Main

The integrin α4β7 functions as a B and T cell adhesion receptor for the mucosal vascular addressin (mucosal addressin cell adhesion molecule 1 (MAdCAM-1)) expressed by postcapillary high endothelial venules (HEVs) in gut-associated lymphoid tissue (GALT), by lamina propria venule sites of effector cell recruitment and by stromal cells in GALT1,2,3. GALT is the major site of B cell activation and humoral immune induction for intestinal immunity. Activated B cells undergo isotype class switching in Peyer’s patches (PPs) and differentiate into migratory immunoglobulin A (IgA)-secreting plasmablasts that use α4β7 to home to mucosal surfaces4 where local production of secretory IgA provides immune protection. Presumably in support of this role, B cells home preferentially to and predominate in murine PPs, contrasting with peripheral lymph nodes (PLNs) where T cells are the majority5. This homing preference correlates with higher surface expression of α4β7 on B cells than on T cells6, but the mechanisms responsible for this differential α4β7 surface expression and its essential role in intestinal immunity have not been defined.

The tyrosine phosphatase Src homology region 2 domain-containing phosphatase 1 (Shp1; encoded by the gene Ptpn6) has essential roles in regulating immune homeostasis. Homozygous gene depletion of Ptpn6, as described in motheathen mice (Ptpn6me/me), results in severe systemic inflammation and death within weeks7. Closely related motheathen viable mutant mice (Ptpn6meV/meV) express wild-type (WT) levels of Shp1 but the catalytic activity of the enzyme is greatly reduced, resulting in a similar yet slightly less severe phenotype8. Shp1 plays essential roles in the regulation of tyrosine phosphorylation in T and B cells9,10. Shp1 can be targeted by its binding to phosphorylated immunoreceptor tyrosine-based inhibitory motif (ITIM) sequences in membrane receptors, often acting in conjunction with ITIM-bearing molecules to inhibit signaling pathways, including those involved in integrin activity. In T cells, for example, Shp1 controls deactivation of the lymphocyte function-associated antigen 1 integrin (αLβ2) to prevent aberrant adhesion of leukocytes to β2 integrin ligands or T cell adhesion to antigen-presenting cells11. In B cells, upon antigen stimulation, the phosphorylated ITIM sequences of sialic acid-binding Ig-like lectin 2 (CD22; also known as Siglec-2) recruit Shp1 to inhibit downstream components of B cell antigen receptor-induced Ca2+ signaling12. However, Shp1 has not been implicated in the control of integrin endocytosis or cell-surface expression.

Here, we report that the pair Shp1/CD22 acts in a cell-intrinsic manner to enhance α4β7 surface abundance, with profound organotypic effects on mucosal immune responses. The findings uncover a selective role for Shp1 in α4β7 endocytosis and surface expression, define CD22 ITIM- and lectin/carbohydrate-dependent mechanisms targeting Shp1 to β7 integrin in B cells, and support the significance of these mechanisms to efficient intestinal antibody responses.

Results

Shp1 augments α4β7 cell-surface display in lymphocytes

In studies of motheaten viable mutant mice (Ptpn6meV/meV) lacking Shp1 activity, immunofluorescence staining with antibodies to the α4β7 heterodimer or to the β7 subunit revealed a substantial (~90%) reduction in the median fluorescence intensity (MFI) of Ptpn6meV/meV splenic naive B cells compared with wild-type controls. The α4 subunit, which forms heterodimers with integrin β1 as well as β7, was reduced ~20% in Ptpn6meV/meV B cells, reflecting the reduction in α4β7. In contrast, we found no reduction for integrins αL, β1 and β2 (Fig. 1a). β7 was also reduced on CD4+ and CD8+ T cells (Fig. 1a). Since the Ptpn6meV/meV phenotype induces profound changes in the phenotype and homeostasis of mature B cells13 and T cells14,15, we repeated the above experiments with heterozygous motheaten viable mice (Ptpn6+/meV). α4, β7 and α4β7 integrins were also reduced on heterozygous Ptpn6+/meV B and T cells, while αL, β1 and β2 were not affected (Fig. 1b). Thus, Shp1 positively regulates the cell-surface abundance of integrin α4β7 on lymphocytes without affecting other integrins.

Fig. 1: Selective reduction of integrin α4β7 on motheathen viable B and T cells.
figure 1

a,b, Left, flow cytometry of WT versus motheathen viable (Ptpn6meV/meV) (a) or Ptpn6+/meV (b) splenic naive B cells, CD4+ T cells and CD8+ T cells stained for αL, β2, β1, α4, β7 or α4β7. Shown are pooled data (means ± s.e.m.) from n = 4 experiments with 4–6 (a) or 6–9 (b) animals per group in total. For each animal within one experiment, the MFI of the integrin staining is expressed as a percentage of the mean MFI of the WT group. Right, representative histogram overlays gated in naive B cells. All groups were compared by two-tailed Student’s t-test. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001.

Full size image

CD22 mediates Shp1-dependent α4β7 effects in B cells

The selective effect of the Ptpn6+/meV phenotype on B cells suggested potential involvement of CD22, a B cell-specific lectin known to recruit Shp1 to the plasma membrane through interactions with its cytoplasmic ITIM sequences12. B cells from Cd22–/– mice showed reduced surface α4, β7 and α4β7 expression by flow cytometry and confocal microscopy, but normal expression of αL and β1 (Fig. 2b,c,e). B cells isolated from the blood, bone marrow, PLNs and PPs of Cd22–/– animals displayed a similar reduction in α4β7 (Fig. 2f). Consistent with selective expression of CD22 by B cells, CD22 deficiency had no effect on CD4+ T cell integrins (Extended Data Fig. 1). Messenger RNA (mRNA) expression of the β7 subunit in wild-type and Cd22–/– B cells was unchanged (Fig. 2g), and subcellular imaging showed similar levels of intracellular β7 in wild-type and mutant cells, far above cell-surface levels (Fig. 2d,e), ruling out an effect of CD22 deficiency on Itgb7 gene expression and protein synthesis. Intracellular Shp1 spots co-localized with cell-surface β7 in wild-type B cells more often than in Cd22–/– B cells (Fig. 2h,i). In wild-type cells, most of these β7–Shp1 interactions aligned with clusters of CD22 (>80%; Fig. 2h,j).

Fig. 2: CD22 mediates Shp1-dependent α4β7 augmentation in B cells.
figure 2

a, Schematic of WT CD22, CD22Y2,5,6F and CD22R130E mutants. b, Left, flow cytometry of WT and CD22 mutant splenic B cells stained for αL, β1, α4, β7 or α4β7. Shown are pooled data (means ± s.e.m.) from n = 3 experiments with 7 animals per group in total, analyzed and presented as in Fig. 1. Right, representative histogram overlays gated in naive B cells. c,d, Immunofluorescence staining of cell-surface (c) and intracellular + cell-surface (d) β7 and CD22 in B cells. The nucleus, cytoplasm, β7 and CD22 are stained blue (DAPI), red (MitoTracker Deep Red), green and magenta, respectively. Scale bars, 2 μM. e, For each cell shown in c and d, we quantified the intensity of the β7 fluorescence. Shown are pooled data (means ± s.e.m.) from three independent experiments, with ~100 cells in total analyzed per condition. The mean β7 fluorescence intensity for the WT cell-surface group was set to 100 and the data are shown as a percentage of this total. NS, not significant. f, Flow cytometry of WT or Cd22–/– B cells isolated from blood, bone marrow (BM), PLNs and PPs and stained for α4β7 or isotype-matched control. Shown are pooled data (means ± s.e.m.) from n = 2 independent experiments, with five animals per group in total. g, mRNA expression of αL, β1, α4 and β7 integrins in B cells from WT and Cd22–/– mice. Shown are pooled data (means ± s.e.m.) from two independent experiments with 3–4 animals per group in total. h, Immunofluorescence staining of intracellular Shp1 and cell-surface β7 and CD22 in B cells. The nucleus, Shp1, β7 and CD22 are stained blue (DAPI), red, green and magenta, respectively. The arrowheads show Shp1–β7 colocalization and the arrows show CD22–Shp1/β7 colocalization. i,j, For each cell shown in h, we calculated Shp1–β7 (i) and CD22–Shp1/β7 (j) proximity indices (see Methods). Shown are pooled data (means ± s.e.m.) from two independent experiments, with 130 (i) and 40 (j) cells in total analyzed per condition. Groups were compared by one-way ANOVA with Dunnett’s multiple comparisons test (b and e) or two-tailed Student’s t-test (f, g and i). NA, not applicable. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001.

Full size image

These data suggest that CD22 mediates the Shp1 augmentation of cell-surface α4β7. If so, we reasoned that mutation of the Shp1-binding cytoplasmic ITIM motifs in CD22 should mimic CD22 deficiency. To address this, we used transgenic animals (CD22Y2,5,6F) in which the CD22 ITIM signaling domains were mutated16 to prevent Shp1 binding and downstream CD22 signaling (Fig. 2a). B cells in CD22Y2,5,6F animals express normal amounts of CD22 (ref. 16). CD22Y2,5,6F B cells displayed surface abundances of α4, β7 and α4β7 as severely reduced as on Cd22–/– B cells, again with no effect on the αL or β1 integrin surface expression (Fig. 2b).

CD22 is a lectin that interacts in cis with α2-6-sialic acid (α2-6-Sia)-decorated glycoproteins such as CD22 (ref. 17) or CD45 (ref. 18), affecting its distribution and motility on the B cell surface19. To assess a potential role for the CD22 lectin–carbohydrate interactions in cell-autonomous α4β7 regulation, we assessed the integrin cell-surface levels of B cells expressing a mutated lectin domain (CD22R130E) that prevents α2-6-Sia binding16 (Fig. 2a). CD22R130E B cells expressed CD22 at wild-type levels16 and displayed a significant reduction in α4β7 levels (Fig. 2b), although the effect was less severe than that of CD22 deficiency or ITIM mutations. Intermediate reduction in α4β7 was also observed in B cells from St6Gal1–/– mice, which lack α2,6-sialyltransferase activity20 and thus CD22 ligands (Extended Data Fig. 2).

α2-6-Sia-dependent CD22–β7 association at the cell surface

The reduction of β7 in the absence of CD22 lectin activity suggests cell-surface interactions between CD22 and β7. Using proximity ligation assay (PLA) in resting B cells, we detected a robust PLA signal between CD22 and β7 (~5–6 spots per cell; Fig. 3a), but there was no interaction of CD22 and β1 (Fig. 3b). A strong CD22-independent PLA signal was observed with anti-β1 and anti-α4 antibodies (detecting α4β1 heterodimer interactions), confirming the activity and specificity of the β1 antibody (Fig. 3c). To assess dependence on sialic acid, we pre-treated purified B cells with Arthrobacter ureafaciens sialidase at a concentration that retained cell viability while reducing Sambucus nigra lectin (α2-6-Sia-binding lectin) staining by ~60% (Extended Data Fig. 3). Sialidase-treated wild-type B cells showed significantly reduced association between CD22 and β7 compared with untreated wild-type cells (Fig. 3a). Together, these data show a cell-surface sialic acid-dependent CD22 association with integrin β7 but not β1.

Fig. 3: Direct physical association of CD22 and β7.
figure 3

a, Association between CD22 and integrin β7 was assessed by PLA using purified B cells from either untreated or sialidase-treated WT animals, or Cd22–/– animals. b, Association between CD22 and integrin β1 was assessed as a negative integrin control. c, Association between α4 and β1 was analyzed as a positive control for the β1 integrin. In a–c, pooled data (means ± s.e.m.) from two independent experiments are shown, with ~200–400 cells analyzed per condition in total. Inset: immunofluorescence stains. Scale bars, 2 μM. Groups were compared by one-way ANOVA with Dunnett’s multiple comparison test (a) or two-tailed Student’s t-test (c). ****P ≤ 0.0001. ND, not detectable.

Full size image

CD22–Shp1 inhibits β7 endocytosis in B cells

Cell-surface receptor expression is regulated by endocytosis and recycling21. We compared the effects of CD22 or Shp1 deficiency on endocytosis of β7 versus β1 integrins and of the transferrin receptor as a control that undergoes endocytic recycling. We assessed internalization by flow cytometry using pHrodo Red, a dye whose fluorescence increases within the acidic pH of endocytic vesicles. We incubated splenocytes at 37 °C to enable endocytosis in the presence of pHrodo Red-conjugated transferrin, anti-β7, anti-β1 or isotype-matched control antibodies. The background was defined by staining of cells with the pHrodo Red constructs incubated at 4 °C to prevent endocytosis. Endocytosis, defined by the internalization-induced (background-corrected) pHrodo Red signal, was normalized to the cell-surface expression of each antigen at 4 °C to yield a relative endocytosis ratio (RER; Fig. 4a).

Fig. 4: CD22 limits integrin β7 endocytosis in B cells via Shp1 and ligand recognition.
figure 4

a, Splenocyte staining with anti-CD19 and anti-IgD (to identify B cells) and pHrodo Red-conjugated tranferrin, HMβ1-1 (anti-β7) or FIB504 (anti-β7) antibodies at either 4 °C (no endocytosis) or 37 °C (endocytosis) was used to calculate endocytosis levels (that is, the MFI of pHrodo Red staining at 37 °C minus the MFI of pHrodo Red staining at 4 °C). Staining at 4 °C with phycoerythrin (PE)-conjugated RI7217 (anti-transferrin receptor 1, (TfR1)), HMβ1-1, FIB504 or the matching phycoerythrin-conjugated isotype control antibodies was used to calculate extracellular TfR1, β1 and β7 levels (that is, the MFI of the antigen-specific phycoerythrin staining at 4 °C minus the MFI of isotype control-phycoerythrin staining at 4 °C). For each molecule (illustrated with β7 in a) and each experiment, the RER was calculated by normalizing endocytosis levels to extracellular levels. b, RER of transferrin, β1 and β7 in WT, St6Gal1–/–, Ptpn6+/meV and Cd22–/– B cells with (+PQ) or without (no Tx) 100 µM primaquine treatment. The mean RER of the WT B cells group without primaquine was set to 1 and all data were normalized to that mean value. Shown are pooled data (means ± s.e.m.) from n = 3 independent experiments, with 6 animals per group in total. All groups were compared by two-way ANOVA with Šidák’s multiple comparisons test. *P ≤ 0.05; **P ≤ 0.01; ****P ≤ 0.0001.

Full size image

CD22-deficient B cells displayed a significantly higher RER for integrin β7 compared with wild-type B cells (1.65-fold higher; P < 0.0001). CD22 deficiency did not alter the basal endocytosis of transferrin or integrin β1 (Fig. 4b), and endocytosis was not observed for the isotype control antibodies (data not shown). Ptpn6+/meV B cells showed a shift similar to that of Cd22–/– cells, but not reaching significance. Treatment of cells with primaquine, to inhibit recycling of proteins back to the plasma membrane22, increased internalization. It also increased the differences between the RERs of the following versus wild-type B cells (RER = 1.18): Ptpn6+/meV (RER = 1.54; P < 0.05); St6Gal1–/– (RER = 1.59; P < 0.01); and Cd22–/– (RER = 2.84; P < 0.0001) (Fig. 4b). These results suggest that loss of CD22 or reduction of Shp1 activity (Ptpn6+/meV) similarly and selectively inhibit β7 endocytosis in B cells.

CD22–Shp1 restrains plasma membrane β7 phosphorylation

Integrin β tails, including that of β7, comprise conserved motifs for tyrosine phosphorylation by Src family kinases, regulating integrin functions23,24,25. These motifs interact with a large number of phosphotyrosine-binding (PTB) domain-containing proteins, including proteins that regulate endocytic trafficking25,26,27,28. To determine whether CD22 restrains phosphorylation of β7 at the cell surface, we compared tyrosine phosphorylation of cell-surface versus intracellular β7 in CD22-deficient versus wild-type B cells. We biotinylated cell-surface proteins by incubation of cells with Sulfo-NHS-Biotin at 4 °C to prevent protein internalization (Fig. 5a). We isolated cell-surface β7 from purified B or T cell lysate using successive immunoprecipitations for β7 with the β7-specific antibody FIB504 and then for biotin with streptavidin-conjugated beads (SA-IP). We recovered cell-surface β7 from the eluate of the SA-IP and intracellular β7 from the biotin-free flowthrough before quantification of β7 and phosphotyrosine (p-Tyr) by immunoblot (Fig. 5a). We found a significant increase of p-Tyr (normalized to β7) in the cell-surface β7 fraction of Ptpn6+/meV and Cd22–/– B cells compared with wild-type controls, whereas phosphorylation of intracellular β7 did not differ in mutant versus wild-type cells (Fig. 5b). The p-Tyr:β7 ratio among the cell-surface β7 fraction of Ptpn6+/meV and Cd22–/– B cells reached levels similar to the intracellular pool. In contrast with the effects of CD22 deficiency on surface β7 phosphorylation in B cells, we found no difference in cell-surface p-Tyr:β7 ratios in Cd22–/– T cells compared with wild-type T cells (Extended Data Fig. 4). In conjunction with the known role of phosphorylation in endocytosis, these results suggest that tyrosine phosphorylation of β7 at the cell surface functions as a switch to enhance β7 endocytosis and that the CD22–Shp1 axis shifts the balance of p-Tyr towards β7 dephosphorylation to maintain the integrin at the cell surface.

Fig. 5: CD22 restrains tyrosine phosphorylation of cell-surface β7 integrin in B cells.
figure 5

a, Schematic of the double immunoprecipitation (IP) experiment. Cell-surface proteins of purified naive B cells were biotinylated with Sulfo-NHS-Biotin at 4 °C for 1 h. Biotinylation was confirmed by flow cytometry using a streptavidin-conjugated probe. Biotinylated B cell lysate was used for immunoprecipitation of β7 integrin (including biotinylated cell-surface β7 and biotin-free intracellular β7). Following elution for β7, a second immunoprecipitation with streptavidin beads was used to recover cell-surface β7 (eluate of the SA-IP) from intracellular β7 (flowthrough of the SA-IP) before immunoblot studies. b, Left, detection of β7 and p-Tyr levels in cell-surface and intracellular β7 fractions (130 kDa) of WT, CD22-deficient (Cd22–/–) and Ptpn6+/meV B cells after the double immunoprecipitation shown in a. Right, quantification of p-Tyr levels normalized to β7 levels (p-Tyr:β7 ratio). Within each experiment, the p-Tyr:β7 of the cell-surface β7 of the WT group was set to 100, and the data are expressed as a percentage of this total. Shown are pooled data (means ± s.e.m.) from n = 2–3 independent experiments. Each dot represents one independent experiment with n = 8 pooled animals for WT and Cd22–/– (that is, n = 24 animals in total for n = 3 independent experimental replicates) and n = 4 pooled animals for Ptpn6+/meV (that is, n = 8 animals in total for n = 2 experimental replicates). The data were analyzed by two-tailed Student’s t-test. **P ≤ 0.01.

Source data

Full size image

CD22–Shp1 enhances B cell homing to GALT

To assess the consequence of CD22–Shp1-dependent α4β7 regulation in GALT, we first assessed the lymphocyte composition of Cd22–/– and wild-type PPs. PPs from mutant and wild-type animals were similar in number and size (Fig. 6a), but mutant PPs had a significant decrease (~70%) in naive B cell numbers compared with wild-type PPs (Fig. 6b). There was no difference in germinal center or activated B cells (CD19+IgD–) or T cell numbers (Fig. 6b and Extended Data Fig. 5). We found no difference in the PLNs, taken as controls, showing the specificity of the effects for GALT (Fig. 6b).

Fig. 6: Functional assays reveal defective PP homing and altered endothelial interactions of Ptpn6+/meV and Cd22–/– B cells.
figure 6

a, Numbers of PPs in WT and Cd22–/– B cells. Shown are pooled data (means ± s.e.m.) from n = 2 experiments with 8 mice per group in total. Inset: representative images of WT and Cd22–/– PPs, indicated by arrowheads. Scale bars, 2 mm. b, Numbers of germinal center/activated B cells (GC/A; CD19+IgD–) and naive B cells (CD19+IgD+) in MLNs, PLNs and PPs of WT and Cd22–/– animals, shown as a percentage of the mean of the WT group. Shown are pooled data (means ± s.e.m.) of n = 2 experiments with 11–12 mice per group in total. c, Schematic of the short-term homing assay. i.v., intravenous. d,e, Left, localization of WT, Cd22–/– and Ptpn6+/meV B cells (d) or WT, Cd22–/–, CD22Y2,5,6F and CD22R130E B cells (e) in PLNs, MLNs and PPs after homing assays. The data are shown as a percentage of the mean localization ratio of the WT group. Shown are pooled data (means ± s.e.m.) of n = 3–5 experiments, with 11–16 mice per group in total. Right, representative dot plots gated on naive B cells. f, Schematic of the in situ video microscopy experiment. Representative video stills are shown of WT B cells (green) and Ptpn6+/meV cells (red). Scale bar, 100 μm. g,j, Numbers of WT versus Ptpn6+/meV (g) or Cd22–/– (j) arresters on PP HEVs were counted per second from the first cell entering the HEVs. The total number of WT B cell arresters at the end of each experiment (that is, the WT maximum; ~40–80 cells on average) was set to 100 and the data are expressed as a percentage of this total. h,k, The behavior (that is, flyer, brief roller or roller; as defined in Extended Data Fig. 8) of WT versus Ptpn6+/meV (h) or Cd22–/– cells (k) entering the HEVs was analyzed in 3–4 representative PP HEVs per experiment. The results are shown as a percentage of the total number of cells entering HEVs (~250–300 cells were analyzed per group). The data in g, h, j and k represent means ± s.e.m. of three independent experiments. i,l, Average rolling velocities (means ± s.e.m.) of representative Ptpn6+/meV (i) or Cd22–/– roller cells (l) versus WT cells from n = 3 experiments. Groups were compared by two-tailed Student’s t-test (a, b, h, i, k and l), one-way ANOVA with Dunnett’s multiple comparisons test (d and e) or paired two-tailed Student’s t-test (g and j). *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001.

Full size image

Next, we assessed the short-term homing of mutant B cells into PPs and mesenteric lymph nodes (MLNs), which express the dedicated α4β7 ligand MAdCAM-1. Recipient wild-type mice received splenocytes from wild-type and mutant donors labeled with different CellTracker dyes and mixed at a 1:1:1 ratio, as confirmed by flow cytometry. Two hours later, we established the cells localized to PPs and MLNs by flow cytometry (Fig. 6c). In agreement with the reduction of α4β7, Ptpn6+/meV B cells, Cd22–/– B cells, CD22Y2,5,6F ITIM mutant B cells and CD22R130E lectin mutant B cells homed poorly to PPs, displaying a similar ~60% reduction in recruitment compared with wild-type B cells (Fig. 6d,e). We observed an intermediate defect for all mutant B cells in homing to the MLN that was consistent with MAdCAM-1 expression on subsets of MLN HEVs, while homing to the PLNs, spleen and bone marrow, which is independent of α4β7, was normal (Fig. 6d,e and Extended Data Fig. 6). CD22 deficiency did not affect CD4+ T cell homing to PPs, while reduced Shp1 activity (Ptpn6+/meV) compromised CD4+ T cell homing, consistent with reduced cell-surface β7 expression. Ptpn6+/meV CD4+ T cells also localized poorly to PLNs (Extended Data Fig. 7). This may reflect the role of Shp1 in modulating β2 integrin activity in T cells11.

α4β7 mediates activation-independent tethering and rolling on PP, as well as chemokine/integrin activation-dependent arrest in combination with lymphocyte function-associated antigen 1 (ref. 2). To assess the effects of the CD22–Shp1 axis on these steps, we visualized B and T cell behavior in PPs by intravital microscopy. We purified B or T cells from wild-type, Ptpn6+/meV and Cd22–/– donors, labeled them with different CellTracker dyes, mixed mutant and wild-type cells at a 1:1 ratio (as confirmed by flow cytometry) and transferred the cells into a recipient wild-type mouse while imaging one PP (Fig. 6f). We captured high-frame-rate videos (40 frames per second) to study the behavior of all cells entering HEVs within the field of view during the recording (~30–40 s). We stratified the cells into four groups (that is, flyers, brief rollers, rollers and arresters) based on their interactions with HEVs. Freely flowing cells that failed to interact detectably, termed flyers, appeared as streaks due to their high velocity during the time of exposure, and passed through the HEVs at a mean velocity of ~1,000 µm s−1 (Extended Data Fig. 8a–c and Supplementary Video 1). Cell capture on the vessel wall could be visualized as soon as the cell appeared round and bright as a result of reduced velocity. Cells attaching briefly (<1 s) before detaching and flying through the HEV were described as brief rollers (Extended Data Fig. 8b,c and Supplementary Video 2). Cells interacting for more than 1 s were called rollers (Extended Data Fig. 8b,c and Supplementary Video 3). Among rollers, cells static for more than 2 s at the end of the recording were considered arresters.

At the end of ~30 s of observation, the numbers of Ptpn6+/meV and Cd22–/– arrested B cells were ~50 and ~60% lower, respectively (P < 0.0001), than wild-type B cells (Fig. 6g,j and Supplementary Videos 4 and 5). The reduced frequency of arrest correlated with inefficient initial capture or tethering, as well as faster rolling (looser interaction) of Ptpn6+/meV and Cd22–/– B cells. We observed a reduced frequency of rolling by Ptpn6+/meV and Cd22–/– B cells compared with wild-type control cells, with a corresponding increase in non-interacting flyers (Fig. 6h,k). Among rollers, the average rolling velocity of Ptpn6+/meV and Cd22–/– B cells was ~2–3 times higher than wild-type controls (Fig. 6i,l and Supplementary Videos 6 and 7). The total number of mutant cells and wild-type B cells entering PP HEVs (including rollers, brief rollers or flyers) was similar in each experiment, thus ruling out distortion of the results by differences in the retention of mutant cells in other organs (Extended Data Fig. 8d). The CD22 effect could be due to a lack of interaction with the HEV-expressed CD22 ligand: however, the parallel alterations in Ptpn6+/meV B cell behavior suggest that the cell-intrinsic reduction of α4β7, which Ptpn6+/meV and CD22-deficient B cells have in common, is probably the major factor involved in reduced interaction efficiency and PP homing. Consistent with this hypothesis, the effects of CD22 deficiency and reduced Shp1 activity (Ptpn6+/meV) on both tethering and rolling velocity were phenocopied by experimental reduction in the surface α4β7 available for interaction (Extended Data Fig. 9). Moreover, reduced α4β7 on Cd22–/– B cells significantly impacted homing to the PP, even in absence of the CD22 ligand on PP HEVs (Extended Data Fig. 10).

We conclude that Shp1-driven CD22 regulation of β7 drives enhanced B cell homing to GALT and substantially augments the adhesive functions of the mucosal lymphocyte integrin α4β7.

CD22 deficiency inhibits intestinal antibody responses

In addition to its roles in gut lymphocyte homing, α4β7 mediates cell–cell interactions of lymphocytes with MAdCAM-1 expressed by follicular dendritic cells in GALT29. It also supports interactions of α4β7-expressing lymphocytes with other ligands, including α4 on adjacent immune cells, fibronectin in the extracellular matrix, and vascular cell adhesion molecule 1 (refs. 1,30). Moreover, α4β7 is highly expressed by B cells involved in gut immune responses, but is downregulated on B cells responding to systemic (non-intestinal) antigens31,32,33. Thus, we reasoned that CD22-dependent α4β7 surface expression might have a selective role in intestinal (versus systemic) antibody responses. In contrast, CD22 suppresses B cell antigen receptor responses in general, and its deficiency has been considered to enhance B cell activation.

To distinguish between these possibilities, we immunized wild-type and CD22-deficient mice with cholera toxin B (CTB) by different routes and compared the humoral responses 2 weeks later. CD22 deficiency caused a clear reduction in antigen-specific IgA and IgG responses to oral CTB, but not to nasal or intramuscular immunization (Fig. 7a,c). Total IgA and IgG concentrations in the serum after oral, nasal and intramuscular immunizations were similar in CD22-deficient versus wild-type animals, ruling out global inhibition or activation of B cell responses (Fig. 7b,d).

Fig. 7: Defects in intestinal responses to oral antigen in CD22-deficient mice.
figure 7

Cohorts of WT or Cd22−/− mice were immunized with CTB via the oral, intranasal (nasal), or intramuscular (muscular) route for 2 weeks. a–d, Serum levels of CTB-specific IgA (a), total IgA (b), CTB-specific IgG (c) and total IgG (d) as measured by ELISA and expressed as the net OD450 or a percentage of the mean OD450 measured in the lowest dilution factor of the WT group. e,f, Small intestine segments were cultured ex vivo for 3 days to titer the quantity of secreted CTB-specific IgA, total IgA and CTB-specific IgA normalized to total IgA (e) or CTB-specific IgG, total IgG and CTB-specific IgG normalized to total IgG (f) produced by ELISA and expressed as a percentage of the WT group mean. Shown in a–f are pooled data (means ± s.e.m.) from n = 2 experiments, with 5–6 mice per group in total. Groups were compared by two-way ANOVA with Šidák’s multiple comparisons test (a–d) or two-tailed Student’s t-test (e and f). *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001.

Full size image

We also analyzed CTB-specific IgA and IgG production from ex vivo cultures of small intestine segments. Intestinal CTB-specific IgA (Fig. 7e) and IgG (Fig. 7f) were decreased in Cd22–/– mice after oral immunization, and CTB-specific IgA was decreased after intramuscular immunization, but there was no significant alteration in small intestine IgA or IgG after intranasal immunization (Fig. 7e,f). These results suggest that the CD22-driven augmentation of α4β7 surface expression selectively amplifies gut-specific, but not systemic, antibody responses.

CD22 improves the antibody response to rotavirus (RV)

RV selectively infects intestinal epithelial cells, leading to strong immune responses in GALT, and the local production of RV-specific IgA plays an important role in protection34. Young pups, which are more susceptible to infection than adults, were infected with RV. We measured total and RV-specific IgA and IgG in the feces daily. Ten days post-infection, wild-type animals started to mount robust RV-specific intestinal IgA and IgG responses (Fig. 8a,b), which preceded a reduction of fecal viral shedding by day 12 post-infection and a full resolution of RV infection by day 15 post-infection (Fig. 8c). In contrast, CD22-deficient animals had a dramatically reduced RV-specific IgA response in the feces (Fig. 8a) and a delay in the resolution, as indicated by residual viral shedding at day 12 post-infection (Fig. 8c). Virus was eliminated by day 15 post-infection in both wild-type and Cd22–/– mice (Fig. 8c).

Fig. 8: Delayed protective immune response to RV infection in CD22-deficient animals.
figure 8

a,b, Five-day-old WT or Cd22−/− pups were orally gavaged with RV strain EW. The production of RV-specific IgA, total IgA and RV IgA normalized to total IgA (a) or RV-specific IgG, total IgG and RV IgG normalized to total IgG (b) in fecal samples was measured by ELISA up to 12 d post-infection (p.i). The mean of the OD450 for the WT group at day 12 post-infection was set to 100 and the data are expressed as a percentage of this total. c, Fecal RV antigen shedding, as measured by ELISA and expressed as the net OD450, normalized to the sample weight. In a–c, pooled data (means ± s.e.m.) from n = 4 independent experiments are shown, with 10–15 animals per group in total. d–h, In separate experiments, 5-day-old WT and Cd22−/− pups were orally inoculated with RV strain EW and sacrificed at day 12 post-infection. d,e, Serum levels of RV-specific IgA and total IgA (d) and RV-specific IgG and total IgG (e) as measured by ELISA. The mean of the maximum OD450 for the WT group was set to 100 and the data are expressed as a percentage of this total. f, Numbers of RV-specific IgA ASCs measured in the spleen, MLN and small intestine lamina propria (SI LP) by ELISPOT. Shown are the numbers of RV IgA ASCs per 106 total cells, expressed as a percentage of the WT group mean. g,h, Ex vivo small intestine fragment cultures. The data are displayed and presented as in Fig. 7e,f to show RV IgA, total IgA and ratios of RV IgA to total IgA (g) and RV IgG, total IgG and ratios of RV IgG to total IgG (h), as measured by ELISA. Shown in d–h are pooled data (means ± s.e.m.) from n = 2 experiments, with 10–11 mice per group in total. Groups were compared by two-way ANOVA with Šidák’s multiple comparisons test (a, b, d and e) or unpaired two-tailed Student’s t-test (c and f–h). **P ≤ 0.01; ***P ≤ 0.001.

Full size image

At the peak of the fecal antibody response (day 12 post-infection), we observed reduced RV-specific IgA titers in the serum of CD22-deficient animals (Fig. 8d), while RV-specific IgG, total IgG and total IgA titers were similar to those in wild-type animals (Fig. 8d,e). Cd22–/– animals also had fewer RV-specific IgA antibody-secreting B cells in MLNs and the small intestine lamina propria than wild-type mice at day 12 post-infection, while there were no differences in the spleen (Fig. 8f). To determine whether the reduction of RV-specific fecal antibody in Cd22–/– mice reflected a deficit in local production in the gut (as opposed to altered antibody secretion in bile), we analyzed RV-specific IgA and IgG levels in ex vivo culture of small intestine segments. We found a twofold decrease in the RV-specific IgA and IgG titers in CD22-deficient animals compared with wild-type controls (Figs. 8g and 8h), which correlated with higher viral shedding in the feces at day 12 post-infection, as mentioned above (Fig. 8c). Together, these data highlight a substantial role of B cell CD22 in the mucosal pathogen response.

Discussion

We report a previously unappreciated contribution for the CD22–Shp1 complex in cell-autonomous regulation of the gut integrin receptor α4β7. We show that Shp1 inhibits α4β7 endocytosis and maintains cell-surface α4β7 expression. We show that B cell-specific Shp1-binding lectin CD22 associates with β7 on the cell surface, specifically targeting Shp1 to β7 and reducing β7 tyrosine phosphorylation. Our results support a model in which this cascade inhibits α4β7 endocytosis and degradation, resulting in higher surface expression of the gut-homing integrin. We present evidence that this novel mechanistic cascade is functionally important not only in B cell homing to GALT but also in intestinal antibody responses. These results define a novel pathway for differential control of integrin function and an unexpected and selective role for a CD22–Shp1 axis in intestinal immunity.

β integrin phosphorylation controls integrin functions including activation23,25,35 and endocytosis28,36,37 by means of PTB domain-containing proteins that interact with integrin β tails such as talin38, kindlin38, Dok-1 (refs. 23,24,25) or phox homology band 4.1/ezrin/radixin/moesin (PX-FERM) domain-containing sorting nexins39,40. Our results suggest that Shp1 acts as a phosphorylation switch that directly or indirectly restrains tyrosine phosphorylation of integrin β7, thus reducing β7 endocytosis. On B cells, CD22 targeting of Shp1 to β7 may also alter the phosphorylation state of other PTB domain- or FERM domain-containing targets in the vicinity of the integrin, potentially affecting endocytic adaptors37 to inhibit endocytosis as well.

β7 cell-surface levels in Cd22–/– B cells are comparable to those on wild-type naive T cells, suggesting that CD22 itself is enough to explain the higher surface expression of α4β7 by normal naive B versus T cells6. It is intriguing to speculate that alternative ITIM-containing proteins may regulate T cell β7, perhaps contributing to the enhanced β7 display by intestinal memory T cells.

On B cells, CD22 can associate in cis with α2-6-Sia-modified membrane glycoproteins, including CD22 itself and CD45. These cis interactions can bring ITIM-bound Shp1 phosphatase into close proximity to target substrates, altering the phosphorylation state and function of associated molecules12. Using PLAs, we showed that β7 but not β1 integrin associates closely with CD22, a specific interaction that is dependent on cell-surface sialic acid and that probably contributes to the magnitude and specificity of the effects. Yet, the reduction of β7 in lectin- (CD22R130E) and St6gal1-deficient cells was less than that in CD22Y2,5,6F cells with mutated CD22 ITIMs. In CD22R130E B cells, CD22 itself is known to be more phosphorylated, to recruit more Shp1 and to have higher lateral mobility than in wild-type B cells16,19. These effects would be expected to allow CD22–Shp1 to access β7 even in the absence of specific targeting, explaining the partial retention of surface levels in the absence of lectin activity (CD22R130E) or CD22 ligand (St6gal1–/–), as well as the intermediate effect on endocytosis seen with St6Gal1–/– B cells. Taken together, these data suggest that CD22−α2-6-Sia interactions recruit CD22−ITIM-docked Shp1 phosphatase to β7 to alter the local tyrosine phosphorylation balance, inhibit β7 endocytosis and enhance α4β7 surface expression.

B cell positioning, activation and cell−cell interactions in the GALT, which all involve contributions from α4β7, drive the production of local and systemic antibodies to intestinal antigens and pathogens. Although CD22-deficient mice show grossly normal systemic B cell development41, we observed a profound defect in the intestinal and systemic antigen-specific IgA and IgG levels in response to oral immunization with CTB in CD22-deficient animals, whereas the IgG response to nasal or intramuscular immunization was preserved. These results closely recapitulate the reported effects of β7 deficiency or α4β7 blockade, which also selectively impair antibody responses to oral versus systemic antigen42. Taking into account that CD22 is known to restrain B cell responses and that its deficiency may enhance them in vivo43, the selective reduction of mucosal immune responses in CD22-deficient animals highlights the critical and dominant importance of the CD22-dependent β7 regulation for optimum mucosal immunity. Accordingly, we found that CD22 deficiency impaired the response of mice to infection with RV, a reovirus that selectively infects small intestinal epithelial cells and is a major cause of childhood diarrheal illness44. CD22 deficiency caused a significant reduction in viral clearance comparable to that reported in mice with a complete lack of B cells45, or lacking IgA46 or GALT47 itself. The protective effect of CD22 deficiency probably reflects global influences of CD22-dependent α4β7 functions in the intestinal immune response.

We previously showed that PP and MLN HEVs, but not PLN HEVs, express functional carbohydrate ligands for CD22 encoded by St6Gal1. St6Gal1–/– PP HEVs are defective at recruiting CD22-expressing B cells in short-term homing assays48, probably as a result of trans interactions between CD22 and α2-6-Sia-decorated glycans on HEVs. Yet, we showed here that even in the absence of HEV CD22 ligands, the reduced α4β7 on Cd22–/– B cells inhibits homing to the PP. In contrast with the effects of a defective CD22–Shp1 axis, which inhibits tethering and speeds up rolling, intravital microscopy in St6Gal1–/– mice revealed that wild-type B cells tether and roll normally on St6Gal1–/– PP HEVs: only definitive arrest is compromised (unpublished observations). We hypothesize that CD22 ligation to α2-6-Sia-modified MAdCAM-1 creates a synapse between CD22, MAdCAM-1 and α4β7, facilitating chemokine-driven activation of α4β7 and arrest of the B cell. An analogous mechanism controls the arrest of neutrophils: heterophilic ligation of endothelial-cell-expressed CD99 to paired immunoglobulin-like receptors enhances the chemokine-induced β2-dependent arrest of neutrophils on intercellular adhesion molecule 1 (ref. 49). Future studies will probably shed light on this complex mechanism. It is fascinating that CD22 has evolved these two separate functions: (1) selective regulation of surface β7 expression; and (2) trans-ligation to PP HEV heterophilic ligands, to serve the same goal of enhancing B cell homing to the GALT.

Methods

Reagents

All of the antibodies used in these studies are listed in Supplementary Table 1 and the Nature Research Reporting Summary linked to this article.

Mice

C57BL/6 WT, BALB/c WT, Cd22–/– (ref. 41) and St6gal1–/– mice20 were bred and housed in the animal facilities of the Veterans Affairs Palo Alto Health Care System, accredited by the Association for Assessment and Accreditation of Laboratory Animal Care. CD22Y2,5,6F (ref. 16) and CD22R130E mice16 were bred and housed at the Department of Biology of the University of Erlangen. Ptpn6meV/meV and Ptpn6+/meV mice8 were bred and housed at the University of California, San Francisco. All animals were housed under standard conditions, maintained in a 12 h/12 h light/dark cycle at 22 ± 2 °C and given food and tap water ad libitum. Unless otherwise stated, 6- to 12-week-old male and female mice were used in all of the experiments. All animal work and housing conditions were approved by the Institutional Animal Care and Use Committee at the Veterans Affairs Palo Alto Health Care System, or by the relevant animal care committees at the University of Erlangen.

Integrin expression and lymphocyte cellularity by flow cytometry

Lymphocytes from the blood, PLNs (inguinal, axillary and brachial lymph nodes), MLNs, bone marrow and PPs (average of five PPs per mouse) of WT and mutant mice were isolated and stained with fluorescently labeled antibodies against CD4, CD3ε, CD19, IgD and either αL, β2, α4, β1, β7, α4β7 or the appropriate isotype control. The Fc receptors were blocked using rat serum and anti-CD16/32 before staining. Dead cells were excluded by staining with 4′,6-diamidino-2-phenylindole (DAPI) or the LIVE/DEAD Fixable Aqua dye (Invitrogen). For the integrin expression studies, the MFI of the isotype control was subtracted from the MFI of the integrin staining. The integrin staining MFI (background corrected) was expressed as a percentage of the WT C57BL/6 (control group) mean MFI. For the cellularity studies, we added CountBright counting beads (Invitrogen) to each sample to calculate the absolute cell number per organ. Samples were acquired on a Fortessa flow cytometer (BD Biosciences) using FACSDiva Software (BD Biosciences; version 8.0.1), and were analyzed using FlowJo (BD Biosciences; 10.3). A representative gating strategy is shown in Supplementary Fig. 1.

In situ video microscopy analyses of lymphocyte interactions with PP HEVs

Naive B cells or CD4+ T cells were isolated from WT and Cd22–/– or Ptpn6+/meV splenocytes using negative selection kits (STEMCELL Technologies). For each experiment, the quality of the isolation was checked by flow cytometry and consisted of ≥98% cells of interest. Lymphocytes were labeled with 2.5 μM CMFDA (Invitrogen) or CMTPX dye (Invitrogen) for 10 min at 37 °C in RPMI without fetal bovine serum (FBS), then washed with RPMI containing 10% FBS. In some experiments, labeled B or CD4+ T cells were incubated with the anti-mouse α4β7 antibody DATK32 (50 μg ml−1)50 in RPMI containing 10% FBS for 45 min at 37 °C. Cells were washed three times to remove unbound antibodies. Before injection into recipient mice, cells were counted by flow cytometry using counting beads. Recipient mice were anesthetized via intraperitoneal injection of ketamine and xylazine. One individual PP of the small intestine was exteriorized and positioned for epifluorescence microscopy and video recording. The same numbers (5–10 × 106 cells) of WT, Cd22–/–, Ptpn6+/meV B or CD4+ T cells were transferred into anesthetized recipients. The interactions of fluorescent cells with PP HEVs were recorded for 30–40 s at 40 frames per second of 25 ms exposure time. All fluorescent cells entering HEVs during the recording time were analyzed on a frame-to-frame basis for the entire duration of the video. Cells passing HEVs for <1 s with no interaction, or with a velocity >300 μm s−1, were considered non-interactive and called flyers. Cells that started binding to HEVs briefly for <1 s before getting released were considered non-interactive and called brief rollers. Cells binding to HEVs for >1 s were considered rollers. At the end of the recording, rollers attached to HEVs with a static binding of >2 s were considered arresters. To assess the mean rolling velocity of each cell, we tracked each cell manually using Imaris (version 9) to define the precise rolling distance from the frame of its first interaction (ffirst) to the first frame of its definitive static binding (flast). The rolling time was (flast – ffirst + 1) × 25 ms.

Short-term homing assay

Donor splenocytes were isolated and labeled with CellTrace Violet (Invitrogen) or CFSE (Invitrogen) in complete RPMI medium, or a combination of CellTrace Violet and CFSE at concentrations optimized for a bar-coding system with two to four donor populations in total. Equal numbers (25–50 × 106 cells) of donor cells were transferred into WT recipient mice by injection into the tail vein. After 1.5 h, lymphocytes from the PLNs and PPs (average of five PPs per mouse) of recipient mice were isolated and stained for flow cytometry to identify T and B cells (see ‘Integrin expression and lymphocyte cellularity by flow cytometry’). The LIVE/DEAD Fixable Aqua dye was used for live/dead staining, and counting beads were added to calculate absolute cell numbers. The efficiency of B and T cell homing to each organ was calculated as a ratio: the number of cells found in each organ was divided by the number of cells injected (input). Then, the results were presented as a percentage of the WT C57BL/6 (control group) mean ratio.

PLA

B cells were isolated from WT and Cd22−/− splenocytes using negative selection kits. In some experiments, B cells were pre-treated with Arthrobacter uereafaciens sialidase (Roche) at 125 mU ml−1 to remove α2-6-Sia modifications. Vehicle-control-treated (phosphate-buffered saline (PBS)) or sialidase-treated purified B cells were cytospun on to a 1 cm2 area of a slide at a concentration of 1.5 × 105 per 100 μl for 7 min at 700 r.p.m. in a Cyto-Tek centrifuge (model number 4324). The cell sections were fixed using 4% paraformaldehyde for 5 min at 4 °C, followed by washing with PBS three times. All subsequent incubations were carried out in a humidity chamber at 37 °C following the Duolink PLA kit (Sigma–Aldrich) instructions. The following combinations of primary antibodies were used: mouse anti-mouse CD22 (Cy34.1) and goat anti-mouse β7 or goat anti-mouse β1; and goat anti-mouse β1 and rat anti-mouse α4 (PS/2) followed by In Situ PLA Probe Anti-Mouse MINUS and Anti-Goat PLUS (Sigma–Aldrich). At the end of the PLA procedure, the sections were stained with DAPI for the identification of nuclei. Images were captured on a Zeiss LSM 880 confocal microscope using the Zeiss ZEN software, a 63× oil-immersion objective and a 1.5× digital zoom. The number of PLA spots per image were quantified using Imaris (version 2.0.0-rc-49/1.51d).

Cellular localization of integrin β7, CD22 and Shp1 by microscopy

B cells were isolated from WT and Cd22–/– splenocytes using negative selection kits. In some experiments, cells were stained with MitoTracker Deep Red FM (Invitrogen) at 0.5 μM for 15 min at 37 °C to identify the cytoplasm. Cells were fixed using 1% paraformaldehyde (Invitrogen) in PBS for 5 min at 4 °C, then stained for extracellular β7 and CD22 with goat anti-mouse β7 and biotin mouse anti-mouse CD22, respectively, in PBS for 40 min at 4 °C, followed by Alexa Fluor 488-conjugated donkey anti-goat (for β7) and Alexa Fluor 594-conjugated streptavidin (for CD22) for 40 min at 4 °C. In some experiments, cells were permeabilized in 1× permeabilization buffer from Invitrogen for 30 min at 4 °C after fixation, stained for Shp1, β7 and CD22 with rabbit anti-human/mouse Shp1, goat anti-mouse β7 and biotin mouse anti-mouse CD22, respectively, in 1× permeabilization buffer for 60 min at 4 °C, washed with fresh 1× permeabilization buffer, then stained with Alexa Fluor 594-conjugated donkey anti-rabbit (for Shp1), Alexa Fluor 488 donkey anti-goat (for β7) and Alexa Fluor 647-conjugated streptavidin (for CD22). At the end of the staining procedure, cells were cytospun and stained with DAPI for the identification of nuclei, and images were captured as described above (see ‘PLA’). For each cell, we quantified the intensity of the β7 fluorescence with ImageJ using the corrected total cell fluorescence. We counted manually the total number of β7 spots at the cell surface found in close proximity with intracellular spots of Shp1 and expressed this number as a percentage of the total β7 spots to yield a Shp1–β7 proximity index (%). Similarly, we counted the total number of CD22 spots at the cell surface found in close proximity to both β7 and Shp1 and expressed this number as a percentage of the total β7/Shp1 hotspots to yield a CD22–Shp1/β7 proximity index (%).

Endocytosis assay

Splenocytes were stained with fluorescently labeled antibodies against CD19, IgD and either CD71 TfR1, β1, β7 or the appropriate isotype control. The Fc receptors were blocked with anti-CD16/32 before the staining. For each sample, staining at 4 °C with phycoerythrin-conjugated RI7217 (anti-TfR1), HMβ1-1 (anti-β1), FIB504 (anti-β7) or the matching phycoerythrin-conjugated isotype control antibodies allowed calculation for extracellular amounts of TfR1, β1, αL and β7 concentrations. We used pHrodo iFL microscale protein labeling kits (Invitrogen) to label the antibody clones HMβ1-1, FIB504 or isotype-matched controls. In parallel, staining of each sample with pHrodo Red-conjugated transferrin (Invitrogen), HMβ1-1, FIB504 or isotype control antibodies at either 4 °C (no endocytosis) or 37 °C (endocytosis) for 1 h was used to calculate the endocytosis efficiency. The same staining protocol was performed in the presence of 100 μM primaquine. Samples were analyzed by flow cytometry. For each antigen and each experiment, the RER was calculated by normalizing endocytosis levels to extracellular levels as follows:

$$\begin{array}{l}{\mathrm{RER}}\,\left({{\mathrm{of}}\,{\mathrm{antigen}}\,{\mathrm{A}}}\right) =\\\quad \frac{\begin{array}{c}{\mathrm{MFI}}\,{\mathrm{of}}\,{\mathrm{anti}}{\mbox{-}}{\mathrm{A}}\,{\mathrm{pHrodo}}\\{\mathrm{Red}}\,{\mathrm{staining}}\,{\mathrm{at}}\,37\,^\circ{\mathrm{C}}\end{array} - \begin{array}{c}{\mathrm{MFI}}\,{\mathrm{of}}\,{\mathrm{anti}}{\mbox{-}}{\mathrm{A}}\,{\mathrm{pHrodo}}\\{\mathrm{Red}}\,{\mathrm{staining}}\,{\mathrm{at}}\,4\,^\circ{\mathrm{C}}\end{array}}{\begin{array}{c}{\mathrm{MFI}}\,{\mathrm{of}}\,{\mathrm{anti}}{\mbox{-}}{\mathrm{A}}\,{\mathrm{phycoerythrin}}\\{\mathrm{staining}}\,{\mathrm{at}}\,4\,^\circ{\mathrm{C}}\end{array}-\begin{array}{c}{\mathrm{MFI}}\,{\mathrm{of}}\,{\mathrm{isotype}}\,{\mathrm{control}}\,{\mathrm{phycoerythrin}}\\{\mathrm{staining}}\,{\mathrm{at}}\,4\,^\circ{\mathrm{C}}\end{array}}\end{array}$$

Double immunoprecipitation and immunoblotting

Purified B or T cells were labeled with EZ-Link Sulfo-NHS-Biotin (Invitrogen) on ice for 60 min. Biotinylated cells were washed with Hanks’ balanced salt solution and lysed in immunoprecipitation buffer (50 mM Tris-HCl (pH 7.5; Sigma–Aldrich), 150 mM NaCl (Sigma–Aldrich), 1% Triton X-100 (Sigma–Aldrich), 0.1% sodium deoxycholate (Sigma–Aldrich), 1 mM EDTA (Invitrogen), cOmplete, Mini, EDTA-free Protease Inhibitor (Roche) and Phosphatase Inhibitor Cocktail 2 and 3 (Sigma–Aldrich) used at 1:100 dilution from stock) for 20 min on ice. Cell lysates were cleared from debris by centrifugation (16,000g; 10 min; 4 °C). After pre-clearing the lysates with unbound Dynabeads protein G beads (Invitrogen), β7 integrin was pulled down by incubation of lysates with Dynabeads protein G bound to rat anti-mouse β7 antibody FIB504 (ref. 51) at 4 °C overnight with agitation. After several washes with immunoprecipitation buffer, the β7 integrin was eluted with 100 mM glycine buffer at pH 2.5 for 15 min at 56 °C with gentle agitation. The acidic pH of the eluate was then quenched with two volumes of 100 mM Tris-HCl (pH 8.0). Biotinylated β7 was pulled down from these eluates with SoftLink Soft Release Avidin Resin (Promega) overnight at 4 °C with agitation. The flowthrough of the streptavidin immunoprecipitation was spun in with 30 kDa molecular weight cut-off ultrafiltration tubes to concentrate the biotin-free β7 fraction, then reduced at 95 °C for 5 min with 1× Laemmli buffer (Bio-Rad) containing 5% β-mercaptoethanol (Sigma–Aldrich). After overnight incubation, the avidin resin was washed three times with Tris-buffered saline (Sigma–Aldrich) containing 0.01% Tween 20 (Sigma–Aldrich), and biotinylated β7 was eluted with 1× Laemmli buffer containing 5% β-mercaptoethanol (95 °C; 5 min). Reduced biotin-free and biotinylated β7 were then subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis and immunoblot analysis. Integrin β7 was detected from immunoblots with rat anti-mouse β7 antibody (clone FIB504) plus horseradish peroxidase-conjugated goat anti-rat antibodies, followed by the addition of enhanced chemiluminescence substrate (Clarity Western ECL Substrate; Bio-Rad). Blots were stripped (Restore PLUS Western Blot Stripping Buffer; Invitrogen) and tyrosine phosphorylation was detected with the mouse anti-p-Tyr antibody (clone PY20) plus horseradish peroxidase-conjugated goat anti-mouse followed by addition of the substrate. Images were captured using the Azure Biosystems 600 imaging system. Imaged blots were quantified using ImageJ.

Real-time quantitative PCR with reverse transcription (RT-qPCR)

Total RNA was isolated from purified B cells using the RNeasy Mini Kit (Qiagen). Analysis of total RNA concentration and integrity was assessed using NanoDrop ND-1000 spectrophotometry. Equal amounts of each total RNA sample were converted into complementary DNA (cDNA) using the High-Capacity cDNA Reverse Transcription Kit (Invitrogen) according to the manufacturer’s instructions. RT-qPCR analysis of genes of interest was performed using SYBR Green on an ABI PRISM 7900HT (Applied Biosystems). PCR reaction mixtures consisted of 1 μl of a 1:5 dilution of template cDNA, 200 nM of each primer and 1× Power SYBR Green PCR Master Mix (Invitrogen) in a final volume of 10 μl. RT-qPCR amplication was conducted using an initial step of 5 min at 45 °C, then 5 min at 95 °C, followed by 40 cycles of denaturation at 95 °C for 30 s, primer annealing at 60 °C for 30 s and extension at 72 °C for 30 s. Melting curve analysis was used to assess the purity of the amplified bands. The sequences for the primer pairs used for Itgal52, Itgb1 (ref. 52), Itgb2 (ref. 52), Itgb7 (ref. 53) and Hprt54 are listed in Supplementary Table 2. Primer pairs for Actb were pre-designed from Qiagen. We used the 2–∆∆Ct method to calculate the relative gene mRNA expression of Itgal, Itgb1, Itgb2 and Itgb7 using Actb and Hprt as housekeeping genes.

CTB immunization studies

Adult C57BL/6 or Cd22−/− mice were immunized with 10 μg CTB (Sigma–Aldrich) diluted in 100 μl PBS for oral gavage (oral route), 10 μl PBS for intranasal administration via pipetting in the nostrils (nasal route) or 50 μl PBS for injection into the mouse hamstring (muscular route). Mice were sacrificed 2 weeks after the immunizations. Serum was collected for enzyme-linked immunosorbent assay (ELISA) studies, and segments of small intestine were used in ex vivo cultures.

RV infection studies

Five-day-old C57BL/6 or Cd22–/– pups were orally gavaged with 104 DD50 (that is, diarrhea dose 50%, the highest dilution of a virus stock that caused diarrhea in 50% of suckling mice) of WT murine RV (strain EW) diluted in M199 medium. Pups were then checked daily for diarrhea by gentle abdominal pressure. Fecal samples from each individual mouse were collected in PBS with calcium/magnesium for ELISAs. In some studies, mice were sacrificed at day 12 post-infection to collect serum for ELISA studies; MLNs, small intestine lamina propria and spleen for enzyme-linked immune absorbent spot (ELISPOT) analysis; and segments of small intestine for ex vivo cultures.

Detection of viral antigen and virus-specific IgA and IgG by ELISA

For the detection of viral antigen in fecal samples55, 96-well ELISA plates were coated with guinea pig anti-RV hyperimmune serum in PBS and incubated at 37 °C for 4 h. The plates were then blocked with 5% non-fat milk in PBS (Blotto) at 37 °C for 2 h. Suspended fecal samples were diluted 1:20 in 1% Blotto, added to the plates and incubated at 37 °C for 1 h. The plates were washed three times with PBS containing 0.05% Tween 20 (PBS-T). Rabbit anti-RV hyperimmune serum in 1% Blotto was added to the plates for 1 h at 37 °C and washed three times. Horseradish peroxidase-conjugated goat anti-rabbit IgG in 1% Blotto was added to the plates and incubated for 1 h at 37 °C. TMB substrate (Sigma–Aldrich) was added after four washes in PBS-T, the plates were developed for 10 min at room temperature, the reaction was stopped by the addition of 0.16 M sulfuric acid, and the optical density measured at a wavelength of 450 nm (OD450) was read with a plate reader. The data are presented as OD450-like numbers after normalization to stool weight.

For the detection of virus-specific fecal and serum IgA and IgG, plates were coated with rabbit anti-RV hyperimmune serum and blocked as described above, then incubated with 1:5 of RV stock (a gift from H. Greenberg at Stanford University) in 1% Blotto overnight at 4 °C. After three washes with PBS-T, stool samples (1:20 in Blotto) or serial dilutions of serum in 1% Blotto were added to the plates. After 2 h incubation at 37 °C, the plates were washed three times in PBS-T and peroxidase-conjugated anti-mouse IgA or IgG was added for 1 h at 37 °C. The plates were washed and developed as described above. For the detection of total fecal and serum IgA or IgG, the plates were coated with anti-mouse IgA, IgG or IgM polyclonal antibodies before using the same protocol as above.

Detection of CTB-specific IgA and IgG by ELISA

ELISA plates were coated with CTB (2 μg ml−1) overnight, blocked with PBS containing 5% bovine serum albumin (BSA), then incubated with serial dilutions of serum or ex vivo small intestine culture samples diluted in PBS containing 2% BSA for 2 h. After four washes with PBS-T, peroxidase-conjugated anti-mouse IgA or IgG in 2% BSA was added for 1 h at 37 °C. TMB substrate was added after four washes in PBS-T, the plates were developed for 5–10 min at room temperature, the reaction was stopped by the addition of 0.16 M sulfuric acid and the OD450 was read with a plate reader. For the detection of total IgA or IgG in serum and ex vivo small intestine culture samples, plates were coated with anti-mouse IgA, IgG or IgM polyclonal antibodies before using the same protocol as described above.

Small intestine fragment cultures

Open segments (1 cm long) of duodenums were weighted and placed in 24-well plates on an elevated mesh filter and cultured for 3 days at 37 °C in complete RPMI medium under hyper-oxygenated conditions56. At the end of the incubation period, intestinal IgA and IgG were measured by ELISA in the supernatants, as described above.

Quantitation of virus-specific antibody-secreting cells (ASCs) by ELISPOT assay

Multiscreen 96-well plates (MAIPS4510; Millipore) were coated with purified RV double-layer particles (a gift from H. Greenberg (Stanford University)) overnight at 4 °C, washed and blocked with RPMI medium containing 10% fetal calf serum for 1 h at 37 °C. Serial tenfold dilutions of spleen, PP or small intestine lamina propria cells were added to the plates and incubated overnight at 37 °C. After six washes, peroxidase-conjugated anti-mouse IgA or IgG was added for 1 h at 37 °C. The plates were washed six times and developed with AEC substrate (Vector Laboratories). To quantity non-specific ASCs, the plates were coated with anti-mouse IgA, IgG or IgM polyclonal antibodies before using the same protocol as above.

Statistical analyses

All data were organized in Microsoft Excel (version 16.42) spreadsheets and we performed all of our statistical analyses using GraphPad Prism (version 8.4.1). For each dataset shown and analyzed, the test used is indicated in the figure caption. Statistical significance is indicated by *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001 and ****P ≤ 0.0001.

Reporting Summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.