Viral RNA–Unprimed Rig-I Restrains Stat3 Activation in the Modulation of Regulatory T Cell/Th17 Cell Balance

Retinoid acid inducible gene-I (Rig-I) is the founding member of the Rig-I–like receptor (RLR) family, one subgroup of pathogen pattern recognition receptors (PRRs) for viral RNA (1, 2). Nevertheless, it was originally identified in all-trans retinoid acid–induced granulocytic differentiation of acute promyelocytic leukemia, a biological setting devoid of viral infection (3). Accordingly, Rig-I acts in two ways: to trigger innate immunity reaction upon being primed by its cognate RNA ligands and to modulate several basic cellular processes in the foreign RNA–unliganded form (hereafter referred to as apo–Rig-I). In the first biological setting, the current model suggests that, upon priming by foreign RNA ligands, the innate immunity–dormant cytoplasmic Rig-I will undergo a series of configuration alterations, thus unmasking its N-terminal caspase recruitment domains (CARDs) for dephosphorylation and K63 ubiquitination (4–6). These two key events permit the formation of 2CARD^Rig-I tetramers, which, in turn, nucleates the formation of filamentous CARD^IPS-1 (also known as MAVS/VISA/Cardif) that triggers transcription activation of type I/III IFNs or other proinflammatory cytokines by activating TRAFs, IRF3/7, and NF-κΒ pathways, etc. (7–10). In contrast, accumulating evidence has shown that viral apo–Rig-I is functionally inert but plays active roles in regulating the proliferation and differentiation of normal and malignant myeloid and hepatic cells (3, 11–13), the phagocytosis of bacteria, and the migration of immune cells (14, 15). Interestingly, although much remains to be explored about how apo–Rig-I functions, these diverse activities are also mediated, at least in part, by N-terminal CARDs (11, 15). Nevertheless, we have shown that the Rig-I/Src association, by which Rig-I restrains the leukemic malignancy of acute myeloid leukemia, was attenuated when Rig-I was primed by the viral RNA ligands (11).

Regulatory T cells (Tregs), as a distinct subset of CD4⁺ T cells marked by Foxp3 expression, represent a major regulatory component of adaptive immunity crucial for the maintenance of immune homeostasis (16, 17). Among a plethora of factors that modulate the differentiation/generation of Tregs from naive CD4⁺ T cells, an alternative differentiation and/or even lineage conversion to another distinct subset of Th cells that express and secrete IL-17, Th17 cells, stands as a major negative mechanism (18–21). As exemplified by the observations that, during the in vitro differentiation of naive CD4⁺ T cells, the sole activation of the Il-6–Stat3 axis converts a Treg-polarized culture condition into a Th17-polarized culture condition (22), proper Stat3 activation is central for balanced Treg/Th17 cell generation (18). Mechanistically, the induction and/or activation of Stat3 dictates RORγt expression in promoting a Th17-specific differentiation program (23), as well as downregulates Foxp3, at least partially via upregulating HIF-1α expression to inhibit Treg generation and function (18, 20, 24).

Consistent with the preferential stimulation of Th cells over Tregs by ligand-primed PRRs (25), mutations resulting in constitutive activation of the RLR–IPS-1 axis have been implicated in a few types of autoimmunopathy (26–29). Particularly, when CD4⁺ T cells are invaded by RNA viruses, the classic Rig-I/IPS-1 axis of innate immunity will activate IRF3 to attenuate Smad3 activation and the differentiation of Tregs (30). Interestingly, a similar role for MDA-5 within Tregs upon foreign RNA priming was reported (31). Nevertheless, under physiological and most inflammatory conditions, CD4⁺ T cells are devoid of cytoplasmic invasion by foreign RNA ligands. The potential immunoregulatory activity of apo–Rig-I in these settings remains largely unexplored. Because splenic T cell overactivation-related colitis occurred in young Rig-I^−/− mice bred in sterile conditions (32), we tested whether a supportive role for apo–Rig-I in Treg differentiation, which is opposite from the role of Rig-I upon RNA ligand priming, may exist.

Wild-type (WT) or Rig-I–knockout (KO) littermates (129 or C57BL/6 strain) and other mice were housed under specific pathogen–free conditions and used according to the guidelines of the Institutional Animal Research Committee at Rui-Jin Hospital, Shanghai Jiao-Tong University School of Medicine. C57BL/6 mice were purchased from Shanghai Ling Chang Biological Science and Technology.

Female WT or Rig-I–KO littermates (8–12 wk old) were used in the acute colitis–induction experiment by administering 3% dextran sulfate sodium (DSS) in drinking water for 7 d.

Naive CD4⁺CD62L⁺CD25⁻CD44⁻ T cells were isolated using a Mouse Naive CD4⁺ T Cell Isolation Kit (STEMCELL Technologies; 19765) and activated with plate-bound anti-CD3 Ab (354720; Corning) and soluble anti-CD28 Ab (2 μg/ml, 102112; BioLegend). Cultures were supplemented with anti–IL-4, anti–IL-12, anti–IFN-γ, IL-6 (40 ng/ml, 406-ML), TGFβ (0.5 ng/ml, 240-B), IL-21 (20 ng/ml, 594-ML-010) and IL-23 (20 ng/ml, 1887-ML-010; all from R&D Systems) for Th17 cell induction and supplemented with anti–IL-4, anti–IL-12, anti–IFN-γ, TGF-β (5 ng/ml, 240-B) and IL-2 (40 ng/ml, 402-ML-020; both from R&D Systems) for Treg induction. Five days after activation, these cultivated cells were restimulated with PMA and ionomycin for 4–5 h for intracellular cytokine analysis by flow cytometry.

For CD45.2 WT or KO to CD45.1 WT transplantation, 1 × 10⁷ CD45.2⁺ bone marrow cells were freshly isolated from WT or Rig-I^−/− mice (B6 strain) and injected into lethally irradiated (9 Gy) CD45.1⁺ WT recipients. For CD45.1 WT to CD45.2 WT or KO transplantation, 1 × 10⁷ CD45.1⁺ bone marrow cells were freshly isolated from WT mice (B6 strain) and injected into lethally irradiated (9 Gy) CD45.2⁺ WT or Rig-I^−/− mice. Mice were sacrificed 8 wk later, and chimeric rates were ∼70%. CD4⁺ T cell differentiation in donor-derived CD45.2⁺ hematopoietic cells and host-derived CD45.1⁺ sections were analyzed in parallel.

Four hundred nanograms of GST–RIG-I, 200 ng of HIS-FOXO1 1–167, or 200 ng of HIS-STAT3 1–320 alone or their mixtures were incubated in 20 μl of binding buffer (50 mM Tris [pH 7.5], 150 mM NaCl, 10% glycerol, 0.5% Triton X-100, and 2 mM EDTA) at 4°C for 5 h. Each solution was loaded onto a Superdex 200 5/150 GL (GE Healthcare) and eluted with elution buffer (50 mM Tris [pH 7.5] and 150 mM NaCl) at a rate of 0.1 ml/min. The presence of RIG-I, HIS-FOXO1 1–167, or HIS-STAT3 1–320 in each collection tube was examined using a Western blotting assay.

The experimental process is similar to that reported by Ivanov et al. (33). Briefly, the intestines were obtained and placed in ice-cold PBS supplemented with 2% FBS. After the removal of residual mesenteric fat tissue, Peyer’s patches were excised, and the intestine was opened longitudinally. The intestine was thoroughly washed in ice-cold PBS four to six times and cut into 1.0-cm pieces. The pieces were incubated twice in 10 ml of 5 mM EDTA in HBSS for 20 min at 37°C and 220 rpm. After intensive vortexing for ∼20 s, the epithelial cell layer containing intraepithelial lymphocytes was removed and passed through a 70-μm cell strainer. After the second EDTA incubation, the pieces were washed in ice-cold PBS, cut into 5-mm pieces, and placed in 10 ml of digestion solution containing 4% FBS and 1 mg/ml each of collagenase (Sigma) and DNase I (1:1000; NEB). After the initial 20 min, the solution was vortexed intensively and passed through a 70-μm cell strainer. The procedure was repeated one more time. Supernatants from a single intestine were combined and resuspended in 5.6 ml of RPMI 1640 medium supplemented with 5% FBS. Percoll gradient separation was performed by centrifugation for 25 min at 600 × g at room temperature. Lamina propria (LP) lymphocytes were collected at the interphase of the Percoll gradient.

Anti-CD4 (562891), anti-CD45.1 (563010), anti-CD45.2 (552950), anti-CD44 (559250), anti-CD62L (553151), anti–Ifn-γ (554411), anti–Il-17A (560220), and anti–Il-4 (554435) Abs were from BD; anti-Foxp3 (12-5773-82), anti-CD25 (15-0251-82), and anti-Ki67 (50-5698-82) Abs were from eBioscience; and anti–H-2Kb Ab (116518) was from BioLegend. Staining of Foxp3, Ifn-γ, and Il-17A was performed using a Fixation/Permeabilization Kit (eBioscience). For intracellular cytokine detection, cell suspensions were treated with 50 ng/ml PMA, 1 ng/ml ionomycin, and 10 μg/ml monensin in complete medium for 4–5 h, followed by surface staining, permeabilization, and intracellular staining. All flow cytometric analyses were performed on an LSR II Fortessa cytometer (BD Biosystems), and the data were analyzed using FlowJo software.

The coimmunoprecipitation and Western blotting assays were performed as previously described (11).

Briefly, primary naive CD4⁺ T cells were freshly isolated from spleen of 8–12-wk-old WT or Rig-I–KO littermates, resuspended in 100 μl of prewarmed mouse T cell nucleofector solution containing 1.5 μg of different plasmids (MigR1, MigR1–Rig-I, MigR1–Rig-I-3A, or MigR1–Rig-I–PxxP mutant), and gently transferred into an Amaxa Nucleofector cuvette for electroporation. The next day, cells were stimulated with anti-CD3, anti-CD28 Abs, Il-2, and Tgfβ for 5 d and processed for intracellular staining and flow cytometry.

RNA samples were isolated using TRIzol Reagent (Invitrogen), according to the manufacturer’s procedures. SYBR Green PCR reagents were purchased from Takara, and RT-PCR was performed following the manufacturer’s protocol in an ABI 7500 PCR machine (Applied Biosystems).

Tregs that were purified from spleens of WT or Rig-I^−/− littermates and WT CD4⁺ T responder cells were cocultured for 4 d. Culture supernatants were collected, and the secretion of Il-2 was measured by ELISA, according to the manufacturer’s instructions (M2000; R&D Systems). Natural Tregs were isolated from the thymuses of WT or Rig-I^−/− littermates and stimulated with PMA and ionomycin for 5 h. The level of Il-10 was measured according to the manufacturer’s instructions (M1000B; R&D Systems).

All proteins were expressed in Escherichia coli BL21 cells (TIANGEN), with RIG-I being encoded in pGEX-4T3 expression vector (GE Healthcare) and FOXO1 1–167 and STAT3 1–320 fragments in pASK-IBA33 plus vectors (IBA). Bacterial cultures were grown at 37°C until reaching an OD₆₀₀ of 0.8. RIG-I expression was induced by addition of 1 mM IPTG, and cells were subsequently incubated at 18°C overnight. FOXO1 1–167 and STAT3 1–320 expression was induced by the addition of 200 μg/l AHT, and cells were incubated at 30°C for 5 h. The collected bacteria were sonicated and centrifuged at 20,000 rpm for 1 h. The lysates were purified with GSTrap or HisTrap (GE Healthcare). Control GST proteins were purchased from Millipore. For pull-down experiments, 100 ng of GST–RIG-I was preincubated or not with 50 ng 5′ppp-dsRNA (InvivoGen) in Buffer A (20 mM Tris-HCl [pH 7.5], 2 mM ATP, 5 mM MgSO₄, and 0.5 mM DTT) at 30°C for 1 h. Then 50 ng/μl HIS-FOXO1 1–167 or HIS-STAT3 1–320 was added to the mixture and incubated for another 2 h. Glutathione particles (Promega) were used to pull-down the GST-fusion protein and the associated protein, which were then analyzed by Western blotting assay.

WT or Rig-I–KO naive CD4⁺ T cells were stimulated under Treg-induction conditions for 3 d. Total RNA was extracted using TRIzol Reagent (15596-018; Life Technologies), following the manufacturer’s instructions. Array hybridization (901229) and scan (00-00212; both from Affymetrix) were performed according to the manufacturer’s instructions. Raw data were normalized using MAS 5.0 algorithm, Gene Spring Software 11.0 (Agilent Technologies, Santa Clara, CA). Microarray data have been submitted to the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/) under accession number GSE97927.

Kaplan–Meier survival analysis, Student t test, or χ² test was used to calculate p values where appropriate. The p values < 0.05 were considered significant.

First, we isolated naive CD4⁺ T cells from the spleens of WT or Rig-I^−/− littermates (Supplemental Fig. 1A) and examined their in vitro differentiation into Tregs. Although no differences between two groups were observed in terms of cell survival and cell cycle (Supplemental Fig. 1B), interestingly, the relative generation of Foxp3⁺Il-17⁻ Tregs was significantly decreased in Rig-I^−/− CD4⁺ T cells compared with WT (Fig. 1A). Notably, this was accompanied by abnormally increased generation of Il-17⁺Foxp3⁻ Th17 cells in Rig-I^−/− CD4⁺ T cells (Fig. 1A). ELISA measurement of Treg-marking cytokine Il-10 and Th17 cell–marking cytokine Il-17 in the supernatants of these cultures supported that Rig-I deficiency in CD4⁺ T cells dampened Treg generation while enhancing Th17 cell generation (Fig. 1B). In agreement with this, more Th17 cells were produced on the Rig-I^−/− background under Th17 cell–polarized culture conditions (Supplemental Fig. 1C, 1D). To determine whether this abnormal Treg/Th17 cell differentiation was due to an intrinsic defect of Rig-I^−/− CD4⁺ T cells or a possible priming of naive CD4⁺ T cells by an uncharacterized microenvironmental alteration in Rig-I^−/− spleens, we performed reciprocal bone marrow CD45⁺ hematopoietic cell transplantation between WT and Rig-I^−/− mice (Supplemental Fig. 1E), thus creating four types of chimeric mice (Supplemental Fig. 1F). The relative generation of Tregs or Th17 cells did not differ between the CD45.1⁺ WT naive CD4⁺ T cells isolated from the WT to WT (2) transplantation recipients and the counterparts from the WT to KO transplantation recipients (Fig. 1C), indicating that this Rig-I deficiency–associated Treg/Th17 cell imbalance was largely of a hematopoietic origin. Conversely, a biased generation of Th17 cells over Tregs was observed in CD45.2⁺ Rig-I^−/− naive CD4⁺ T cells isolated from KO to WT transplantation chimeric recipients compared with the WT naive CD4⁺ T cells from the WT to WT (1) transplantation recipients (Fig. 1D). Moreover, the generation of Tregs by Rig-I^−/− naive CD4⁺ T cells isolated from the KO to WT chimeric mice was significantly restored by the reintroduction of a foreign Rig-I (Fig. 1E). Taken together, these results indicated that Rig-I within CD4⁺ cells exerts a regulatory effect to maintain the Treg/Th17 cell balance.

Next, we sought to determine the mechanisms behind this Rig-I deficiency–associated Treg/Th17 cell imbalance. It has been clearly shown that activation of the Il-6–Stat3 axis switches Treg differentiation toward Th17 cell differentiation in the presence of TGF-β signaling (20), and that STAT3 expression in T cells is essential for the induction of colitis (18). Coincidently, ingenuity pathway analyses of cDNA array profiling data for WT or Rig-I^−/− CD4⁺ T cells that were polarized to Treg differentiation revealed that colitis- and Jak-Stat–related pathways were among the top ones affected by Rig-I deficiency (Fig. 2A), whereas GSEA indicated that Tgfβ signaling strength during in vitro Treg differentiation was not greatly disturbed by Rig-I deficiency (Supplemental Fig. 2A).

Indeed, the expression of three of four classic Stat3-regulated genes was upregulated in Rig-I^−/− CD4⁺ T cells (Fig. 2B) (34–37). The mRNA level of Il-22, a Th17 cell–derived cytokine whose transcription was activated by Stat3 (38), was increased by Rig-I deficiency in Treg- or Th17 cell–polarized conditions (Supplemental Fig. 2B). Interestingly, although the mRNA level of Foxp3 was decreased by Rig-I deficiency in Th17 cell– or Treg-polarized CD4⁺ T cells, Stat3 mRNA levels were not altered by Rig-I deficiency (Supplemental Fig. 2C). Western blotting assay showed that a substantial amount of Rig-I protein was readily detected in the in vitro–induced WT Tregs or Th cells (Fig. 2C) and that the p-Stat3–Y level, but not the total Stat3 level, was highly elevated in Rig-I^−/− CD4⁺ T cells compared with WT CD4⁺ T cells (Fig. 2C). In accordance, we detected a similar alteration in the p-Stat3–Y level in the freshly isolated intestinal CD4⁺ T cells (Fig. 2D). As expected, knockdown of Stat3 in Rig-I^−/− CD4⁺ T cells at least partially restored Treg differentiation over Th17 cell differentiation in vitro (Fig. 2E), implicating a functional contribution of this unrestrained Stat3 overactivation to the imbalanced Treg/Th17 cell generation of Rig-I^−/− CD4⁺ T cells.

Because the mRNA and protein levels were not altered by Rig-I deficiency, we used a coimmunoprecipitation (co-IP) assay in 293T cells to screen for a possible physical association of human RIG-I with STAT family members and FOXP3, which indicated a specific association of RIG-I with endogenous STAT3 (Supplemental Fig. 2D). This association of Rig-I with Stat3 was further verified by co-IP assay between these endogenous proteins in freshly isolated splenic CD4⁺ T cells (Fig. 2F). Thus, these results indicated that Stat3 is a crucial and direct target of the Rig-I regulatory effect on Treg/Th17 cell generation.

Proper Stat3 activation is dynamically modulated by its phosphorylation via Jak1/2 and dephosphorylation via Shp-1/2 (39). Interestingly, co-IP assay of CD4⁺ T cells exposed to Il-6 stimulation showed that the association of Shp2/Stat3, but not Shp1/Stat3, was specifically decreased by Rig-I deficiency, whereas the association of Jak1/Stat3, but not Jak2/Stat3, was increased by Rig-I deficiency (Fig. 3A). This result indicated that by specifically regulating upstream and downstream regulatory steps, Rig-I restrains Stat3 phosphorylation. As revealed by domain-mapping experiments (Supplemental Fig. 3A, 3B), although N-terminal CARDs of RIG-I were responsible for the association with STAT3, the C-terminal part of RIG-I mediated its association with SHP2. These observations suggested a mechanism whereby Rig-I served as a chaperone-like molecule to facilitate the formation of a trimeric complex. In support of this, a co-IP assay showed that, in freshly isolated murine CD4⁺ T cells or in 293T cells overexpressing human RIG-I, STAT3, and SHP2, at least a portion of Rig-I, Stat3, and Shp2 coexisted in the same protein complex (Fig. 3B, Supplemental Fig. 3C). Moreover, the overexpression of RIG-I promoted the STAT3/SHP2 association in a dose-dependent manner (Fig. 3C). In contrast, the domain-mapping experiments showed that RIG-I and JAK1 occupied the same region of STAT3: aa 689–770 at the C terminus (Supplemental Fig. 3D, 3E). Furthermore, RIG-I overexpression inhibited JAK1/STAT3 association in a dose-dependent manner (Fig. 3D), suggesting that Rig-I also competitively blocks the recognition of Stat3 by Jak1.

Next, we examined whether the status of the classic innate immunity IPS-1 pathway would be altered by Rig-I deficiency when CD4⁺ naive T cells were cultured under Treg-polarized conditions. Of note, Irf3 dimer formation, which is indicative of the activation of the Rig-I–IPS-1 axis in CD4⁺ T cells, was undetectable after WT or Rig-I^−/− CD4⁺ naive T cells were induced to Treg differentiation (Supplemental Fig. 4A, 4B), excluding the possibility that the Rig-I–IPS-1 axis is actively involved in the differentiation programs of Tregs in the biological settings devoid of viral infection (30). Furthermore, as indicated by a previous observation that IPS-1 did not primarily affect CD4⁺ T cell differentiation (40), IPS-1 overexpression or knockdown did not alter the Treg-differentiation pattern of Rig-I^−/− naive CD4⁺ T cells (Supplemental Fig. 4C, 4D).

In contrast, the regulatory activities of Rig-I on Stat3 described above presumed that apo–Rig-I would be able to physically associate with Stat3. In support of this, the pull-down assay and the gel filtration of the purified bacterially expressed proteins demonstrated a direct, stable, and RNA priming–independent physical association between full-length RIG-I and the STAT3 fragment (Fig. 4A, 4B, Supplemental Fig. 4E). To further discriminate between the Rig-I/Stat3 association and the Rig-I/IPS-1 association, we created a Rig-I (3A) mutant in which the three amino acids within the N-terminal CARDs crucial for Rig-I CARD tetramer formation and/or for the formation of the filamentous IPS-1 CARDs were mutated (8, 41). As expected, the Rig-I N284 (3A) mutant failed to activate the transcription of Ifn-β (Fig. 4C). Nevertheless, the Rig-I mutant (3A) was able to associate with Stat3 in a similar manner to WT Rig-I (Fig. 4D). Finally, delivery of this Rig-I (3A) mutant was able to rescue the differentiation deficiency of Tregs from Rig-I^−/− naive CD4⁺ T cells in vitro as the WT Rig-I or Rig-I PxxP mutant (that lost the ability to inhibit Src/AKT association) did (Fig. 4E) (11), indicating that the regulation of Treg differentiation by Rig-I/Stat3 association was independent of Rig-I/IPS-1 signaling, as well as the Src/AKT pathway.

As described previously (32), Rig-I^−/− mice exhibit increased susceptibility to the development of colitis under sterile breeding conditions; however, whether a Treg/Th17 cell imbalance occurred in the intestines in situ remains undetermined. To address this, WT and Rig-I^−/− mice were treated or not with 3% DSS in drinking water for 7 d. As described previously, young Rig-I^−/− mice spontaneously developed colitis, and DSS treatment induced a more severe colitis in Rig-I^−/− mice than in WT mice, as reflected by shortened colons, exacerbated intestinal bleeding and necrosis, and disrupted membrane structure in colon sections (data not shown). In suggestion of an immunological origin for this colitis, we detected a massive in situ CD3⁺ T cell infiltration into the inflamed intestine of young Rig-I^−/− mice that was exacerbated by DSS treatment and paralleled the colitis severity (Fig. 5A). Interestingly, a consistent decrease in the generation of Tregs, as well as an increased generation of Th17 cells, was observed in spleen, LP, and mesenteric lymph nodes (MLNs) of Rig-I^−/− mice compared with WT littermates with or without DSS treatment (Fig. 5B). In a naive WT CD4⁺ cell transfer–mediated colitis model in Rag-1^−/− mice, the cotransplantation of highly purified Rig-I^−/− CD4⁺CD25⁺ cells failed to prevent the loss of body weight and the progression of colonic inflammation (Fig. 5C, 5D), indicating a deficient function of Rig-I^−/− Tregs. To test the pathological role of this decreased Treg generation in association with Rig-I^−/− colitis, we performed “rescue experiments” by transplanting WT Tregs into Rig-I^−/− mice. Notably, Treg supplementation into Rig-I^−/− mice significantly remedied the overproduction of Th17 cells in situ (Fig. 5E) and ameliorated DSS-induced colitis (Fig. 5F, 5G), verifying a crucial functional contribution of Rig-I deficiency–induced Treg reduction to the occurrence of colitis.

Finally, we tested the effect of a Stat3-specific inhibitor, Stattic, on the imbalanced Treg/Th17 cell differentiation in vivo and colitis observed in Rig-I^−/− mice. As expected, in Treg-inducible cultures, the addition of Stattic significantly promoted Treg differentiation from Rig-I^−/− naive CD4⁺ T cells while decreasing Th17 cell differentiation in a dose-dependent manner (Fig. 6A, 6B). Likewise, the administration of Stattic in vivo significantly ameliorated DSS-induced colitis in Rig-I^−/− mice (Fig. 6C), which was accompanied by an increased generation of Tregs but a decreased generation of Th17 cells in MLNs, LP, and spleen (Fig. 6D). Taken together, these results further support the notion that the restriction of Stat3 activity by apo–Rig-I is crucial for maintaining Treg/Th17 cell balance and intestinal homeostasis.

Once PRRs, such as TLRs or RLRs, within CD4⁺ T cells recognize their cognate ligands, the Treg/Th17 cell balance is tilted by the classic signaling pathways that mediate the activation of innate immunity (25, 30, 42–44), thus constituting a direct link between innate immunity recognition and activation with the initiation and procession of adaptive immunity. However, how an apo-PRR regulates CD4⁺ T cell progeny generation remains largely unexplored. In this study, we have shown a direct regulatory role for apo–Rig-I in restraining Stat3 activation within CD4⁺ T cells, which is involved in the maintenance of Treg/Th17 cell balance. It is unknown how these two distinct functional modes, as a PRR of viral RNA and as an authentic signal transducer of CD4⁺ T cells, can be combined in a single Rig-I molecule. Nevertheless, it can be postulated that the appearance and evolution of Rig-I have been delimited by the regulatory machineries of innate immunity and adaptive immunity, because RLR family members, especially Rig-I, evolutionarily appeared only after the acquisition of adaptive immunity (since jawed fish) (45).

In contrast to the intrinsic suppressing role of RNA ligand–primed Rig-I in Treg differentiation by antagonizing TGF-β signaling strength (30), apo–Rig-I exerts an opposing role to maintain Treg/Th17 cell balance, at least in part by restraining Stat3 activation; this Treg/Th17 cell–balancing role of apo–Rig-I seems largely independent of the modulation of TGF-β signaling strength (Supplemental Fig. 2A). The physiological significance of this Treg/Th17 cell–balancing role for apo–Rig-I during various types of immunological responses remains largely unexplored, although in this study we provide evidence that this role of apo–Rig-I contributes, at least in part, to the suppression of colitis (Fig. 5). Moreover, in light of a previous study showing that proper activity of Tregs is highly beneficial to the process of innate immunity that prevents in situ viral infections from spreading (46), this protective role of apo–Rig-I in Treg generation seems reconciled with the execution of an effective immunological response against viral infection.

It is well established that Stat3 activation governs the cell differentiation and functionality of Tregs and Th17 cells (18, 23, 47). Although it was previously shown that Rig-I functionally associates with and promotes Stat1 activation in hepatic cells and myeloid cells (12, 13, 48), in this study, we provide evidence that apo–Rig-I probably possesses stronger affinity for Stat3 than for Stat1 (in 293T cells) and that endogenous Rig-I and Stat3 indeed physically associate within CD4⁺ T cells (Fig. 2). It is likely that Rig-I regulates Stat3 and Stat1 in different modes. Although Rig-I/Stat1 association was shown to promote type I IFN–induced Stat1 activation by blocking Stat1/Shp1 association (12), we provide evidence that apo–Rig-I/Stat3 association promoted Stat3/Shp2 association by blocking Jak1/Stat3 association (Fig. 3). Unlike the role of apo–Rig-I in competitively hindering Src/AKT association (11), N-terminal CARDs, but not the PxxP motif, are needed for this apo–Rig-I/Stat3 physical association (Fig. 4). Finally, it is interesting to consider that, in both functional modes of Rig-I, with or without RNA ligand priming, N-terminal CARDs are essential. In this regard, we provide evidence that, in the in vitro–induced Tregs, Rig-I deficiency in CD4⁺ T did not significantly alter Irf3 activation status, and IPS-1 overactivation or knockdown did not alter the generation of Rig-I^−/− Tregs (Fig. 4), excluding the possibility that an ectopic alteration of the classic innate immunity pathway by apo–Rig-I was involved in the maintenance of Treg/Th17 cell balance. Verifying this, Rig-I with CARDs mutant that failed to associate with IPS-1 retained the ability to associate with Stat3 and to rescue the deficient in vitro generation of Rig-I^−/− Tregs. Nevertheless, it remains to be explored whether the crucial motives of apo–Rig-I CARDs are special in their physical association with Stat3 and how they differentiate their association with Src and Stat3.

We thank the Animal Experiment Research Center, Rui-Jin Hospital, Shanghai Jiao-Tong University School of Medicine, for feeding the mice.

The authors have no financial conflicts of interest.