microRNA-142 guards against autoimmunity by controlling Treg cell homeostasis and function

Regulatory T (Treg) cells are critical in preventing aberrant immune responses. Posttranscriptional control of gene expression by microRNA (miRNA) has recently emerged as an essential genetic element for Treg cell function. Here, we report that mice with Treg cell–specific ablation of miR-142 (hereafter Foxp3CremiR-142fl/fl mice) developed a fatal systemic autoimmune disorder due to a breakdown in peripheral T-cell tolerance. Foxp3CremiR-142fl/fl mice displayed a significant decrease in the abundance and suppressive capacity of Treg cells. Expression profiling of miR-142–deficient Treg cells revealed an up-regulation of multiple genes in the interferon gamma (IFNγ) signaling network. We identified several of these IFNγ-associated genes as direct miR-142-3p targets and observed excessive IFNγ production and signaling in miR-142–deficient Treg cells. Ifng ablation rescued the Treg cell homeostatic defect and alleviated development of autoimmunity in Foxp3CremiR-142fl/fl mice. Thus, our findings implicate miR-142 as an indispensable regulator of Treg cell homeostasis that exerts its function by attenuating IFNγ responses.


Introduction
Regulatory T (T reg ) cells are vital in maintaining immune self-tolerance and restraining aberrant immune responses against infections [1,2]. Foxp3, an X chromosome-linked member of the forkhead box/winged helix family of transcription factors, is a master regulator of the genetic program that governs development and suppressive activity of T reg cells. Humans and mice that carry loss-of-function Foxp3 mutations develop a fatal autoimmune disease due to impaired T reg cell activity [3][4][5][6][7]. The majority of T reg cells are generated in the thymus (tT reg cells) through a selection process that favors cells with a strong functional T cell receptor a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 (TCR) avidity toward self-antigens. In contrast, peripheral T reg (pT reg ) cells arise from naive CD4 + T cells upon encounter of non-self-antigens in the context of appropriate cytokine stimulation [8][9][10]. Harnessing the power of T reg cells to control immunological responses has a great potential for human therapy because, on one hand, T reg cells can promote transplantation tolerance, but on the other, can hinder antitumor immunity.
Posttranscriptional regulation of gene expression by microRNA (miRNA), a class of small (approximately 22 nucleotides) noncoding RNA, recently emerged as a critical genetic element that is essential for T reg cell function. T reg cell-specific knockouts (KOs) of either Drosha or Dicer genes, encoding 2 endonucleases required for mature miRNA generation from precursor transcripts, phenocopy mice with Foxp3 ablation, and develop severe systemic autoimmunity because of a defect in T reg cell activity [11][12][13]. Furthermore, deletion of Dicer at the double positive (DP) thymocyte stage in mice significantly diminished the frequency of tT reg cells, suggesting that miRNA-dependent gene control is also required for normal development of T reg cells [11]. Thus, the current pressing challenge in the field is to determine how specific miRNA gene(s) exert control of the T reg cell genetic program. The present report addresses this goal by examining the role of miR-142 in T reg cell development and function.
miR-142 is predominantly expressed in cells of hematopoietic origin and encodes 2 abundant mature miRNA molecules-miR-142-5p and miR-142-3p-which arise from the opposite strands of the hairpin-like miR-142 precursor. Using genetic loss-of-function studies, miR-142 was previously implicated in the regulation of ontogenesis and function of several immune cell types. Our earlier report determined that deletion of this miRNA gene in mice results in aberrant B lymphopoiesis and impaired humoral immunity [14]. In addition, miR-142-deficient mice develop thrombocytopenia stemming from defective megakaryocyte maturation [15] and exhibit dysregulation of dendritic cell (DC) function [16]. Disruption of 2 miR-142 paralog genes in zebra danio using zinc-finger nucleases was reported to cause aberrant neutrophil differentiation [17]. In the T-cell compartment, miR-142 is required for the homeostasis of peripheral T effector (T eff ) cells, but is apparently dispensable for conventional T (T conv ) cell development in the thymus [14,18,19]. A recent study by Anandagoda and colleagues has demonstrated that miR-142 is essential for the immunosuppressive activity of T reg cells, but failed to reveal a significant role for this miRNA in T reg cell development and homeostasis [20]. The authors suggest that posttranscriptional repression of the cAMP-hydrolyzing enzyme Pde3b by miR-142-5p isoform plays a key role in the regulation of the T reg cell suppressive function.
Here, we show that mice with a conditional deletion of miR-142 in T reg cells develop severe autoimmune disease due to a profound defect in T reg cell homeostasis and function. In addition, our findings suggest that miR-142 plays an important role in tT reg cell development. We have determined that the miR-142-3p isoform and its capacity to silence multiple interferon gamma (IFNγ)-associated genes play a critical role in mediating the regulatory activity of miR-142 in T reg cells. Global ablation of IFNγ rescues the T reg cell defect and autoimmunity in Foxp3 Cre miR-142 fl/fl mice, thus providing further evidence for the essential role of the miR-142-IFNγ signaling pathway in the regulation of T reg cell homeostasis and function.

miR-142 is dynamically expressed in T cells, and its global ablation results in a T reg cell defect
Our efforts to define the role of miR-142 in the regulation of T-cell tolerance stem from an unexpected observation that global miR-142 ablation results in a marked T reg cell defect. We found that germline miR-142 KO mice display a significant drop in T reg cell numbers in both thymus and secondary lymphoid organs (S1A-S1C Fig), indicating that miR-142 plays an important role in T reg cell development and homeostasis. The observed defect was specific to T reg cells, because thymic development of T conv cells was largely normal in the germline miR-142 KO mice [14,18,19]. Our expression profiling experiments revealed that the mature miR-142-3p isoform is abundantly expressed in thymic and pT reg cells, whereas the mature miR-142-5p isoform is present in significantly lower amounts (S1D Fig). The level of miR-142-3p expression in T reg cells is roughly comparable to the expression of this miRNA in naive and activated T eff cells. Of note, miR-142 is dynamically expressed during T-cell development: its abundance gradually increases following T-cell maturation in the thymus and reaches a peak at the single positive (SP) thymocyte stage; however, mature T cells that egress from the thymus into periphery display a somewhat reduced miR-142 expression in comparison to SP thymocytes (S1D Fig).

Lethal autoimmune disease in mice with T reg cell-specific miR-142 deletion
To determine the role of miR-142 in T reg cell function, we created mice with a T reg cell-specific ablation of this miRNA (hereafter Foxp3 Cre miR-142 fl/fl mice). Foxp3 Cre deleter/reporter mice [21] that were used to generate the T reg cell-specific excision of miR-142 conditional allele (miR-142 fl/fl ) express Cre recombinase as a fusion protein with yellow fluorescent protein (YFP) under the control of endogenous Foxp3 promoter, providing an effective way to purify T reg cells by flow cytometry. As expected, CD4 + YFP + T reg cells isolated from Foxp3 Cre miR-142 fl/fl mice were virtually devoid of mature miR-142-3p, whereas expression of this miRNA in CD4 + T conv cells remained unchanged (S1E Fig). Despite appearing normal at birth, Foxp3 Cre miR-142 fl/fl mice exhibited an apparent failure to thrive: they had a very short life span (Fig 1A), were smaller in size (Fig 1B), and had significantly lower body weight than wild-type (WT; Foxp3 Cre ) littermates ( Fig 1C). Gross morphological analysis revealed that Foxp3 Cre miR-142 fl/fl mice developed a systemic lymphoproliferative autoimmune disorder that was characterized by significant lymphadenopathy (Fig 1D and S1J Fig), mild splenomegaly ( Fig 1D, S1F and S1H Fig), thymic involution (S1G and S1I Fig), dermatitis (Fig 1B), and massive immune cell infiltration into various peripheral organs ( Fig 1E). Taken together, the phenotypic findings from Foxp3 Cre miR-142 fl/fl mice resemble the autoimmune pathology observed in mice with severe T reg cell defects (e.g., scurfy mutants or Foxp3 KO mice [3][4][5][6][7]), suggesting that miR-142 is involved in the regulation of T reg cell homeostasis or stability.

T reg cell defect and aberrant T-cell activation in Foxp3 Cre miR-142 fl/fl mice
In agreement with this notion, we observed a marked reduction in T reg cell population in the Foxp3 Cre miR-142 fl/fl spleen (Fig 2A). Furthermore, the absolute number of thymic T reg cells in Foxp3 Cre miR-142 fl/fl mice was also significantly lower, despite a slight increase in the frequency of T reg cells in the atrophied thymus (S2A and S2B Fig). Because T reg cells play a crucial role in suppressing self-destructive responses elicited by autoreactive T cells, we examined the activation status of peripheral T cells in Foxp3 Cre miR-142 fl/fl mice by flow cytometry. In comparison to WT mice, the number of CD4 + and CD8 + T cells with memory/effector phenotype (CD44 hi CD62L lo ) was significantly higher in the spleen and lymph nodes of Foxp3 Cre miR-142 fl/fl mice, whereas the pool of naive (CD44 lo CD62L hi ) CD4 + and CD8 + T cells in the KO animals diminished considerably (Fig 2B and 2C, S2C and S2D Fig). Foxp3 Cre miR-142 fl/fl mice exhibited an expansion of peripheral CD4 + and CD8 + T-cell compartments (S2G Fig), most likely due to a failure of miR-142-deficient T reg cells to suppress proliferation of hyperactivated T lymphocytes. In addition, both CD4 + and CD8 + T cells isolated from Foxp3 Cre miR-142 fl/fl mice displayed a sharp increase in production of IFNγ (Fig 2D and 2E, S2E and S2F Fig), a pro-inflammatory cytokine that promotes type 1 T helper cell 1 (Th1) responses. In contrast, no obvious changes in interleukin (IL)-4 (secreted by Th2 cells) or IL-17 (secreted by Th17 cells) production by splenic CD4 + T cells were observed in Foxp3 Cre miR-142 fl/fl animals (S2H Fig). Thus, our findings suggest that conditional miR-142 deletion impairs T reg cell function and subsequently drives severe dysregulation of T eff responses.

miR-142 is essential for the homeostasis and suppressive activity of T reg cells
To determine how miR-142 ablation affects suppressive function of T reg cells, we cocultured purified miR-142-deficient and miR-142-sufficient T reg cells together with WT T conv cells that were induced to proliferate by antigen receptor stimulation. We found that miR-142-deficient T reg cells exhibited a comparable capacity to restrain the proliferation of activated T conv cells as miR-142-sufficient T reg cells in this classical in vitro T reg cell suppression assay (Fig 3A, S3A  Fig). Similar in vitro findings were previously reported for miR-146a and miR-181a/b-1, two other miRNAs that are implicated in the regulation of T reg cell activity [22,23]. Nevertheless, adoptive transfer of miR-142-deficient T reg cells failed to attenuate development of systemic inflammatory syndrome in a mouse model of acute graft-versus-host disease (aGVHD). Lethally irradiated BALB/c mice (H2 d ) transplanted with allogeneic CD4 + T conv cells from Results are shown as mean ± SD. P values were determined by log-rank test (A) or 2-tailed Student t test (C); ���� , P < 0.0001. The underlying raw data can be found in S1 Data file. HE, hematoxylin-eosin; SD, standard deviation; T reg , regulatory T.
https://doi.org/10.1371/journal.pbio.3001552.g001 C57BL/6 donors (H2 b ) rapidly developed sublethal aGVHD that was manifested by hunched posture, severe diarrhea, and significant weight loss in the host mice (Fig 3B and 3C). The concomitant transfer of WT T reg cells together with allogeneic donor T cells reduced the severity of aGVHD symptoms, as indicated by diminished diarrhea and less pronounced body weight loss in the host mice ( Fig 3C). In contrast, adoptive transfer of miR-142-deficient T reg cells failed to protect the host mice from the development of aGVHD and resulted in mortality, perhaps by exacerbating the inflammatory response (Fig 3B and 3C). The failure of miR-142-deficient T reg cells to suppress aGVHD indicated an essential role of miR-142 in the regulation of T reg cell suppressive activity in vivo. In addition, a significantly lower frequency of donor miR-142-deficient T reg cells in the spleens of host mice 17 days posttransplantation (Fig 3D) suggested the survival or homeostatic defect of miR-142-deficient T reg cells.
To test whether miR-142 ablation affects survival of T reg cells, we compared induction of apoptosis in miR-142-sufficient and miR-142-deficient T reg cells in response to TCR stimulation. We found that purified splenic miR-142 KO T reg cells were more prone to activationinduced cell death (Fig 3E), although resting T reg cells from Foxp3 Cre miR-142 fl/fl spleen displayed no obvious difference in Annexin V staining (S3B Fig). To further establish the cell- . Results are shown as mean ± SD. P values were calculated using 2-tailed Student t test. �� , P < 0.01; ��� , P < 0.001; ���� , P < 0.0001. The underlying numerical raw data can be found in S1 Data file. The underlying flow cytometry raw data can be found at the Figshare repository. FACS, fluorescence activated cell sorting; IFNγ, interferon gamma; SD, standard deviation; T reg , regulatory T.
https://doi.org/10.1371/journal.pbio.3001552.g003 frequency of splenic YFP + miR-142-deficient T reg cells in the mosaic mice ( Fig 3F, S3C Fig), indicating that miR-142 plays a cell-autonomous role in T reg cell homeostasis. In addition, in vivo labeling experiments using 5-bromo-2-deoxyuridine (BrdU) revealed that miR-142-deficient T reg cells have almost 2-fold lower proliferation capacity than WT T reg cells ( Fig 3G). Collectively, these findings implicate miR-142 in the regulation of homeostatic maintenance and proliferation of T reg cells.
Analysis of cell surface markers on pT reg cells from Foxp3 Cre miR-142 fl/fl mice by flow cytometry suggested that miR-142-deficient T reg cells display an activated/effector phenotype [24]. Consistent with this notion, miR-142-deficient T reg cells exhibited an up-regulation of several T-cell activation markers, including glucocorticoid-induced tumor necrosis factor receptor family-related gene (GITR), interleukin-2 receptor α (CD25), and integrin α E (CD103) (Fig 3H, S3D Fig). On the other hand, the levels of programmed death-1 (PD-1) receptor, which protects T reg cells from apoptosis [25], were significantly lower on miR-142deficient T reg cells ( Fig 3H, S3D Fig), although expression of other immune checkpoint molecules such as cytotoxic T-lymphocyte associated protein 4 (CTLA4) and inducible T-cell costimulator (ICOS) was not affected.

Specific deletion of miR-142 in T reg cells results in global derepression of miR-142-3p targets
To uncover the molecular mechanism by which miR-142 controls T reg cell function, we performed global transcriptome analysis of splenic CD4 + YFP + miR-142-sufficient and miR-142-deficient T reg cells using RNA sequencing (RNA-seq). We identified a total of 1,520 genes showing statistically significant change in expression due to miR-142 ablation, 988 of which were up-regulated and 532 of which were down-regulated in miR-142-deficient T reg cells ( Fig 4A, S1 Table). Analysis of differentially expressed genes in miR-142-deficient T reg cells with the SylArray and ToppFun software algorithms [26,27] revealed a significant enrichment of miR-142-3p (85 out of 429 annotated target genes), but not miR-142-5p, direct targets among the up-regulated genes ( Fig 4B), suggesting that impaired homeostasis and function of T reg cells in Foxp3 Cre miR-142 fl/fl mice are most likely driven by the loss of mature miR-142-3p expression. This conclusion aligns well with significantly more abundant expression of miR-142-3p in T reg cells (S1D Fig) and a large number of previously published observations from a broad spectrum of miR-142-deficient immune cell types [14][15][16][17][18][28][29][30]. Nevertheless, the possibility that miR-142-5p contributes to the regulation of T reg cell biology (as was recently proposed by Anandagoda and colleagues [20]) either by controlling the expression of a limited number of target mRNAs or by exerting its regulatory effect entirely at the translational level cannot be excluded, although our results suggest that the T reg cell defect in Foxp3 Cre miR-142 fl/fl mice is primarily mediated by miR-142-3p.
Pathway enrichment analysis of the differentially expressed genes identified significant overrepresentation of biological processes that are associated with innate and adaptive immune responses (S2 Table). In line with these findings, miR-142-deficient T reg cells display an up-regulated expression of diverse cytokine, chemokine, and immune receptor genes (S4A Fig). One potential caveat of this finding is that the aberrant inflammation observed in Foxp3 Cre miR-142 fl/fl mice can conceivably induce pro-inflammatory changes in the gene expression of T reg cells. In addition, expression of several T reg cell signature genes was altered by the loss of miR-142 (S4B Fig). Interestingly, the transcripts up-regulated in miR-142-deficient T reg cells were enriched for genes from the IFNγ signaling network according to the Enrichr software algorithm [31] and the Gene Set Enrichment Analysis (GSEA) [32] (Fig 4C, S4C Fig,  S2 Table). In agreement with our bioinformatic findings, we found that miR-142-deficient T reg cells produced significantly more IFNγ in comparison with WT cells (Fig 4D and 4E Because miR-142 deletion impairs T reg cell homeostasis, we examined the effect of IFNγ stimulation on T reg cell survival and found that IFNγ treatment significantly enhanced apoptosis of miR-142-deficient T reg cells in vitro (Fig 4G and 4H).

Ifng ablation rescues the T reg homeostatic defect and alleviates development of the systemic autoimmune disorder in Foxp3 Cre miR-142 fl/fl mice
To determine how dysregulated IFNγ signaling impacts the homeostasis of miR-142-deficient T reg cells, we generated Foxp3 Cre miR-142 fl/fl Ifng −/− double KO mice by crossing Foxp3 Cre miR- spleens with or without IFNγ treatment (10 ng/ml). Results are shown as mean ± SD. P values were calculated using 2-tailed Student t test. � , P < 0.05; �� , P < 0.01; NS, not significant. The underlying numerical raw data can be found in S1 Table and S1 Data file. The underlying flow cytometry raw data can be found at the Figshare repository. FACS, fluorescence activated cell sorting; IFNγ, interferon gamma; KO, knockout; MFI, mean fluorescence intensity; SCS, seed complementary sequence; SD, standard deviation; T reg , regulatory T; WT, wild-type.
https://doi.org/10.1371/journal.pbio.3001552.g004 142 fl/fl mice with Ifng −/− mice. Intriguingly, our analysis revealed a complete rescue of the T reg homeostatic defect in the double KO mice (Fig 5A). Furthermore, Ifng deletion alleviated development of the systemic lymphoproliferative and fatal autoimmune disease in Foxp3 Cre miR-142 fl/fl mice. This conclusion was supported by observations of improved survival (Fig 5B), body weight normalization ( Fig 5C) and marked rescue of the lymphadenopathy (Fig 5D), splenomegaly (Fig 5D and 5E), and thymic involution (Fig 5F) defects in Foxp3 Cre miR-142 fl/fl Ifng −/− mice. Of note, the double KO mice displayed an abatement of tissue inflammation, evidenced by a significant decrease in the amount of immune cell infiltration into peripheral tissues (Fig 5G). These  (Fig 5H), suggesting that the miR-142-IFNγ signaling axis is primarily important for maintaining T reg cell homeostasis. Despite the restoration of T reg cell frequency in Foxp3 Cre miR-142 fl/fl Ifng −/− mice, the remaining hyperactivation of CD4 + T conv cells in the double KO mice implies a potential role for miR-142 in the regulation of T reg cell immunosuppressive activity.

miR-142-3p targets multiple IFNγ-associated genes
Several of the IFNγ-associated genes that were derepressed in miR-142-deficient T reg cells, including Ifngr2, Stat1, Irf1, and Gbp3, are predicted by the TargetScan algorithm [33] as putative direct targets of miR-142-3p (Fig 4C, S5A Fig). We validated these bioinformatic predictions by examining the high-throughput sequencing of RNAs isolated by cross-linking immunoprecipitation (HITS-CLIP) database that was previously compiled by Rudensky and colleagues [34]. In activated CD4 + T cells, 2 of the 4 (Ifngr2 and Gbp3) target genes showed detectible peaks of Ago2 binding activity in their respective 3 0 UTRs, corresponding to predicted binding sites for miR-142-3p (S5A Fig). An extended analysis of the HITS-CLIP database revealed hypoxia-induced factor 1 alpha (Hif1a) as a potential miR-142-3p target ( Fig  6A). Our interest in this transcription factor was prompted by a previous report that implicated Hif1a in the regulation of T reg cell homeostasis and suppressive function through the control of IFNγ expression [35]. Using 3 0 UTR luciferase reporter assay, we confirmed the capacity of miR-142-3p to attenuate Hif1a and Ifngr2 expression by binding directly to evolutionary conserved seed sequences in their respective 3 0 UTRs (Fig 6B, S5B Fig). In line with these findings, we detected a significant increase in Hif1α and IFNγR2 protein levels in miR-142-deficient T reg cells, whereas Hif1α expression in CD4 + T conv cells remained unchanged (Fig 6C, S5C and S5D Fig). Of note, we did not observe changes in Hif1a mRNA expression in T reg cells lacking miR-142 expression.
In addition, we examined the capacity of miR-142-3p to regulate the expression of phosphodiesterase 3b (Pde3b), which was recently identified as a critical miR-142-5p target gene modulating T reg cell activity [20]. In agreement with the previously published results, we detected an approximately 3-fold increase of Pde3b mRNA expression in miR-142-deficient T reg cells (S1 Table). Analysis of the Pde3b 3 0 UTR by the TargetScan algorithm revealed 2 putative poorly conserved miR-142-5p binding sites and a single predicted poorly conserved miR-142-3p binding site (S5E Fig). Using the 3 0 UTR luciferase reporter assay, we determined that overexpression of miR-142 precursor can modestly attenuate Pde3b expression. Interestingly, we observed that mutation of the miR-142-3p binding site in the Pde3b 3 0 UTR can completely negate the silencing effect of miR-142 on the Pde3b reporter (S5B and S5F Fig). Thus, our findings have uncovered Pde3b as a novel miR-142-3p target gene, providing additional evidence for the critical role of miR-142-3p isoform in the regulation of T reg cell biology.
Hif1a haploinsufficiency partially rescues the T reg cell homeostasis defect in Foxp3 Cre miR-142 fl/fl mice Next, we sought to determine the contribution of the miR-142-Hif1a axis in T reg cell homeostasis and function. We lowered the Hif1a gene dose in miR-142-deficient T reg cells by crossing Foxp3 Cre miR-142 fl/fl mice with mice carrying a conditional Hif1a allele (Hif1a fl/fl ). The resultant Foxp3 Cre miR-142 fl/fl Hif1a +/fl mice expressed normal levels of Hif1α protein in T reg cells (S6A Fig) and displayed a partial rescue of T reg cell abundance (Fig 6D, S6B Fig). Furthermore, Foxp3 Cre miR-142 fl/fl Hif1a +/fl mice exhibited a partial reduction in the frequency of activated/ effector CD4 + T cells in the periphery (Fig 6E and 6F). Nevertheless, Hif1a haploinsufficiency

Discussion
Although miRNA-mediated posttranscriptional control of gene expression is recognized as crucial for T reg cell development and function [11][12][13], our understanding of how specific miRNA genes govern T reg cell responses is incomplete. Here, we report that mice with T reg cell-specific miR-142 deletion develop a fatal systemic autoimmune disease due to a severe defect in T reg cell homeostasis and suppressive activity. Furthermore, we found that

PLOS BIOLOGY
The role of miR-142 in Treg cell biology constitutive miR-142 ablation results in aberrant thymic T reg cell development. Thus, our findings have uncovered an indispensable role for miR-142 in the control of T reg cell-mediated immunological tolerance. We propose that miR-142 plays a dominant role among miRNAs involved in the regulation of T reg cell activity because the phenotype of Foxp3 Cre miR-142 fl/fl mice closely resembles the autoimmune pathology observed in mice with global disruption of miRNA biogenesis in T reg cells [11][12][13]. This notion should be confirmed by further conditional KO studies of several miRNA genes that were previously implicated in the regulation of T reg cell function, including miR-146a, miR-155, and miR-27 [22,36,37].
The role of miR-142 in T reg cell function was previously examined by Anandagoda and colleagues using a conditional KO mouse model [20]. In contrast with the findings from this report, we determined that deletion of miR-142-3p and not miR-142-5p (as proposed by Anandagoda and colleagues) is the major driver of the T reg cell defect and subsequent systemic autoimmunity in Foxp3 Cre miR-142 fl/fl mice. This conclusion is strongly supported by the observed global derepression of miR-142-3p target genes in miR-142-deficient T reg cells, whereas the levels of the majority of miR-142-5p targets did not significantly change. Our inference of a critical role for miR-142-3p in T reg cells is well aligned with the fact that miR-142-3p is more abundantly expressed in T reg cells than miR-142-5p and a large body of literature that assigns the main regulatory role in immune cells to miR-142-3p [14][15][16][17][18][28][29][30]. Our data revealing that miR-142-3p can directly bind and regulate the phosphodiesterase Pde3b gene, through which, as Anandagoda and colleagues suggest [20], miR-142 controls T reg cell immunosuppressive activity, further validates miR-142-3p as the key miR-142 isoform in T reg cells.
Our findings indicate that miR-142-3p controls T reg cell homeostasis and function by attenuating IFNγ production and signaling. Interestingly, dysregulation of IFNγ responses is increasingly recognized as a critical factor that negatively impacts T reg cell activity. For example, excessive IFNγ production was reported to drive functional "fragility" of T reg cells in the context of antitumor immunity [38]. Additionally, conditional deletion of the E3 ubiquitin ligase von Hippel-Lindau (Vhl) gene was shown to impair T reg cell function through IFNγ dysregulation [35]. Moreover, unrestrained Stat1 activation and subsequent IFNγ production by Socs1-deficient T reg cells was linked to a severe failure of immunological tolerance [22]. Finally, a loss of functional activity by Dicer-deficient T reg cells was associated with excessive IFNγ production [13]. miRNAs, despite eliciting a moderate effect on the expression of their target genes, often have a significant impact on cellular physiology through coordinated and coherent targeting of multiple key molecules in a signaling cascade [39]. In agreement with this notion, we found that miR-142 is predicted to control expression of several IFNγ-associated genes, including Ifngr2, Gbp3, Stat1, Irf1, and Hif1a and validated some of these as bona fide miR-142-3p targets. We propose that coordinated derepression of these target genes in miR-142-deficient T reg cells drives the observed dysregulation of IFNγ responses. In support of this hypothesis, we found that genetic blockade of IFNγ production rescues the homeostatic defect in miR-142deficient T reg cells and prevents development of systemic lymphoproliferative and autoimmune disorder in Foxp3 Cre miR-142 fl/fl mice. Despite a complete restoration of normal T reg cell frequency in Foxp3 Cre miR-142 fl/fl Ifng −/− mice, the partial rescue of peripheral T eff cell hyperactivation in these mice suggests a possibility that miR-142-mediated control of T reg cell function is not limited to the attenuation of IFNγ signaling and likely involves additional molecular targets. Another caveat of our genetic epistasis experiments using Foxp3 Cre miR-142 fl/fl Ifng −/− mice is a potential non-cell-autonomous effect of IFNγ deletion on miR-142-deficient T reg cells. Because we found a significant IFNγ up-regulation in both miR-142-deficient T reg cells and miR-142-sufficient T eff cells in Foxp3 Cre miR-142 fl/fl mice, the global IFNγ ablation might potentially impact miR-142-deficient T reg cell function in cell-extrinsic manner via changes in their IFNγ-rich inflammatory environment. Future analysis of miR-142-sufficient and miR-142deficient T reg cells from animals that lack IFNγ expression such as Foxp3 Cre miR-142 fl/fl Ifng −/− and female Foxp3 Cre/+ miR-142 fl/fl mice will be required to corroborate the intrinsic effect of IFNγ on miR-142-deficient T reg cells and uncover the additional signaling pathways through which miR-142 mediates its regulatory functions in T reg cells.
Our investigation of Hif1a, a validated miR-142-3p target, revealed an important role for the miR-142-Hif1a axis in the regulation of T reg cell homeostasis. We observed that lowering the Hif1a gene dose in miR-142-deficient T reg cells partially rescued the T reg homeostatic defect and modestly reduced the hyperactivation of peripheral T eff cells in Foxp3 Cre miR-142 fl/fl mice. The role suggested by our findings for Hif1a as a negative regulator of T reg cell homeostasis is consistent with the conclusions of 2 previous reports [35,38]. However, Hif1a haploinsufficiency had little impact on the dysregulated IFNγ production in miR-142-deficient T reg cells and failed to prevent development of fatal autoimmunity in Foxp3 Cre miR-142 fl/fl mice. This outcome is not surprising given the fact that the T reg cell number in Foxp3 Cre miR-142 fl/fl Hif1a +/fl mice is only partially restored. The failure of Hif1a haploinsufficiency to fully rescue the T reg cell defect in Foxp3 Cre miR-142 fl/fl mice is probably linked to the abundance of miR-142-3p target genes in the IFNγ signaling pathway. The existence of multiple miR-142-3p targets likely makes the dysregulated state of IFNγ signaling in miR-142-deficient T reg cells refractory to changes in the expression of a single target. This conclusion is supported by studies in miR-142-3p-deficient zebra danio, which display impaired myelopoiesis due to an aberrant activation of the IFNγ signaling pathway [17]. This developmental defect in miR-142-3p-deficient zebra danio could be rescued by a compound knockdown of stat1a and irf1b genes, whereas silencing of either factor alone was insufficient to restore normal neutrophil differentiation. Of note, based on the observations of dysregulated IFNγ signaling in miR-142-deficient immune cells from zebra danio and rodents, the role of miR-142 in suppressing IFNγ signaling appears to be evolutionary conserved. In contrast, the function of miR-142-Pde3b signaling axis is unlikely to be evolutionary preserved, because miR-142-3p and miR-142-5p binding sites in the mouse Pde3b 3 0 UTR are poorly conserved.
In summary, our results have established miR-142 as a central regulator of T reg cell development, homeostasis, and suppressive activity that mediates its function in T reg cells in part by limiting IFNγ production and responsiveness. Besides advancing our understanding of the T reg cell biology, these novel insights may open a new avenue for targeted pharmacological manipulation of T reg cell activity in cancer immunotherapy and autoimmune disease settings.

Mice
C57BL/6J (stock#000664), B6/CD45.1 (stock#002014), Hif1a fl/fl (stock#007561), and Foxp3 YFP-Cre (stock#016959) mice were purchased from the Jackson Laboratory (Bar Harbor, Maine, USA). BALB/c mice were obtained from Charles River Laboratories (Wilmington, Massachusetts, USA). miR-142 fl/fl mice were described previously [14]. Foxp3 Cre miR-142 fl/fl mice were generated by crossing miR-142 fl/fl mice with Foxp3 YFP-Cre deleter/reporter mice [21]. Mice were kept in a specific pathogen-free facility at the City of Hope Animal Resource Center, and all animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of the City of Hope. Our approved animal protocols that are relevant for this work are the following: IACUC#13021, IACUC#13020, and IACUC#03008.
All flow cytometry raw data generated in this study can be found at the Figshare data repository (https://figshare.com/projects/microRNA-142_guards_against_autoimmunity_by_ controlling_Treg_cell_homeostasis_and_function/128960).

Global transcriptome profiling by RNA-seq
CD4 + YFP + T reg cells were sorted using BD FACSAria II machine from single-cell splenocyte suspensions derived from 8-to 11-week-old Foxp3 Cre and Foxp3 Cre miR-142 fl/fl female mice (n = 3 per group). Total RNA from purified WT and miR-142-deficient T reg cells was isolated using miRNeasy kit (QIAGEN, Hilden, Germany) and subjected to RNA-seq. RNA quality was determined with an Agilent Bioanalyzer (RNA integrity number (RIN) > 7.5 for all samples). Library was prepared according to the manufacturer's protocol using Illumina TruSeq RNA Library Prep Kit v2 (San Diego, Califonia, USA) and subsequently loaded on an Illumina HiSeq 2500 for parallel sequencing. The 51 base pair single-ended sequence reads were mapped to the mouse reference genome (mm10) using the alignment program HISAT (https://daehwankimlab.github.io/hisat2). Gene expression was measured from alignment bam files by read counting function featureCounts in the Bioconductor package Rsubread (http://bioconductor.org/packages/Rsubread). The unstranded raw counts were then normalized using a trimmed mean of M values (TMM) method implemented in the Bioconductor package edgeR (https://bioconductor.org/packages/edgeR). A total of 11,658 genes having counts per million (CPM) values higher than 1 in at least 3 samples were included in the downstream differential expression analysis. Differentially expressed genes were tested using 3 statistical methods in edgeR, including the generalized linear model (GLM) quasi-likelihood F (QLF) test, the likelihood ratio (LR) test, and the exact test based on quantile-adjusted conditional maximum likelihood (qCML) methods. A complete list of genes and the statistical test results (miR-142-deficient (KO) versus miR-142-sufficient (WT) T reg cells) are shown in S1 Table. Statistical P values were adjusted by Benjamini-Hochberg method for false discovery rate (FDR) controls. The volcano plot visualizing the distribution of differentially expressed genes was based on QLF test results. The RNA-seq raw data were deposited in the Gene Expression Omnibus under the accession number GSE190192.

Sylamer analysis
Analysis of miRNA seed enrichment in the 3 0 UTRs of genes that are differentially expressed in miR-142 KO T reg cells was performed by the Web-based SylArray software algorithm (http://www.ebi.ac.uk/enright-srv/sylarray)) [26].

Histopathology
For histological sectioning, mouse tissues were collected and placed into 10% formalin, fixed for 24 hours, washed, and transferred to 70% ethanol before standard paraffin embedding.
Tissue sections were stained with hematoxylin and eosin and examined by an experienced veterinary pathologist.

Statistical analysis
All statistical analyses were performed using Prism 6 (GraphPad, San Diego, California, USA) software. Statistical analyses were performed using 2-tailed Student t test or ANOVA. Results were considered significant when P � 0.05.