Dendritic cell-restricted antigen presentation initiates the Follicular Helper T cell program but cannot complete ultimate effector differentiation

Abstract

Follicular Helper T cells (T_FH) are critical for germinal center (GC) formation. The processes that drive their generation and effector potential remain unclear. Here, we define requirements for MHCII antigen presenting cells (APCs) in murine T_FH formation by either transiently ablating or restricting antigen presentation to dendritic cells (DCs). We find that cognate interactions with DCs are necessary and sufficient to prime CD4⁺ T cells towards a CXCR5⁺ICOS⁺Bcl6⁺ T_FH intermediate. However, in the absence of additional APCs, these T_FH fail to produce IL-21. Furthermore, in vitro priming of naïve T cells by B cells engenders optimal production of IL-21, which induces a GC B cell transcriptional profile. These results support a multi-step model for effector T_FH priming and GC initiation, in which DCs are necessary and sufficient to induce a T_FH intermediate that requires additional interactions with distinct APCs for full effector function.

INTRODUCTION

Germinal centers (GCs) are transient structures formed during thymus-dependent (TD) immune responses, where expanding B cells undergo affinity maturation, yielding memory B and plasma cells that mediate protective humoral immunity. The exact interactions dictating whether B and helper T cells assume GC status versus alternative fates remain unclear. A key event in GC formation is the Bcl-6 dependent differentiation of Follicular Helper T (T_FH) cells (1–2). T_FH cells express inducible co-stimulator (ICOS) and CXC chemokine receptor 5 (CXCR5), enabling their migration to the B cell follicle (3–5) and their continue differentiation into GC T_FH (6). Moreover, T_FH express costimulatory molecules and produce IL-21 that are essential for GC B cell survival and differentiation (7–8).

The sources of MHCII⁻restricted antigen presentation required for T_FH differentiation are controversial. Several lines of evidence suggest that B cells provide unique signal(s) required for complete T_FH differentiation. Thus the absence of B cells, the presence of B cells with irrelevant specificity, or an inability of B cells to express ICOSL impairs T_FH generation (2, 9–11). Moreover, T cells unable to stably conjugate with B cells display defective T_FH differentiation (12). The notion of a unique B cell signal has been challenged by findings that suggest instead that only prolonged, high avidity antigen presentation is needed for GC T_FH differentiation that can be mediated by DCs (13) . Indeed, co-stimulation via DCs promotes T cell migration into the B cell follicle (14), suggesting that DC antigen presentation contributes to the T_FH program.

We have dissected the steps of T_FH differentiation by examining the requirement for MHCII-expressing DCs. We find that cognate interactions with conventional DCs are both necessary and sufficient to prime CD4⁺ T cells towards a previously unappreciated T_FH intermediate that expresses Bcl6, CXCR5 and ICOS. However, DC priming alone is insufficient to generate PD1^hi GC T_FH and IL-21 production, indicating the need for subsequent, distinct APC interactions for commitment to the T_FH program. Furthermore, we show that naïve T cells primed in vitro by B cells express more IL-21 than do those primed by DCs. Moreover, addition of IL-21 to in vitro B cell activation cultures that mimic cognate T cell help induces a GC B transcription profile. Thus, we conclude that T_FH differentiation is a multi-step process in which conventional DCs are critical for the initial priming events; however, they are insufficient for imparting complete effector potential.

MATERIALS AND METHODS

Mice & Immunizations

C57BL/6J, OTII and CD11c-DTR mice were purchased from Jackson Laboratories. CD11c/A_β^b mice were bred in house (15). 6–14 week old mice were housed under pathogen free conditions, in accordance with the University of Pennsylvania Animal Care and Use Guidelines. CD4⁺ OTII T cells were enriched by negatively selecting out CD8⁺, B220⁺, MHCII⁺ and FcγRII⁺ cells using magnetic bead selection (Qiagen) and labeled with CFSE (Invitrogen). OTII were transferred i.v. 1 day prior to i.p. immunization with 50 μg NP₁₅-OVA (Biosearch Technologies) in alum. Some mice were given OVA_323–329 peptide (GenWay Biotech). DC depletion and Toxoplasma infection in CD11c DTR mice as described (16).

Flow cytometry

Antibodies were purchased from Biolegend, eBioscience, BD Pharmingen or Invitrogen. DAPI or AQUA™ (Invitrogen) was used to identify live cells. FoxP3 fixation and permeabilization kit was used for intracellular Bcl6 staining (eBioscience). Cells were acquired or sorted on an LSR II cytometer or FacsAria II, respectively (BD Biosciences). Data was analyzed using FlowJo software (TreeStar).

Immunohistochemistry

Spleens immersed in O.C.T. (Tissue Tek) were flash frozen using 2-methylbutane / liquid nitrogen. 8 μm sections were fixed with cold acetone and stained with PNA and antibodies to IgD and CD90.1. Sections were imaged with a Zeiss LSM510META NLO laser scanning confocal system.

In vitro stimulations

OTII cells were polarized towards T_FH lineage as published (9) with CD11c^hi or CD23⁺ cells at a ratio of 1 APC:10 T cells and harvested on day 4. B cell stimulations were performed as described (17) with anti-CD40 (10 ng/ml; BD Biosciences) and IL-21 (100ng/ml; R&D).

QPCR

QPCR was conducted as previously described (17). 18S or GAPDH were used as housekeeping gene for T and B cells, respectively.

ELISPOT

Splenocytes were incubated on 10 ug/ml of NP33-BSA (Biosearch Technologies) coated plates and developed with biotin-anti-mouse IgM or -IgG1 (Southern Biotech) followed by ExtrAvidin-Alkaline Phosphatase and NBT/BCIP substrate (Sigma). Spots were enumerated on CTL-ImmunoSpot (Cellular Technologies).

RESULTS AND DISCUSSION

Dendritic Cells are necessary for TFollicular Helper formation

Using a loss of function approach, we determined whether DC-T cell interactions are necessary for initiation of T_FH cells priming in vivo. We transiently ablated DCs by Diptheria toxin (DT) in CD11c-DTR mice (18), and infected the mice with Toxoplasma gondii as described (16). T_FH cells were present in untreated mice at day 7 post-infection, but were significantly reduced in DT-treated CD11c-DTR animals (Fig. 1A). Normal DC numbers were required for optimal generation of CD4⁺CXCR5⁺ cells (WT 1.71 ± 0.15; CD11c-DTR 0.67 ± 0.13; p<0.01) as well as induction of ICOS (Fig 1B). To verify that these results were not limited to toxoplasmosis, we transferred 10⁵ CD90.1⁺ ovalbumin(OVA)-specific OTII CD4 T cells into CD90.2⁺ WT or CD11c-DTR recipient mice followed by DT treatment and i.p. immunization with NP-OVA/alum. Again, there was a profound decrease in the number of OTII T_FH generated (data not shown). The few CD4⁺CXCR5⁺ T cells formed after DC depletion could potentially migrate into the follicle and initiate GC via other co-stimulatory molecules, including CD40L. We conclude that DCs are necessary to generate T_FH cells, as expected from their role in priming other T_H lineages (19).

Figure 1.

DCs are necessary for initiating T_FH priming. WT and CD11c^DTR mice were treated with DT and infected with T. gondii as described (16). (A) Numbers of T_FH (CD4⁺CXCR5⁺ICOS⁺) were calculated at day 7 p.i. Data are pooled from two experiments (n>8 per group) (B) Representative FACS plots of CXCR5 and ICOS expression on CD4⁺ T cells in WT or CD11c^DTR treated with DT. ** denotes statistical significance of p<0.01 in a 2-tailed t test at α=0.05

Dendritic cells are sufficient for the generation of a Bcl6⁺ T_FH intermediate

Using a gain of function approach, we asked if cognate DC interactions are sufficient to generate T_FH cells. We used the transgenic CD11c/A_β^b mouse model, in which MHCII antigen presentation is limited to conventional CD11c^hi DCs (15). As CD11c/A_β^b mice lack endogenous CD4 T cells, we transferred CFSE labeled 10⁶ OTII cells. Splenic OTII cells proliferated similarly in both WT and CD11c recipients following i.p. immunization with NP-OVA/alum (Fig. 2A, 2B). Further, the numbers of OTII derived CD62L^loCXCR5⁺ T_FH generated in WT (WT-T_FH) or CD11c/A_β^b mice (CD11c-T_FH) were not significantly different, and both subsets displayed equivalent expression of ICOS (Fig 2B, 2C). Moreover, transfer of 10⁵ OTII cells yielded similar results (Supp. Fig. 1A). Furthermore, OTII cells migrated into the follicle in both mice strains (Fig. 2E). Since Bcl6 is necessary and sufficient for differentiation of CXCR5⁺ T_FH cells (1–2), we asked if antigen presentation restricted to DCs effectively drove Bcl6 induction. Compared to naïve OTII T cells, WT-T_FH or CD11c-T_FH expressed higher but equivalent levels of Bcl6 message (Fig. 2D) and protein (Fig. 2C). Thus, cognate interactions with DCs are sufficient to skew naïve CD4⁺ T cells towards Bcl6⁺ T_FH.

Figure 2.

Cognate interactions by dendritic cells are sufficient for generation of T_FH cells. 10⁶ CFSE labeled CD90.1⁺ OTII T cells were transferred into either WT or CD11c/A_β^bmice and mice were immunized with NP-OVA/alum. (A) Representative FACS gating strategy for identification of splenic OTII T cells (CD19⁻ TCRβ⁺ CD4⁺ CD90.1⁺). OTII T_FH cells were identified as CD62L⁻ CXCR5⁺ cells. CFSE dilution of transferred OT-II cells was assessed (middle panel). (B) Total number of splenic OTII and OTII T_FH present in immunized WT or CD11c/A_β^b recipients at day 8 p.i. or in non-immunized WT mice at d8 post-transfer (n=4–5 mice, representative of >5 experiments). (C) Representative histogram overlay of ICOS and Bcl6 staining intensity on OTII T_FH generated in WT or CD11c/A_β^b mice and naïve endogenous CD4⁺ T cells (CD19⁻ TCRβ⁺ CD4⁺CD62L⁺) present in WT mice (D) Bcl-6 mRNA in FACS sorted OT-II T_FH generated in WT or CD11c/A_β mice at d7 p.i.. relative to naive OTII cells (representative of 3 experiments) (E) Confocal micrographs of splenic sections to identify CD90.1⁺ OTII along with PNA⁺ GCs and IgD⁺ naïve B cells in the follicle. Scale bar denotes 50 μm.

Cognate DC-T interactions cannot induce GC status in T cells

MHCII expression by B cells is necessary for antibody class switching (15, 20) and presumably for GC formation. Consistent with this view, GC B cells were absent both phenotypically (Supp. Fig. 2A, 2B) and histologically GCs (Fig. 2E) in immunized CD11c/A_β^b mice. Further, CD11c/A_β^b mice generated few NP-specific IgG1 ASCs, despite an intact IgM ASC response (Supp. Fig 2C), as reported before (15, 20). PD1 expression is thought be an indicator of chronic antigen exposure in the GC, as correlative evidence links the presence of GC B cells with PD1 expression on T_FH cells (4). Thus, we examined the expression of PD1 on T_FH cells. Amongst the CXCR5⁺ cells, a small subset of the WT-T_FH cells expressed high levels of PD1 as well as GL7, which were absent within the CD11c-T_FH cells (Fig. 3A, B) consistent with GC T_FH phenotype (Fig. 3C, (6)). Moreover, the transfer of 10⁷ polyclonal WT CD4⁺ T cells did not alter the observed phenotype (data not shown), indicating that the absence of endogenous helper T cells does not impact the evolution of this response in the CD11c/A_β^b mice. Since T_FH down regulate P-selectin glycoprotein ligand 1 (PSGL1) (11) and upregulate and B- and T-lymphocyte attenuator (BTLA) (6), thus we examined their expression on CD11c-T_FH and WT-T_FH. Compared to naïve WT T cells, both CD11c-T_FH and WT-T_FH had significantly upregulated BTLA and down regulated PSGL1 (data now shown). However, CD11c-T_FH expressed more PSGL1 than WT-T_FH (data not shown) (11) while BTLA expression was comparable. Collectively, we conclude that not only do GC B cells require T_FH cells for their induction, GC T_FH also rely on interactions with other APCs, including activated B cells, to complete their phenotype.

Figure 3.

DC restricted antigen presentation is insufficient for the generation of GC T_FH 10⁶ OTII T cells were transferred into WT and CD11c/A_β^b mice and mice were immunized with NP-OVA/alum. (A) Representative FACS gating strategy to identify PD1⁺CXCR5⁺ within activated OTII T_FH (CD19⁻TCRβ⁺CD4⁺CD90.1⁺CD62L⁻) in immunized WT or CD11c/A_β^b on d7 p.i. as well as naïve WT CD4⁺ T cells as a staining control. The PD1⁺ cells were parsed into PD1^hi (GC T_FH) and PD1^int populations. (B) The total number of OTII GC T_FH in WT or CD11c/A_β^b mice on d7 p.i. (n=4–5) (C) Histogram overlay of GL7 staining on various subsets gated in (A) (n=4–5 mice). Results are representative of 3 experiments. ** denotes statistical significance of p<0.01 in a 2-tailed t test at α=0.05.

IL-21 expression is profoundly reduced in T_FH cells primed by DCs alone

IL-21 is a key T_FH effector cytokine that mediates differentiation of GC B cells (6–8). Strikingly, CD11c-T_FH primed solely by DCs expressed minimal IL-21 compared to WT-T_FH (Fig. 4A). To determine if B cells were intrinsically more adept at inducing IL-21 producing T_FH, we polarized naïve OTII T cells in vitro towards the T_FH lineage with splenic CD11c^hi DCs or CD23⁺ B cells as APCs as described (9). While both B cells and DCs induced expression of Bcl6, B cells induced greater IL-21 transcript in polarized T_FH cells (Fig. 4B). Together, these data suggest that B cells provide qualitatively different signal(s) than DCs, perhaps via expression of co-stimulatory molecules that facilitate optimal IL-21 production by T_FH cells.

Figure 4.

Reduced expression of IL-21 in T_FH cells primed in CD11c/A_β^b mice. (A) Expression of IL-21 transcript in FACS sorted OT-II T_FH generated in WT or CD11c/A_β^b mice at d7 p.i. normalized to naïve OTII cells. Results are representative of 3 experiments. (B) Relative expression of Bcl6 and IL-21 in OTII cells polarized towards T_FH lineage using splenic CD11c⁺ or CD23⁺ cells as APCs compared to Th0 condition. Results are representative of 2 experiments. (C) QPCR analysis of Bcl6 and Blimp1 on CD23⁺ cells stimulated with F(ab)₂ fragments of αIgM alone or in combination with αCD40 with or without IL-21 for 3 days. Results pooled from 3 experiments.

Consistent with a previous study (9), in vitro primed T_FH expressed little to no IL-17 transcript (Supp. Fig. 1E), further confirming their T_FH character. Moreover, the superior ability of DCs to induce IL-17 when compared to B cells was recapitulated in vivo, as OTII T_FH and OTII T_eff primed solely via conventional DCs in CD11c/A_β^b mice expressed more IL-17 compared to their WT counterparts (Supp. Fig. 1F). Together, these data highlight the critical role played by non-DC APCs, presumably B cells, in enforcing progression to T_FH status while inhibiting other lineage potential.

Sustained antigen presentation on conventional DCs is insufficient to rescue GC T_FH or production of IL-21

Recent findings suggest that sustained antigen presentation restores the formation of PD1^hiCXCR5⁺ T_FH in the absence of B cells (13). We performed a similar experiment in the CD11c/A_β^b mice. We transferred 10⁵ OTII cells, immunized with NP-OVA the next day and then administered 10 μg of OVA_323–339 peptide i.v. on day 3 p.i.. Consistent with Deenick et al. (2010), we observed greater expansion of OTII cells in the CD11c/A_β^b mice given exogenous peptide (Supplemental Fig. 1A) on day 7 p.i.. However, neither PD1^hi GC T_FH nor IL-21 production were observed in CD11c/A_β^b mice that received extra peptide (Supp. Fig. 1B–D). Whether these differences arise because of irradiation chimeras used by Deenick et. al. (2010) or due to the lack of other MHCII⁺ APCs in the CD11c/A_β^b mice is unclear. Thus, sustained antigen presentation by conventional DCs is insufficient to induce GC T_FH cells and a second, distinct, antigen presenting cell population is required.

In conjunction with BCR and CD40 ligation, IL-21 induces a GC B transcription profile

Recent studies suggest that IL-21R ligation yields maximal Bcl6 expression in B cells, thus fostering GC formation in vivo (7–8). However, in the presence of anti-CD40 in vitro, IL-21 induces B Lymphocyte-induced maturation protein 1 (Blimp1) (21) skewing B cells towards a plasma cell fate. To simulate antigen-driven, T-dependent activation of Follicular B cells in vitro, we ligated both the BCR and CD40 with anti-IgM and anti-CD40 antibodies respectively, then assayed for the induction of Blimp1 and Bcl6 transcripts. Compared to B cells stimulated with only anti-IgM or anti-IgM plus anti-CD40, IL-21 markedly augmented the transcriptional profile associated with adoption the GC fate. Both AID (data not shown) and Bcl6 were substantially upregulated, whereas Blimp1 transcripts, which were elevated following BCR and CD40 ligation alone, were lowered in the presence of IL-21 (Fig 4C). We conclude that concurrent stimulation of BCR, CD40 and IL-21R potently induces Bcl6 expression, which in turn represses Blimp1 (22). This reciprocal interaction is key to skew the B cells towards a GC fate and away from the short-lived plasma cell fate, which can also inhibit T_FH differentiation (23).

Collectively, our data suggest that complete differentiation of T_FH is most effectively accomplished via successive, cognate interactions with distinct APCs. Herein, we demonstrate that DCs are necessary and sufficient to initiate the T_FH program, arguing against the notion that B cells are critical for priming (2, 9–11). Priming solely via DCs in the CD11c/A_β^b system leads to accumulation and persistence of a novel intermediate or pre-T_FH, which express CXCR5, ICOS and Bcl6, but lacks IL-21. Thus, optimal induction of IL-21 as well as further differentiation towards GC T_FH requires cognate interactions with other APCs, including activated B cells. Thus, we propose that as poised CXCR5⁺ICOS⁺T_FH relocate from the DC-rich T cell zone to B cell-rich follicles, their interactions with activated B cells via costimulatory ligands including PDL2, ICOSL increases the likelihood of adopting a T_FH cytokine profile, including IL-21 production (24–25). Whether B cells provide qualitatively distinct signal(s) to T_FH for optimal effector cytokine production remains unclear.

Since CD4⁺ T cells require continued interactions with APCs for their expansion and effector functions (26), the assumption of alternate effector fates likely reflects the receipt of chronologically ordered signals during the early primary response. This suggests a step-wise model for GC T_FH formation. T cells primed by DCs differentiate into a novel pre-T_FH subset. These cells can then migrate towards the T-B border, where cognate interactions with antigen primed B cells induce optimal IL-21 production, thus acquiring effector potential characteristic of T_FH subset while inhibiting other fates, such as Th17. IL-21, in turn, reinforces a GC transcriptional program in the activated B cells. Thus, fully-differentiated T_FH are only generated when a humoral response is appropriate and antigen-specific T cell help is very specifically directed towards antigen-experienced B cells.

Abbreviations used in the paper

DC

Dendritic Cells

PSGL1

P selectin glycoprotein ligand 1

BTLA

B- and T-lymphocyte attenuator

PD1

Programmed Death Ligand 1