 2015). Here we show that IgM and IgD are each sufficient to drive GC responses, but IgD is less efficient at promoting SLPC fate in response to endogenous antigens, implying that IgDhi IgMlo B cells may similarly be shunted away from SLPC responses. This discrepancy is unlikely to be attributable to differences in antigen capture and presentation by the IgM and IgD BCRs, as GC entry is highly T cell-help dependent. Instead, we provide evidence to suggest that reduced antigen-dependent BCR signals are transduced in vivo by IgD. We show that, in the absence of Lyn, Ets1 downregulation and unswitched PC expansion require the IgM BCR. IgM−/−Lyn−/− mice phenocopy BtkloLyn−/− mice; both strains are protected from Ets1-downregulation, PC expansion, and anti-dsDNA autoantibody production in vivo (Mayeux et al., 2015; Whyburn et al., 2003). This suggests that robust Btk-dependent Ets-1 downregulation in response to endogenous antigens relies upon efficient antigen sensing by the IgM BCR and is important to drive unswitched PC expansion in the absence of Lyn. It will be important to explore whether a similar mechanism accounts for impaired IgG1+ SLPC responses by Lyn-sufficient B cells with low or absent IgM expression.

We propose that the purpose (and consequence) of IgM downregulation on autoreactive B cells is to limit their direct differentiation into SLPCs in response to chronic endogenous antigen stimulation, and prevent secretion of auto-antibodies in the context of an acute humoral immune response (Figure 7—figure supplement 2A). However, IgD is sufficient to drive germinal center differentiation without the contribution of IgM. This has been well-demonstrated both in the present study as well as prior work (Lutz et al., 1998). Indeed, not only are autoreactive IgDhi IgMlo B cells competent to enter the germinal center, they appear to do so with greater efficiency, perhaps due in part to improved survival (Sabouri et al., 2016; Sabouri et al., 2014). It is worth emphasizing that selection pressures and tolerance mechanisms operating within the germinal center must independently ensure that somatic hypermutation both abolishes germ-line-encoded autoreactivity and prevents de novo acquisition of autoreactivity. Goodnow and colleagues recently showed that autoreactive BCRs can indeed be ‘redeemed’ by somatic hypermutation, providing proof-of-principle that entry of autoreactive B cells into the GC does not necessarily pose a risk to the organism (Sabouri et al., 2014).

Why, though, are autoreactive B cells preserved in the periphery and not deleted? Why is IgD necessary at all? One suggestion is that such BCRs are retained in the pre-immune B cell compartment to fill holes in the repertoire, and that IgD is essential for their survival; indeed, it has been shown that loss of IgD expression, with or without compensatory upregulation of IgM, leads to loss of mature naïve B cells in competition (Roes and Rajewsky, 1993; Sabouri et al., 2016). We similarly find that IgD-only B cells have a profound competitive advantage in the follicular B cell compartment relative to IgM-only (but not WT) B cells, implying an obligate survival function for the IgD isotype BCR. This is consistent with a well-appreciated pro-survival function of the BCR as demonstrated by Rajewsky and colleagues (Kraus et al., 2004; Lam et al., 1997). Since IgD expression is needed to keep IgMlo B cells alive, this may explain why IgD expression facilitates a broad dynamic range of IgM expression across the B cell repertoire (Figure 1B,C), allowing B cells to fine-tune the intensity of their SLPC responses according to their autoreactivity. IgD expression on IgMlo cells could serve to mediate antigen capture and participation in T-dependent immune responses, while simultaneously limiting SLPC responses as we show here.

We demonstrate here for the first time that IgD BCRs sense endogenous antigen more weakly than IgM BCRs in vivo, are inefficient at driving SLPC responses to endogenous antigens, and thereby may function to divert autoreactive follicular B cells from direct PC differentiation. We propose that this property of the IgD BCR limits generation of auto-reactive antibodies in the context of immediate humoral immune responses. Taken together with recent work from the Jumaa and Goodnow groups, we propose a unified model of how IgM downregulation in the face of high IgD expression regulates the fate of naïve autoreactive B cells in vivo while retaining their contribution to the mature BCR repertoire.

Materials and methods

Key resources table
Reagent type (species)
or resource
DesignationSource or referenceIdentifiersAdditional information
Strain, strain background
(Mus musculus)
C56BL/6The Jackson Laboratory;
Taconic
JAX:000664; TAC:BCNTac
Strain, strain background
(M. musculus)
Balb/CThe Jackson LaboratoryJAX:000651
Strain, strain background
(M. musculus)
Nur77-eGFPMMRRC UC DavisMMRRC:012015-UCDCharacterized in
PMID:22902503
Strain, strain background
(M. musculus)
IgHELPMID:3261841MD-4
Strain, strain background
(M. musculus)
sHELPMID:3261841ML-5
Strain, strain background
(M. musculus)
IgM-/-PMID:9655395
Strain, strain background
(M. musculus)
IgD-/-PMID:8446604
Strain, strain background
(M. musculus)
CD40L-/-PMID:7964465
Strain, strain background
(M. musculus)
Unc93b 3d/3dPMID:16415873
Strain, strain background
(M. musculus)
TLR7-/-PMID:15034168
Strain, strain background
(M. musculus)
BaffTgMMRRC UC DavisRRID:MMRRC_036508-UCDDescribed in
PMID:15972664
Strain, strain background
(M. musculus)
B6.IgHaThe Jackson LaboratoryJAX:001317B6.Cg-Gpi1a
Thy1a Igha/J
Strain, strain background
(M. musculus)
Lyn-/-PMID:9252121
Strain, strain background
(M. musculus)
MB1-CrePMID:16940357
Strain, strain background
(M. musculus)
MyD88 fl/flPMCID:PMC2847796
Strain, strain background
(M. musculus)
Nr4a1-/-PMID:7624775JAX:006187
Biological sample
(Ovis aires)
Sheep Red Blood CellsRocklandR406-0050
AntibodyAnti-B220-A647
(rat monoclonal)
BD Pharmingen557683(1:200)
AntibodyAnti-B220-APC-e780
(rat monoclonal)
eBioscience47-0452-82(1:400)
AntibodyAnti-B220-FITC
(rat monoclonal)
Tonbo35–0452 U100(1:200)
AntibodyAnti-B220-Pacific Blue
(rat monoclonal)
Tonbo75–0452 U100(1:200)
AntibodyAnti-B220-PE
(rat monoclonal)
BD Pharmingen553090(1:200)
AntibodyAnti-B220-PE-Cy7
(rat monoclonal)
BD Pharmingen552772(1:200)
AntibodyAnti-B220-PerCP-Cy5.5
(rat monoclonal)
Tonbo65–0452 U100(1:200)
AntibodyAnti-CD5-APC
(rat monoclonal)
Tonbo20–0051 U100(1:100)
AntibodyAnti-CD19-PE-Cy7
(rat monoclonal)
Biolegend115520(1:150)
AntibodyAnti-CD19-PerCP-Cy5.5
(rat monoclonal)
BD Pharmingen551001(1:150)
AntibodyAnti-CD21-A647
(rat monoclonal)
Biolegend123424(1:100)
AntibodyAnti-CD21-Pacific Blue
(rat monoclonal)
Biolegend123414(1:100)
AntibodyAnti-CD23-A647
(rat monoclonal)
Biolegend101612(1:200)
AntibodyAnti-CD23-FITC
(rat monoclonal)
BD Pharmingen553138(1:200)
AntibodyAnti-CD23-Pacific Blue
(rat monoclonal)
Biolegend101616(1:100)
AntibodyAnti-CD23-PE
(rat monoclonal)
BD Pharmingen553139(1:200)
AntibodyAnti-CD23-PE-Cy7
(rat monoclonal)
eBioscience25-0232-82(1:200)
AntibodyAnti-CD69-APC
(hamster monoclonal)
Biolegend104514(1:100)
AntibodyAnti-CD69-PE-Cy7
(hamster monoclonal)
Tonbo60–0691 U100(1:100)
AntibodyAnti-CD86-Pacific Blue
(rat monoclonal)
Biolegend105022(1:100)
AntibodyAnti-CD93 (AA4.1)-PE-Cy7
(rat monoclonal)
Biolegend136506(1:100
AntibodyAnti-CD138-PE
(rat monoclonal)
Biolegend142504(1:100)
AntibodyAnti-CD138-PE-Cy7
(rat monoclonal)
Biolegend142513(1:100)
AntibodyAnti-CXCR4-Biotin
(rat monoclonal)
BD Pharmingen551968(1:100)
AntibodyAnti-ETS1
(rabbit monoclonal)
Epitomics; abcamEPI:3123–1; AB:109212(1:10,000); concentrated
lot from L.A. Garrett-Sinha
AntibodyAnti-Fas-PE-Cy7
(hamster monoclonal)
BD Pharmingen557653(1:200)
AntibodyAnti-GAPDH
(mouse monoclonal)
EMD MilliporeAB2302(1:2000)
AntibodyGL-7-A647
(rat monoclonal)
BD Biosciences561529(1:400)
AntibodyAnti-IgA-Biotin
(rat monoclonal)
Biolegend407003(1:400)
AntibodyAnti-IgD (goat
polyclonal serum)
MD Biosciences2057001(1:50-1:400)
AntibodyAnti-IgD[a]-Biotin
(mouse monoclonal)
BD Pharmingen553506(1:300)
AntibodyAnti-IgD-APC-e780
(rat monoclonal)
eBioscience47-5993-80(1:500)
AntibodyAnti-IgD-HRP
(rat monoclonal)
American Research Products09-1008-4(1:2000)
AntibodyAnti-IgD-Pacific Blue
(rat monoclonal)
Biolegend405712(1:300)
AntibodyAnti-IgD-PE
(rat monoclonal)
eBioscience12-5993-82(1:800)
AntibodyAnti-IgG1[a]-Biotin
(mouse monoclonal)
BD Pharmingen553500(1:300)
AntibodyAnti-IgG1[b]-Biotin
(mouse monoclonal)
BD Pharmingen553533(1:300)
AntibodyAnti-IgG2a[a]-Biotin
(mouse monoclonal)
BD Biosciences553502(1:1000)
AntibodyAnti-IgG2a[b]-Biotin
(mouse monoclonal)
BD Biosciences553504(1:1000)
AntibodyAnti-IgG2c-Biotin
(goat polyclonal)
SouthernBiotech1079–08(1:1000)
AntibodyAnti-Igκ (goat polyclonal)SouthernBiotech1050–011–20 μg/mL
AntibodyAnti-Igκ-F(ab')2
(goat polyclonal)
SouthernBiotech1052–011–20 μg/mL
AntibodyAnti-Igκ-FITC
(goat polyclonal)
SouthernBiotech1050–02(1:300)
AntibodyAnti-Igκ-FITC
(rat monoclonal)
BD Pharmingen550003(1:300)
AntibodyAnti-Igk-PerCP-Cy5.5
(rat monoclonal)
BD Pharmingen560668(1:300)
AntibodyAnti-Igλ-FITC
(rat monoclonal)
BD Pharmingen553434(1:300)
AntibodyAnti-Igλ-PE
(rat monoclonal)
Biolegend407307(1:300)
AntibodyAnti-IgM-F(ab')2
(goat polyclonal)
Jackson ImmunoResearch115-006-0201–20 μg/mL
AntibodyAnti-IgM-HRP
(goat polyclonal)
SouthernBiotech1020–05(1:2000); secondary
for ELISA
AntibodyAnti-IgM[a]-Biotin
(mouse monoclonal)
Biolegend408603(1:100)
AntibodyAnti-IgM[a]-FITC
(mouse monoclonal)
Biolegend408606(1:100); (1:400)
for plasma cell
AntibodyAnti-IgM[a]-PE
(mouse monoclonal)
BD Pharmingen553517(1:100); (1:400)
for plasma cell
AntibodyAnti-IgM[b]-Biotin
(mouse monoclonal)
Biolegend406204(1:100)
AntibodyAnti-IgM[b]-PE
(mouse monoclonal)
Biolegend406208(1:100); (1:400)
for plasma cell
AntibodyAnti-IgM-APC
(rat monoclonal)
eBioscience17-5790-82(1:100)
AntibodyAnti-MHC-2-APC
(rat monoclonal)
Tonbo20–5321 U100(1:1000)
AntibodyAnti-Mouse-IgG(H + L)-HRP
(goat polyclonal)
SouthernBiotech1031–05(1:5000); secondary
for western blots
AntibodyAnti-Nur77-PE
(mouse monoclonal)
eBioscience12-5965-80(1:100)
AntibodyAnti-pERK
(rabbit monoclonal)
Cell Signaling Technology4377S(1:80)
AntibodyAnti-pS6
(rabbit monoclonal)
Cell Signaling Technology4856S(1:100)
AntibodyAnti-Rabbit-IgG-APC
(donkey polyclonal)
Jackson ImmunoResearch711-136-152(1:100); secondary
for pERK/pS6
AntibodyAnti-Rabbit-IgG(H + L)-HRP
(goat polyclonal)
SouthernBiotech4050–05(1:5000); secondary
for western blots
Sequence-based reagentNr4a1 forward primerElim Biopharmgcctagcactgccaaattg
Sequence-based reagentNr4a1 reverse primerElim Biopharmggaaccagagagcaagtcat
Sequence-based reagentGAPDH forward primerElim Biopharmaggtcggtgtgaacggatttg
Sequence-based reagentGAPDH reverse primerElim Biopharmtgtagaccatgtagttgaggtca
Peptide, recombinant
protein
NP-RSABiosearchN-5054–100Conj. ratio: 10
Peptide, recombinant
protein
NP-BSABiosearchN-5050H-100Conj. ratio: 23
Peptide, recombinant
protein
Streptavidin-HRPSouthernBiotech7100–05(1:5000)
Peptide, recombinant
protein
Streptavidin-APCTonbo20–4317 U500(1:100-1:400)
Peptide, recombinant
protein
Streptavidin-Pacific BlueLife TechnologiesS11222(1:200)
Peptide, recombinant
protein
Streptavidin-PerCP-Cy5.5BD Pharmingen551419(1:400)
Peptide, recombinant
protein
Streptavidin-FITCBiolegend405202(1:100-1:200)
Peptide, recombinant
protein
CXCL12Peprotech300-28A
Peptide, recombinant
protein
NP-PEBiosearchN-5070–1(1:400)
Chemical compound, drugPoly-L-LysineSigmaP2636-100MG100 μg/mL in 0.1 M Tris-HCl pH7.3
Chemical compound, drugPoly dA-dTSigmaP0883-50UN0.2 U/mL in 0.1 M Tris-HCl pH7.3
Chemical compound, drugLPSSigmaL8274
Chemical compound, drugCpGInvivoGentlrl-1826b
Chemical compound, drugPam3CSK4InvivoGentlrl-pms
Commercial assay, kitIndo-1, AMLife TechnologiesI-1223(1:1000)
Commercial assay, kitLive/Dead Fixable Near-IR
Dead Cell Stain Kit
InvitrogenL10119(1:1000)
Commercial assay, kitECL Luminol;
Oxidizer Reagents
Perkin Elmer0RT2751; 0RT2651
Commercial assay, kit3,3',5,5'-Tetramethylbenzidine,
Slow Kinetic Form
SigmaT4319-100MLELISA substrate
Software, algorithmFlowJoFlowJo LLCVersion 9.9.4
Software, algorithmPrismGraphPadVersion 7.0b
Software, algorithmCanopyEnthoughtVersion 1.4.1.1975
Software, algorithmBinning program in Figure 1FOtherSource code provided
in this publication
OtherBD Microtainer Capillary
Blood Collector
Fisher365967
OtherPtC-Rhodamine
(DOPC/CHOL Liposomes)
FormuMaxF60103F-R(1:1000); used in
PtC-specific B1a staining
OtherNuPAGE 4–12%
Bis-Tris Protein Gels
InvitrogenNP0335BOX
OtherImmobilon-P PVDF
Membrane
EMD MilliporeIPVH00010
OtherAssay Plate, 96 Well, No Lid, VinylCostar2595Used for ELISA

Mice

Nur77-eGFP mice and Lyn−/− mice have been previously described (Chan et al., 1997; Zikherman et al., 2012). BAFF Tg mice were originally generated in the Nemazee lab, and express BAFF under the control of the myeloid promoter human CD68 (Gavin et al., 2005) (source: MMRRC UC Davis). IgD-/- and IgM−/− mice were previously described, and the former were generously shared by Dr. Hassan Jumaa (Lutz et al., 1998; Nitschke et al., 1993). TLR7−/−, CD40L−/−, Unc93b13d/3d, Nr4a1−/−, and MB1-Cre MyD88fl/fl mice were previously described (Hobeika et al., 2006; Hou et al., 2008; Lee et al., 1995; Lund et al., 2004; Renshaw et al., 1994; Tabeta et al., 2006). All strains were backcrossed to the C57BL/6 genetic background for at least six generations. Mice were used at 5–12 weeks of age for all functional and biochemical experiments unless otherwise noted. Germ-free (GF) and specific pathogen free (SPF) C57BL/6 mice used for direct comparison were purchased from Taconic and Jackson Laboratory and kept in microisolators under GF or SPF (ISOcage P - Bioexclusion System, Tecniplast) conditions. All GF mice were housed in closed caging systems and provided with standard irradiated chow diet, acidified water and housed under a 12 hr light cycle; 7-week-old males mice were used. GF and control mice were generously provided by Drs. Sergio Baranzini and Anne-Katrin Proebstel at UCSF. All other mice used in our studies were housed in a specific pathogen free facility at UCSF according to the University Animal Care Committee and National Institutes of Health (NIH) guidelines.

Antibodies and reagents

Fluorescently-conjugated or biotin-conjugated antibodies to B220, CD5, CD19, CD21, CD23, CD69, CD86, CD93 (AA4.1), CD95 (Fas), CD138, CXCR4, IgA, IgM, IgM[a], IgM[b], IgD, IgD[a], IgG1[a], IgG1[b], Igκ, Igλ, GL-7, MHC Class II, and fluorescently-conjugated streptavidin were from Biolegend, eBiosciences, BD Biosciences, Tonbo, or Life Technologies. NP-PE was from Biosearch Technologies. 100 nm Rhodamine PtC liposomes were from FormuMax. Antibodies for intra-cellular staining, pErk Ab (clone 194g2) and pS6 Ab (2F9), were from Cell Signaling Technologies, and Nur77 Ab (clone 12.14) conjugated to PE was from eBioscience. Goat anti-mouse IgM F(ab’)2 was from Jackson Immunoresearch. Goat anti-mouse Igκ and goat F(ab’)2 anti-mouse Igκ were from Southern Biotech. Anti-IgD was from MD Biosciences. CXCR4 ligand (CXCL12/hSDF-1α) was from Peprotech. LPS (Cat. L8274) was from Sigma. CpG DNA (ODN 1826 Biotin; Cat. tlrl-1826b) and Pam3CSK4 (Cat. tlrl-pms) were from InvivoGen.

Flow cytometry and data analysis

Cells were stained with indicated antibodies and analyzed on a Fortessa (Becton Dickson) as previously described (Hermiston et al., 2005). Data analysis was performed using FlowJo (v9.7.6) software (Treestar Incorporated, Ashland, OR). Statistical analysis and graphs were generated using Prism v6 (GraphPad Software, Inc). Figures were prepared using Illustrator CS6 v16.0.0. Median Igk levels (Figure 1F) were calculated for 200 equally sized ‘bins’ spanning the Nur77-eGFP spectrum using Canopy v1.4.1 (Enthought); source code is provided, and parameters can be modified to condense any 2D FACS plot for comparisons of multiple samples.

Statistics and replicates

We define biological replicates as independent analyses of cells isolated from different mice of the same genotype, and we define technical replicates as analyses of cells isolated from the same mouse and used in the same experiment. When technical replicates were used, we averaged them to calculate a value for the biological replicate. All reported values and statistics correspond to biological replicates only, and all ‘n’ values reported reflect the number of biological replicates. Wherever MFIs are directly compared, we collected all samples in a single experiment to avoid potential error that could arise from fluctuations in our flow cytometers. In some instances, MFI ratios were compiled from different experiments for statistical purposes, but qualitative findings were consistent experiment to experiment. As our panels reproducibly generated flow plots with well-defined populations, population sizes were calculated from data compiled from different experiments. When comparing two groups, we employed Welch’s t test, which is more stringent than Student’s t test and does not assume that the groups have the same standard deviation. We used paired difference t tests when studying parameters in allotype-heterozygous mice that are sensitive to cell-extrinsic factors (e.g. dose of immunogen or severity of autoimmune disease). When comparing three genotypes (e.g. receptor levels on WT vs. IgM−/−vs. IgD−/− B cells), we used one-way ANOVA and Tukey’s multiple comparison test to calculate p values.

Intracellular Nur77, pErk and pS6 staining

Ex vivo or following stimulation, cells were fixed in 2% paraformaldehyde, permeabilized with 100% methanol, and stained with anti-Nur77, anti-pErk, or anti-pS6 followed by lineage markers and secondary antibodies if needed.

Intracellular plasma cell staining

Cells were fixed in 2% paraformaldehyde and permeabilized in BD Perm/Wash. Stains for antibody isotypes and allotypes were prepared in BD Perm/Wash.

Intracellular calcium flux

Cells were loaded with 5 μg/mL Indo-1 AM (Life Technologies) and stained with lineage markers. Cells were rested at 37 C for 2 min, and Indo-1 fluorescence was measured immediately prior to stimulation to calculate basal calcium, and for at least two minutes following addition of stimulus.

In vitro B cell stimulation

Splenocytes or lymphocytes were harvested into single cell suspension, subjected to red cell lysis using ACK buffer, and plated at a concentration of 7.5 × 105 cells/200 μL in round bottom 96 well plates in complete RPMI media with stimuli for 18 hours prior to analysis. In vitro cultured cells were stained using fixable near IR live/dead stain (Life technologies) per manufacturer’s instructions.

ELISA

Serum antibody titers for total IgM, total IgD, anti-dsDNA IgG2a, and NP-specific IgG1 were measured by ELISA. For total IgM and total IgD, 96-well plates (Costar) were coated with 1 μg/mL anti-IgM F(ab’)2 (Jackson) or 1 μg/mL anti-IgD (BD 553438), respectively. Sera were diluted serially, and total IgM and total IgD were detected with anti-IgM-biotin (eBioscience) and SA-HRP (Southern Biotech) or anti-IgD-HRP (American Research Products). dsDNA plates were generated by serially coating plates with 100 μg/mL poly-L-lysine (SigmaAldrich) and 0.2 U/mL poly dA-dT (SigmaAldrich) in 0.1 M Tris-HCL pH 7.6. Sera were diluted serially on dsDNA plates, and autoantibodies were detected with IgG2[a]-biotin, IgG2a[b]-biotin, and SA-HRP. NP plates were generated by coating 96-well plates with 1 μg/mL NP-BSA (Biosearch, conjugation ratio 23) in PBS. Sera from NP-RSA immunized and unimmunized mice were diluted serially, and NP-specific IgG1 was detected with IgG1[a]-biotin, IgG1[b]-biotin, and SA-HRP. All ELISA plates were developed with TMB (Sigma) and stopped with 2N sulfuric acid. Absorbance was measured at 450 nm. For total IgM, total IgD, and dsDNA IgG2a, OD values or OD ratios were calculated and displayed. For NP-specific IgG1, titers were calculated at OD = 0.2 and normalized such that the average IgG1[a]/IgG1[b] ratio in immunized Balb/c-C57BL/6 F1 mice equals 1.0.

B cell purification and western blots

Splenic B cells were purified by negative selection with a MACS kit according to manufacturer’s protocol (Miltenyi Biotech, cat# 130-090-862). Purified B cells were lysed in 2N 4X SDS and boiled at 95 C for 5 min. 500,000 cells per lane were loaded onto NuPAGE 4–12% Bis-Tris gels (Invitrogen NP0335), run for 50 min at 200V, and transferred onto PVDF membranes using an XCell II Blot Module (Invitrogen). Membranes were blocked with 3% BSA. Proteins were detected with rabbit monoclonal anti-Ets1 (Epitomics, custom lot provided by Lee Ann Garrett-Sinha), mouse anti-mouse GAPDH (Santa Cruz Biotechnology 32233), and HRP-conjugated secondary antibodies goat anti-rabbit IgG (H + L) and goat anti-mouse IgG (H + L) (SouthernBiotech). Membranes were developed with Western Lightning Plus ECL (Perkin Elmer 0RT2651 and 0RT2751) and imaged using a ChemiDoc Touch Imaging System (Bio-Rad). Quantifications were performed using Image Lab v. 5.2.14 (Bio-Rad).

Immunizations

Mice were immunized i.p. with 200 uL of 10% sheep red blood cells (Rockland R406-0050) diluted in PBS. Mice were sacrificed 5 days after immunization, and serum and spleens were harvested. Mice were immunized i.p. with 100 ug of NP-conjugated rabbit serum album (NP-RSA, Biosearch N-5054–10, conjugation ratio 10) prepared in PBS with Alhydrogel 1% adjuvant (Accurate Chemical and Scientific Corp. 21645-51-2). NP-RSA immunized mice were sacrificed 7–8 days following immunization, and serum and spleens were harvested. To ensure that our calculated values accurately reflect the magnitude and variability of immune responses induced by immunization, we analyzed all mice that had larger plasma cell and germinal center compartments than unimmunized mice. One IgM+/− mouse immunized with SRBCs (Figure 7E–F) was excluded because it did not display an expanded plasma cell compartment; no other mice were excluded from analysis.

Quantitative PCR

Splenocytes were lysed in TRIzol (Invitrogen) and stored at −80 C. cDNA was prepared using a SuperScriptIII kit (Invitrogen). Quantitative PCR reactions were run on a QuantStudio 12K Flex thermal cycler (Applied Biosystems) with FastStart Universal SYBR Green Master Mix (Roche). Nr4a1 (forward: gcctagcactgccaaattg; reverse: ggaaccagagagcaagtcat) and GAPDH (forward: aggtcggtgtgaacggatttg; reverse: tgtagaccatgtagttgaggtca) primers were used at 250 nM each.

References

  1. 1
  2. 2
  3. 3
    Role of the mu immunoglobulin heavy chain transmembrane and cytoplasmic domains in B cell antigen receptor expression and signal transduction
    1. JH Blum
    2. TL Stevens
    3. AL DeFranco
    (1993)
    The Journal of Biological Chemistry 268:27236–27245.
  4. 4
    Enhanced B-1 cell development, but impaired IgG antibody responses in mice deficient in secreted IgM
    1. M Boes
    2. C Esau
    3. MB Fischer
    4. T Schmidt
    5. M Carroll
    6. J Chen
    (1998)
    Journal of Immunology 160:4776–4787.
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
    Induction of early response genes by cross-linking membrane Ig on B lymphocytes
    1. PR Mittelstadt
    2. AL DeFranco
    (1993)
    Journal of immunology 150:4822–4832.
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
  70. 70
  71. 71
  72. 72
  73. 73
  74. 74
  75. 75
  76. 76
  77. 77
  78. 78

Decision letter

  1. Facundo D Batista
    Reviewing Editor; Ragon Institute of MGH, MIT and Harvard, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for submitting your work entitled "Differential sensing of endogenous antigens and control of B cell fate by IgM and IgD B cell receptors" for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. As we feel the additional work needed to address the issues raised by the reviewers would take more than two months to complete, we are returning your submission to you now in case you wish to submit elsewhere for speedy publication. However, if you address these points and wish to resubmit your work to eLife, we would be happy to look at a revised paper. Please note that it would be treated as a new submission with no guarantees of acceptance.

Reviewer #2:

The manuscript entitled "Differential sensing of endogenous antigens and control of B cell fate by IgM and IgD B cell receptors" by Noviski et al. follows a series of manuscripts started back in 2012 with the publication of the Nur77-GFP mice (Zikherman et al., 2012). Nur77-GFP mice have thus emerged as a useful tool to study B cell antigen receptor (BCR) and T cell antigen receptor (TCR) signaling in vivo.

In the present manuscript, the authors aimed to answer the long-standing question why mature B cells express two BCR isotypes, IgM and IgD, with identical antigen-specificity, and how these two isotypes impose B cell tolerance in vivo.

A major issue for me is the clarity of the present manuscript. Although the initial question is clear and well defined, a good rational flow along the experiments, figures and the text is missing. Also the way the data are presented is highly confusing and prevents comparison between figures. The manuscript clearly needs an effort to synthesize and clarify the message.

The main message of the manuscript is that IgD-BCR signal strength is reduced when compared to IgM-BCR signal strength only in vivo upon recognition of autoantigens driving differentiation (Figure 1 and MZ versus B1a cells under competitive conditions, Figure 4) or for the generation of SLPC (Figure 7). In contrast, in vitro (Figure 2), upon immunization (Figure 7) and upon recognition of auto-antigens in the Lyn−/− background (Figure 5A-B) the signals of IgD-BCR are as competitive as the signals from IgM-BCR. Taken together, these results are confusing and point out to microenvironment, cell-differentiation stage, co-receptor expression and crosstalk between receptors in a given situation as reasons modulating the signals through the BCR. To conclude that "IgD-BCR sensing antigen is impaired" is an over-interpretation of the results, a not-justified generalization and neglects a major part of the results exposed in this manuscript.

1) The whole study is based on the observation that Follicular B cells in the spleen are GFP+, indicating that they have been activated at some point during development (T2/T3 stage) by any of the receptors that induce the activation of the Nur77 locus (BCR, CD40, TLR9, TLR4…). The authors claimed already back in the first publication (Zikherman et al., 2012) that GFP positivity results from BCR signals upon encounter with autoantigens because GFP positivity is significantly reduced in BCR transgenic settings (IgHEL Tg), and again increased in the presence of antigen (IgHEL Tg/sHEL Ag Tg). However, if the encounter with autoantigens drives GFP positivity, why are there not GFP+ cells within the immature cells in the BM? At any given moment, some immature B cells are signaling through the BCR during B cell selection, one should expect a proportion of GFP+ cells in the BM. Moreover, it has been broadly demonstrated that BCR driven signals are crucial for B cell development, why are there not GFP+ B cells during B cell development? Is it possible that GFP positivity is the result of a combination of different signals happening from the T2/T3 stage on, including also BCR signals in the IgHEL Tg/sHEL Ag Tg settings? The authors discard CD40; TLR3,7,9 (Unc93b1 mice, this information should be found at least in the figure legend); TLR7; IL4; BAFF and CXCR4 (Figure 1—figure supplement 1 and Zikherman et al., 2012). However, an obvious candidate is TLR4, no data are shown using either LPS stimulation in vitro, TLR4−/− mice, Myd88−/− mice or even better by housing the mice in pathogen free/sterile conditions. Indeed amount of mRNA of TLR4 increases from the T1 stage on to get a maximum at the Fo stage. Other receptors which levels of mRNA are upregulated from the transitional stage on are TLR6, TLR1, CD22….

2) In line with the comment before, why most of Fo cells (around 90%) are GFP+ also in the WT settings? This high percentage is expected in the IgHEL Tg/sHEL Ag Tg setting. However, in the WT situation, if GFP positivity is the consequence of autoantigen encounter, this would mean that most of B cell emigrants from the BM carry autoreactive receptors. Do the authors suggest that tolerance in mice is established mainly in the peripheral checkpoint (T2/T3)? This will be in contrast with the data in humans in which both checkpoints reduced the percentage of self-reactive BCR to half, and only 40% of new emigrants carry self-reactive BCRs, percentage that is reduced to 20% after the peripheral checkpoint.

Taken together the last two points, it seems true that autoantigens are sufficient to drive Nur77-GFP expression as claimed by the authors (IgHEL Tg/sHEL Ag Tg setting already reported in 2012), but I am not convinced at this stage that all GFP+ cells in the Fo stage are GPF+ because BCR signaling as consequence of autoantigen encounters. I rather think that GFP accumulation is the result of several activation signals from diverse receptors, including the BCR but not exclusively.

3). Because an inverse correlation between GFP levels and IgM expression is observed, the authors propose that IgM is downregulated as consequence of autoantigen encounter and therefore, activation of autoreactive B cells is prevented. The slope that correlates IgM and GFP expression in the IgM-only mice is reduced compared to the slope in WT, why is this? In other words, why in the absence of IgD, IgM cannot be optimally downregulated? In this the case also upon anti-IgM stimulation ex vivo?

In contrast to IgM, IgD levels are invariant with GFP expression, this would mean that in IgD mice GFP high cells could be easily re-activated driving autoimmunity, is this the case? The authors claim that these cells cannot be optimally activated, but the total GFP levels are the same, proving that activation was equally strong. Can IgD be downregulated upon stimulation ex vivo in the IgD-only B cells?

4) In Figure 1F, the authors showed that for a given GFP signal more IgD-BCRs are needed than IgM-BCRs, proposing that IgD-BCRs do not signal properly in vivo (supposedly, in response to antigens). Taking in consideration recent data regarding the nanoscale distribution of IgM and IgD-BCRs in protein islands, how are the levels of co-receptors such as CD19, CD22, CD20 in WT, IgM-only and IgD-only? Are these co-receptors excluded or included in the IgM protein island in the absence of IgD? Are these coreceptors modulating differently IgM-BCR and IgD-BCR signals? Differences in co-receptor levels and/or composition of these protein islands might explain why IgD stimulation in vitro, which happens solely by the BCR, is strong (Figure 2).

5) What exactly does this mean for the authors, "sensing of endogenous antigen"? or just "sensing of antigen"?

6) It is unclear to me the significance of the experiments shown in Figure 3. Both B1a and MZ B cells express preferentially IgM-BCRs. Are they generated to the same extent in IgD-only mice? If yes, is this not probing that IgD-BCR in vivo signals as efficient as IgM-BCR at least to drive differentiation of these cells? However, IgD-only B1a and M Z cells showed reduced levels of Nur77-GFP ex vivo. Again, how is the expression of co-receptors and especially of innate-like receptors in IgD-only versus WT or IgM-only B1a and MZ B cells? In these two populations all cells are GFP+ cells, it seems counterintuitive that all cells that populate the periphery are autoreactive… In the IgD-only B1a cells there is indeed a GFP- population, the authors should compare the repertoire of these GFP- cells to GFP+ cells to support they claims.

7) As cited by the authors, a link between BCR signal strength and cell fate has been established some years ago, in which weak BCR signals will allow MZ B cell generation but prevent B1 cell generation. In contrast, strong signals might favor B1 generation/maintained. No major differences were observed in B1 or MZ cell generation between IgM-only, IgD-only and WT. However, under competitive conditions, IgD-only B cells fill up preferentially the MZ B cell compartment and are unable to differentiate to B1a cells. How are these cells phenotypically? MZ are BM-derived while B1a cells are mainly fetal liver derived and long lived. It has been proposed that in B2 cells IgD provide survival signals. Is it possible that B1 cells required survival signals via IgM and that IgD cannot provide survival signals or self-renewal signals? Is it really a matter of signal strength or of signal quality? How is this piece of data related to autoantigen recognition?

8) Which bibliography supports the following sentence "endogenous auto-antigens may differ significantly from model antigens"? The authors used HEL as model antigen and as endogenous auto-antigen, is the model antigen HEL not differing from endogenous auto-antigens?

9) Similar to the results obtained in vitro, the results in the Lyn−/− mice support that both IgM and IgD-BCR provide sufficient signals for B cell activation (CD86, CD69) and for GC differentiation against the theory of the authors that "IgD-BCR sensing" of antigen is impaired.

10) Regarding the production of autoantibodies in Figure 5, why the production of anti-dsDNA antibodies is also increased in the WT cells that co-exist with the IgM-only B cells? Indeed, at 25 weeks the basal levels are elevated in all the mice compared to all the other situations. It is possible that other receptors, such as germline-encoded receptors, cooperate only with IgM-BCR to break tolerance and not with IgD? This will be independent of BCR strength but dependent of crosstalk between receptors. This crosstalk has been described at least for TLR4/IgM-BCR and RP105/IgM-BCR.

11) In Figure 6A, how are the levels of Ets1 in IgM-only? And in Figure 6B, how is intracellular IgD in WT or IgM-only mice? I do not agree with the interpretation of Figure 6E, IgM−/−Lyn−/− have elevated IgD serum levels when compared to Lyn−/−.

12) Why are IgD-only less competitive to enter the GC but disfavored to produced SLPC responses? Is this because BCR-strength as proposed by the authors or by the effect of isotype-specific interactions with co-receptors?

Reviewer #3:

In this manuscript the authors present an extensive body of work comparing the development and responsiveness of polyclonal B cells capable of expressing either IgM or IgD BCRs prior to undergoing class switching. Whilst there is an extensive literature comparing the functional capabilities of these two (normally) co-expressed classes of BCR, this particular study examines a variety of readouts not previously examined, in particular the apparent sensing of endogenous self-antigens. Despite some concern about the different strain origins of the original IgM and IgD KO lines, the authors appear to control for this adequately throughout the study.

Overall I found the experiments to be well-performed and thoughtfully interpreted. The data presented here go some way to resolve recent controversies in the field and as such provide a valuable source of new information to help resolve the longstanding question of why naive, resting B cells co-express IgM and IgD BCRs.

I have just a couple of specific points:

1) In the SRBC-induced responses, did the IgM KO mice possess a general deficiency in PC production apart from IgG1-switched PCs e.g. in producing unswitched PCs?

2) Similarly, was the production of switched (e.g. IgG1+) GC B cells different between the IgM-only and IgD-only B cells?

It would be important to add this information to clarify the nature of the defect involved.

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for submitting your work entitled "IgM and IgD B cell receptors differentially respond to endogenous antigens and control B cell fate" for consideration at eLife. Your article has been favorably evaluated by Tadatsugu Taniguchi (Senior Editor) and three reviewers, one of whom is a member of our Board of Reviewing Editors.

This manuscript provides novel insights into different functions between IgM and IgD BCRs. There are some remaining issues that need to be addressed before acceptance, as outlined below in the separate reviews. Specifically, we request that you recognize that the Nur77-GFP signals in B cells might be the result of BCR stimulation by a broader collection of non-inflammatory antigens in vivo, not only self-antigens. They might be globally named as endogenous antigens but not exclusively self-antigens, or endogenous self-antigens.

Specific comments.

Reviewer #1:

The authors have satisfactorily addressed most the reviewers’ comments and the manuscript has improved.

Reviewer #2:

The authors have satisfactorily addressed all the points I raised in my initial review.

Reviewer #3:

I have read with enthusiasm and interest the revised version of the manuscript entitled "Differential sensing of endogenous antigens and control of B cell fate by IgM and IgD B cell receptors" by Noviski et al. I appreciate the determination of the investigators to address the issues that arose during the revision process and thank the authors for their efforts.

One of my major points in the previous revision round was whether the Nur77-GFP positive B cells were indeed positive because activation of the BCR by self-antigens (as claimed by the authors) or by a combination of signals by the BCR and/or additional receptors expressed on the surface of B cells.

The authors have appropriately answered part of this question by showing that some of these receptors do not activate expression of Nur77 (TLR3, TLR7, TLR9, CXCR4). Other receptors, such as TLR4, TLR9 and TLR1/2 do activate expression of Nur77-GFP (Figure Author response image 6). However, the authors demonstrate here (Figure 1—figure supplement 3) that upregulation of Nur77-GFP by these receptors is equal between IgM-only and IgD-only B cells (although the concentration of ligands chosen are too high). The authors also show in the new version, that endogenous levels of Nur77 transcripts in B cells are equally elevated between Germ-free mice and SPF, arguing that Nur77 signals in B cells do not from microbial stimulation (although these signals might come from the endotoxin present in the chow, or did the authors used endotoxin-free chow?). Moreover, the levels of endogenous Nur77 transcripts were equally elevated between MyD88KO and MyD88WT peritoneal B1a cells. Here, I would like to ask why peritoneal B1a cells were chosen and not splenic B cells as in Figure 1—figure supplement 1D-E?

However, I do not agree that the Nur77-GFP signals in B cells are the consequence of self-antigen (self-antigen defined by an antigen encoded by the DNA of the self-organism) stimulation by the BCR. I disagree with the authors using self-antigen (subsection “Cell-intrinsic skewing of B cell development by the IgM and IgD BCRs”, first paragraph), endogenous auto-antigen (subsection “Either IgD or IgM BCR is sufficient to mediate polyclonal B cell activation and germinal center differentiation in Lyn−/− mice”, first paragraph) and antigen encounters in vivo (subsection “IgD BCRs sense endogenous antigens less efficiently than IgM BCRs”) as synonymous. I fully agree that Nur77-GFP signals are upregulated upon BCR stimulation by antigen and that antigen is necessary to drive Nur77-GFP expression in vivo. But I do not agree that self-antigen is necessary to drive Nur77-GFP expression in vivo. My arguments are:

• Nur77-GFP signals are highly upregulated when B cells leave the sterile/defined microenvironment of the BM. Nur77-GFP expression timely occurs when cells encounter "antigens" in the periphery, and those antigens activate the BCR. But these antigens are not necessarily self-antigens, they can be food-antigens, LPS (from the chow, antibodies against LPS are broadly recognized), and other non-pathological or non-inflammatory antigens recognized by the BCR. These antigens might activate B cells to upregulate Nur77, but in the absence of T cell help or an inflammatory milieu will not drive an immune response.

• The results with the IgHEL BCR Tg mice (Author response image 1) are fully compatible with the above-described scenario. These mice express a clonal BCR not able to recognize any self-antigen, but also, not able to recognize any of the possible non-inflammatory antigens that will be encountered in the periphery, for this reason the signal of Nur77 is lower. This interpretation is also compatible with higher level of Nur77-GFP in HEL Tg cells exposed to cognate HEL antigen that in WT. Related to this, the results show in Author response image 6 are also compatible with this scenario since Nur77-GFP upregulation is reduced in IgHEL BCR Tg mice in respond to LPS (small reduction), CpG (clear at lower concentration), and to Pam3CSK4 (visible at both concentrations). A possible interpretation of these results is that the HEL Tg BCR is not able to recognize epitopes in those substances that, together with the corresponding TLR, aid to activate Nur77-GFP in B cells.

• The results with the VH3H9 Tg mice (Author response image Figure 3) are also fully compatible with the above-described scenario, since the κ+ cells (not recognizing endogenous DNA) also up-regulate (even to the same levels than λ+ cells) Nur-77 upon leaving the BM. Most probably because they encounter non-inflammatory antigens in the periphery.

Taken together, I fully agree that the present manuscript is improved upon revision and most of my requests have been clarify. However, I still request that the authors recognize that the Nur77-GFP signals in B cells might be the result of BCR stimulation by a broader collection of non-inflammatory antigens in vivo, not only self-antigens. They might be globally named as endogenous antigens but not exclusively self-antigens, or endogenous self-antigens. The main conclusion of the manuscript is unaltered, namely that IgD-BCR signals differently upon encounter of non-inflammatory antigens in vivo, but to narrow this conclusion only to self-antigens is not proven and not justified.

https://doi.org/10.7554/eLife.35074.051

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

Reviewer #2:

The manuscript entitled "Differential sensing of endogenous antigens and control of B cell fate by IgM and IgD B cell receptors" by Noviski et al. follows a series of manuscripts started back in 2012 with the publication of the Nur77-GFP mice (Zikherman et al., 2012). Nur77-GFP mice have thus emerged as a useful tool to study B cell antigen receptor (BCR) and T cell antigen receptor (TCR) signaling in vivo.

In the present manuscript, the authors aimed to answer the long-standing question why mature B cells express two BCR isotypes, IgM and IgD, with identical antigen-specificity, and how these two isotypes impose B cell tolerance in vivo.

A major issue for me is the clarity of the present manuscript. Although the initial question is clear and well defined, a good rational flow along the experiments, figures and the text is missing.

We have made manuscript edits to clarify the flow of logic in our Introduction (last paragraph), and to transition more smoothly between sections throughout the manuscript.

The structure of the paper reflects the nature of the question we sought to address, namely differences between the response of the IgM and IgD BCRs to a diverse repertoire of bona fide endogenous antigens in vivo. Rigorously studying in vivo antigen sensing in the context of a polyclonal repertoire requires examining a broad array of BCR-dependent physiological readouts. The paper begins by establishing the specificity of the Nur77-eGFP reporter as a read-out of endogenous antigen-dependent signaling in B cells (below in the rebuttal we review new data establishing specificity and validity of this reporter). We next introduce this reporter into mice lacking either IgM or IgD to determine how endogenous antigens-dependent signals are transduced by these BCR isotypes. We observe that IgD drives less reporter expression per receptor than IgM, and we proceed to rule out differences in downstream signal transduction as an explanation. Next, we observe that signaling in response to residual endogenous antigens is impaired in IgD-only cells (as read out by basal calcium and ex vivo GFP induction). Indeed, we show that marginal zone (MZ) and B1a cells that express IgD but lack IgM induce less Nur77-eGFP than WT and this is not attributable to repertoire differences. Subsequent figures examine sensing of endogenous antigens in relevant in vivo contexts: B cell development and autoimmune disease. In the context of autoimmune disease, we identified a defect in rapid plasma cell responses by IgD-only cells. The final figure builds upon this finding, and identifies a major defect in rapid IgG1+ plasma cell induction by IgD-only cells in the context of immunization with T-dependent antigens, but intact generation of germinal center responses.

Also the way the data are presented is highly confusing and prevents comparison between figures.

We anticipated that the use of IgM−/−, IgD−/−, and IgM−/−IgD−/− mice (containing mixed populations of IgM-only, WT, and IgD-only B cells) along with IgM−/−, WT, and IgD−/− mice might be challenging to keep track of. For this reason, we intentionally generated a consistent color scheme to aid in interpretation of data, and we provided schematic diagrams to clarify our logic (Figure 6—figure supplement 2, Figure 7—figure supplement 2). We labeled all graphs with both the genotype of the mouse, and the “genotype” of the B cells specifically to aid in clarity. In general, cells that express IgM alone are blue and cells that express IgD alone are red (WT cells are grey or orange depending on allotype). We have revised our figures to remove as many deviations from this color scheme as possible, specifically in Figure 5E, F, Figure 7B, D, Figure 7—figure supplement 1B, C, D.

The manuscript clearly needs an effort to synthesize and clarify the message.

The main message of the manuscript is that IgD-BCR signal strength is reduced when compared to IgM-BCR signal strength only in vivo upon recognition of autoantigens driving differentiation (Figure 1 and MZ versus B1a cells under competitive conditions, Figure 4) or for the generation of SLPC (Figure 7).

Yes, this message is the one we hoped to communicate. Our evidence for this is also grounded in Manuscript Figure 3 demonstrating reduced Nur77-eGFP expression in IgD-only relative to IgM-only MZ and B1a B cells. We also show defect in SLPC generation and autoantibody production by IgD-only cells in response to endogenous antigens in the Lyn−/− model (Figures 5, 6). So, in sum, Figures 3, 4, 5, 6, and 7 provide evidence to support this message.

In contrast, in vitro (Figure 2), upon immunization (Figure 7) and upon recognition of auto-antigens in the Lyn−/− background (Figure 5A-B) the signals of IgD-BCR are as competitive as the signals from IgM-BCR.

This is in part an error of interpretation. We showed (Figure 2.) that in vitro IgD was equally if not more efficient at signal transduction than IgM in response to antibody cross-linking, in contrast to our in vivo observations. We suggest that coupling of the IgD BCR to downstream signaling machinery is intact but sensitivity to endogenous antigens is specifically impaired. We reconcile these results by suggesting that responses to endogenous antigens differ from those triggered by antibody-mediated crosslinking of the receptors in vitro. In response to endogenous antigens, we show that polyclonal B cell activation and GC B cell fate are intact, while autoantibody and SLPC generation are defective (Figures 5, 6). In response to T-dependent immunizations, we show impaired IgG1 SLPC generation, but intact GC B cell fate (Figure 7). We reconcile these observations by suggesting that the stringent requirement for strong BCR signaling for SLPC (but not GC) generation accounts for these observations.

Taken together, these results are confusing and point out to microenvironment, cell-differentiation stage, co-receptor expression and crosstalk between receptors in a given situation as reasons modulating the signals through the BCR. To conclude that "IgD-BCR sensing antigen is impaired" is an over-interpretation of the results, a not-justified generalization and neglects a major part of the results exposed in this manuscript.

We examine a wide variety of readouts known to be affected by BCR signal strength (ex vivo calcium, ex vivo GFP, in vivo GFP, MZ/B1 fate, SLPC and GC fate in Lyn−/−, SLPC and GC fate in response to immunization), and we cite literature describing how BCR signal strength affects the propensity of a B cell to adopt each fate. The common thread is that IgD-only cells are inefficient at adopting the fates that require the strongest in vivo BCR signaling in response to endogenous antigens. It is worth emphasizing that IgM and IgD are BCR isotypes that signal in response to antigen. Therefore, it should not be extremely controversial to propose in vivo antigen-dependent signaling differences as an explanation for phenotypes in mice expressing each isotype in isolation.

We acknowledge that complex in vivo phenotypes may reflect cooperation between IgM or IgD BCRs and some to-be-defined signal, but it is not possible or reasonable to exclude every known receptor expressed on B cells in this manuscript. That said, we provide important new data excluding a role for microbial stimulation of pattern-recognition receptors as well as TLR-MyD88-dependent signaling in regulating Nur77 expression in vivo by using two independent approaches (germ free mice and conditional deletion of MyD88 in B cells, Figure 1—figure supplement 1D, E, F). Table 1 lists data to exclude a role for: commensal flora and signaling downstream of MyD88, endocytic TLRs, Jak-Stat, CXCR4, CD40, and BAFF in the regulation of Nur77 expression in B cells in vivo under steady-state conditions. We further show in new data that a range of TLR ligands do not specifically require either IgM or IgD to signal and therefore would not account for our phenotypes (Figure 1—figure supplement 3). In the fifth paragraph of our revised Discussion, we explicitly state that the most likely mechanistic explanation for our findings are structural differences between the IgM and IgD isotypes, which could either directly influence transduction of endogenous antigen-driven signals or influence pairing with and access to relevant co-receptors such as CD19 and CD22 that in turn modulate BCR signaling.

We have no wish to over-interpret our data, but one cannot expect a model that synthesizes the data (clarity of message as requested above) without any interpretation or conclusions. Some degree of interpretation is necessary to help the reader make sense of a complex set of in vivo phenotypes. We have made every effort in our revised manuscript to make the clear distinction between description and interpretation of data.

Our manuscript should also be read against a background literature which has failed to identify significant non-redundant functions of these co-expressed BCR isotypes in vivo. We have done so for the first time and present a compelling model to explain why IgM and IgD are co-expressed on naïve B cells, and how down-regulation of the former but not the latter on self-reactive cells could function as a vital tolerance mechanism.

1) The whole study is based on the observation that Follicular B cells in the spleen are GFP+, indicating that they have been activated at some point during development (T2/T3 stage) by any of the receptors that induce the activation of the Nur77 locus (BCR, CD40, TLR9, TLR4…). The authors claimed already back in the first publication (Zikherman et al., 2012) that GFP positivity results from BCR signals upon encounter with autoantigens because GFP positivity is significantly reduced in BCR transgenic settings (IgHEL Tg), and again increased in the presence of antigen (IgHEL Tg/sHEL Ag Tg).

We make use of the Nur77-eGFP reporter in our manuscript precisely because our goal is to understand response of a polyclonal repertoire of IgM and IgD isotypes to endogenous antigens in vivo (rather than to model antigens). Major points 1 and 2 raised by this reviewer question whether Nur77-eGFP is a valid reporter of antigen-dependent signaling by B cells in vivo. We argue that this degree of skepticism is not justified. We provide data in this rebuttal and in our revised manuscript to further validate this reporter precisely because it aids in interpretation of our cellular phenotypes. As the reviewer notes and we have shown genetically in previous work (Zikherman et al., 2012), endogenous antigen is necessary and sufficient for Nur77-eGFP expression in vivo in mature B cells under steady state conditions (Manuscript Figure 1A, Author response image Figure 1).

Author response image 1
Antigen is necessary and sufficient for Nur77-eGFP expression in B cells; IgHEL-sHEL interaction drives higher Nur77-eGFP expression than the wild type BCR repertoire.

Panel (A) take from Manuscript Figure 1A. Panel (B) depicts quantification of mean Nur77-eGFP MFI -/- SEM in N=5 biological replicates.

https://doi.org/10.7554/eLife.35074.041

In our rebuttal and revised manuscript, we present substantial new data to exclude the contribution of other immunoreceptor pathways as regulators of Nur77-eGFP expression in B cells under steady-state conditions, especially those mediated by microbial stimuli. We have included a new table that summarizes all published and new figure panels that address specificity of reporter expression in B cells in response to immunoreceptor signals in vitro and in vivo (Table 1). These data collectively demonstrate that in vivo expression of Nur77-eGFP in B cells under steady state conditions is specifically regulated by antigen and regulators of BCR signaling (e.g. CD45, Lyn), but not by other immunoreceptors (including each receptor listed by the reviewer above: CD40, TLR9, TLR4) and therefore represents a VALID readout of antigen-dependent BCR signaling in vivo. We now proceed to address specific reviewer questions about the function of this reporter in our point-by-point response below.

However, if the encounter with autoantigens drives GFP positivity, why are there not GFP+ cells within the immature cells in the BM? At any given moment, some immature B cells are signaling through the BCR during B cell selection, one should expect a proportion of GFP+ cells in the BM. Moreover, it has been broadly demonstrated that BCR driven signals are crucial for B cell development, why are there not GFP+ B cells during B cell development?

We believe the explanation for low levels of Nur77-eGFP upregulation at early stages of B cell development is two-fold:

1) Immature B cells are more refractory to BCR-induced Nur77 upregulation than transitional and mature B cells. Ex vivo stimulation of reporter B cells at sequential stages of development shows that immature BM B cells are less responsive to anti-IgM stimulation than more mature B cell subsets (Author response image 2). This may be due in part to differences in receptor expression (lower levels of surface IgM, absence of IgD), but may also reflect differences in signaling machinery or chromatin landscape. Importantly, we see identical phenomenology comparing two independently generated Nur77-eGFP BAC Tg reporters (Moran et al., 2011; Zikherman et al., 2012).

Author response image 2
In vitro anti-IgM stimulation upregulates Nur77-eGFP expression in a stage-specific manner.

BM and splenic B cells from our Nur77-eGFP reporter mice (Zikherman et al., 2012) as well as independent reporter line (Moran et al. JEM 2011) were stimulated in vitro with media alone (shaded gray histogram) or 10μg/ml anti-IgM (solid black line) for 3 hours, and subsequently stained to detect subset markers as well as GFP.

https://doi.org/10.7554/eLife.35074.042

2) Highly self-reactive BM B cells are efficiently silenced via receptor editing and deletion. Indeed, when we force expression of a highly self-reactive BCR Tg, we observe that immature BM B cells in fact do upregulate Nur77-eGFP expression (Author response image 3). VH3H9 is a heavy chain (HC) originally cloned from the MRL/lpr lupus mouse model and biases the entire BCR repertoire towards DNA-reactivity, particularly when paired with endogenous Vλ1 light chain (LC) (Erikson et al., 1991). This HC has been knocked-in to the endogenous BCR locus to generate site directed (sd) VH3H9 Tg mice in which the majority of B cells express this HC paired with endogenous LCs. This model system is extremely powerful because bona-fide DNA-reactive B cells can be tracked through development in a polyclonal repertoire on the basis of surface Vλ1 expression. B cells expressing dsDNA-specific Vλ1 BCRs persist in the periphery in sd-VH3H9 mice despite extensive receptor editing and counter-selection, in part because Vλ1 light chain expression is itself the result of extensive receptor editing and “traps” self-reactivity in the repertoire (Erikson et al., 1991; Fields et al., 2005). We have generated reporter mice harboring the site-directed (sd) VH3H9 Tg in the endogenous BCR locus. dsDNA-reactive Vλ1 B cells turn on GFP early in development and express higher levels in maturity, consistent with their chronic self-antigen stimulation (Author response image 3, manuscript in preparation from our laboratory). These data support the Nur77-eGFP reporter as a sensitive marker of B cell self-reactivity, and argue that highly self-reactive B cells do upregulate reporter expression during BM development, but these cells normally represent a small proportion of the repertoire and are efficiently eliminated in WT mice.

Because these points are an important response to the reviewer query about the sensitivity of the Nur77-eGFP reporter to antigen-dependent signaling during B cell development, but are not directly relevant to the manuscript under review, we present these data in the rebuttal, but have not incorporated them into the revised manuscript. Moreover, data in Author response image 1B and Author response image 3 are taken from a manuscript in preparation in our laboratory.

Author response image 3
Self-reactive B cells upregulate Nur77-eGFP at early stages of B cell development.

Nur77-eGFP BAC Tg was crossed to the Vh3H9 HC site directed Tg. BM and splenic cells from these mice (red) and unrestricted repertoire reporter mice (gray) were harvested and stained ex vivo to identify B cell subsets as well as surface light chain expression. Histograms depicting GFP expression in K+ and L+ B cells subsets are representative of > 9 biological replicates.

https://doi.org/10.7554/eLife.35074.043

Is it possible that GFP positivity is the result of a combination of different signals happening from the T2/T3 stage on, including also BCR signals in the IgHEL Tg/sHEL Ag Tg settings? The authors discard CD40; TLR3,7,9 (Unc93b1 mice, this information should be found at least in the figure legend); TLR7; IL4; BAFF and CXCR4 (Figure 1—figure supplement 1 and Zikherman et al., 2012). However, an obvious candidate is TLR4, no data are shown using either LPS stimulation in vitro, TLR4−/− mice, Myd88−/− mice or even better by housing the mice in pathogen free/sterile conditions. Indeed amount of mRNA of TLR4 increases from the T1 stage on to get a maximum at the Fo stage. Other receptors which levels of mRNA are upregulated from the transitional stage on are TLR6, TLR1, CD22….

We have modified the figure legend.

As suggested by this reviewer, we have directly addressed the role of microbial stimulation of TLR pathways in regulating Nur77 expression in B cells in vivo at steady-state.

1) Specifically, our revised manuscript examines both germ-free mice and mice with conditional deletion of MyD88 in B cells (MB1cre MyD88 fl/fl mice) in order to demonstrate that Nur77 expression is INDEPENDENT of commensal flora and TLR /MyD88-dependent signaling pathway at steady-state in vivo. Since re-deriving reporter mice under germ-free conditions and breeding the reporter into the MyD88 conditional background would each be extremely time consuming and expensive, we instead assayed endogenous Nur77 protein in B cells and endogenous Nr4a1 transcript in splenocytes from these mice. We first took advantage of Nr4a1−/− mice to confirm specificity of intracellular staining for endogenous Nur77 protein (Author response image 4A, Figure 1—figure supplement 1D, F). Because qPCR primer sets to detect Nr4a1 expression flank a neo cassette in Nr4a1−/− mice, they also serve as a specificity control for Nr4a1 primer sets (Figure 1—figure supplement 1E). We next verified that endogenous Nur77 protein and Nr4a1 transcript serve as sensitive measures of Nur77-eGFP reporter expression in B cells under steady state conditions (Author response image 4B, C). We demonstrate that neither endogenous Nur77 protein nor endogenous Nr4a1 transcript are upregulated under steady state conditions in splenic B cells / splenocytes in SPF mice relative to germ-free mice (Figure 1—figure supplement 1D, E). We further show that endogenous Nur77 protein levels are equivalent in WT and MB1-cre MyD88 fl/fl B1a cells (Figure 1—figure supplement 1F). We also confirm successful deletion of MyD88 in B cells in these mice (Author response image 5).

Author response image 4
Endogenous Nr4a1 transcript and Nur77 protein correlate with Nur77-eGFP reporter expression in B cells.

(A) Intracellular staining of endogenous Nur77 protein in splenic B cells from Nr4a1+/+, -/-, and -/- mice. Graph depicts mean -/- SEM of N=3 mice / genotype. (B) Nur77-eGFP reporter B cells were stained to detect endogenous Nur77 protein intra-cellularly. Graph depicts mean MFI of endogenous Nur77 in splenic B cells with low or high GFP expression (15% lowest and 15% highest) -/- SEM of N=3 mice. (C) Mature naive reporter B cells were sorted for 3% highest and 3% lowest GFP expression and subjected to RNASeq. Graph depicts mean FPKM of Nr4a1 transcripts -/- SEM of N=3 biological replicates.

https://doi.org/10.7554/eLife.35074.044
Author response image 5
Efficient conditional deletion of MyD88 in MB1-Cre MyD88 fl/fl B cells (“MyD88 cKO”).

Splenocytes from MyD88 cKO or control mice were stimulated for 2 hours with LPS (333 ng/ml). Graph depicts mean MFI of Intracellular staining to detect Nur77 upregulation in splenic B cells -/- SEM in N=4 mice. Nur77-deficient mice are a staining control.

https://doi.org/10.7554/eLife.35074.045

2) In our Nature 2012 paper (Figure 3) and in Figure 1A of this manuscript we show that forced expression of the IgHEL BCR Tg in the absence of endogenous cognate antigen nearly eliminates all Nur77-eGFP expression in resting naïve splenic B cells (Author response image 1). This demonstrates that antigen is necessary to drive Nur77eGFP expression in vivo. We present new data to show that this loss of Nur77-eGFP expression is not due to impaired TLR-dependent signaling in these B cells as they are fully competent to respond to TLR ligands in vitro (Author response image 6). Together this argues that since microbial sensing by IgHEL B cells is intact, and yet Nur77-eGFP expression is lost, endogenous antigen is necessary for Nur77-eGFP expression, and Nur77-eGFP expression reflects endogenous antigen rather than indirect loss of a TLR-dependent signaling pathway.

Author response image 6
Naïve B cells are competent to upregulate Nur77-eGFP in response to canonical TLR ligands even in the absence of endogenous antigen.

Splenocytes from non-BCR Tg and IgHEL BCR Tg reporter mice were stimulated overnight with low and high doses of ligands for TLR4 / rp105 (A), TLR9 (B), and TLRs1/2 (C). Histograms depict GFP fluorescence in B220+ splenocytes. As previously reported, Nur77-eGFP expression is low in the absence of endogenous antigen (IgHEL BCR Tg), but these cells are nevertheless responsive to a broad range of TLR ligands.

https://doi.org/10.7554/eLife.35074.046

3) Lastly, in Figure 1—figure supplement 3A, we compare TLR-dependent signal transduction by IgM-deficient and IgDdeficient B cells in vitro and show that signaling downstream of LPS, CpG, and Pam3CSK4 through TLR4/Rp150, TLR9, and TLRs1/2 respectively are not dependent upon either the IgM or the IgD BCR. This allows us to conclude that differences between IgM-deficient and IgD-deficient B cells in vivo are not due to defective TLR-dependent signaling.

Taken together, data presented here and summarized in Table 1 supports specificity and validity of Nur77-eGFP reporter as a marker of endogenous antigen-dependent signaling in vivo under steady state conditions.

We discuss CD22 separately below.

2) In line with the comment before, why most of Fo cells (around 90%) are GFP+ also in the WT settings? This high percentage is expected in the IgHEL Tg/sHEL Ag Tg setting. However, in the WT situation, if GFP positivity is the consequence of autoantigen encounter, this would mean that most of B cell emigrants from the BM carry autoreactive receptors. Do the authors suggest that tolerance in mice is established mainly in the peripheral checkpoint (T2/T3)? This will be in contrast with the data in humans in which both checkpoints reduced the percentage of self-reactive BCR to half, and only 40% of new emigrants carry self-reactive BCRs, percentage that is reduced to 20% after the peripheral checkpoint.

Taken together the last two points, it seems true that autoantigens are sufficient to drive Nur77-GFP expression as claimed by the authors (IgHEL Tg/sHEL Ag Tg setting already reported in 2012), but I am not convinced at this stage that all GFP+ cells in the Fo stage are GPF+ because BCR signaling as consequence of autoantigen encounters. I rather think that GFP accumulation is the result of several activation signals from diverse receptors, including the BCR but not exclusively.

We have indeed proposed that most mature B cells harbor self-reactivity to endogenous antigens (albeit mild) that is proportional to Nur77-eGFP reporter expression, and that Nur77-eGFP expression in those cells is a result of endogenous antigen recognition. This was a central part of the message of our Nature 2012 paper. Our evidence in support of this contention is three-fold:

a) Restricting the normal diverse BCR repertoire to force recognition of a non-murine antigen (via IgHEL BCR Tg) and thereby eliminate endogenous antigen recognition results in near-complete loss of Nur77-eGFP expression in B cells in vivo (Figure 1A; Zikherman et al., 2012: Figure 3B, C, Author response image 1). Notably, this occurs despite the fact that surface BCR expression in IgHEL BCR Tg cells is higher than that found on WT B cells, and despite the fact that these cells are competent to respond to TLR ligands (Author response image 6). Reintroducing model antigen (sHEL Tg) genetically into this system is sufficient to upregulate Nur77-eGFP in B cells and recapitulate GFP expression seen in a normal diverse mature BCR repertoire. Importantly, the GFP MFI seen in IgHEL BCR Tg B cells exposed to high affinity cognate antigen sHEL in vivo is considerably higher than that seen in WT B cells with a diverse repertoire (Author response image 1B). This suggests that WT B cells do see endogenous antigens, but at considerably lower intensity / affinity than that encoded by the IgHEL / sHEL model.

b) We have shown in previously published work that Nur77-eGFP B cells harbor BCRs with higher ANA-reactivity as measured by ELISA (Zikherman et al., 2012: Figure 4E). Similar findings of autoreactivity among IgMlo mature B cells from mice and humans (corresponding to B cells with higher GFP and endogenous Nur77 expression) have been reported by independent groups in mice and humans (Kirchenbaum et al., 2014; Quach et al., 2011).

c) Moreover, in published data and the present revised manuscript and rebuttal, we systematically exclude a contribution by: commensal microbes, MyD88-dependent signaling, CD40-dependent signaling, CXCR4-dependent signaling, Jak-Statdependent signaling, and BAFF-dependent signals (see Table 1, Figure 1—figure supplement 1).

The reviewer wishes to reconcile our argument that Nur77-eGFP expression in naïve mature B cells reflects reactivity to endogenous antigens with data reported by Wardemann, Nussenzweig and colleagues (Wardemann et al., 2003). Wardemann at al. show that anti-nuclear-reactivity of mature human B cells measured by immunofluorescence assay (IFA) are seen in only about 20% of cells. This study of necessity uses an arbitrary cutoff and an artificial readout (ANA IFA) to identify “autoreactive B cells”. Our readout (GFP fluorescence) is a continuous, linear, and sensitive reporter of bona fide reactivity to endogenous antigens without any arbitrary cutoff other than detection of fluorescence by flow cytometer. We propose that the degree of endogenous antigen recognition by the naïve mature repertoire is “weak / mild” in comparison to genetic models of B cell autoreactivity (see Author response image 1, 3), but is detectable by the Nur77-eGFP reporter and is above the baseline seen with a bona fide non-self-reactive BCR (IgHEL Tg). The self-reactivity detected by ANA IFA used by the Nussenzweig laboratory may be distinct from true endogenous antigen recognition and/or this assay may have a lower sensitivity for detection of self-reactivity. Indeed, Ignacio Sanz and others have characterized a naturally-occurring human BCR heavy chain (VH34-4) that recognizes an epitope on the surface of RBCs that would presumably not be detected by an ANA assay (Quach et al., 2011; Reed et al., 2016).

Indeed, there is an extensive B cell literature that is consistent with our contention that self-reactivity (mild) is a feature of the normal mature B cell repertoire, and may actually be selected for. Prior work has supplied indirect evidence for antigen-dependent positive selection of B cells (Cancro and Kearney, 2004; Cariappa and Pillai, 2002; Rosado and Freitas, 1998; Su et al., 2004; Thomas et al., 2006; Wang and Clarke, 2003). Studies have shown that self-ligands can drive selection of B cells into the mature primary immune repertoire in the context of BCR Tg models (Hayakawa et al., 2003; Wang and Clarke, 2003). Analysis of AgR repertoire diversity has revealed a restricted repertoire in mature relative to immature B cells, suggesting active selection by antigen (Freitas et al., 1989; Gu et al., 1991; Levine et al., 2000).

3). Because an inverse correlation between GFP levels and IgM expression is observed, the authors propose that IgM is downregulated as consequence of autoantigen encounter and therefore, activation of autoreactive B cells is prevented. The slope that correlates IgM and GFP expression in the IgM-only mice is reduced compared to the slope in WT, why is this? In other words, why in the absence of IgD, IgM cannot be optimally downregulated? In this the case also upon anti-IgM stimulation ex vivo?

We have explicitly addressed this point in our revised manuscript (subsection “Dual expression of IgM and IgD BCRs is required to establish a broad dynamic range of BCR responsiveness across the repertoire”, last paragraph; Discussion, eighth paragraph). Here we clarify further: previous work from the Rajewsky lab identified a critical survival function for the BCR, as B cells die upon acute deletion of the BCR (Kraus et al., 2004; Lam et al., 1997). Moreover, recent work from Goodnow lab describes a novel ENU-generated mutant that lacks IgD expression but does not compensatorily upregulate surface IgM (dmit mutant), in contrast to the IgD−/− mice studied in our manuscript (Sabouri et al., 2016). The dmit mutant B cells have a major defect in survival, suggesting that not only is BCR expression required for B cell survival, but the amount matters. In Figure 1D, we show that IgM-only (IgD−/−) B cells have 60% as much total receptor as WT cells. Since these IgM-only cells express just one receptor, any downregulation in IgM leads to an equivalent downregulation in total surface receptor. WT cells express two isotypes, and we propose that the high level of IgD on all WT cells gives them “room” to downregulate IgM more extensively without losing BCR-mediated survival. We propose here and in our manuscript that IgM downregulation in IgD−/− mice is limited by survival of such cells. It is not possible to accurately recapitulate tonic survival function of the BCR in vitro.

In contrast to IgM, IgD levels are invariant with GFP expression, this would mean that in IgD mice GFP high cells could be easily re-activated driving autoimmunity, is this the case?

This would be the case if IgD BCR were sensitive to endogenous antigen stimulation. Our manuscript argues that IgD is less sensitive than IgM to endogenous antigens. We propose that IgD’s reduced sensitivity towards self-antigens permits high surface expression on autoreactive cells (thereby facilitating their survival) without promoting aberrant activation in response to endogenous antigens. As IgD can respond robustly to anti-kappa stimulation in vitro, we propose that IgD is maintained on autoreactive cells to permit both survival of these cells and to allow for activation upon encounter with exogenous antigens. Our data in Figure 7 show that this subsequent activation through IgD has limits, particularly with regard to IgG1+ PC generation. The Goodnow lab has proposed (Model in Figure 7—figure supplement 2) that promotion of GC fate in this context allows autoreactive IgDhiIgMloGFPhi cells to mutate away from self-reactivity(Reed et al., 2016; Sabouri et al., 2016; Sabouri et al., 2014).

The authors claim that these cells cannot be optimally activated, but the total GFP levels are the same, proving that activation was equally strong.

GFP levels are the same, but it takes twice as much IgD BCR than IgM BCR to achieve comparable Nur77-eGFP expression (Figures 1B, D, E, F). Moreover, IgD-only B1a and MZ B cells have higher BCR expression but lower GFP expression than IgM-only comparators (Figure 3, Figure 3—figure supplement 1). Furthermore, when we reduce selection pressures on the MZ repertoire (BAFF Tg), resulting in comparable surface expression of IgM and IgD, we unmask a markedly reduced expression of Nur77-eGFP in IgD-only MZ B cells (Figure 3D, Figure 3—figure supplement 1A). In most of our manuscript, we focus on BCR-dependent phenotypes that are present in spite of high surface IgD receptor expression.

Can IgD be downregulated upon stimulation ex vivo in the IgD-only B cells?

IgD can be downregulated on the surface of both WT and IgD-only cells in response to anti-IgD or anti-kappa stimulation. In preliminary studies, we didn’t observe significant differences between the isotypes with regards to receptor downregulation in response to anti-kappa stimulation, so we did not pursue this further.

4) In Figure 1F, the authors showed that for a given GFP signal more IgD-BCRs are needed than IgM-BCRs, proposing that IgD-BCRs do not signal properly in vivo (supposedly, in response to antigens). Taking in consideration recent data regarding the nanoscale distribution of IgM and IgD-BCRs in protein islands, how are the levels of co-receptors such as CD19, CD22, CD20 in WT, IgM-only and IgD-only? Are these co-receptors excluded or included in the IgM protein island in the absence of IgD? Are these coreceptors modulating differently IgM-BCR and IgD-BCR signals? Differences in co-receptor levels and/or composition of these protein islands might explain why IgD stimulation in vitro, which happens solely by the BCR, is strong (Figure 2).

We agree that differential association with B cell coreceptors that directly modulate BCR signal transduction (e.g. CD19, CD22) may contribute to differential signaling by IgM and IgD. Our revised Discussion incorporates this possibility (fifth paragraph). We studied surface expression of CD22, CD19, and CD45 on IgM-only, IgD-only, and WT B cells via FACS staining and found no differences (Author response image 7), but we did not assess receptor co-localization by microscopy. We believe that further exploring this possibility is beyond the scope of the current manuscript, but it is a fruitful avenue to pursue in the future.

Author response image 7
BCR, co-receptor and CD45 expression on IgM or IgD-deficient B cells.

Splenocytes from IgM-/-, IgD-/-, or WT ice were stained to detect surface expression of CD19, CD22, CD45, CD79b, and kappa light chain. Histograms depict CD23+ splenocyte gate. Differences in surface kappa expression between genotypes consistent with Figure 1D, Figure 1—figure supplement 2B, C.

https://doi.org/10.7554/eLife.35074.047

5) What exactly does this mean for the authors, "sensing of endogenous antigen"? or just "sensing of antigen"?

When referring to antigen sensing, we mean that the BCR not only binds to an antigen but also transmits a biochemical signal (can be detected by basal calcium tone, by Nur77-eGFP expression). The B cell then evaluates this signal and can either remain quiescent or undergo respond (proliferation, differentiation, etc.) “Sensing of endogenous antigen” refers to the BCR sensing an antigen in the absence of infection or immunization. As we describe above, the relevant signals in the populations we study are not microbial in origin. We have clarified our use of this term in the Introduction of our revised manuscript (last paragraph).

6) It is unclear to me the significance of the experiments shown in Figure 3. Both B1a and MZ B cells express preferentially IgM-BCRs. Are they generated to the same extent in IgD-only mice? If yes, is this not probing that IgD-BCR in vivo signals as efficient as IgM-BCR at least to drive differentiation of these cells? However, IgD-only B1a and MZ cells showed reduced levels of Nur77-GFP ex vivo.

As shown in Figure 4—figure supplement 1, IgD-only mice generate significantly fewer B1a cells and more MZ cells than WT and IgM-only mice. This altered B cell fate is markedly exaggerated in a competitive setting (Figure 4). Extensive and cited literature suggests that MZ fate requires weak BCR signals while B1a cell fate requires strong BCR signals. Thus, we demonstrate that IgD-only B cells exhibit impaired BCR signaling in response to endogenous antigens in vivo (Figure 3), and exhibit B cell fates consistent with this (Figure 4). Since signal strength has been shown to be important in the development of these subsets, we use this result as support for our claim that IgD senses endogenous antigens more weakly than IgM does.

Again, how is the expression of co-receptors and especially of innate-like receptors in IgD-only versus WT or IgM-only B1a and MZ B cells?

Again, how is the expression of co-receptors and especially of innate-like receptors in IgD-only versus WT or IgM-only B1a and MZ B cells?

The phenotype of B1a and MZ B cells from IgD-only, IgM-only and WT mice is grossly normal, as is coreceptor expression (Author response image 8). We have not examined surface expression of TLRs on these cell populations but we demonstrate that splenic B cell Nur77-eGFP expression is unaffected by commensal flora, and that MyD88-deficiency does not alter Nur77 expression in B1a cells (new Figure 1—figure supplement 1D, E, F).

Author response image 8
Grossly normal surface phenotype of B1a and MZ B cells in WT, IgM-/- and IgD-/- mice.

(A) Plots show CD19+ PerC cells gated to identify CD5+ B1a cells (reduced in IgM-/- as shown in Figure 4—figure supplement 1). Histograms depict CD5 and CD19 expression in B1a cells. (B) Plots show B220+ splenocytes gated to identify CD21hiCD23lo MZ B cells. Histograms depict CD21 and CD23 expression in MZ B cells.

https://doi.org/10.7554/eLife.35074.048

In these two populations all cells are GFP+ cells, it seems counterintuitive that all cells that populate the periphery are autoreactive. In the IgD-only B1a cells there is indeed a GFP- population, the authors should compare the repertoire of these GFP- cells to GFP+ cells to support they claims.

The B1a B cell receptor (BCR) repertoire is highly restricted and selected on the basis of germ-line-encoded reactivity to self-antigens such as phosphatidylcholine (PtC), which is exposed on the membranes of dying cells (Baumgarth, 2016; Bendelac et al., 2001; Yang et al., 2015). Indeed, the Herzenberg lab has recently shown that a small number of PtCspecific CDR3s represent over a third of the entire B1a repertoire implying active selection for this specificity (Yang et al. 2015). Consistent with this, we show in recently published work that B1a cells express more GFP than B2 cells and moreover that B1a cells with PtC-specific BCRs have the highest levels of Nur77-eGFP expression (Huizar et al., 2017). The latter result is shown in Author response image 9, excerpted from that paper for ease of reference. As suggested by the reviewer, this data essentially compares the repertoire of GFP-low and GFPhigh B1a cells and reveals marked enrichment of PtC-specific BCRs among GFPhi B1a cells.

Author response image 9
Nur77-eGFP expression identifies selfreactive PtC-specific B-1a cells in vivo.

(Figure 2A,B reproduced from Figure 2 of Huizar et al., 2017, Immunohorizons, published under the Creative Commons Attribution 4.0 International Public License (CC BY4.0; https://creativecommons.org/licenses/by/4.0/)). (A) PerC cells from Nur77-eGFP reporter mice were stained immediately ex vivo on ice with surfacemarkers and PtC-rhodamine liposomes to identify B cell subsets. Representativeplots show PtC+ cells in B-2 and B-1a PerC subsets. (B) Representative histogramsdepict GFPexpression in PtC+ and PtC- B-1a cells.

https://doi.org/10.7554/eLife.35074.049

7) As cited by the authors, a link between BCR signal strength and cell fate has been established some years ago, in which weak BCR signals will allow MZ B cell generation but prevent B1 cell generation. In contrast, strong signals might favor B1 generation/maintained. No major differences were observed in B1 or MZ cell generation between IgM-only, IgD-only and WT. However, under competitive conditions, IgD-only B cells fill up preferentially the MZ B cell compartment and are unable to differentiate to B1a cells. How are these cells phenotypically?

Please see response to point 6 above as well as Figure 4—figure supplement 1 demonstrating cell fate differences even in the absence of competition. See also Author response image 8 which specifically depicts MZ and B1a phenotypes.

MZ are BM-derived while B1a cells are mainly fetal liver derived and long lived. It has been proposed that in B2 cells IgD provide survival signals. Is it possible that B1 cells required survival signals via IgM and that IgD cannot provide survival signals or self-renewal signals? Is it really a matter of signal strength or of signal quality? How is this piece of data related to autoantigen recognition?

Our data does not directly distinguish between generation and survival of B1a cells. We have clarified our manuscript text to make this explicit (subsection “Cell-intrinsic skewing of B cell development by the IgM and IgD BCRs”, end of first paragraph). However, antigen-dependent BCR signal strength is essential for normal steady-state numbers of B1a cells in mature mice (despite their fetal origin) as described in literature cited in our manuscript. It is highly likely that such strong BCR signaling is antigen-dependent at least in part because the B1a repertoire is highly self-reactive. Therefore, it is reasonable to argue that impaired antigen-dependent signaling correlates with and likely accounts for loss of IgD-only B1a cells in IgM-deficient mice and in IgM−/− mice. Our in vivo data does not distinguish between signal strength and “signal quality”. Our revised manuscript addresses this point explicitly (see response to point 4 above).

8) Which bibliography supports the following sentence "endogenous auto-antigens may differ significantly from model antigens"? The authors used HEL as model antigen and as endogenous auto-antigen, is the model antigen HEL not differing from endogenous auto-antigens?

We study soluble HEL model antigen as a starting point to validate function of our Nur77eGFP reporter, but our ambition in this study was really to understand how the IgM and IgD isotypes in a diverse repertoire of B cells respond to bona fide endogenous antigens.

The 9G4 anti-idiotype antibody recognizes naïve human B cells expressing the IgHV4-34 heavy chain segment. These cells account for 5-10% of circulating naïve human B cells, have been reported to recognize the I/i blood group antigen expressed on the surface of circulating red blood cells as well as O-linked carbohydrates on the B220 isoform of CD45, and exhibit downregulation of IgM consistent with self-reactivity. These bona fide endogenous antigens differ from model antigens such as HEL by their chemical structure (carbohydrate vs. protein) and form (cell surface vs. soluble), and likely by their affinity for the BCR. We have presented this example explicitly in both the Introduction (fourth paragraph) and Discussion (fourth paragraph) of our revised manuscript along with relevant citations (Quach et al., 2011; Reed et al., 2016).

9) Similar to the results obtained in vitro, the results in the Lyn−/− mice support that both IgM and IgD-BCR provide sufficient signals for B cell activation (CD86, CD69) and for GC differentiation against the theory of the authors that "IgD-BCR sensing" of antigen is impaired.

IgD is not incapable of transmitting allantigen-dependent signals. Rather, we argue that it does so less efficiently than IgM especially in response to endogenous antigens. Indeed, if this were not the case, IgD would not be able to mediate B cell development and immune responses as reported in the original description of IgM−/− mice by Frank Brombacher and colleagues (Lutz et al., 1998). Rather, we argue that IgD-only B cells are particularly defective at adopting cell fates requiring strong BCR signaling. Our manuscript has identified a range of in vivo phenotypes which IgD alone is insufficient to drive (e.g. high Nur77-eGFP expression, B1a cell fate, unswitched PC expansion in Lyn−/− mice, autoantibodies in Lyn−/− mice). By contrast, GC fate is not compromised. This is consistent with intact affinity maturation in response to TD-immunogens as reported by Brombacher and colleagues. We propose this is because weak BCR signals are sufficient to drive GC fate, in contrast to SLPC fate. Indeed, there is an extensive literature supporting this model (cited in our manuscript). Similarly, we argue that BCR signaling transmitted through IgD in the absence of Lyn is sufficient to upregulate activation markers and GC fate, but apparently not sufficient to drive unswitched PC or autoantibody production. The model we propose is that IgD-dependent signaling in vivo in response to endogenous antigens is not completely defective, but is rather less sensitive than that mediated by IgM. In other words, our paper describes which BCR signaling thresholds IgD can overcome and which ones it cannot.

10) Regarding the production of autoantibodies in Figure 5, why the production of anti-dsDNA antibodies is also increased in the WT cells that co-exist with the IgM-only B cells? Indeed, at 25 weeks the basal levels are elevated in all the mice compared to all the other situations.

We have noted a reduced penetrance of autoantibody production in IgM−/− Lyn−/− mice relative to IgD−/− Lyn−/− mice. In order to compensate for this, we collected additional serum samples from IgM−/− Lyn−/− mice aged beyond 24 weeks (between initial manuscript submission and current revision included as new data in this revision). This allowed us to obtain more samples with detectable anti-dsDNA IgG2a and validate our finding that IgDonly cells are indeed protected from giving rise to anti-dsDNA IgG2a-secreting plasma cells relative to WT B cells in the same animal (Figure 5E). We believe these additional data further strengthen our conclusions. Our favored explanation for the reduced penetrance of autoantibodies in IgM−/− Lyn−/− mice relative to IgD−/− Lyn−/− mice is that the frequency of cells competent to initiate anti-dsDNA IgG2a PC production regulates disease progression. This was recently shown to be important for GC initiation in an autoimmune mouse model (Degn et al., 2017). We speculate that it might also play a role in PC differentiation, either through generation of an inflammatory environment or through a feed-forward mechanism where circulating autoantibodies drive further autoantibody generation. In addition, non-cell-autonomous differences in serum IgM levels between may contribute.

It is possible that other receptors, such as germline-encoded receptors, cooperate only with IgM-BCR to break tolerance and not with IgD? This will be independent of BCR strength but dependent of crosstalk between receptors. This crosstalk has been described at least for TLR4/IgM-BCR and RP105/IgM-BCR.

We understand that the reviewer is referring to the following publications:

Otipoby, K. L., et al., 2015. The B-cell antigen receptor integrates adaptive and innate immune signals.Proc Natl Acad Sci U S A 112: 12145-12150.

Ogata et al., 2000. The Toll-like Receptor Protein Rp105 Regulates Lipopolysaccharide Signaling in B Cells. J Exp Med. 2000 Jul 3; 192(1): 23–30.

Schweighoffer E, Nys J, Vanes L, Smithers N, Tybulewicz VLJ. 2017. TLR4 signals in B lymphocytes are transduced via the B cell antigen receptor and SYK. The Journal of experimental medicine214:1269-1280

TLR4 signals in response to LPS. The TLR protein RP105 also regulates responses to LPS in B cells. We show new data to establish that IgD-only as well as IgM-only B cells respond efficiently to LPS as well as other canonical TLR ligands (new Figure 1—figure supplement 3A). Therefore, these signals do not uniquely cooperate with the IgM or the IgD BCR and would not account for phenotypes described in our manuscript. We address this explicitly in our revised manuscript (Discussion, sixth paragraph).

11) In Figure 6A, how are the levels of Ets1 in IgM-only? And in Figure 6B, how is intracellular IgD in WT or IgM-only mice?

Based on how Figure 6A is presented, we assume that the reviewer meant to ask about Ets1 levels in IgD-only (IgM−/− Lyn+/+) B cells. We generated additional lysates and found that IgD-only B cells have normal Ets1 levels relative to WT B cells at steady state (Author response image 10). We did not detect appreciable numbers IgD+ plasma cells in WT and IgM-only mice.

Author response image 10
Ets1 protein expression is unaltered in splenic B cells the absence of IgD.

Representative western blot of splenic B cells probed for Ets1 on left. Graph depicts mean normalized protein expression -/- SEM of N=3 biological replicates.

https://doi.org/10.7554/eLife.35074.050

I do not agree with the interpretation of Figure 6E, IgM−/−Lyn−/− have elevated IgD serum levels when compared to Lyn−/−.

The message of Figure 6 is that WT B cells exhibit massive expansion of unswitched plasma cells on the Lyn−/− background, while IgD-only B cells do not and are protected from this phenotype. We believe that the reviewer has made a typo and misinterpreted our claims because we do not show IgD levels in Lyn−/−. Rather, in Figure 6E we compare serum IgD between IgM−/− Lyn−/− and IgM−/−. The point of this figure is to show that there is no difference in serum IgD between these genotypes; the dilution curves between the two genotypes are overlaid across the linear part of the curve showing concentration of IgD is unaltered between these genotypes. This corresponds to lack of expansion of IgD plasma cells in IgM−/− Lyn−/− mice (Figure 6B, C). This is in contrast to expansion of IgM plasma cells and increased serum IgM (approximately 10x) in Lyn−/− mice (Figure 6B, C, D). We believe the reviewer mistakenly thought we were claiming that there was a difference.

12) Why are IgD-only less competitive to enter the GC but disfavored to produced SLPC responses? Is this because BCR-strength as proposed by the authors or by the effect of isotype-specific interactions with co-receptors?

See above response to points 4 and 9.

Reviewer #3:

In this manuscript the authors present an extensive body of work comparing the development and responsiveness of polyclonal B cells capable of expressing either IgM or IgD BCRs prior to undergoing class switching. Whilst there is an extensive literature comparing the functional capabilities of these two (normally) co-expressed classes of BCR, this particular study examines a variety of readouts not previously examined, in particular the apparent sensing of endogenous self-antigens. Despite some concern about the different strain origins of the original IgM and IgD KO lines, the authors appear to control for this adequately throughout the study.

Overall I found the experiments to be well-performed and thoughtfully interpreted. The data presented here go some way to resolve recent controversies in the field and as such provide a valuable source of new information to help resolve the longstanding question of why naive, resting B cells co-express IgM and IgD BCRs.

I have just a couple of specific points:

1) In the SRBC-induced responses, did the IgM KO mice possess a general deficiency in PC production apart from IgG1-switched PCs e.g. in producing unswitched PCs?

2) Similarly, was the production of switched (e.g. IgG1+) GC B cells different between the IgM-only and IgD-only B cells?

It would be important to add this information to clarify the nature of the defect involved.

We have examined IgM and IgD PC generated in response to SRBC immunization (new Figure 7—figure supplement 1C, D). We find that both IgM-only and IgD-only mice can generate unswitched PCs. IgD-only B cells and WT B cells produce similar numbers in absence of competition, while IgD-only B cells produce more in competitive setting. This trend correlates very well with the relative numbers of MZ B cells of each genotype (i.e. expanded number of IgD-only B cells especially in IgM+/- mice –Figure 4—figure supplement 1B and Figure 4D). MZ B cells efficiently generate short-lived unswitched plasma cells even in response to TD-antigens (Phan et al., 2005; Song and Cerny, 2003). We therefore suspect that IgG1 PCs emanate from the Fo B cell compartment, while the unswitched PC response may originate in the MZ B cell compartment. Normal SLPC generation by IgM-deficient MZ B cells in response to immunization would be consistent with previously reported normal T-independent II responses in IgM-deficient mice (Lutz et al., 1998). However, we cannot exclude a contribution of follicular-derived PCs that have failed to class switch to IgG1. We have modified our manuscript accordingly (subsection “IgM BCR is required for efficient generation of short-lived IgG1+ plasma cells but is dispensable for GC B cell fate”, third paragraph).

In pilot experiments at the time point analyzed, we observed very low numbers of switched GC B cells in all genotypes in response to sheep exposed on the membranes of dying cells RBC immunization; the vast majority were unswitched (see the second paragraph of the aforementioned subsection).

[Editors' note: the author responses to the re-review follow.]

This manuscript provides novel insights into different functions between IgM and IgD BCRs. There are some remaining issues that need to be addressed before acceptance, as outlined below in the separate reviews. Specifically, we request that you recognize that the Nur77-GFP signals in B cells might be the result of BCR stimulation by a broader collection of non-inflammatory antigens in vivo, not only self-antigens. They might be globally named as endogenous antigens but not exclusively self-antigens, or endogenous self-antigens.

We are extremely grateful for the thoughtful editorial and review process. We have addressed all reviewer queries. Specifically, we have incorporated the possibility that Nur77-eGFP expression downstream of BCR engagement may be triggered not only by “self” antigens, but also by non-inflammatory antigens that are not germline encoded. We have made this clear in our revised manuscript. Please see response to reviewer #3 below for detailed description of manuscript edits that address this point.

Reviewer #3:

I have read with enthusiasm and interest the revised version of the manuscript entitled "Differential sensing of endogenous antigens and control of B cell fate by IgM and IgD B cell receptors" by Noviski et al. I appreciate the determination of the investigators to address the issues that arose during the revision process and thank the authors for their efforts.

One of my major points in the previous revision round was whether the Nur77-GFP positive B cells were indeed positive because activation of the BCR by self-antigens (as claimed by the authors) or by a combination of signals by the BCR and/or additional receptors expressed on the surface of B cells.

The authors have appropriately answered part of this question by showing that some of these receptors do not activate expression of Nur77 (TLR3, TLR7, TLR9, CXCR4). Other receptors, such as TLR4, TLR9 and TLR1/2 do activate expression of Nur77-GFP (Author response image 6). However, the authors demonstrate here (Figure 1—figure supplement 3) that upregulation of Nur77-GFP by these receptors is equal between IgM-only and IgD-only B cells (although the concentration of ligands chosen are too high).

We have performed these assays with lower doses of LPS and CpG and see comparable upregulation of Nur77-eGFP as well as activation markers by IgM-/- and IgD-/- B cells. Lower doses of Pam3CSK4 were insufficient to activate B cells or upregulate Nur77-eGFP. We also corrected an error: CpG dose units are nM, not µM. We have now revised Figure 1—figure supplement 3 to depict lower doses of stimuli.

The authors also show in the new version, that endogenous levels of Nur77 transcripts in B cells are equally elevated between Germ-free mice and SPF, arguing that Nur77 signals in B cells do not from microbial stimulation (although these signals might come from the endotoxin present in the chow, or did the authors used endotoxin-free chow?). Moreover, the levels of endogenous Nur77 transcripts were equally elevated between MyD88KO and MyD88WT peritoneal B1a cells. Here, I would like to ask why peritoneal B1a cells were chosen and not splenic B cells as in Figure 1—figure supplement 1D-E?

We have now modified Figure 1—figure supplement 1 to include qPCR data analyzing Nr4a1 transcript in whole splenocytes taken ex vivo from B cell-conditional MyD88KO and MyD88WT mice (Figure 1—figure supplement 1G).

Since submitting our revised manuscript to eLife we optimized our Nr4a1 primer design to improve specificity. This lowers non-specific background signal in Nr4a1-/- samples and has enabled us to detect endogenous Nr4a1 transcript directly ex vivo with greater sensitivity and dynamic range. Therefore, we reanalyzed Nr4a1 transcript expression in germ-free and SPF as well as B cell-conditional MyD88 KO and WT splenocytes alongside Nr4a1-/- controls. We have replaced Figure 1—figure supplement 1E to display Nr4a1 transcript in germ-free and SPF splenocytes analyzed immediately ex vivo, and we replaced Figure 1—figure supplement 1D to depict endogenous protein in splenic B cells from the same mice also analyzed directly ex vivo.

Figure 1—figure supplement 1D-G now collectively show no statistically significant differences in Nur77 protein or Nr4a1 transcript analyzed directly ex vivo between germ-free and SPF mice, as well as between B cell-conditional MyD88 KO and WT samples.

However, I do not agree that the Nur77-GFP signals in B cells are the consequence of self-antigen (self-antigen defined by an antigen encoded by the DNA of the self-organism) stimulation by the BCR. I disagree with the authors using self-antigen (subsection “Cell-intrinsic skewing of B cell development by the IgM and IgD BCRs”, first paragraph), endogenous auto-antigen (subsection “Either IgD or IgM BCR is sufficient to mediate polyclonal B cell activation and germinal center differentiation in Lyn−/− mice”, first paragraph) and antigen encounters in vivo (subsection “IgD BCRs sense endogenous antigens less efficiently than IgM BCRs”) as synonymous. I fully agree that Nur77-GFP signals are upregulated upon BCR stimulation by antigen and that antigen is necessary to drive Nur77-GFP expression in vivo. But I do not agree that self-antigen is necessary to drive Nur77-GFP expression in vivo. My arguments are:

• Nur77-GFP signals are highly upregulated when B cells leave the sterile/defined microenvironment of the BM. Nur77-GFP expression timely occurs when cells encounter "antigens" in the periphery, and those antigens activate the BCR. But these antigens are not necessarily self-antigens, they can be food-antigens, LPS (from the chow, antibodies against LPS are broadly recognized), and other non-pathological or non-inflammatory antigens recognized by the BCR. These antigens might activate B cells to upregulate Nur77, but in the absence of T cell help or an inflammatory milieu will not drive an immune response.

• The results with the IgHEL BCR Tg mice (Author response image 1) are fully compatible with the above-described scenario. These mice express a clonal BCR not able to recognize any self-antigen, but also, not able to recognize any of the possible non-inflammatory antigens that will be encountered in the periphery, for this reason the signal of Nur77 is lower. This interpretation is also compatible with higher level of Nur77-GFP in HEL Tg cells exposed to cognate HEL antigen that in WT. Related to this, the results show in Author response image 6 are also compatible with this scenario since Nur77-GFP upregulation is reduced in IgHEL BCR Tg mice in respond to LPS (small reduction), CpG (clear at lower concentration), and to Pam3CSK4 (visible at both concentrations). A possible interpretation of these results is that the HEL Tg BCR is not able to recognize epitopes in those substances that, together with the corresponding TLR, aid to activate Nur77-GFP in B cells.

• The results with the VH3H9 Tg mice (Author response image 3) are also fully compatible with the above-described scenario, since the κ+ cells (not recognizing endogenous DNA) also up-regulate (even to the same levels than λ+ cells) Nur-77 upon leaving the BM. Most probably because they encounter non-inflammatory antigens in the periphery.

Taken together, I fully agree that the present manuscript is improved upon revision and most of my requests have been clarify. However, I still request that the authors recognize that the Nur77-GFP signals in B cells might be the result of BCR stimulation by a broader collection of non-inflammatory antigens in vivo, not only self-antigens. They might be globally named as endogenous antigens but not exclusively self-antigens, or endogenous self-antigens. The main conclusion of the manuscript is unaltered, namely that IgD-BCR signals differently upon encounter of non-inflammatory antigens in vivo, but to narrow this conclusion only to self-antigens is not proven and not justified.

We gratefully acknowledge the possibility that endogenous antigens that signal through the BCR of naïve B cells in vivo may not exclusively represent “self” antigens encoded by the host genome or generated by enzymes encoded by the host genome. We have therefore modified the text of our revised manuscript to carefully eliminate such an implication. We have now consistently used the term “endogenous” rather than “self” antigens in order to encompass non-inflammatory exogenous antigens (such as food antigens) that may signal through the BCR. Specifically, we have modified citations in the first paragraph of the subsection “Cell-intrinsic skewing of B cell development by the IgM and IgD BCRs” and “Either IgD or IgM BCR is sufficient to mediate polyclonal B cell activation and germinal center differentiation in Lyn-/- mice”. We have gone on to also edit references throughout the manuscript in order to clarify the same point. Finally, we have made this point explicitly in our revised Discussion:

“These antigens may not be exclusively restricted to those that are germline-encoded or generated by host enzymes; rather they may include non-inflammatory antigens taken up through the gastrointestinal tract and recognized by specific BCRs.”

References:

Baumgarth, N. 2016. B-1 Cell Heterogeneity and the Regulation of Natural and Antigen-Induced IgM Production. Frontiers in immunology 7:324.

Bendelac, A., M. Bonneville, and J.F. Kearney. 2001. Autoreactivity by design: innate B and T lymphocytes. Nat Rev Immunol 1:177-186.

Cancro, M.P., and J.F. Kearney. 2004. B cell positive selection: road map to the primary repertoire? J Immunol 173:15-19.

Cariappa, A., and S. Pillai. 2002. Antigen-dependent B-cell development. Curr Opin Immunol 14:241-249.

Degn, S.E., C.E. van der Poel, D.J. Firl, B. Ayoglu, F.A. Al Qureshah, G. Bajic, L. Mesin, C.A. Reynaud, J.C. Weill, P.J. Utz, G.D. Victora, and M.C. Carroll. 2017. Clonal Evolution of Autoreactive Germinal Centers. Cell 170:913-926 e919.

Erikson, J., M.Z. Radic, S.A. Camper, R.R. Hardy, C. Carmack, and M. Weigert. 1991. Expression of anti-DNA immunoglobulin transgenes in non-autoimmune mice. Nature 349:331-334.

Fields, M.L., B.D. Hondowicz, G.N. Wharton, B.S. Adair, M.H. Metzgar, S.T. Alexander, A.J. Caton, and J. Erikson. 2005. The regulation and activation of lupus-associated B cells. Immunological reviews 204:165-183.

Freitas, A.A., M.P. Lembezat, and A. Coutinho. 1989. Expression of antibody V-regions is genetically and developmentally controlled and modulated by the B lymphocyte environment. Int Immunol 1:342-354.

Gu, H., D. Tarlinton, W. Muller, K. Rajewsky, and I. Forster. 1991. Most peripheral B cells in mice are ligand selected. J Exp Med 173:1357-1371.

Hayakawa, K., M. Asano, S.A. Shinton, M. Gui, L.-J. Wen, J. Dashoff, and R.R. Hardy. 2003. Positive selection of anti-thy-1 autoreactive B-1 cells and natural serum autoantibody production independent from bone marrow B cell development. J Exp Med 197:87-99.

Huizar, J., C. Tan, M. Noviski, J.L. Mueller, and J. Zikherman. 2017. Nur77 Is Upregulated in B-1a Cells by Chronic Self-Antigen Stimulation and Limits Generation of Natural IgM Plasma Cells. Immunohorizons 1:188-197.

Kraus, M., M.B. Alimzhanov, N. Rajewsky, and K. Rajewsky. 2004. Survival of Resting Mature B Lymphocytes Depends on BCR Signaling via the Igα/β Heterodimer. Cell 117:787-800.

Lam, K.P., R. Kuhn, and K. Rajewsky. 1997. In vivo ablation of surface immunoglobulin on mature B cells by inducible gene targeting results in rapid cell death. Cell 90:1073-1083.

Levine, M.H., A.M. Haberman, D.B. Sant'Angelo, L.G. Hannum, M.P. Cancro, C.A. Janeway, Jr., and M.J. Shlomchik. 2000. A B-cell receptor-specific selection step governs immature to mature B cell differentiation. Proc Natl Acad Sci U S A 97:2743-2748.

Lutz, C., B. Ledermann, M.H. Kosco-Vilbois, A.F. Ochsenbein, R.M. Zinkernagel, G. Kohler, and F. Brombacher. 1998. IgD can largely substitute for loss of IgM function in B cells. Nature 393:797-801.

Moran, A.E., K.L. Holzapfel, Y. Xing, N.R. Cunningham, J.S. Maltzman, J. Punt, and K.A. Hogquist. 2011. T cell receptor signal strength in Treg and iNKT cell development demonstrated by a novel fluorescent reporter mouse. The Journal of experimental medicine 208:1279- 1289.

Phan, T.G., S. Gardam, A. Basten, and R. Brink. 2005. Altered migration, recruitment, and somatic hypermutation in the early response of marginal zone B cells to T celldependent antigen. J Immunol 174:4567-4578.

Quach, T.D., N. Manjarrez-Orduno, D.G. Adlowitz, L. Silver, H. Yang, C. Wei, E.C. Milner, and I. Sanz. 2011. Anergic responses characterize a large fraction of human autoreactive naive B cells expressing low levels of surface IgM. J Immunol 186:4640-4648.

Reed, J.H., J. Jackson, D. Christ, and C.C. Goodnow. 2016. Clonal redemption of autoantibodies by somatic hypermutation away from self-reactivity during human immunization. The Journal of experimental medicine 213:1255-1265.

Rosado, M.M., and A.A. Freitas. 1998. The role of the B cell receptor V region in peripheral B cell survival. Eur J Immunol 28:2685-2693.

Sabouri, Z., S. Perotti, E. Spierings, P. Humburg, M. Yabas, H. Bergmann, K. Horikawa, C. Roots, S. Lambe, C. Young, T.D. Andrews, M. Field, A. Enders, J.H. Reed, and C.C. Goodnow. 2016. IgD attenuates the IgM-induced anergy response in transitional and mature B cells. Nat Commun 7:13381.

Sabouri, Z., P. Schofield, K. Horikawa, E. Spierings, D. Kipling, K.L. Randall, D. Langley, B. Roome, R. Vazquez-Lombardi, R. Rouet, J. Hermes, T.D. Chan, R. Brink, D.K. Dunn-Walters, D. Christ, and C.C. Goodnow. 2014. Redemption of autoantibodies on anergic B cells by variable-region glycosylation and mutation away from self-reactivity. Proc Natl Acad Sci U S A 111:E2567-2575.

Song, H., and J. Cerny. 2003. Functional heterogeneity of marginal zone B cells revealed by their ability to generate both early antibody-forming cells and germinal centers with hypermutation and memory in response to a T-dependent antigen. The Journal of experimental medicine 198:1923-1935.

Su, T.T., B. Guo, B. Wei, J. Braun, and D.J. Rawlings. 2004. Signaling in transitional type 2 B cells is critical for peripheral B-cell development. Immunol Rev 197:161-178.

Thomas, M.D., B. Srivastava, and D. Allman. 2006. Regulation of peripheral B cell maturation. Cell Immunol 239:92-102. Wang, H., and S.H. Clarke. 2003. Evidence for a ligand-mediated positive selection signal in differentiation to a mature B cell. J Immunol 171:6381-6388.

Wardemann, H., S. Yurasov, A. Schaefer, J.W. Young, E. Meffre, and M.C. Nussenzweig. 2003. Predominant autoantibody production by early human B cell precursors. Science 301:1374-1377.

Yang, Y., C. Wang, Q. Yang, A. Kantor, H. Chu, E. Ghosn, G. Qin, S. Mazmanian, J. Han, and L. Herzenberg. 2015. Distinct mechanisms define murine B cell lineage immunoglobulin heavy chain (IgH) repertoires. eLife 4:e09083.

Zikherman, J., R. Parameswaran, and A. Weiss. 2012. Endogenous antigen tunes the responsiveness of naive B cells but not T cells. Nature 489:160-164.

https://doi.org/10.7554/eLife.35074.052

Article and author information

Author details

  1. Mark Noviski

    Biomedical Sciences (BMS) Graduate Program, University of California San Francisco, San Francisco, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-8072-1059
  2. James L Mueller

    Rosalind Russell and Ephraim P. Engleman Arthritis Research Center, Division of Rheumatology, Department of Medicine, University of California San Francisco, San Francisco, United States
    Contribution
    Formal analysis, Investigation
    Competing interests
    No competing interests declared
  3. Anne Satterthwaite

    Department of Immunology, UT Southwestern Medical Center, Dallas, United States
    Contribution
    Resources, Writing—review and editing
    Competing interests
    No competing interests declared
  4. Lee Ann Garrett-Sinha

    Department of Biochemistry, University at Buffalo, The State University of New York, Buffalo, United States
    Contribution
    Resources, Writing—review and editing
    Competing interests
    No competing interests declared
  5. Frank Brombacher

    1. International Center for Genetic Engineering and Biotechnology (ICGEB), Cape Town, South Africa
    2. Institute of Infectious Diseases and Molecular Medicine, Division of Immunology, Faculty of Health Sciences, University of Cape Town & Medical Research Council (SAMRC), Cape Town, South Africa
    Contribution
    Resources, Writing—review and editing
    Competing interests
    No competing interests declared
  6. Julie Zikherman

    Rosalind Russell and Ephraim P. Engleman Arthritis Research Center, Division of Rheumatology, Department of Medicine, University of California San Francisco, San Francisco, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Investigation, Methodology, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    julie.zikherman@ucsf.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0873-192X

Funding

National Science Foundation (1650113)

  • Mark Noviski

National Institute of Arthritis and Musculoskeletal and Skin Diseases (R01 A127648)

  • Julie Zikherman

Rheumatology Research Foundation

  • Julie Zikherman

National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS K08 AR059723)

  • Julie Zikherman

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Anthony DeFranco, Arthur Weiss, and Jason Cyster for valuable feedback and suggestions for this manuscript, and Al Roque for help with mouse husbandry. We thank Renuka Nayak, Anne-Katrin Proebstel, and Eric Wigton for sharing valuable mouse reagents. Funding for this project was provided by the Rheumatology Research Foundation (JZ), NIAMS K08 AR059723 and R01 A127648 (JZ), and NSF Graduate Research Fellowship 1650113 (MN).

Ethics

Animal experimentation: This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All of the animals were handled according to approved institutional animal care and use committee (IACUC) protocols of the University of California, San Francisco. The protocol was approved by the IACUC committee of the University of California, San Francisco (Protocol number: AN171020-01).

Reviewing Editor

  1. Facundo D Batista, Ragon Institute of MGH, MIT and Harvard, United States

Publication history

  1. Received: January 13, 2018
  2. Accepted: March 8, 2018
  3. Accepted Manuscript published: March 9, 2018 (version 1)
  4. Version of Record published: April 12, 2018 (version 2)

Copyright

© 2018, Noviski et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 3,517
    Page views
  • 570
    Downloads
  • 4
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, Scopus, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Immunology and Inflammation
    2. Microbiology and Infectious Disease
    Adi Naamati et al.
    Tools and Resources
    1. Immunology and Inflammation
    Mark Austin Hanson et al.
    Research Article Updated