Murine lupus susceptibility locus Sle2 activates DNA-reactive B cells through two sub-loci with distinct phenotypes.

The NZM2410-derived Sle2 lupus susceptibility locus induces an abnormal B-cell differentiation, which most prominently leads to the expansion of autoreactive B1a cells. We have mapped the expansion of B1a cells to three Sle2 sub-loci, Sle2a, Sle2b and Sle2c. Sle2 also enhances the breach of B-cell tolerance to nuclear antigens in the 56R anti-DNA immunoglobulin transgenic (Tg) model. This study used the Sle2 sub-congenic strains to map the activation of 56R Tg B cells. Sle2c strongly sustained the breach of tolerance and the activation of anti-DNA B cells. The production of Tg-encoded anti-DNA antibodies was more modest in Sle2a-expressing mice, but Sle2a was responsible for the recruitment for Tg B cells to the marginal zone, a phenotype that has been found for 56R Tg B cells in mice expressing the whole Sle2 interval. In addition, Sle2a promoted the production of endogenously encoded anti-DNA antibodies. Overall, this study showed that at least two Sle2 genes are involved in the activation of anti-DNA B cells, and excluded more than two-thirds of the Sle2 interval from contributing to this phenotype. This constitutes an important step toward the identification of novel genes that have a critical role in B-cell tolerance.

INTRODUCTION

Sle2 is one of three major quantitative trait loci that increase susceptibility to lupus nephritis in the NZM2410 mouse model1. Analysis of congenic strains combining these three loci on a C57BL/6 (B6) genetic background has shown that Sle2 increased the frequency of fatal disease from 41% in B6.Sle1.Sle3 to 98% in B6.Sle1.Sle2.Sle3 mice2. Sle2 expression on a B6 background is associated with a number of B cell defects, including an expansion of the B1a cell compartment, especially in the peritoneal cavity (PerC). Using congenic recombinants, we have determined that the expansion of B1a cells mapped to three sub-locus, Sle2a, Sle2b, and Sle2c, with the major contribution provided by the telomeric Sle2c3. Sle2 also increased production of polyreactive IgM antibodies (Ab)4, which may be at least in part related to the expansion of the B1a cell compartment.

The 56R immunoglobulin heavy chain (HC) transgenic (Tg) anti-nuclear autoreactive B cells represent one of the best characterized models of B cell tolerance relevant to systemic lupus erythematosus (SLE) 5,6 Autoreactive anti-nuclear specificities are created by the pairing of the 56R HC (IgMa allotype) with a number of endogenous light chains. Contrary to the BALB/c genetic background in which 56R Tg autoreactive B cells are effectively tolerized through at a variety of checkpoints, the B6 background is more permissive and induces the production of Tg-encoded anti-DNA Abs7. The breach of tolerance by 56R Tg B cells is greatly enhanced by the MRL/lpr lupus-prone genetic background6. Sle2 also enhances the activation and differentiation of 56R Tg autoreactive B cells, in that B6.Sle2.56R mice produced significantly more Tg-encoded anti-DNA Abs than the B6.56R controls8. Furthermore, the activation of autoreactive 56R Tg B cells by Sle2 involved their preferential recruitment to the marginal zone (MZ) compartment8. MZB cells in non-autoimmune mice are enriched for autoreactive specificities9, and preferential recruitment to this compartment may represent a venue by which autoreactive B cells escape tolerance checkpoints.

The present study was conducted to map the activation of 56R Tg B cells within the Sle2 locus using the sub-congenic strains that were produced to map the expansion of B1a cells3. We have found that Sle2c, and to a lesser extent Sle2a, expression enhanced the breach of tolerance of anti-DNA 56R Tg cells. Furthermore, we found that Sle2a but not Sle2c promoted the recruitment of autoreactive B cells to the MZ. Finally, Sle2a induced the activation and differentiation of B cells, including autoreactive ones, expressing endogenous HCs. Overall, these results showed that at least two gnes are involved in the Sle2 activation of anti-DNA autoreactive B cells, and excluded more than two-thirds of the Sle2 interval from contributing to this phenotype. This constitutes an important step toward the identification of novel genes that play a critical role in B cell tolerance to nuclear antigens.

RESULTS

Two Sle2 sub-loci enhanced Ab production from 56R Tg B cells

Since their initial production and the characterization of their involvement in the accumulation of B1a cells3, the Sle2a and Sle2c intervals have been fine-mapped (Fig. 1). Sle2a is now defined as a 1.5 – 4 Mb interval of NZW origin which contains a maximum of 24 expressed genes (Table 1) plus 16 additional predicted genes. The localization of Sle2c has been refined to a 10 –15 Mb NZB interval, and it is has been renamed Sle2c1 to distinguish it from a more telomeric locus, Sle2c2 (Xu et al., submitted). In this report, Sle2c1 will be referred to as Sle2c for simplicity. The Sle2b interval in the central part of Sle2 is the largest one and it potentially overlaps with Sle2c in their respective telomeric and centromeric recombination regions.

FIGURE 1

Genetic map of the Sle2 congenic strains used in this study. The location of the markers defining the termini of each recombinant is indicated on the top. The NZB/NZW derivation of the region is also shown. NZM2410 (NZW or NZB)-derived intervals are indicated by solid boxes, with the area of recombination between the NZM2410 and B6 genomes indicated by lines on each side. The location of each marker corresponds to the NCBIM37 built. For SNPs located within a known gene, this gene is indicated in parenthesis.

Table 1

List of expressed genes in the Sle2a interval

The shaded rows show the genes of known NZW derivation, flanked by the region of recombination between the NZW and B6 genomes on each side.

Genome coordinates1	Gene	B cell expression2
53043659-53172767 (−)	Abca1	++
53217938-53230729 (−)	4930412L05Rik	nt
53259068-53263987 (+)	4930522O17Rik	nt
53275604-53283104 (−)	AI427809	nt
53453285-53635350 (+)	Slc44a1	−
53575855-53577569 (−)	1700060J05Rik	nt
53644343-53719881 (+)	Fsd1l	+
53726870-53778657 (+)	Fktn	low
53792577-53801584 (+)	Tal2	+
53838917-53874891 (+)	Tmem38b	−
53873524-53873642 (+)	n-R5s185	nt
54957920-55096435 (+)	Zfp462	+
55362915-55405109 (+)	Rad23b	+
55540015-55545338 (−)	Klf4	−
56233489-56237393 (−)	2310081O03Rik	nt
56752877-56754315 (−)	Actl7b	−
56756285-56757797 (+)	Actl7a	−
56762552-56815203 (−)	Ikbkap	+
56815217-56822473 (+)	BC026590	nt
56823807-56878060 (−)	Ctnnal1	−
56879795-56960309 (−)	D730040F13Rik	nt
56970045-57003263 (−)	6430704M03Rik	nt
57004844-57156309 (−)	Epb4.1l4b	−
57203713-57314709 (−)	Ptpn3	low

Genome coordinates are calculated from NCBIM37. + or − indicate strand orientation

Expression deduced from http://biogps.gnf.org

Activation of autoreactive 56R Tg B cells was first assessed by the presence of serum anti-DNA IgM Abs in the three sub-congenic strains as compared to B6.56R. Samples from B6.Sle2.56R were used as positive controls. B6.Sle2a.56R and B6.Sle2c.56R produced significantly more anti-ssDNA IgM than B6.56R mice, while there was no difference between B6.Sle2b.56R and B6.56R (Fig. 2A). Interestingly, both B6.Sle2a.56R and B6.Sle2c.56R produced significantly more anti-ssDNA IgM than B6.Sle2.56R mice (p = 0.006 and p = 0.001, respectively), suggesting that Sle2 also contains a suppressive locus located outside the Sle2a and Sle2c regions. As expected, the majority anti-ssDNA IgM was of the Tg-encoded IgMa allotype in B6.Sle2.56R and B6.Sle2c.56R mice (Fig. 2B). This was not the case, however, for B6.Sle2a.56R mice, in which the majority of anti-ssDNA IgM Abs carried the endogenous IgMb allotype (Fig. 2C). Similar results were obtained for anti-dsDNA IgM (Fig. 2D–F), although B6.Sle2a.56R mice produced significantly more anti-dsDNA IgMa Ab than B6.56R mice, but significantly less than B6.Sle2c.56R mice (p = 0.01). An ELISPOT analysis confirmed that the B6.Sle2c.56R splenocytes contained significantly more anti-ssDNA IgMa Ab forming cells (AFCs) than B6.56R (Fig. 2G). On the other hand, B6.Sle2a.56R splenocytes contained significantly more anti-ssDNA IgMb AFCs than B6.56R or any of the two other strains (Fig. 2I). These results show that the activation of 56R Tg autoreactive B cells maps to Sle2c, and to Sle2a to a lesser extent, with an additional contribution of Sle2a to the activation of endogenous HC autoreactive B cells.

FIGURE 2

Two Sle2 sub-loci activate 56R B cells to produce anti-DNA Ab. Serum anti-ssDNA (A–C) and anti-dsDNA (D–F) IgM, with the left panels showing total IgM, the middle panels 56R Tg-encoded IgMa, and the right panels endogenous IgMb. G–F. Splenic anti-ssDNA IgMa and IgMb AFCs. Each data point represents one mouse.

To further characterize the activation of autoreactive B cells and to determine the respective contribution of the Tg versus endogenous HC, we compared the in vitro Ab secretion from splenic B cells, either spontaneously or after stimulation by LPS. B6.Sle2c.56R B cells spontaneously produced low but significantly higher levels of total IgMa that B6.56R B cells (Fig. 3A), while B6.Sle2a.56R B cells spontaneously produced significantly more total IgMb (Fig. 3B). IgG was undetectable in the supernatant of unstimulated cells (data not shown). As expected, LPS stimulation induced plasma cell differentiation and confirmed the results observed for serum IgM. Stimulated B6.Sle2c.56R B cells produced significantly more total IgMa that B6.56R B cells (Fig. 3C). A higher level of total IgMb was observed in all three Sle2 strains, but the highest level was in B6.Sle2a.56R (Fig. 3D). Finally, B6.Sle2.56R and B6.Sle2c.56R B cells produced significantly more total IgG than B6.56R B cells, but the difference was not significant for B6.Sle2a.56R B cells (Fig. 3E). The in vitro production of anti-ssDNA IgMa and IgMb followed the same pattern as what we observed in serum samples (Fig. 3F–G). The different results obtained for total IgMb and anti-ssDNA IgMb indicate that both Sle2a and Sle2c enhanced Ab production by B cells carrying the endogenous HC, but only Sle2a supports the production of autoreactive Ab from these cells.

FIGURE 3

In vitro Ab production in the B6.Sle2.56R congenic strains. Total Ig and anti-ssDNA IgMa or IgMb were measured in the supernatant of splenocytes cultured for 5 d without (A–B) or with 5 ug/ml LPS (C–G). Total IgMa (A) and IgMb (B) in the supernatant of unstimulated cells. Total IgMa (C), IgMb (D) and IgG (E) in the supernatant of LPS-stimulated cells. Anti-ssDNA IgMa (F) and IgMb (G) in the supernatant of LPS-stimulated cells. Graphs show means and SEM of 4–6 mice per strain.

Sle2a induced the recruitment 56R Tg B cells to the MZB compartment and Sle2c increased their activation

The effect of the Sle2 sub-loci expression on the distribution of peripheral B cells and their activation was analyzed by flow cytometry. A significantly higher ratio of perC B1a versus B2 IgMa 56R Tg B cells was found in B6.Sle2.56R mice as compared to B6.56R mice (Fig. 4A). The perC B1a/B2 IgMa 56R Tg B cell ratio was however similar between the three Sle2 sub-congenic strains and B6.56R. As expected, the proportion of endogenous perC B cells with a B1a phenotype was much higher in B6.Sle2.56R than in B6.56R mice (Fig. 4B). This was also the case for both B6.Sle2a.56R and B6.Sle2c.56R, although to a lower level than with the full interval. This confirmed that the Sle2 sub-loci have additive effect to promote the recruitment of B cells into the B1a compartment, and that it also applied to Tg B cells.

FIGURE 4

Flow cytometric analysis of B cells in the B6.Sle2.56R congenic strains. A–B. PerC B1a/B2 cell ratios with B1a cells defined as B220int CD5+ and B2 cells defined as B220hi CD5−. The graph in A shows the ratios for cells gated on IgMa and the graph in B for the cells gated on IgMb. C–I. Analysis of splenic B cells. Percentage of total B220+ cells (C) and IgMa+ B220+ cells (D). E. Total CD21hi CD23− MZB cells expressed as the percentage of AA4.1− B220+ mature B cells. F. Tg MZB cells expressed as the percentage of IgMa+ B220+ cells. G. Percentage of large (FSChi) IgMa+ B220+ B cells. H. Class II MHC I-ab expression, measured as geometric mean fluorescence intensity (mfi) on IgMa+ B cells. I. Percentage of CD86+ IgMa+ B cells.

All five 56R congenic strains showed similar splenocyte numbers (data not shown), percentages (Fig. 4C) and numbers (data not shown) of B220+ cells, as well as percentages (Fig. 4D) and numbers (data not shown) of IgMa Tg B cells (about 80% of the total B cells). The distribution of these splenic B cells differed however between strains. As previously reported, the MZB compartment was greatly expanded in the 56R strains (Fig. 4E, compare to non-Tg strains on the right of the graph), and this expansion was further enhanced by Sle2 expression. Both Sle2a and Sle2c expression expanded the MZB compartment in 56R Tg mice (Fig. 4E), but only Sle2a was associated with the expansion of the 56R Tg IgMa MZB subset that has been described for the whole Sle2 interval8. Interestingly, in addition to be recruited more frequently to the MZ, total B220+ IgMa Tg B cells were also larger in B6.Sle2a.56R IgMa than in the other 56R congenic strains (Fig. 4G). Finally, the increased activation of 56R Tg IgMa B cells that has been described for B6.Sle2.56R mice8 mapped to Sle2c and not Sle2a, as shown for both class II I-ab (Fig. 4H) and CD86 (Fig. 4I). As for Ab production, the B cells in B6.Sle2b.56R mice were very similar to that of B6.56R, except for an increased I-ab expression (Fig. 4H). This indicated that the Sle2b interval can be excluded from contributing to the activation of autoreactive B cells.

The effects of Sle2a expression on the recruitment of the 56R Tg B cells to the MZ was examined by histology. We used CD1d expression to differentiate CDd1hi MZB cells from CD1dlo follicular (FO) B cells, as this staining scheme has been validated in the NZM2410 model10. The MZ looked markedly expanded in size and cell density in B6.Sle2a.56R spleens as compared to either B6.56R or B6.Sle2c.56R (Fig. 5A). The B6.Sle2a.56R MZB cells were all from Tg origin with no B6.Sle2a.56R IgMb cell being detected by histology in the MZ (data not shown). A morphometric analysis demonstrated that the MZ area contained significantly more B cells in B6.Sle2a.56R spleens than in either B6.56R or B6.Sle2c.56R spleens (Fig. 5B). The same result was obtained when the percentage of CD1d+ B cells was computed (Fig. 5B). We have previously shown that the majority of the MZB cells are located inside the follicles in NZM2410 and B6.Sle1.Sle2.Sle3 mice10. This was not the case for B6.Sle2a.56R mice in which the majority of CD1dhi B cells were located inside the MZ, outside the Moma1+ macrophages (Fig. 5A and D). Furthermore, the proportion of CD1d+ B cells that were located in the FO was not different between 56R strains (data not shown) indicating that Sle2a is not responsible for the NZM2410 intra-follicular location of the MZB cells. Finally, the percentage of CD1d− B cells was also elevated in the MZ of B6.Sle2a.56R mice (Fig. 5E). Overall, the results demonstrate that Sle2a expression greatly enhances the recruitment of autoreactive B cells to the MZB compartment, considering either lineage marker expression or tissue location.

FIGURE 5

Sle2a promotes the expansion of MZB cells in 56R Tg mice. A. Representative follicles from B6.56R, B6.Sle2a.56R, and B6.Sle2c.56R spleen sections. B220-PB (blue) and CD1d-PE (red) identify B220+ CD1dhi/+ magenta MZB cells and B220+ CD1dlo/− blue FOB cells, and their location relative to Moma-1-FITC (green)-stained metallophilic macrophages that separate the MZ from the FO zones. Note the intensity of CD1d expression in the B6.Sle2a.56R specimen. B. Proportion of B220+ B cells located in the MZ outside of the Moma-1+ ring. C. Proportion of B220+ B cells that expressed CD1dhi/+ regardless of their location. D. Proportion of B220+ B cells in the MZ that are CD1dhi/+. E. Proportion of B220+ B cells in the MZ that are CD1dlo/−. Each data point corresponds to a representative follicle from one mouse from the indicated strains.

DISCUSSION

The mechanisms by which B cell tolerance is maintained in normal individuals but breached in autoimmune conditions has been of a great interest to immunologists. Most of the studies have used B cell receptor transgenic mouse models and a number of checkpoints have been defined from these studies6. The well-documented effect of genetic background on the enforcement of these tolerance checkpoints implies that natural genetic variation regulates the fate of autoreactive B cells. Very little is known, however, on the identity of the genes that are involved. The lpr mutation that impairs Fas expression has been the first documented natural genetic variation that regulates B cell tolerance11. Very little progress has been made since with mouse models. In humans, recent progress has been made with recent association between the presence of autoAbs and loss-of-function rare variants of the sialic acid acetylesterase (SIAE) gene12 or specific common alleles of the IRF7/PHRF1 locus13.

We have used a congenic dissection approach to identify the genes responsible for lupus susceptibility in the NZM2410 model14, and we have identified two SLE-susceptibility loci, Sle1 and Sle2, that regulate B cell tolerance. Within Sle1, Sle1b regulates B cell tolerance to chromatin15. The double Tg HEL model16 has demonstrated that Sle1b impairs B cell tolerance, and differential expression of splice isoforms of Ly108, a gene located in the Sle1b interval, were shown to regulate the deletion of autoreactive immature B cells17. These results have identified Ly108 as a novel gene that regulates B cell tolerance to nuclear antigens. Sle2 affects B cell development and functions in a B cell intrinsic manner4,18. Its involvement in B cell tolerance was formally demonstrated with the 56H HC Tg model8. Therefore, the B6.Sle2.56R strain represents a good model to indentify the genes in the Sle2 locus that amplify the breach of tolerance to nuclear antigens. We have already identified three regions in Sle2, Sle2a, Sle2b, and Sle2c, that contribute to B1a cell expansion3. B1a cells are the major source of natural Abs and their repertoire is intrinsically autoreactive19. It is therefore possible that the same genetic variations regulate B1a cell numbers and B cell tolerance. We have assessed Ab production, B cell distribution and activation, and splenic histology to analyze the fate of B cells in B6.Sle2a.56R, B6.Sle2b.56R, and B6.Sle2c.56R mice. These approaches concurred in identifying Sle2c and Sle2a as being responsible for the breach of tolerance of 56R Tg autoreactive B cells. The Sle2c interval induced the production of Tg-encoded anti-DNA Ab to a level similar or higher than the whole interval, as well as the activation of 56R Tg B cells. This identified Sle2c as the major contributor to B cell breach of tolerance within Sle2. The interval that we used in this study is large (~ 24 Mb), with too many genes to consider a candidate gene approach. We have produced Sle2c recombinants to map the B1a cell expansion and identified a mutation in the promoter of the Cdkn2c gene that greatly reduces the expression of this cell cycle inhibitor and is associated by B1a cell expansion (Xu et al. submitted). One may hypothesize that defects in cell cycle regulation may affect the fate of autoreactive B cells as it has been shown for genes regulating apoptosis6. The Sle2c recombinants will be used to further map the activation of the 56R Tg B cells and determine if it co-localizes with B1a cell expansion.

The involvement of Sle2a in 56R Tg B cell activation and differentiation is more complex. The production of Tg-encoded anti-DNA Ab in Sle2a-expressing mice is intermediate between that of B6.Sle2c.56R and B6.56R mice. However, Sle2a is clearly the locus that drives the selection of the 56R Tg B cells to the MZB cell compartment. This MZB cell selection is not sufficient to recapitulate the entire Sle2 phenotype for 56R Tg B cells, indicating that once the autoreactive B cells are selected to the MZ, another step is necessary to fully activate them. B1a cells and MZB cells are functionally related and are both enriched for autoreactive specificities20, suggesting that the same gene might be responsible for the expansion of both compartments. One can hypothesize that the presence of the 56R HC Tg polarizes the B cell selection away from the B1a cell to the MZB cell compartment, but the molecular mechanism is the same. In addition, Sle2a was surprisingly associated with the differentiation of B cells producing total and anti-DNA Ab carrying endogenous HCs. Further experiments will be necessary to determine if this phenotype results from early bone-marrow selection events or a differential peripheral expansion of Tg versus endogenous HC clones. It is not possible to determine at this point whether the two major Sle2a phenotypes, MZB expansion and preferential endogenous HC activation, are regulated by the same gene. None of the Sle2a genes represents an obvious candidate gene (Table 1), and there are a large number of uncharacterized expressed sequences and predicted genes, indicating that a systematic approach will be necessary to indentify the causative gene.

The Sle2b interval was associated with a modest contribution to the B1a cell expansion3, and we have shown here that it is not involved in the activation of anti-nuclear autoreactive B cells. This region is of interest because it contains the Ifna gene cluster, and we have shown that a lower expression of type 1 interferons in B6.Sle2 mice was associated with increased autoAb levels and B1a cell numbers21. Our results with the 56R Tg model indicate that this is unrelated to the activation of anti-nuclear autoreactive B cells.

Overall, we have determined that at least two genes are responsible for the Sle2 activation of anti-nuclear autoreactive B cells and have greatly reduced the genomic location of these two genes to a third of the original whole interval. None of the genes located in either of the two loci have a known involvement in B cell tolerance, which ensures that this study will lead to the discovery of new mechanisms of B cell tolerance.

MATERIALS AND METHODS

Mice

The B6.Sle2 and B6.Sle2 sub-congenic strains have been previously described14,3. The termini of the Sle2a and Sle2c1 intervals were mapped with a panel of SNPs that are polymorphic between NZW and B6, and NZB and B6, respectively (Fig. 1). SNP genotyping was performed by direct sequencing. B6 mice were originally obtained from the Jackson Laboratories. B6.Sle2.56R and B6.56R breeders were obtained from Dr. Chandra Mohan, (UTSW). B6.Sle2a.56R, B6.Sle2b.56R, and B6.Sle2c.56R were produced by intercrossing the corresponding B6.Sle2 sub-congenic strains to B6.56R and by selecting for homozygozity at the termini shown in Fig. 1 and the presence of the 56R HC Tg as previously reported22. Mice used in this study were males and females aged between 9 and 12 mo of age. All experiments were conducted according to protocols approved by the University of Florida Institutional Animal Care and Use Committee.

Antibody measurements

Total IgM and IgG was measured as previously described23. Total IgMa or IgMb was detected in plates coated with goat anti-κ and revealed by polyclonal rabbit anti-IgMa or IgMb (Nordic Immunological Labs) followed by goat anti-rabbit-AP (Sigma-Aldrich). Anti-ssDNA and dsDNA IgM, IgMa or IgMb were detected in the same fashion on plates coated with dsDNA or heat-denatured ssDNA (50 μg/ml), as previously described24. Serum dilutions were 1:5,000 for total IgM, IgG, IgMa or IgMb, and 1:500 for anti-DNA IgM, IgMa or IgMb. In vitro Ab secretion was measured from magnetic bead-purified CD43− B cells (105 per well) cultured in complete RPMI 1640 (Cellgro) containing 10% FCS with or without 5 ug/ml LPS (Sigma) for 5 d. Ab levels were measured by ELISA in supernatants diluted 1:10 for unstimulated cells and 1:100 for LPS stimulated cells. ELISPOTs were performed as previously described24. Briefly, cells serially diluted in RPMI 1640 supplemented with 5% FCS were incubated on multiscreen filter plates (Millipore) coated with ssDNA for 5–7 h at 37°C. Bound secreted Ab was detected with polyclonal rabbit anti-IgMa or IgMb followed by goat anti-rabbit-AP. AFCs were counted using a Bioreader 4000 Pro-x (Bio-Sys).

Flow cytometry

PerC lavages or RBC-depleted splenocytes were stained as previously described23 with the following conjugated mAbs or their isotype controls, purchased from BD Biosciences or eBioscience: AA4.1 (CD93), B220 (RA3-6B2), CD5 (53–7.3) CD23 (B3B4), CD86 (GL1), CD138 (281–2), IgMa (DS-1), IgMb (AF6-78) and I-Ab (AF6). The 7E9 Ab was used to detect both NZW and B6 alleles of CD2125. Stained cells (100,000 per sample) were analyzed on a FACSCalibur™ (BD Biosciences, Mountain View, CA), and dead cells were excluded based on forward and side scatter characteristics.

Histology

Immunofluorescence staining was performed on frozen sections as previously described26, with Moma-1-FITC (Serotec), CD1d-PE (1B1), and IgMa-Biotin-SA-PB or B220-PB (RA3-6B2). Quantitation of the number and location of the cells with MetaMorph 7.5 image analysis was performed by a single operator without knowledge of the samples’ identities. For each mouse, the numbers of CD1d+ B220+ cells and CD1d− B220+ selected on each side of the Moma-1+ ring were computed for one or two representative follicles with homogeneous staining on 100X amplification images.

Statistical analysis

Unpaired one-tailed t tests, or Mann-Whitney tests when the data were not distributed normally, were used to compare the B6.56R and the B6.Sle2.56R congenic mice. Dunnett’s multiple comparison tests were used when appropriate. Data was analyzed with Graphpad Prism 4.0 software. Means, the standard errors of the mean (SEM), and the levels of statistical significance (*: P < 0.05, **: P < 0.01; ***: P < 0.001) are shown in the figures.