STAT6 controls the stability and suppressive function of regulatory T cells

Signal transducer and activator of transcription 6 (STAT6) promotes tumorigenesis by decreasing the Forkhead box P3+ (Foxp3+) cell frequency allowing for the infiltration of inflammatory cells during the early stages of colitis‐associated cancer (CAC). In this study, we dissected the role of STAT6 in the generation of inducible in vitro regulatory T cells (iTregs) and peripheral in vivo Tregs (pTregs) under inflammatory conditions. In in vitro assays, when STAT6 was lacking, iTregs preserved a stable phenotype and expressed high levels of Foxp3 and CD25 during long expansion periods, even in the presence of IL‐6. This effect was associated with increased in vitro suppressive ability, over‐expression of programmed death‐1 (PD‐1), CTLA‐4, and Foxp3, and decreased IFN‐γ expression. Furthermore, iTregs developed during STAT6 deficiency showed a higher demethylation status for the FOXP3 Treg‐specific demethylated region (TSDR), coupled with lower DNA methyltransferase 1 (DNMT1) mRNA expression, suggesting that STAT6 may lead to Foxp3 silencing. Using a mouse model of CAC, the STAT6‐/‐ pTregs expressed a more activated phenotype at the intestine, had higher suppressive capacity, and expressed more significant levels of PD‐1 and latency‐associated peptide of TGF‐β (LAP) associated with their ability to attenuate tumor development. These data suggest that STAT6 signaling impairs the induction, stability, and suppressive capacity of Tregs developed in vitro or in vivo during gut inflammation.


Introduction
Regulatory T cells (Tregs) prevent inappropriate immune responses by suppressing immune effector cells. In addition to controlling immunological tolerance to self-and non-selfregulating immune homeostasis, Tregs suppress inflammation [1]. There are two main types: those produced in the thymus ("natural" Tregs-nTregs or thymic Tregs, tTregs) and those that differentiate from naïve T cells in the periphery (pTregs) or cell cultures ("adaptive or induced" Tregs-iTregs). Tregs express the T cell co-receptor CD4 and the interleukin-2 (IL-2) receptor αchain, CD25; therefore, their phenotype is CD4+CD25+. The specific expression of the transcription factor Forkhead box P3 (Foxp3) allows for the development and function of these cells. Foxp3 is vital to maintaining the suppression of the immune system [2].
In colorectal cancer (CRC), the contribution of tumor infiltrating Tregs to patient prognosis is still not completely defined. A high number of Tregs in CRC stroma has been associated with a better prognosis [3]; however, Tregs infiltration in the colon correlates with lymph node metastasis and increased malignancy [4]. Tregs from CRC patients usually express molecules that correlate with suppression, such as programmed death receptor 1 (PD-1), CD103, Tim-3, LAG-3, TGF-β, IL-10, CD25, and CTLA-4 [5]. CD8+ T cells are inhibited by adenocarcinoma-associated Tregs that regulate PD-1 [6]. However, Tregs modify their phenotype in the tumor microenvironment in a disease-stage-dependent manner [7]. Still, the signals that participate in that process are not entirely understood.
Signal transducer and activator of transcription 6 (STAT6) participates in the cellular response to IL-4 and IL-13, which generate CD4+ Th2 cells. Our previous studies have demonstrated that STAT6 is an essential regulator of Tregs' function during colitis-associated colorectal cancer (CAC) [8]. STAT6 knockout (STAT6-/-) mice are highly resistant to CAC development and present a high number of CD4+Foxp3+CD25+ cells in the colon, circulation, and spleen, including over-expressions of TGF-β, IL-10, and Foxp3, compared with wild-type (WT) mice, during the early stages of CAC development [8]. The increased frequency of Tregs during STAT6 deficiency coincided with minor intestinal damage, decreased cell proliferation in the early stages of injuryinduced tumor formation, and increased apoptosis in advanced tumors [8,9]. We detected an inverse Tregs frequency between WT and STAT6-/-mice as CAC progressed. At an early stage of CAC induction, STAT6−/− mice had twice the number of CD4+Foxp3+CD25+ Tregs in the circulation and the spleen compared with WT [8], suggesting that Tregs expansion is negatively controlled by a STAT6-dependent mechanism, which probably limits the number of Tregs at sites of inflammation. However, whether STAT6 is involved in the process of Foxp3 expression, and Tregs' stability and suppressive function remains unclear.
To further characterize the contribution of STAT6 to Tregs development and stability, in this study, we developed doubled transgenic STAT6-/-Foxp3 EGFP mice. In vitro generated STAT6-/-iTregs expressed high levels of Foxp3 and CD25 and maintained their phenotype and suppressive function for extended periods compared to WT counterparts. The strength of Foxp3 expression was associated with the methylation status of the Foxp3 Tregspecific demethylated region (TSDR) locus. STAT6 deficiency elevated the numbers of in vivo intestinal CD4+Foxp3+PD1+ and CD4+Foxp3+LAP+ cells, maintaining their functional stability and suppressive activity under inflammatory conditions in the early stages of intestinal tumorigenesis.

Regulatory T cell development is not altered in STAT6-/-FOXP3 EGFP mice
We analyzed if the STAT6 mutation affects the development and function of Foxp3 Tregs. We crossed BALB/C knock-in mice co-expressing enhanced green fluorescent protein (EGFP) and the regulatory T cell-specific transcription factor Foxp3 (B6.Cg-Foxp3tm2Tch/J mice, called Foxp3 EGFP mice here) with STAT6-/mutant mice to generate mice homozygous for the STAT6 mutation and the Foxp3 EGFP allele (STAT6 −/-Foxp3 EGFP mice) Fig. 1A and B.
Due to Foxp3 EGFP mice co-expressing EGFP and the regulatory T cell specific transcription factor Foxp3, the Foxp3 EGFP allele accurately recapitulates Foxp3 function, and EGFP expression does not affect the expression of the Foxp3 protein [10]. Consistent with this, FACS analysis showed that WT Foxp3 EGFP and STAT6 −/-Foxp3 EGFP mice had equivalent circulating percentages of CD4+Foxp3 EGFP + cells (Fig. 1C). Only a single Foxp3 allele is active in each female cell due to random X chromosome inactivation, and only a single allele is present in male cells. In accordance, male mice hemizygous for Foxp3 EGFP allele and homozygous Foxp3 EGFP female mice remained as healthy as their wildtype littermates through the period of observation (6 months). The frequencies of CD4+Foxp3 EGFP + cells were similar between male and female STAT6 −/− Foxp3 EGFP mice and their littermate controls (Fig. 1D). Even though the frequency of CD4 T cells was similar between groups (Fig. 1E), we showed that there is a reduced frequency of CD8 T cells in the spleen of STAT6 −/-/Foxp3 EGFP mice vs. WT Foxp3 GFP (7.4 ± 0.2%. vs. 8.9 ± 0.4 p < 0.05) (Fig. 1F). This reduced frequency of CD8 T cells was previously reported and attributed to impaired antigen-presenting cell function [11].
In thymus, the frequencies of CD4 and CD8 T single-positive cells were largely unaltered in STAT6 −/− Foxp3 EGFP mice ( Fig. 1G and H). Most EGFP+ thymocytes found in WT Foxp3 EGFP and STAT6 −/− /Foxp3 GFP mice corresponded to CD4 single-positive cells and represented 0.95 and 0.99% of total thymocytes, respectively. Since Foxp3 has been implicated in the thymic development of CD4+Foxp3+ cells (natural Treg cells), we evaluated this population for a STAT6-deficient background. WT Foxp3 GFP and STAT6 −/− /Foxp3 GFP mice had similar percentages in this population (Fig. 1I). Because the transcription factor Helios is a marker of thymus-derived Tregs [12], we analyzed thymocytes for the Figure 1. Generation and genetic characterization of STAT6-/-Foxp3 EGFP mice. The double transgenic mice were obtained by mating Foxp3EGFP mice with STAT6-/-animals. Foxp3EGFP mice were crossed to STAT6-/-mice to generate homozygous mice for the STAT6-/-mutation and the Foxp3EGFP allele. (A) Representative genotyping result for STAT6-/-mice with PCR. STAT6-/-homozygous mice (lines I, III, and IV showed the presence of a neomycin-resisting gene, 380 bp). Heterozygous mice (line II). Genomic DNA (gDNA) was used as a template from BALB/c mice (lane VI shows the absence of transgenes, 275 bp). (B) Characterization of mouse samples with PCR analysis to detect enhanced green fluorescent protein (EGFP) expression. Control mice (lanes I and II) and Foxp3EGFP mice (lines III and IV). (C) Cells from WT Foxp3EGFP and STAT6-/-Foxp3EGFP mice were obtained, incubated with anti-CD4 and anti-CD8 antibodies, and analyzed by FACS. The lymphocyte region was first identified and gated according to its forward scatter (FSC) and side scatter (SSC) characteristics, and the CD4+ region was subgated and captured 10 000 CD4+ cells. Representative results of Treg-cell detection; percentages of CD4+Foxp3EGFP+ cells in the CD4+ gate are displayed. (D) Analysis of CD4+Foxp3EGFP+ T cell subsets in females and males STAT6-/-Foxp3EGFP mice. The percentage of splenic CD4+ (E) and CD8+ (F) T cells in WT Foxp3EGFP and STAT6-/-/Foxp3EGFP mice. Analysis of thymic CD4 single-positive (G), CD8 single-positive (H), and CD4+Foxp3+ (I) T cell subsets in the thymus of the indicated genotype. (J) Representative flow cytometry plots (left) and bar graphs (right) show the percentage of thymic CD4+Foxp3+Helios+ T cells for the indicated group. Bars represent the mean from 3 to 8 spleens (E, F) or thymus (G-J), and each data point represents an individual mouse. Statistical significance was determined by unpaired t-test, * p < 0.05. expression of Helios in CD4+Foxp3+ cells. Double-reporter mice exhibited a similar frequency of Helios+Foxp3+ cells (Fig. 1J). Our results show no differences in tTregs or nTreg development during STAT6 deficiency.
In order to delineate the functionality of iTregs during STAT6 deficiency, we analyzed the expression of PD-1 and cytokines associated with activation in Tregs. Tregs are characterized by the expression of PD-1 and are significant producers of the suppressive cytokines IL-10 and TGF-β. PD-1 prompts Foxp3 expression and enhanced Treg suppressive activity [13,14] In the absence of STAT6, iTregs from day 8 of culture significantly increased PD-1 expression than their WT counterparts ( Fig. 2 E), in accordance, STAT6-/-Foxp3 EGFP cells showed a significant decrease in IFN-γ expression ( Fig. 2 H), suggesting a more suppressive phenotype. We also analyzed the percentage of CD4+Foxp3+ cells expressing latency-associated peptide (LAP) (Fig. 2F). TGF-β is first generated as pro-TGF-β. The expression of pro-TGF-β can be determined by the N-terminal portion of the pro-TGF-β, the LAP [15]. Despite the increase in PD-1 expression on STAT6-/-Foxp3 EGFP cells compared to that in WT Foxp3 EGFP cells, the expression of LAP decreased in STAT6-/-Foxp3 EGFP cells (Fig. 2F), suggesting that STAT6 is critical for pro-TGF-β expression in iTregs. In addition, no differences in IL-10 expression were observed (Fig. 2G).
iTregs were cultured with the pro-inflammatory cytokine IL-6 for 3 days to analyze if iTregs generated during STAT6 absence are stable even during inflammatory conditions. The frequencies of CD4+Foxp3+CD25+ cells were maintained in iTregs developed under STAT6 deficiency when compared with control iTregs (without cytokines) (83 ± 3.9 vs. 72 ± 5.3 %, NS) ( Fig. 2I and J). On the contrary, the percentage of CD4+Foxp3+CD25+ cells dropped significantly in iTregs from Foxp3 EGFP mice in the presence of IL-6 (70 ± 5.1 vs. 49 ± 7.6%, p < 0.05) ( Fig. 2I and J). These data suggest that the absence of STAT6 impacts iTregs differentiation, maintaining Foxp3 and CD25 expression for a long time and promoting overexpression of PD-1 while IFN-γ expression is decreased.

STAT6 deficiency results in a reduced pro-inflammatory cytokine profile during iTreg expansion
Due to our results that iTregs induced under STAT6 deficiency expressed higher levels of Foxp3 and CD25 for a longer time, even during inflammatory conditions than their WT counterparts, we decided to evaluate cytokine production during different days of iTreg cell induction (day 5) and expansion (days 8 and 15) ( Fig. 3A-D). We detected the early production of inflammatory cytokines such as IL-17 A, TNF-α, and IL-6, in the supernatants of day 5 of cultures of WT Foxp3 EGFP cells but not in STAT6-/-Foxp3 EGFP cells ( Fig. 3A-D). Interestingly, during Tregs expansion (day 15), the secretion of IL-4 increases in cultures from WT Foxp3 EGFP cells (Fig. 3E), probably by Tregs fragility. During the expansion, STAT6-/-Foxp3 EGFP cells increased the mRNA relative expression of Foxp3 and CTLA-4 in contrast to WT Foxp3 EGFP cells, both molecules critically required for the function of Tregs in vivo ( Fig. 3F and G). Producing pro-inflammatory cytokines such as IL-6 likely inhibits the optimal induction of Foxp3 expression. We hypothesized that WT Foxp3 EGFP cells lose stability due   to increased STAT6 signaling. In order to assess this possibility, we determined the phosphorylation of STAT6 on iTregs from different times of culture (Fig. 3H). Phosphorylated STAT6 was restricted to WT Foxp3 EGFP cells during day 15 of expansion; in contrast, its level was absent during induction ( Fig. 3H). This result suggests that STAT6 is activated during long expansion, impairing Tregs stability.

iTregs expanded in STAT6 deficiency show increased suppression on responder T cells proliferation
Given that the expression of Foxp3, CD25, and PD-1 was increased in iTregs from STAT6-/-Foxp3 EGFP versus WT Foxp3 EGFP cells at 8 days of expansion, we decided to evaluate if iTregs developed under these conditions may show different suppressive capacity. We assessed the suppressive ability of iTregs sorted from cultures over naïve splenocytes stimulated with α-CD3, which were exposed to different ratios of iTreg cells. When total splenocytes were stimulated with α-CD3, both CD4+ and CD8+ T cells strongly proliferated as expected ( Fig. 4A and B). When Tregs came from STAT6-/-Foxp3 EGFP culture, we observed an increased suppressive capacity over CD4+ and CD8+ cells, with a more remarkable effect on CD8+ T cells, compared with wild-type Tregs ( Fig. 4A-D). Taken together, these data suggest that STAT6 has a negative impact on the suppressive function of WT Foxp3 EGFP iTregs.

Demethylation of TSDR FOXP3 gene is higher in iTregs expanded in the absence of STAT6
Complete demethylation within the noncoding TSDR of the FOXP3 gene has been demonstrated in mouse and human nTregs and is an essential element regulating Foxp3 expression stability [16,17]. As iTregs from STAT6-/-Foxp3 EGFP mice displayed prolonged stability and suppressive phenotype, we analyzed the methylation status of the TSDR of naïve CD4+ T cells, thymic nTregs and iTregs at day 8 of expansion for Foxp3 EGFP and STAT6-/-Foxp3 EGFP genotypes ( Fig. 5A and B), (Supporting Information S. 1). Our data show a lower percentage of methylated CpGs (of a total of 9) in the TSDR for six independent cultures of iTregs in the absence of STAT6 (46%) compared to both WT iTregs (66%) and naïve CD4+ T cells (93%) (Fig. 5A and B). This translates into a significantly higher demethylation status of TSDR in STAT6-/-Foxp3 EGFP iTregs (54%) compared to both other groups (34% and 7% for WT iTregs and naïve CD4+ cells respectively) (Fig. 5B), which could be related to their enhanced stability and suppressive functions. Furthermore, there was a significant decrease for DNA methyltransferase 1 (DNMT1) mRNA expression for the STAT6-/-Foxp3 EGFP iTregs (38%), which could contribute to the observed lower TSDR methylation status (Fig. 5C).
Given that nTregs are characterized by complete demethylation of the TSDR region and stable expression of Foxp3 [18], we sought to evaluate if the absence of STAT6 could modify the methylation status of the TSDR in nTregs. Thymic Foxp3 EGFP and STAT6-/-Foxp3 EGFP cells had an almost complete demethylation status of the TSDR (Fig. 5A and B), showing very low average TSDR methylation values (11.1 ± 4.5% vs. 2.8 ± 2.8%, NS), demonstrating that STAT6 did not impact the nTreg development.

STAT6 impairs the optimal Tregs response during in vivo inflammation
Previously, we determined that STAT6 deficiency induces resistance to tumorigenesis, arresting tumor-promoting inflammation in CAC through an azoxymethane (AOM)/dextran sulphate sodium salt (DSS) regiment [8]. One potential mechanism contributing to protection against cancer in STAT6-/-mice is the increased recruitment of CD4+Foxp3+CD25+ regulatory cells in the colon, circulation, and spleen compared to WT mice during the early stages of CAC development. Since inflammation is the primary driver of tumor initiation in CAC, increasing suppressive and stability of Tregs during STAT6 deficiency in vivo may decrease the proliferation of inflammatory cells. To test this hypothesis, we subjected WT Foxp3 EGFP or STAT6-/-Foxp3 EGFP mice to an AOM/DSS regimen and analyzed the CAC progression at day 20 (early stage) and day 72 (late stage of tumor development, where adenoma-like lesions are observed) as an approximation of different stages of tumor progression. As previously reported [8], the WT Foxp3 EGFP AOM mice displayed both increased numbers of tumors as well as increased tumor load at day 72 (10 ± 1), whereas STAT6-/-Foxp3 EGFP AOM animals developed very scarce tumors (2 ± 1) (Supporting Information S. 2). Thus, STAT6-/-Foxp3 EGFP mice are as resistant to developing CAC as regular STAT6-/-mice. We next examined the frequency of Tregs in the spleen of WT Foxp3 EGFP AOM and STAT6-/-Foxp3 EGFP AOM mice. We found an increased Tregs frequency in the spleen of STAT6-/-Foxp3 EGFP AOM animals at day 20 (early stage of tumor development) compared to the control and WT Foxp3 EGFP AOM mice (7.1± 0.3 vs. 5± 0.3, p < 0.01) ( Fig. 6A and B). To determine whether the Tregs from the STAT6-/-Foxp3 EGFP AOM animals displayed different markers of suppression compared with those expressed by WT Foxp3 EGFP AOM mice, we analyzed the expression of PD-1 and CD103. As shown in Fig. 6C and D, we observed a higher expression of PD-1 on CD4+Foxp3+ cells in the spleens of STAT6-/-Foxp3 EGFP AOM animals. However, we did not observe differences in CD103 expression in the same population ( Fig. 6E and F). Given that a proportion of Foxp3+ Tregs could be generated extrathymically from T-conventional cells, we decided to analyze if STAT6 deficiency negatively regulates expansion of natural Tregs. As Helios is a marker of thymic-derived Treg (tTreg) [19], we evaluated its expression in splenic CD4+Foxp3+ cells from WT Foxp3 EGFP AOM and STAT6-/-Foxp3 EGFP AOM mice. A significant increase in CD4+Foxp3+Helios+ was observed in STAT6-/-Foxp3 EGFP mice compared to WT Foxp3 EGFP mice ( Fig. 6G and H), suggesting that STAT6 may participate in natural/thymic Tregs recruitment during homeostatic and inflammatory conditions.
Given that the increase in Tregs during STAT6 deficiency in the early stages of CAC induction has been related to decrease mucosal inflammation, reduced histological damage, and delayed tumorigenesis, we decided to evaluate the suppressive capacity of these cells. To assess whether Tregs developed in WT Foxp3 EGFP AOM and STAT6-/-Foxp3 EGFP AOM mice during the early stages of tumor development can induce suppression, we sorted Foxp3+ cells from spleens at day 20 of CAC induction. Tregs were cultured in different ratios with naïve splenocytes stimulated with α-CD3. Interestingly, when Tregs came from STAT6-/-Foxp3 EGFP AOM mice and the Treg/T responder cells (Tresp) ratio was diluted, we observed a potent inhibition of the proliferative capacity of CD4+ ( Fig. 6I and J) and CD8+T (Fig. 6K and L) cells at a ratio of 1:14. These results confirm that STAT6 deficiency increases the stability and suppressive function of in vitro iTregs and in vivo Tregs under inflammatory conditions, indicating that STAT6 lack allows a more robust Foxp3+ Tregs response.

STAT6 impacts intestinal Tregs stability during in vivo inflammation
Deficiency in Tregs stability, function, and generation in the gut causes inflammation and increases the risk of CAC [20]. Intestinal Tregs can produce high levels of inhibitory cytokines to suppress intestinal immune responses and maintain intestinal homeostasis [20]. To determine the effect of STAT6 deficiency on intestinal Tregs generation and stability during carcinogenesis in our model, we examined the frequency of Tregs in the colon of WT Foxp3 EGFP AOM and STAT6-/-Foxp3 EGFP AOM mice during the early stage of tumor development (Fig. 7A and B). We did not find differences in the percentage of CD4+Foxp3+CD25+ Tregs in the gut between both experimental groups. However, Tregs from STAT6-/-Foxp3 EGFP AOM animals showed an increased expression in PD-1 (25 ± 0.3 vs. 19 ± 0.6 %, p < 0.01) and LAP (24 ± 0.4 vs 16 ± 2.2 %, p < 0.01) compared to the WT Foxp3 EGFP AOM mice (Fig. 7C-F). Even though we did not observe differences in IL-10 cytokine expression ( Fig. 7I and J), the expression of IFN-γ in CD4+Foxp3+ cells from STAT6-/-Foxp3 EGFP AOM mice was significantly decreased, suggesting a stable Tregs phenotype ( Fig. 7G and H). Our results demonstrated that deletion of STAT6 during in vivo inflammation successfully induces a phenotype of more suppressive intestinal Tregs.

Discussion
In this study, we examined the role of STAT6 in Tregs expansion, survival, and suppressive function during iTreg development and an in vivo inflammatory scenario. The lack of constitutive STAT6 signaling in CD4 naïve T cells improves the stability and suppressive potential of iTregs. The biological significance of this finding lies in the scenario of CAC, where a significant increase in suppressive and activated Tregs occurs during the early stages of tumor development, controlling the inflammation [8].
In the present report, we generated a STAT6-/-Foxp3 EGFP double transgenic mouse which readily allowed the isolation of naive T cells and followed Tregs recruitment to the intestine. Lack of STAT6 does not affect the differentiation of thymic cell populations like CD8+ and CD4+ single positive cells, Tregs, and Helios+ Tregs during homeostatic conditions. However, splenic CD8+ cells were slightly reduced in STAT6-/-Foxp3 EGFP mice. This reduced frequency of CD8 T cells was previously reported and attributed to impaired antigen-presenting cell function [11]. The lack of STAT6 signaling in STAT6-KO mice might delay DC maturation, decreasing antigen-cross presentation to prime CD8 T cells [21].
In vitro, Tregs differentiation requires IL-2 and TGF-β; however, iTregs are unstable and rapidly lose Foxp3 expression [17]. Our results showed that STAT6 deficiency in naïve CD4+ T cells promotes the permanence of Foxp3 expression and suppressive function after iTreg differentiation for up to 15 days of expansion. The demethylation of a conserved region rich in CpG within the first intron of the FOXP3 locus, called TSDR, determines the stability and maintenance of Tregs [16]. A silencing region in the Foxp3 transcript, with a specific binding site for STAT6, preventing Foxp3 mRNA expression has been identified [22]. Recent reports have suggested that specific modifications of DNA and histones are required to differentiate Tregs. Cui et al. [23] reported that histone deacetylases 9 (HDAC9) repressed Foxp3 transcription by decreasing chromatin accessibility in a STAT6-dependent manner. Interestingly, HDAC are transcriptional repressors recruited by methylated CpG [24]. STAT6 has also been reported as a global repressor of gene expression during T helper cell differentiation by recruiting histone methyltransferases and promoting repressive marks such as Histone 3 lysine 4 methylation (H3K4me3) and H3K27me3 [25]. We found that in the absence of STAT6, iTregs display higher partial demethylation of the TSDR Foxp3 gene and expressed higher levels of Foxp3 mRNA. Furthermore, we found that STAT6 deficiency caused an impaired DNMT1 expression in iTregs. DNMT1 binds to and modulates the methylation status of the promoter [26] and intronic [27] regulatory elements of the FOXP3 gene in naïve CD4+ cells (but not in stable nTregs). On the other hand, TGF-β can induce Foxp3 expression through a reduction in DNMT1 activity [27]. This could explain the observed prolonged stability of Foxp3 expression in STAT6-/-Foxp3 EGFP iTregs during expansion in the absence of TGF-β, whose removal triggers iTreg instability [28]. Thymic nTreg development is not affected by STAT6 deficiency as we did not find significant differences in nTreg proportions nor TSDR methylation state. iTregs and nTregs follow differential development pathways: while iTregs are unstable and show only partial demethylation at the TSDR, nTregs display a more stable profile, and their development depends on extensive demethylation at the promoter and TSDR regions of the FOXP3 gene [29].
Our results show that STAT6 appears responsible for the impairment of iTreg stability during cell culture. Nevertheless, how could STAT6 be activated during in vitro Treg expansion? We detected STAT6 phosphorylation during iTreg long-time expansion due to a concurrent increase of IL-4 in the supernatants of Foxp3 EGFP iTregs up to day 15. One source of IL-4 that builds up during iTreg expansion could be the Tact population (CD4+CD25+Foxp3-), which is increased and sustained over time only in WT Foxp3 EGFP cultures. These results agree with evidence that IL-4 strongly inhibits Foxp3 expression and TGF-β-mediated Treg differentiation from naive CD4+ T cells cultured under iTregpolarizing conditions ( [23,30]).
There are several possibilities as to what mechanisms are involved in the increased suppressive capacity of STAT6-/-Tregs. IL-10 secretion is a paramount mechanism for Tregs suppression. However, in our study, STAT6+/+Tregs and STAT6-/-Tregs showed similar IL-10 production; therefore, it is unlikely that IL-10 is involved in this regard. Unstable Tregs are characterized by functional fragility and loss of suppressive activity. IFN-γ exogenously provided or produced by WT iTregs could lead to Tregs fragility inducing loss of suppressive activity in vitro and limiting Tregs expansion [31].
On the other hand, the small number of Foxp3-T cells that might be present throughout the induction could be responsible for the higher levels of inflammatory cytokines (IL-6, TNF, and IL-17A), leading to a more fragile Tregs phenotype in STAT6+/+ cells. In addition, under STAT6 deficiency, we observed an increased expression of PD-1. PD-1 is a co-inhibitory receptor expressed on Tregs and functions by inhibiting the co-stimulation of T cells after binding to PD ligand 1 (PD-L1) [13]. PD-1 is a critical homeostatic regulator for Tregs. STAT6-/-iTregs maintain stability in the expression of Foxp3 and the functional properties of Tregs for a long time, even during culture with an inflammatory cytokine such as IL-6. PD-1 expression by STAT6-/-iTregs, may provide inhibitory signals affecting selectively the production of cytokines and causing cell cycle blockade.
Bacterial infiltration in the intestine with concomitant mucosal inflammation facilitates the development of intestinal adenomas [32]. During CAC development, Tregs develop and coexist with Th cells that mediate immune responses to foreign or self-antigens under inflammatory conditions. In a murine model of sporadic cancer, we previously showed that STAT6 deficiency increases the accumulation of Tregs in the colon in early CAC, which was associated with a remarkable reduction in tumor growth. Increased mucosal inflammation, histological damage, and tumorigenesis were restored to levels observed in WT mice when an early inhibition/depletion of Treg cells was performed in STAT6-/-mice [8]. We hypothesized that with STAT6 deficiency, increased Tregs populations induce resistance against tumor-promoting inflammation, and tumorigenesis is therefore stopped. Therefore, in this publication, we focus on analyzing Tregs only during the early stages of tumor development, where stability and suppressive function could be important for containing inflammation.
Our results show that intestinal and peripheral CD4+Foxp3+ cells developed during the early stages of CAC (but also during in vitro induction) under STAT6 deficiency are more suppressive and highly expressed PD-1. Evidence shows that in vivo, PD-1 expression prevents the conversion of Tregs into pro-inflammatory effector memory T cells, making them more resistant to apoptosis [33,34]. In a model of colitis induced by the transfer of Th17 cells, co-transfer of IL-10-producing C-C chemokine receptor type 5 (CCR5)+PD-1+ type 1 Tregs strongly inhibits colitis, whereas IL-10-producing control T cells lacking CCR5 and PD-1 are less efficient [35]. In our model, STAT6 deficiency leads to an improved expression of suppressive markers such as LAP (pro-TGF-β) and PD-1 but also lower production of IFN-γ by colonic Tregs, which is considered a hallmark of fragile Tregs phenotype and would limit Tregs ability to control inflammation [31].
Tregs are broadly considered harmful in the context of cancer; however, we have shown that they could be helpful if they are more suppressive during the initial stages of inflammationinduced tumorigenesis, particularly in a STAT6-/-background. The suppression studies suggest that in the context of active inflammation during CAC, Tregs developed under STAT6 deficiency build a high capacity to suppress CD4+ and CD8+ T cells. This fact would likely result in a net beneficial effect. However, this report has not evaluated the effects of Tregs STAT6 deficiency in late-stage disease. Previous studies demonstrated that the accumulation of Foxp3+ cells in patients with CAC is related to a better prognosis [36], so the inhibition of STAT6 signaling could enhance Tregs activity and suppressive function and contribute to improved survival. In addition, we found that in the in vivo AOM/DSS CAC model, the inhibition of STAT6 phosphorylation by AS1517499 during late stages of tumorigenesis induced a 29% reduction in tumor growth compared to untreated CAC mice [37]. Nevertheless, IL-10, IL-17, and TGF-β expression, cytokines related to Tregs activity were significantly reduced in CAC mice treated with STAT6 inhibitors. However, in this kind of experiment, AS1517499 could inhibit STAT6 signaling in multiple cells, including tumor cells, so determining its effect on Tregs was impossible.
Other questions remain to be explored regarding the STAT6 functions in Tregs development. For example, why does LAP decrease in iTregs but increase in intestinal Tregs during the absence of STAT6? What happens at the intestinal level? TGFβ1 is secreted as an inactive precursor molecule combined with LAP and needs to be bound by integrins, such as αvβ8 and αvβ6 [38]. Importantly, the expression of integrin β8 is highly upregulated in intestinal CD103+ DCs [39], and consequently CD103+ DCs deficient in αvβ8 have a compromised capacity for Treg induction in the mesenteric lymph nodes [40]. Additionally, intestinal WT Foxp3 EGFP Tregs may have a different origin. We have shown a higher proportion of CD4+CD25+Helios+ cells in the spleen of STAT6-deficient mice, suggesting a differential nTreg recruitment and nTreg/pTreg balance in the periphery for both groups.
Our work demonstrates the central role of STAT6 in controlling the expression of the Tregs master regulator Foxp3. These results translate into excellent stability in culture and increased suppressive function even under inflammatory conditions of iTregs developed under STAT6 loss and suggest the need to evaluate these cells in therapy for CAC and other inflammation-driven chronic diseases.

Mice
A total of 8-to 10-week-old BALB/c (WT) mice were bred in our animal house and kept in cages according to our institutional guidelines. Foxp3 EGFP knock-in mice (B6.Cg-Foxp3tm2Tch/J), coexpressing Foxp3 and enhanced GFP under the same promoter, were donated by Rafael Saavedra and maintained under the same conditions in our animal house. All experiments were carried out on age and sex-matched animals. Animal experimentation protocols were approved by the Facultad de Estudios Superiores Iztacala (FES-I) Bioethics Committee for Animal Research.

Generation of STAT6-/-FOXP3 EGFP mice
Screening of mice for the presence of a neomycin-resistance gene insertion in STAT6 -/-was performed by PCR using the following oligonucleotides: Mutant, 5´AAT CCA TCT TGT TCA ATG GCC GAT C 3´; Common, 5´ACT CCG GAA AGC CTC ATC TT 3´; Wild Type, 5´AAG TGG GTC CCC TTC ACT CT 3´. DNA-PCR analysis for genotyping Foxp3 EGFP animals was performed using the following oligonucleotides: 5´CAC CTA TGC CAC CCT TAT CC 3á nd 5´ATT GTG GGT CAA GGG GAA G 3´. The primer pair specific for the Foxp3 gene amplifies a 275-bp fragment, and the primer pair specific for the Foxp3 EGFP allele (FoxP3 and EGFP) amplifies a 325-bp fragment.

Surface and intracellular staining and flow cytometry
Immunofluorescence labeling was performed by incubating cells with allophycocyanin (APC) anti-mouse CD4, Brilliant violet (BV) 605 TM anti-mouse CD8, and BV 711 TM anti-mouse CD25 anti-bodies (BioLegend, San Diego, CA, USA) diluted in FACS buffer (Dulbecco's phosphate-buffered saline (DPBS), 1% fetal bovine serum (FBS) and 0.1% sodium azide) at 4°C for 30 min. Foxp3+ cells were detected directly by EGFP fluorescence (which is coexpressed under the FOXP3 promoter in Foxp3 EGFP and STAT6-/-Foxp3 EGFP strains). For the intracellular staining (anti-mouse Helios, BioLegend), cells were previously fixed and permeabilized with the True-Nuclear TM Transcription Factor Buffer Set (BioLegend) according to the manufacturer's instructions. Immunostaining was followed by washing and resuspension in DPBS. Cells were then analyzed on Attune TM NxT (ThermoFischer, Waltham, MA, USA) cytometer. Dead cells were excluded by 7aminoactinomycin D (7-AAD) staining (ThermoFischer). A minimum of 10 000 living cells were gated and analyzed unless stated otherwise. Supporting Information Fig. 3 shows the gating strategy. The flow data analysis was carried out using the FlowJo TM v10 software (Tree Star, Inc., Ashland, OR, USA).

Analysis of iTreg stability in the presence of IL-6 cytokine
iTregs generated from naïve CD4+ cells from FoxP3EGFP and STAT6-/-FoxP3EGFP mice at day 5 of differentiation were gated on living cells (negative for 7-AAD (ThermoFischer) fluorescence) and sorted in a FACS Aria TM (ThermoFischer) according to positive EGFP fluorescence signal. Isolated iTregs (>95%) were expanded for 3 days in RPMI complete in the presence or absence of rhIL-6 (30 ng/mL). Cells were then washed and analyzed by flow cytometry, and the proportions of CD4+CD25+Foxp3+ (Tregs) cells were obtained for each group.

In vitro suppression assays
Total splenocytes from Balb/c mice (Tresp) were isolated as described above, incubated with CellTrace TM Violet (ThermoFischer) for 20 min according to the manufacturer's instructions, and seeded on a 96-well plate pre-treated with 5 μg/mL antimouse CD3 (BioLegend). In vitro iTregs at day 8 of expansion and in vivo pTregs from total splenocytes of wild-type Foxp3 EGFP or STAT6-/-Foxp3 EGFP mice were sorted in a FACS Aria TM (Ther-moFischer) according to positive EGFP fluorescence signal. Isolated Tregs (>95%) were then seeded in wells harboring Tresp at different Treg: Tresp proportions for a total of 100 000 cells per well. Cells were cultured for 3 days at 37°C, 5%CO 2 in RPMI complete. Determination of proliferation was performed by flow cytometry. Total cells were gated based on 7-AAD (ThermoFischer) exclusion (living cells) and then sub-gated based on CD4 or CD8 expression. Peaks for CellTrace TM violet staining were classified in non-divided or divided cells, and percentages were obtained. CD4+ Tregs were excluded from the analysis based on the absence of CellTrace TM violet staining Supporting Information S.3.

RNA extraction and quantitative RT-PCR
Total RNA was extracted from 1 × 10 6 iTreg cells at day 8 of expansion with the All-In-One DNA/RNA/Protein Miniprep Kit (BioBasic, Markham, ON, Canada) according to the manufacturer's instructions. Total RNA was analyzed and quantified based on 260 and 280 nm absorbance readings (A260/A280> 1.8) in an Inplem TM nanophotometer (Fischer Scientific; Hampton, NH, USA). A total of 50-100 ng total RNA were reverse transcribed with the M-MLV reverse transcriptase and oligo(dT)18 (Ther-moFischer) to generate first-strand Complementary DNA (cDNA) according to the manufacturer's instructions. A total of 1 μL cDNA was then used for real-time PCR with the qPCR Master Mix with SYBR Green (GoldBio, St Louis, MO, USA) and the gene-specific oligonucleotides listed on Supporting Information Table 1 in a CFX 96-well one-touch real-time PCR system TM (Bio-Rad, Hercules, CA, USA). mRNA expression values were obtained according to the 2 − CT method using 18S as a housekeeping gene. Obtained values were further normalized as the fraction of the average expression in the wild-type Foxp3 EGFP group for each evaluated gene.

FOXP3 TSDR DNA methylation analysis
Total genomic DNA (gDNA) was isolated from 1 × 10 6 iTreg cells at day 8 of expansion, CD4+ naïve T cells, or nTregs with the DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany). nTregs were derived from Foxp3 EGFP or STAT6-/-Foxp3 EGFP CD4+ thymocytes previously isolated with the CD4+T cell isolation kit (Miltenyi Biotec), and sorted by FACS according to positive EGFP fluorescence. A total of 250-1500 ng gDNA were bisulfite-treated with the EpiTect Bisulfite kit (Qiagen) according to the manufacturer's instructions. Control-untreated and bisulfite-treated gDNA were analyzed and quantified based on 260 and 280 nm absorbance readings (A260/A280 > 1. 8) in an Inplem TM nanophotometer (Fischer Scientific). A total of 25-50 ng of bisulfite-treated gDNA was used for qPCR in a LightCycler 480 (Roche) with the SYBR green Master Mix (Amplicon) and bisulfite-treated specific oligonucleotides for the TSDR of the murine FOXP3 gene [41] (Supporting Information Table S1). qPCR protocol was as follows: 94°C, 3 min, 94°C, 15 s, 55°C, 30 s 40×, 72°C, 30 s high-resolution melting from 60°C to 94°C (melting curve). qPCR products were analyzed by agarose gel electrophoresis to corroborate the expected size (291 bp) and were then purified with the MinElute PCR Purification Kit (Qiagen) if no unspecific bands were observed, otherwise specific size DNA band was excised from the gel and DNA was purified with the PureLink TM Quick Gel Extraction Kit (ThermoFischer). Isolated PCR products were then Sanger-sequenced with forward and reverse PCR-specific oligonucleotides. Sequences were then aligned and analyzed for transformed (C→T, demethylated) and untransformed (methylated) CpG sites using the BISMA (Bisulfite Sequencing DNA Methylation Analysis) web platform suite [42].

Colitis-associated cancer induction
Mice were administered an intraperitoneal (i.p.) injection of AOM (12.5 mg/kg) (Sigma, USA). After 5 days, 2% DSS (molecular weight, MW: 40 000, Alfa Aesar, Canada) was added to drinking water for 7 days. Mice were then left to rest with normal drinking water for 14 days. After that, mice were subjected to two more cycles of DSS administration for 7 days, followed by 14 resting days after each cycle. The mice were sacrificed on day 20 (early stage) and day 72 (late stage) after AOM injection to evaluate tumor development during CAC. Mice were evaluated twice weekly for body weight, stool consistency, rectal bleeding, and blood in the stool. After sacrifice, the colon was removed, weighed, and inspected macroscopically for tumors.

Isolation of intestinal lymphocytes
Lamina propria lymphocytes (LPLs) were isolated by collagenase digestion and a Percoll gradient [43]. Briefly, control and AOM/DSS-treated mice (day 20) were sacrificed, and the colon was dissected and washed with saline solution to remove feces. Colon was cut longitudinally and washed with Hank's balanced salt solution (HBSS; Biowest, Nuaille, France) supplemented with antibiotic (penicillin/streptomycin 100×; Cytiva, Marlborough, MA, USA), cut into 2-cm pieces and digested in complete RPMI supplemented with collagenase (2 mg/mL) and DNAse I (40 μg/mL) (Sigma-Aldrich, St Louis, MO, USA) for 2 h at 37°C, 250 rpm. The digested cell suspension was passed through 100 μm and 40 μm filters to remove debris and centrifuged (300 × g, 10 min, 4°C). Cells were washed twice and resuspended in 5 mL Percoll (30%). The cell suspension was layered on top of 8 mL Percoll (70%), followed by centrifugation without brake at room temperature (400 × g, 30 min). The LPLs located as an interphase between 30% and 70% Percoll were aspired and washed twice with RPMI for immunostaining and flow cytometry.

Statistical analysis
Data analysis was performed by one-way or two-way analysis of variance followed by Bonferroni's multiple comparisons tests using GraphPad Prism 8 (San Diego, CA, USA). In the case of only two-group comparisons, unpaired two-tailed t-tests were done. The data are expressed as the mean ± standard error (SEM). * p < 0.05, ** p < 0.01, *** p < 0.001.

Conflicts of interest:
The authors declare no financial or commercial conflict of interest.
Ethics approval: All mice were maintained and bred in a specific pathogen-free animal facility and maintained in a pathogenfree environment at the Facultad de Estudios Superiores Iztacala Data availability statement: The datasets used and analyzed during the current study are available from the corresponding author upon reasonable request.