Self-tolerance by clonal anergy of B cells is marked by an increase in IgD and decrease in IgM antigen receptor surface expression, yet the function of IgD on anergic cells is obscure. Here we define the RNA landscape of the in vivo anergy response, comprising 220 induced sequences including a core set of 97. Failure to co-express IgD with IgM decreases overall expression of receptors for self-antigen, but paradoxically increases the core anergy response, exemplified by increased Sdc1 encoding the cell surface marker syndecan-1. IgD expressed on its own is nevertheless competent to induce calcium signalling and the core anergy mRNA response. Syndecan-1 induction correlates with reduction of surface IgM and is exaggerated without surface IgD in many transitional and mature B cells. These results show that IgD attenuates the response to self-antigen in anergic cells and promotes their accumulation. In this way, IgD minimizes tolerance-induced holes in the pre-immune antibody repertoire.
Clonal anergy is an enigmatic mechanism for actively acquired tolerance, a process in which self-reactive cells remain in the lymphocyte repertoire of secondary lymphoid tissues but are deficient in generation of effector progeny1,2. Anergy is best characterized in mouse and human peripheral B cells expressing high cell surface levels of IgD and low levels of IgM B cell receptors (BCR), which account for 10–50% of the mature pre-immune B cell repertoire, depending on an arbitrary cut-off for low surface IgM (refs 3, 4, 5, 6, 7). Retaining anergic B cells bearing self-binding antibodies in the secondary lymphoid organs presents a risk of autoimmunity8, as the diminished proliferation and antibody secretion that characterizes anergic B cells is potentially reversible2,9. Pathological proliferation of B cells that would normally be anergic also leads to common adult malignancies, exemplified by a large subgroup of chronic lymphocytic leukaemia cases10, and by the over-representation of B cells using self-reactive VH4-34 heavy chains, which are normally anergic, within the poor prognosis subset of diffuse large B cell lymphoma11. By contrast, physiological proliferation of B cells that were initially anergic has been shown to occur when these cells bind a foreign antigen recognized by T-follicular helper cells and produce germinal centre (GC) progeny and IgG antibodies that have been hypermutated away from self-reactivity12,13. The molecular nature of B cell anergy that precedes any reactivation into proliferation nevertheless remains unresolved, in particular whether or not anergy is explained by binding antigen primarily through IgD antigen receptors.
Anergic cells selectively inhibit trafficking of nascent IgM but not IgD through the trans-Golgi network to the cell surface14. A similar change in IgM trafficking occurs in malignant B cells in chronic lymphocytic leukaemia15 and during normal maturation of B cells in the spleen16. This altered trafficking may be explained by the IgD juxtamembrane and transmembrane segments—one of the few evolutionarily conserved domains of IgD (ref. 17)—associating preferentially with the CD79αβ subunits needed for IgM and IgD trafficking and signalling on the cell surface18,19,20,21. Immature B cells begin by expressing only IgM, but IgD co-expression progressively increases as they become transitional and mature B cells in the spleen due to increased expression of Zfp318 (ref. 22), which facilitates alternative mRNA splicing of the heavy chain variable (VDJH) exon to either IgM or IgD heavy chain constant (C)-region exons. This arrangement is evolutionarily preserved in most species of fish, amphibians, reptiles, birds and mammals17,23, yet mice lacking IgD have normal B cell development and only slightly delayed antibody responses24,25. Likewise, comparison of mice that express only IgM or only IgD reveals no discernable difference in the capacity of these alternative receptors to promote B cell development, tolerance, activation or antibody secretion in vivo26,27.
Conflicting results from tissue culture experiments support the idea that IgM and IgD signal differently. Early experiments suggested that IgM signals drive apoptosis whereas IgD signals induce proliferation, so that immature IgM-only B cells would pass through a window of obligate tolerance susceptibility28,29. However, these differences might reflect the use of different antibody ligands to engage IgM and IgD, rather than an intrinsic difference in the receptor isotypes. In transfected B lymphoma cells, a single haptenated antigen signalled through IgD more strongly and durably than through IgM30. By contrast, hen egg lysozyme (HEL) antigen signalling induced CD86 identically through IgD or IgM on splenic B cells expressing one or other isotype21. However, when the same IgMHEL and IgDHEL BCRs were expressed separately in a pro-B cell line with a partially crippled BLNK (SLP65) intracellular signalling adaptor, soluble monovalent HEL antigen signalled an increase in intracellular calcium when it bound IgM but induced no calcium signalling when it bound IgD, whereas multivalent HEL-antigen signalled through both isotypes31. Extrapolating the findings from the BLNK mutant cell line, it was concluded that the predominant expression of IgD on anergic cells prevents any response to monovalent self-antigens, ascribing the in vivo state of anergy to the change in BCR isotype31.
Here we directly address the role of IgD on anergic B cells in vivo with three complementary approaches, by analysing anergic B cells in mice either lacking IgD, with a novel point mutation in IgD, or inactivation of the IgD-splicing factor Zfp318. We find little intrinsic signalling difference between the two receptor isotypes, but IgD has an essential function when co-expressed with IgM, attenuating the in vivo response to self and promoting accumulation of mature anergic B cells to increase their availability to encounter foreign antigens and potentially form GCs.
Calcium signalling by IgD and IgM
We first tested the proposal that IgD is unable to trigger an acute elevation of intracellular calcium in response to monomeric antigens like soluble HEL (ref. 31), potentially explaining the unresponsive state of anergic B cells. The intracellular calcium increase elicited by monomeric HEL was directly compared in splenic B cells from MM4 and DD6 transgenic mice, which respectively express the IgMHEL or IgDHEL antigen receptors studied in ref. 31 comprising identical variable regions and different constant regions. In contrast to the findings made in BLNK-mutant pro-B cells31, when tested here in mature B cells with normal BLNK both isotypes signalled an acute and sustained calcium response (Fig. 1), although the initial rise in calcium was slightly decreased in cells with IgDHEL antigen receptors.
IgD enhances formation of GC cells from anergic B cells
Although IgD expression ceases on activated and GC B cells, we investigated if IgD had any role in the reactivation of anergic B cells using an animal model where anergic B cells have been shown to be reactivated by foreign antigen to form GC progeny, hypermutate their V-regions and rapidly evolve antibodies with lower affinity for self-antigen12. HEL-specific anergic B cells were obtained from MD4:ML5 animals that co-express IgDHEL with IgMHEL, or from MM4:ML5 animals where the heavy chain transgene lacks the IgD exons and encodes only IgMHEL. The mice also carried the ML5 transgene producing circulating HEL at sufficient concentration to induce B cell anergy1,26. Naïve (non-anergic) control B cells with or without IgD were obtained from MD4 or MM4 single-transgenic mice. After labelling with CFSE dye, equal numbers of splenic B cells bearing the CD45.1 congenic marker were adoptively transferred into C57BL/6 (CD45.2) mice. The recipients were immunized with HEL coupled to sheep red blood cells, the latter eliciting T follicular helper cells to form GCs (HEL-SRBC). At the time of transfer, IgD+ anergic B cells had three times more HEL-specific antigen receptors on their surface compared with anergic IgM-only B cells (red versus orange histograms, Fig. 2a).
Half of the recipients were analysed after 2.5 days. The progeny of naïve control B cells had divided and diluted CFSE comparably with or without IgD (Fig. 2b, unshaded blue versus green histograms), 90% diluting by more than 4 cell divisions (Fig. 2c). By comparison, 75% of IgD+ and 55% of IgD-deficient anergic cell progeny had divided >four times (red and orange histograms, Fig. 2b,c). The total number of progeny from IgD-deficient anergic cells was decreased to only 20% of naïve controls by day 2.5 (Fig. 2d), whereas progeny of IgD+ anergic cells accumulated to the same number as controls.
The remainder of the recipients were analysed after 5 days. Fas+GL7+GC B cells in the spleen averaged 2.6% of B cells and were not significantly different between the groups of recipients, and CD45.1+ HEL-binding donor GC B cells accounted for 9% of GC cells averaged across the groups. However, anergic IgD+ B cells formed more GC progeny than naïve IgD+ counterparts, as observed previously12, but the opposite was observed for anergic B cells without IgD, which formed many fewer GC progeny than their naïve counterparts (orange bar, Fig. 2e). By contrast, IgD-deficiency did not alter the number of GC progeny formed from naïve B cells on day 5 (green bar).
IgD promotes accumulation of mature anergic B cells
One explanation for the results above was that IgD assists short-lived anergic B cells to survive in the spleen long enough to respond to foreign antigen and T cell help. Consistent with this hypothesis, mature CD93−CD23+ anergic B cells accumulated to much lower frequency in the spleen of IgD-deficient MM4:ML5 double transgenic animals compared with MD4:ML5 animals (Fig. 2f,g). To exclude differences in transgene integration site between the MD4 and MM4 animals, we also prevented IgD co-expression in MD4:ML5 transgenic mice by breeding with a Zfp318 null allele22. Accumulation of anergic mature B cells was diminished in Zfp318−/− MD4:ML5 transgenic mice where the B cells only expressed IgMHEL (Fig. 2h).
IgD promotes accumulation independently of effects on IgM
Diminished accumulation of mature anergic B cells in the experiments above could reflect either a difference between surface IgM and surface IgD in the way they trigger responses to self-antigen, or a loss of the attenuating effect of IgD on surface IgM levels, or both. In an ENU mutagenesis screen, we discovered a surface IgD-deficient C57BL/6 mouse strain with unaltered levels of surface IgM. The dimit strain (C57BL/6JAnu-Ighddmit/Apb) has acquired a single T>A point mutation resulting in an Ile81Lys substitution within the Cδ1 exon (Fig. 3a). Ile81 flanks Cys79 that forms the intradomain disulphide bond and corresponds to an obligatory hydrophobic residue at position 5 within β-sheet F of all Ig CH1 domains, part of a folding nucleus of inward-pointing hydrophobic residues in β-sheets B, C, E and F32,33. Ile81 mutation to charged, hydrophilic lysine would be expected to prevent folding of the IgD CH1 domain into the conformation needed to pair with immunoglobulin light chains.
Surface IgD progressively increased on wild-type spleen B cells between immature (CD93+CD23− transitional 1, T1), semi-mature transitional cells (CD93+CD23+ T2 / T3 subset), and mature follicular cells (CD93− CD23+)34, but was decreased to less than 5% of normal levels in Ighddmit/dmit homozygotes (Fig. 3b,c). The mean surface IgM and the broad cell to cell variance was nevertheless identical in Ighddmit/dmit homozygotes and wild-type controls (Fig. 3c). This contrasts with the aberrantly high cell surface IgM on B cells in IgD knockout or Zfp318 knockout mice22,24,25, where IgD no longer competes with IgM for assembly with CD79αβ (refs 18, 20). The loss of IgD and normal surface IgM caused an overall 65% decrease in mean BCR numbers on mature B cells (Fig. 3c).
The IgD point mutation enabled testing if loss of surface IgD affected mature B cell accumulation independently from IgD’s inhibitory effect on surface IgM levels. In heterozygous Ighddmit/+mice, allelic exclusion resulted in 50% of the immature T1, T2 and T3 B cells expressing the wildtype Ighd allele and 50% expressing the mutant allele, the latter distinguished as IgDlow cells by flow cytometry (Fig. 3b). By contrast, B cells expressing the surface IgD-deficient allele contributed poorly to the mature B cell repertoire and accounted for only 24% of cells (Fig. 3d). A similar requirement for surface IgD for the accumulation of mature B cells relative to immature B cells was observed in the bone marrow, blood and spleen of mixed bone marrow chimeras reconstituted with a mixture of CD45.1 and CD45.2 bone marrow (Fig. 3e). The 52% decrease in mature B cells with the Ighddmit mutation appears less than the decrease in mature anergic B cells in MD4:ML5 transgenic mice with the Zfp318 mutation or the decrease in MM4:ML5 anergic cells lacking IgD. This may reflect the fact that the latter two lose two functions of IgD: surface IgD display and IgD attenuation of surface IgM expression, whereas the Ighddmit mutation selectively disrupts surface IgD.
IgD attenuates anergic B cell mRNA response to self-antigen
We next performed a global RNA analysis of MD4 and MM4 naïve and anergic mature B cells to test for any evidence that surface IgD and IgM differ in their capacity to elicit responses to self-antigen. Two variables differ between MD4:ML5 and MM4:ML5 anergic B cells: (1) surface IgM is lower on MD4:ML5 cells because IgD competes with IgM for CD79 assembly and surface expression; (2) the total cell surface pool of antigen-binding receptors is higher on MD4:ML5 cells, because IgD is not downregulated on anergic cells. The null hypothesis, that there is no difference in the response to self elicited by surface IgM and IgD, would therefore predict that there should be no difference in the induction of anergy RNAs or there should be less induction of some or all anergy RNAs in MM4:ML5 cells because of their 65% fewer receptors for self-antigen. The opposite result—a higher response to self by some or all of the anergy-induced RNAs in MM4:ML5 cells without IgD—would demonstrate that IgM is better at eliciting that part of the response than IgD, and since IgD decreases surface IgM that could further exaggerate any difference in the activity of the surface receptors themselves. Much higher induction of all anergy genes in MM4:ML5 cells, and little or no response to self-antigen in MD4:ML5 cells, would be predicted based on the hypothesis that IgD cannot signal to monomeric HEL (ref. 31).
Anergic and naïve mature CD93−CD23+ HEL-binding splenic B cells were sorted from MD4:ML5 or MM4:ML5 mice and from MD4 or MM4 single-transgenic controls. As an additional comparison, mature HEL-binding B cells that expressed only IgDHEL were sorted from DD6:ML5 or DD6 transgenic mice, which carry the same transgene as MD4 but missing the IgM exons26. RNA from independent donors was measured on Agilent microarrays carrying 59,305 different oligonucleotide probes for protein-coding and non-coding RNAs (Supplementary Data 1). Analysis of constitutively expressed mRNAs in B cells, exemplified by Cd19, Cd79b, B2m and Gapdh, showed comparable expression regardless of BCR isotype or exposure to self-antigen (Fig. 4a and Supplementary Fig. 11a).
In B cells that lose surface Ig completely, tonic BCR signals through PI3-kinase are lost as evidenced by an increase in mRNAs for four PI3-kinase repressed genes, Rag1, Cdkn1b, Bcl2l11 and Aicda (ref. 35). Hence it was conceivable that the 65% decrease in surface BCRs on anergic MM4:ML5 cells compared with anergic MD4:ML5 cells or naïve MM4 cells (Fig. 2a) might also increase these mRNAs, but there was no evidence that this was the case (Fig. 4a and Supplementary Fig. 11b).
We next analysed 18 mRNAs previously identified as being consistently induced in anergic MD4:ML5 B cells using first-generation Affymetrix arrays36. Twelve of these mRNAs (Egr1, Egr2, Gfi1, Nab2, Cd83, Tgif1, Lck, Cd72, Nrgn, Ccnd2, Crisp3 and Pcp4) were consistently increased in anergic B cells relative to naïve B cells, regardless of whether the B cells co-expressed IgM and IgD or only expressed one isotype (Fig. 4a and Supplementary Fig. 11c). Despite 62% fewer surface antigen receptors on IgD-deficient anergic cells, Egr1, Egr2, Gfi1 and Nab2 were paradoxically induced to 1.25–2.3 fold higher levels in IgD-deficient MM4:ML5 anergic cells compared with IgD+ MD4:ML5 anergic cells. The Egr2 paralogue, Egr3, and two other genes shown previously to be induced in anergic B cells, Nr4a1 (NUR77(ref. 7)) and Sdc1 (CD138 (refs 37, 38)), were also induced to higher levels in MM4:ML5 IgD-deficient anergic B cells (Fig. 4a and Supplementary Fig. 11d).
To obtain a comprehensive transcriptional landscape of B cell anergy with and without IgD, we analysed all 59,305 oligonucleotide probes on the arrays. Using limma39, we focussed on 33,653 probes that were positive for expression in anergic B cells (Supplementary Data 2). Of these, 97 probes had strong evidence of at least two-fold increased expression in MD4:ML5 anergic B cells compared with MD4 naïve counterparts, including 13 previously described anergy genes (Table 1 and Fig. 4b). In this set of 97 probes, 91 and 78% had moderate or strong evidence of induction in MM4:ML5 and DD6:ML5 anergic B cells, respectively, compared with only 1 and 3% of all 33,653 expressed probes (Fig. 4c and Table 1). An additional 123 probes had moderate evidence of induction in MD4:ML5, and the majority were also induced in anergic B cells with only IgM or only IgD (Table 1). Thus, when chronically stimulated by monomeric self-HEL antigen in vivo, IgD expressed with IgM or on its own induces the great majority of the transcriptional response, in contrast to recent conclusions from a BLNK-deficient pro-B cell line31.
Of 97 and 123 probes with strong and moderate evidence, respectively, of induction in anergic cells co-expressing IgD and IgM, 34 and 29% had moderate evidence of ≥125% increased expression in MM4:ML5 anergic cells lacking IgD compared with MD4:ML5 anergic cells with IgD (Fig. 4d, Table 1). By contrast, only 4% of the 33,653 expressed probes were increased ≥125% in MM4:ML5 samples. The hyper-induced gene set in IgD-deficient cells included mRNAs encoding transcription regulators (Egr1, Egr2, Egr3, Gfi1, Lef1, Ahr, Myb, Myc, Nr4a1, Sox4, Hmgn3, Apex1), apoptosis inducers (Casp4), and cell surface proteins (Sdc1, Il18r1, Dlk1). Reciprocally, there was no evidence for diminution of a subset of the anergy response without IgD: only 5% of anergy induced genes were decreased without IgD, which is comparable to 4% of all expressed probes on the array showing similar evidence for decreased expression (Table 1). Thus, IgD was not required for induction of the anergy mRNA program but, paradoxically, its co-expression increased the number of self-antigen binding receptors nearly 3-fold (Fig. 2a) yet attenuated one third of the mRNA response to self-antigen.
Hyper-induction of EGR1, EGR2 and SDC1 proteins
We validated the observations above by antibody staining and flow cytometry, focussing on SDC1 because it is a cell surface protein induced in B cells with chronically elevated intracellular calcium38, and on the transcription factors EGR1 and EGR2, which are induced in anergic B cells by the ERK and calcium-calcineurin-NFAT signalling pathways, respectively4,36,40. Anergic B cells from MD4:ML5 animals had significantly increased EGR1 and EGR2 intracellular staining compared with naïve MD4 B cells, which served as a negative control B cell population because they have very low Egr1 and Egr2 mRNA, (Fig. 5a,b). Consistent with the microarray findings, there was exaggerated induction of both proteins in IgD-deficient anergic cells from MM4:ML5 animals (Fig. 5b, orange versus red columns).
Cell surface staining confirmed that SDC1 was specifically induced on immature, transitional and mature anergic B cells from MD4:ML5 animals compared with naïve MD4 B cells (Fig. 5c,d, red versus blue columns), and was induced to higher levels on IgD-deficient anergic B cells from MM4:ML5 mice (Fig. 5d, orange columns). A similar level of SDC1 was present on IgMlow splenic B cells in C57BL/6 mice with a diverse antibody repertoire (Fig. 5c), consistent with evidence that surface IgM downregulation identifies anergic self-reactive B cells in the normal repertoire3,4,5,6,7.
IgD attenuates SDC1 induction independently of effect on IgM
To extend these observations to the normal B cell repertoire, we focussed on SDC1 because cell surface antibody staining could readily detect its induction on individual B cells above background autofluorescence. The increase in SDC1 and decrease in surface IgM varied from cell to cell in a highly correlated way in the immature T1, transitional T2/3, and mature follicular B cell subsets (Fig. 6a). While surface IgM was unaltered in B cells from homozygous Ighddmit mice, SDC1 was induced to higher mean levels on the T2/3 and mature B cell subsets with little surface IgD compared with wild-type controls with normal surface IgD (Fig. 6b). When mature B cells were separated into four quartiles based on the amount of surface IgM, mean SDC1 was higher on surface IgD-deficient cells in each quartile (Fig. 6c). SDC1 levels were also analysed in heterozygous Ighddmit/+ mice, where allelic exclusion enabled comparison within individual animals between the 50% of transitional IgMlow T3 B cells expressing the mutant Ighd allele and the 50% with wild-type Ighd. SDC1 was induced to higher levels on transitional B cells lacking surface IgD (Fig. 6d). Thus surface IgD acts cell-autonomously to attenuate induction of the SDC1 anergy marker on many semi-mature and mature B cells in the normal repertoire, independently of IgD’s inhibitory effect on surface IgM levels.
The experiments above advance understanding of self-tolerance and the role of IgD in six ways. First, they define 220 induced RNA species that provide a comprehensive definition of the anergic state in vivo, enabling future studies of how anergy is distorted by susceptibility genes in autoimmunity or by mutations in chronic lymphocytic leukaemia10 or lymphoma11. Second, the global RNA analysis reveals that IgD differs from IgM by attenuating a large subset of the anergy response to self-antigen. Third, the results of the RNA analysis and of calcium signalling did not support previous conclusions that IgD signals for B cell activation while IgM signals for tolerance28,29, that IgD signals in a stronger and more sustained way than IgM (ref. 30), nor that IgD is unable to signal to monovalent ligands31. Fourth, flow cytometric staining for SDC1 reinforces the proposal that IgM downregulation on many B cells in the normal repertoire reflects an active anergic response to self3 and yields a method to analyse and sort anergic B cells without engaging surface IgM with antibodies3,4 nor requiring crosses to a transgenic Nr4a1:GFP reporter strain7. Fifth, we discover a new mouse mutant that eliminates surface IgD but retains the inhibitory effect of IgD on surface IgM display, and show that the selective loss of surface IgD results in exaggerated induction of SDC1 on IgMlow anergic B cells. Finally, the experiments reveal that IgD promotes accumulation of anergic B cells as mature follicular B cells, thus minimizing tolerance-induced holes in the pre-immune repertoire of antibodies available in secondary lymphoid tissues.
Loss of surface IgD on transitional IgMlow T3 B cells decreases the number of surface BCRs and would be expected to diminish their response to self-antigen if IgM and IgD were functionally equivalent, but paradoxically it increased the anergy response reported by SDC1. Plasma cells with high levels of SDC1 are absent in Prdm1 (Blimp-1) knockout mice but the subset of SDC1low mature B cells analysed here are not plasma cells because they form independently of Blimp-1 (ref. 41) and have high B220 and CD19 that are both downregulated on plasma cells. SDC1 is also induced on self-reactive B cells binding the Smith ribonucleoprotein37. While MD4 B cells did not express SDC1 in the absence of self-HEL (ref. 37), SDC1 was spontaneously induced on MD4 naïve B cells when their intracellular calcium was chronically elevated by genetic deficiency of inositol 1,4,5-trisphosphate 3-kinase b (Itpkb)38. Repeated cycles of self-antigen binding in anergic cells raise cytosolic calcium to 100–400 nM in an oscillatory manner, and calcium reverts to the 50 nM baseline of naïve B cells when anergic cells are transplanted into recipients lacking the self-antigen40 or when the self-antigen is outcompeted by a non-stimulatory ligand for the BCR (ref. 42). Consistent with the range of SDC1 induction between B cells in non-transgenic mice and its correlation with surface IgM downregulation, intracellular calcium is elevated to varying extents among polyclonal B cells in proportion to their expression of an Nr4a1:GFP reporter transgene7. Thus SDC1 serves as a convenient reporter for the varying degrees of self-reactivity among B cells in the pre-immune repertoire.
The attenuated self-antigen response when IgD is coexpressed with IgM in vivo helps clarify earlier conflicting findings. A small decrease in the proportion of mature B cells was observed comparing a pair of transgenic strains expressing a polyreactive BCR as IgM-only or as IgM and IgD (ref. 43). However, another pair of transgenic mice with a slightly different H-chain V-region displayed more mature B cells when bearing IgM-only than with IgM+IgD, although in these animals transitional IgM-only B cells appeared more susceptible to deletion when exposed to a tolerizing injection of human IgG antigen44. Mature B cells were depleted by a tolerizing injection of TNP-dextran in IgM-only IgSP6 transgenic mice, but not when these mice were crossed with an IgD-only IgSP6 transgenic strain so that IgD was co-expressed with IgM (ref. 45). From these results it was hypothesized that IgD attenuates tolerogenic signalling by IgM (ref. 46), consistent with the findings here. Interpretation of these effects was confounded by the absolute requirement for endogenous light chain specificities for IgSP6 B cells to mature, and by rescue of their maturation by injection of TNP-ficoll47.
A key question is how much the attenuation of the anergy response by IgD demonstrated here reflects an intrinsic decrease in the ability of IgD to signal in response to self-antigens and how much reflects IgD inhibition of surface IgM for assembly with CD79αβ and trafficking to the cell surface18,19,20,21. The experiments with the IgD dmit point mutant in Figs 3 and 6 show that exaggerated response to self-antigen still occurs when IgD attenuates surface IgM normally but IgD is no longer expressed on the surface of anergic cells. Thus IgD attenuates the response to self-antigen in two ways: IgD is less able to elicit a subset of the response than IgM on the cell surface, and IgD decreases the display of surface IgM.
A simple explanation for the decreased signalling by IgD on the surface has recently been proposed based on findings that IgD was unable to signal an acute calcium response by monovalent but not multivalent HEL antigen in a pro-B cell line where the BLNK cytoplasmic signalling adaptor was expressed as an oestrogen receptor-chimera that partially crippled BLNK function by binding to heat shock proteins31. Four lines of evidence here indicate this black and white difference between IgD and IgM does not extrapolate to primary B cells with normal BLNK. First, in primary B cells IgD was capable of inducing an acute calcium response to monomeric HEL in vitro (Fig. 1). The initial calcium peak was consistently slightly lower in IgDHEL B cells than in IgMHEL B cells, and conceivably this subtle difference may be exaggerated in cells with defective BLNK. Secondly, IgD was competent to induce 78% of the anergy-induced RNA species in response to monomeric HEL in vivo, including calcium-induced genes like Egr2. Third, anergic IgMlow B cells in mice with a normal antibody repertoire make antibodies binding to DNA and other polymeric self-antigens4,6,7, making it unlikely that IgD attenuation of SDC1 on a large fraction of these cells reflects recognition of only monovalent self-antigens. Instead, IgD appears to acquire the capacity to attenuate the response when it is co-expressed with IgM. This might reflect competition between the two receptor isotypes for particular signalling molecules or membrane domains18,19,21,48,49.
The need for IgD to promote anergic B cell accumulation in the spleen and blood may represent a mechanism to minimize holes in the antibody repertoire created by self-tolerance. Deletion of self-reactive B cells in the bone marrow, either by apoptosis or receptor editing, removes antibody specificities from the repertoire before they can be tested for binding to foreign antigens. By contrast, antibodies on anergic B cells can be hypermutated in germinal centres and selected for variants that have lost binding to self but retain binding to foreign microbes13.
All mice were on a C57BL/6 background, male, mean age 157 days (range 84–287 d), maintained in specific pathogen free conditions under a protocol approved and monitored by the ANU Animal Experimentation and Ethics Committee. MD4:ML5, MM4:ML5 and DD6:ML5 transgenic mice were as described1,3,26. The Ighddmit strain was isolated by screening ENU-mutagenised C57BL/6 mice for low surface IgD on blood B cells. The ENU-induced mutation was identified by exome sequencing using the SureSelect Mouse Exome kit (G7550A-001: Agilent, CA) and 100 bp paired-end sequencing on an Illumina HySeq2500 following the manufacturers’ protocols. Allele-specific genotyping in C57BL/6 backcrossed progeny was performed using Amplifluor assays (Chemicon, Temecula, CA) and primers designed using the Assay architect online tool (http://apps.serologicals.com/AAA/mainmenu.aspx; sequences available on request). Zfp318 knockout mice were as described22. For ethical purposes, each experiment used the minimum number of animals per group needed to provide statistical power to detect a two-fold difference between groups, and each experiment was performed multiple times. Animals from the same parents were randomly assigned to experimental groups based on their genotype, but were not blinded to the investigator. No animals were excluded from analysis.
Flow cytometry and sorting
Preparation of spleen cells and staining for immature and mature B cells as described previously12,22, employing Efluor 780 fixable viability dye (eBioscience) and the following antibodies: B220, RA3-6B2 coupled to Alexafluor700 or APC-Cy7 (Becton Dickinson, 1/200); Fas, Jo-2 coupled to PE (Becton Dickinson), GL7 coupled to FITC (Becton Dickinson); CD38, clone 90 coupled to PE (Biolegend), CD45.1, A20 coupled to Pacific Blue or Alexfluor700 (Biolegend); CD45.2, 104 coupled to PerCP (Pharmingen); CD93, AA4.1 coupled to FITC, APC or biotin (eBioscience, 1/100) or PE (Biolegend); CD23, B3B4 coupled to PE (BD), PE-Cy7 (eBioscience, 1/200), or Pacific Blue (Biolegend); IgD, 11-26c coupled to FITC or biotin (eBioscience, 1/100), PE (Southern Biotech), PerCP-Cy5.5 or BV510 (Biolegend, 1/200); IgM, II/41 coupled to APC or FITC (BD) or PE-Cy7 (eBioscience, 1/100); Streptavidin coupled to PE-Cy7 or Qdot 605 (eBioscience). Each antibody conjugate and lot was titrated and used at dilutions yielding saturated staining.
In Ig-transgenic mice, edited B cells were eliminated by gating out surface IgKappahi HELlow B cells after staining with HEL followed by Hy9-APC (1/800) and then anti-kappa monoclonal 187.1 conjugated to PerCP-Cy5.5 (Becton Dickinson, 1/100). Staining for SDC1 employed biotinylated monoclonal antibody 281–2 followed by streptavidin-brilliant violet 605 (Biolegend). EGR1 and EGR2 staining was performed on fixed and permeabilised cells using the Foxp3 staining kit (eBioscience), using rabbit monoclonal antibody T.126.1 (Thermo Scientific) followed by anti-rabbit IgG CF405M polyclonal antibody (Sigma Aldrich) and Erongr2-Phycoerythrin (eBioscience). CFSE labelling, adoptive transfer and immunization with HEL-SRBC was as described12.
For microarray analysis, aliquots comprising 105 selected cells were sorted into 15 ml Falcon tubes containing 2 ml foetal calf serum, on ice. The cells were washed in PBS, resuspended in 1,400 μl PBS, transferred to 1.5 ml Eppendorf tubes, centrifuged at 900 g for 6 min at 4 °C, supernatants removed, and tubes with cell pellets snap frozen in liquid N2, stored at −80 °C, and all tubes shipped together on dry ice to Miltenyi Biotec, Germany for processing and microarray hybridization as a single batch.
Cell pellets were lysed and RNA isolated on NucleoSpin RNA II kits (Macherey-Nagel) using standard protocols. RNA quality was analysed on an Agilent 2,100 Bioanalyzer and samples with RIN >6.4 were used. Cy3 labelled cRNA was produced by linear T7-based amplification using the Agilent Low Input Quick Amp Labelling Kit following the manufacturer’s protocol, and yields measured with a Nanodrop ND-1,000 Spectrophotometer. 600 ng Cy3-labelled fragmented cRNA was hybridized for 17 h at 65 °C to Agilent Whole Mouse Genome Oligo Microarrays 8 × 60 K according to the Agilent 60-mer microarray processing protocol. After washing, the arrays were scanned on an Agilent Microarray Scanner System. The image files were analysed by two independent methods, Agilent Feature Extraction Software and limma, with similar results. Using Agilent Feature Extraction Software, signal intensities from individual features were normalized by dividing by the median signal of all features on the same array, yielding the relative mRNA units shown for selected probes in Fig. 1a and Supplementary Fig. 11, and for all probes in Supplementary Data 1.
Using limma39 as detailed in Supplementary Methods, probe intensities were corrected for background fluorescence using the normal-exponential convolution model implemented by limma followed by cyclic loess normalization. Two low quality samples were excluded from further limma analysis and from Agilent analysis: one of three MM4 naïve and one of three MM4:ML5 anergic samples. To focus on B cell expressed genes, probes with a positive feature score in all MD4:ML5 or MM4:ML5 samples were selected, yielding 33,653 anergic B cell expressed probes (Supplementary Data 2). These were analysed for differential expression between anergic and naïve cells, or between anergic cells with and without IgD. The estimated log2 fold change and 95% confidence interval (CI) were computed. A minimum threshold was set for differential expression, with 2-fold chosen for comparison of anergic and naïve B cells, and 1.25-fold for comparing anergic cells with or without IgD. Probes with CIs that overlapped this threshold but had an estimated log fold change beyond the threshold were classed as having moderate evidence of differential expression. Probes where the entire CI fell outside the threshold were classed as having strong evidence of differential expression.
Microarray data that support the findings of this study have been deposited in GEO with the primary accession code GSE82091. All other data that support the findings of this study are available from the corresponding author on request.
How to cite this article: Sabouri, Z. et al. IgD attenuates the IgM-induced anergy response in transitional and mature B cells. Nat. Commun. 7, 13381 doi: 10.1038/ncomms13381 (2016).
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This work was supported by NIH grant U19 AI100627 and NHMRC grants 585490, 1016953 and 1081858 to C.C.G., NHMRC CJ Martin Fellowship 595989 to J.H.R., an Endeavour Award from the Australian Government to Z.S. and the National Collaborative Research Infrastructure Scheme Australian Phenomics Facility.
This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/