TLR4 regulates IFN-γ and IL-17 production by both thymic and induced Foxp3⁺ T_regs during intestinal inflammation

TLR4 regulates inflammatory cytokine production from thymic- and peripherally-derived T_reg cells during intestinal inflammation.

Abstract

T_regs play a crucial role in the maintenance of intestinal immune homeostasis. However, significant numbers of Foxp3⁺ T_regs accumulate in the inflamed lesions in experimental colitis and in IBD patients. T_reg production of the proinflammatory cytokines IFN-γ and/or IL-17 may arguably explain their ineffectiveness in suppressing intestinal inflammation. However, it remains unknown whether iT_reg and tT_reg produce proinflammatory cytokines and how TLR signaling regulates this process. Here, we found that Foxp3⁺T_regs were increased in the intestines of B6.TLR4^−/− and B6.IL-10^−/− mice when compared with WT B6 mice. TLR4^−/− and IL-10^−/− resulted in more T_regs within inflamed intestines. The majority of Foxp3⁺ T_regs in the spleen was Helios⁺Nrp1⁺, whereas most Foxp3⁺ T_regs in the intestinal LP were Helios⁻Nrp1⁻. More Helios⁺Nrp1⁺ T_regs expressed IFN-γ and/or IL-17 than did Helios⁻Nrp1⁻ T_regs in the spleen and intestine, which was increased with TLR4^−/−. TLR4 signaling in T cells and APCs inhibited Foxp3⁺ induction via MyD88-dependent, TRIF-independent pathways, which was negatively regulated by SOCS3. Collectively, these data demonstrate Helios⁺Nrp1⁺ tT_regs and Helios⁻Nrp1⁻ iT_regs produce proinflammatory cytokines in the intestines during inflammation, which was regulated by TLR4 signaling.

Introduction

The gastrointestinal tract provides a major entry for dietary and microbial antigens into the body. Among the substantive microbial population, both beneficial commensal microbiota and potential pathogens reside in the gut. A key mechanism of regulation in the intestinal tract lies with T_regs, capable of surveying a wide array of immune responses to maintain intestinal homeostasis and protect the intestine from inflammation. The majority of T_regs is defined by expression of the transcription factor Foxp3, which is critical for T_reg development and stability [1]. The catastrophic inflammation observed in the absence of Foxp3 [1, 2] demonstrates that T_regs are vital for immune regulation and restraint of the inflammatory effects of effector T cells. Foxp3⁺ T_regs arise in two separate origins, differing in location and TCR repertoire [3]. CD4⁺ T cells with TCR affinities toward self-antigens become Foxp3⁺ T_regs in the thymus and are denoted tT_regs, whereas Foxp3⁺ T_regs can also be induced in the periphery after antigen presentation and are denoted iT_regs [3]. T_regs^−/− results in chronic intestinal inflammation, and colitis can be prevented and cured by transfer of T_regs, showcasing a key role for T_regs in the maintenance of intestine homeostasis [4]. Paradoxically, however, Foxp3⁺ T_regs are increased in inflamed intestinal tissues in animal models and patients with IBD [5–7], and most of them express inflammatory cytokines, such as IFN-γ and IL-17 [5, 8, 9]. Anti-TNF-α treatment suppresses inflammation in a majority of patients with IBD with a concomitant decrease in local mucosal T_regs and increase in peripheral T_regs [10, 11], indicating that inflammation drives T_reg expansion and subsequently promotes accumulation in inflamed lesions. T_reg expression of a number of factors has been shown to contribute to T_reg function and stability [12–15]. However, how TLR signaling regulates T_reg expression of proinflammatory cytokines in inflamed intestine is still not understood completely.

Although a majority of T_regs continue to express high levels of Foxp3 after transfer into noninflamed hosts, many reports demonstrate the plasticity of the T_reg differentiation and maintenance program [16, 17]. Foxp3⁺ T_regs can fully convert into Th1, Th17, and/or follicular Th cells in the intestine, especially during inflammation [5, 18]. This is manifested by the loss of Foxp3 and acquisition of T-bet (Th1), retinoic acid receptor-related orphan receptor-γt (Th17), or B cell lymphoma 6 (follicular Th) by T_regs [18, 19]. However, it is unclear whether tT_reg or iT_regs have different propensities for acquiring proinflammatory function and how microbial exposure regulates this process. The interaction of the commensal microbiota with TLRs plays an important role in innate immune responses, e.g., the activation of TLRs on DCs and macrophages results in production of inflammatory cytokines and chemokines. Conversely, the intestinal microbiota can promote immune regulation through the activation of TLRs on CD4⁺ T cells, independent of APCs [20, 21]. TLR4^−/− or TLR9^−/− in vivo, coupled with IL-10^−/−, result in spontaneous colitis that is significantly worse than that seen in IL-10^−/− mice, demonstrating that TLR4 and TLR9 have regulatory roles in CD4⁺ T cells [22, 23]. Notably, T_regs in TLR4^−/−IL-10^−/− mice express very high levels of inflammatory cytokines, including GM-CSF, IFN-γ, and IL-17, and have decreased suppressive capability [23]. Activation of MyD88, a downstream signaling molecule of most TLRs, has also been shown to be important for T_regs to control intestinal inflammation in vivo [24]. In this report, we demonstrate that LPS influences the generation and expansion of T_regs through TLR4 signaling on APCs and CD4⁺ T cells in a MyD88-dependent but TRIF-independent manner. Helios⁺Nrp1⁺ tT_regs and Helios⁻Nrp1⁻ iT_regs produce proinflammatory cytokines in the intestines. TLR4 signaling influences T_regs expressing proinflammatory cytokines; however, it does not appear to favor either T_reg subset. Furthermore, T_reg expansion under intestinal inflammation occurs in the intestine but not in the periphery.

MATERIALS AND METHODS

Mice

B6, B6.TRIF^LPS2, and B6.RAG1^−/− mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA). CBir1 Tg mice, B6.MyD88^−/−, B6.TLR4^−/−, B6.IL-10^−/−, B6.TLR4^−/−IL-10^−/− [23], and LysMCre-SOCS3^fl/fl mice [25] were maintained in the animal facilities at UTMB, The University of Chicago, and The University of Alabama at Birmingham. Age-matched and cohoused mice of 8–12 weeks were used in these experiments. All experiments were reviewed and approved by the Institutional Animal Care and Use Committees of UTMB, The University of Chicago, and The University of Alabama at Birmingham.

Antibodies and reagents

Fluorochrome-conjugated anti-mouse CD4 (RM4-5), Helios (22F6), IL-17A (TC11-18H10), and IFN-γ (XMG1.2) antibodies and rTGF-β were purchased from BioLegend (San Diego, CA, USA). Anti-mouse Nrp1 was purchased from R&D Systems (Minneapolis, MN, USA). Anti-mouse Foxp3 (FJK-16s) was purchased from eBioscience (San Diego, CA, USA). Live/Dead Fixable Dead Cell stain kits were used to gate on live lymphocytes (Life Technologies, Carlsbad, CA, USA). LPS (0111:B4) was purchased from Sigma-Aldrich (St. Louis, MO, USA). FTY720 was purchased from Cayman Chemical (Ann Arbor, MI, USA).

Isolation of LP cells

As described previously [5], large intestines were removed, sliced, and digested by collagenase IV. The cells were resuspended in 40% Percoll and overlaid carefully onto 70% Percoll. The interface containing the LP lymphocytes was collected. Lymphocytes were restimulated with PMA (50 ng/ml), ionomycin (750 ng/ml), and monensin for 5 h before labeling with antibodies for flow cytometry.

CD4⁺ T cell purification, in vitro polarization of T_regs, and BMDC generation

CD4⁺ T cells were isolated by using anti-mouse CD4 magnetic beads (BD Biosciences, San Diego, CA, USA). Naive CD62L⁺CD4⁺ T cells were sorted by FACSAria (BD Biosciences), based on CD62L and CD4 coexpression. CD4⁺ T cells from B6, MyD88^−/−, or TRIF^LPS2 mice were cultured with plate-bound αCD3 (1 μg/ml) and αCD28 under T_reg-polarizing conditions (5 ng/ml TGF-β) or in the absence of polarizing cytokines. Irradiated splenocytes from B6, MyD88^−/−, or TRIF^−/− mice were cultured with CD4⁺ T cells from B6 mice with soluble αCD3 (1 μg/ml), with or without T_reg polarization, or with CD4⁺ T cells from CBir1 Tg mice with CBir1 flagellin peptide under T_reg polarization.

BMDCs were generated from B6 or SOCS3 conditional knockout femurs. BM cells were cultured with GM-CSF (BioLegend) for 8 days and cultured with CD4⁺ T cells from CBir1 Tg mice with CBir1 flagellin peptide under T_reg polarization.

Statistical analysis

Levels of significance were determined by Student's t-test in Prism 5.0 (GraphPad Software, La Jolla, CA, USA). P < 0.05 was considered to be statistically significant.

RESULTS

TLR4 regulates expansion and inflammatory cytokine expression by Foxp3⁺ T_regs in the intestines of colitic mice

Our previous studies demonstrated that during intestinal inflammation, Foxp3⁺ T_regs accumulated in the inflamed lesions, and substantial numbers of such T_regs expressed the proinflammatory cytokines IFN-γ and IL-17 in the inflamed colon [5, 23]. However, the factors responsible for driving T_reg expansion and expression of these proinflammatory cytokines and whether tT_regs or iT_regs are capable of producing these cytokines remain unclear, as recent reports have offered conflicting results [17, 26]. We have shown previously that IL-10^−/− and TLR4^−/− resulted in severely aggravated intestinal inflammation, thereby suggesting a critical role of TLR signaling in controlling proinflammatory activities [23]. To assess how TLR4 regulates tT_reg and iT_reg expression of proinflammatory cytokines in the presence and absence of inflammation, we analyzed Foxp3⁺ T_reg expression of proinflammatory cytokines from WT B6, TLR4^−/−, IL-10^−/−, and TLR4^−/−IL-10^−/− mice. We have reported previously that TLR4^−/−IL-10^−/− mice developed more severe colitis compared with IL-10^−/− mice, and there was no inflammation in the intestines of WT and TLR4^−/− mice [23]. Examination of cytokine production by CD4⁺ T cells showed that IFN-γ-producing Th1 cells and IL-17-producing Th17 cells were increased in spleen of colitic IL-10^−/− mice compared with that of WT mice and were amplified further in the intestinal LP (Fig. 1). TLR4^−/− alone did not affect CD4 T cell cytokine production in WT mice. There was, however, a considerable increase in the number of cells that produced IFN-γ and IL-17 in colitic IL-10^−/− mice, which was increased further in TLR4^−/−IL-10^−/− mice.

Figure 1. The absence of TLR4 increases CD4 cell inflammatory cytokine production.

(A) Total spleen CD4 cells from TLR4^−/−, IL-10^−/−, TLR4^−/−IL-10^−/− [double-knockout (dKO)], and WT mice were analyzed for IFN-γ and IL-17 production. (B) IFN-γ and IL-17 expression from total CD4⁺ T cells in the spleen and intestine. Bar charts reflect mean ± sem. *P < 0.05, **P < 0.01, and ***P < 0.001 compared with WT mice. Data are representative of three or more experiments of two or more mice/group with similar results. FACS plots are gated upon live CD4⁺ cells.

Interestingly, Foxp3⁺ T_regs were increased in intestinal LP but not in the spleen of colitic IL-10^−/− mice compared with WT mice (Fig. 2). However, the presence of Foxp3⁺ T_regs was increased in spleen and LP of TLR4^−/− mice compared with that of WT mice, which was increased further in TLR4^−/−IL-10^−/− mice. These data are consistent with the hypothesis that TLR4 signaling inhibits T_reg expansion. In addition to the increase of Foxp3⁺ T_regs, there was an increase in Foxp3⁺ T_regs that produce IL-17, but not IFN-γ, in the intestines of TLR4^−/− mice. Interestingly, there were more Foxp3⁺ T_regs expressing IFN-γ and IL-17 in the intestines of IL-10^−/− mice, and TLR4 appeared to play a role in controlling Foxp3⁺ T_reg cytokine production, as the levels of Foxp3⁺ T_regs expressing IFN-γ and IL-17 are increased further in TLR4^−/−IL-10^−/− mice compared with those in IL-10^−/− mice (Fig. 2).

Figure 2. The absence of TLR4 increases intestinal T_reg proinflammatory cytokine production.

(A) CD4 T cells were isolated from the spleens and intestines of TLR4^−/−, IL-10^−/−, TLR4^−/−IL-10^−/−, and WT mice and analyzed for IFN-γ and IL-17 production by flow cytometry gated on Foxp3⁺ cells. (B) IFN-γ and IL-17 expression from total CD4⁺Foxp3⁺ T cells in the spleen and intestine. Bar charts of Foxp3⁺IFN-γ⁺, Foxp3⁺IL-17⁺, and Foxp3⁺IFN-γ⁺IL-17⁺ are calculated based on total CD4⁺Foxp3⁺ T cells and reflect mean ± sem. *P < 0.05, **P < 0.01, and ***P < 0.001 compared with WT B6 mice. Data are representative of three or more experiments of two or more mice/group with similar results. FACS plots are gated upon live CD4⁺ cells.

Helios⁺Nrp1⁺ and Helios⁻Nrp1⁻ Foxp3⁺ T_regs produce IFN-γ and IL-17 in the intestines of colitic mice

Although it has been shown that T_regs can convert into proinflammatory cytokine-producing T cells [5], it is unclear whether tT_reg and iT_reg or only iT_regs produce proinflammatory cytokines, as tT_regs are thought to be relatively stable [26]. As Helios and Nrp1 have been reported to be expressed predominantly, although not absolutely, by tT_regs [27, 28], we chose the combined expression of Helios and Nrp1, which is the best identifier for tT_regs to investigate the tT_regs that produce proinflammatory cytokines. Consequently, the absence of Helios and Nrp1 represents peripherally iT_regs [29]. We assessed the difference in proinflammatory cytokine production from Foxp3⁺ T_regs between Helios⁺Nrp1⁺ tT_regs and Helios⁻Nrp1⁻ iT_regs in the context of TLR4^−/− and inflammation. We found that Helios⁺Nrp1⁺ and Helios⁻Nrp1⁻ T_regs produced IL-17 and IFN-γ in the spleen and LP under steady-state conditions (Figs. 3–6). Analysis of intestinal T_regs in WT mice revealed a higher presence of Helios⁻Nrp1⁻ T_regs versus Helios⁺Nrp1⁺ T_regs than was found in the spleen. Although TLR4^−/−IL-10^−/− mice had higher frequencies of T_regs in the intestines, the ratio of Helios⁻Nrp1⁻ T_regs to Helios⁺Nrp1⁺ T_regs was conserved between strains (Figs. 3A, 4B, 5A, and 6B). In WT mice, Helios⁺Nrp1⁺ T_regs had a higher propensity to produce IFN-γ or IL-17 in the intestines, as opposed to the spleen, whereas Helios⁺Nrp1⁺ T_regs only produced IFN-γ with little IL-17. Helios⁺Nrp1⁺ T_regs produced higher levels of IFN-γ compared with that of Helios⁻Nrp1⁻ T_regs in the spleen and LP, and LP Helios⁺Nrp1⁺ T_regs produced more IL-17 than did Helios⁻Nrp1⁻ T_regs (Fig. 5B), whereas Helios⁻Nrp1⁻ T_regs and Helios⁺Nrp1⁺ T_regs produced IL-17 at similar levels in the spleen (Fig. 3B). In IL-10^−/− mice, IFN-γ and IL-17 production by Helios⁻Nrp1⁻ T_regs and Helios⁺Nrp1⁺ T_regs in the spleens and LP was increased compared with that in WT mice; however, the pattern remained the same; i.e. Helios⁺Nrp1⁺ T_regs produced a higher level of IFN-γ compared with Helios⁻Nrp1⁻ T_regs, and Helios⁻Nrp1⁻ T_regs and Helios⁺Nrp1⁺ T_regs produced IL-17 at similar levels (Figs. 3–6). Although IL-17 and IFN-γ production by Helios⁻Nrp1⁻ T_regs and Helios⁺Nrp1⁺ T_regs was not different from WT in TLR4^−/− mice, TLR4^−/− in IL-10^−/− mice significantly increased production of IL-17 and IFN-γ by Helios⁻Nrp1⁻ T_regs and Helios⁺Nrp1⁺ T_regs. Notably, the Foxp3⁺ T_regs expressing IFN-γ and IL-17 were predominantly Helios⁺Nrp1⁺ T_regs in the TLR4^−/−IL-10^−/− mice. Collectively, these data demonstrated that whereas the absence of TLR4 signaling does not affect T_reg proinflammatory cytokine production alone, it augments T_reg proinflammatory cytokine production during intestinal inflammation in the absence of IL-10.

Figure 3. Helios⁺Nrp1⁺ T_reg and Helios⁻Nrp1⁻ T_regs differentially express IFN-γ in the spleen.

(A) T_regs (CD4⁺Foxp3⁺) were separated into Helios⁺Nrp1⁺ and Helios⁻Nrp1⁻ T_regs, and analyzed for IFN-γ and IL-17 expression from the spleen of TLR4^−/−, IL-10^−/−, TLR4^−/−IL-10^−/−, and WT mice. (B) Helios⁺Nrp1⁺ T_regs [double-positive (DP) tT_reg] and Helios⁻Nrp1⁻ T_regs [double-negative (DN) iT_reg] from the spleen of TLR4^−/− IL-10^−/− and WT mice were analyzed for IFN-γ and IL-17 production. Bar charts are calculated based on total double-positive tT_reg or double-negative iT_reg and reflect mean ± sem. *P < 0.05, and **P < 0.01. Data are representative of three or more experiments of two or more mice/group. FACS plots are gated upon live CD4⁺Foxp3⁺ cells.

Figure 4. T_regs are increased systemically in the absence of TLR4 and IL-10.

(A) Spleen CD4 cells from TLR4^−/−, IL-10^−/−, TLR4^−/−IL-10^−/−, and WT mice were counted for Foxp3 expression and IFN-γ and IL-17 in T_regs. (B) T_regs (CD4⁺Foxp3⁺) were analyzed for Helios and Nrp expression. DP tT_reg, Helios⁺Nrp1⁺ T_regs; DN iT_reg, Helios⁻Nrp1⁻ T_regs. *P < 0.05, **P < 0.01, and ***P < 0.001 compared with WT B6 mice. Data are representative of three or more experiments of two or more mice/group.

Figure 5. Intestinal Helios⁺ Nrp1⁺ T_regs and Helios⁻Nrp1⁻ T_regs differentially produce IFN-γ and IL-17.

(A) T_reg (CD4⁺Foxp3⁺) cells were separated into Helios⁺Nrp1⁺ and Helios⁻Nrp1⁻ T_regs and analyzed for IFN-γ and IL-17 expression from the intestines of TLR4^−/−, IL-10^−/−, TLR4^−/−IL-10^−/−, and WT mice. (B) Helios⁺Nrp1⁺ T_regs (DP tT_reg) and Helios⁻Nrp1⁻ T_regs (DN iT_reg) from the intestines of TLR4^−/−IL-10^−/− and WT mice were analyzed for IFN-γ and IL-17 production. Bar charts are calculated based on total double-positive tT_reg or double-negative iT_reg and reflect mean ± sem. *P < 0.05, **P < 0.01, and ***P < 0.001. Data are representative of three or more experiments of two or more mice/group. FACS plots are gated upon live CD4⁺Foxp3⁺ cells.

Figure 6. Intestinal T_regs are increased in the absence of TLR4 and IL-10.

(A) Intestinal CD4 cells from TLR4^−/−, IL-10^−/−, TLR4^−/−IL-10^−/− (dKO), and WT mice were counted for Foxp3 expression and IFNγ and IL-17 in T_regs. (B) T_regs (CD4⁺Foxp3⁺) were analyzed for Helios and Nrp expression.*P < 0.05, and **P < 0.01 compared with WT B6 mice. Data are representative of three or more experiments of two or more mice/group.

Foxp3⁺ T_regs expand in the intestine during intestinal inflammation

We then investigated whether T_reg expansion occurred systemically or within the tissue at the interface of microbial exposure and interaction. We used the CBir1 T cell transfer model to examine the location of T_reg expansion. The transfer of CD4⁺ T cells from CBir1 Tg mice, which are specific for an immunodominant commensal bacterial antigen CBir1 flagellin into immunodeficient RAG^−/− mice, induces colitis in the recipients [5, 30]. CBir1 CD4 T cells were i.v. transferred into recipient RAG^−/− mice and treated orally with FTY720 on the same day of cell transfer and every other day thereafter for 4 weeks, which inhibits lymphocyte egress from the tissues [31]. In FTY720-treated mice, CD4⁺ T cells activated in the intestines are expected to remain within the intestinal compartment. Consistent with our previous observations [5], both FTY720-treated and control CBir1 T cell-reconstituted RAG^−/− mice developed colitis (Fig. 7A and B). Analysis of the mice receiving CBir1 CD4⁺ T cells revealed substantial numbers of Th1 cells, Th17 cells (data not shown), as well as Foxp3⁺ T_regs in the spleen and LP (Fig. 7C). The intestine contained the highest percentage of Foxp3⁺ T_regs producing IFN-γ, IL-17, or both IFN-γ and IL-17. Treatment with FTY720 greatly decreased the Foxp3⁺ T_regs in spleen but not in the mesenteric lymph node and intestine. Notably, the proportion of IFN-γ⁺Foxp3⁺ T_regs was increased greatly in the intestine after FTY720 treatment. IL-17⁺Foxp3⁺ T cells also increased to a lesser extent. Collectively, these data demonstrate that the expansion and subsequent production of proinflammatory cytokines by Foxp3⁺ T_regs occur primarily in the inflamed intestine with eventual systemic circulation.

Figure 7. Expansion and acquisition of proinflammatory cytokine production by T_regs occur in the intestine.

CBir1 CD4 T cells (0.25×10⁶) were i.v. transferred into RAG2^−/− mice. Recipient mice were gavaged with PBS or FTY720 every 2 days. (A) Four weeks post-T cell transfer, the histopathology of intestines was determined. Images are representative of two experiments of four mice/group with similar results. (B) Pathological scores are shown. (C) T_regs were isolated from the spleens and intestinal LP of CBir1 CD4 T cell recipient mice and analyzed for Foxp3, IFN-γ, and IL-17 production by flow cytometry. FACS plots are gated on total live CD4⁺ T cells. Data are representative of two experiments of four mice/group with similar results.

TLR4 signaling down-regulates development of Foxp3⁺ T_regs and T_reg expression of IFN-γ in a MyD88-dependent and TRIF-independent manner

We next investigated how TLR4 signaling regulates Foxp3⁺ T_regs. As TLR4^−/− mice had a higher frequency of T_regs in the intestine, where CD4⁺ T cells would have the most contact with bacteria and TLR ligands, we questioned whether TLR4 signaling inhibited induction of Foxp3⁺ T_regs and whether TLR4 signaling on CD4⁺ T cells or APCs was responsible for down-regulation of Foxp3⁺ T_reg induction. We first cultured naive CD62L⁺CD4⁺ T cells from CBir1 Tg mice with splenic APCs under T_reg conditions with TGF-β in the presence or absence of a series of doses of LPS. As shown in Fig. 8A, LPS inhibited development of Foxp3⁺ T_regs in a dose-dependent manner under T_reg-polarizing conditions and had no discernible effect on Foxp3 expression on unpolarized T cells in the absence of TGF-β (Fig. 8B). Treatment with LPS led to an increase in IFN-γ expression in Foxp3⁻ effector T cells, both in the presence and absence of TGF-β, but inhibited Foxp3⁺IFN-γ⁻ and Foxp3⁺IFN-γ⁺ T_regs.

Figure 8. LPS inhibits Foxp3 T_regs.

(A) CD62L⁺CD4⁺ CBir1 CD4 T cells were cocultured with irradiated splenocytes with CBir1 peptide and 5 ng/ml TGF-β to induce Foxp3 induction. Cells were also treated with or without varying concentrations of LPS and analyzed by flow cytometry after 5 days. (B) CD62L⁺CD4⁺ CBir1 CD4 T cells were cocultured with irradiated splenocytes (APC) with CBir1 peptide in the absence of polarization (Th0) or with 5 ng/ml TGF-β (T_reg). Cells were also treated with or without LPS (1000 ng) and analyzed by flow cytometry after 5 days. (C) CD62L⁺CD4⁺ T cells were cultured with plate-bound αCD3 (1 μg) and αCD28 (2 μg/ml) in the absence of polarization (Th0) or with 5 ng/ml TGF-β (T_reg). Cells were also treated with or without LPS (1000 ng) and analyzed by flow cytometry after 5 days. (D) Bar graph reflects total numbers of Foxp3⁺ cells generated in culture of C and represents mean ± sem. *P < 0.05, n = 4. FACS plots are gated upon live CD4⁺ cells.

As CD4⁺ T cells express TLR4 and respond to its ligand LPS [22], we cultured naive CD62L⁺CD4⁺ T cells with plate-bound anti-CD3 and anti-CD28 with TGF-β in the presence or absence of LPS to determine the role of LPS-TLR4 signaling in T cells on T_reg differentiation. As shown in Fig. 8C, LPS-TLR4 signaling in T cells inhibited T_reg development. Although LPS can promote the overall proliferation of CD4⁺ T cells [24], we observed that the total number of Foxp3⁺ T_regs was decreased after LPS treatment (Fig. 8D), thus excluding the possibility that hyperproliferation of non-T_reg Foxp3⁻ cells was diluting the frequency of Foxp3⁺ T_regs. It has been reported previously that TLR4 signaling on Th1 cells can decrease IFN-γ production [22]. Whereas we observed little difference in IFN-γ expression in Foxp3⁻ cells, we found that direct treatment with LPS strongly decreased IFN-γ production in T_regs (Fig. 8C).

TLR4 signals through two separate pathways: the MyD88 pathway and the TRIF pathway [32]. To determine the mechanism behind TLR4 inhibition of Foxp3⁺ T_regs, we used MyD88^−/− or TRIF^−/− CD4⁺ T cells to identify which pathway was responsible for the effects directly in T cells. After treatment with LPS and TGF-β, MyD88^−/− naive CD4⁺ T cells exhibited no discernible change in induction of Foxp3⁺ T_regs, whereas LPS down-regulated Foxp3⁺ T_reg induction in TRIF^−/− naive CD4⁺ T cells (Fig. 9A), revealing that CD4⁺ T cell TLR4 inhibition of Foxp3⁺ T_regs depends on MyD88 but not TRIF.

Figure 9. LPS partially inhibits T_reg generation through MyD88 signaling on CD4 cells and APCs.

(A) MyD88^−/−, TRIF^−/−, or WT CD4 cells were cultured with plate-bound αCD3 (1 μg) and αCD28 (2 μg/ml) with 5 ng/ml TGF-β (T_reg) in the presence or absence of LPS (1 μg) and analyzed by flow cytometry after 5 days. (B) MyD88^−/−, TRIF^−/−, or WT irradiated splenocytes (APC) were cultured with WT CD4 cells with plate-bound αCD3 (1 μg) and 5 ng/ml TGF-β in the presence or absence of LPS (1000 ng) and analyzed by flow cytometry after 5 days. FACS plots are gated upon live CD4⁺ cells.

As LPS signaling on APCs drastically inhibited Foxp3⁺ T_regs at a much greater level compared with TLR4 signaling solely on CD4⁺ T cells (Fig. 8), we then examined the role of the MyD88 and TRIF pathways in APCs by coculturing WT CD4⁺ T cells with MyD88^−/− or TRIF^−/− APCs under T_reg conditions in the presence or absence of LPS. As addition of LPS in the culture of naive CD4⁺ T cells with WT APCs under T_reg conditions almost completely inhibited induction of Foxp3⁺ T_regs, addition of LPS did not result in a decrease of induction of Foxp3⁺ T_regs in the absence of APC MyD88 signaling (Fig. 9B). In contrast, when WT naive CD4⁺ T cells were cultured with TRIF^−/− APCs, LPS inhibited induction of Foxp3⁺ T_regs at a similar level to that with WT APCs. Collectively, these data demonstrated that LPS modulates Foxp3⁺ T_regs in a MyD88-dependent, TRIF-independent manner. This effect is mediated directly on CD4⁺ T cells, as well as indirectly through APCs.

SOCS3 mediates LPS-TLR4 inhibition of Foxp3⁺ T_reg development

LPS-TLR4 signaling stimulates SOCS3 expression and STAT3 activation, and SOCS3 serves as a key negative regulator of the LPS-TLR4 pathway through inhibition of STAT3 [33]. Therefore, SOCS3 has a crucial role in modulating STAT3 activation and inflammatory gene expression. We next investigated whether SOCS3 regulates LPS inhibition of Foxp3⁺ T_regs. We first generated BMDCs from WT and LysMCre-SOCS3^fl/fl mice and stimulated them with LPS. SOCS3^−/− BMDCs expressed higher levels of activation of STAT3 (Fig. 10A) and produced higher levels of IL-1β, IL-6, IL-12, IL-23, and TNF-α compared with levels in WT BMDCs (Fig. 10B). To determine whether SOCS3 regulates LPS-TLR4 inhibition of Foxp3⁺ T_regs, we cultured WT naive CD4⁺ T cells with WT and SOCS3^−/− BMDCs with anti-CD3 under T_reg conditions with TGF-β. To delineate the effect of SOCS3, we selected a low dose of 1 ng/ml LPS that only partially inhibited induction of Foxp3⁺ T_regs. SOCS3^−/− in APCs enhanced LPS down-regulation of Foxp3⁺ T_regs (Fig. 10C and D). Collectively, these data indicated that SOCS3 negatively regulates LPS inhibition of Foxp3⁺ T_reg development.

Figure 10. SOCS3 negatively regulates LPS inhibition of Foxp3⁺ T_regs.

BMDCs from SOCS3^−/− or WT mice were derived with GM-CSF after 8 days. (A) BMDCs were treated with medium or LPS (10 ng/ml) for 4 h. Cell lysates were subjected to immunoblotting with antiphospho-STAT3 (pSTAT3), anti-STAT3, and GAPDH antibodies. (B) Primary BMDCs from WT and SOCS3^−/− mice were treated with medium or LPS (10 ng/ml) for 4 h. mRNA was analyzed for IL-1β, IL-6, IL-12p40, IL-23p19, and TNF-α by quantitative RT-PCR. Representative of three independent experiments. *P > 0.01, and **P > 0.001. (C) BMDCs were cocultured with CBir1 CD4 cells with TGF-β (5 ng/ml) in the presence or absence of LPS (1 ng) and analyzed by flow cytometry after 5 days. (D) Bar graph reflects percentage of Foxp3⁺ cells generated in culture of C and represents mean ± sem. **P < 0.01, n = 6. FACS plots are gated upon live CD4⁺ cells. NT, non-treated. (E) T_regs were sorted from Foxp3-GFP CBir1 CD45.2 CD4 T cells after culture with 5 ng/ml TGF-β in the presence or absence of LPS (1 ng/ml) and WT or SOCS3^−/− BMDCs. FACS-sorted GFP⁺ T_regs were cultured with CFSE-labeled CBir1 CD45.1 CD4 T cells (1:4) and irradiated APCs with 1 μg/ml CBir1 peptide. Proliferation of CD45.1 responder CD4 T cells was measured after 4 days. FACS plots are gated upon live CD45.1⁺ CD4⁺ cells (dashed lines, naive T cell alone; solid lines, with T_regs).

To assess whether the functionality of T_regs is affected after generation under the influence of LPS and SOCS3, we cultured naive CD45.2⁺ Foxp3-GFP CBir1 CD4 T cells with WT and SOCS3^−/− BMDCs under T_reg conditions with TGF-β in the presence or absence of LPS (1 ng/ml). After sorting GFP⁺ T_regs, we assessed their ability to suppress effector T cell proliferation by coculturing with CFSE-labeled splenic CD45.1 CBir1 CD4 T cells and measured effector T cell proliferation via CFSE dilution after 4 days. As illustrated in Fig. 10E, LPS treatment during T_reg induction had no discernible effect on T_reg function. Whereas the frequency of T_reg generation is strongly inhibited by LPS, the T_regs that arise continue to function normally. SOCS3^−/− in APCs during T_reg generation also does not have significant bearing on the function of T_regs after induction of Foxp3.

DISCUSSION

IBD is dependent on the microbiota for providing signals to modulate the immune system and control inflammation. An inability to regulate the intestinal environment properly coupled with constant exposure to microbial ligands induces chronic inflammation. In both patients with IBD and experimental colitis, high levels of Foxp3⁺ T_regs are found to produce proinflammatory cytokines within the inflamed intestine [5, 6, 8], potentially intensifying inflammation or stifling regulation. Recently, it has been shown that microbial ligands modulate CD4⁺ T cell activation. Although CD4⁺ T cells respond in an antigen-specific manner, the presence of TLR ligands can potentiate CD4⁺ T cell responses [34, 35]. TLR4^−/− on CD4⁺ T cells aggravates colitis in IL-10^−/− mice by increasing T_reg and T effector cell production of IFN-γ and IL-17 [22, 23]. As T_regs can arise from the thymus or in the periphery, there is a question of whether there are differences in immunoregulation and cytokine production between T_reg subsets. It has been reported that Helios and Nrp1 are expressed predominantly by tT_regs, although not absolutely. Helios⁺Nrp1⁺Foxp3⁺ is considered to be the best indicator of tT_regs, whereas Helios⁻Nrp1⁻ Foxp3⁺ is the best indicator of iT_regs [27, 28, 36, 37]. We report here that Helios⁺Nrp1⁺ and Helios⁻Nrp1⁻ Foxp3⁺ T_regs can be induced to express IFN-γ and/or IL-17 under homeostatic conditions and inflammatory conditions, a process that is negatively regulated by TLR4 in a MyD88-dependent but TRIF-independent manner.

It has been observed consistently that Foxp3⁺ T_regs accumulate in inflamed tissue of animal models of colitis and IBD patients [6, 38]. Whereas it would appear that the influx of T_regs is a response at controlling extensive inflammation, a substantial number of them produce proinflammatory cytokines. This highlights the challenge of using expanded T_reg populations to treat autoimmune diseases, which might result in the uncontrolled production of proinflammatory cytokines by T_regs [39–41]. Although recent reports demonstrated that encounter with microbiota resulted in the peripheral generation of T_regs, and microbiota antigens play an important role in shaping the colonic T_reg population [42, 43], it is still highly debatable as to whether only iT_regs can convert into proinflammatory cytokine-producing T cells or whether tT_regs can also do so under certain conditions in the intestine. In contrast to some previous reports [44] indicating that tT_regs were stable and resistant to conversion into Th1 or Th17 cells, our current findings revealed that Helios⁺Nrp1⁺Foxp3⁺ T_regs produced IFN-γ and IL-17 at a higher frequency and even in greater numbers than Helios⁻Nrp1⁻ Foxp3⁺ T_regs. These data corroborate previous findings that tT_regs are apt to produce IL-17 after treatment with IL-6 [45], while continuing to express Foxp3. Although severe inflammation was not a prerequisite for T_reg conversion, as it can be seen in the intestine of healthy WT mice, the degree of inflammation did contribute to the acquisition of proinflammatory cytokines in T_regs. Given our previous observations that the acquisition of proinflammatory cytokines can occur in tandem with the loss of Foxp3 expression from “ex-T_regs”; at this point, it remains unclear how the chronic inflammation in TLR4^−/−IL-10^−/− mice affects stability of Foxp3 expression in tT_regs or iT_regs. It has been shown that T_reg expression of a number of factors contributes to T_reg function and stability through improving T_reg-mediated immunoregulation, including OX40 [12, 13], which enhances T_reg responsiveness to IL-2, and TNFRII [14, 15], which enhances T_reg function in the presence of TNF-α. However, it is currently not known completely how TLR4 and its signaling components regulate such conversion. Furthermore, it remains unclear whether microbial antigens and pathogen-associated molecular patterns are able to impair Foxp3 and T_reg stability and promote conversion to other subsets, perhaps as a result of consistent or repeated TLR stimulation.

Interestingly, although we observed an increase in intestinal Foxp3⁺ T_regs in TLR4^−/− mice housed in the animal facilities at The University of Chicago and at UTMB, there was no observable difference in Foxp3⁺ T_reg numbers between the colons of WT and TLR4^−/− mice housed in the animal facilities at Massachusetts General Hospital in our previous studies [23], indicating that different commensal bacteria differentially stimulate gut TLR4 signaling and host immune responses. The acquisition or loss of particular bacterial populations may have altered the propensity for induction of Foxp3 in the context of TLR4 signaling, as it has been reported that identical strains of mice often have considerable phenotypic variation in their intestinal microbiota when relocated to different facilities.

During colitis, antigen-specific T_regs induced by microbial signals produce proinflammatory cytokines that accumulate not only in the inflamed intestine but also in the spleen (Fig. 7). That raises important questions: where do T_reg expansion and conversion into proinflammatory cytokine-producing T cells occur? Are they induced systemically or locally in the mucosa where they are activated after encountering their cognate microbial antigens in the intestinal lumen? Treatment with FTY720, which inhibits lymphocyte egress from lymph nodes [31], decreased total Foxp3⁺ T_regs as well as proinflammatory, cytokine-expressing Foxp3⁺ T_regs in the spleens of colitic RAG^−/− mice that received CBir1 Tg T cells. However, FTY720 treatment did not affect Foxp3⁺ T_regs in the LP, indicating that Foxp3⁺ T_reg expansion and production of proinflammatory cytokines occurred in the intestine, where they were activated by their cognate microbiota antigens, and migrated into systemic lymphoid organs.

Activation of CD4⁺ T cells requires the presentation and coreceptor binding by APCs. Thus, the effects of LPS cannot be isolated solely on CD4⁺ T cells or on the APCs. When APCs are taken into account, LPS strongly abrogates Foxp3⁺ T_reg development under T_reg conditions. With the decrease in Foxp3⁺ T_regs, increases in IFN-γ and IL-17 are also observed in Foxp3⁻ cells, thereby indicating that polarizing cytokines are being produced by APCs upon LPS exposure, which, in turn, inhibits induction of Foxp3⁺ T_regs and skews CD4⁺ T cell differentiation. The effect of LPS directly upon Th1 cells has been detailed previously [22], with LPS able to suppress IFN-γ. We found that the presence of LPS during T_reg differentiation via αCD3/αCD28 was able to suppress Foxp3 induction. Additionally, LPS was able to suppress IFN-γ selectively in T_regs (Fig. 8). It is likely that the absence of TLR4 signaling on T cells partially contributes to the increases in cytokine production from T_regs, as seen in TLR4^−/− and TLR4^−/−/IL-10^−/− mice (Figs. 2B and 4A).

The LPS-TLR4 pathway has been shown to stimulate SOCS3 expression and STAT3 activation, with SOCS3 serving as a crucial negative regulator of LPS-TLR4 responses by inhibiting activation of STAT3 and NF-κB, thus limiting inflammatory cytokine production [33]. By using BMDCs from myeloid-specific SOCS3^−/− mice, we demonstrated that SOCS3 negatively regulated LPS-TLR4 inhibition of T_reg Foxp3 expression, in that SOCS3^−/− further enhanced LPS inhibition of Foxp3⁺ T_regs. This is most likely mediated by higher levels of inflammatory cytokine production in SOCS3^−/− DCs in response to LPS, such as IL-1, IL-6, IL-12, and IL-23, which have been shown to be able to inhibit Foxp3⁺ T_regs [16, 25, 46].

In summary, our data demonstrated that Helios⁺Nrp1⁺ tT_regs and Helios⁻Nrp1⁻ iT_regs are able to produce IFN-γ and/or IL-17 in the intestine and that intestinal inflammation promotes T_reg accumulation and proinflammatory cytokine production. TLR4 negatively regulates T_reg generation through a MyD88-dependent, TRIF-independent manner. TLR4 regulation of Foxp3⁺ T_regs acts indirectly through APCs, as well as directly through CD4⁺ T cells. These data thus provide insights into T_reg regulation under TLR4-mediated conditions.

AUTHORSHIP

A.T.C. performed experiments and wrote the manuscript. S.Y., H.L., and H.L.E-M. performed intestinal cell isolations. A.T.S. contributed reagents and technical advice. H.Q. performed SOCS3 experiments. Z.L., C.O.E., and C.R.N. contributed reagents and to the manuscript review. Y.C. conceived of experiments, supervised the study, and wrote the manuscript.

DISCLOSURES

The authors declare no conflicts of interest.