Chromatin remodeling by the NuRD complex regulates development of follicular helper and regulatory T cells

Significance

Production of high-affinity antibody responses after infection or vaccination requires precise control of germinal center B cells by follicular helper T cells and follicular regulatory T cells. Although the Bcl6 transcription factor plays a central role in follicular T cell differentiation, the molecular basis of Bcl6 control has been clouded in uncertainty. Here we report that Bcl6-dependent control reflects the formation of a macromolecular complex between Bcl6 and the Mi-2β-nucleosome remodeling deacetylase complex (Mi-2β-NuRD). The repressive activity of this intranuclear complex potentiates the follicular T cell phenotype and inhibits alternative T cell fates. Identification of this intracellular complex may facilitate new targeted approaches to the treatment of autoimmune disorders.

Abstract

Lineage commitment and differentiation into CD4⁺ T cell subsets reflect an interplay between chromatin regulators and transcription factors (TF). Follicular T cell development is regulated by the Bcl6 TF, which helps determine the phenotype and follicular localization of both CD4⁺ follicular helper T cells (T_FH) and follicular regulatory T cells (T_FR). Here we show that Bcl6-dependent control of follicular T cells is mediated by a complex formed between Bcl6 and the Mi-2β-nucleosome-remodeling deacetylase complex (Mi-2β-NuRD). Formation of this complex reflects the contribution of the intracellular isoform of osteopontin (OPN-i), which acts as a scaffold to stabilize binding between Bcl6 and the NuRD complex that together regulate the genetic program of both T_FH and T_FR cells. Defective assembly of the Bcl6–NuRD complex distorts follicular T cell differentiation, resulting in impaired T_FR development and skewing of the T_FH lineage toward a T_H1-like program that includes expression of Blimp1, Tbet, granzyme B, and IFNγ. These findings define a core Bcl6-directed transcriptional complex that enables CD4⁺ follicular T cells to regulate the germinal center response.

The germinal center (GC) response is a highly dynamic process in tissues where high levels of dying cells provide a battery of self-antigens that can activate autoreactive antibody responses (1). Generation of high-affinity antibodies and avoidance of autoimmune responses after microbial infection or vaccination requires precise control of the GC reaction, depending, to a large degree, on the combined activities of CD4⁺ follicular helper T (T_FH) and follicular regulatory T (T_FR) cells (2–6). T_FH cells that arise from naive CD4⁺ T cells induce GC formation and help B cells to produce protective antibody responses to invading pathogens through generation of memory B cells and long-lived plasma cells (2, 3, 7). T_FR cells that originate from FoxP3⁺ Treg precursors dampen T_FH-driven GC responses and can prevent the emergence of autoreactive B cells and consequent autoantibody production (4–6). While T_FH and T_FR cells have opposing functions, shared expression of the Bcl6 TF serves to repress alternative differentiation pathways (5, 6). Although engagement of the T cell antigen receptor (TCR) and costimulatory receptor inducible T cell costimulator (ICOS) has been implicated in this process (4–6, 8), the conserved genetic and epigenetic mechanisms that ensure Bcl6-directed differentiation of this critical pair of follicular T cells remain largely unknown.

Differentiation of CD4⁺ T cells following engagement of the TCR and costimulatory receptors is determined by changes in gene expression, which in part reflect chromatin modifications that shape transcription, differentiation, and cellular replication. Regulation of gene expression during differentiation of T_FH and T_FR cells depends on Bcl6-dependent recruitment of corepressor complexes that help shape the chromatin landscape surrounding Bcl6 target loci, including Prdm1 (encoding Blimp1) and other genes that may promote alternative T-helper (T_H)-cell fates (9, 10).

The Mi-2β-nucleosome-remodeling deacetylase complex (Mi-2β-NuRD) couples a histone deacetylase and a nucleosome-stimulated ATPase to several corepressors, including a family of metastasis-associated (MTA) proteins (11, 12), which can repress transcription following interactions with site-specific DNA binding proteins (11). Previous studies have indicated that B cell development may reflect recruitment of Mi-2β-NuRD to Bcl6 target loci by MTA3, a cell-type-specific subunit of the Mi-2β-NuRD complex (12). Recent analysis of the Bcl6 secondary repression domain (Bcl6-RD2) has also suggested that MTA3 may interact with Bcl6 in CD4⁺ T_FH cells (13). However, whether Bcl6, MTA3, and Mi-2β-NuRD form a complex in T_FH and T_FR cells and the impact of a putative Bcl6–MTA3–Mi-2β-NuRD complex on follicular T cell differentiation during an immune response is unknown.

Our recent analysis of CD4⁺ T-helper responses has revealed that expression of the intracellular isoform of osteopontin (OPN-i) is essential for the differentiation of both follicular T cell subsets –T_FH and T_FR cells (4). For example, analysis of T_FH cells indicates that engagement of ICOS on T_FH and T_FR cells promotes nuclear translocation of OPN-i, binding to Bcl6 via the RD2 domain and protection of the Bcl6–OPN-i complex from proteasomal degradation to allow sustained T_FH/T_FR responses following initial lineage commitment (4).

Here we analyze the transcriptional events that confer commitment to the two major follicular T cell lineages. We noted a surprising and profound defect in early T_FH/T_FR lineage commitment by OPN-i–deficient cells despite intact Bcl6 protein levels. Analyses of the complex formed by OPN-i, Bcl6, and Mi-2β-NuRD revealed that the OPN-i protein acts as a scaffold that supports the formation of a complex between Bcl6 and MTA3 that mediates the genetic programming of T_FH and T_FR cells (SI Appendix, Fig. S1). Additional interrogation of the biologic activity of this complex revealed that OPN-i–dependent recruitment of the Bcl6–Mi-2β-NuRD complex to Bcl6 target loci is a critical step in the transcriptional repression of the Prdm1 locus and commitment to the T_FH and T_FR cell genetic program.

Results

OPN-i Deficiency Impairs T_FH and T_FR Early Commitment.

To define the impact of OPN-i deficiency on early commitment of T_FH and T_FR cells, we used Spp1^flstop mice bearing a mutated Spp1 allele that allows expression of the OPN-i isoform after Cre-mediated recombination. These Spp1^flstopCre⁺ mice are termed OPN-i-knock-in (OPN-i-KI) mice, while Spp1^flstopCre^– mice are OPN-knockout (OPN-KO) mice (4). We then isolated CD4⁺ T cells from OPN-i-KI or OPN-KO mice that coexpress the OT-II [ovalbumin (OVA)-specific] TCR transgene. Since T_FH commitment occurs as early as 72 h in vivo (8), we analyzed T_H cell differentiation at 2.5 d after transfer of these CD4⁺ T cells along with B cells into Rag₂^−/−Prf1^−/− mice followed by immunization with NP₁₃-OVA in Complete Freunds’ Adjuvant (CFA) (Fig. 1). Bcl6 protein levels were not affected by OPN-i deficiency at this early time point (Fig. 1A) (4). However, OPN-KO CD4⁺ T cells displayed a marked impairment in T_FH commitment, as reflected by reduced proportions of CD4⁺ T cells expressing CXCR5 compared with OPN-i-KI cells (Fig. 1B). A bifurcation between T_FH and other effector T_H cells, particularly T_H1 cells, occurs during early T_H cell fate determination (8). Analysis of the T_H1-cell-associated phenotype of these differentiating cells revealed that a substantially increased proportion of OPN-KO CD4⁺ T cells expressed the Tbet, Ly6C, and granzyme B triad, which characterize a T_H1-like phenotype (14). As a consequence, OPN-KO CD4⁺ T cells displayed an increased ratio of triad⁺ CD4⁺ T cells to CXCR5⁺ CD4⁺ T cells compared with their OPN-i-KI counterparts (Fig. 1B), suggesting that OPN-i deficiency might impair early T_FH commitment and precede the reduced Bcl6 protein levels that occur later in the response (4).

Fig. 1.

OPN-i deficiency impairs T_FH and T_FR early commitment. (A and B) FACS analysis of T_H cell differentiation at day 2.5 after transfer of OT-II CD4⁺ T cells and B cells into Rag₂^−/−Prf1^−/− mice followed by immunization with NP₁₃-OVA in CFA. (A) Histogram overlays of intracellular protein expression (gated on FoxP3^– CD4⁺ T cells). (B) Plots of non-T_FH and T_FH phenotype (gated on FoxP3^– CD4⁺ T cells) and mean ratios of non-T_FH to T_FH cells are shown for OPN-i-KI and OPN-KO mice (n = 3–4 for each group). GzmB, granzyme B. (C and D) Treg from CD45.2⁺ WT, OPN-i-KI, or OPN-KO mice were transferred along with naive CD45.1⁺ CD4⁺ T cells into Tcra^−/− mice followed by immunization with NP₁₃-OVA in CFA. Analysis of CD45.2⁺ Treg cells (gated on FoxP3⁺) 3 d postimmunization. Histogram overlays (C) and quantitation of mean fluorescence intensity (MFI) (D) of each protein (n = 3 for each group). Data shown are representative of three independent experiments (*P < 0.05 and **P < 0.01). Error bars indicate mean ± SEM.

Bcl6-dependent differentiation of T_FH cells includes repression of an alternative Blimp1-associated non-T_FH program (Fig. 1) (9, 15). We therefore asked whether OPN-i deficiency altered the Bcl6−Blimp1 balance during early CD4⁺ T_H cell differentiation. We used Blimp1-YFP reporter mice to generate Blimp1-YFP×OPN-KO mice and Blimp1-YFP×OPN-i-KI mice. Analysis of T_FH differentiation at day 2.5 postimmunization revealed that the proportions of Blimp1⁺ CD4 effector T cells (FoxP3⁻) were considerably higher in OPN-KO mice than OPN-i-KI mice, despite unimpaired Bcl6 protein expression (SI Appendix, Fig. S2). Moreover, higher frequencies of Blimp1⁺ FoxP3⁺ CD4⁺ Treg were also noted in OPN-KO mice compared with OPN-i-KI mice (SI Appendix, Fig. S2), opening the possibility that OPN-i deficiency might impair Bcl6-dependent repression of Blimp1 transcription in both T_FH and T_FR cells.

We then asked whether early T_FR differentiation was also affected by OPN-i deficiency using the approach described above (Fig. 1 A and B). We transferred CD25^hi Treg cells from WT, OPN-KO, or OPN-i-KI mice along with naive CD4⁺ T cells from CD45.1 congenic mice into Tcra^−/− mice followed by immunization with NP₁₃-OVA in CFA. After 2.5 d, OPN-KO but not OPN WT or OPN-i-KI Treg displayed elevated expression of Blimp1 and Tbet but reduced expression of CXCR5 by FoxP3⁺ T cells (Fig. 1 C and D), suggesting that OPN-i deficiency skewed Treg away from the conventional follicular phenotype. Taken together, these results suggested that OPN-i might regulate early T_FH and T_FR commitment, in part through enhanced Bcl6-dependent repression of alternative genetic programs that might depend on Blimp1 expression.

OPN-i Interacts with MTA3 to Promote the Formation of a Bcl6–MTA3–NuRD Complex.

An interaction between Bcl6 and the Mi-2β-NuRD complex via the MTA3 corepressor contributes to Bcl6 transcriptional repressive activity and B cell fate (12). Previous studies have also indicated that, in response to TCR and ICOS signals, a pool of OPN-i translocates into the nucleus to interact with Bcl6 via the Bcl6-RD2 domain (4). The above findings that a complex formed by OPN-i and Bcl6 might regulate early T_FH and T_FR commitment led us to ask whether OPN-i–dependent formation of a Bcl6–MTA3–Mi-2β-NuRD complex might mediate Bcl6-dependent T_FH and T_FR differentiation. We first analyzed 293T cells transfected with vectors expressing HA-tagged MTA3, OPN-i, and Flag-tagged Bcl6 followed by immunoprecipitation with anti-Flag antibody. Consistent with previous reports (4, 12), immunoblot analysis revealed that Bcl6 interacted with both MTA3 and OPN-i, either directly or indirectly. We also noted that the Bcl6–MTA3 association was enhanced in direct proportion to increasing concentrations of OPN-i (Fig. 2A, Left). Deletion of the N-terminal portion of the Bcl6-RD2 region (Δ120–300), which disrupts the Bcl6–OPN-i interaction (4), prevented OPN-i–dependent enhancement of the Bcl6–MTA3 association (Fig. 2A, Right). Moreover, enhanced binding of Bcl6 to MTA3, a component of the Mi-2β-NuRD complex, was associated with increased binding of Bcl6 to Mi-2β (Fig. 2A), a central component of the Mi-2β-NuRD complex (11, 12).

Fig. 2.

Interaction of OPN-i with MTA3 increases Bcl6–MTA3–NuRD complex formation. (A) Cotransfection of 293T cells with vectors expressing Flag-Bcl6 WT or Flag-Bcl6-RD2 deletion mutant and HA-MTA3 without or with increasing concentrations of OPN-i was followed by immunoprecipitation (IP) with anti-Flag antibody (Ab) and immunoblotting (WB) with indicated Abs. (B) Cell lysates of purified CD44^hiCD25^–CD4⁺ T cells from OT-II×OPN-i-KI or OT-II×OPN-KO mice 3 d after NP₁₃-OVA immunization were immunoprecipitated with anti-Bcl6 Ab or rabbit IgG control before immunoblotting with Abs to Bcl6, MTA3, OPN, and Mi-2β. (C) Interaction of OPN-i and the MTA3 component of NuRD complex in CD4⁺ T cells. Nuclear lysates of purified CD4⁺CD44⁺ T cells from mice at day 5 postimmunization with NP₂₆-KLH in CFA were immunoprecipitated with anti-MTA3 and immunoblotted with Abs to Bcl6 and OPN. Input, immunoblot analysis of an aliquot of lysate without IP. (D) MTA3–OPN-i interaction depends on ELM2 domain of MTA3. 293T cells were cotransfected with vectors expressing HA-MTA3 WT or its deletion mutants (diagramed above) with or without OPN-i, followed by IP with anti-HA and immunoblotting with anti-HA and anti-OPN Abs. (E) Deletion of the ELM2 domain impairs the Bcl6–MTA3 interaction. 293T cells were cotransfected with plasmids expressing Flag-Bcl6, HA-MTA3 WT, or its deletion mutants with or without OPN-i, followed by IP with anti-Flag Ab and immunoblotting with the indicated Abs. Data shown are representative of three independent experiments.

We then asked whether OPN-i–mediated enhancement of Bcl6–MTA3–Mi-2β-NuRD complex formation noted above might be apparent in primary CD4⁺ T cells. Analysis of Bcl6-associated proteins expressed by CD4⁺ T cells from OT-II×OPN-i-KI mice compared with CD4⁺ T cells from OT-II×OPN-KO mice 3 d after immunization indicated that OPN-i deficiency greatly reduced the association of Bcl6 with MTA3 and Mi-2β in OPN-KO CD4⁺ T cells, despite unaltered Bcl6 protein expression (Fig. 2B). These results suggested that OPN-i promoted formation of the Bcl6–MTA3–Mi-2β-NuRD complex.

The regulatory activity of the Mi-2β-NuRD complex depends on the activity of its corepressor components, including MTA3 family members, which may demarcate distinct forms of Mi-2β-NuRD that control cell-type-specific transcription (11, 16). We noted that MTA3 bound to both Bcl6 and OPN-i within the nucleus of CD4⁺ T cells (Fig. 2C). We further defined the OPN-i interaction with MTA3 according to mutational analysis (Fig. 2D). We found that a specific interaction between OPN-i and the ELM2 domain of MTA3 promoted binding of the complex to Bcl6. Thus, MTA3-ELM2 deletion mutants (but not MTA3-WT or MTA3-BAH mutants) failed to bind to Bcl6, as judged by anti-Flag (Bcl6) immunoprecipitation (Fig. 2E). These findings suggest that binding of OPN-i to both Bcl6 and MTA3 allows OPN-i to function as a scaffold or bridge to promote the association of Bcl6 with the Mi2β-NuRD complex (SI Appendix, Fig. S1).

OPN-i Promotes Bcl6–MTA3-Dependent Repression of Prdm1/Ifnγ Expression by T_H1 Cells.

Repression of Blimp1 and other non-T_FH genes by Bcl6 plays a central role in T_FH commitment and maintenance of the T_FH phenotype (9, 10). To determine whether the OPN-i–dependent association between Bcl6 and MTA3–Mi-2β-NuRD noted above contributed to Bcl6 transcriptional repression of canonical T_H1 genes, we asked whether forced expression of Bcl6 alone or with MTA3 in T_H1 cells [which do not express significant levels of Bcl6 or MTA3 (4)], might reprogram this CD4⁺ T_H subset. We therefore infected in-vitro–differentiated T_H1 cells [after 5 d culture as described previously (17)] with retroviruses expressing Bcl6, MTA3, or both Bcl6 and MTA3. Quantitative RT-PCR analysis of T_H1-associated gene expression showed that retroviral coexpression of Bcl6 and MTA3, but not expression of either retrovirus alone, substantially repressed both Prdm1 and Ifnγ expression (Fig. 3A). The specificity of this response was confirmed by the finding that transduction of these T_H1 cells with a retrovirus expressing the N-terminal Bcl6-RD2 deletion mutant (Δ120–300), which impairs the MTA3–Bcl6 interaction (Fig. 2A), failed to repress Prdm1 or Ifnγ expression even at the highest dose tested (Fig. 3B).

Fig. 3.

OPN-i promotes Bcl6–MTA3-dependent repression of Prdm1 and Ifnγ expression in T_H1 cells, which requires the MTA3 ELM2 domain. (A) OT-II T_H1 cells were transduced with retroviral vectors expressing either Flag-Bcl6 or HA-MTA3 alone [multiplicity of infection (MOI) = 10]; or a mixture containing constant Flag-Bcl6 concentrations (MOI = 10) combined with increasing concentrations of HA-MTA3 (wedge: 2.5, 5, 10). (B) OT-II T_H1 cells were transduced with a mixture of retroviral vectors expressing constant HA-MTA3 (MOI = 10) combined with Flag-Bcl6 at increased MOI (wedge: 2.5, 5, 10), or with Flag-Bcl6-RD2 deletion mutant (MOI = 10). (C) OT-II T_H1 cells were transduced with a mixture of retroviral vectors expressing constant concentrations of Flag-Bcl6 (MOI = 10) with HA-MTA3 ELM2 deletion mutant (MOI = 10). (D) OT-II T_H1 cells were transduced with retroviral vectors expressing [Flag-Bcl6 (MOI = 5) + HA-MTA3 (MOI = 5)] or OPN-i (MOI = 5), or Flag-Bcl6 + HA-MTA3 + OPN-i (each at a suboptimal MOI of 5). qRT-PCR was performed after 2.5 d. Gene expression was normalized to expression of the control gene Rps18 (encoding ribosomal protein S18) and presented as relative to cells transduced with control virus, set as 1. Data shown are representative of three independent experiments (*P < 0.05, **P < 0.01, and ***P < 0.001). Error bars indicate mean ± SEM.

In view of our findings that localized the interaction of OPN-i with MTA3 to the MTA3-ELM2 domain (Fig. 2 D and E), we tested the functional impact of this interaction on the CD4⁺ T cell genotype in vitro. We observed that transduction of T_H1 cells (after differentiation from CD25^–CD4⁺ T cells from OT-II×OPN-i-KI mice) with a retrovirus expressing Bcl6 and the MTA3 protein but not the MTA3-ELM2 deletion mutant suppressed Prdm1 or Ifnγ expression (Fig. 3C). Moreover, limiting concentrations of Bcl6 and MTA3, which did not repress Prdm1 or Ifnγ, fully repressed these genes in the presence of OPN-i (Fig. 3D), consistent with earlier findings that coexpression of OPN-i enhances the repressive efficiency of Bcl6–MTA3. These findings together are consistent with the ability of OPN-i to promote the biochemical association of Bcl6 with MTA3–Mi-2β-NuRD (Fig. 2).

Promotion of T_FH and T_FR Differentiation in Vivo Requires an Interaction Between OPN-i and MTA3.

The specificity of the OPN-i–Bcl6–MTA3 interaction described above is supported by findings that deletion of the ELM2 domain of MTA3 disrupts binding of OPN-i to MTA3 (Fig. 2D), impairs MTA3 binding to Bcl6 (Fig. 2E), and decreases Bcl6–MTA3-dependent repression of Prdm1/Ifnγ expression (Fig. 3C). We then tested the physiological relevance of the OPN-i–Bcl6–MTA3 interaction to T_FH and T_FR differentiation in vivo using a retroviral reconstitution system (4). We transduced in-vitro–activated OT-II CD4⁺ T cells with a retroviral vector expressing GFP alone (empty vector, EV) or GFP plus either WT MTA3 (MTA3) or the MTA3-ELM2 deletion mutant (delELM2). Since MTA3 and delELM2 are expressed within the same bicistronic IRES retroviral vector as GFP, their expression is correlated with GFP levels. We transferred sorted GFP⁺ cells into Tcra^−/− hosts followed by immunization with NP₁₃-OVA in CFA (Fig. 4A and SI Appendix, Fig. S3A). T_FH differentiation and associated GC B cell formation were increased for OT-II CD4⁺ T cells transduced to express WT MTA3 compared with CD4⁺ T cells transduced with EV (Fig. 4B). In contrast, transduction of OT-II CD4⁺ T cells with the delELM2 mutant resulted in decreased numbers of T_FH and GC B cells (Fig. 4B). Consequently, both the total (anti-NP₂₃) and high-affinity (anti-NP₄) NP-specific antibody responses were markedly impaired (Fig. 4C). Although Bcl6 levels were not altered among cells expressing GFP alone, MTA3, or the delELM2 mutant, transduction with the delELM2 mutant exerted a “dominant negative” impact on T_FH cell function, as judged by increased Tbet, Ly6C, and Blimp1 expression (Fig. 4 D and E).

Fig. 4.

OPN-i–mediated promotion of T_FH differentiation requires intact OPN-i−MTA3 interaction. (A and F) Schematic diagrams of experimental protocols. Purified naive OT-II×OPN-i-KI CD4⁺ T cells were activated in vitro, transduced with retroviral vector encoding GFP alone (EV) or GFP plus WT MTA3 (MTA3) or deletion mutant MTA3 (delELM2). GFP⁺ CD4⁺ T cells (A) and GFP^hi or GFP^med-lo CD4⁺ T cells (F) were then sorted and transferred into Tcra^−/− hosts followed by immunization with NP₁₃-OVA in CFA. B–E were analyzed from protocol A, and G and H from protocol F. (B and G) FACS analysis of T_FH cells (gated on FoxP3^– CD4⁺ T cells) and GC B cells (gated on B220⁺ cells) 10 d postimmunization. (D and H) Histogram overlays of Bcl6, Tbet, Ly6C, and Blimp1 expression in donor CD4⁺ T cells (gated on FoxP3^– CD4⁺ T cells). (E) Quantitation of MFI of each protein and frequency of Ly6C⁺CD4⁺FoxP3^– cells in D. Data shown are representative of two independent experiments. (C) Ectopic expression of the delELM2 mutant in OT-II CD4⁺ T cells impairs the Ab response postimmunization. Anti-NP₂₃ or anti-NP₄ antibody titers were determined from mice transferred with OT-II CD4⁺ T cells expressing empty control, WT MTA3, or delELM2 mutant followed by immunization, as in A. ***P < 0.001, and ns, no significance. Error bars indicate mean ± SEM.

The above finding that CD4⁺ T cells expressing the delELM2 mutant failed to differentiate into functional T_FH cells led us to ask whether the delELM2 MTA3 mutant could compete with endogenous MTA3 to interfere with T_FH differentiation. To address this, we transduced OT-II CD4⁺ T cells with a retroviral vector expressing GFP alone or GFP plus the delELM2 mutant, and then transferred sorted GFP^hi or GFP^med-lo CD4⁺ T cells separately into Tcra^−/− mice followed by immunization with NP₁₃-OVA in CFA (Fig. 4F and SI Appendix, Fig. S3B). Consistent with these results (Fig. 4B), T_FH differentiation was decreased for OT-II CD4⁺ T cells transduced to express the delELM2 mutant compared with cells expressing GFP alone, which was associated with reduced GC B cells (Fig. 4G). The frequencies of T_FH and GC B cells were also not affected by GFP levels in OT-II CD4⁺ T cells expressing GFP alone. In contrast, the extent of T_FH differentiation and GC B cell formation in mice transferred with OT-II CD4⁺ T cells expressing delELM2 was negatively correlated with levels of delELM2 (GFP) and associated with increased expression of non–T_FH-associated markers (Tbet, Ly6C, and Blimp1) (Fig. 4 G and H). These results suggest that expression of the ectopic delELM2 mutant might impede T_FH differentiation by competing with the endogenous MTA3 in a dose-dependent manner.

Using a retroviral reconstitution system similar to that described above, we evaluated the physiological contribution of the OPN-i–Bcl6–MTA3 interaction to the formation of T_FR cells. We transduced CD45.2⁺ WT CD25⁺CD4⁺ T cells with retroviral vectors expressing GFP alone (EV) or GFP plus WT MTA3 (MTA3) or delELM2 mutant (delELM2), then transferred each group of cells together with CD45.1⁺ CD25^–CD4⁺ T cells into Tcra^−/− mice, followed by immunization of hosts with NP₁₃-OVA in CFA (Fig. 5A). T_FR differentiation was increased for CD45.2⁺ Treg transduced to express WT MTA3 compared with cells transduced with EV (Fig. 5B and SI Appendix, Fig. S4). In contrast, expression of the delELM2 mutant in CD45.2⁺ Treg decreased T_FR differentiation to levels comparable to cells expressing GFP alone, which was associated with increased T_FH differentiation and GC B cell formation (Fig. 5B and SI Appendix, Fig. S4). These results indicate that an OPN-i−MTA3 interaction required for Bcl6−MTA3−NuRD complex formation in vitro is also essential for T_FH and T_FR differentiation in vivo.

Fig. 5.

OPN-i–mediated promotion of T_FR differentiation requires intact OPN-i−MTA3 interaction. (A) Schematic diagram of experimental procedure. Purified CD45.2⁺ Treg were activated in vitro, transduced with retroviral vector encoding GFP alone (EV), or GFP plus WT MTA3 (MTA3) or deletion mutant MTA3 (delELM2). GFP⁺ Treg were then sorted and transferred into Tcra^−/− hosts along with CD45.1⁺ naive CD4⁺ T cells followed by immunization with NP₁₃-OVA in CFA. (B) Frequency of T_FR, T_FH, and GC B cells in SI Appendix, Fig. S4. Data shown are representative of two independent experiments. *P < 0.05, **P < 0.01, ns, no significance. Error bars indicate mean ± SEM.

OPN-i Promotes Binding of Bcl6 and MTA3 to Bcl6 Target Genes and Regulates Bcl6 Transcriptional Activity.

To gain insight into the genetic mechanisms that underpinned OPN-i–mediated enhancement of T_FH and T_FR differentiation, we asked whether OPN-i–promoted binding of the Bcl6−MTA3−NuRD complex reflected increased binding to Bcl6 target loci. We focused on Prdm1, since Bcl6-dependent repression of Prdm1 is a key element in the determination of T_FH and T_FR cell fate (Fig. 3 and SI Appendix, Fig. S2). We noted that the mRNA levels of Prdm1 were substantially up-regulated in OPN-KO T_FH and T_FR cells compared with OPN-i-KI cells 3 d postimmunization, despite unaltered Bcl6 mRNA levels in these cells (Fig. 6 A and B). We performed a chromatin immunoprecipitation (ChIP)-qPCR analysis of the Bcl6 and MTA3 occupancy on the conserved Bcl6 response element (BRE) within Prdm1. We observed that Bcl6 binding to the Prdm1 BRE region was substantially decreased in OPN-KO T_FH cells, consistent with a failure of Bcl6 to repress Prdm1 in OPN-i–deficient T_FH cells (Fig. 6A). Moreover, analysis of OPN-KO T_FH cells revealed an almost complete loss of MTA3 bound to the Prdm1 BRE locus (Fig. 6C), further suggesting severely impaired recruitment to this canonical Bcl6 target gene. Since Bcl6 transcriptional repression is mediated in part by recruiting histone deacetylases to target loci via the Mi-2β-NuRD complex, we asked whether OPN-i deficiency influenced the histone acetylation status surrounding Bcl6-bound loci. There was almost no acetylated H3 (AcH3) at the Prdm1 BRE locus of OPN-i-KI T_FH cells, consistent with a repressive chromatin status. In contrast, these loci displayed increased levels of AcH3 in OPN-KO T_FH cells (Fig. 6C), consistent with an active chromatin locus in the absence of OPN-i. Taken together, these results indicate that OPN-i is required for efficient binding of Bcl6 and MTA3 to a major Bcl6 target gene as well as associated repression of this locus.

Fig. 6.

OPN-i promotes binding of Bcl6 and MTA3 to Bcl6 target genes and regulates Bcl6 transcriptional activity. (A and B) qRT-PCR analysis of Prdm1 and Bcl6 in sorted pure (>95%) T_FH cells (A) or T_FR cells (B) from OPN-i-KI and OPN-KO mice at day 3 postimmunization with NP₁₃-OVA in CFA. Gene expression was normalized to expression of the control gene Rps18 (encoding ribosomal protein S18) and expressed as relative to T_FH or T_FR cells from OPN-i-KI mice, set as 1. (C) In-vitro–differentiated T_FH cells from OT-II×OPN-i-KI or OT-II×OPN-KO mice were cross-linked, chromatin prepared, and ChIP-PCR analyses performed for Bcl6, MTA3, and Acetylated H3 (AcH3) at the BRE of Prdm1 gene. Data, shown as the percent of input, reflecting enriched binding at the indicated loci, are representative of three independent experiments (mean ± SEM). *P < 0.05, **P < 0.01, ns, no significance.

Discussion

Specification of T cell fate reflects the concerted action of chromatin regulators and transcription factors in response to signals emanating mainly from the TCR and costimulatory receptors. Our studies suggest that the functional differentiation of T_FH and T_FR cells is mediated, in part, by recruitment of the Mi-2β-NuRD complex to specific Bcl6 target loci. The formation of this complex in differentiating CD4⁺ T cells requires the scaffold-like contribution of OPN-i to the binding of Bcl6 to the MTA3–Mi-2β-NuRD complex and formation of a biologically active corepressor complex.

The transcriptional activity of Bcl6 in other cell types may also reflect recruitment of different corepressor complexes to different Bcl6 domains and the formation of target-specific complexes. For example, an interaction between the Bcl6-BTB domain and the BCOR/SMART corepressors promotes GC B cell differentiation without a significant effect on the T_FH cell response (18). In contrast, previous studies of T_FH differentiation have underlined the significance of an interaction between the Bcl6-RD2 domain and MTA3 (13) as well as a second interaction with OPN-i (4). Here we identify OPN-i as a critical bridging intermediary that facilitates binding between Bcl6 and MTA3 and promotes the formation of a Bcl6–NuRD complex that is equipped to direct both T_FH and T_FR cell differentiation. The extended and flexible structure of OPN-i, a member of the SIBLING protein family (19), may permit interactions with a variety of partners, including the Mi-2β-NuRD macromolecule in the nucleus, as described here, as well as with proteasomal complexes in the cytosol, as noted previously (20), to promote Bcl6-directed differentiation of follicular T cells.

These findings provide insight into the epigenetic mechanisms that govern lineage commitment of the follicular T cell pair that regulates GC antibody and autoantibody responses. Our findings also help clarify the differentiative relationship between the T_FH and T_FR cell lineages. Although T_FH and T_FR cells coexpress Bcl6 as well as several surface receptors, the shared genetic elements responsible for follicular differentiation of these two CD4⁺ T cell lineages have been obscure. We have reported previously that T_FH and T_FR cells may share an ICOS-dependent pathway that promotes the formation of an intranuclear complex between Bcl6 and OPN-i that protects the Bcl6 protein from proteasomal degradation (4). Here we identify an additional role for OPN-i in T_FH differentiation, i.e., integration of Bcl6 with Mi-2β-NuRD to form biologically active complexes that enhance T_FH and T_FR lineage differentiation. Regulation of Prdm1 and other canonical target genes by this complex may account for core features shared by differentiated T_FH and T_FR cells that reside in the germinal centers and lymphoid tissue follicles. Formation of this Bcl6 complex may represent a critical downstream consequence of the ICOS-dependent pathway that favors the differentiation of follicular T cells from CD4⁺ precursors (4, 8). Since the ratio of T_FH to T_FR cells has a direct impact on the intensity and quality of GC antibody responses (4, 21), a detailed correlation between the T_FH/T_FR ratio and the intensity and quality of the B cell response at defined intervals after immunization is necessary to fully evaluate the impact of this Bcl6-containing complex on the immune response.

Our finding that Bcl6 transcriptional activity in T_FH cells depends on its association with the complex described here also suggests that anti-Bcl6-based ChIP-seq analysis of T_FH cells may lack the specificity necessary to precisely define the genetic program of T_FH (and T_FR) cells. A precedent for this comes from analysis of early B cell differentiative steps that are regulated by Ikaros–Mi-2β-NuRD complexes. These studies indicate that combined occupation of target loci in early B cells by Ikaros and Mi-2β-NuRD, but not by Ikaros alone, is essential for functional control of early B cell differentiation genes (22). Recent genomewide Bcl6 ChIP-seq analysis of human GC T_FH cells has indicated that Bcl6 binds to over 8,500 target loci that localize predominantly to promoter regions (23), while analysis of murine T_FH cells has revealed about 5,100 Bcl6 binding peaks localized mainly to intron and intergeneic regions (24). It is likely that a more precise identification of the key target genes that control T_FH and T_FR differentiation may come from identification of the genetic loci that are cooccupied by both Bcl6 and the partner Mi-2β-NuRD complex identified in this study.

Our findings are also relevant to understanding pathways that lead to autoimmunity. Aberrant or altered interactions with target gene loci by the Bcl6–OPN-i–Mi-2β-NuRD complex in T_FH and T_FR cells are likely to be associated with dysregulated differentiation of these cells, and potential autoimmune or inflammatory sequelae. Analysis of the chromatin landscape surrounding genes targeted by the Bcl6–OPN-i–Mi-2β-NuRD complex in follicular CD4⁺ T cells from autoimmune-prone and autoimmune-resistant mouse strains may reveal new disease susceptibility loci and a molecular foothold for new approaches to these disorders.

Methods

Mice.

C57BL/6J (B6), Tcra^−/−, OT-II transgenic [B6.Cg-Tg(TcraTcrb)425Cbn/J], Blimp1-YFP reporter [B6.Cg-Tg(Prdm1-EYFP)1Mnz/J] (Jackson Labs), Rag2^−/−Prf1^−/−, B6SJL (CD45.1) (Taconic Farms), Spp1^flstopCre⁺, and Cre^– littermates (4) were housed in pathogen-free conditions and used at 7–12 wk of age. Experiments were performed in an unblinded fashion, with both sexes included for all experiments. All experiments were performed in compliance with federal laws and institutional guidelines as approved by Dana-Farber Cancer Institute’s Animal Care and Use Committee.

Statistical Analyses.

Statistical analyses were performed using two-tailed, unpaired Student’s t test or Mann–Whitney test with the assumption of equal sample variance, with GraphPad Prism V6 software. Error bars indicate mean ± SEM. A P value < 0.05 was considered to be statistically significant (*≤ 0.05, **≤ 0.01, ***≤ 0.001). No exclusion of data points was used. Sample size was not specifically predetermined, but the number of mice used was consistent with previous experience with similar experiments.

Additional methods are provided in SI Appendix.