Using defined in vitro conditions, we have examined the differentiation of naïve CD4+ T cells into cells with characteristics of Tfh cells, including the ability to provide efficient help for GC formation in vivo. These in vitro generated Tfh-like cells are notable for their high expression of BTLA, ICOS, CD84, IL-21 and IL-2, low expression of other effector cytokines, and the presence of a subpopulation with increased CXCR5 and PD-1, all of which are comparable to Tfh cells isolated ex vivo. Furthermore, in comparison to Th0, Th1, Th2, and Th17 populations, these cells exhibit greater expression of Bcl6.
With some exceptions, Tfh cells are found to reside within the GC as well as along the boundaries of the T-B cell zones, and may require prolonged B cell or other secondary APC interactions to acquire full effector cell function (Baumjohann et al., 2011
; Cannons et al., 2010
; Crotty, 2011
; Deenick et al., 2010
; Goenka et al., 2011
; Qi et al., 2008
). Although our Tfh-like cells are generated in vitro, the ability of these cells to potently induce GCs when transferred in vivo strongly argues that they contribute to functional Tfh cells. For this reason, we refer to them as “Tfh-like”, but alternatively they may be considered a “pre-Tfh” cell due to their resemblance to Tfh cells early in humoral responses, or a “Th21” cell, due to their cytokine expression being primarily limited to IL-21. Whether these in vitro differentiated cells represent bona fide Tfh cells, versus a Tfh cell precursor, may require further evaluation. However, these cells have permitted examination of questions about early Tfh differentiation and epigenetic modifications, which, importantly, we were able to confirm with primary Tfh cells isolated directly ex vivo from mice. Strikingly, both ex vivo and in vitro cells exhibit evidence for multi-potential cytokine production, accompanied by epigenetic modifications consistent with plasticity of expression of regulatory transcription factors.
We have used this system to probe the requirements for differentiation of Tfh-like populations. Similar to studies on IL-17a (Veldhoen et al., 2009
), we found that IMDM also enhanced generation of IL-21 expressing cells, suggesting that IMDM components enhance production of multiple cytokines. We further found a requirement for STAT3 for in vitro production of IL-21 in these culture conditions. Although the requirements for IL-6 and IL-21 for Tfh cells and GC formation in vivo are controversial (Eddahri et al., 2009
; Eto et al., 2011
; Nurieva et al., 2008
; Poholek et al., 2010
; Vogelzang et al., 2008
), blockade of IL-6 and IL-21 prevented IL-21 production in Tfh-like culture conditions (data not shown). Whether other cytokines or pathways may substitute in vivo or in vitro will require further examination. It is notable that patients expressing a dominant-negative mutation of STAT3 still develop GC responses, suggesting that there may be multiple pathways of Tfh differentiation, an idea that would be consistent with the concept of Th cell plasticity.
Because prolonged interactions with B cells or other APCs have been implicated in development of Tfh cells in vivo, we utilized mitomycin C-treated T-depleted splenocytes as APCs to preserve the antigen presenting function of B cells (Webb et al., 1985
). Although in our hands isolated splenic CD11c+
DCs and T-depleted splenocytes induced similar amounts of IL-21 and CXCR5, T cells activated with T-depleted splenocytes induced twice the percentage of GC B cells (data not shown). Thus, although DCs can confer Tfh phenotypes upon CD4+
T cells, the presence of B cells and/or prolonged antigen stimulation may be important for Tfh differentiation and function.
Supporting this idea, SAP-deficiency did not intrinsically alter the initiation of a Tfh differentiation program, although it prevented Tfh-like cells from functioning in vivo, consistent with defects in T-B cell interactions (Cannons et al., 2010
; Qi et al., 2008
). These results were not due to reduced efficiency; increasing the number of transferred cells did not improve GC formation (data not shown). However, whether these observations result from an inability of Sh2d1a−/−
T cells to undergo a late differentiation process is not clear. Indeed, although IL-21 (Spolski and Leonard, 2010
), ICOS (Tafuri et al., 2001
), CXCR5 (Ansel et al., 1999
), and PD-1 (Agata et al., 1996
) are induced by initial T cell activation, Tfh cells may specifically maintain or increase their expression, perhaps due to prolonged or secondary antigen encounter (Kerfoot et al., 2011
; Kitano et al., 2011
). Our findings are consistent with the concept that Tfh cell differentiation is a multistep process that requires secondary interactions after initial activation by DCs (Baumjohann et al., 2011
; Choi et al., 2011
; Deenick et al., 2010
; Goenka et al., 2011
Recent studies suggest that Tfh cells exhibit features overlapping with other effector Th cell populations (Spolski and Leonard, 2010
). Consistent with this idea, we found that in vitro Tfh-like cells can be induced to produce other effector cytokines. Likewise, other effector CD4+
T cell populations, including Th1, Th2, and Th17, can be induced to express Tfh markers. Our results provide evidence for reprogramming capacity at least between early Tfh-like cells and other effector Th cell populations. This reciprocal plasticity in cytokine production does not appear to result solely from the presence of non-differentiated cells in the culture, since many of the Tfh-like cells produced both IL-21 and other effector cytokines upon exposure to secondary polarization. Notably, our findings were confirmed with sorted CXCR5+
Tfh cells isolated directly ex vivo, which also could be induced to produce IFN-γ, IL-4, and IL-17 after exposure to polarizing cytokines.
This potential for plasticity is reinforced by evaluation of the chromatin landscape of the genes encoding the master regulators of Th cell differentiation. Although variability was observed in certain markings, Tfh-like cells consistently displayed active marks on the Tbx21, Gata3 and Rorc loci, even though the vast majority of these cells expressed IL-21 in the absence of other cytokines. Importantly, positive chromatin modifications on Tbx21, Gata3 and Rorc were also observed in sorted in vitro CXCR5+PD-1+ Tfh-like cells, as well as Tfh cells sorted directly ex vivo.
Conversely, in most non-Tfh-like cells, including naïve cells, Bcl6
exhibited active chromatin marks. The universal presence of H3K4me3 marks on Bcl6
may reflect the ability of multiple Th cell populations (including naïve cells) to initially express markers of Tfh cells upon activation. Even though recent data demonstrate an early bifurcation of Tfh and non-Tfh cells (Choi et al., 2011
), our data suggest these populations retain the potential to express both Bcl6 and Blimp1, and therefore maintain the ability to adopt Tfh or other Th properties in the proper context. Indeed, Bcl6 is co-expressed with and regulates other Th cell-specific transcription factors, suggesting it is the balance of transcription factors that determines cell phenotype (O’Shea and Paul, 2010
; Oestreich et al., 2011
; Zhou et al., 2009
; Zhu et al., 2010
). The presence of active chromatin modifications on genes encoding Th effector lineage-specific transcription factors suggest that Tfh-like cells have a global reciprocal reprogramming capacity with other effector Th cell populations. Although we cannot rule out these findings could reflect heterogeneity in our cultured cells or even in Tfh cells sorted from immunized mice at the early time points used in this study (six days post-immunization), our results suggest Tfh cells constitute a state of differentiation for CD4+
effector cells that can be associated with multiple cytokine-producing capacities.
Our findings therefore argue that CD4+ T cells can gain distinct cytokine expression profiles of given effector Th cell populations upon activation by DCs, yet still maintain the potential to adopt Tfh cell characteristics. This flexibility would permit CD4+ T cells to migrate to the B cell follicle for induction of GCs, even after expression of other cytokine producing fates, allowing them to orchestrate appropriate antibody class switching and production of specific immunoglobulin subclasses suited for a given infection. Conversely, cells that adopt Tfh characteristics, such in the context of prolonged or repeated stimulation, may still possess the potential to produce other effector cytokines, thereby allowing appropriate contouring of humoral responses. Our work, therefore, suggests flexibility in effector T cell responses that permit CD4+ T cells to effectively coordinate integrated immune responses to pathogens. Whether such plasticity can be exploited to better induce specific antibody isotypes may be important for development of optimal long-term humoral immunity in response to vaccines and other immune challenges.