Defective Migration of CCR10-Deficient IgA⁺ Plasma Cells into Small and Large Intestines.

To understand how CCR10 is involved in regulating IgA responses in the intestine, we used a strain of CCR10-KO/EGFP-knock-in mice in which the coding region of enhanced GFP (EGFP) replaced most of the CCR10 coding region from its translation start site down, allowing us to use the knocked-in EGFP as a reporter for endogenous CCR10 expression (17). First, we determined effects of CCR10 KO on intestinal IgA production under normal specific pathogen-free conditions. In heterozygous CCR10-KO/EGFP-knock-in (CCR10^+/EGFP) mice, nearly all mature IgA⁺ plasma cells (B220⁻CD19⁻) of intestines were EGFP⁺ (Fig. 1A), consistent with the previous report that they expressed CCR10 (9). Intestinal IgA⁺ plasma cells of homozygous CCR10^EGFP/EGFP were also EGFP⁺, and their numbers were not different from those of WT (CCR10^+/+) or CCR10^+/EGFP mice (Fig. 1A). Mean fluorescence intensities of EGFP signals in the IgA⁺ cells of CCR10^EGFP/EGFP mice were two times those of CCR10^+/EGFP mice (Fig. 1A), suggesting that both alleles were coexpressed and the knocked-in EGFP could reliably report CCR10 expression. Levels of total IgA antibodies in feces were similar in CCR10^EGFP/EGFP, CCR10^+/EGFP, and WT mice (Fig. 1B), consistent with the normal numbers of intestinal IgA⁺ plasma cells and results of a recent report (16). The IgA⁺ plasma cells of CCR10^EGFP/EGFP and control mice also had the similar capacity to produce IgA in vitro (Fig. S1).

Because no other chemokine receptor could substitute for the CCR10-mediated migration of IgA⁺ cells toward its mucosal ligand CCL28 in vitro (16), it is unclear why the CCR10 KO did not have much effect, particularly in the large intestine, where the ligand (CCL25) for another major gut-homing molecule, CCR9, is not expressed (9). We then tested whether CCR10-deficient IgA plasma cells were defective in migration into intestines by using an in vivo migration assay in which similar numbers of EGFP⁺IgA⁺ cells of CCR10^+/EGFP and CCR10^EGFP/EGFP mice were cotransferred into WT mice. Notably, significantly fewer EGFP⁺IgA⁺ cells of the CCR10^EGFP/EGFP donor (CD45.1⁻CD45.2⁺) were recovered from large and small intestines of the recipients than those of the CCR10^+/EGFP donor (CD45.1⁺CD45.2⁺) 60 h after the transfer (Fig. 1 C and D). These results suggest that CCR10 is required for efficient migration of IgA⁺ cells into the intestines and no other molecule could fully compensate for its function in this process. Therefore, the apparently normal levels of intestinal IgA⁺ cells and antibodies in CCR10^EGFP/EGFP mice result from a compensation mechanism other than functional redundancy of any other homing receptor with CCR10.

Enhanced Generation of IgA⁺ Cells Within Increased Numbers of ILFs in CCR10^EGFP/EGFP Mice.

One possible mechanism to compensate for the impaired intestinal localization of IgA⁺ cells in CCR10^EGFP/EGFP mice is to increase their generation. ILFs, the dynamic B-cell–rich follicles in intestines, are likely a site for the compensational generation of IgA⁺ cells. Supporting this notion, there were significantly more ILFs in intestines of CCR10^EGFP/EGFP mice than of CCR10^+/EGFP mice based on an en face whole-mount staining method that identifies clusters of B220⁺ cells of the ILFs (Fig. 2A) (18, 19). Consistent with the increased numbers of ILFs, there were higher percentages of B220⁺ B cells in lymphocytes isolated from LP of intestines (LPLs) of CCR10^EGFP/EGFP than of CCR10^+/EGFP mice, whereas percentages of CD3⁺ T cells were similar (Fig. 2B and Fig. S2A). Percentages of B220⁺IgA⁺ plasmablasts were also higher in the LPLs of CCR10^EGFP/EGFP than of CCR10^+/EGFP mice, consistent with the enhanced generation of IgA⁺ cells (Fig. 2B). Most of the B220⁺IgA⁺ cells were EGFP⁻ (Fig. S2B), suggesting that CCR10 is up-regulated in a later phase of the IgA⁺ cell differentiation. Immunofluorescent staining of intestinal sections detected similar densities of B220⁺ cells in the LP regions of CCR10^EGFP/EGFP and CCR10^+/EGFP mice (Fig. 2C), consistent with the notion that the higher percentages of B cells in intestines of CCR10^EGFP/EGFP mice are mainly a result of their increased ILF numbers.

We then assessed numbers of IgA⁺ cells in ILFs by immunofluorescent microscopy. Much more IgA⁺ cells were found within ILFs of CCR10^EGFP/EGFP than in those of CCR10^+/EGFP mice (Fig. 2D). The majority of IgA⁺ cells in the ILFs were EGFP⁺, suggesting that they were meant to express CCR10. In contrast, the density of IgA⁺ cells in the intestinal LP areas of CCR10^EGFP/EGFP mice was similar to, if not slightly lower than, that in the CCR10^+/EGFP controls (Fig. S3). Based on the increased numbers of ILFs and their enhanced generation of IgA⁺ cells, we estimated that there were 11 times more IgA⁺EGFP⁺ plasma cells in ILFs of CCR10^EGFP/EGFP mice than in those of CCR10^+/EGFP controls (Fig. S4A). IgA-producing capacities were similar in ILF and LP IgA⁺ plasma cells of CCR10^+/EGFP and CCR10^EGFP/EGFP mice (Fig. S4B). Although the increased IgA⁺ plasma cells of ILFs account only for approximately 2% of IgA⁺ plasma cells of LP in CCR10^EGFP/EGFP mice, contribution of the ILF-originated IgA⁺ plasma cells to the intestinal IgA production might be much greater than this number suggested, as the mature IgA⁺ plasma cells generated in ILFs would likely migrate out continuously and end up in LP for a long-term production of IgA antibodies. Together, these results demonstrate that the impaired migration of CCR10-deficient IgA⁺ cells into intestines is compensated for (at least in part) by their enhanced generation within increased numbers of ILFs. There were same numbers of Peyer patches and same percentages of IgA⁺ cells within them in CCR10^EGFP/EGFP and CCR10^+/EGFP mice (Fig. S5), suggesting that they are not involved in the compensational generation of IgA⁺ cells.

Reduced Hypermutation in IgA Antibodies of CCR10^EGFP/EGFP Mice.

ILFs support a T-cell–independent IgA isotype switch (3), which generates antibodies carrying fewer hypermutations than with a T-cell–dependent process. We therefore assessed mutation rates in the Ig heavy chain (IgH) alleles of intestinal IgA⁺ cells of CCR10^EGFP/EGFP mice. Compared with the WT controls, significantly higher percentages of the CCR10-KO IgA⁺ cells carried germline sequences whereas much fewer of them carried high numbers of mutations (i.e., >5) in the assayed IgH region (Fig. 3A). On average, mutation rates were 8.75 and 4.56 nucleotides per 1 kb in the CCR10^EGFP/EGFP and CCR10^+/EGFP samples, respectively (the basal mutation rate of this assay was 0.59 nucleotides per kb). Therefore, although production of IgA was compensated for in CCR10^EGFP/EGFP mice, quality of the antibodies might be different from that of WT mice.

Involvement of Commensal Bacteria in Compensational Generation of IgA⁺ Cells in ILFs of CCR10^EGFP/EGFP Mice.

The reduced IgA hypermutation could have impacts on intestinal homeostasis. Consistent with this, CCR10^EGFP/EGFP mice had slightly higher loads of microflora in the mucosa of large intestines than CCR10^+/EGFP mice (Fig. 3B), although compositions of several major groups of commensal bacteria were mostly normal, and titers of fecal IgA antibodies against them were not significantly affected (Fig. S6), suggesting a minor disturbance of intestinal homeostasis of the microflora. As development of ILFs and IgA isotype switch are induced by the colonization of commensal bacteria in intestines (3, 6), the higher levels of intestinal microflora in CCR10^EGFP/EGFP mice are likely associated with the enhanced generation of ILFs and IgA. To test this, we treated the mice with a mixture of antibiotics at ages of 5 wk, when many ILFs are fully developed in the large but not small intestine (6). A 2-wk treatment drastically reduced development of ILFs in small intestines; ILF numbers become similar in the treated CCR10^EGFP/EGFP and CCR10^+/EGFP mice (Fig. 3C). On the contrary, the antibiotic treatment did not significantly reduce numbers of ILFs in colons; in which ILF numbers were still higher in the treated CCR10^EGFP/EGFP than in CCR10^+/EGFP mice (Fig. 3C). However, there were essentially no IgA⁺ cells in the colonic ILFs of the treated CCR10^EGFP/EGFP mice, the same as in the CCR10^+/EGFP mice (Fig. 3D), indicating that the enhanced generation of IgA⁺ cells within ILFs of CCR10^EGFP/EGFP mice is stimulated by commensal bacteria and could be reduced by decreasing the commensal bacteria with antibiotic treatment. The antibiotics did not decrease titers of fecal IgA antibodies, even after a longer period of treatment (Fig. S7), likely because intestinal IgA⁺ plasma cells are long-lasting (i.e., half-life of >16 wk) even in absence of stimulation from commensal bacteria (8).

Dysregulated Distribution of CCR10-Deficient IgA⁺ Cells in Intestines and Internal Lymphoid Tissues.

The compensation process made it difficulty to dissect how intrinsic expression of CCR10 on IgA⁺ cells is important for their migration and maintenance in the intestines using the CCR10^EGFP/EGFP mice. We then performed competitive bone marrow (BM) transfer experiments in which similar numbers of BM cells of CCR10^EGFP/EGFP and CCR10^+/EGFP mice were cotransferred into lethally irradiated WT mice. Markedly, whereas the donor cells of both types reconstituted BM of the recipients with similar efficiencies 7 to 8 wk after the transfer (Fig. 4A), IgA⁺ cells of the CCR10^EGFP/EGFP donor origin were profoundly under-represented in small and large intestines (Fig. 4B). In contrast, there were more IgA⁺ cells of CCR10^EGFP/EGFP than CCR10^+/EGFP origin in internal lymphoid tissues such as spleens and BM (Fig. 4B). These results demonstrate that (i) intrinsic expression of CCR10 by IgA⁺ cells is critically important for their efficient migration and/or maintenance in the intestines, and (ii), in its absence, the IgA⁺ cells abnormally accumulate in the internal tissues. Further supporting the second notion, significantly higher percentages of EGFP⁺IgA⁺ cells were also found in BM, spleens, and mesenteric lymph nodes (MLNs) of unmanipulated CCR10^EGFP/EGFP than of CCR10^+/EGFP mice (Fig. 4C and Fig. S8). Serum levels of IgA were also slightly higher in CCR10^EGFP/EGFP mice than in CCR10^+/EGFP mice, whereas those of IgG, as controls, were similar (Fig. 4 D and E). Relative abundance of monomeric and dimeric forms of IgA in sera of CCR10^EGFP/EGFP and CCR10^+/EGFP mice was similar (Fig. S9), consistent with a previous finding that the monomer is a dominant form of serum IgA, as the dimers (or polymers) could be transported into intestines through bile ducts even if they were generated in internal tissues and secreted into blood (20). The increased internal tissue-derived IgA antibodies likely also contribute to the intestinal IgA antibody pool. Taken together, our data demonstrate that absent expression of CCR10 on IgA⁺ cells results in their dysregulated distribution in intestines and internal tissues, but enhanced generation of IgA⁺ cells in ILFs and possibly other lymphoid tissues helps to offset the defect in intestines. Consistent with this, there were similar frequencies of total IgA antibody-secreting cells (IgA-ASCs) in intestines of CCR10^EGFP/EGFP and CCR10^+/EGFP mice based on an enzyme-linked immunosorbent spot (ELISPOT) assay, whereas higher frequencies of IgA-ASCs were found in BM cells of CCR10^EGFP/EGFP than of CCR10^+/EGFP mice (Fig. 4F).

Differential Effects of CCR10 KO on Intestinal IgA Responses to T-Dependent Antigen Stimulation and Bacterial Infection.

To further understand the compensation process involved in intestinal IgA responses in CCR10^EGFP/EGFP mice, we then tested how CCR10 KO affects intestinal IgA responses to T-dependent and -independent antigen stimulations, which should be different if the compensatory generation of IgA⁺ cells is mainly through a T-independent process. Indeed, in a T-dependent model in which mice were orally challenged with the protein antigen CT, titers of CT-specific IgA antibodies were up to fivefold lower in CCR10^EGFP/EGFP mice than in CCR10^+/EGFP mice (Fig. 5A). In contrast, there was no difference in titers of fecal IgA specific to the infection in Rag1^−/− mice that were reconstituted with naive CCR10^EGFP/EGFP vs. CCR10^+/EGFP B cells and orally infected with bacteria Citrobacter rodentium, a model of T-independent IgA responses (Fig. 5B) (21). These results support the notion that the compensatory generation of IgA in CCR10^EGFP/EGFP mice is mainly through a T-independent process and is therefore not effective in response to T-dependent antigen stimulation. In a more natural but complicated setting, Citrobacter-infected CCR10^EGFP/EGFP mice, which are capable of T-dependent and -independent IgA responses, had a slight delay in the Citrobacter-specific IgA production at an early time point after infection compared with the CCR10^+/EGFP controls but caught up at all later time points (Fig. 5C) and did not have defects in clearance of the infection (Fig. S10), suggesting that any impaired intestinal migration of Citrobacter-specific IgA⁺ cells was largely compensated. Consistent with the notion that CCR10-deficient IgA⁺ plasma cells abnormally accumulate in internal tissues, serum levels of the Citrobacter-specific IgA, but not IgG, were higher in the infected CCR10^EGFP/EGFP mice than in the CCR10^+/EGFP mice at most time points, even long (i.e., > 3 mo) after the infection and clearance of the bacteria during the memory phase (Fig. 5D and Fig. S11).

Impaired Maintenance of Long-Lived Citrobacter-Specific IgA-Producing Plasma Cells in CCR10^EGFP/EGFP Mice.

Confirming the long-term accumulation of Citrobacter-specific IgA⁺ plasma cells in internal tissues, higher frequencies of Citrobacter-specific IgA⁺ spots were detected in BM cells of CCR10^EGFP/EGFP mice than in those of CCR10^+/EGFP mice 3 mo after their infection, as determined by an ELISPOT assay (Fig. 5E), Interestingly, frequencies of the Citrobacter-specific IgA⁺ spots in LPL of the infected CCR10^EGFP/EGFP mice were significantly lower than in the CCR10^+/EGFP controls 3 mo after infection (Fig. 5E), revealing an important role of CCR10 for the long-term maintenance of IgA-producing plasma cells in intestines. Therefore, although the defective intestinal migration of CCR10-deficient IgA⁺ cells could be compensated for under homeostatic conditions or in the early phase of infection by the enhanced generation of IgA⁺ cells as a result of stimulation from presence of the commensal or pathogenic bacteria, such a mechanism did not exist to offset the impaired maintenance of the long-lived Citrobacter-specific IgA-producing plasma cells after the bacteria are cleared.

Severely Impaired Memory IgA Response to Citrobacter Infection in CCR10^EGFP/EGFP Mice.

The impaired maintenance of long-lived Citrobacter-specific IgA⁺ plasma cells in CCR10^EGFP/EGFP mice might affect memory responses to reinfection of the bacteria. To test this, we reinfected the Citrobacter-infected CCR10^EGFP/EGFP and CCR10^+/EGFP mice 5 mo after their infection and clearance. In striking contrast to the generally normal primary IgA response, the memory IgA response to the Citrobacter infection in CCR10^EGFP/EGFP mice was severely impaired (Fig. 6A). Two weeks after the reinfection, the CCR10^+/EGFP mice reached the peak level of memory response, with the titer of Citrobacter-specific IgA in feces four times higher than that of the CCR10^EGFP/EGFP mice (Fig. 6A). Four weeks after the reinfection, the fecal IgA level went down from its peak in the CCR10^+/EGFP mice but kept increasing in the CCR10^EGFP/EGFP mice, resulting in their comparable IgA levels at this time point (Fig. 6A). However, the IgA production in CCR10^EGFP/EGFP mice never reached the peak level of CCR10^+/EGFP mice. In addition, the significant difference in the fecal IgA level between the CCR10^EGFP/EGFP and CCR10^+/EGFP mice reappeared by end of the week 5 and thereafter because the CCR10^+/EGFP mice maintained a high level of the Citrobacter-specific IgA for a long time but the CCR10^EGFP/EGFP mice showed a rapid decrease (Fig. 6A). On the whole, the fecal IgA production pattern of the memory response in CCR10^EGFP/EGFP mice was not significantly different from that of the primary response, whereas the IgA memory response was much faster and lasted much longer than the primary response in the CCR10^+/EGFP mice (compare Fig. 5C vs. Fig. 6A). This demonstrates that CCR10^EGFP/EGFP mice are almost devoid of the IgA memory response.

By using an ELISPOT assay, we confirmed that the severely impaired production of fecal Citrobacter-specific IgA antibodies in the reinfected CCR10^EGFP/EGFP mice was a result of reduced numbers of Citrobacter-specific IgA-secreting plasma cells in the intestines. At day 14 after the reinfection, numbers of Citrobacter-specific IgA-secreting plasma cells in large intestines of CCR10^EGFP/EGFP mice was drastically (15 fold) lower than in CCR10^+/EGFP controls (Fig. 6B, Top). The numbers were also significantly (threefold) reduced in small intestines of the CCR10^EGFP/EGFP mice (Fig. 6B, Bottom). The different extents of effect of CCR10 deficiency in the small and large intestines are likely caused by differential expression of other chemokine molecules involved in the migration of IgA⁺ cells in these tissues, such as the CCR9 ligand CCL25 that is expressed only in the small intestine (9).

As seen in the primary response, there were much higher levels of Citrobacter-specific IgA in blood of the reinfected CCR10^EGFP/EGFP mice than in CCR10^+/EGFP mice most of the time (Fig. 6C). During a period between 1 and 2 mo after reinfection, serum levels of the Citrobacter-specific IgA were similar in the CCR10^+/EGFP and CCR10^EGFP/EGFP mice (Fig. 6C), likely because of the higher production of IgA in intestines of the CCR10^+/EGFP mice (Fig. 6A). However, the difference in titers of serum IgA between CCR10^EGFP/EGFP and CCR10^+/EGFP mice reappeared after the intestinal IgA production went down (>3 mo past reinfection; Fig. 6C).

Impaired Maintenance of Citrobacter-Specific IgA⁺ Memory B Cells in Intestines of CCR10^EGFP/EGFP Mice.

The severely impaired IgA memory response in CCR10^EGFP/EGFP mice suggests that any possible compensational generation of IgA⁺ cells could not offset the impaired migration and maintenance of CCR10-deficient IgA-producing plasma cells during the memory response, likely because generation of IgA⁺ cells from naive B cells in ILFs would require a much longer time than from the IgA⁺ memory B cells. In addition, it is possible that maintenance of the IgA⁺ memory B cells in infected CCR10^EGFP/EGFP mice is also defective even before the reinfection. Supporting this possibility, CCR10 (EGFP) was expressed on a fraction of IgA⁺ memory-like B cells (CD19⁺CD38^hiCD138⁻) of intestines, but not of other internal lymphoid tissues such as BM and MLN and blood (Fig. 7A) (22–24). Although frequencies of total CCR10⁺ and CCR10⁻ IgA⁺ memory-like B cells in intestines were not significantly different in CCR10^EGFP/EGFP vs. CCR10^+/EGFP mice (Fig. S12), this could be because the effect of CCR10 KO on the memory cells is masked under homeostatic conditions, as with the IgA⁺ plasma cells. Indeed, when Citrobacter-specific IgA⁺ memory B cells were assessed by using an ELISPOT assay in CCR10^EGFP/EGFP and CCR10^+/EGFP mice 3 mo after their infection and clearance of Citrobacter, the Citrobacter-specific IgA⁺ memory B-cell numbers were significantly lower in intestines of the CCR10^EGFP/EGFP mice (Fig. 7B), whereas the ELISPOT assay still detected the same frequencies of total IgA⁺ memory B cells in intestines of CCR10^EGFP/EGFP and CCR10^+/EGFP mice (Fig. 7C). CCR10^EGFP/EGFP and CCR10^+/EGFP mice also had similar frequencies of total and Citrobacter-specific IgA⁺ memory B cells in their BM (Fig. 7 B and C), indicating that CCR10 KO specifically impaired intestinal maintenance of the Citrobacter-specific IgA⁺ memory B cells.

Mice.

CCR10-KO/EGFP-knock-in mice on C57BL/6 (B6) background (CD45.2⁺) were described previously (17), and crossed to B6 mice bearing CD45.1 alleles (Jackson Laboratory) to obtain CCR10^EGFP/EGFP and CCR10^+/EGFP mice bearing CD45.1 and/or CD45.2 alleles. Rag1^−/− mice were from Jackson Laboratory. Mice were used at 8 to 12 wk of age unless indicated otherwise. All mouse experiments were performed in specific pathogen-free conditions in accordance with protocols approved by the Pennsylvania State University Institutional Animal Care and Use Committee.

Antibodies and Chemicals.

PE-Cy7–conjugated anti-CD45.1 (A20), PE-Cy7–conjugated anti-CD3ε (145-2c11), biotin-conjugated anti-IgA (11-44-2), purified or Alexa Fluor 647-conjugated anti-B220 (RA3-6B2), PE- or PE-Cy7–conjugated anti-CD19 (1D3), and PE-Cy5–conjugated anti-CD38 (clone 90) antibodies were purchased from eBioscience. APC-conjugated anti-CD45.2 (clone 104), biotin-conjugated anti-CD138 (281-2), purified rat anti-mouse CD16/CD32 (2.4G2) antibodies, and PE-Texas Red–conjugated streptavidin were from BD Bioscience. PE-conjugated anti-IgA (11-44-2) was from Southern Biotech. Alexa Fluor 568-conjugated streptavidin was from Molecular Probes. Peroxidase-conjugated AffiniPure goat anti-rat IgG (H + L) was from Jackson ImmunoResearch. CT was from List Biological.

BM Transfer.

Cell sorter-isolated EGFP⁻ BM cells of CCR10^+/EGFP (CD45.1⁺CD45.2⁺) and CCD10^EGFP/EGFP (CD45.1⁺CD45.2⁻) mice were 1:1 mixed and injected i.v. into lethally irradiated (950 rad) WT B6 (CD45.1⁻D45.2⁺) mice (total 10⁶ cells per mouse). The recipients were analyzed 7 to 8 wk after the transfer.

Oral Immunization with CT.

On days 1 and 3, mice were deprived from food for 8 h and gavaged with two doses of CT (10 μg in 0.5 mL PBS solution containing 3% sodium bicarbonate), and analyzed on 7, 14, and 21 d after the first immunization.

Citrobacter Infection.

Mice were fasted for 8 h and then orally infected by gavage with 2 × 10⁹ cfu of C. rodentium strain DBS100 (ATCC 51459; American Type Culture Collection) in a total volume of 200 μL per mouse in both the primary and secondary infection. The secondary infection was performed on mice that were infected with Citrobacter 5 mo earlier and confirmed cleared of the infection. To assess T-independent IgA responses, Rag1^−/− mice were transferred with 10⁶ EGFP⁻B220⁺IgM⁺ B cells of naive CCR10^+/EGFP or CCR10^EGFP/EGFP mice 1 d before they were infected with Citrobacter. The cfu count of the bacteria used in the infection was estimated by absorbance at the wavelength of 600 nm and confirmed by serially diluting and plating the inoculums on MacConkey agar plates (Becton Dickinson). For quantification of bacterial titers after the infection, feces were homogenized in PBS solution, serially diluted, and plated on MacConkey agar plates. Colonies formed were counted after an overnight culture.

Antibiotic Treatment.

Mice were fed with an antibiotic mixture of 500 mg/L ampicillin (Sigma), 500 mg/L vancomycin (Sigma), 1 g/L neomycin (USB), and 1 g/L metronidazole (Sigma) in drinking water as reported previously (30).

Assessment of Somatic Hypermutation Rates in IgH Alleles.

Genomic DNA were prepared from cell sorter-purified intestinal EGFP⁺sIgA⁺ cells or B220⁺IgM⁺ spleen cells and analyzed for the hypermutation frequencies in a 3′ region franking V_HJ558/D/J_H4 rearrangements as previously described (31).

Quantification of Commensal Bacteria.

Quantification of commensal bacteria was performed as previously described (32, 33).

Western Blot Quantification of Monomeric and Dimeric IgA Antibodies.

Serum samples were run on 6% nonreducing SDS/PAGE gel, transferred to nitrocellucose, blotted with alkaline phosphatase-conjugated goat anti-mouse IgA (Santa Cruz Biotechnology), and developed with the enhanced chemiluminescence substrate (GE Healthcare). Fluorescent signals from products of the enhanced chemiluminescence substrate were detected by using the FLA-7000 system (Fujifilm), and intensities of monomeric and dimeric IgA bands were quantified by using Multi gauge software.

Whole-Mount Staining of Intestines for Detection of B220⁺ Clusters of ILFs.

Whole-mount staining of intestines for detection of B220⁺ clusters of ILFs was performed as previously reported (18, 19).

Immunofluorescent Microscopy.

Intestines were flushed with PBS solution, opened longitudinally along the mesenteric border, and fixed for 2 h at 4 °C in a fresh 4% paraformaldehyde solution. After three washes in PBS solution, the tissues were incubated in a 30% sucrose solution overnight at 4 °C, and then frozen in OCT compound (Sakura Finetek). The frozen tissues were cut into 6-μm sections and stained with fluorescently labeled antibodies in an antibody amplifier (ProHisto) according to the manufacturer's instructions. The stained sections were covered with mounting medium containing DAPI (Vector) and examined under a FluoView FV1000 confocal microscope (Olympus). Images were processed with FV10-ASW software (Olympus).

In Vivo Migration Assay.

Intestinal LP lymphocytes were isolated from CCR10^+/EGFP (CD45.1⁺CD45.2⁺) and CCD10^EGFP/EGFP (CD45.1⁻CD45.2⁺) mice. Similar numbers of EGFP⁺sIgA⁺ cells of the two kinds of mice were mixed and injected i.v. into C57BL/6 WT recipient mice (CD45.2⁻CD45.2⁺). Sixty hours after the injection, lymphocytes were isolated from large and small intestines of the recipients separately and FACS analyzed for EGFP, sIgA, CD45.1, and CD45.2 expression. Percentages of the CCR10^+/EGFP vs. CCR10^EGFP/EGFP donor cells in gated EGFP⁺sIgA⁺ populations were determined based on the CD45.1 and CD45.2 expression.

Cell Isolation.

LP lymphocytes were isolated from large and small intestines as described (34), with some modifications. In brief, intestines were flushed with cold HBSS containing 15 mM Hepes (CMF/Hepes). After removal of fat, connective tissues, and Peyer patches, the intestines were opened longitudinally, cut into 5-mm segments, and washed four or five times with cold CMF/Hepes. The intestinal segments were then incubated for 15 min at 37 °C with shaking on a shaker (Model: GyromaxTM 737, Amerex Instrument Inc., Lafayette, CA) at 200 rpm in HBSS containing 15 mM Hepes, 5 mM EDTA, and 10% FBS (Equitech-Bio), followed by intense vortexing to remove the epithelium and intraepithelial lymphocytes. This step was repeated three or four times until no more epithelium shedding occurred. The remaining pieces were washed with RPMI medium (Mediatech) containing 10% FBS, minced, and then digested for 60 min at 37 °C with shaking at 200 rpm in RPMI medium containing 5% FBS, 0.6 mg/mL collagenase (Worthington), and penicillin and streptomycin (Gibco). Dissociated cells from the digestion were washed once with PBS solution, resuspended in 5 mL of 100% Percoll (GE Healthcare), underlaid with 4 mL of 40% Percoll, and centrifuged for 20 min at 850 × g at room temperature. LP lymphocytes were recovered from the interphase of the Percoll gradient, washed twice, and resuspended in FACS buffer (PBS solution containing 3% FBS and 0.05% sodium azide) or RPMI medium. To isolate lymphocytes from Peyer patches, Peyer patches were gently homogenized with a tissue homogenizer, passed through a 70-μm cell strainer (BD Biosciences), and enriched by Percoll gradient centrifugation. BM cells were isolated by flushing tibias and femurs. Spleen and MLN cells were prepared by pressing the tissues through cell strainers using the end of a sterile plunger of a 5-mL syringe. Peripheral blood lymphocytes were isolated by gradient centrifugation by using Lympholyte-Mammal (Cedarlane Laboratories).

Flow Cytometry.

All FACS analyses were performed on an FC500 series system (Beckman Coulter), and data were analyzed with FlowJo software (TreeStar).

ELISA.

Samples of feces and blood were analyzed for the total and Citrobacter-specific IgA or IgG antibodies. Feces were collected and homogenized in 1 mL of TBS–BSA solution (50 mM Tris, 140 mM NaCl, 1% BSA, pH 8.0) per 100 mg feces. After centrifugation at 16,000 × g for 5 min, the supernatants were collected and kept at −80 °C until the ELISA. Blood samples were collected by cheek puncture, put at room temperature for 30 min and 4 °C for 2 h, and centrifuged at 2,500 × g for 10 min. Resulting sera were collected after the centrifugation and kept at −80 °C until analysis.

Total IgA and IgG levels in the fecal and serum samples were determined by using a Mouse IgA ELISA Quantitation Set and Mouse IgG ELISA Quantitation Set (Bethyl Laboratories), respectively, following the manufacturer's instructions.

For quantification of bacteria-specific IgA or IgG antibodies, 96-well ELISA plates (Costar) were coated overnight at 4 °C with 100 μL of soluble proteins (10 μg/mL in PBS solution) prepared from centrifugation supernatants (16,000 × g, 1 h at 4 °C) of sonicated C. rodentium, Clostridium perfrigens (ATCC 13124D-5), Eubacterium rectale (ATCC 33656), Lactococcus lactis (ATCC 11454), or Escherichia coli (ATCC 43890) (24 cycles of 5 s burst plus 10 s resting, on ice). The wells were then washed five times and blocked for 2 h at room temperature with 200 μL of TBS–BSA solution, followed with two washes again. Wells were then incubated for 1 h at room temperature with 100 μL of samples diluted in the TBST–BSA solution. Wells were again washed five times, and incubated for 1 h at room temperature with HRP-conjugated anti-mouse IgA or HRP-conjugated anti-mouse IgG (Bethyl Laboratories). After five washes, wells were developed with 100 μL of TMB substrate solution (BD Biosciences) for 5 to 15 min, and then stopped with 100 μL of 2 N H₂SO₄ solution. The optical density was read at 450 nm. A set of twofold serially diluted fecal extracts or serum samples was always included to produce a standard curve in every experiment. The sample served as the standard was kept at −80 °C as aliquots to avoid freezing and thawing, and the same standard was used in every experiment. Relative concentrations of bacteria-specific IgA or IgG antibodies in tested samples were calculated by comparing to the standard curve and expressed as relative units.

ELISPOT.

The frequency of total or Citrobacter-specific IgA–secreting plasma cells in intestinal lymphocyte preparations or BM cells was determined by ELISPOT, similar as previously described (35, 36) with a minor modification. In brief, MultiScreen IP filtration plates (Millipore) were prewetted for 1 min with 15 μL of 35% ethanol, rinsed three times with sterile PBS solution, and then coated with 100 μg/mL of 1:50-diluted anti-IgA coating antibody (Bethyl Laboratories) or C. rodentium soluble proteins in PBS solution overnight at 4 °C, followed by five washes with PBS solution. The plates were blocked for 2 h at 37 °C with RPMI medium supplemented with 10% FBS, 50 μM β-mercaptoethanol, glutamine, and penicillin and streptomycin (RPMI-10). Cells to be tested were then transferred to the plates in serial dilutions and cultured overnight at 37 °C in the RPMI-10 medium. The plates were washed five times with PBS solution containing 0.01% Tween-20, and incubated for 1 h at 37 °C with 1:5,000 diluted HRP-conjugated anti-IgA detection antibody (Bethyl Laboratories), followed by five washes. The reactions were developed by using TMB substrate for ELISPOT assay (Mabtech) and stopped by washing under running water. Spots corresponding to antibody-secreting cells were counted under a dissecting microscope. The results were normalized to total cell inputs.

Detection of total or Citrobacter-specific memory B cells was also performed as previously described (35, 36) with minor modifications. Briefly, BM cells or intestinal lymphocyte preparations were cultured at 5 ×10⁵ cells/mL and stimulated with 5 μg/mL PWM (Sigma-Aldrich), 6 μg/mL type B CpG (ODN-1668; InvivoGen), 2 μg/mL LPS (Sigma-Aldrich), and 100 U/mL IL-2 in the RPMI-10 medium for 6 d in 24-well plates. After stimulation, cells were washed three times with RPMI medium and used in the ELISPOT assays. Frequencies of the Citrobacter-specific IgA memory B cells were calculated from the numbers of IgA⁺ spots normalized to the initial cell inputs.

Statistical Analyses.

All data are expressed as means ± SEM. Statistical significance was determined by two-tailed Student t tests, with P < 0.05 as the threshold of significance.