Mouse Liver B Cells Phagocytose Streptococcus pneumoniae and Initiate Immune Responses against Their Antigens

An adaptive immune response is initiated when naive T cells encounter specific Ags on APCs in secondary lymphoid tissues, such as the spleen (1). These APCs include dendritic cells, macrophages, and B cells, among which dendritic cells are considered by far the most important for this priming of naive T cells (2), while B cells are not because an extremely limited number of B cells could specifically bind to a certain Ag. In addition, most natural Ags, such as bacteria and viruses, are particulate (3). The common acceptance of B cells as lacking phagocytic activity has limited opportunities to discuss their presentation of Ag-derived peptides to T cells (4, 5).

However, because Li et al. reported that B cells in early vertebrates have phagocytic and bactericidal activities (6), several lines of evidence have shown that mammalian B cells also phagocytose particulate Ags. After ingesting bacteria, B cells further contribute to bacterial clearance by producing nonspecific IgM (7). Human B cells ingesting Salmonella typhimurium via their surface IgM were shown to present Ag derivatives to CD4⁺ T cells and to produce IgM specific for their LPSs (8, 9), although few researchers have assessed the effects of B cell phagocytosis on the adaptive immune responses.

In the process of high-affinity Ab production by B cells, some of the helper CD4⁺ T cells are required to have been activated by the peptide, which is either identical or physically associated with those Ags recognized by the B cells. Those activated B cells move to the T zone–B zone boundary in secondary lymphoid tissues, where they may encounter their cognate T cells and begin to proliferate. Some B cells move to the germinal center, where somatic hypermutation and class-switch recombination take place (10, 11). In this process, enzymes of activation-induced cytidine deaminase (AICDA), uracil-DNA glycosylase (UNG), and apurinic/apyrimidinic endonuclease 1 (APE1) sequentially act on the base–excision repair pathway in B cell bodies (12), and those B cells that bind Ags with the highest affinity are selected for survival and further differentiation (4, 13).

Streptococcus pneumoniae is a major cause of bacterial pneumoniae, otitis, and meningitis in humans. Its virulence is attributed in part to capsules surrounding the cell membrane, which enable the bacteria to resist phagocytosis by dedicated phagocytes (14). Because clinical data have shown that deficiency in the classical complement pathway increases susceptibility to S. pneumoniae infection (15), opsonization with serum rich in complement and immunoglobulin is required for elimination from the host (16). We previously showed that opsonization with complement also increased the ingestion of Escherichia coli by mouse B cells (17); however, the effect of complement on the ingestion of S. pneumoniae by B cells has not yet been elucidated.

S. pneumoniae possesses multiple protein and polysaccharide Ags on the cell surface. Once protein Ags, such as pneumococcal surface protein A (PspA), are internalized into an APC, they are degraded into constituent peptides, which then become associated with MHC class II molecules for transport to the cell surface and are further presented to T cells as thymus-dependent Ags. In contrast, polysaccharide Ags, such as pneumococcal polysaccharide (PPS), are not degraded within the APC for MHC class II association. PPS induces multivalent cross-linking of specific membrane immunoglobulins on B cells, and it produces IgM as a thymus-independent type 2 Ag. Nevertheless, some reports have suggested that PPS also induces the production of anti-PPS IgG dependently on T cells (18).

Using fluorescence-labeled S. pneumoniae, we investigated the effects of B cell phagocytosis on adaptive immune responses, such as Ag-specific Ab production.

This study was conducted according to the permissions (number: 17044) and guidelines of the Institutional Review Board for the Care of Animal Subjects at the National Defense Medical College, Japan. C57BL/6 mice 8–10 weeks of age were purchased from Japan SLC (Hamamatsu, Japan). Because several studies have revealed that males are more susceptible to infection with S. pneumoniae than females, both in humans and mice, we used only male mice (19, 20).

Liver and spleen mononuclear cells (MNCs) were obtained as previously described (21). Peritoneal MNCs were obtained by lavage with 4 ml of PBS, with 3.5 ml of this lavage solution collected. Fresh serum was concomitantly obtained from the vena cava.

B cell–deficient muMT (B6.129S2-Ighm^tm1Cgn/J) mice were obtained from Jackson Laboratory. This strain was developed by disrupting one of the membranous exons of the gene encoding immunoglobulin heavy chain of the class μ with a neomycin resistance cassette (22). We confirmed homogenous mutation of that gene and lacking B cells in the mature mice before conducting the experiments using primers of 5′-CCG TCT AGC TTG AGC TAT TAG G-3′ (common), 5′-GAA GAG GAC GAT GAA GGT GG-3′ (wild-type), and 5′-TTG TGC CCA GTC ATA GCC GAA T-3′ (mutant reverse).

S. pneumoniae serotype 3 (6303) was purchased from American Tissue Culture Collection (Manassas, VA) and grown in Todd-Hewitt broth (BD Pharmingen, San Diego, CA) to the mid-log phase, at which point it was stored at −80°C. FITC (F1907) and pHrodo Red succinimidyl ester (P36600) were purchased from Life Technologies Corporation (Eugene, OR) to prepare FITC–S. pneumoniae and pHrodo–S. pneumoniae, respectively. These fluorescent S. pneumoniae were prepared according to the manufacturer’s instructions. Regarding pHrodo–E. coli, we used commercially available products (P35361; Life Technologies). Heat-killed S. pneumoniae was prepared with a high-temperature and high-pressure steam sterilizing autoclave (121°C, 2 atm, 15 min).

We obtained Abs conjugated with FITC (anti-CD43, anti-IgM, anti-MHC class II, and rat IgG2b isotype), PE (anti-IgM, anti-B220, anti-CD21, and anti-CD4), PE-Cy5 (anti-IgM and anti-B220), allophycocyanin (anti-F4/80, anti-CD86, rat IgG2a, and rat IgG2b), biotin (anti-B220 and anti-IgM), and PerCP-Cy5.5 (streptavidin) from Thermo Fisher Scientific (Waltham, MA). We also purchased Abs conjugated with FITC (anti-CD80, anti-CD40, Armenian Hamster IgM, and its IgG), allophycocyanin (anti-CD20 and rat IgG2a isotype), PE-Cy7 (anti-CD19 and anti-CD23), Pacific Blue (anti-CD5), and AmCyan (anti-CD11b) from BioLegend (San Diego, CA). Anti-FITC, PE, and Cy5 microbeads were purchased from Miltenyi Biotec (Bergisch Gladbach, Germany).

We cultured liver, spleen, and peritoneal MNCs (5 × 10⁵ cells/200 μl) with pHrodo– or FITC–S. pneumoniae at a cell/bacteria ratio of 1:100 in the presence of 5% fresh serum from naive mice or wild-type, with or without small amounts of cytochalasin D (Sigma, St. Louis, MO), 30 μg/ml of anti-C3 Ab (Cedarlane, Burlington, NC), or the same amounts of rat IgG2a (Thermo Fisher Scientific). They were cultured for 1 h at 37°C under 5% CO₂. In some experiments, we added culture with 5% heat-inactivated serum in place of fresh serum. In other experiments, to quench the fluorescence of the labeled bacteria on B cells, we washed cultured cells with PBS followed by staining with 400 μg/ml trypan blue (23). After blocking with anti-CD16/32 Ab (Invitrogen), the MNCs were stained with Abs for an analysis with a Novocyte (Agilent Technologies, Prague, Czech Republic) or FACSCanto II (BD Bioscience, Franklin Lakes, NJ) or further stained with DAPI for a fluorescence microscopic examination (BX-8000; Keyence, Osaka, Japan). Flowcytometry data were further analyzed with the FlowJo software program (version 10; BD Biosciences).

For immunofluorescence staining, spleen sections were treated with FITC-βTCR Ab (Thermo Fisher Scientific) and rat anti-B220 Ab (biotin; Thermo Fisher Scientific). After washing, they were stained with donkey anti-rat IgG (Alexa Flour 594; Life Technologies). Liver sections were primarily stained with hamster anti-CD3 Ab (biotin-conjugated; Thermo Fisher Scientific) and rat anti-B220 Ab (PE), followed with Alexa Fluor 488–conjugated streptavidin (Thermo Fisher Scientific) and donkey anti-rat IgG.

B and T cells were enriched by positive selection using a MACS system (Miltenyi Biotec). Purities of the obtained PE-IgM⁺ and biotin-CD4⁺ cells were >95% and 90%, respectively.

To prepare B cells ingesting pHrodo–S. pneumoniae, we negatively sorted B cells (EasySep Mouse B Cell Isolation Kit; Veritas, Tokyo, Japan) using an IMag system (BD Biosciences), cultured them with pHrodo-labeled bacteria (cell/bacteria = 1:200), and then stained them with anti–FITC-IgM Ab. To prepare B cells capturing FITC–S. pneumoniae, we cultured MNCs with labeled bacteria (cell/bacteria = 1:200) in culture medium containing 10% FBS and 5% normal mouse serum in RPMI 1640 for 1 h and then stained them with anti–PE-IgM Ab. Using SH800 (Sony, Tokyo, Japan), we sorted those cells into pHrodo–S. pneumoniae⁺FITC-IgM⁺ and FITC–S. pneumoniae⁺PE-IgM⁺ cells (pHrodo–S. pneumoniae⁺FITC-IgM⁺ >80%, pHrodo–S. pneumoniae⁻FITC-IgM⁺ >90% purity, FITC–S. pneumoniae⁺PE-IgM⁺ >90%, FITC–S. pneumoniae⁻PE-IgM⁺ >90%, respectively). B cells ingesting pHrodo–E. coli were obtained in the same way (>80% purity).

We cultured magnetically sorted B cells (5 × 10⁵ cells in 200 μl of medium) with heat-killed S. pneumoniae at a cell/bacteria ratio of 1:100 for 48 h. In some experiments, those B cells were cultured for 1 h with 5% serum, 5% serum plus 1% cytochalasin D, or none, and then washed. They were cultured for 4 days with the same number of CD4⁺ cells (2.5 × 10⁵ cells in 200 μl of medium), which were magnetically sorted from liver or spleen MNCs in mice that had been administered heat-killed S. pneumoniae (1 × 10⁸ CFUs) 7 days earlier.

In other experiments, such spleen CD4⁺ T cells from immunized mice were cultured with equal numbers of FITC–S. pneumoniae^+/−IgM⁺ cells or pHrodo–S. pneumoniae^+/−IgM⁺ cells (2.5 × 10⁵ cells in 200 μl of medium) for 5 d. To examine the T cell proliferation, we stained the T cells with a CellTrace Far Red Cell Proliferation Kit (Thermo Fisher Scientific) before culture. Cultured cells were stained with PerCP-βTCR (Thermo Fisher Scientific) before the flow cytometric analysis.

To prepare specimens for transmission electron microscopy (TEM), IgM⁺ cells cultured with heat-killed S. pneumoniae were embedded in gelatin. They were fixed with 1% glutaraldehyde, postfixed with 1% osmium tetroxide, dehydrated in ethanol solutions, cleared in propylene oxide, and embedded in Epon. Thin sections were double stained with uranyl acetate and lead citrate and then examined with a JEM-1010 (JEOL, Tokyo, Japan).

Mice were i.v. administered with 5 × 10⁵ cells of liver IgM⁺ cells, pHrodo–S. pneumoniae^+/−IgM⁺ cells, or pHrodo–E. coli⁺IgM⁺ cells. Blood was sampled from the tail vein before and after the administration (at 0, 3, 7, 14, and 21 d) and centrifuged at 12,000 rpm for 5 min (H1200F; KOKUSAN, Omiya, Saitama, Japan). Plasma was stored at −80°C until the assay.

Regarding the cell trace study, pHrodo–S. pneumoniae⁺IgM⁺ cells were stained with a CellTrace Far Red Cell Proliferation Kit to evaluate their distribution in the spleen and liver, followed by i.v. injection into mice. Seven days later, the obtained livers and spleens were embedded in OCT compound (Sakura Finetek, Tokyo, Japan) and frozen at −80°C.

ELISA plates were coated with 5 μg/ml (50 μl/well) PPS (serotype 3; American Tissue Culture Collection) or PspA (PP0962; Bioclone Inc., San Diego, CA) in PBS overnight at 4°C. Plates were washed three times with PBS and 0.1% Tween 20 and then blocked with PBS and 1% BSA for 30 min at room temperature (20–25°C). Three-fold dilutions of plasma samples, starting at 1/50 plasma dilution, in PBS and 0.05% Tween 20 were then added for 1 h at room temperature, and plates were washed three times with PBS and 0.1% Tween 20. Horseradish peroxidase–conjugated polyclonal goat anti-mouse IgM, IgG, IgG1, IgG2a, IgG2b, or IgG3 detection Ab (Bethyl Laboratories, Montgomery, TX) was then added, and plates were incubated at room temperature for 1 h. After washing three times, tetramethylbenzidine substrate was added for 30 min for color development. Color was read at an absorbance of 450 nm on a microplate reader (SpectraMax iD3; Molecular Devices, Tokyo, Japan). Total Ig isotypes were determined according to the manufacturer’s instructions.

IL-2, IL-6, IL-12 (p40), IFN-γ, IL-4, and IL-5 levels in the culture medium were measured using ELISA kits from BD Biosciences (San Diego, CA). IL-21 and TGF-β1 were determined using the kit from Proteintech Group (Rosemont, IL).

Total RNAs were isolated from cultured cells using an RNeasy Mini kit 50 (QIAGEN, Valencia, CA), and their concentrations were determined with the ABI PRISM 7000 sequence detection system (Life Technologies Corporation, Carlsbad, CA). cDNAs were synthesized from 500 ng of total RNA by reverse transcription using a SuperScript III First-Strand Synthesis device (Thermo Fisher Scientific). Quantitative real-time RT-PCR was performed on a LightCycler 480 System (Roche, Mannheim, Germany) with SYBR Green PCR reagents (LightCycler 480 SYBR Green I Master Version 12; Roche). The primers were designed by Takara Bio (Tokyo, Japan), and the data were normalized for the GAPDH expression in each sample. For mouse activation-induced cytidine deaminase 1 (AICDA1), antisense primer (5′-TCCTGCTCACTGGACTTCG-3′) and sense primer (5′-GCGTAGGAACAACAATTCCAC-3′) were used; for AICDA2, antisense (5′-GCCACCTTCGCAACAAGTCT-3′) and sense (5′-CCGGGCACAGTCATAGCAC-3′); for UNG, antisense (5′-CCATGGGGATTTGTCAGG-3′) and sense (5′-ACAGTGAGGACGGCGTTG-3′); for APE1, antisense (5′-AAAGAAAGGTTTGGATTGGGTAA-3′) and sense (5′-CTGACCAGTACTGATGGGTGAG-3′); and for GAPDH, antisense (5′-GTGGAGTCATACTGGAACATGTAG-3′) and sense (5′-AATGGTGAAGGTCGGTGTG-3′) were used.

Statistical analyses were performed using the JMP Pro software program, 14.0.0 (SAS, Cary, NC). The survival rates were compared using Wilcoxon rank test, and other statistical analyses were performed using the Mann–Whitney U and F tests. Results are given as the mean ± SEM. Differences were considered to be significant at p < 0.05.

When mouse liver MNCs were incubated with pHrodo–S. pneumoniae for 1 h, flow cytometric analyses revealed that limited percentages of their CD19⁺IgM⁺ B cells were positive for the labeled bacteria (Fig. 1A, 1B). Because TEM showed that magnetically sorted liver IgM⁺ B cells contained heat-killed S. pneumoniae in their phagosomes (Fig. 1C), we assumed this CD19⁺IgM⁺pHrodo–S. pneumoniae⁺ population to be phagocytic. This population, which was depicted as merged spots of green (FITC-IgM) and red (pHrodo–S. pneumoniae) in fluorescence microscopy (Fig. 1D), increased doubly by adding fresh serum from naive mice (Fig. 1B). However, the phagocytic population was not increased on culture with heat-inactivated serum or fresh serum with anti-C3 Ab added (Fig. 1B), although the expression of complement receptor CD21 was lower in liver B cells cultured with fresh serum than in those B cells with heat-inactivated serum or no serum (Fig. 1E).

Mouse peritoneal B cells have a B-1 subtype with conspicuous phagocytic activity, which motivated us to examine the activity of the hepatic counterpart. We found that liver CD43⁺B220^dimIgM⁺ B-1 cells had higher phagocytic activity than CD43⁻B220^highIgM⁺ B-2 cells, as well as peritoneal B-1 cells, when liver and peritoneal MNCs were cultured with pHrodo–S. pneumoniae under the same condition (Fig. 2A). In addition, the expression of CD11b (Fig. 2B) and IgD (Fig. 2C) in liver CD5⁺B220^low-highCD19⁺ B-1a cells and CD5⁻B220^low-highCD19⁺ B-1b cells differed from those in their peritoneal B cell counterparts, suggesting that liver B-1 cells were distinct from peritoneal B-1 cells.

Unlike liver B cells, splenic B cells scarcely showed positivity for pHrodo–S. pneumoniae irrespective of their subtypes of CD21^lowCD23^high follicular (FO) B cells, CD21^highCD23^dim marginal zone (MZ) B cells, or CD43⁺B220^dimIgM⁺ B-1 cells, even in the presence of mouse serum (Fig. 3A). TEM and fluorescence microscopic observation also showed that most IgM⁺ cells were merely attaching to FITC–S. pneumoniae on their cell surface when splenic MNCs were incubated with FITC–S. pneumoniae (Supplemental Fig. 1A, 1B). This may be because of their having a smaller size than liver B cells (Fig. 3B). However, splenic B cells internalized S. pneumoniae more efficiently than had been expected from the results with pHrodo–S. pneumoniae; when splenic B cells were cultured with FITC–S. pneumoniae in the presence of fresh serum, the FITC-positive populations in total IgM⁺ cells, FO B cells, MZ B cells, and B-1 cells were 5.7% ± 0.9%, 3.2% ± 0.5%, 22.3% ± 3.8%, and 13.3% ± 2.1%, respectively (Fig. 3C, upper panels), whereas these populations declined with the addition of trypan blue to 2.2% ± 0.6%, 1.8% ± 0.3%, 0.8% ± 0.1%, and 8.0% ± 1.7%, respectively (Fig. 3C, lower panels). Consistently, the addition of phagocytosis-inhibiting reagents of cytochalasin D to these cultures tended to decrease the FITC–S. pneumoniae⁺ population in cultured B cells dose-dependently (Supplemental Fig. 1C). Considering that pHrodo reagents characteristically emit fluorescence in an acidified environment, splenic B cells appear to internalize S. pneumoniae to some degree without subsequently subjecting the bacteria to processing/acidification.

We adopted liver B cells for assessing the role of phagocytosis on Ab production in vitro. Liver IgM⁺ cells were negatively enriched from liver MNCs followed by culture with pHrodo–S. pneumoniae, and these IgM⁺ cells were largely positive for CD19 and CD20 (>95%), indicating them to be B cells (Supplementary Fig. 2A). In addition, thus obtained pHrodo–S. pneumoniae⁺IgM⁺ cells were confirmed to scarcely express markers of professional phagocytes, such as F4/80 and CD11c (Supplemental Fig. 2B). After incubation of these pHrodo–S. pneumoniae-positive or -negative B cells for 72 h, B cells ingesting S. pneumoniae produced larger amounts of anti-PPS IgM than nonphagocytic (pHrodo–S. pneumoniae⁻) B cells, whereas the specificity for PPS, as shown by anti-PPS IgM/total IgM, was lower in the culture of pHrodo–S. pneumoniae⁺ liver B cells (Fig. 4A). Neither of these liver B cells produced IgM specific for PspA (data not shown). Activation markers of CD86, CD40, and MHC class II were markedly (partially for CD80) enhanced in the pHrodo–S. pneumoniae⁺ B cells, as well as in the pHrodo–S. pneumoniae⁻ population (Fig. 4B), suggesting that most B cells were activated by culture with S. pneumoniae, regardless of ingestion.

To assess the systemic effect of B cells ingesting S. pneumoniae on acquired immune responses, we transferred liver IgM⁺ B cells either positive or negative for pHrodo–S. pneumoniae or pHrodo–E. coli⁺ liver B cells into wild-type mice and examined the plasma Ab concentrations (Fig. 5). We found that transfer of pHrodo–S. pneumoniae⁺ liver B cells induced higher titers of anti-PspA IgG, especially anti-PspA IgG2a and IgG2b, in the recipients than that of pHrodo–S. pneumoniae⁻ B cells or B cells ingesting other bacteria (pHrodo–E. coli⁺ B cells). The pHrodo–S. pneumoniae⁺ B cells also showed higher levels of IgG3 specific for PPS than the pHrodo–S. pneumoniae⁻ B cells, but no marked increases in other IgG subtypes or total IgG were noted. Transfer of pHrodo–S. pneumoniae⁺ B cells did not increase anti-PPS or nonspecific IgM to levels higher than other B cell transfers after 7 d either.

The earlier-mentioned experiments suggest that phagocytic B cells initiate adaptive immune responses in the mouse body. Those responses are considered to develop in secondary lymphoid tissues, such as the spleen. Assuming that phagocytic B cells also enter those tissues and interact with Th cells there, we examined the distribution of S. pneumoniae–phagocytosing B cells in the liver and spleen of recipients 7 d after i.v. injection of liver pHrodo–S. pneumoniae⁺IgM⁺ cells labeled with CellTrace Violet. Fluorescence microscopy detected their distribution in the T cell area and B cell follicles of the spleen, although too few B cells were detected to assess the trend in their local existence (Fig. 6A). Traced B cells were also distributed in the liver of recipients, albeit apart from T cells. Hepatic T cells were scarcely distributed, but they at least did not form clusters even after adaptive transfer of phagocytic B cells (Supplemental Fig. 3A).

We next evaluated the ability of splenic and hepatic CD4⁺ T cells to amplify humoral immune responses by measuring their cytokine production when cultured with liver IgM⁺ B cells ingesting heat-killed S. pneumoniae. Before this experiment, we obtained liver or spleen helper CD4⁺ T cells from mice that had been primed with heat-killed S. pneumoniae 7 d earlier. Thereafter, those CD4⁺ T cells were cultured for 4 d with liver IgM⁺ B cells from wild-type mice in medium containing heat-killed S. pneumoniae. We found that spleen, but not liver, CD4⁺ T cells from those primed mice produced certain amounts of IL-2, and this IL-2 production was diminished by adding cytochalasin D (Fig. 6B). Those primed splenic CD4⁺ T cells more actively proliferated than did the liver CD4⁺ T cells when cultured with liver B cells and heat-killed S. pneumoniae (Fig. 6C). In addition, spleen CD4⁺ cells also produced detectable levels of IL-21 when they were cultured under the same conditions (Fig. 6D). These results suggest that spleen, but not liver, T cells engage in acquired immune responses on encountering phagocytic B cells.

We next examined the effect of S. pneumoniae–phagocytic liver B cells on the splenic Th cell activation. When phagocytic or nonphagocytic liver B cells for pHrodo–S. pneumoniae were cultured with the primed spleen CD4⁺ T cells, the pHrodo–S. pneumoniae⁺ liver B cells stimulated spleen CD4⁺ cells to produce a larger amount of IL-2 than pHrodo–S. pneumoniae⁻ B cells (Fig. 7A). The pHrodo–S. pneumoniae⁺ B cells induced proliferation of the primed CD4⁺ cells more efficiently than the nonphagocytic B cells (Fig. 7B). The pHrodo–S. pneumoniae⁺ B cells also induced more IL-6, IL-12, and IFN-γ production from the spleen CD4⁺ cells than the nonphagocytic B cells, while they induced neither IL-4, IL-5, nor TGF-β production (Fig. 7A). These results suggest that phagocytic liver B cells potently activate splenic Th cells to produce Th1 cytokines. These cytokine productions by primed CD4⁺ cells were also observed on culture with liver B cells positive for pHrodo–E. coli (Fig. 7A). Compared with pHrodo–S. pneumoniae⁺ B cells, pHrodo–E. coli⁺ B cells induced substantial IL-6 production, whereas their IL-2 production (Fig. 7A) and induction of T cell proliferation (Fig. 7B) were not as remarkable as that of pHrodo–S. pneumoniae⁺ B cells.

Previous findings of acquired immune responses for soluble Ags have revealed that B cells specifically binding to Ags and encountering their cognitive T cells is a prerequisite for their clonal expansion (10, 11). These responses occur in secondary lymphoid tissues, such as the spleen, where B cells proliferate and differentiate into Ab-secreting cells after capturing soluble Ags. For the particulate Ags of S. pneumoniae, spleen B cells did not apparently induce them to undergo processing after incorporation (Fig. 3A, 3C); however, whether clonal expansion occurs in phagocytic B cells that digest the engulfed bacteria with acidification is unclear. We compared the functions between these liver and splenic B cells under coculture with splenic Th cells, focusing on their potential for Ab secretion. For this assessment, we enriched B cells from spleen MNCs to culture with FITC–S. pneumoniae and sorted IgM⁺ cells as FITC–S. pneumoniae⁺ or FITC-S. pneumoniae⁻, according to the gating strategy (Supplementary Fig. 3B). We then compared them with pHrodo–S. pneumoniae⁺ liver B cells. We examined the gene expressions of enzymes important for somatic hypermutation and class switching, such as AICDA1, AICDA2, UNG, and APE1, in the phagocytic liver B cells (pHrodo–S. pneumoniae⁺IgM⁺ cells) and spleen B cells merely capturing bacteria (FITC–S. pneumoniae⁺IgM⁺ cells) after culture for 5 d with spleen CD4⁺ T cells of the primed mice (Fig. 8A). Neither pHrodo–S. pneumoniae⁺, pHrodo–S. pneumoniae⁻, nor pHrodo–E. coli⁺ liver B cells showed a significant expression of these mRNA. In contrast, FITC–S. pneumoniae⁺ spleen B cells showed a significant expression of these mRNA compared with FITC–S. pneumoniae⁻ spleen B cells. In the setting of this experiment, we also found that pHrodo–S. pneumoniae⁺ liver B cells scarcely induced IL-21 production from the spleen CD4⁺ T cells, while FITC–S. pneumoniae⁺ spleen B cells induced much more marked IL-21 production (Fig. 8B).

Finally, we transferred these pHrodo–S. pneumoniae⁺ liver B cells or FITC–S. pneumoniae⁺ spleen B cells into B cell–deficient muMT mice (Supplemental Fig. 4A) and determined their humoral immune responses. Transfer of these B cells increased the total IgM level in the plasma of recipient muMT mice (Supplemental Fig. 4B). In contrast, anti-PspA IgG was detected in those that received FITC–S. pneumoniae⁺IgM⁺ spleen B cells, albeit at low levels (14 d after transfer; (Fig. 8C). These findings suggest that unlike spleen B cells, which were merely incorporating, but not processing, S. pneumoniae, phagocytic liver B cells per se did not take part in either somatic hypermutation or isotype switching.

This study showed that mouse liver B cells phagocytose S. pneumoniae in the presence of their serum. Unlike spleen B cells, liver B cells digest these bacteria in their phagolysosomes with acidification, which enables the detection and sorting of those B cells phagocytosing S. pneumoniae coated with pHrodo reagents (pHrodo–S. pneumoniae⁺IgM⁺ cells). We confirmed their initiation of humoral immune responses via adoptive transfer experiments, in which the recipient naive mice showed higher titers of Ag-specific Abs in the plasma than mice transferred with nonphagocytic B cells (pHrodo–S. pneumoniae⁻IgM⁺ cells) or B cells phagocytosing other bacteria (pHrodo–E. coli⁺IgM⁺ cells).

In in vitro experiments using spleen CD4⁺ T cells obtained from mice that had been administered heat-killed S. pneumoniae 7 d before, bacterium-phagocytic liver B cells potently activated spleen Th cells primed with S. pneumoniae. However, these B cells did not express mRNA of enzymes important for somatic hypermutation or isotype switching under coculture with Ag-primed spleen CD4⁺ T cells. Bacterium-phagocytic liver B cells may activate Th cells, leading to the effective production of Ag-specific Abs in mice. However, an in vitro culture study showed that these phagocytic liver B cells were unable to effectively produce Ag-specific Abs by themselves, even with Ag-primed spleen CD4⁺ T cells. Conversely, spleen B cells, which would not process the incorporated bacterium for sterilization, may effectively proliferate or differentiate into Ab-secreting B cells.

S. pneumoniae is coated with a gelatinous capsule of PPS to protect against phagocytosis by host cells (24), but this structure is deactivated by opsonization with fresh serum containing Abs and complements (15, 16). We found that the cultured liver B cells effectively phagocytosed S. pneumoniae by adding fresh mouse serum. This enhancing phagocytosis is mainly attributed to the opsonization by complement, because neither heat-inactivated serum nor serum with anti-C3 Ab added increased the phagocytic population (Fig. 1A). Therefore, complement appears to play a central role in the opsonization of S. pneumoniae in the humoral immune systems, allowing these bacteria to be efficiently incorporated by professional phagocytes and B cells (25, 26).

In this study, we found that liver B-1 cells had a markedly higher phagocytic activity than liver B-2 cells, as well as peritoneal B-1 cells, which are known to have potent phagocytic activity. We cannot completely deny the possibility that peritoneal B cells contaminated the liver or migrated to the liver, especially the latter, because peritoneal B cells had a much higher percentage of B-1 subtypes than liver B cells. Nevertheless, we speculate that most liver B-1 cells are distinct from peritoneal B-1 cells, because we thoroughly washed the livers with PBS to avoid contamination of peritoneal cells, and the expression of CD11b and IgD in liver B-1a and B-1b cells differed from that in their peritoneal counterparts (Fig. 2B, 2C).

We also uncovered that these phagocytic B cells preferentially produced anti-PPS IgM by themselves, findings that are in agreement with those of Souwer’s study (8), where human B cells forced to ingest Salmonella typhimurium also produced IgM specific for their LPS. Likewise, So et al. (7) showed that human B cells ingesting Neisseria gonorrhea produced certain amounts of gonococcal-specific IgM, although IgMs specific for irrelevant Ags were produced in greater quantities. They suggested that this broad and polyspecific Ig response may be attributed to bacterial degradation in B cell bodies, which allows for innate immune detection through TLR9 to drive their polyclonal IgM production (6). Similarly, in our study, specificity for PPS among totally produced IgM was lower in the phagocytic population than the nonphagocytic population, likely because phagocytic B cells obtain larger amounts of PPS, which may induce them to produce IgM nonspecifically by cross-linking these receptors multivalently (18).

Up to the present time, few researchers have focused on the effects of B cell phagocytosis on adaptive immune responses. Souwer et al. (8) showed an efficient Ag presentation to CD4⁺ T cells by B cells after internalization of Salmonella. Recently, Martínez-Riaño et al. (27) studied that B cell phagocytosis of particulate Ags induces humoral responses by revealing that Rho^−/− mice, whose B cells have decreased phagocytic activities, have lower levels of high-affinity class-switched Abs in serum than wild-type mice. We demonstrated in this study that adoptively transferred phagocytic B cells increased the plasma titers of anti-PspA IgG to levels in recipient mice higher than those induced by nonphagocytic B cells. Protein Ag of PspA undergoes degradation in phagolysosomes of APCs for Ag presentation to helper CD4⁺ T cells, and our results suggested the possibility that phagocytic liver B cells would degrade the body of S. pneumoniae to functionally present PspA derivatives and initiate humoral immune responses.

In fact, some elevation of the anti-PspA IgG level was observed in plasma of recipient mice on the transfer of B cells ingesting E. coli, which may have caused confusion regarding the specificity of Ab acquired by transfer of B cells that had ingested S. pneumoniae. However, we assume that the elevation of the anti-PspA IgG level induced by S. pneumoniae–ingesting B cells was due to their Ag presentation and not to other mechanisms, such as induction of inflammatory cytokines, based on in vitro experiments showing that they induced proliferation of primed splenic CD4⁺ T cells and IL-2 production more efficiently than B cells ingesting E. coli (Fig. 7).

We observed significant increases of IgG2a and IgG2b, but not of IgG1, among IgG subtypes specific for PspA in plasma of recipients adoptively transferred with phagocytic B cells. Those B cells, cultured with the primed CD4⁺ T cells, induced a higher production of IFN-γ but lower production of IL-21 than nonphagocytic B cells. Phagocytic B cells induced little IL-4 or TGF-β production by these primed Th cells. Among these cytokines, IFN-γ may also be produced in the lymphoid tissues of recipient mice to promote isotype switching to IgG2a, but not IgG1, in collaboration with thymus-independent Ags of PPS (28–31).

In spleen, Ag-binding B cells meet T cells at the border between the T cell area and the B cell follicle. Some T cells, especially FO T cells, produce IL-21 in response to IL-6 (32), which enhances B cell proliferation and differentiation (33). In isotype switching, IL-21 has been shown to promote Ag-specific IgG1 production (34, 35). Our in vitro experiment revealed that bacterium-phagocytic liver B cells induced the production of certain amounts of IL-6 but little IL-21 from spleen CD4⁺ T cells. In addition, we detected no gene expression for enzymes important for somatic hypermutation or isotype switching of AICDA, UNG, or APE1 in those B cells after culture with primed spleen CD4⁺ T cells. Furthermore, muMT mice adoptively transferred with bacterium-phagocytic liver B cells did not show increased levels of anti-PspA IgG at all, suggesting that bacterium-phagocytic B cells would not be dedicated in somatic hypermutation or isotype switching.

In contrast, spleen B cells merely capturing S. pneumoniae (FITC–S. pneumoniae⁺IgM⁺ cells) consistently showed gene expressions for enzymes involved in affinity maturation at higher levels than seen in noncapturing B cells, suggesting their potential ability to initiate this process in lymphoid tissues. Those B cells induced more IL-21 from CD4⁺ T cells than those not capturing the bacteria in vitro, as well as detectable levels of anti-PspA IgG in the plasma of recipient muMT mice.

Why liver and spleen B cells have different functions, both in innate and acquired immunities, remains unclear, but it may be due in part to differences in their B cell composition, such as there being fewer MZ B cell subtypes producing immunosuppressive cytokines in the liver than in the spleen (36). Because some researchers have shown that B cell lymphopoiesis was negatively regulated by cathepsin L (37), their lysosomal formation itself may have had some influence on their response toward the production of Ag-specific Abs. Further studies are required to clarify the mechanism underlying the differences in the immunological functions of these cells.

In conclusion, our study showed that liver B cells phagocytosing S. pneumoniae may perform Ag presentation to CD4⁺ T cells in secondary lymphoid tissues with Th1 cytokine production, thereby leading to an increase in IgG specific for the Ags. However, these bacterium-phagocytosing B cells per se did not take part in either somatic hypermutation or isotype switching.

The authors have no financial conflicts of interest.