T-bet⁺ B cells Dominate the Peritoneal Cavity B Cell Response During Murine Intracellular Bacterial Infection

Abstract

T-bet⁺ B cells have emerged as a major B cell subset associated with both protective immunity and immunopathogenesis. T-bet is a transcription factor associated with type I adaptive immune response to intracellular pathogens, driving an effector program characterized by the production of IFNγ. Murine infection with the intracellular bacterium, E. muris, generates protective extrafollicular T cell-independent T-bet⁺ IgM-secreting plasmablasts, as well as T-bet⁺ IgM memory cells. Although T-bet is a signature transcription factor for this subset, it is dispensable for splenic CD11c⁺ memory B cell development, but not for class switching to IgG2c. In addition to the T-bet⁺ plasmablasts found in the spleen, we show that antibody secreting cells can also be found within the mouse peritoneal cavity; these cells, as well as their CD138-negative counterparts, also expressed T-bet. A large fraction of the T-bet⁺ peritoneal B cells detected during early infection were highly proliferative, expressed CXCR3 and CD11b, but unlike in the spleen, they did not express CD11c. T-bet⁺ CD11b⁺ memory B cells were the dominant B cell population in the peritoneal cavity at 30 days post-infection, and although they expressed high levels of T-bet, they did not require B cell-intrinsic T-bet expression for their generation. Our data uncover a niche for T-bet⁺ B cells within the peritoneal cavity during intracellular bacterial infection, and identify this site as a reservoir for T-bet⁺ B cell memory.

Introduction

Immune responses to infection generate antibody-secreting effector cells and B cell memory, which together facilitate antibody responses upon re-exposure to the same antigenic stimulus. A canonical B cell response requires T and B cell cognate interactions, wherein T cell-dependent antigens, primarily proteins, activate germinal center reactions within or near secondary lymphoid organ B cell follicles (1). These germinal centers produce memory and long-lived plasma cells, favoring the selection of high-affinity affinity matured B cell receptors (BCRs). It has become increasingly clear that the generation of T-bet⁺ antibody secreting cells (ASCs) and memory B cells (MBCs) is a normal and effective response during chronic, repeated, or antigenically-variable pathogen challenges (2). Indeed, MBCs expressing the transcription factor T-bet have been identified following viral, intracellular bacterial, and parasitic infections, and have been associated with autoimmune diseases and aging (3–5). The distribution and function of T-bet⁺ B cells, also known as Age-related B cells (ABCs), has been studied during viral infections, but less is known as to how these B cells respond during intracellular bacterial infections.

Our laboratory has identified a population of T-bet⁺ CD11c⁺ MBCs that are generated in the apparent absence of germinal centers following infection with the intracellular bacterium Ehrlichia muris (6, 7). Studies of E. muris infection have demonstrated an essential role for humoral immunity in host defense, mediated by T-bet⁺ plasmablasts (PBs) and T-bet⁺ MBCs (6, 8). It has been proposed that IFNγ-driven T-bet expression confers MBCs with an antibody-secreting “poised” epigenetic state, arming them with the capacity to generate robust responses to secondary challenge and to contribute to autoimmune pathogenesis (9–11). These studies suggest that ASC generation during the type I recall response is dependent upon B cell-intrinsic T-bet expression.

T-bet⁺ B cells have been studied extensively in secondary lymphoid organs such as the spleen (12, 13), as well as in the peripheral blood (14), and liver (15). However, peritoneal T-bet⁺ B cells have not been described. B cell subsets within the peritoneal cavity include B1 B cells, which possess broad antigen reactivity, in some cases to self-antigens, and a smaller subset of B2 B cells. The BCR repertoires of B1 B cells are specialized for protection against challenge with microbial or T cell-independent host antigens such as bacterial polysaccharides, outer membrane porin proteins, or cellular debris (16). The classification of B1 B cells is complicated by their selective expression of definitive surface markers, and is dependent upon cell locale and activation state, a topic reviewed extensively elsewhere (17). Peritoneal cavity B1 B cells are commonly categorized by the expression of CD11b, and the inhibitory glycoprotein CD5. B1a and B1b B cells can be distinguished by the presence or absence of CD5, respectively. Although CD5 expression and Toll-like receptor (TLR) signals have been shown to be required for the activation and differentiation of B1 cells, subsequent re-organization of the BCR can lead to the downregulation of CD5 expression and increased BCR signaling (18). The same TLR signals, coupled with BCR activation, have been shown to drive the expression of T-bet in B cells (19, 20). These findings raise the possibility that during intracellular bacterial infection, T-bet⁺ B cells are part of the B1 B cell response.

In this study we show that intracellular bacterial infection elicits a peritoneal cavity B cell response characterized by T-bet expression, high levels of proliferation, and expression of some phenotypic markers characteristic of B1 B cells. Kinetic analysis of the peritoneal B cell population revealed a shift in the ratio of CD5-positive and -negative B cells, concomitant with T-bet⁺ B cell expansion. We also show that the peritoneal cavity is a reservoir for a persistent population of T-bet-expressing CD11b⁺ memory B cells. These peritoneal cavity T-bet⁺ memory B cells did not require B cell-intrinsic T-bet for their generation, although B cells that lacked T-bet were CXCR3-deficient. Our findings reveal the murine peritoneal cavity to be a site of T-bet⁺ B cell expansion during infection, as well as a site of T-bet⁺ MBC persistence. Our study provides important new information regarding the dynamic changes that occur within the peritoneal cavity during ehrlichial infection, and suggests that T-bet⁺ B cells contribute to control of intracellular bacterial infection at this site.

Materials and Methods

Mice.

C57BL/6J, Tbx21^fl (B6.129-Tbx21^tm2Srnr/J), Mb1^cre (B6.C(Cg)-Cd79a^tm1(cre)Reth/EhobJ), and Rosa26^eYFP (B6.Cg-Gt(ROSA)26Sor^{tm3(CAG-EYFP)Hze}/J) mice were obtained from The Jackson Laboratory (Bar Harbor, ME). T-bet-cre^ERT2 mice were generated by Dr. Lin Gan at the University of Rochester, Rochester, NY. All mice were housed in a specific pathogen-free environment in the SUNY Upstate Medical University Animal Care Facility (Syracuse, NY) in accordance with federal guidelines and with approval of the Institutional Animal Care and Use Committee. All mice used for experiments were at least 6 weeks old, with male and female mice used excluding the transfer experiments which used tamoxifen and therefore only female mice.

Bacterial infection and quantitation.

Mice were infected intraperitoneally (i.p.) with 5x10⁴ copies of E. muris, as previously described (21). Quantitative real-time polymerase chain reaction (PCR) was used to determine the E. muris copy number in infected liver tissue to confirm infection (22). The PCR products were analyzed with an Applied Biosystems Step-One Real-Time PCR System (Applied Biosystems, Foster City, CA). The copy number of the E. muris genus-specific disulfide bond formation protein (dsb) gene was determined using known quantities of dsb amplicon as standards. We have made the simplifying assumption that bacterial copy number and numbers of viable bacteria were equivalent in our experimental model.

Peritoneal cavity wash and parallel adoptive transfer.

Peritoneal cavity cells were collected by washing the cavity with 10 mL sterile wash buffer with a 12 mL syringe and 18G needle, agitation of the abdomen, and retrieval of fluid back into the syringe. Cells were washed and resuspended with sterile wash buffer, counted on a particle counter (Beckman Coulter), and transferred i.p. to mice synchronously infected with E. muris. Transfer was validated following negative selection of peritoneal macrophages by FACS or culture in flasks without appreciable differences in bacterial load, but performed for the reporter experiments shown with whole suspension to ensure maximum cell viability.

BrdU administration and staining for flow cytometry.

5’-bromo-2’-deoxyuridine (BrdU) detection of proliferation was performed as previously described (6). Briefly, mice were administered 0.8 mg of BrdU by i.p. injection, and maintained on BrdU in drinking water (0.8 mg/mL, plus 10% dextrose), ad libitum, from days 4-8 post-infection. Peritoneal cavity wash cells were stained for surface markers, and fixed using a BD Cytofix/Cytoperm solution, for 30 minutes at 4°C. Prior to pelleting the cells, BD Perm/Wash solution was added. Cells were then resuspended in 10% dimethyl sulfoxide (DMSO) in BD Perm/Wash solution and incubated for 10 minutes at 4°C. Cells were then fixed again with BD Cytofix/Cytoperm solution for 5 minutes at 4°C. DNase I in 1x Phosphate Buffered Saline (PBS) was then incubated on the cells for 1 hour at 37°C. The cells were then stained with FITC-conjugated anti-BrdU (Bu20a) diluted in BD perm/wash solution for 30 minutes at room temperature. Cells were then washed and analyzed as below.

Flow cytometry and antibodies.

Cells were treated with anti-CD16/32 (2.4G2), stained with fixable viability Live/Dead stain in Aqua or Far Red (Invitrogen), and washed prior to incubation with the antibodies listed in Supplementary Table 1. The cells were stained at 4°C for 30 min, washed, and analyzed. For intracellular staining, surface-stained cells were fixed/permeabilized for 50 minutes at 4°C using BD Pharmingen Transcription Factor Buffer set Fixation/permeabilization buffer, washed, stained at 4°C for 30 minutes and analyzed. Unstained cells were used to establish the flow cytometer voltage settings, and single-color positive controls were used to adjust compensation. Data were acquired on a BD Fortessa flow cytometer with Diva software (BD bioscience), and were analyzed with FlowJo software for Mac v. 10.7.1 (Becton Dickinson and Co.). FlowSOM (23) and Cluster Explorer plugins were provided by FlowJo via their website.

Sorting and transcriptional analysis of peritoneal cavity B cells.

CXCR3⁺ and CXCR3-negative live singlet IgM⁺ IgD^−neg CD19⁺ CD138^−neg peritoneal cavity B cells were sorted directly into RLT+ buffer (Qiagen, catalog #1053393) on a BD FACS Aria II, and snap-frozen on dry ice. The FACS gating strategy is included in Supplementary Figure 1a. Total RNA was extracted using the standard Illumina preparation kit according to the manufacturer’s recommendations, and quality and quantity assessed on an Agilent Bioanalyzer (Agilent, Santa Clara, CA). Library preparation was performed using the Illumina Library Preparation kit according to the manufacturer’s protocol. Libraries were sequenced on an Illumina NextSeq 550 instrument using the NextSeq 500/550 Mid Output Kit v2.5 (150 Cycles). Sequencing parameters were set for 150 cycles, 75 cycles for each paired-end read. Prior to sequencing, library quality and concentration were assessed using an Agilent 4200 TapeStation. All extraction, quality control, and sequencing was performed at the SUNY Upstate Molecular Analysis Core. The results of RNA-Seq have been deposited in the NCBI Gene Expression Omnibus database under accession code GSE190651 at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE190651.

Gene expression analysis/visualization.

Sample demultiplexing was performed within the BaseSpace Sequence Hub (Illumina, CA) according to the manufacturer’s recommendations, using the default settings. Gene expression pseudoalignment was performed using Kallisto (24) following quality assessment using FastQC (25) and MultiQC (26). Transcript pseudoalignment was performed against a mouse reference library using Kallisto and the Ensembl GRCm39 cDNA FASTA (27). Data normalization, visualization, and differential gene expression were performed using the statistical computing environment R version 3, RStudio version 1.3.1093 (28) and Bioconductor (29). Transcript-level counts and abundance were summarized to genes using the TxImport package (30). Data were filtered and normalized using the Trimmed Mean of M-values (TMM) method in EdgeR package (31). Differential gene expression analysis was performed using linear modeling and Bayesian statistics in the R package Limma with Benjamini-Hochberg correction for multiple testing (32). Gene Set Enrichment Analysis (GSEA) was carried out using the GSEAbase package (33). For the anti-IgM-stimulation gene set, the GSEA plot compares gene set (CXCR3⁺ vs CXCR3^−neg) expression enrichment against a publicly available dataset, which compared control vs. 12 hr. anti-IgM stimulated murine B cells (GSE13547_3163_200_UP) (34). GO analysis was carried out using the PANTHER Overrepresentation test (release 20210224) using all genes in the Mus musculus upload_1 GO Ontology database (35), Fisher’s exact test and Bonferroni correction.

Statistical analysis.

Statistical analysis was performed using Prism version 9.1.0 (GraphPad Software, La Jolla, CA). Statistical significance was represented as shown, where ns indicates p > 0.05, *: p < 0.05, **: p < 0.01, ***: p < 0.001, and ****: p < 0.0001. Individual analyses performed and sample size (n) are indicated in the figure legend. Statistical tests were chosen based on descriptive statistics for the specified dataset. The column in each column plot indicates the arithmetic mean of the dataset and upper and lower bounds indicate standard deviation of the dataset. For absolute cell counts, frequencies among leukocytes were used to determine absolute counts from 4–16 um particle counter values. For GSEA, the false discovery rate (FDR) q value denotes the estimated probability that the normalized enrichment score represents a false positive finding, while the normalized enrichment score (NES) means the enrichment score for the gene set after normalization across all analyzed gene sets.

Graphics

The working model and cell transfer schematics were created with BioRender.com.

Results

The peritoneal cavity is a site of T-bet⁺ B cell responses during early infection

We have demonstrated that Ehrlichia muris infection generates both CD11c⁺ T-bet⁺ plasmablasts and memory B cells (6, 8). Our earlier published studies indicated that the T-independent CD11c⁺ PBs detected in the spleen on day 10 post-E. muris infection had characteristics of both marginal zone (MZ) and B1 B cells, and produced the majority of antigen-specific splenic IgM, and nearly all antigen-specific IgM (8). Studies from our laboratory also demonstrated that the protective anti-E. muris IgM was highly polyreactive, and independent of both MZ-characteristic Notch signaling and secondary lymphoid organ development (36). Other studies have demonstrated that peritoneal cavity B1 cells can migrate to the spleen and become antibody-secreting cells (37). These observations suggested that peritoneal cavity B1 cells participate in the E. muris B cell response. We therefore examined the peritoneal cavity to determine if B1 B cells expressed T-bet during E. muris infection.

Peritoneal cavity B cells detected during E. muris infection were found to express T-bet, and these peritoneal B cells, which expressed B220⁺ IgM^hi IgD^−neg CD23^neg CD11b⁺, exhibited higher intracellular expression of T-bet than ASCs present at the same time post-infection (Figure 1a–c), consistent with our previous observations of splenic B cell and ASC populations (38). Although the T-bet⁺ B220⁺ IgM^hi IgD^−neg CD23^neg CD11b⁺ cells may be of B1 origin, definitive proof will require additional studies (18). To address when T-bet expression is initiated in peritoneal B cells, we assessed the kinetics of cellularity and T-bet expression within the peritoneal cavity and compared this to the spleen. While the absolute number of T-bet⁺ B cells was consistently higher in the spleen, due to the higher cellularity of the organ (Figure 1d–i), the frequency of T-bet⁺ B cells among leukocytes increased over time post-infection in both the peritoneal cavity and the spleen (Figure 1j, k). These data reveal the peritoneal cavity to be a site of a T-bet⁺ B cell response, implicating T-bet-expressing B cells as participants in the early T-independent response to E. muris.

Figure 1: Activated B cells and ASCs in the peritoneal cavity express T-bet during intracellular bacterial infection.

(A-C): Peritoneal cavity cells from E. muris-infected C57BL/6 mice were analyzed on day 8 post-infection. Plots show T-bet expression among CD11b⁺ CD19^hi IgM^hi, IgD-negative Dump (CD4/CD8/Ter119/Gr-1/NK1.1)-negative live leukocytes (A), and CD138⁺ cells among live leukocytes (B). (C) The histogram shows mode-normalized frequency vs. fluorescent intensity for anti-T-bet staining. The actPBC T-bet⁺, T-bet^−neg and ASC gates include 12013, 6508, and 535 events, respectively. Representative data are shown for 3 biological replicates.

(D-E): Cells from E. muris-infected mice were enumerated on days 0, and 4-9 post-infection from the spleen (D) and peritoneal cavity (E). Each datum represents an individual mouse. Differences between timepoint groups were assessed by One-way ANOVA, alpha = 0.05. p = 0.1531 (D), p <0.0001 (E).

(F-G): Cells were analyzed on days 0, and 4-9 post-infection. Graphs show the number of CD19 and/or B220⁺ dump-negative live leukocytes from the spleen (F) and peritoneal cavity (G). Each dot represents an individual mouse. Differences between timepoint groups were assessed by One-way ANOVA, alpha = 0.05. p = 0.0074 (F), p = 0.0325 (G).

(H-K): Cells were analyzed on days 0, and 4-9 post-infection. The plots show the number of T-bet⁺ CD19 and/or B220⁺ cells among live dump-negative leukocytes from the spleen (H) and peritoneal cavity (I). Frequencies of the same population are shown for the spleen (J) and peritoneal cavity (K). Each dot represents an individual mouse. Differences between timepoint groups were assessed by One-way ANOVA, alpha = 0.05. p <0.0001 (H), p =0.0006 (I), p <0.0001 (J), p = 0.0007 (K). ns, p > 0.05, *p < 0.05, **p < 0.01, ****p < 0.0001.

The frequency of T-bet⁺ B cells was inversely associated with B cell CD5⁺ expression in the peritoneal cavity

B1 cells are typically categorized on the basis of CD5 expression, where B1b cells lack CD5, and B1a cells are CD5-positive (17). To determine if T-bet was expressed preferentially in CD5-negative or CD5⁺ B cells, we examined the frequency of CD5-negative and -positive T-bet⁺ B cells in the peritoneal cavity in the same mice shown in Figure 1. Although T-bet⁺ B cells were largely absent prior to day 8 post-infection in the spleen (Figure 1h, j), the peritoneal cavity contained a substantial population of activated T-bet⁺ B cells within 5 days post-infection (Figure 2a, b). Moreover, although the number of CD5⁺ and CD5-negative activated B cells followed a similar trend over time, the ratio of CD5⁺ to CD5-negative peritoneal cavity B cells decreased as infection progressed (Figure 2c–e). The proportion of T-bet⁺ B cells expressing CD5 also decreased over time, while the frequency of CD5-negative activated B cells expressing T-bet dramatically increased during the same period (Figure 2f–g). These findings suggest that T-bet expression accompanies the expansion or local recruitment of CD5-negative B cells, or leads to the downregulation of CD5.

Figure 2: Expression of T-bet among activated peritoneal cavity B cells coincides with a loss of CD5.

(A-B): Peritoneal cavity cells from E. muris-infected WT mice were analyzed on days 0, and 4-8 post-infection. Plots show the number (A) and frequency (B) of T-bet⁺ cells among CD11b⁺ or CD43⁺ CD19⁺ IgM^hi, IgD-negative live leukocytes. Each dot represents an individual mouse. Differences between timepoint groups were assessed by One-way ANOVA, alpha = 0.05. p = 0.0015 (A), p < 0.0001 (B).

(C-D): Peritoneal cavity cells were analyzed on days 0, and 4-8 post-infection. Plots show the number of CD5⁺ (C) and CD5-negative cells (D), or mean frequency of CD5⁺ and CD5-negative cells (E), among CD11b⁺ or CD43⁺ CD19⁺ IgM^hi, IgD-negative live leukocytes. Each dot represents an individual mouse. Differences between timepoint groups were assessed by One-way ANOVA, alpha = 0.05. p = 0.009 (C), p = 0.0231 (D), p <0.0001 (E).

(F): Peritoneal cavity cells were analyzed on days 0, and 4-8 post-infection. The plot shows the mean frequency of CD5⁺ and CD5-negative cells among T-bet⁺ CD19⁺ live leukocytes. Differences between timepoint groups were assessed by One-way ANOVA, alpha = 0.05. p = 0.0014 (CD5⁺ and CD5^−neg).

(G): Peritoneal cavity cells were analyzed on days 0, and 4-8 post-infection. The plot shows the proportion of T-bet⁺ cells among CD5⁺ and CD5-negative CD11b⁺ or CD43⁺ CD19⁺ IgM^hi, IgD-negative live leukocytes. Differences between timepoint groups were assessed by One-way ANOVA, alpha = 0.05. p = 0.0009 (CD5⁺), p <0.0001 (CD5^−neg). ns, p > 0.05, *p < 0.05, **p < 0.01, ****p < 0.0001.

(H-J): Peritoneal cavity cells were analyzed on day 8 post-infection. Cluster Explorer representation of FlowSOM analysis of CD11b⁺ CD19⁺ IgM^hi, IgD-negative live leukocytes by flow cytometry. Event (cell) number (E), T-bet and CD19 expression (F), and transformed relative expression levels of each marker among color-coded clusters (G) are shown. The data are representative of 3 biological replicates.

To characterize whether the CD5-negative B cells we observed in the peritoneal cavity were heterogeneous, we used the unsupervised clustering and visualization algorithm FlowSOM, which evaluates flow cytometry data using a self-organizing map (23). B cells from the peritoneal cavity were found as three phenotypic clusters. The majority of B cells on day 8 post-infection were found within one cluster of cells characterized by T-bet, CD11b, and B220 expression, but lacking CD5 (Figure 2h–j). This primary cluster did not differ appreciably from the other clusters in the expression of CD19, suggesting that paucity of CD5 expression was not due to CD19 deficiency in T-bet-expressing cells. This observation is important, because changes in CD19 signaling can have significant effects on the development and responsiveness of CD5-expressing B cells (39). Other phenotypic clusters identified included two T-bet-negative CD5-expressing populations, the larger of which expressed more B220, and a small cluster of T-bet- and CD5-negative B cells expressing intermediate levels of B220. Our data are consistent with the explanation that E. muris elicits a CD5-negative activated B cell response in the peritoneal cavity that coincides with an increase in B cell T-bet expression.

CXCR3⁺ T-bet-expressing B cells in the peritoneal cavity are transcriptionally distinct and are enriched for genes involved in cell proliferation

We next sought to determine whether T-bet⁺ and T-bet-negative B cells in the peritoneal cavity during E. muris infection are transcriptionally divergent, and to identify biological pathways that are differentially utilized in the T-bet⁺ B cells at this location. Therefore, we performed RNA-sequence analysis on activated peritoneal cavity CD19^hi IgM^hi IgD^−neg CD138^neg cells. Because intracellular staining for T-bet requires fixation, which can damage nucleic acid, we identified T-bet-expressing B cells using the cell surface marker CXCR3, an analysis which does not require fixation. Although not all T-bet⁺ B cells (roughly half) in the peritoneal cavity on day 8 post-infection expressed CXCR3 (Figure 3a), in ours and others’ studies effectively all CXCR3⁺ B cells have been shown to express T-bet (Supplementary Figure 1b–d) (13, 40). CXCR3⁺ and CXCR3-negative CD19^hi IgM^hi IgD^−neg CD138^neg cells isolated from the peritoneal cavity were enriched by flow cytometric cell sorting and mRNA transcripts were compared. A total of 4631 genes were differentially expressed between the cell populations on day 8 post-infection, using an adjusted p value below 0.05, between CXCR3⁺ and CXCR3-negative B cells. Genes critical for DNA replication, cell division, and metabolism were the most highly represented and most differentially expressed in the enriched gene set (Figure 3b), indicating that the CXCR3⁺ subset of B cells are actively proliferating.

Figure 3: T-bet⁺ CXCR3⁺ peritoneal cavity B cells are transcriptionally enriched for proliferation, metabolic activity, but not B1a characteristics on day 8 post-infection.

(A) Peritoneal cavity cells were analyzed on day 8 post-infection. Flow cytometric analysis of CXCR3 expression among T-bet⁺ peritoneal cavity B cells gated for singlet live dump-neg (CD4/CD8/Ter119/Gr-1/NK1.1) CD19^hi IgM^hi CD138-neg T-bet⁺ cells is shown. Data representative of 3 biological replicates are shown.

(B-D): RNA sequencing analysis of bulk-sorted CXCR3⁺ and negative CD19^hi IgM^hi IgD^neg CD138^neg cells isolated from the peritoneal cavity of 3 biological replicates per group (n = 6). Enrichment is interpreted as average gene expression in CXCR3⁺ - CXCR3^−neg.

(B) Gene Ontology analysis of DEGs was performed using the PANTHER Overrepresentation test. Redundant/subsets of family processes omitted for clarity.

(C) Gene Set Enrichment Analysis was run for enrichment against a publicly available dataset (GSE13547_3163_200_UP), comparing control vs. 12 hr. anti-IgM stimulated murine B cells.

(D) Volcano plot representation of DEGs between CXCR3⁺ and CXCR3-negative populations.

(E) Mice were treated with BrdU i.p. on day 4 post-infection; thereafter, BrdU was administered in drinking water ad libitum from days 4-8, at which time peritoneal cavity wash was collected. A flow cytometric analysis of anti-BrdU staining among CXCR3⁺ or CXCR3^−neg cells is plotted as the frequency of BrdU⁺ cells among the population shown. Both populations were pre-gated as live IgD/CD138/CD23^−neg CD19/IgM^hi CD11b⁺ cells. Each dot represents an individual mouse. Statistical significance was determined using two-tailed unpaired t tests. ns, p > 0.05, *p < 0.05, **p < 0.01, ****p < 0.0001.

To identify functional differences between the CXCR3⁺ and CXCR3-negative B cells, we next identified groups of genes commonly enriched between our dataset and publicly-available gene sets. Gene Set Enrichment Analysis revealed an antigen-stimulated transcriptional profile of the CXCR3⁺ population, and enhanced expression of Tbx21 (T-bet); moreover, the CXCR3⁺ population was highly enriched for genes also upregulated following BCR stimulation, relative to the CXCR3-negative B cell population (Figure 3c). These data suggest that the CXCR3⁺ T-bet⁺ peritoneal cavity B cells were stimulated by antigen and were differentiating into pre-plasmablasts, an event also observed at a later point in the E. muris-infected spleen and liver (8, 15).

Beyond the abundance of proliferation-associated genes, CXCR3⁺ B cells in the peritoneal cavity showed enhanced expression of genes known to be expressed following IFNγ signaling, including Ighg2c (IgG2c); (9, 41). Expression of genes involved in cell trafficking, such as Ccr6, and the integrin-encoding Itga4 (LFA-4 subunit), were also enriched in the CXCR3⁺ subset, indicating that this population may be migratory (Figure 3d). Numerous B1a cell-associated genes were enriched for expression in the CXCR3-negative population, including the inhibitory molecules Cd5 and Siglecg (42, 43). Additionally, expression of genes encoding canonical B1a immunoglobulin heavy chain variable sequences Ighv11-2 and Ighv12-3 was enriched in the CXCR3-negative B cell population. Enrichment for these sequences is known to be associated with binding to phosphatidylcholine, a component of cell membranes, and is a characteristic of natural antibody-producing B1a cells (44, 45). The CXCR3⁺ and CXCR3-negative populations did not, however, differ in expression of other genes commonly used to identify B1 cells, including Itgam (CD11b) and Spn (CD43), suggesting that these populations share surface markers due to a common environment or origin. These data implicate T-bet gene regulation in the apparent antigen stimulation-induced loss of canonical B1a peritoneal B cells, and/or in the preferential expansion CXCR3⁺ B cells.

To confirm the proliferative nature of the CXCR3⁺ peritoneal B cell population, we administered the thymidine analog bromodeoxyuridine (BrdU) to mice from days 4 through 8 following infection with E. muris, and assessed activated peritoneal cavity B cells on day 8. While very few CXCR3-negative B cells incorporated BrdU, nearly all CXCR3⁺ B cells were BrdU⁺ (Figure 3e), confirming that this population had undergone cell division early post-infection.

Early T-bet⁺ peritoneal cavity B cells were observed to emigrate from the infection site.

We identified expression of CXCR3, a known target gene of T-bet in B cells (13), on a large proportion of the T-bet⁺ B cells in the peritoneal cavity during infection. Given that CXCR3 endows lymphocytes with the capacity to traffic to sites of inflammation in response to its ligands, CXCL9, CXCL10, and CXCL11 (46), and because B1 cells with a similar surface phenotype can migrate from the peritoneal cavity to the spleen in response to stimulation (37, 47), we hypothesized that peritoneal cavity T-bet⁺ B cells were capable of migrating to the spleen, as a means of enhancing the response at that location.

To address this hypothesis, we utilized tamoxifen-inducible T-bet-Cre^ERT2 mice that were bred to mice expressing ROSA26-stop-flox EYFP (48). In this experimental model, all cells expressing T-bet at the time of tamoxifen administration are induced to irreversibly express enhanced yellow fluorescent protein (EYFP). These T-bet reporter mice were infected with E. muris and fed tamoxifen-containing chow from days 6-9 post-infection. Peritoneal cavity cells were harvested on day 9, transferred intraperitoneally to synchronously-infected wild-type mice, and assessed 40 hours later (Figure 4a). Although many of the EYFP-labeled T-bet⁺ B cells remained in the peritoneal cavity 2 days post-transfer, 33% of the transferred reporter cells detected had migrated to the spleen within this time period, after adjusting for the difference in cellularity between the anatomical sites (Figure 4c, d, Supplementary Table 2). These results suggest that some T-bet⁺ B cells migrate from the peritoneal cavity to the spleen during early E. muris infection, as occurs in response to LPS stimulation (37, 47). Thus, during E. muris infection, specialized T-bet⁺ B cell subsets may maintain residency within the tissue wherein they are activated, while others may contribute to the splenic response.

Figure 4: Some T-bet⁺ peritoneal cavity B cells migrated to the spleen during E. muris infection.

T-bet-cre^ERT2 x Rosa26^EYFP mice were fed tamoxifen chow from days 6-9 post-infection; on day 9 post-infection peritoneal cavity cells were analyzed by flow cytometry and transferred at a 1:1 ratio to the peritoneal cavity of mice infected on the same day. Two days later, the recipient mice were sacrificed and their B cells were analyzed by flow cytometry. Cells (3 × 10⁶) were stained from each site and analyzed to repletion on the cytometer. Correction for organ cellularity and sampling is shown in Supplementary Table 2.

(A) Flow cytometric analysis of T-bet expression among donor peritoneal cavity B cells on day 9 post-infection gated for singlet live dump-neg (CD4/CD8/Ter119) cells is shown. A representative of 2 biological replicates is shown.

(B) Flow cytometric analysis of T-bet expression among recipient peritoneal cavity B cells on day 11 post-infection gated for singlet live dump-neg (CD4/CD8/Ter119) cells is shown. A representative of 2 biological replicates is shown.

(C) Flow cytometric analysis of T-bet expression among recipient splenic B cells on day 11 post-infection gated for singlet live dump-neg (CD4/CD8/Ter119) cells is shown. A representative of 2 biological replicates is shown.

T-bet-expressing CD11b⁺ memory B cells dominate the peritoneal cavity B cell pool during chronic E. muris infection.

The observation that T-bet⁺ B cells remained in the peritoneal cavity following cell transfer led us to ask whether these cells were maintained as memory cells following resolution of acute infection. We have previously reported that T-bet⁺ plasmablasts preceded the generation and long-term maintenance of T-bet⁺ memory B cells in the spleen during E. muris infection (38). We therefore assessed the peritoneal B cell population 30 days following E. muris infection. We first observed that the CD5⁺ B cell compartment did not fully re-establish itself as the most prevalent peritoneal cavity B cell subset following resolution of acute infection. Instead, the peritoneal cavity B cell pool consisted primarily of T-bet⁺ CD11b⁺ PD-L2⁺ CD80⁺ CD73⁺ cells with a MBC phenotype (Figure 5a). These cells are nearly identical to the T-bet⁺ memory B cells we and others have described in the spleen and liver during E. muris infection (6, 15). One notable difference was that T-bet⁺ MBC in the peritoneal cavity did not express CD11c, a characteristic marker of splenic T-bet⁺ MBC (49). This difference in CD11 expression has been noted in other studies, and has been suggested to reflect the influence of peritoneal cavity-restricted activation or maturation on B cells (50, 51).

Figure 5: The peritoneal cavity is a site of T-bet⁺ MBC and ASC residence during chronic E. muris infection.

(A) Peritoneal cavity cells from E. muris-infected WT mice were analyzed on day 30 post-infection. Representative dot plots show gating of T-bet⁺ and T-bet-negative B cells pre-gated as singlet, live, B220 and/or CD19⁺ cells. The histograms show mode-normalized frequency vs. fluorescence intensity for the marker indicated. Representative data are shown for 3 biological replicates.

(B) Singlet, live peritoneal cavity cells from day 5 and day 30 post-infection were analyzed for expression of CD5 following gating for T-bet⁺ B cells as gated in (A). The graph shows mean fluorescent intensity for CD5 staining. Each dot represents an individual mouse. Statistical significance was determined using two-tailed unpaired t tests. Representative data are shown for two experiments. ns, p > 0.05, *p < 0.05, **p < 0.01, ****p < 0.0001.

(C) Singlet, live peritoneal cavity cells from day 30 post-infection were analyzed for expression of CD138 and sub-gated by CD5 expression. The histogram shows mode-normalized frequency vs. fluorescent intensity for anti-T-bet staining. The data are representative of 3 biological replicates.

(D) Singlet, live peritoneal cavity cells from naïve mice (left plots) and day 30 post-infection (right plot) were analyzed for expression of CD138 (ASCs), B220 and CD19 (B cells), T-bet⁺ among B cells (T-bet⁺ B cells), IgD^−neg CD23^−neg CD138^−neg IgM⁺ CD11b⁺ B cells (ActPBCs), and T-bet⁺ among actPBCs (T-bet⁺ actPBCs).

CD5 expression was observed within a subset of the T-bet⁺ B cell population, and this subset, making up an average of 7.9% of the T-bet⁺ B cells on day 30, exhibited significantly higher expression of CD5 than T-bet⁺ B cells on day 5 post-infection (Figure 5b). A substantial population of CD138⁺ ASCs was also present in the peritoneal cavity on day 30 post-infection, and these cells almost exclusively expressed IgM (Figure 5c). These ASCs, like the MBC, were also B220⁺ CD19^lo/neg, and IgD^−neg (data not shown). While the majority of the ASCs were CD5-negative, the CD5⁺ subset exhibited much higher T-bet expression. Our data suggest that CD5 and T-bet expression may both occur following activation and accompanying PB differentiation.

Compared to naïve mice, the peritoneal cavity of mice 30 days post-infection had higher overall levels of leukocytes and B cells, with T-bet⁺ B cells making up a large share of both total B cells and actPBCs (Figure 5d). These observations demonstrate that the peritoneal cavity immunological landscape undergoes substantial changes following acute infection and is a reservoir for both T-bet⁺ ASCs and type I memory cells.

Peritoneal cavity CD11b⁺ memory B cells do not require B cell-intrinsic T-bet expression for their generation.

Our previous work demonstrated that the generation of CD11c⁺ memory B cells in the spleen did not require B cell-intrinsic T-bet expression (49). To determine whether the generation of CD11b⁺ MBCs in the peritoneal cavity could likewise occur in the absence of B cell T-bet, we generated B cell conditional T-bet KO mice by crossing T-bet^flox mice to Mb1^cre mice, and infected the resulting strain with E. muris. As in the spleen, the peritoneal cavities of infected B cell specific T-bet-deficient mice contained a distinct population of CD11b⁺ CD73⁺ PD-L2⁺ CD80⁺ MBC 30 days after infection (Figure 6a–c). Furthermore, the frequency of CD73⁺ PD-L2⁺ CD80⁺ MBC in the peritoneal cavity did not significantly differ between B cell T-bet-deficient mice and control mice (Figure 6d). As expected, CXCR3 expression was absent on B cells in T-bet-deficient mice (Figure 6c, e). These results suggest that intrinsic expression of T-bet is not required for the generation of MBCs in the peritoneal cavity, similar to our previous observations in the spleen (49).

Figure 6: B cell-intrinsic T-bet expression is not required for the generation of CD11b⁺ MBCs in the peritoneal cavity.

(A) Peritoneal cavity cells from E. muris-infected Mb1Cre^−/− T-bet^fl/fl (Control) and Mb1Cre^+/− T-bet^fl/fl (T-bet KO) mice were analyzed by flow cytometry on day 30 post-infection. Representative dot plots show gating of B cells pre-gated as singlet, live, B220 and/or CD19⁺ cells. Representative data from 8 biological replicates acquired in 2 independent experiments are shown.

(B) Total singlet live leukocytes isolated from the peritoneal cavity of control (left) and B cell T-bet KO (right) mice were analyzed on day 30 post-infection. The analysis was performed without B cell gating to demonstrate positive T-bet staining compared to B cell-specific T-bet deficiency. Data representative of 8 biological replicates from two experiments is shown.

(C) Peritoneal cavity B cells, gated as in (A), were analyzed on day 30 post-infection. Expression of B cells from control and B cell T-bet KO mice are represented by maroon and blue lines, respectively. The histograms show mode-normalized frequency vs. fluorescence intensity for the markers indicated below the histograms. Data representative of 8 biological replicates from two experiments are shown.

(D-E): Peritoneal cavity cells from Mb1Cre^−/− T-bet^fl/fl (Control) and Mb1Cre^+/− T-bet^fl/fl (T-bet KO) mice were analyzed on day 30 post-infection. The plots show the frequency (left plot) and number (right plot) of CD73⁺ CD80⁺ PD-L2⁺ MBC (D), and CXCR3⁺ (E) cells among singlet, live, B220 and/or CD19⁺ cells. Each dot represents an individual mouse. Statistical significance was determined using two-tailed unpaired t tests. The data are representative of two experiments. ns, p > 0.05, *p < 0.05, **p < 0.01, ****p < 0.0001.

Discussion

We report here that T-bet is expressed in B cells in the peritoneal cavity during intracellular bacterial infection, and that T-bet expression coincides with a loss of CD5 expression among activated B cells at this site. CXCR3⁺ T-bet⁺ B cells in the peritoneal cavity were found to be proliferative and lacked enrichment for canonical B1a variable region genes, relative to CXCR3-negative B cells. T-bet-expressing cells were observed to migrate to the spleen during acute infection. Moreover, we identified the peritoneal cavity as a reservoir for a sizeable pool of T-bet-expressing CD11b⁺ memory B cells, although these cells did not require T-bet expression for their generation. Our study therefore reveals the murine peritoneal cavity to be a site of a T-bet⁺ B cell response during intraperitoneal infection, and a site of T-bet⁺ MBC persistence.

Our findings demonstrate that E. muris infection causes a shift in the peritoneal cavity B cell landscape, fostering the dominance of T-bet⁺ B cells, generation of T-bet⁺ ASC, and retention of T-bet⁺ B cell memory. A working model of two pathways by which we propose T-bet⁺ B cell subsets could arise in the peritoneal cavity are shown in in Figure 7. As part of a B1 B cell-centric or residential model, CD5⁺ or CD5-negative B1 B cells are activated and downregulate CD5, express T-bet and CXCR3, and differentiate to become T-bet⁺ ASCs and MBCs in situ. During B1 cell activation, expression of CD5, and its subsequent downregulation, are required for calcium mobilization, NF-kB pathway activation, and differentiation to IgM-secreting B1 plasmablasts (18, 52). Our work raises the possibility that this form of cell activation could precede the development of T-bet-expressing activated B cells capable of generating ASC and memory responses to intracellular bacterial infection. The innate stimulus that activates the immune system during E. muris infection has not yet been identified, and although TLR9 recognizes bacterial DNA, there is little evidence that this or other TLR pathways are used (53, 54). Notwithstanding, MyD88 and NF-kB signaling appear to be important for controlling E. muris infection (55), suggesting that a yet to be identified innate receptor contributes to activation of innate myeloid-derived cells and B cells in a manner congruous to that observed during other infections that generate T-bet⁺ B cells, and in autoimmune settings (20, 56).

Figure 7: Working model of hypotheses for future testing.

Two models of B cells development and differentiation with the peritoneal cavity during E. muris infection are shown. The first B-1-centric/residential model, shown in (A), describes a scenario in which CD5⁺ or CD5-negative B-1 cells in the peritoneal cavity are activated, begin to express T-bet and CXCR3, and proliferate. These cells then differentiate into T-bet^lo ASCs and T-bet^lo and T-bet^hi MBCs. Some differentiated cells may migrate to the peritoneal cavity.

In the B-2-centric model, shown in (B), B2-derived B cells expand locally or traffic to the peritoneal cavity from one or more sites, after which these cells differentiate into T-bet-expressing ASCs and MBC, as in (A).

Our study also raises the possibility that the population of T-bet⁺ B cells in the peritoneal cavity expanded from B2 cells resident in the peritoneal cavity, or immigrated from other locations known to be sites of T-bet⁺ B cell generation, such as the spleen and liver (14, 15). This hypothesis is illustrated in the B2-centric or migratory model in the working model diagram. Indeed, there does not exist a consensus in the field of B1 B cell biology on a definitive immunophenotype for B1 cells, and the use of both conventional confirmatory markers we used in our study, CD11b and CD43, has been a point of contention in relevant literature. Furthermore, peritoneal cavity resident B2 cells have been proposed to adopt an intermediate B1 and B2 phenotype and functional capacity within the peritoneal cavity (57). Despite the anatomical location and absence of follicular, naïve, plasma cell, and myeloid-derived cell surface markers, high expression of CD19 and IgM indicates that these cells are activated B cells, but does not reveal their lineage. Accordingly, it is possible that crosslinking of the BCR, combined with innate cytokine signals from IFNγ and IL-12, is sufficient to induce T-bet expression in B2 cells resident in the peritoneal cavity or recruited from other sites such as the bone marrow.

During acute infection, our observation of reduced expression of Siglecg and Cd5 in CXCR3⁺ B cells in the peritoneal cavity supports the idea that this population is activated, as siglec G and CD5 have both been shown to have inhibitory functions in B cells (43, 58). The enrichment for genes involved in replication and metabolism also supports the hypothesis that loss of CD5 following infection with E. muris may permit these cells to respond to BCR crosslinking and cellular activation, as described in other studies (18, 58). In support of this model, B cell-specific T-bet-deficient mice have reduced frequencies of splenic plasmablasts during the T-independent phase following E. muris infection (Supplementary Figure 2), although this was associated with a transient reduction in bacterial load during early infection in the B cell-specific T-bet-deficient mice, which may have limited the PB response.

Consistent with the apparent activation of B cells within the peritoneal cavity, we observed early B cell migration from the peritoneal cavity to the spleen upon peritoneal cell transfer. Several studies have observed B1 B cell migration from the peritoneal cavity to the spleen, although context-specific changes in integrin expression and signaling following TLR4-dependent activation with extracellular pathogens do not occur during intracellular E. muris infection (59). Additionally, peritoneal cavity residence has been shown to imprint B cells with an inherent propensity for serosal cavity residence, therefore it will be important to identify the stimulus for migration during ehrlichial infection (60, 61). It is also possible that some B cells migrated to other sites that were not examined, such as the lymph nodes, but we have not observed T-bet-expressing B cells within the lymphatics during E. muris infection.

It will be important to examine whether the T-bet⁺ CD11b⁺ MBC we identified in the peritoneal cavity during E. muris remain resident, as described for resident MBCs identified within the lung and spleen following murine influenza infection (14, 62). It is also possible that T-bet-expressing B cells become resident in the peritoneal cavity due to persistent antigen at this site. Our earlier observations indicate that despite the resolution of hypercellularity, the peritoneal cavity continues to harbor E. muris during chronic infection, regardless of the route of infection (63). We have not confirmed the specificity nor antigen-dependence of the T-bet⁺ B cells in the peritoneal cavity, and these are important questions for future investigation. Neither have we formally confirmed the quiescent state of the MBCs. However, our previous studies of splenic T-bet⁺ B cells have clearly defined these cells as quiescent memory B cells; that is, they did not proliferate under steady-state conditions, and they were required for a secondary response to antigen (7). Given their phenotypic similarities, our data strongly suggest that the actPBCs are also quiescent memory cells, although direct evidence is not yet available. Nevertheless, the maintenance or continual renewal of T-bet⁺ B cells at sites of antigen persistence may help to explain how activated and memory B cells are maintained during chronic infection and autoimmunity.

The findings presented here have wide applicability, given the proposed contribution of T-bet-expressing B cells in autoimmune disorders and immunity in humans. It will be important to determine if T-bet is expressed in B cells within the peritoneal cavity in humans, particularly during infection with Mycobacterium tuberculosis. Indeed, peritoneal tuberculosis is associated with CD4⁺ T cells capable of producing IFNγ in the peritoneal cavity, the same cytokine that drives T-bet expression in B cells (9, 64). The identification of T-bet expression in B cells within a novel site generates a multitude of questions for future investigation. Thus, our studies have implications for a greater understanding of bacterial immunopathogenesis and immune function within serosal cavities.

Key points:

• Peritoneal cavity B cells express T-bet following infection

• Peritoneal cavity CXCR3⁺ and CXCR3^−neg B cells are transcriptionally distinct

• T-bet⁺ MBCs dominate the peritoneal cavity B cell pool during chronic infection

Glossary

MBC

Memory B cell

PB

plasmablast