B cell antigen receptor signal strength and peripheral B cell development are regulated by a 9-O-acetyl sialic acid esterase

The Siae gene (sialate:O-acetyl esterase) can generate two alternatively spliced variants. One form contains a signal peptide and can encode a protein that enters the secretory pathway (originally called Lse [lysosomal sialic acid acetylesterase]), whereas the other lacks this signal peptide and encodes a cytosolic esterase (25–27). Although the Lse protein was thought to localize to lysosomes, a site that would not be suitable for the modification of Siglec ligands, we have observed in transfected cells that this protein exhibits, at best, a limited lysosomal localization and is secreted and can bind to the surface of transfected cells (Figs. S1 and S2, available). The enzyme, therefore, has the potential to remove acetyl groups from 9-O-acetylated Siglec ligands in a post Golgi vesicle or at the cell surface. Although some antibodies against Siae have been generated, they lack the specificity to detect Siae in murine cells and tissue sections. The actual location of the Siae protein in murine B cells is therefore yet to be conclusively established.

Exon 2 of the Siae gene is unique to Lse. An engineered in-frame deletion of exon 2 in a murine Lse complementary DNA resulted in a protein that lacked esterase activity (Fig. 1 A). Genomic deletion of exon 2 was achieved as described in the Materials and methods (Fig. 1 B). After germline transmission, homozygous mutant mice were generated and were found to be viable. Truncated exon 2–deficient Siae mRNA could be detected in KO mice (Fig. S3, available). Cytosolic esterase mRNA continues to be transcribed in these mutant mice. We refer to these KO animals as Siae^Δ2/Δ2 mice.

Because Siae has the potential to remove 9-O-acetyl residues from α2–6-linked sialic acid containing Siglec ligands, we predicted that B cells from mice lacking this esterase might exhibit enhanced BCR signaling similar to that noted in CD22-null mice (13–16). Mice were first bred into the C57BL/6 background for 10 generations. B cells from WT and Siae^Δ2/Δ2 mice were gated on, and the accumulation of cytoplasmic calcium after ligation of the BCR was analyzed using flow cytometry. As seen in Fig. 2, BCR cross-linking resulted in an accelerated and enhanced calcium flux. A similar result was seen when purified splenic B cells from mutant and WT mice were analyzed (Fig. S4, available). These data suggested that in the absence of functional Siae, BCR signal strength is markedly enhanced and that this alteration in signal strength is an intrinsic property of mutant B lymphocytes.

The defect in Siae could result in the increased acetylation of α2–6-linked sialic acid on N-glycans in B cells and, thus, attenuate the ability of glycoproteins on B cells to ligate CD22 and generate inhibitory signals. We sought to examine if there was a defect in CD22 signaling in Siae^Δ2/Δ2 mice. After BCR cross-linking, CD22 was isolated by immunoprecipitation, and immunoprecipitates were examined for CD22 tyrosine phosphorylation and for associated SHP-1 using Western blot assays. As seen in Fig. 3 (left, experiment 1; and right, experiment 2), tyrosine phosphorylation of CD22 after BCR signaling was reduced in Siae^Δ2/Δ2 mice in spite of similar levels of surface CD22 expression in WT and mutant mice (Figs. S5 and S6, available). Recruitment of SHP-1 by CD22 was clearly impaired after BCR ligation of Siae^Δ2/Δ2 B cells as compared with WT B cells. This result showing diminished SHP-1 recruitment in Siae^Δ2/Δ2 B cells is consistent with the accelerated enhancement in BCR signal strength observed in Siae^Δ2/Δ2 mice.

It remained to be demonstrated whether a defect in Siae would result in enhanced 9-O-acetylation of α2–6-linked sialic acid in Siae^Δ2/Δ2 mice. 9-O-acetylation of sialic acid has not been previously observed on B cells, and it is likely that this modification occurs only transiently on WT B lymphocytes. A well established reagent for the detection of 9-O-acetylated α2–6-linked sialic acid is a fusion protein containing the Fc portion of human IgG and the influenza C hemagglutinin esterase treated with diisopropyl fluorophosphate (28). Modification of the catalytic serine nucleophile in this fusion protein by diisopropyl fluorophosphate permits the viral protein to bind to 9-O-acetyl sialic acid but not to cleave it. This reagent, designated CHE-FcD, binds to N-linked glycans that are decorated with 9-O-acetylated α2–6-linked sialic acid moieties. As predicted, a consistent increase in 9-O-acetylation was detected in Siae^Δ2/Δ2 B cells compared with WT B cells (Fig. 4).

In keeping with the increased magnitude of BCR signaling, antigen receptor ligation induced enhanced proliferation of B cells in Siae^Δ2/Δ2 mice (Fig. S7, available). Thymidine uptake studies uncovered a delayed stimulus strength–dependent inhibition of proliferation (Fig. S8) that might be related to a modest increase in ongoing spontaneous apoptosis of follicular B cells in mutant mice (Fig. S9). Four different CD22-null mutant mice have been described by different groups (13–16), and although enhanced antigen receptor signaling was observed in all four mutants, BCR-induced proliferation was enhanced in some mutants and diminished in others. Enhanced cell death has also been reported in some CD22 mutants (14, 16).

Enhanced BCR signaling in the absence of negative regulators of the BCR such as Aiolos or CD22 has been linked to the relative absence of marginal zone (MZ) B cells (29–31). As seen in Fig. 5 A, immunohistochemistry revealed a marked reduction in splenic MZ B cells in Siae^Δ2/Δ2 mice, although some reduction in follicle size was also apparent. Flow cytometric analysis revealed that IgM^hiIgD^hi CD21^hi MZ B cell precursors (MZP), as well as IgM^hiIgD^loCD21^hi MZ B cells, were markedly reduced in Siae^Δ2/Δ2 mice (Fig. 5 B). An analysis of the absolute numbers of B cells in Siae^Δ2/Δ2 mice (Table I) revealed a striking decrease in MZP and MZ B cells and a more modest reduction in follicular B cells. These results are consistent with the ability of Siae to negatively regulate BCR signaling.

Follicular B cells occupy two niches: the follicular niche in conventional secondary lymphoid organs and the perisinusoidal niche in the BM (32, 33). In mutant mice, including Aiolos^−/− mice, CD22^−/− mice, and CD72^−/− mice, all of which exhibit enhanced BCR signal strength, there is a selective loss of recirculating follicular phenotype B cells from the perisinusoidal niche even as these cells efficiently seed the follicular niche (14–16, 34, 35). We examined follicular phenotype B cell populations in the perisinusoidal niche in Siae^Δ2/Δ2 mice and noted a marked and selective loss of recirculating B cells in the BM (Fig. 5 C and Table II, fraction F), again paralleling a phenotypic alteration previously noted in CD22-null mice (14–16).7 We analyzed peritoneal B-1 B cell populations in Siae^Δ2/Δ2 and WT mice and noted an increase in IgM^hi CD43⁺ CD5⁻ B1b B cells but decreased numbers of IgM^hi CD5⁺ B1a B cells in Siae^Δ2/Δ2 mice (Fig. 5 D). The CD22 mutants showed increased B1 cells (13), increased B1a cells (15), or normal numbers (16).

To determine whether the alteration in peripheral B cell populations reflected a cell-intrinsic defect in Siae function, we reconstituted Rag-1^−/− mice with hematopoietic stem cells from WT and Siae^Δ2/Δ2 BM. Siae^Δ2/Δ2 lymphocytes in reconstituted mice exhibited a defect in MZ B cell development and a reduction of perisinusoidal BM B cells (Fig. 5 E), which establishes that these developmental abnormalities represent lymphocyte-intrinsic defects. No gross defect in T cell development was discovered in the absence of Siae (Fig. 5 F), although there was a small decrease in T cell numbers both in the thymus and the spleen (Tables III and IV). Other B cell–deficient mice have been reported to have a mild decrease in T cell numbers (36). The phenotype of Siae mutant mice in which a subtle alteration may have been generated in CD22 ligands is very different from that of the ST6GalI-null mouse, which completely lacks α2–6-linked sialic acid and which presents with defects in B cell activation and in thymocyte development (37, 38). Murine CD22, unlike human CD22, has a markedly higher affinity for Neu5Gc (sialic acid with an N-glycolyl moiety in the C5 position; the major form in the mouse) as compared with Neu5Ac (sialic acid with an N-acetyl moiety in the C5 position) (39–42). The enzyme CMP-Neu-5Ac hydroxylase (CMAH) converts CMP-Neu5Ac to CMP-Neu5Gc, and mice lacking this enzyme not only synthesize N-glycans with a subtly different form of sialic acid that binds CD22 poorly but they also exhibit a poorly understood down-regulation of Siae expression and an increase in 9-O-acetylation of Neu5Ac (43). Cmah-null mice as well as Cmah/Siae double mutant mice would therefore be expected to exhibit BCR hyperreactivity, the relative loss of MZ B cells, and a relative loss of perisinusoidal B cells. Indeed, B cells from Cmah mutant mice have been reported to exhibit an enhanced proliferative response to BCR cross-linking (44). We show here that B cells from Cmah-null mice have higher levels of 9-O-acetyated sialic acid as revealed by CHE-FcD binding (Fig. S10, available). Both Cmah-null mice and Cmah/Siae double mutant mice present with enhanced BCR activation as measured by the release of intracellular calcium after antigen receptor ligation and, as predicted, present with marked reductions in MZ B cells and perisinusoidal B cells (Figs. S10 and S11). A combination of two changes in the structure of sialic acid that each compromise CD22 binding results in a phenotype that is similar to that seen in mice with enhanced 9-O-acetylation of sialic acid alone but is quite distinct from the phenotype of mice with the complete loss of α2–6-linked sialic acid.

Many phenotypic features of Siae^Δ2/Δ2 mice resemble those seen in the absence of CD22. These include enhanced BCR signaling, the loss of MZ B cells, and a reduction in perisinusoidal B cells. Subtle alterations in responses to synthetic T-dependent and T-independent antigens have been described in CD22-null mice, and an altered spectrum of responsiveness to similar antigens was also noted in Siae^Δ2/Δ2 mice. Immunization with DNP-KLH resulted in diminished IgG2a, IgG2b, and IgG3 responses in Siae mutant mice (Fig. S12, available). Siae^Δ2/Δ2 mice spontaneously developed high levels of certain class-switched serum immunoglobulins, including IgE and IgG_2b (Fig. 6 A), as well as high titers of antinuclear antibodies and circulating immune complexes as early as at 20 wk of age (Fig. 6 B).10 These mice also develop an immune complex glomerulonephritis (Fig. 6, C and D). Anti-DNA antibodies develop in CD22-null mice after 9 mo of age, but glomerular immune complex deposits were not observed in these mice on a C57BL/6 background (45).

It has been suggested that CD22 may function as an enforcer of peripheral tolerance (45), and a possible role for CD22 in setting signaling thresholds in the context of tolerance has been previously postulated (46). Lyn-null mice (17–19), as well as conditional SHP-1 mice (21), develop a lupuslike phenotype, supporting the notion that CD22, possibly other Siglecs, SHP-1, and Lyn are part of an inhibitory axis setting higher thresholds for B cell activation by self-antigens as well as by exogenous antigens (46). Siae may be an important component of this inhibitory axis (Fig. 7).

In this article, we show that mice with an engineered deletion in Siae phenocopy many of the alterations seen in CD22 -null mice (13–16). These common features include an enhancement of BCR-induced release of calcium from internal stores, the loss of MZ B cells, a reduction in BM perisinusoidal B cells, alterations in B cell proliferation that are dependent on the degree and duration of BCR cross-linking, some increase in follicular B cell apoptosis, and the spontaneous development of antinuclear antibodies.

Our results suggest that in Siae mutant mice, terminal α2–6-linked sialic acid moieties on N-linked glycans are 9-O-acetylated and, as a result, CD22 cannot readily inhibit BCR signaling. Although these results imply that ligation of CD22 by α2–6 sialic acid–containing ligands contributes to inhibitory signaling, the precise role that sialoglyconjugates play in the function of CD22 remains to be determined with certainty. The phenotypes of CD22 KO mice clearly suggest that a major in vivo function of this Siglec is to attenuate BCR signaling, but other studies have suggested that CD22 may potentially contribute to both inhibitory and activating functions (47). Mutating or blocking the sialic acid binding site of CD22 in a cell line context led to a loss of CD22 inhibitory signaling as indicated by an enhanced calcium flux after BCR ligation (11, 48). Knockin mice harboring an identical mutation in the sialic acid binding site of CD22 exhibited some, but not all, of the phenotypic changes seen in CD22 null mice. Strikingly, an enhanced calcium flux was not observed in this knockin mouse after BCR ligation (22). Background-dependent modifier genes do contribute to differences seen in CD22-dependent phenotypes in KO and knockin animals, and there is growing evidence from SNP genotyping that even inbred mice strains are heterogenous at several loci (49). Nevertheless, no satisfactory explanation can be provided at present for the discrepant results of site directed mutagenesis-based studies of CD22 function.

ST6GalI is the sialyl transferase that transfers sialic acid in an α2–6-linked form from CMP-Sia to the preterminal galactose on N-glycans in the Golgi. In mice lacking ST6GalI, the absence of ligands for CD22 might have been expected to result in enhanced BCR signaling. Instead, these mice exhibit a gross defect in B cell signaling after ligation of the BCR and of other mitogenic receptors in B cells (37). One interpretation provided for these data is that in the absence of α2–6-linked sialic acid in the ST6GalI-null mouse, CD22 fails to multimerize and cannot be sequestered from the BCR (50). It is possible that in the absence of ST6GalI, inhibition of the BCR by CD22 is enhanced because of an abnormal facilitation of CD22 and IgM colocalization (23). However, the basis for this enhanced colocalization is not known, the reason for the relatively broad defect in signaling in ST6GalI mutant mice is not entirely clear, and is it not fully understood why these mice have defects in thymocyte development (38).

Mice that lack both CD22 and ST6GalI exhibit enhanced BCR signaling similar to that seen in CD22-null mice (23, 51). Although these data have been interpreted to indicate a restoration in the absence of CD22 of BCR signaling that was diminished in the absence of ST6Gal I, both studies on cd22^−/−/ST6GalI^−/− mice do not actually show a restoration of BCR signaling to WT levels but demonstrate that BCR signaling is enhanced in double mutant mice to levels similar to those seen in cd22^−/− B cells. The loss of CD22 may be considered to be dominant. These data are therefore compatible with the possibility that CD22 can provide inhibitory signals constitutively, presumably by providing ITIM tyrosines as a potential substrate for phosphorylation by Lyn and other Src family kinases. Nevertheless, CD22 might provide optimal inhibitory signals only in the presence of its sialoglycoconjugate ligands. The total absence of α2–6-linked sialic acid may represent a relatively drastic alteration that could result in aberrant capping of N-acetyllactosamine termini of many glycans (52) and might affect B cell function adversely in poorly understood ways. A subtle alteration of the structure of CD22 ligands, such as enhanced 9-O-acetylation of sialic acid in mice with mutant Siae, or the combined loss of Neu5Gc and enhanced 9-O-acetylation of sialic acid in the Cmah-null mouse and the Cmah-Siae double mutant mouse both appear to provide somewhat different insights compared with those obtained from the ST6Gal1 KO mouse harboring a more global alteration in CD22 ligand structure.

In contrast to Siae mutant mice that have detectable anti-DNA antibodies when they are 5 mo of age, cd22^−/− mice, also on a C57BL/6 background, develop anti-DNA antibodies only when they are at least 9 mo old, and these latter mice do not develop kidney deposits of antibody and complement (45). Collectively, the spontaneous secretion of increased levels of class switched immunoglobulins and antichromatin antibodies, the accumulation of glomerular immune complex deposits, and the apparent activation-dependent attrition of follicular B cells in Siae mutant mice suggest a stronger phenotype in these mice than that observed in mice lacking CD22. These data suggest indirectly that Siae might not only attenuate CD22 function but may also possibly regulate an additional Siglec or Siglecs in B cells, although direct evidence for such a hypothesis is lacking. A second inhibitory Siglec in murine B cells, Siglec G, has been previously described (53). However, although Siglec G–deficient mice exhibit a marked increase in B-1a B cells, IgG autoantibodies are not increased. It remains to be determined whether this or any other Siglec, other than CD22, is regulated by Siae. It also remains to be determined if an engineered defect in Cmah leads to an autoimmune state.

It has been previously suggested that CD22, Lyn, and SHP-1 set a threshold for B cell activation (46). The enhanced BCR signaling noted in mice that lack CD22 and in Siae mutant mice suggests that loss of tolerance may result from a mechanism that is distinct from models invoking clonal deletion, receptor editing, and anergy, all of which might be presumed to be enhanced, not abrogated, by an increase in BCR signal strength in B cells. These well established mechanisms of tolerance have been studied extensively using transgenic and knockin models that use relatively high-affinity BCRs for model self-antigens. The analysis of anti–HEL-soluble HEL double transgenic mice in a cd22-null background did not result in a break in tolerance to HEL (46, 54). We expand here on a suggestion originally made by Neuberger and coworkers (45) and also discussed by Cornall et al. (46) that CD22, and possibly other Siglecs, might set thresholds for the activation of B cells in the periphery. Weakly self-reactive B cells may not be activated by self-antigens because CD22 helps set an activation threshold in the periphery. If the strength of activation of B cells by self-antigens were to cross a threshold, the processing and presentation of self-antigens could be facilitated, and the unwanted entry of relatively weakly self-reactive B cells into the germinal center reaction might also occur. There are many potential ways in which, during an infection, a self-structure may be physically complexed with a foreign protein. In a viral infection, for instance, host chromatin may contain proteins of viral origin bound to host antigens including host nucleic acids. If the BCR threshold signaling is not set high enough (because of defects in CD22 or Siae for instance), self-antigens might not only activate weakly self-reactive B cells but may also deliver physically linked foreign proteins to endosomes in B cells, thus permitting the receipt of T cell help for somatic mutation. It is perhaps significant that in human autoimmune diseases in which antibodies are known to be relevant, pathogenic autoantibodies tend to be somatically mutated and class switched. Siglecs, assisted by Siae, presumably play a key role in keeping weakly self-reactive B cells quiescent. It is clear that analyses of responses to one specific T cell–dependent synthetic immunogen are not apparently consonant with the spontaneous immunoglobulin and autoantibody levels observed in Siae mutant mice, and there is a need for more detailed immunization studies in these mutant mice and in controls.

Elucidating the full complexity of the biological functions of 9-O-acetylation of sialic acid will require the analysis of a mutant organism that lacks the sialic acid O-acetyl transferase in the Golgi complex that can transfer an acetyl moiety from acetyl CoA to the 7-OH or 9-OH position of terminal sialic acid residue on glycoconjugates. Although more than one sialic acid O-acetyl transferase may exist (55), to date no such O-acetyl transferase has been molecularly identified. Our data indicate that modulation of 9-O-acetylation of sialic acid, a modification which was first described more than half a century ago (1), regulates inhibitory signaling in B lymphocytes and participates in the maintenance of immunological tolerance. Furthermore, our ongoing unpublished studies involving deep resequencing of the SIAE gene in human controls, and patients with autoimmunity have revealed a strong association of rare loss-of-function variants of SIAE with autoimmune disease, suggesting that the regulatory function of Siae is not restricted to rodents (unpublished data).

Siae was identified over a decade ago as a gene that is regulated during B cell development (26). We have now shown that this enzyme regulates the cell surface acetylation of sialic acid. It remains to be determined where exactly in B cells Siae is located and whether the expression and catalytic activity of this enzyme is regulated during antigen-mediated B cell activation.

8–12-wk-old C57BL/6 and B6.129S7-Rag1 mice (The Jackson Laboratory) were used in this study. Animal procedures were approved by the subcommittees on research animal care at Massachussetts General Hospital and at the University of California, San Diego.

Exon 2 of Siae encoding the N terminus of the Lse form of Siae was deleted. The Siae targeting vector was assembled from a 129/Sv genomic clone by inserting the 0.7-kb BamHI–XbaI fragment, containing exon 2 of Siae, into the BamHI site of the pflox vector (56). Adjacent 129/Sv Siae genomic sequences were added by subcloning the 1.5-kb XbaI–NheI fragment into the XhoI site and the BamHI–BamHI 9-kb fragment into the SalI site of pflox. 10 µg of NotI-linearized targeting vector was electroporated into 150 µg/ml R1 ES cells and G418-resistant transfectants, which was positive by PCR for homologous recombination and retaining all three loxP sites, and was transfected with the pCreHygro expression vector. After 4 d of 2-µM gancyclovir selection, subclones were isolated and those harboring the Siae-floxed allele were detected by Southern blotting with BamHI–SpeI genomic digestion and a loxP probe. Two ES cell clones were used to generate chimeric mice in C57BL/6 host embryos. Offspring were genotyped by Southern blotting with BamHI–SpeI tail DNA hybridized to the 3′ probe (3S0.6). Heterozygous offspring were mated to C57BL/6 ZP3-Cre mice. Female offspring were then mated to produce homozygous Siae^Δ2/Δ2 mice. PCR primers used for genotyping were as follows: PS-LseG-16, 5′-TTTAGGAGCAAGGGGTTGGCCAAGA-3′; PS-LseG-B, 5′-GGTTTCCTGACCTTGGCACAAGGT-3′; and LOX-R, 5′-CGGTACCCGGGGATCAATTCGAG-3′. PS-LseG-16 and PS-LseG-B yield the 420-bp WT band and PS-LseG-16 and LOX-R the 340-bp mutant band. The mice used in these studies were backcrossed into the C57BL/6 background for 10 generations and then maintained by intercrossing.

The following murine monoclonal antibody conjugates were used: Allophycocyanin (APC)-RA3-6B2 (anti-CD45R/B220, rat IgG_2a, κ), FITC-S7 (anti-CD43, rat IgG_2a, κ), FITC-7G6 (anti-CD21/CD35, rat IgG_2b), FITC-Cy34.1 (anti-CD22.2, mouse [DBA/1] IgG₁, κ), APC-53-7.3 (anti-CD5, rat IgG_2a, κ), APC-M1/70 (anti-CD11b, rat IgG_2b, κ), r-phycoerythrin (R-PE)-GK1.5 (anti-CD4, rat IgG_2b, κ), and FITC-53-6.7 (anti-CD8a, rat IgG_2a,κ; all from BD); and APC and R-PE-1B4B1 (anti-IgM, rat IgG), and R-PE-11-26 (anti-IgD, rat IgG_2a, κ; SouthernBiotech).

Single cell suspensions were made from spleen, BM (one femur and tibia), thymus, and peritoneal washings using standard methodology. Red cells were lysed with 2 ml ACK lysing buffer (Cambrex). The lysis buffer was neutralized by adding 10 ml PBS and 0.2% BSA (PBA). Before staining, 1 × 10⁶ cells were reacted with 2.5 µg of 2.4G2 (anti-CD16/CD32 [Fcγ III/II receptor], rat IgG_2b, κ; BD). Surface staining was performed using appropriate dilutions of antibodies in 12 × 75-mm round-bottom polystyrene tubes in a volume of 200 µl for 30 min in the dark at 4°C.

Flow cytometric analysis was performed as previously described (57) on a dual laser FC500 (Beckman Coulter), a MoFlo (Dako), and a FACSAria (BD). Unstained cells were used to set voltage and single color positive controls were used for electronic compensation. Viable cells were determined by forward and side scatter characteristics and 3–5 × 10⁴ gated events were collected. Gates were set as previously described (33). Processed samples were analyzed using RXP (Beckman Coulter), and FloJo v8.8 (Tree Star, Inc.) analysis software.

Rag-1^−/− mice were irradiated with 7.5 Gy and reconstituted with 5 × 10⁶ WT or mutant adult BM cells via tail vein injection. Reconstitution of the BM and spleen was assessed 7 wk later by flow cytometry.

The CHE-FcD probe, a fusion protein composed of the extracellular domains of the influenza C hemagglutinin esterase (CHE) which binds 9-O-acetylated sialic acids, and the Fc portion of human IgG₁(Fc), treated with diisopropylfluorophosphate (D) was generated as previously described (28, 58). The chimeric CHE Fc-D protein was precomplexed with Cy5 or FITC-F(ab′)₂ goat anti–human IgG (Jackson ImmunoResearch Laboratories; 1 µl of a 1:10 dilution of CHE-FcD in PBA with 2 µl of the secondary antibody in a total volume of 50 µl PBA) for 2 h at 4°C in the dark. 10⁶ cells in 50 µl PBA were preincubated for 45 min at 37°C, added to the precomplex, and incubated on ice for an additional 1.5 h. The cells were washed once, reacted with 2.4G2, an Fcγ III/II receptor-blocking antibody, and surface stained as described earlier. Staining of acetylated sialic acid on CD4⁺ T cells served as a positive control (28).

In brief, Indo-1 AM (acetoxymethyl ester; Invitrogen) was added to 3 × 10⁶ splenocytes in 500 µl HBSS (Invitrogen) and 10% FCS (HBSS-10). The final concentration of Indo-1 was 1 µM. The cells were incubated in a 37°C water bath for 30 min, held at room temperature for 5 min, and washed with HBSS containing 2 mM Hepes buffer and no serum (HBSH-0). All manipulations including the incubation were in the dark. Cells were then reacted with 2.4G2, an Fcγ III/II receptor-blocking antibody, surface stained as described earlier, and washed with HBSH-0, all in the dark at 4°C. The cells were resuspended in the residual volume and an additional 50 µl HBSH-0 was added to make a total volume of ∼150 µl. The cells were then split into three 50-µl aliquots and the volume made up to 100 µl with HBSH-0. The cells were held in the dark at 4°C before analysis. Before calcium flux measurements, the cells were filtered and incubated at RT for 10 min. The cells were warmed to 37°C in a heat block for 6 min and placed in the flow cytometer in a chamber maintained at 37°C. After a flow rate of ∼500 cells/s was established, acquisition was begun to record a 30-s baseline. After addition of stimulatory antibody (Fab₂ fragments of anti-IgM), acquisition was continued for another 180 s with the cells at 37°C. Between samples, distilled water was run at a high flow rate for 2 min to flush the lines. The flow cytometer used was a FACS Vantage SE (BD) operating with a laser (Innova Enterprise Model 621; Coherent, Inc.).

Total serum immunoglobulins of various isotypes were analyzed using an ELISA approach (Alpha Diagnostic International, San Antonio, TX). ELISA was also used to quantify serum autoantibodies (anti-dsDNA, anti-ssDNA, anti-histone, anti-Sm) and circulating immune complexes (all from Alpha Diagnostic).

Immunofluorescence was performed as previously described (59) with some modifications. Sections for immunofluorescence were blocked with 20% normal goat serum in 1% BSA/PBS for 20 min and stained with FITC-AffiniPure F(ab′)₂ goat anti–mouse IgG (H + L; Jackson ImmunoResearch Laboratories) diluted 1:200 in 5% normal goat serum/1% BSA/PBS for 15 min in the dark at room temperature. After rinsing three times in PBS for 5′ each, sections were mounted using 25 µl of Vectashield (Vector Laboratories). Immunohistochemistry was performed as described previously (60). Paraffin sections were stained with hematoxylin and eosin and the periodic acid Schiff reagent using standard methods.

The method used by Poe et al. (22) was adhered to with some modifications. 10⁷ murine splenocytes or sorted 5 × 10⁵ B cells were lysed before and after BCR ligation using 1% NP40 in PBS. Lysates were immunoprecipitated using anti-CD22 (CD-22 rabbit IgG; Epitomics, Inc.) and protein A Sepharose, and after separation on an SDS/polyacrylamide gel and Western transfer, membranes were probed with antibodies to CD22, SHP-1, and phosphotyrosine (anti-CD22.2, clone Lyb-8.2 [BD]; anti-SHP1/2 rabbit polyclonal IgG, and antiphosphotyrosine, clone 4G10 [Millipore]). Relative expression of CD22 and SHP-1 was quantified using Quantity One software (Bio-Rad Laboratories).

Fig. S1 shows that Siae is secreted from stably transfected U2OS cells. Fig. S2 shows that Siae is expressed on the surface of transfected U2OS cells and is also partially present in lysosomes. Fig. S3 shows the presence of truncated Siae mRNA in mutant mice as revealed by RT-PCR. Fig. S4 shows accelerated and enhanced BCR-induced calcium release in purified B lymphocytes from Siae mutant mice. Figs. S5 and S6 show that surface expression of CD22 on splenocytes in Siae^Δ2/Δ2 mice is unaltered compared with WT mice. Fig. S7 shows that Siae mutant splenocytes proliferate more than WT B cells. Fig. S8 shows a stimulus strength–dependent inhibition of B cell proliferation after BCR cross-linking. In Fig. S9, mutant resting B splenocytes show a mild increase in apoptosis. In Fig. S10, peripheral B cell populations and the response to BCR cross-linking of the Cmah-null mouse are depicted, and in Fig. S11, similar studies are presented that were performed on Siae/Cmah double mutant mice. Fig. S12 shows antibody responses to synthetic T cell–dependent and T cell–independent antigens in Siae mutant and WT mice.

We acknowledge initial studies of the mouse phenotype by Dawne Page, Jamey Marth, and Laura Martin. Joanne Yetz-Aldape, Suzan Lazo-Kallanian, and John Daley are thanked for help with flow cytometry. We thank Eveline Schneeberger for advice on renal pathology. We thank Stacey Beganny and Jonathan Foley for help with densitometry, and Emily Forbes and Johnathan Whetstine for help with confocal microscopy.

These studies were supported by grants from the National Institutes of Health to S. Pillai (AI064930 and AI069458) and A. Varki (GM32373 and P01HL057345).

The authors have no conflicting financial interests.