Systemic autoimmunity in murine STAT3 gain-of-function syndrome is characterized by effector T cell expansion, in the absence of overt regulatory T cell dysfunction

Abstract

Germline gain-of-function (GOF) mutations in the transcriptional factor signal transducer and activator of transcription 3 (STAT3) promote early-onset multisystemic autoimmunity. To investigate how increased STAT3 promotes systemic inflammation, we generated a transgenic knock-in strain expressing a pathogenic human mutation STAT3^K392R within the endogenous murine locus. As predicted, STAT3^K392R mice develop progressive lymphoid hyperplasia and systemic inflammation, mirroring the human disease. However, whereas the prevailing model holds that increased STAT3 activity drives human autoimmunity by dysregulating the balance between regulatory T cells (Treg) and Th17 cell differentiation, we observed increased Th17 cells, in the absence of major defects in Treg differentiation or function. In addition, STAT3^K392R animals exhibited a prominent accumulation of IFN-γ producing CD4⁺ and CD8⁺ T cells. Together, these data provide new insights into this complex human genetic syndrome and highlight the diverse cellular mechanisms by which dysregulated STAT3 activity promotes breaks in immune tolerance.

INTRODUCTION:

Patients with gain-of-function (GOF) mutations in gene signal transducer and activator of transcription 3 (STAT3) develop a complex syndrome of early-onset autoimmunity. Amongst the diverse clinical manifestation are type 1 diabetes, autoimmune enteropathy, inflammatory lung disease, polyarthritis, autoimmune cytopenias, and postnatal short stature (1–3). The Janus kinase (JAK)-STAT signaling pathway regulates transcriptional responses to extra-cellular cytokines and growth factors. The complex phenotypes of STAT3 GOF syndrome can therefore be attributed to both STAT3 activation downstream of multiple cytokine receptors and to ubiquitous STAT3 expression in hematopoietic and non-hematopoietic lineages. Because of this complexity, our understanding of STAT3 GOF pathophysiology remains poor.

Several hypotheses have been proposed to explain STAT3 GOF autoimmunity. Chief among these, subjects with STAT3 GOF syndrome exhibit reduced regulatory T cell (Treg) numbers (1, 2), although immunophenotyping has been limited to peripheral blood and at least one STAT3 GOF patient developed multisystem autoimmunity despite normal Treg numbers and Treg suppressor function (2). Thus, additional immune mechanisms may facilitate STAT3 GOF disease, including expansion of pathogenic Th17 cells, dysregulation of Th17/Treg balance, enhanced pro-inflammatory cytokine signaling, or tissue-specific impacts on organ function (4, 5).

Ultimately, the rarity of STAT3 GOF syndrome and limited availability of human clinical samples have prevented a detailed interrogation of underlying immune mechanisms. To address these limitations, we generated and phenotyped a new transgenic strain expressing a pathogenic human STAT3 mutation (STAT3^K392R) within the endogenous murine Stat3 locus.

Materials and methods:

Murine model:

STAT3^K392R knock-in murine model described in Supplemental Figure 1. Murine studies performed in specific pathogen-free (SPF) environment in accordance with IACUC approved protocols. To establish mixed bone marrow (BM) chimeras, 6 × 10⁶ CD45.1 WT and CD45.2 Stat3^WT/GOF BM (50%:50% ratio) was injected retro-orbitally into lethally irradiated (450cGy × 2 doses) CD45.1 WT recipients. Resulting BM chimeras were sacrificed at 14 weeks post-transplant.

Histopathology:

Formalin-fixed, paraffin-embedded tissue sections from 12 m.o. Stat3^WT/GOF and WT littermates were stained with H&E. Pathology was scored by a board-certified veterinary pathologist blinded to genotype.

Flow cytometry:

Flow cytometry of splenocyte, lymph node (LN), and bone marrow suspensions was performed as described (6). Anti-murine antibodies include: CD19 (1D3), CD95 (Jo2), CD45.1 (A20) and CD45.2 (104) from BD Biosciences; B220 (RA3-6B2), CD4 (GK1.5 and RM4-4), CD8 (53-6.7), CD19 (6D5), CD21 (7E9), CD23 (B3B4), CD24 (M1/69), CD25 (PC61), CD44 (IM7), CTLA-4 (IC10-4B9), Helios (22F6), TACI (8F10) from BioLegend; CD8 (53-6.7), FoxP3 (FJK-16s), IFN-γ (XMG1.2), IL-17A (17B7) from eBioscience; CD8 (53–6.7) from Life Technologies; Live/Dead (L34962) from ThermoFisher; PNA (Fl-1071) from Vector Labs; goat anti-mouse IgM-, IgG-, IgA-HRP conjugated, unlabeled, or isotype, IgG2c (1079–02) from Southern Biotechnology.

ELISA:

Performed as described (6) using the following reagents: calf thymus dsDNA (Sigma-Aldrich D3664-5X2MG); Sm/RNP (Arotec Diagnostic Limited ATR01-10); IgM, IgA, or IgG (total Ig titers); and goat anti-mouse IgG-, IgG2c-, IgG3-HRP conjugated Ab (Southern Biotech).

In vitro T cell differentiation assay:

WT and Stat3^WT/GOF naïve CD4⁺ T cells were purified using the “Naïve CD4+ T Cell Isolation Kit” (Miltenyi Biotec, Cat. 130-104-453), cultured in supplemented RPMI (Th0, Th1, and Treg) or supplemented IMDM (Th17) media, and stimulated with plate-bound anti-CD3 (2.5ug/mL; BioXCell, 145-2C11) and soluble anti-CD28 (1ug/mL; SouthernBiotech, PV-1). T cell differentiation conditions: Th1, anti-IL-4 (10ug/mL; BioXCell, 11B11), murine IL-2 (50ng/mL; PeproTech, Cat. 212–12), and murine IL-12 (10ng/mL; R&D Sysemts, Cat. 419-ML-010); Treg, anti-IL-4, anti-IFNγ (10ug/mL; BioXCell, R4-6A2), murine IL-2, and human TGF-β (2.5ng/mL; PeproTech, Cat. 100–21); classical Th17, anti-IL-4, anti-IFNγ, human TGF-β, and murine IL-6 (30ng/mL; PeproTech, Cat. 216–16); pathogenic Th17, anti-IL-4, anti-IFNγ, human TGF-β, murine IL-1β (20ng/mL; Miltenyi Biotec, Cat. 130-094-053), and murine IL-23 (20ng/mL; R&D Systems, Cat. 1887-ML-010); Th0, anti-IL-4, anti-IFNγ, and IL-2. On day 5, cells were stimulated with phorbol-12-myristate-13-acetate (PMA; 50ng/mL; Sigma-Aldrich, Cat. 5.00582), ionomycin (1ug/mL; MilliporeSigma, Cat. 407950), and GolgiPlug Protein Transport Inhibitor (1:1000; BD Biosciences, Cat. 555029) and then stained for flow cytometry.

In vitro Treg suppression assay:

CD45.1⁺CD25⁻CD4⁺ WT T cells and CD45.2⁺CD25⁺CD4⁺ Treg (WT and Stat3^WT/GOF) were purified from spleen and LN using “CD4⁺CD25⁺ Regulatory T Cell Isolation Kit” (Miltenyi Biotec, Cat. 130-091-041). 1×10⁵ Cell Trace Violet (CTV)-labelled CD45.1⁺CD25⁻CD4⁺ T cells were stimulated with 5×10⁵ irradiated CD4^neg cells and 0.5μg/mL anti-CD3 monoclonal antibodies (BioXCell, Cat. # BE0001-1) plus titrated ratios of Tregs. Proliferation of CD45.1⁺CD25⁻CD4⁺ T cells was quantified by CTV dilution on Day 4.

EAE model:

9–11 week WT and Stat3^WT/GOF mice were injected s.c. with 200μL MOG_35–55/CFA emulsion plus 100ng Pertussis Toxin on days 0 and 1 (Hooke Laboratories; EK-2110), and monitored daily for weight loss and clinical score (scored 0–5). On day 17, CNS was processed in digestion buffer (HBSS, 10% FBS, 1.5mg/mL collagenase, 10μg/mL DNAse; 37°C for 45min), followed by immune cell isolation on a 38%/70% Percoll gradient.

RESULTS:

Spontaneous systemic autoimmunity in a murine model of STAT3 GOF syndrome

We generated a knock-in mouse model allowing conditional expression of pathogenic Stat3^K392R in the endogenous locus via Cre-mediated replacement of WT Stat3 exon 13 with the K392R-expressing mutant allele (Supplemental Figure 1). In the current study, we focused our analyses on the impact of global Stat3^K392R expression by crossing Stat3^K392R and CMV-Cre animals (7) to induce germ-line recombination of the mutant allele (hereafter Stat3^GOF model).

When heterozygous Stat3^WT/GOF mice were intercrossed to generate experimental animals, homozygous Stat3^GOF/GOF offspring were born at frequencies below expected mendelian ratios, suggesting embryonic or perinatal lethality (Fig. 1A). In addition, Stat3^GOF/GOF homozygous mice exhibited accelerated mortality with only 50% of mice surviving beyond 12 weeks (Fig. 1B). Although heterozygous Stat3^GOF expression did not result in reduced survival, male Stat3^WT/GOF mice manifested reduced weight gain with age, reflecting either Stat3^GOF-dependent growth failure (8) or the impact of chronic illness (Fig. 1C). Cohorts of heterozygous Stat3^WT/GOF mice, sacrificed at 6 and 12 months, exhibited splenomegaly and progressive lymph node (LN) expansion (Fig. 1D, E). In addition, histopathologic evaluation of organ inflammation in 12-month-old Stat3^WT/GOF heterozygous mice identified widespread inflammatory lesions within salivary glands, lungs, liver, and pancreas of affected animals (Fig. 1F–J). In summary, murine expression of a pathogenic human STAT3 GOF variant promotes gene-dose dependent systemic inflammation.

Figure 1: Stat3^K392R expression promotes multi-system autoimmunity in transgenic murine model:

(A) Genotype of 48 pups derived from heterozygous Stat3^WT/GOF breeders. *, P<0.05 by chi-square goodness-of-fit. (B) Kaplan-Meier survival curve showing accelerated mortality of homozygous Stat3^GOF/GOF animals. ****, P<0.0001, by log-rank test. (C) Body weight in male (left) and female (right) mice. (D) Spleen weight. (E) Total LN cell count. (F-I) Representative images showing widespread organ inflammation in 12 m.o Stat3^WT/GOF mice (right), compared with age-matched controls (left). Arrows depict inflammatory infiltrates of predominantly lymphoid cells within the lungs adjacent to pulmonary blood vessels (F), portal tracts in the liver (G), periductal regions of salivary glands (H), and islets and intralobular ducts of the pancreas (I). 10X magnification; bars 100μm. (J) Blinded pathology score. (C, D, E, J) *, P<0.05; **, P<0.01 by two-tailed Student’s t test. Each data point indicates individual animal.

Preserved development of functional Treg in Stat3^WT/GOF mice.

To gain a greater understanding of the cellular drivers of STAT3 GOF autoimmunity, we immunophenotyped aged Stat3^WT/GOF heterozygous mice. In parallel with widespread organ inflammation, we observed increased activated CD44-expressing CD4⁺ and CD8⁺ T cells in spleen and LN of Stat3^WT/GOF animals (Fig. 2A, B). Surprisingly, no decrease in Foxp3⁺ Treg in Stat3^WT/GOF animals was noted. Rather, the number of Tregs was modestly expanded in Stat3^WT/GOF mice in proportion with the expansion of CD44⁺ effector CD4⁺ T cells (Fig. 2C–E). In addition, CD25 and CTLA-4 expression was preserved on the surface of Stat3^WT/GOF Foxp3⁺ Treg, as was the ratio of thymic derived vs. induced Treg, defined by Helios expression (Fig. 2F, G) (9). Finally, in keeping with this unaltered surface phenotype, Stat3^WT/GOF Treg exhibited normal suppressive activity in vitro. (Fig. 2H, I). These data were unexpected, since STAT3-mediated induction of suppressor of cytokine signaling 3 (SOCS3) limits STAT5 phosphorylation, a known transcriptional regulator of Treg differentiation (10, 11). Consistent with this model, naïve CD4⁺ T cells from Stat3^WT/GOF mice exhibited reduced differentiation into CD25⁺Foxp3⁺ Treg in vitro (Fig. 2J). To reconcile these data, we established mixed bone marrow chimeras to assess whether the Stat3^GOF allele impacts Treg differentiation in competitive settings. While Stat3^GOF expression did not block Treg development, WT CD4⁺ T cells exhibited a modest competitive advantage in vivo (Fig. 2K–M). Thus, despite these more subtle impacts on regulatory T cell biology, functional Tregs can develop in murine STAT3 GOF syndrome suggesting important contributions of additional immune lineages.

Figure 2: Normal Treg development and function in Stat3^GOF mice:

(A, B) No. effector (CD44⁺) CD4⁺ and CD8⁺ T cells in spleen (A) and LN (B) of WT and Stat3^WT/GOF mice. (C, D) % Foxp3⁺ (C) and no. Foxp3⁺CD4⁺ T cells (D) in spleen and LN. (E) CD44⁺CD4⁺ Teff to Foxp3⁺CD4⁺ Treg ratio in spleen and LN of 6 m.o. WT and Stat3^WT/GOF mice. (F) Left panels: Histograms of surface CD25 and CTLA-4 expression on WT (dashed line) and Stat3^WT/GOF (solid line) Treg. Gray histogram: WT Foxp3⁻CD4⁺ non-Treg. Right: CD25 and CTLA-4 MFI (normalized to WT non-Treg). (G) Histogram showing splenic Treg Helios (left panel) and %Helios⁺ Treg (right panel) in indicated strains. (H) Histogram showing CD4⁺ T cell proliferation by Cell-Trace Violet (CTV) dilution in WT CD4⁺ T cells co-cultured WT and Stat3^WT/GOF Treg. (I) % CD4⁺ T cell proliferation at different Treg ratios. (J) In vitro Treg differentiation. Data combined from two replicate experiments. (K) Representative FACS plots showing CD25⁺Foxp3⁺ Treg in CD45.1 (WT) vs. CD45.2 (WT; Stat3^WT/GOF) CD4⁺ T cells in control vs. Stat3^WT/GOF chimeras. Number indicates % in gate. (L) % CD25⁺Foxp3⁺ Treg in CD45.1 (WT) vs. CD45.2 (WT; Stat3^WT/GOF) CD4⁺ T cells. (M) Selection of CD45.1⁺ WT vs. CD45.2⁺ Stat3^WT/GOF CD4⁺ T cells into the Treg compartment (CD45.1/CD45.2 ratio in Treg compartment normalized to ratio in non-Treg CD4⁺ T cells). (A-L) *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001; NS, not-significant; by one-way ANOVA and Tukey’s multiple comparison test (A,B,F,J), ANOVA plus Šidák’s correction multiple comparison test, with a single pool variance (C,D) and by two-tailed Student’s t test (E,G,I,L,M). Each data point equals individual WT (open circle) and Stat3^WT/GOF (solid circle) animal.

Enhanced STAT3 activity promotes polyclonal B cell expansion in the absence of serum autoantibodies.

Humoral autoimmunity, in particular autoimmune cytopenia, is a frequent manifestation of STAT3 GOF syndrome. STAT3 functions downstream of several key cytokines impacting B cell activation, including IL-6, IL-10, and IL-21. Based on these data, we anticipated that Stat3^GOF expression would drive polyclonal B cell activation, and possibly spontaneous germinal center (GC) formation and class-switched autoantibody production. However, patients with STAT3 GOF syndrome frequently exhibit hypogammaglobulinemia, highlighting the complex roles for STAT3 in regulating B cell function.

Onset of systemic autoimmunity in 12 m.o. Stat3^WT/GOF mice was accompanied by increased total splenic and LN B cells, although enhanced STAT3 activity exerted no major impact on splenic B cell development, except for a modest reduction in transitional T1 and T2 cells and expanded follicular mature (FM) B cells (Supplemental Fig. 2A–C). We also observed no spontaneous GC formation and no increase in spleen, LN, or bone marrow plasma cells in aged Stat3^WT/GOF mice (Supplemental Fig. 2D, E). Moreover, although total serum IgM and IgG titers were modestly increased in aged animals, 12 m.o. Stat3^WT/GOF mice lacked anti-dsDNA and anti-Sm/RNP autoantibodies, despite multi-organ inflammation at this age (Supplemental Fig. 2F,G).

Expansion of IFN-γ⁺ Th1 and IL17A⁺ Th17 cells in diseased Stat3^GOF mice

Co-ordinated cytokine signals promote differentiation of naïve CD4⁺ T cells into distinct effector subsets. Th17 differentiation is facilitated by TGF-β and IL-6 cytokines, which induce STAT3-dependent activation of the lineage-defining transcription factor retinoid-related orphan receptor γt (RORγt), implicating Th17 cell expansion in STAT3 GOF autoimmunity (1, 2). To validate that murine Stat3^K392R functions as a GOF allele, we performed in vitro CD4⁺ T cell differentiation assays and confirmed that Stat3^GOF drove increased Th17 differentiation while modestly reducing Th1 differentiation (Fig. 3A). To address the potential T cell lineages promoting systemic inflammation in vivo, we measured T cell cytokine production in aged Stat3^WT/GOF mice. Surprisingly, IL-17-producing CD4⁺ T cells were not markedly increased in diseased Stat3^WT/GOF animals. Rather, onset of autoimmunity in 12-month-old Stat3^WT/GOF mice was characterized by a prominent accumulation of IFN-γ-producing CD4⁺ and CD8⁺ T cells (Fig. 3B–E).

Figure 3: Increased IFN-γ-producing CD4⁺ and CD8⁺ T cells in Stat3^GOF mice:

(A) In vitro differentiation of WT (open circles) and Stat3^WT/GOF (black circles) CD4⁺ T cells under classical Th17 (cTh17), pathogenic Th17 (pTh17), and Th1 conditions. Data combined from two replicate experiments. **, P<0.01; ****, P<0.0001; by one-way ANOVA and Tukey’s multiple comparison test. (B) Flow plots showing splenic CD4⁺ (left) and CD8⁺ (right) T cell IFN-γ and IL-17A production. No. equals % in gate. (C, D) % IFN-γ⁺ (C) and IL-17A⁺ (D) CD4⁺ T cells in spleen and LN of WT and Stat3^WT/GOF mice. (E) % IFN-γ⁺ CD8⁺ T cells. (B-E) **, P<0.01; ***, P<0.001; ****, P<0.0001; NS, not-significant; by one-way ANOVA and Tukey’s multiple comparison test. Each data point equals individual WT (open circle) and Stat3^WT/GOF (solid circle) animal.

Enhanced STAT3 promotes Th17 differentiation during EAE.

Finally, we sought to reconcile the limited Th17 expansion in aged Stat3^WT/GOF mice with our murine T cell differentiation data and with increased Th17 differentiation in human STAT3 GOF syndrome (12). Th17 cells exhibit significant plasticity in vivo, including trans-differentiation into IFN-γ-producing cells during autoimmune inflammation (13, 14). For this reason, we quantified T helper cell differentiation during acute autoimmune inflammation using the EAE model of multiple sclerosis. Although neurologic disease is not a common manifestation of human STAT3 GOF syndrome, we reasoned that EAE could be used to model the impact of the Stat3^GOF allele on T cell biology during autoimmunity. Following MOG immunization, WT and Stat3^WT/GOF mice developed similar disease severity (Fig. 4A). In keeping with reduced Treg differentiation in competitive settings, CNS Foxp3⁺ Tregs were moderately reduced in Stat3^WT/GOF animals, although the intra-CNS Teff:Treg ratio was not altered (Fig. 4B), suggesting that Stat3^GOF does not prevent the differentiation and/or migration of Foxp3⁺ Treg to sites of organ inflammation. Notably, Stat3^WT/GOF animals exhibited an expansion of IL-17A-producing Th17 cells post-MOG immunization, with no increase in Th1 or Th1/Th17 double-positive CD4⁺ T cells (Fig. 4C, D). Thus, despite the relative expansion of IFN-γ-producing Th1 cells in aged animals with spontaneous disease, enhanced STAT3 promotes Th17 differentiation during acute autoimmunity.

Figure 4: Stat3^GOF mutation drives expansion of Th17 cells in CNS during EAE.

(A) EAE clinical score in WT and Stat3^WT/GOF mice. (B) CNS %CD44⁺ Teff (left), Foxp3⁺ Treg (middle), and Teff:Treg ratio (right) (C) Flow plots showing IFN-γ and IL-17A production by CNS CD4⁺ T cells. (D) % IFN-γ⁺, IL-17A⁺, and IFN-γ⁺IL-17A⁺ (double positive; DP) CNS CD4⁺ T cells. (B, D) *, P<0.05; ***, P<0.001; NS, not-significant; by two-tailed Student’s t test (B) and ANOVA plus Šidák’s correction multiple comparison test, with a single pool variance (D). Each data point equals individual WT (open circle) and Stat3^WT/GOF (solid circle) animal.

DISCUSSION:

Targeted animal models carrying human disease-associated mutations are an important strategy in understanding the pathogenesis of Mendelian immune disorders. Here, we generated a new murine strain expressing a human pathogenic STAT3 GOF variant and confirmed that this mutation drives spontaneous, multisystem autoimmunity on the non-autoimmune C57BL/6 background.

STAT3 regulates the balance of Th17 and Treg differentiation. IL-6-dependent STAT3 phosphorylation, in combination with TGF-β, drives Th17 differentiation, while inhibiting FOXP3 expression to theoretically shift the balance from Treg to pathogenic Th17 cells (15). In addition, increased STAT3 activation promotes SOCS3 expression, which inhibits STAT5 signals downstream of the IL-2 receptor (IL-2R). Multiple lines of evidence link STAT5 with normal Treg function, including reduced Treg in STAT5a/b-deficient mice (11), enhanced suppressive function of Tregs expressing constitutively active STAT5b (16), and patients with STAT5b loss-of-function (LOF) mutations exhibiting systemic inflammation reminiscent of individuals with forkhead box protein 3 (FOXP3) mutations (17). The overlap in STAT3 GOF and STAT5b LOF clinical phenotypes extends beyond immune dysregulation to include growth hormone (GH)-resistant short stature (17). Together, these data strongly supported an important role for Treg dysfunction in STAT3 GOF autoimmunity.

Despite this prevailing model, we observed no defect in Treg differentiation or in vitro suppressive activity in Stat3^WT/GOF mice. Rather, progressive STAT3-driven autoimmunity was accompanied by a parallel expansion of activated CD44⁺ Treg and restoration of Teff:Treg balance. An important caveat is that reduced Stat3^GOF Treg differentiation in competitive settings may implicate relative Treg dysfunction as contributing to organ-specific autoimmunity in STAT3 GOF patients. However, while not precluding these more subtle impacts on Treg biology, our findings compliment a recent study from the Anderson group, in which the identical STAT3^K392R allele accelerated the development of type 1 diabetes (T1D) on the non-obese diabetic (NOD) murine background (18). Consistent with our data, Treg development was normal in NOD.Stat3^K392R mice and NOD.Stat3^K392R Treg were able prevent diabetes in transfer models.

Finally, despite the increase in Th17 relative to Th1 differentiation in vitro, a notable feature of the Stat3^GOF model is the prominent accumulation of IFN-γ-producing Th1 cells. These data are strikingly consistent with a similar increase in Th1 cells, but modest Th17 expansion, in NOD.Stat3^K392R animals (18). Importantly, Th17 cells exhibit extensive plasticity in vivo with ex-Th17 cells forming a major source for CD4⁺ IFN-γ production during chronic inflammation. This trans-differentiation into unconventional Th1 cells is driven by STAT3-dependent IL-23 receptor signaling, suggesting a mechanism for Th1 expression during chronic Stat3^GOF inflammation (13, 14). Alternatively, IL-23R signaling can directly drive the differentiation of colitogenic Th1 cells during murine inflammatory bowel disease, in the absence of initial Th17 differentiation (19). Testing these models for Th1 expansion in STAT3 GOF syndrome will require additional IL-17A fate-mapping experiments. In addition, whether additional cellular lineages (such as CD8⁺ T cells or B cells) contribute to autoimmunity in STAT3 GOF syndrome remains to be determined but will be a major focus of future studies using this Cre-driven Stat3^WT/GOF model.

In summary, these findings highlight the complexity of STAT3 regulation of immune tolerance and emphasize the challenges in predicting in vivo biology from ex vivo human assays. The ability to model cytokine- and lineage-specific roles for STAT3 in autoimmunity using this new Stat3^GOF model holds the promise of advancing our understanding of both rare and common autoimmune syndromes.

Key points:

1. Human STAT3 gain-of-function mutation promotes autoimmunity in a murine model.

2. Preserved regulatory T cell development and function in STAT3 GOF mice.

3. Enhanced STAT3 promotes in vivo expansion of Th1 and Th17 CD4⁺ T cells.