IL4, IL21, and IFNγ interact to govern TBET and CD11c expression in TLR-activated B cells

Abstract

TBET and CD11c expression in B cells is linked with IgG_2c isotype switching, virus-specific immune responses, and humoral autoimmunity. However, the activation requisites and regulatory cues governing TBET and CD11c expression remain poorly defined. Herein we reveal a relationship between TLR engagement, IL4, IL21, and IFNγ that regulates TBET expression in B cells. We find that IL21 or IFNγ directly promote TBET⁺ expression in the context of TLR engagement. Further, IL4 antagonizes TBET induction. Finally, IL21, but not IFNγ, promotes CD11c expression independent of TBET. Using influenza virus and H. polygyrus infections, we show that these interactions function in vivo to determine whether TBET⁺ and CD11c⁺ B cells are formed. These findings suggest that TBET⁺ B cells seen in health and disease share the common initiating features of TLR driven activation within this circumscribed cytokine milieu.

INTRODUCTION

Although initially implicated in CD4 T cell differentiation, T-Box Expressed in T cells (TBET) is a key transcriptional regulator in many immune cells. B cell intrinsic TBET expression fosters switching to IgG_2a (1–3), an isotype associated with both T_H1-driven antibody responses and humoral autoimmunity (4, 5). Moreover, TBET is required for the generation of age associated B cells (ABCs), which are transcriptionally distinct from other B cell subsets and have also been associated with both viral clearance and humoral autoimmunity (6–8). Finally, many TBET⁺ B cells express CD11c, a phenotype associated with viral or bacterial infections, autoimmunity, and neoplasia (7, 9–12). Despite growing appreciation for the importance of TBET-expressing B cell subsets, the signals that yield B lineage effectors characterized by TBET expression–as well as how these regulate appropriate versus pathogenic outcomes–remain poorly defined. Candidates include cell-intrinsic cues from adaptive and innate receptors, including the BCR and toll-like receptors (TLRs), as well as signals from T follicular helper (T_FH) cells. In this regard, several T_H1 cytokines, including IL12, IL18, and IFNγ, can induce TBET in activated B cells (4, 5). Nonetheless, the roles and interactions of canonical T_FH cytokines–IL21, IL4, and IFNγ–in regulating TBET expression have not been systematically interrogated (13–15).

Here, we show that mouse and human B cells integrate signals from IL4, IL21, and IFNγ to regulate TBET expression. In the context of TLR engagement, both IL21 and IFNγ directly drive follicular (FO) B cells to express TBET in vitro. However, IL4 antagonizes IL21-driven TBET upregulation, but enhances IFNγ-induced TBET expression. Moreover, IL21 but not IFNγ promotes CD11c expression. Consistent with these in vitro results, the in vivo frequencies of germinal center (GC) and memory (B_MEM) B cells expressing TBET or CD11c vary based on the prevailing cytokine milieu. Finally, using viral and helminthic infections in single and double cytokine KO mice, we show that the relative abundance of these cytokines determines GC and B_MEM cells generated during ongoing immune responses express TBET and CD11c. Together, these findings reveal a previously unappreciated interplay of IL4, IL21, and IFNγ that, in concert with innate sensors, controls TBET and CD11c expression in B cells.

MATERIALS & METHODS

Mice

Tbx21^−/−, Stat6^−/−, Tbx21^f/fCd19^Cre/+, C57BL/6, and BALB/c mice were maintained and used in accordance with the University of Pennsylvania Institutional Animal Care and Use Committee (IACUC) guidelines. University of Pennsylvania IACUC approved all animal experiments. Il4^−/− mice were a gift from Dr. Paula Oliver. Ifng^−/− mice were a gift from Dr. Edward Behrens. Il4^−/−Ifng^−/− double deficient mice were bred in house. Il21r^−/− and Il21tg spleens and sera were shipped overnight on ice from Dr. Warren Leonard. All mice were 2–6 months of age.

Infections

Mice were infected by oral gavage with 200 infectious larvae of Heligmosomoides polygyrus (HP) as previously described (16). Mice were infected by intranasal (i.n.) infection with 30 TCID50 of influenza strain A/PR/8/34 (PR8) (ATCC).

In vitro cultures

Mouse CD23⁺ splenic B cells were enriched by magnetic positive selection (Miltenyi Biotec), labeled with either Violet Cell Trace (VCT, Invitrogen) or CFSE (eBioscience), and cultured as previously described (17). Human PBMCs were isolated from blood samples obtained from healthy donors that expressed written informed consent and after ethical approval by the U Penn IRB. All investigations were conducted according to the principles expressed in the Declaration of Helsinki. Human B cells were enriched by CD27 microbead negative selection followed by CD19 microbead positive selection (Miltenyi Biotec), labeled with CFSE, and cultured with indicated stimuli for 5 days. Mouse or human IL21, IL4, and IFNγ were used at 25, 10, and 10 ng/mL respectively (Shenandoah). ODN2006 was used at 1uM (Invivogen).

Flow Cytometry

FACS reagents were purchased from BioLegend (BL), Beckton-Dickenson (BD), or eBioscience (eBio): T-BET (4B10, BL), CD11c (N418, BL), IgM (R6-60.2, BD), CD38 (90, eBio), CD138 (281-2, BL), IgD (11–26c.2a, BL), CD4 (RM4-5, BL), B220 (RA3-6B2, BL), CD62L (MEL-14, eBio), TCR-β (H57-597, BL), CD19 (6D5, BL), CXCR5 (L138D7, BL); PD-1 (RMP1-30, BL); CD8 (53-6.7, eBio), CD4 (H129.19, BL); F4/80 (BM8, eBio); Ly-6G/GR1 (RB6-8C5, eBio); CD43 (S7, BD); CD21/CD35 (CR2/CR1, BL); CD23 (B3B4, eBio); CD93 (AA4.1, BL); PNA-FITC (Sigma); Zombie Aqua (BL). FACS analyses were performed as described (17).

Serum antibody titers

ELISAs were performed as previously described (17) using α-mouse IgG_2a, IgG_2b, IgG_2c, or IgG₁ HRP antibodies (SouthernBiotech).

Quantitative PCR analysis & Transcriptional profiling

Quantitative PCR experiments were performed as previously published (17) using the following probes: Il4 (Mm00445260_m1), Ifng (Mm00801778_m1), Il21 (Mm00517640_m1), Tbx21 (Mm00450960_m1), Aicda (Mm00507774_m1). Transcriptional profiling data generated as previously described (18) and have been deposited on the Gene Expression Omnibus (GEO) database for public access (Accession # GSE77145, http://www.dtd.nlm.nih.gov/geo/query/acc.cgi?acc=GSE77145).

Statistics

Student’s t-test was used to generate all P-values, * P <0.05, ** P <0.01, *** P <0.001, **** P <0.0001. Data represented as box and whisker plots with mean depicted as “+”.

RESULTS & DISCUSSION

IL21, IL4, and IFNγ differentially regulate TBET and CD11c expression

In preliminary in vitro studies, we established that IL21 drives TBET expression in mouse FO B cells responding to TLR9, but not BCR or CD40 signals (Fig. 1A). To explore these interactions further, we cultured FO B cells with IL4, IL21, or IFNγ in the presence of TLR7 or TLR9 agonists. Both Tbx21 transcripts and TBET protein increased markedly in FO B cells cultured with IL21 or IFNγ, but IL4 influenced these outcomes differently. IL4 blocked IL21-driven TBET upregulation, but enhanced IFNγ-mediated TBET upregulation (Fig. 1B & Supplemental Fig. 1A).

Figure 1. IL4 and IL21 act in a cell intrinsic manner to regulate TBET expression in vitro.

Magnetically enriched CD23⁺ splenic B cells were cultured in vitro with various combinations of α-Ig-μ (IgM), α-CD40 (40), IL4 (4), IL21 (21), and IFNγ (γ). Mouse data are representative of 3 independent experiments. (A) WT or Cd19^cre/+Tbx21^f/f B cells treated for 48hrs and probed for TBET (ΔMFI=WT-mutant). (B) Tbx21 mRNA levels in WT cells treated for 20hrs. (C) WT, Il21r^−/−, or Stat6^−/− B cells were labeled with either CFSE (green plots) or Violet Cell Trace (VCT, purple plots), treated with ODN1826 and indicated cytokines for 48h, then stained for CD11c and TBET. (D) Magnetically enriched CD27⁻CD19⁺ human B cells were labeled with CFSE, treated for 108h, and probed for TBET on live, CFSE⁻ cells. (E) Frequency of TBET⁺ B cells from each treatment across 6 healthy, adult donors.

To determine whether IL21 and IL4 directly regulate TBET in B cells, either Il21r^−/− or Stat6^−/− B cells were co-cultured with wild type (WT) B cells and stimulated as above. Since IL21R is required for IL21 signaling and STAT6 is the key signal transducer of IL4 (19, 20), we reasoned that co-culturing these mutants with WT cells would reveal any secondary trans effects. In order to track both cell origin and division, WT or KO cells were labeled with VCT or CFSE, respectively (Supplemental Fig. 1B). While IL21-induced TBET expression in WT B cells, the co-cultured Il21r^−/− B cells remained TBET negative (Fig. 1C, top row). Analogously, although IL21-driven TBET upregulation in WT B cells was reversed by IL4, co-cultured Stat6^−/− cells were refractory to this negative effect (Fig. 1C, bottom row). Similar results were obtained using the TLR7 agonist, CL097 (not shown). Importantly, in all cases, IFNγ treatment induced TBET irrespective of Il21r or Stat6 deficiency (Fig. 1C). To assess whether similar relationships exist in human B cells, we cultured CD27⁻CD19⁺ PBMCs as above. TLR9 stimulation alone upregulated TBET in these cultures. It is not clear whether intrinsic effects of TLR signaling or trans effects induced by these signals underlies this observation. Nonetheless, IFNγ significantly increased TBET expression, and IL4 completely blocked TBET in all cultures except those with IFNγ (Fig. 1D & 1E). In toto, these results show that in the context of TLR signaling, IL4, IL21, and IFNγ interact to regulate TBET expression in both mouse and human B cells.

The converse effect of IL4 on IFNγ- vs IL21-induced TBET expression suggests that unique, TBET-associated programs are facilitated by each cytokine. We interrogated this possibility in several ways. First, since previous studies have linked TBET with CD11c expression (7), we asked whether IFNγ or IL21 influence CD11c differently. The results show that IL21 drives CD11c expression, but IFNγ does not (Fig. 1C). Further, as with TBET, IL4 blocks IL21-induced CD11c expression. Finally, IFNγ drives TBET expression, and is not appreciably influenced by either IL4 or IL21 (Supplemental Fig. 1C). These findings indicate that IL21 and IFNγ drive TBET and CD11c expression through distinct mediators, and that TBET expression per se is insufficient for CD11c induction. To further interrogate differential TBET expression driven by IL21 versus IFNγ, as well as to distinguish TBET-dependent and -independent effects of each cytokine, we performed genome-wide transcriptional profiling on WT or Tbx21^−/− B cells stimulated with either IFNγ or IL21. Principal components analysis shows that 82.7% of variance in these data was explained by the cytokine employed, while Tbx21 genotype accounted for 6.3% of the variance (Supplemental Fig. 1D). Further, each cytokine induces a unique transcriptional profile, including some TBET-dependent shifts in gene expression (Supplemental Fig. 1E & Supplemental Table 1). Thus, IFNγ and IL21 drive TBET associated phenotypes in B cells.

Together, these results show that in the context of TLR engagement, the aggregate of IFNγ, IL21, and IL4 signals determines whether B cells express TBET. TLR engagement but not BCR crosslinking (Fig. 1A) appears necessary to position B cells for TBET expression upon subsequent IFNγ or IL21 signaling. We obtained similar results with the TLR2/4 ligand LPS (not shown), suggesting pathways common to most TLRs, and perhaps other innate receptors, provide these key initial signals. We speculate that these signals alter gene loci accessibility for subsequent cytokine cues. Indeed, prior reports that CD11c⁺ or TBET⁺ B cells emerge in responses to a variety of viral and bacterial infections are consistent with this idea (6, 9). Moreover, the differential effects of IL4 on IL21 versus IFNγ suggest a complex interplay of STAT-dependent transcriptional regulation. The clear dose-response relationship of IL4-mediated effects is consistent with the idea that competitive relationships are involved (Supplemental Fig. 1F). Although IL4 and IL21 both require common-γ chain receptor to initiate STAT signal transduction (21), our Stat6^−/− coculture data (Fig. 1C) indicate that competition for membrane proximal receptor components is unlikely to explain these findings. If this were the case, then Stat6^−/− cells would also be subject to IL4’s repressive effects. Instead, downstream events are more likely candidates, including differential occupation of transcriptional regulatory sites, and altered stoichiometric relationships among the JAK-STAT proteins involved.

Relative abundance of IL21, IL4, and IFNγ regulate TBET expression in vivo

Our in vitro findings suggest that IFNγ, IL4, and IL21 interact to modulate TBET and CD11c expression in B cells. As an initial assessment of whether this relationship exists in vivo, we surveyed GC B and B_MEM cells for TBET expression in C57BL/6 (B6) versus BALB/c mice (Supplemental Fig. 1G), because these strains display inherent T_H1 versus T_H2 skewing, respectively (22). We reasoned that if TBET expression is promoted by milieus rich in IFNγ, but repressed in those with plentiful IL4 and little IFNγ, then the frequencies of TBET⁺ B cells in these two strains should differ. In agreement with this prediction, while most GC B cells in B6 mice are TBET⁺, BALB/c have a lower frequency of TBET⁺ GC B cells (Fig. 2A). Importantly, CD11c protein expression was restricted to B6 B_MEM cells (Fig. 2B) and not GC B cells (Supplemental Fig. 1H). These findings are consistent with the notion that IFNγ and IL4 levels regulate TBET expression in GC B cells. To probe the impact of IL21 on this overall relationship, we next asked whether extra-physiological levels of IL21 would foster accumulation of TBET⁺CD11c⁺ B cells. Profound increases in both TBET and CD11c expression were seen in all splenic B cells in Il21tg mice (Fig. 2C), which is consistent with our in vitro results suggesting that IL21 drives both TBET and CD11c expression. Although the partially activated state of B cells in these mice confounds conventional phenotyping strategies, nearly all mature B cells in the Il21tg bear a CD23⁻CD21⁻ phenotype (Supplemental Fig. 1I) identical to the TBET-dependent ABC subset (17, 23). Finally, consistent with TBET’s role in fostering class switch recombination to IgG_2a/c, we observed a marked increase of IgG_2a/c but not IgG₁ serum antibody titers in Il21tg compared to WT mice (Fig. 2D).

Figure 2. TBET⁺CD11c⁺ cells delineate a B_MEM cell subset and accumulate in Il21tg mice.

(A–B) GC B and B_MEM cells were analyzed for TBET and CD11c expression by FACS. GC B and B_MEM cell gating strategies are in Supplemental Fig. 1G. All panels are representative of 3 independent experiments with ≥ 3 mice per strain. (A) TBET staining on GC B cells from C57BL/6 (B6, n=14) or BALB/c (n=23) mice with frequency enumeration. (B) TBET and CD11c staining on B_MEM cells from B6 mice. (C) TBET and CD11c staining on splenic B-2 cells from WT and Il21tg mice. (D) Serum IgG₁ or IgG_2a/c (IgG_2a + IgG_2c) levels in WT and Il21tg mice were determined by ELISA. Values are means ± S.E.M. from 5 WT and 7 Il21tg mice.

Together, our in vitro and in vivo observations prompt a model in which the relative availability of IL4, IL21, and IFNγ govern the likelihood of establishing B_MEM cells expressing TBET and CD11c. Further, they suggest that abundant IFNγ will drive a TBET⁺CD11c⁻ phenotype regardless of IL4 or IL21 levels, but that in the absence of IFNγ, the TBET⁺CD11c⁺ phenotype is reciprocally regulated by IL21 versus IL4. We therefore evaluated these predictions by tracking the immune responses to either influenza virus or Heligmosomoides polygyrus (HP) in mice where cytokine availability could be experimentally manipulated.

Influenza virus infection generates TBET⁺CD11c⁺ B_MEM in the absence of both IL4 and IFNγ

Influenza virus infection yields a well-characterized T-dependent and T_H1-skewed response, in which responding T_FH cells produce copious IFNγ as well as IL21 and IL4 (13). Thus, we reasoned that IFNγ would induce TBET expression in GC B and B_MEM cells, but in the absence of IFNγ, IL4 would prevent TBET expression. Accordingly, WT or Ifng^−/− mice were infected with the A/Puerto Rico/8/1934 H1N1 influenza virus strain (PR8). As expected, WT animals mounted a robust GC B cell response to PR8 (Fig. 3A), and these GC B cells expressed TBET (Fig. 3B, sort strategy & Tbx21 expression Supplemental Figures 1J & 1K). In contrast, GC B cells in Ifng^−/− mice failed to express TBET even though the magnitude of the GC B cell response was similar to WT. Assuming that T_FH cells are the major source of cytokine, we confirmed that both WT and Ifng^−/− mice made substantial numbers of T_FH cells (Supplemental Figures 1J & 1L) and their capacity to make IL4 and IL21 was unperturbed (Figures 3C & 3D). These results are consistent with the idea that in the absence of IFNγ, IL4 blocks TBET expression in response to IL21. To directly test this, we infected Il4^−/−Ifng^−/− double deficient mice with PR8. While Il4^−/−Ifng^−/− mice mounted a blunted GC B cell response (Fig. 3A), these cells nonetheless express T-BET (Fig. 3B & Supplemental Fig. 1K).

Figure 3. Influenza virus infection drives TBET⁺CD11c⁺ B_MEM cell formation in the absence of both IFNγ and IL4.

Splenocytes were harvested from non-infected (−) or day 10 post i.n. 30 TCID50 PR8 infection (+) WT (n=21, black bars), Ifng^−/− (n=10, white bars), or Il4^−/−Ifng^−/− (n=13, gray bars) mice across 3–7 experiments with ≥3 mice per group. GC B, B_MEM, and T_FH cell gating strategies are in Supplemental Figures 1G & 1J. (A) Enumeration of GC B cells. (B) TBET staining on GC B cells. (C) Il4 and (D) Il21 mRNA levels from sorted naïve CD62L⁺ CD4 T (T_N, n=9) or T_FH cells. (E) Proportions and (F) numbers of TBET⁺CD11c⁺ B_MEM cells.

Although the splenic Plasma Cell (PC) numbers were reduced in Ifng^−/− mice, B_MEM cell numbers remained intact across genotypes (Supplemental Figures 1M & 1N). However, the composition of the B_MEM cell pool differed according to genotype (Figures 3E & 3F). While WT mice generated some TBET⁺CD11c⁺ B_MEM cells, Ifng^−/− mice produced few if any above non-infected controls, likely reflecting IL4’s dominance in the absence of IFNγ. Consistent with this interpretation, Il4^−/−Ifng^−/− mice generated the most TBET⁺CD11c⁺ B_MEM cells. Lastly, CD11c expression was restricted to B_MEM cells and not GC B cells (Supplemental Fig. 1O). Overall, these findings confirm and extend our in vitro findings, since the same interplay of cytokines directs TBET expression among B effectors in vivo. Further, our observations suggest that TBET⁺CD11c⁺ B_MEM cells will be fostered in immune responses where IL4 is limited.

Il4 deficiency yields TBET⁺CD11c⁺ B_MEM independent of IFNγ in Heligmosomoides polygyrus infection

Results with influenza virus infection are consistent with the notion that IFNγ drives TBET expression irrespective of concomitant IL4 or IL21, and that eliminating IFNγ creates a situation where the relative levels of IL4 and IL21 govern the TBET⁺CD11c⁺ phenotype. However, this subtractive approach does not necessarily show that in responses where IFNγ is normally absent, the sole determinant of TBET expression is IL4 availability. Accordingly, we asked whether IL4 deficiency is sufficient to permit TBET expression in GC B and B_MEM cells during a T_H2 response, using HP. This intestinal helminth induces IL4 and IL21 production by T_FH cells, which drives a robust IgG₁ response (14). Thus, we hypothesized that in the absence of IL4, IL21 would be sufficient to induce TBET expression in GC B and B_MEM cells. To test this idea, we infected WT or Il4^−/− mice with HP, and probed GC B cells for TBET. As expected, WT mice mounted a GC B cell response that lacked TBET expression, which correlated with increased serum IgG₁ titers. Conversely, although blunted in magnitude, Il4^−/− mice initiated a TBET⁺ GC B cell response with decreased serum IgG₁ titers compared to WT (Figures 4A–C, Supplemental Figures 1J & 1P). To eliminate the possibility that excess IFNγ in Il4^−/− mice explains these phenotypes, we infected Il4^−/−Ifng^−/− mice with HP. The GC B cell response in Il4^−/−Ifng^−/− mice was similar to WT levels (Fig. 4A) but maintained TBET expression independently of IFNγ (Fig. 4B & Supplemental Figures 1J & 1P). Isotype representation varied with TBET expression: whereas WT mice produced >95% IgG₁, over half of the serum antibodies in Il4^−/−Ifng^−/− and Il4^−/− mice were IgG_2b and IgG_2c (Fig. 4C). Further, while Il4^−/−Ifng^−/− mice mounted a higher T_FH cell response (Supplemental Figures 1Q), both Il4^−/− and Il4^−/−Ifng^−/− mice produced less IL21 (Fig. 4D). Regardless, the magnitude of the PC and B_MEM cell response remained intact across genotypes (Supplemental Figures 1R & 1S). However, we again observed alterations in the B_MEM pool according to cytokine availability. Whereas HP-infected WT mice did not generate TBET⁺CD11c⁺ B_MEM cells, both Il4^−/− and Il4^−/−Ifng^−/− mice did—again suggesting IL21 drives a unique TBET⁺ phenotype (Figures 4E & 4F). While prior reports showed CD11c mRNA in GC B cells defined by CD95 and PNA (24), we observed CD11c protein expression only in B_MEM cells (Supplemental Fig. 1T). This seeming disparity may indicate that CD11c transcripts in GC B cells go untranslated, as well as the further resolution of GC and B_MEM by CD38 in our gating strategy. Overall, the HP infection data support our model, inasmuch as in the absence of IFNγ we observe both TBET and CD11c expression that is modulated by IL4. Further, the consistent relationships observed in both types of infection argue that this is a feature common to most humoral immune responses.

Figure 4. Activated B cells express T-BET independent of IFNγ in IL4 limiting conditions.

Splenocytes and sera were harvested from non-infected (−) or day 14 post oral gavage (+) of 200 H. polygyrus in WT (n=20, black bars), Il4^−/− (n=24, white bars), or Il4^−/−Ifng^−/− (n=11, gray bars) mice across 3–6 experiments with ≥3 mice per group. GC B, B_MEM, and T_FH cell gating strategies are in Supplemental Figures 1G & 1J. (A) Enumeration of GC B cells. (B) TBET staining on GC B cells. (C) Serum concentrations of IgG₁ and IgG_2c + IgG_2b. (D) Il21 mRNA levels from sorted T_FH cells. (E) Proportions and (F) numbers of TBET⁺CD11c⁺ B_MEM cells.

In toto, our findings reveal a novel cytokine network that governs TBET expression in the context of TLR stimulation. In the absence of IFNγ, IL4 and IL21 reciprocally regulate TBET and CD11c expression both in vitro and in vivo. Since immune responses are rarely monolithic with regard to these three cytokines (13, 25), distinct or multifunctional T_FH cells likely generate a diverse set of B effectors. Consequently, altering the cytokine milieu affects the isotypes generated (Fig. 4C) and the composition of the B_MEM pools (Figures 3F & 4F) while maintaining the magnitude of the response.

It is tempting to speculate that the TBET⁺CD11c⁺ B cells reported in autoimmunity, viral infections, and aging share a common underlying origin involving TLR engagement coupled with either copious IFNγ or abundant IL21 with little IL4. Indeed, both TLR7 and IL21 deficiencies ameliorate disease in humoral autoimmunity models (26, 27), and poor IL4 production has been observed in T_FH cells from aged mice (28). Thus, understanding this interplay among IL4, IL21, and IFNγ might better define the etiology of humoral autoimmune syndromes where such cells are implicated (7, 12, 29). Lastly, while it is clear that IFNγ and IL21 differentially induce CD11c expression (Fig. 1C), the functional consequences of expressing this integrin remain elusive. Importantly, the restriction of CD11c expression to B_MEM cells is consistent with prior B_MEM subsetting studies in human tonsils and may thus define a tissue-homing population (30). Accordingly, further studies are needed to assess the role of these different TBET⁺ B_MEM cells in both health and disease.