Functionally Responsive Self-Reactive B Cells of Low-Affinity Express Reduced Levels of Surface IgM

SUMMARY

Somatic gene rearrangement generates a diverse repertoire of B cells, including B cell receptors (BCR) possessing a range of affinities for self-Ag. Newly generated B cells express high and relatively uniform amounts of surface IgM (sIgM), while follicular (FO) B cells express sIgM at widely varying levels. It is plausible, therefore, that down-modulation of sIgM serves as a mechanism to maintain weakly self-reactive B cells in a responsive state by decreasing their avidity for self-Ag. We tested this hypothesis by performing comparative functional tests with FO IgM^hi and IgM^lo B cells from the unrestricted repertoire of wildtype (WT) mice. We found that FO IgM^lo B cells mobilized Ca²⁺ equivalently to IgM^hi B cells when the same number of sIgM molecules was engaged. In agreement, FO IgM^lo B cells were functionally competent to produce an antibody response following adoptive transfer. The FO IgM^lo cell population had elevated levels of Nur77 transcript, and was enriched with nuclear-reactive specificities. Hybridoma sampling revealed that these BCR were of low affinity. Collectively, these results suggest that sIgM down-modulation by low-affinity, self-reactive B cells preserves their immunocompetence and circumvents classical peripheral tolerance mechanisms that would otherwise reduce diversity within the B cell compartment.

INTRODUCTION

Somatic recombination of V(D)J gene segments generates an enormously diverse B cell repertoire with a high incidence of self-reactivity [1, 2]. Negative selection of autoreactive BCR largely occurs during the immature B cell stage following Igκ rearrangement [3-6]. Autoreactive BCR can either be edited through secondary rearrangements within the κ/λ loci or removed completely by clonal deletion [7-9]. B cells are subject to additional tolerance checkpoints in the periphery resulting in arrested development, anergy and exclusion from the follicles [10-19]. Despite these multiple tiers of censorship, B cells with some degree of self-reactivity appear to be present in the mature pool [20-23].

Much of our knowledge regarding how mechanisms of central and peripheral tolerance shape the B cell repertoire derives from mouse models involving both conventional and site-directed immunoglobulin transgenes (Ig-Tg) [4, 5, 24]. These Ig-Tg models facilitate the study of self-tolerance by providing large and relatively uniform populations of B cells expressing BCR with known self-reactive specificities. At the same time, these models have shortcomings that include somatically mutated BCR, high-affinity for self-Ag and severely reduced interclonal competition by nonautoreactive B cells. In light of these limitations, and with the advancement of new technologies, several groups are revisiting these tolerance mechanisms in the context of diverse WT repertoires [20, 21, 25-31].

Of particular interest is the role of anergy in maintaining peripheral self-tolerance in B cells. Unlike central tolerance mechanisms, through which the BCR is either changed or removed, the original BCR specificity is preserved in anergic B cells. A salient and unifying feature observed in Ig-Tg models of B cell anergy is down-modulation of surface IgM (sIgM) [11, 12, 18, 32]. Therefore, we sought to address whether there was evidence that WT FO B cells with low levels of sIgM had encountered self-Ag in vivo, and whether such cells remained immunocompetent. To this end, we assessed FO IgM^lo B cells for a signature molecule of BCR engagement. We also evaluated the potential of these cells to respond to BCR and TLR stimuli in vitro and to produce an antibody response in vivo. Our results support the view that B cell encounters with self-Ag in vivo induce sIgM down-modulation and functional preservation of low-affinity, self-reactive B cells within the FO repertoire.

RESULTS

The amount of surface IgM varies widely among follicular B cells

It is a common observation that the amount of surface IgM (sIgM) varies widely among follicular (FO) B cells of wildtype (WT) mice. To exclude the possibility that this might be due to differences in cell size, we assessed the distribution of sIgM on electronically gated FO B cells within tightly restricted forward and side scatter profiles. In this, and all experiments of our study, we utilized fluorescently-coupled, monovalent Fab reagents generated from the high-affinity rat anti-mouse IgM (μ-specific) mAb b7-6 [33] to avoid BCR cross-linking, internalization and B cell activation. The gating scheme used for identification of size-restricted FO B cells is presented in Figure 1A and 1B. As shown in Figure 1C, the size-restricted FO B cell population from B6 mice still produced the characteristic broad distribution of fluorescence intensity when stained with Fab b7-6, indicating that size alone cannot account for the varying levels of sIgM expression. In addition, FO IgM^lo B cells possessed significantly reduced quantities of intracellular IgM in comparison to both FO IgM^int and FO IgM^hi B cells (Figure 1D). The difference in intracellular Igμ (~74 kDa) protein expression between FO IgM^lo and IgM^hi B cells was also confirmed by western blot analysis (Figure S1) [34].

Figure 1. Surface and intracellular IgM expression by FO B cells.

(A and B) Scheme for identification of size-restricted FO B cells (B220^pos CD23^hi) electronically gated for a narrow distribution of forward and side scatter (area of ■). (C) Representative histogram overlay of electronically gated IgM^lo (dashed), IgM^int (solid) and IgM^hi (dotted) FO B cells. Percent of total FO B cell population (gray shaded area) +/− SEM is also shown. (D) Mean fluorescence intensity (MFI) +/− SEM for intracellular IgM expression by the FO B cell populations (n=7). Data are combined from 2 independent experiments. Significance was determined using a two-tailed paired student t-test (*** p<0.0001).

Surface IgM^lo follicular B cells are BCR responsive

To determine if FO IgM^lo B cells from B6 mice possessed classical features of anergy, such as elevated basal Ca²⁺ and an impaired Ca²⁺ flux following sIgM aggregation [12, 35, 36], we loaded spleen cells with the fluorescent Ca²⁺ indicator Indo-1. Splenocytes were then stained for additional markers to discriminate the mature FO B cell compartment, and Fab b7-6 was used to segregate these cells according to sIgM status. Retrospective analysis revealed a trend for increased basal Ca²⁺ concentration in the FO IgM^lo B cell population prior to stimulation, with some variation among experiments (Figure 2). At a fixed dose of GαMμ, B cells with low levels of sIgM fluxed less Ca²⁺ than FO B cells with either intermediate or high levels of sIgM (Figure 2A). In addition, FO IgM^int B cells reproducibly mobilized less Ca²⁺ than IgM^hi cells, but more than IgM^lo cells, suggesting that the magnitude of Ca²⁺ flux might be proportional to the number of receptors cross-linked.

Figure 2. BCR responsiveness of FO B cells expressing different levels of surface IgM.

(A) Representative Ca²⁺ traces of mature FO B cells (B220^pos CD23^hi CD24^int) expressing low (green), intermediate (blue), or high (red) levels of sIgM following stimulation with GαMμ (2.5 μg/ml). Inset shows representative gating of IgM^lo and IgM^hi populations, each comprising 10% of the population, for all panels shown in this figure. (B) FO IgM^lo B cells stimulated with increasing concentrations of GαMμ (2.5, 10, 25 and 50 μg/ml) in comparison to anergic Ars/A1 B cells (solid black line) stimulated with GαMμ (50 μg/ml). (C and E) Ca²⁺ mobilization by mature FO B cells expressing low (green) or high (red) levels of sIgM upon stimulation with GαMδ. (D and F) Ca⁺² mobilization of mature FO B cells expressing low (green) or high (red) levels of sIgM upon stimulation with GαMκ. Data are representative of 3 or more independent experiments per panel.

We next sought to determine whether the observed hyporesponsiveness of the FO IgM^lo B cell population was the result of insufficient receptor engagement. As shown in Figure 2B, FO IgM^lo B cells were capable of mobilizing intracellular Ca²⁺ to progressively greater degrees in response to increasing concentrations of stimulatory GαMμ. The ability of FO IgM^lo B cells to mobilize Ca²⁺ in response to an increased concentration of GαMμ stood in contrast to the behavior of anergic Ars/A1 B cells, which did not flux Ca²⁺ in response to the highest concentration, despite expressing similar levels of sIgM (Figure 2B and data not shown).

We also analyzed Ca²⁺ mobilization following stimulation with antibodies against Igδ and Igκ chains. In contrast to their reduced responsiveness to a fixed quantity of GαMμ, FO IgM^lo and IgM^hi B cells responded similarly to GαMδ (Figure 2C and 2E). FO IgM^lo B cells express significantly though only slightly lower levels of sIgδ than do FO IgM^hi B cells (Figure S2). As expected, FO IgM^lo B cells responded somewhat more weakly to GαMκ than did FO IgM^hi B cells when differences in baseline Ca²⁺ levels are taken into consideration (Figure 2D and 2F). Collectively, these data indicate that FO IgM^lo B cells are responsive to BCR stimulation and suggest that their reduced Ca²⁺ flux in response to anti-μ may be due to a lower degree of sIgM engagement.

Equivalent IgM stimulation induces indistinguishable Ca²⁺ fluxes

Although the preceding experiments indicated that the FO IgM^lo B cell subset was responsive to anti-μ stimulation, it remained possible that IgM^lo B cells are less responsive than IgM^hi cells on a per sIgM molecule basis. Addressing this question precisely required a modified approach to attain equivalent sIgM ligation on IgM^hi and IgM^lo cells. This was achieved by labeling the entire FO B cell pool with limiting concentrations of Bio-Fab b7-6, such that a low concentration of Bio-Fab b7-6 would engage an equivalent number of IgM molecules on IgM^hi B cells as a high concentration would engage on IgM^lo B cells. Titration of the Bio-Fab b7-6 resulted in both dose-dependent staining and Ca²⁺ mobilization of the entire FO B compartment upon addition of fluorochrome coupled streptavidin (Figure S3A and S3B). The fluorescent streptavidin served to both aggregate the Bio-Fab b7-6 and to provide a measure of the degree of sIgM ligation, which in turn, could be associated with the magnitude of Ca²⁺ mobilization (Figure 3A). To validate our approach, we measured residual unoccupied sIgM on the gated populations with DyLight 488-Fab b7-6. In these experiments, B cells that were originally treated with a high concentration of Bio-Fab b7-6 had less residual sIgM than did B cells with a similar fluorescence intensity that were originally treated with a low concentration of Bio-Fab b7-6 (Figure 3B). This staining procedure produced a clear separation in residual sIgM staining for B cell populations labeled with different concentrations of Bio-Fab b7-6 (Figure S3C). When B cells expressing different levels of total sIgM were stimulated through an equivalent number of IgM receptors in this way, they fluxed Ca²⁺ indistinguishably (Figure 3C). These data indicate that in response to anti-μ stimulation, sIgM molecules on FO IgM^lo cells are equally competent as those on IgM^hi cells for mediating a Ca²⁺ flux.

Figure 3. Similar Ca²⁺ responses by FO B cells expressing different levels of sIgM upon equivalent sIgM engagement.

(A) Representative staining of FO B cells (B220^pos CD23^hi) pre-labeled with low (red, 0.5 μg/ml), intermediate (blue, 2 μg/ml), or high (green, 5 μg/ml) concentrations of Bio-Fab b7-6 following addition of DyLight 649 conjugated streptavidin (1 μg/ml). (B) Residual unoccupied sIgM on B cells within the defined fluorescence (gated population) was detected with DyLight 488-Fab b7-6. (C) Ca²⁺ traces of B cells expressing the defined fluorescence following addition of DyLight 649 conjugated streptavidin. Data are representative of 3 independent experiments with similar results.

In vivo IgG1κ production by follicular IgM^lo B cells

To determine whether FO IgM^lo B cells were functionally competent in vivo, we modified an adoptive transfer approach previously described by Heiser et al. [37]. FACS purified mature FO B cells from A/J donors were segregated into IgM^hi and IgM^lo populations and transferred into A/J Igκ-deficient recipients, which were subsequently immunized with GαMκ in alum (Figure S4). The GαMκ acts as a polyclonal, T-cell dependent immunogen enabling us to assess the potential of Igκ⁺ B cells to produce an antibody response in vivo. In this experiment, FO IgM^lo B cells produced similar quantities of IgG1κ relative to FO IgM^hi cells (Figure 4A). Additional experiments utilizing the immunogen Protein L, an Igκ-binding surface protein expressed by Peptococcus magnus [38], yielded similar results (Figure 4B). We conclude that mature IgM^lo cells are functionally competent to produce an antibody response in vivo following immunization.

Figure 4. In vivo antibody production by FO IgM^lo B cells.

Total IgG1κ on day 14 following transfer of FACS purified mature FO IgM^lo or IgM^hi B cells into A/J κ^−/− recipients. (A) Transfer of 2.5 × 10⁵ FO IgM^lo or IgM^hi B cells and immunization with 50 μg of GαMκ in alum. (B) Transfer of 5 × 10⁴ mature FO IgM^lo or IgM^hi B cells and immunization with 50 μg of Protein L in alum. In both experiments, mature FO B cells were defined as B220^pos CD23^hi CD24^int, and the IgM^lo and IgM^hi populations constituted the lowest and highest 10% of the FO B cell population respectively. Data are representative of 2-3 independent transfer experiments.

Follicular IgM^lo B cells are enriched for nuclear-reactive specificities

We then addressed whether autoreactive cells were enriched in the mature FO IgM^lo compartment. B6 FO B cell populations comprising the highest and lowest 10-15% of sIgM expressing cells were isolated and subsequently cultured with LPS to induce differentiation and antibody secretion. After 7 days, concentrations of IgMκ and IgG3κ in culture supernatants were determined. We focused on μ and γ3 isotypes because γ1, γ2a and γ2b were nearly undetectable (data not shown). The quantity of IgMκ produced by FO IgM^lo B cells was generally less than that of FO IgM^hi cells (Figure 5A), while the amount of IgG3κ produced by the two populations did not differ significantly (Figure 5B). Similar results were observed using FACS purified FO B cells isolated from pooled peripheral lymph nodes, suggesting that the increased IgMκ produced by FO IgM^hi B cells was unlikely to be accounted for by contaminating marginal zone B cells (data not shown). These results indicate that mature FO IgM^lo B cells are capable of antibody secretion and class-switch recombination following stimulation through TLR4.

Figure 5. Anti-nuclear antibody production by FO IgM^lo B cells.

(A) Total IgMκ and (B) IgG3κ in culture supernatants quantified by interpolation using isotype-specific standards. Results are presented as the mean +/− SEM of 3 combined experiments. Significance was determined using a two-tailed paired student t-test (*p=0.0493). Immunoassay for Igκ directed against (C) calf chromatin and (D) ssDNA. Eu³⁺ counts were plotted versus normalized total Igκ in supernatant (x-axis). 10⁵ FACS purified mature FO (B220^pos CD23^hi CD93^neg) IgM^lo or IgM^hi B cells were cultured in vitro for 7 days with 25 μg/ml LPS. Results of three independent experiments are presented with symbols matching individual samples. (E) Immunoassay for ssDNA-reactive IgM secreted from B cell hybridomas derived from mature FO IgM^lo and IgM^hi B cells. Mature FO B cells were defined as B220^pos CD23^hi CD24^int cells. IgM^lo and IgM^hi populations constituted the lowest and highest 10% of the mature FO B cell population respectively. The two dashed black lines show binding by anti-DNA IgM (A35-1.20 and 2A7-1.3) isolated from an autoimmune B6.Nba2 mouse and illustrate low affinities of the experimental mAb. (F) Graphical representation of B6 FO IgM^lo or FO IgM^hi derived mAbs that were reactive with ssDNA, calf chromatin or BSA. P-value was determined using Fischer's exact test for association between mAbs population of origin and reactivity for the model antigens.

We then tested culture supernatants for the presence of Igκ reactive with either ssDNA or a crude preparation of bovine chromatin in solid phase immunoassays. Binding results were normalized with respect to the quantity of total Igκ in supernatants. In three independent experiments, supernatants from FO IgM^lo B cells produced substantially more Igκ reactive against ssDNA and chromatin than did IgM^hi cells (Figure 5C and 5D). Near linear binding curves hinted at low-affinity binding, as expected for self-reactive cells that survive in the periphery. The difference in self-reactivity between antibody produced by IgM^hi and IgM^lo cells, shown in Figure 5C and 5D, is likely to be an underestimate because a larger proportion of the Igκ secreted by the IgM^hi cells was multivalent IgM (Figure 5A). These findings indicate that the mature FO IgM^lo compartment is enriched in cells with nuclear-reactive BCR.

To test our interpretations at the monoclonal level, we generated hybridomas from sorted mature IgM^hi and IgM^lo FO B cells that were stimulated in vitro with LPS (Figure S4). The mAb derived from the FO IgM^lo and IgM^hi populations were screened for ssDNA-reactivity, which could now be assessed more precisely as a function of IgM concentration in cloned samples. As predicted, the frequency of hybridomas producing IgM reactive with ssDNA was highest in the IgM^lo population (Figure 5E). In addition, these binding assays with defined quantities of IgM demonstrated that the mAbs derived from FO IgM^lo and IgM^hi B cells of B6 mice were only weakly reactive with ssDNA and thus of low affinity. Moreover, when the ssDNA binding data were combined with those for two additional antigens, chromatin and BSA, the FO IgM^lo derived mAb panel was significantly enriched for polyreactivity (Figure 5F). In sum, the results of our mAb studies were in agreement with those generated using the polyclonal populations, and suggest an enrichment of low-affinity, self-reactive BCR specificities within the FO IgM^lo compartment.

Disproportional IgM internalization by follicular IgM^lo B cells

To determine whether FO IgM^lo B cells from B6 mice internalize IgM more readily in the steady state than do FO IgM^hi B cells, we stained splenocytes with FITC-Fab GαMμ, followed by Bio-Fab b7-6. At various time-points (0-60 min) the cells were fixed, stained for additional markers and the amount of residual sIgM expression revealed with a fluorescently-coupled streptavidin. This approach enabled stable discrimination of FO B cells according to their initial sIgM status and independent tracking of sIgM. As shown in Figure 6, the FO IgM^lo B cell population retained a significantly reduced proportion of labeled IgM on the cell surface compared to either the FO IgM^int or FO IgM^hi B cell populations. Moreover, the FO IgM^int population also exhibited a reduced proportion of labeled IgM on the cell surface relative to the FO IgM^hi population. These results are consistent with our interpretation that BCR engagement of self-Ag is highest on FO IgM^lo cells.

Figure 6. Surface IgM internalization by FO B cells expressing different levels of IgM.

The proportion of sIgM internalization was assessed on mature FO B cells (B220^pos CD23^hi CD24^int) expressing low (▼), int (▲) or high (■) levels of sIgM. Splenocytes were first labeled with FITC-Fab GαMμ, followed by Bio-Fab b7-6. After washing, the cells were incubated at 37°C for 0, 15, 30 or 60 min and immediately fixed in 1.6% paraformaldehyde. The cells were then stained with additional markers (B220, CD23 and CD24) to identify the mature FO B cell population and the remaining (noninternalized) sIgM revealed with PE-conjugated strepavidin. FITC staining remained stable as demonstrated in a prior experiment using FACS purified, IgM^hi and IgM^lo FO B cells stained with FITC-Fab GαMμ (data not shown). Data are combined from 2 independent experiments (n=7) in which each sample was performed in triplicate. Significance was determined using a two-tailed paired student t-test (*** p<0.0001).

Follicular IgM^lo B cells possess elevated levels of Nur77 transcript ex vivo

Collectively, our results were consistent with the hypothesis that the low level of sIgM on some FO B cells is due to receptor down-modulation upon engagement of self-Ag in vivo. If so, FO IgM^lo B cells should possess a biochemical signature of activation such as up-regulation of Nur77, an early response gene that is induced following BCR engagement [39]. To assay for Nur77 expression, we performed qPCR analyses on B6 FO B cell populations expressing different levels of sIgM. As a positive control, we assessed Nur77 transcript abundance within the marginal zone B cell compartment because this subset is enriched for polyreactive cells [40]. As shown in Figure 7A, qPCR revealed significantly more Nur77 transcripts in mature FO IgM^lo cells than in mature FO IgM^hi B cells. This result was confirmed using two independent primer sets. FO IgM^int B cells from B6 mice also had more Nur77 transcripts than FO IgM^hi cells. As expected, marginal zone B cells had relatively high transcript levels as well. These results indicate that FO IgM^lo cells had encountered Ag in vivo, and likely self-Ag in light of the specificity studies described above.

Figure 7. Nur77 transcript and protein expression in FO B cells expressing high and low levels of surface IgM.

(A) Quantitative PCR for Nur77 transcript in FACS purified mature B cells. Data are comprised of 3-4 B cell samples using 2 different PCR primer sets. Expression is relative to HPRT and was normalized to FO IgM^hi B cells. (B) Representative Nur77 FACS staining following 4 h of continuous in vitro culture with media (solid) or 50 μg/ml GαMμ (dashed). Isotype staining control (IgG1κ) for GαMμ culture (gray shaded). (C) Nur77 MFI + SEM expression by FO B cells from B6 (n=4) or Ars/A1 (n=1) mice following acute stimulation with decreasing concentrations of GαMμ. Isotype control staining was subtracted. (D) Representative Nur77 histogram overlay of FO B cells following acute stimulation with 3 μg/ml GαMμ. IgM^lo (light gray), IgM^int (dark gray) and IgM^hi (black). (E) Nur77 MFI expression of FO B cells expressing low (▼), int (▲) or high (■) levels of sIgM following acute stimulation with decreasing concentrations of GαMμ. Isotype control staining was subtracted. Data are representative of three independent experiments comprising ≥ 3 mice of either genotype. Significance was determined using a two-tailed paired student t-test (**p< 0.01, *p<0.05).

Follicular IgM^lo B cells up-regulate Nur77 protein following anti-μ stimulation

We then assayed for Nur77 protein expression in B6 FO B cells by flow cytometry, but were unable to detect staining above background levels ex vivo (data not shown). However, following a 4 h continuous in vitro stimulation of splenocytes with GαMμ, Nur77 protein was readily detectable in mature FO B cells (Figure 7B). In the absence of anti-μ stimulation, a low level of Nur77 staining above background was also observed, which is consistent with the spontaneous activation of B cells upon in vitro culture reported by Snyder et al. [41].

We next sought to determine whether Nur77 protein induction correlated with the strength of anti-μ stimulus. To this end, we modified our previous in vitro culture system in which the anti-μ stimulus was present throughout the culture period. Instead, we pulsed splenocytes with varying concentrations of GaMμ on ice and washed the cells to remove unbound stimulatory antibody prior to in vitro culture. The absence of stimulatory antibody in the culture limited signaling to a single round of BCR aggregation, thus enabling the degree of Nur77 induction to be directly correlated with the concentration of the anti-μ stimulus. In this regard, the approach paralleled our Ca²⁺ flux assays.

In contrast to anergic Ars/A1 B cells, which constitutively expressed high levels of Nur77 protein that remained relatively unchanged following GαMμ stimulation, mature FO B cells were induced to synthesize Nur77 regardless of sIgM status (Figure 7C). Furthermore, IgM^hi cells responded to a greater extent at low concentrations of anti-μ than did IgM^int or IgM^lo cells, in agreement with our Ca²⁺ mobilization results (Figure 7D). FO IgM^lo B cells were capable of robust Nur77 induction, eventually reaching the same maximum observed with IgM^hi B cells at high concentrations of anti-μ (Figure 7E). These data demonstrate that FO IgM^lo B cells from B6 mice were responsive to anti-μ stimulation and capable of Nur77 protein induction. Additionally, they indicate that, as in the case of Ca²⁺ mobilization, induction of Nur77 was proportional to the amount of sIgM on the FO B cell and to the strength of the anti-μ stimulus.

DISCUSSION

In this study, we demonstrate that the population of mature FO B cells expressing low levels of sIgM is enriched with clones that are weakly reactive with nuclear antigens. Relative to FO IgM^hi cells, FO IgM^lo cells had elevated levels of Nur77 transcripts, indicating that they had likely encountered self-Ag in vivo. Nevertheless, they appeared to be functionally competent. They were able to mobilize Ca²⁺ equivalently to FO IgM^hi cells when an equivalent number of sIgM molecules were engaged. And upon adoptive transfer and immunization, they generated an IgG antibody response. Collectively, these findings support the view that sIgM is down-modulated in response to weak interactions with self-Ag in vivo, enabling low-affinity self-reactive B cells to avoid negative selection and contribute to BCR diversity in the mature immunocompetent repertoire.

It is possible that there is functional heterogeneity within the IgM^lo population. For example, the IgM^lo population may contain anergic B cells, which are unable to flux Ca²⁺ even in response to high concentrations of anti-μ. However, in view of our results showing equivalent Ca²⁺ fluxes in FO IgM^hi and IgM^lo cells upon equivalent receptor engagement with anti-μ, anergic B cells would have to constitute only a minor fraction of the IgM^lo population. Alternatively, anergic B cells could be present at substantial frequencies in both IgM^lo and IgM^hi populations, provided that the frequencies are similar within both. In this vein, the possibility that WT anergic B cells may be present within both the IgM^hi and IgM^lo repertoires is supported by findings of Taylor et al. [21]. In their study, WT B cells rendered anergic in vivo by membrane-bound ovalbumin had no specific sIgM phenotype.

We also must consider the possibility that some FO B cells may express low levels of sIgM due purely to variation in V gene promoter strength or transcript stability, or at the protein level due to differences in rate of translation or in efficiency of pairing between specific heavy and light chain V regions. Any one of these could influence the level of sIgM independent of BCR specificity. While our experiments were unable to formally exclude the contributions of these to the heterogeneity in sIgM levels, we think the extent to which these mechanisms contribute to the observed IgM^lo phenotype is likely to be modest at best. This is because FO IgM^lo and IgM^hi cells both express high levels of sIgD and because developing B cells express relatively uniform levels of sIgM through the transitional 2 developmental stage [42, 43]. Additionally, we did not observe a difference in the quantity of IgM produced by FO IgM^lo and IgM^hi derived hybridomas (data not shown), which otherwise would have provided support for these alternative mechanisms. Collectively, these observations support our interpretation that a substantial fraction of the IgM^lo B cell population had down-modulated sIgM following encounter with self-Ag in vivo.

Although we evaluated the IgM^lo and IgM^hi B cell populations for binding against three model self-Ags, it is unclear exactly which Ags induce sIgM down-modulation in vivo. While we did detect ssDNA-reactive IgM in our hybridoma screen, a majority of the IgM^pos samples from both the IgM^lo and IgM^hi populations lacked appreciable binding to this antigen. From each population, we also tested 15 mAb that were negative for binding in our initial screen for reactivity against HEp-2 cells and observed negligible reactivity (data not shown). Therefore, it is possible that the IgM^lo B cell population possesses BCR that recognize a wide array of self-Ag including those expressed by B cells themselves and perhaps Ag associated with commensal microbiota.

We found that FO IgM^lo B cells underwent a robust Ca²⁺ flux in response to stimulation with anti-Igδ. At present, it is not clear whether human FO B cells expressing varying levels of sIgM have similar response characteristics. Prior studies of IgD^pos IgM^neg (B_ND) or IgM^lo B cell subsets in human peripheral blood indicated that these populations are hyporesponsive to IgD stimulation [22, 23]. These differences suggest that IgD signaling may be differentially regulated in IgM^lo/neg B cells of the two species. Alternatively, this discrepancy might be attributed to the use of different staining and stimulation reagents. In our study, Fab reagents were exclusively used during cell isolation and for phenotypic identification during Ca²⁺ mobilization assays. Regardless of potential differences in responses to anti-δ by human and mouse IgM^lo B cells, IgM^lo cells of both species were found to be enriched in self-reactive B cells (Figure 5 and [22, 23]).

Our results are in close agreement with those of a recent report by Zikherman et al., who approached the same general problem from a different angle [20]. These investigators demonstrated a positive correlation between BCR signal strength and expression of a Nur77-GFP reporter. They also reported that FO B cells with the highest levels of GFP trended towards lower levels of sIgM expression and were enriched with nuclear-reactive antibody specificities. Although our focus on FO IgM^lo B cells enabled us to arrive at many of the same conclusions, it appears that the populations analyzed in the two studies were not completely overlapping. First, sIgM expression did not correlate as closely with elevated basal Ca²⁺ as did Nur77-GFP reporter expression. Additionally, we did not detect elevated levels of Nur77 protein in the FO IgM^lo B cell subset directly ex vivo. This discrepancy could be the result of sensitivity limitations in detecting Nur77 protein by flow cytometry. Alternatively, it could be due to differences in the lifespan of Nur77 versus GFP protein. In support of the latter explanation, the half-life of the GFP reporter was reported to be longer than that of Nur77 protein in CD8+ T cells [44]. This suggests that Nur77 may be post-transcriptionally regulated. Notably, because the GFP gene was driven by the Nur77 promoter in the Zikherman et al. study [20], they too assessed transcription not translation of Nur77.

The notion that lymphocytes are capable of modifying their activation threshold following encounter with antigen was initially suggested by Grossman and Paul, and is referred to as the tunable activation threshold (TAT) model [45, 46]. Adoptive transfer studies utilizing Tg (5C.C7) CD4⁺ T cells support the TAT model and suggest that signaling is modulated in response to self-Ag in part by upregulation of CD5 [47-49]. In this model, T cells avoid clonal deletion and attain tolerance to the steady state level of self-Ag without a concomitant reduction in surface TCR. In contrast, anti-nuclear (HKIR) B cells maintain tolerance towards self-Ag through down-modulation of surface BCR [50, 51]. Thus, in response to steady-state self-Ag, T and B cells tune activation thresholds in distinct ways. In this vein, we did not observe elevated levels of CD5 on sIgM^lo cells (Figure S5).

While the presence of functional B cells with weak reactivity for endogenous self-Ags could be viewed as a failure in peripheral tolerance, the low affinity of these cells suggests otherwise. An alternative perspective is that maintenance of polyreactivity in the repertoire provides a benefit by enhancing the immune system's ability to respond to diverse pathogens. In this vein, we speculate that sIgM down-modulation is an important physiologic mechanism for preserving polyreactive BCR in the immunocompetent repertoire. This position is supported by a report demonstrating polyreactivity (including anti-DNA activity) among gp120-specific mAb capable of neutralizing human immunodeficiency virus (HIV) [52]. It is plausible that polyreactivity enhances Ab or BCR binding of the widely spaced HIV spike proteins, as suggested by the authors. Finally, we speculate that the ability of B cells with low-affinity receptors for viral antigens to efficiently transport virus-like particles from the lung to the spleen, as shown by Bessa et al., might have been facilitated by polyreactive B cells possessing an IgM^lo phenotype [53]. These reports provide insight into potential benefits afforded by maintaining BCR with degrees of polyreactivity in the immunocompetent B cell repertoire.

MATERIALS AND METHODS

Mice

C57BL/6 and A/J mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and maintained in the Biological Resource Center at National Jewish Health (Denver, CO). Mice carrying the p-azophenylarsonate (Ars) specific canonical Ig μδ and κ transgenes (Ars/A1) were maintained on the B6 genetic background [12]. Mice deficient in Igκ due to a targeted deletion [54] were bred to the A/J genetic background (A/J Igκ-deficient) for >10 backcross generations [37] and were used as recipients for adoptive transfer experiments. Mice used for experiments were generally 2-4 months old. All mice were handled and bred with Institutional Animal Care and Use Committee approval in accordance with institutional guidelines.

Antibodies

The antibody/fluorescent reagents used were: 1) from Biolegend (San Diego, CA): mAb RA3-6B2 (B220), mAb 1B1 (CD1d), mAb 93 (CD16/32), mAb M1/69 (CD24) and mAb 11-26c.2a (IgD); 2) from eBioscience (San Diego, CA): mAb B3B4 (CD23), mAb 12.14 (Nur77), and mAb P3.6.2.8.1 (IgG1 isotype); 3) from Pierce (Rockland, IL): Dylight 649 conjugated streptavidin; 4) in-house: mAb 493 (CD93 [55]), mAb 187.1 (Igκ [56]), and mAb b7-6 (IgM [33]). Antibodies were either directly coupled to a fluorochrome or were resolved with a fluorochrome-coupled streptavidin.

Fab Anti-μ Generation/Conjugation

Fab fragments of mAb b7-6 were generated according to the procedure of Udhayakumar et al. [57]. Fab b7-6 was chemically coupled to EZ-Link NHS-LC-Biotin, DyLight 488, or DyLight 649 (Pierce) according to the manufacturer's instructions.

Antibody Immunoassays

Europium (Eu³⁺)-based fluoroimmunometric assays were described previously [58, 59]. For quantification of total IgG1κ produced in vivo, 96-well europium plates (Greiner Bio-One, Frickenhausen, Germany) were coated overnight at 4°C with Rat anti-mouse kappa (RαMκ) (mAb 187.1, generated in-house) followed by treatment with blocking buffer (2% BSA, 1% gelatin in PBS) for 2 h at 37°C. Serum samples were diluted in blocking buffer and incubated at 37°C for 2 h or overnight at 4°C. IgG1κ antibodies were detected using biotin-Rat anti-mouse IgG1 (mAb RMG1-1, Biolegend), and quantities were interpolated based on standard binding curves generated using IgG1κ (MOPC 31C) myeloma Ig (Sigma-Aldrich, St. Louis, MO). For total IgM, IgMκ or IgG3κ produced in vitro, europium plates were coated overnight at 4°C with either GαMμ (Sigma), mAb b7-6 (generated in-house), or Goat anti-mouse IgG3 (Bethyl, Montgomery, TX) followed by treatment with blocking buffer for 2 h at 37°C. Culture supernatant was diluted in blocking buffer and incubated at 37°C for 2 h or overnight at 4°C. Total IgM antibodies were detected using biotin-Donkey anti-mouse IgM (Jackson Immunoresearch, West Grove, PA), IgMκ and IgG3κ antibodies detected using biotin-RαMκ (in-house), and quantities were interpolated based on standard binding curves generated using IgMκ (TEPC 183) and IgG3κ (FLOPC 21) myeloma Igs (Sigma-Aldrich). Calf chromatin, ssDNA, and BSA binding assays were performed as previously described [58] and bound antibodies were detected using biotin-RαMκ or biotin-Donkey anti-mouse IgM

Flow Cytometry

Single-cell suspensions from spleen were prepared by mechanical dissociation through 40 μm cell strainers (BD, Franklin Lakes, NJ). Cells were depleted of erythrocytes (RBC lysis buffer; Sigma-Aldrich) and washed twice with PBS containing 2% FCS + 0.1% NaN₃ (staining buffer). The cells were surface stained with a combination of conjugated antibodies diluted in staining buffer following a standard protocol with 30 min incubations on ice in the presence of anti-CD16/32 (mAb 93) to block Fc receptors. Following surface staining, cells were fixed for 15 min in the dark at room temperature in staining buffer containing paraformaldehyde at 1.6%. Intracellular staining for IgM was performed using Perm/Wash Buffer™ (BD Biosciences) according to the manufacturer's instructions. Flow cytometric acquisitions were performed on the LSRII™ (BD Biosciences) flow cytometer. Data were analyzed using FlowJo 9.2 (Tree Star, Ashland, OR)

Cell Sorting

Single-cell suspensions were prepared as described above except cells were stained in phosphate buffered saline PBS containing 2% FCS. After surface staining, cells were resuspended to ~20-25 ×10⁶ cells/ml in RPMI 1640 supplemented with Na⁺ pyruvate, L-glutamine, antibiotics, and 5% FCS. Cells were separated using the SY3200 (Sony Biotechnology Inc. Champaign, IL) based on B220, CD23, and either CD24 or CD93 surface expression. Mature follicular (B220^pos CD23^hi CD24^int or B220^pos CD23^hi CD93^neg) B cells were segregated into 2 or 3 populations based on sIgM expression. FO IgM^lo and IgM^hi cells were defined as the 15-20% brightest or dullest staining cells unless otherwise noted in the figure legends. IgM^int cells typically comprised 25-30% of the total FO B cell population. Marginal zone B cells were identified as B220^pos CD23^neg CD93^neg cells expressing high levels of CD1d and IgM. Cells were sorted into tubes containing RPMI 1640 supplemented media containing 10% FCS (hereafter referred to as complete media).

Surface IgM Internalization Assay

Splenocytes were stained (5 × 10⁷ cells/ml) with anti-CD16/32 (clone 93) and FITC-conjugated Fab goat anti-mouse IgM (GαMμ, Jackson Immunoresearch) on ice for 10 min. Biotin-conjugated Fab rat anti-mouse IgM (Bio-Fab b7-6, generated in-house) was then added and the cells stained for an additional 5 min on ice. The splenocytes were washed with ice-cold complete media and 100 μl (2.5 × 10⁶ cells) aliquots transferred into 96-well plates. Following incubation at 37°C for various time-points (15, 30, or 60 min), cells were fixed with 1.6% paraformaldehyde and the remaining biotin labeled sIgM revealed with PE-conjugated strepavidin. The percentage of remaining biotin-labeled sIgM was calculated according to the formula [(PE fluorescence at T_n ÷ PE fluorescence at T₀) × 100]. PE fluorescence of Fab-bio b7-6 labeled cells not stained with SA-PE was used for background subtraction.

Ca²⁺ Assays

Splenocytes (15-20 × 10⁶/ml) were loaded with 5 μM Indo-1 AM (Molecular Probes, Eugene, OR) in Hyclone IMDM (Logan, UT) containing 2.5% FCS for 30 min in a 37°C water bath. Cells were washed twice in PBS + 2% FCS and surface stained for 10 min at room temperature with conjugated mAb. Cells were resuspended in IMDM, warmed to 37°C, and stimulated with either F(ab')₂ GαMμ (Jackson Immunoresearch), GαMκ (Bethyl), or GαMδ antiserum (eBiosciences). Alternatively, Indo-1 loaded splenocytes were surface stained with Bio-Fab b7-6 for 5 min at room temperature prior to the addition of DyLight 488 conjugated Fab b7-6 (generated in-house) and remaining Abs. The cells were stained for an additional 10 min at room temperature prior to being washed, resuspended in IMDM, and warmed to 37°C. Receptor aggregation in these assays was induced following addition of DyLight 649 conjugated strepavidin (Pierce). For all assays, baseline Ca²⁺ was established (35 seconds) prior to the addition of the stimulatory agonist, and cellular events were acquired for an additional 5 min.

Adoptive Transfer and Immunizations

A/J Igκ-deficient mice were injected i.v. with FACS purified mature FO IgM^lo or IgM^hi B cells from A/J donors. Recipient mice were subsequently immunized i.p. with 50 μg of GαMκ (Bethyl) or 50 μg recombinant Protein L (Pierce) precipitated in alum as previously described [37].

In Vitro Culture of FACS Purified FO B cells

Following FACS purification, cells were washed once in complete media and viable cells were counted using Trypan blue dye (Life Technologies, Carlsbad, CA) exclusion. Cells were plated at 5 × 10⁴/ml in 2 ml of complete media containing 25 μg/ml of LPS (E.coli 0111:B4) (InvivoGen, San Diego, CA) and cultured for 7 days at 37°C in 5% CO₂.

Hybridoma Generation/Screening

FACS purified mature FO IgM^lo and IgM^hi B cells were cultured in complete media containing 25 μg/ml LPS for 3 days prior to fusion with the Sp2/0 mIL-6 partner [60]. Hybridomas were selected with hypoxanthine and azaserine (Sigma), and wells with viable cells from plates averaging between 0.12 and 0.005 clones/well from a Poisson analysis were expanded and assessed for total IgM production and reactivity against ssDNA, calf chromatin and BSA. The results presented are from a single fusion in which the FO IgM^lo and IgM^hi derived hybridomas were screened simultaneously.

Quantitative RT-PCR

RNA was isolated from FACS sorted mature FO IgM^lo and IgM^hi B cells using the PicoPure RNA Isolation Kit (Life Technologies) and RNase-Free DNase I Kit (Qiagen, Valencia, CA) according to the manufacturer's instructions. Equal volumes of RNA were used for cDNA generation using the SuperScript III First-Strand Synthesis System (Life Technologies). Real-time PCR was performed using the ABI 7300 and SYBR Green Master Mix (Life Technologies). HPRT primers were described previously [61] and Nur77 primers were designed with MacVector (Cary, NC) (F1 5’-GCTTCGTGTCAGCACTATGGGG-3’ + R1 5’-TCCACAGGGCAATCCTTGTTTG-3’ or F2 5’-GTTGATGTTCCCGCCTTTGC-3’ + R2 5’-TGAGGCAAAAGATGCGCTGC-3’). Nur77 transcript abundance was relative to HPRT expression and normalized to that of the FO IgM^hi group.

In vitro Nur77 Protein Induction Assay

Splenocytes were resuspended at 2.5 × 10⁷/ml in complete media, and 100 μl aliquots were transferred into 96-well plates. Following centrifugation, the cell pellets were resuspended with ice-cold complete media containing 2-fold diluted concentrations of F(ab')₂ GαMμ ranging from 50 μg/ml to 0.78 μg/ml or media alone. Cells were incubated for 10 min on ice, washed with ice-cold complete media to remove excess stimulatory GαMμ fragments and then cultured for 4 hr in complete media. Intracellular staining for Nur77 was then performed using eBioscience's Foxp3/Transcription Factor Staining Buffer Set™ according to the manufacturer's instructions.

Statistics

Statistical analyses were performed using PRISM 5.0.