Report

Preferential Generation of Follicular B Helper T Cells from Foxp3+ T Cells in Gut Peyer's Patches

See allHide authors and affiliations

Science  13 Mar 2009:
Vol. 323, Issue 5920, pp. 1488-1492
DOI: 10.1126/science.1169152

Abstract

Most of the immunoglobulin A (IgA) in the gut is generated by B cells in the germinal centers of Peyer's patches through a process that requires the presence of CD4+ follicular B helper T(TFH) cells. The nature of these TFH cells in Peyer's patches has been elusive. Here, we demonstrate that suppressive Foxp3+CD4+ T cells can differentiate into TFH cells in mouse Peyer's patches. The conversion of Foxp3+ T cells into TFH cells requires the loss of Foxp3 expression and subsequent interaction with B cells. Thus, environmental cues present in gut Peyer's patches promote the selective differentiation of distinct helper T cell subsets, such as TFH cells.

The production and secretion of immunoglobulin A (IgA) by the host is critical for the maintenance of a vast community of commensal bacteria in the intestinal lumen with minimal penetration of the gut epithelium (1, 2). Most of this IgA synthesis requires germinal center (GC) formation in Peyer's patches (PPs), aggregations of lymphoid follicles in the gut. In GCs, activated B cells express activation-induced cytidine deaminase (AID) and switch from making IgM to IgA (35). GC development in PPs requires bacteria in the gut; germ-free mice have small GCs, probably induced by bacterial components in food (1). T cells, by providing cytokines and costimulatory molecules to B cells, are also required for GC induction. Mice that lack T cells are devoid of GCs, and GC formation can be rescued by the adoptive transfer of CD4+ T cells (1, 6, 7).

We investigated the origin of TFH in PPs by examining the contribution of Foxp3+CD4+ T cells (810) obtained from Foxp3EGFP reporter mice (11), which express green fluorescent protein (GFP) under the control of the Foxp3 promoter. Total CD4+ T cells, Foxp3/GFP+ CD4+ T cells (hereafter called Foxp3+ T cells) or Foxp3/GFP CD4+ (hereafter called Foxp3 T cells) that were isolated from the spleen and lymph nodes (LNs) of Foxp3EGFP reporter mice were adoptively transferred into T cell–deficient, CD3ϵ–/– mice (12). These spleen- and LN-derived Foxp3+ T cells were phenotypically heterogeneous and expressed variable levels of the α chains of interleukin-2 (IL-2) receptor and IL-6 receptor on their surfaces (fig. S1). Transfer of total CD4+ T cells or Foxp3+ or Foxp3 T cells into CD3ϵ–/– mice reconstituted the percentage and number of B220 IgA+ plasma cells in the lamina propria (LP) of the small intestine and of peanut agglutinin–positive (PNA+) Fas+ GC-phenotype B cells in PPs after 4 weeks (Fig. 1, A and B). To our surprise, however, activated B cells formed typical GC clusters (large groups of AID-expressing cells) only when Foxp3+ T cells were transferred, whereas, in mice injected with Foxp3 T cells, the AID-expressing B cells were scattered throughout the PPs (Fig. 1C). Furthermore, formation of GCs was induced by Foxp3+ T cells in PPs more effectively than by unfractionated CD4+ T cells or by a 1:1 mixture of Foxp3+ and Foxp3 T cells (Fig. 1C). Thus, Foxp3+ T cells preferentially induced GC formation in PPs and IgA-producing cells in the gut.

Fig. 1.

Induction of GCs in PPs by Foxp3+ T cells. (A) Representative flow cytometric profiles of cells stained for the indicated markers. B220 is used as a marker for B cells. Numbers on plots indicate the frequency of cells in the gate. (B) Total numbers of GC B cells isolated from the PPs and IgA plasma cells from the LP of the small intestine of CD3ϵ–/– mice 4 weeks after reconstitution with the indicated T cells. Means ± SEM for three to five mice per group. (C) Immunofluorescent microscopy analysis of AID, DAPI (4′,6′-diamidino-2-phenylindole) staining for DNA, and CD3 (upper row) or GFP (lower row) expression in horizontal sections of the PPs from CD3ϵ–/– mice 4 weeks after reconstitution with the indicated T cell populations.

How do the Foxp3+ T cells, which are known to suppress T cells, act as B helper T cells? There were many more CD3+ T cells than GFP+ T cells in the PPs of CD3ϵ–/– mice that received Foxp3+ T cells (Figs. 1C and 2A). Indeed, 80% of Foxp3+ T cells lost GFP expression in PPs, whereas ∼50% of CD3+ T cells remained GFP+ in the LP of the small intestine, the large intestine, LNs, and spleen (Fig. 2B). The majority of GFP+ cells transferred to CD3ϵ–/– host mice, as in nonmanipulated Foxp3EGFP mice, were located in the T cell zone, close to the dendritic cells (DCs), which were stained with CD11c (Fig. 2A). In contrast, most GFP CD3+ T cells were located within the B cell follicles, among the AID-expressing cells. Thus, a large fraction of the transferred Foxp3+ T cells lost Foxp3 expression and preferentially localized to B cell follicles in the PPs. These results suggest that the environment of the PPs down-regulates Foxp3 expression or fosters the accumulation of T cells that have lost Foxp3 expression elsewhere, or both.

Fig. 2.

Foxp3+ T cells down-regulate Foxp3 expression and migrate into B cell follicles of PPs. (A) Immunofluorescent microscopy analysis of PPs horizontal sections, stained as indicated, from the CD3ϵ–/– mice 4 weeks after reconstitution with Foxp3+ T cells and from 2-month-old Foxp3EGFP mice. (B) Representative flow cytometric profiles of GFP (Foxp3) expression on CD3+ CD4+ T cells from the indicated tissues. Numbers indicate the percentage of cells in the gate. (C) Immunofluorescent microscopy analysis of horizontal serial PPs sections stained as indicated from the secondary CD3ϵ–/– recipient mice, 4 weeks after reconstitution with the indicated T cells. (D) Total CD4+ T cells and the ratio of GC B cells to CD4+ T cells in the PPs of CD3ϵ–/– mice 4 weeks after reconstitution with the indicated T cells. Means ± SEM for three to five mice per group.

To assess whether T cells that had down-regulated expression of Foxp3 were among the helper cells able to induce GC formation, we performed serial transfer experiments. Foxp3+ T cells that had down-regulated Foxp3 (designated Foxp3+ ) were sorted from the spleen and LNs of recipient mice and then transferred into CD3ϵ–/– recipients. Subsequent GC formation, as indicated by AID-expressing clusters, was comparable to that observed after the initial transfer of Foxp3+ T cells (Fig. 2C, compare with Fig. 1C). Furthermore, the induction of GCs in PPs and the generation of IgA-producing cells in the gut by Foxp3+ T cells were more efficient than either of these resulting from the transfer of unfractionated CD4+ T cells (Fig. 2D and fig. S2). These results indicate that Foxp3 expression is plastic and that cells that have previously expressed Foxp3 are effective precursors for TFH cells that are capable of inducing GC formation in PPs.

TFH cells express surface markers, such as the chemokine receptor CXCR5 and the costimulatory molecules CD40 ligand (CD40L) and inducible T cell costimulator (ICOS) (1317). We examined these markers on CD4+ T cells that either maintained Foxp3 expression or down-regulated Foxp3 expression after transfer of Foxp3+ T cells into CD3ϵ–/– mice. Over half of the T cells with down-regulated Foxp3 in PPs expressed CXCR5, and these cells were located in the light zone of GCs where the CXCR5 ligand, CXCL13, is abundantly expressed (16) (Fig. 3, A and B). In addition, many CXCR5+ Foxp3 T cells expressed CD40L, ICOS, programmed death-1 (PD-1), and CD28 (Fig. 3A and fig. S3A), all of which are costimulatory molecules necessary for TFH cell generation and function (17).

Fig. 3.

Differentiation of Foxp3 T cells derived from Foxp3+ T cells into TFH cells in PPs. (A) Flow cytometric analysis of CD4+ T cells from PPs of CD3ϵ–/– mice 4 weeks after reconstitution with Foxp3+ T cells. Numbers of graphs indicate the frequency of cells in the gate. (B) Immunofluorescent microscopy analysis of consecutive vertical sections from PPs of CD3ϵ–/– mice 4 weeks after reconstitution with Foxp3+ T cells and of 2-month-old Foxp3EGFP mice, stained as indicated to reveal the localization of T cells in the CXCL13-rich zone (dashed lines) of the GCs. (C) Expression of indicated molecules in sorted T cell populations assessed by real-time reverse transcription polymerase chain reaction. Relative amounts of mRNA normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) are shown. Means ± SEM of three independent experiments. SPL, spleen. (D) Flow cytometric analysis of Foxp3 expression (as measured by GFP) on CD4+ T cells from PPs of CD3ϵ–/– mice 4 weeks after the reconstitution with a 1:1 mixture of Foxp3+ and Foxp3 T cells. (E) The percentage of Foxp3+- or Foxp3-derived T cells expressing the indicated surface molecules as determined by flow cytometry. Cells were pooled from three mice. Data represent one of the two experiments with consistent results.

We determined the gene expression profile of TFH cell–like T cells that had down-regulated Foxp3 by dividing the Foxp3 T cells isolated from PPs of CD3ϵ–/– mice transferred with Foxp3+ T cells into CXCR5+ and CXCR5 subsets. We then compared the gene expression profiles in these subsets with those in control Foxp3+ and CXCR5+ TFH cells in PPs, as well as with those in Foxp3+ and Foxp3 T cells from spleen and LNs of CD3ϵ–/– mice into which Foxp3+ T cells were transferred. The Foxp3+ T cells, especially those isolated from PPs, expressed abundant transcripts for IL-10, a major anti-inflammatory cytokine in the gut (18) (Fig. 3C). In contrast, cyclic nucleotide phosphodi-esterase 3B (Pde3b), which is repressed by Foxp3 (19), was highly expressed in Foxp3CXCR5+ T cells derived from Foxp3+ T cells and in control TFH cells (fig. S3B). In addition, the Foxp3 CXCR5+ cells in the PPs expressed transcripts of two signature genes of TFH cells: the transcription factor Bcl-6 and IL-21 (Fig. 3C), a critical cytokine for both generation of TFH cells and T cell–dependent antibody responses (17, 20, 21). The transcripts for IL-4, which also contributes to GC formation (16), were also highly expressed by Foxp3 CXCR5+ T cells derived from the Foxp3+ T cells, as well as by control TFH cells (fig. S3B). Although TH17 CD4+ Thelper cells (which produce IL-17) have been reported to share attributes of TFH cells (21), we detected low mRNA expression for TH17-specific molecules RORγt and IL-17A in Foxp3 CXCR5+ T cells, as well as in TFH cells in PPs (22, 23) (fig. S3B). Thus, a fraction of Foxp3+ T cells became TFH-like cells in the PPs after down-regulating Foxp3 expression.

Unexpectedly, Foxp3+ T cells preferentially acquired a TFH cell phenotype even when we transferred a 1:1 mixture of Foxp3+ and Foxp3 T cells (Fig. 3, D and E, and fig. S3C). Furthermore, in secondary 1:1 transfers of Foxp3+ T cells mixed with control Foxp3 T cells, CXCR5 was expressed predominantly in T cells derived from Foxp3+ T cells (fig. S4A).

In vitro activation of Foxp3+ T cells but not of Foxp3 T cells induced CXCR5 expression, especially on Foxp3+ T cells (fig. S4B). Together, these results suggest that Foxp3+ T cells induce enhanced GC formation in PPs because of their ability to differentiate into TFH cells.

To determine whether B cells are required for the down-regulation of Foxp3 that preceded Foxp3+ T cell differentiation into TFH cells, we injected Foxp3+ T cells into recombination-activating gene–2–/– (RAG-2–/–) recipient mice, which lack both T and B cells. Four weeks after the transfer, 90% of the T cells in PPs were Foxp3 (Fig. 4A). In the absence of B cells, however, Foxp3 T cells failed to generate CXCR5+ PD-1high [elevated programmed death–1 (PD-1) expression] TFH cells, which suggested that, although Foxp3 down-regulation does not require B cells, B cells are required for acquisition of the TFH cell phenotype. The cytokines IL-6 and IL-21, abundantly produced by gut DCs and T cells, respectively (fig. S4C), down-regulate Foxp3 expression (24). Activation of Foxp3+ T cells in vitro, particularly in the presence of IL-6, decreased Foxp3 expression (Fig. 4B). Addition of IL-6 also enhanced expression of IL-21, a cytokine that induces TFH cell differentiation, an effect that was partially dependent on autocrine production of IL-21 and transforming growth factor–β (TGFβ)(Fig. 4B and fig. S4D).

Fig. 4.

Differentiation of TFH cells, but not Foxp3 down-regulation, requires B cells and depends on CD40-CD40L interaction. (A) Flow cytometric analysis of the expression of GFP (Foxp3) on CD4+ T cells from PPs of RAG-2–/– mice 4 weeks after reconstitution with Foxp3+ T cells. (B) Flow cytometric analysis of the indicated molecules in GFP+ or GFP T cells cultured for 5 days with beads coated with antibodies against CD3 and CD28 in the presence of IL-2 and with the indicated cytokines or monoclonal antibodies. Data represent one of the three experiments with consistent results. (C) Representative flow cytometric profiles of CD4+ T cells isolated from the indicated tissues of CD3ϵ–/– mice that received Foxp3+ T cells and CD40L-blocking antibodies (Abs). (D) Immunofluorescent microscopy analysis of representative spleen sections from CD3ϵ–/– mice 4 weeks after reconstitution with the T cell populations indicated. Mice were immunized with SRBCs and Pam2CSK4 (a synthetic lipoprotein) on day 14 and 21 and analyzed 1 week later. Sections from the spleen of wild-type (WT) mice without [phosphate-buffered saline (PBS) only] or with immunization are shown as controls. (E) Flow cytometric analysis of CD4+ T cells from spleens of mice analyzed in (D).

The interaction of CD40 on B cells or DCs with CD40L on T cells is critical for the formation of GCs (25, 26). To test whether this interaction is required for the generation of TFH cells from Foxp3+ T cells in PPs, we injected mice with a CD40L-specific antibody. In antibody-treated mice, Foxp3+ T cells still lost Foxp3 expression but did not gain expression of CXCR5, CD40L, and PD-1 (Fig. 4C). Consistent with these observations, blocking CD40-CD40L interactions resulted in a reduction of GC formation in PPs and of B220 IgA+ plasma cells in the LP (fig. S5, A and B). Thus, the differentiation of Foxp3+ T cells into TFH cells requires B cells and CD40 expression, presumably by either B cells or DCs or both.

Foxp3+ T cells were converted into TFH cells only in PPs; neither TFH cells nor GCs could be detected in spleen or LNs of CD3ϵ–/– mice adoptively transferred with Foxp3+ T cells (Fig. 4, D and E, and fig. S6). Furthermore, immunization with sheep red blood cells (SRBCs) in the presence of bacterial components failed to induce GCs in spleen of CD3ϵ–/– mice that received Foxp3+ T cells. In contrast, control mice and CD3ϵ–/– mice that had received Foxp3 T cells generated GCs in the spleen after SRBC immunization (Fig. 4, D and E). Thus, the precursors of PP TFH cells are enriched in the Foxp3+ T cell population, whereas other T cells, like Foxp3 T cells, can differentiate into TFH cells in the spleen on experimental systemic immunization.

Our studies demonstrate that Foxp3+ T cells in PPs can differentiate efficiently into cells with characteristics of TFH cells, which then participate in the induction of GCs and IgA synthesis in the gut. How can we then explain the preferential generation of PP TFH cells from the peripheral Foxp3+ T cell population? Two possible scenarios can be conceived. First, TFH cell differentiation per se may require a Foxp3-dependent molecular program. This is unlikely, however, because adoptively transferred Foxp3 T cells efficiently generated TFH cells in the spleen on immunization with SRBCs and because scurfy T cells, which contain a mutated Foxp3 gene, could generate TFH-phenotype cells in PPs, albeit less efficiently than control T cells (fig. S7). A more likely scenario is that PP TFH cell differentiation may be controlled by the same signals that promote Foxp3 expression in gut T cells (2730), such as antigen recognition through the T cell receptor (TCR). Consistent with this idea, adoptive transfer of ovalbumin-specific OT-II TCR transgenic RAG-2–/– CD4+ T cells into CD3ϵ–/– hosts led to generation of Foxp3+ T cells in the gut LP and TFH cells in the PPs if mice were fed ovalbumin (fig. S8). These data suggest that, depending on the environment, TCR stimulation induces either IL-10–producing “suppressor” or IL-21–producing “helper” T cells (fig. S9). Despite the presence of the same antigens that previously induced Foxp3 expression, the IL-6–, IL-21–, and activated B cell–rich environment of PPs results in many Foxp3+ T cells differentiating into TFH cells. These studies have implications for how the suppression of inflammatory reactions and induction of IgA synthesis occur in the gut.

Supporting Online Material

www.sciencemag.org/cgi/content/full/323/5920/1488/DC1

Materials and Methods

Figs. S1 to S9

References

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article