A SNP uncoupling Mina expression from the TGFβ signaling pathway

Abstract Introduction Mina is a JmjC family 2‐oxoglutarate oxygenase with pleiotropic roles in cell proliferation, cancer, T cell differentiation, pulmonary inflammation, and intestinal parasite expulsion. Although Mina expression varies according to cell‐type, developmental stage and activation state, its transcriptional regulation is poorly understood. Across inbred mouse strains, Mina protein level exhibits a bimodal distribution, correlating with inheritance of a biallelic haplotype block comprising 21 promoter/intron 1‐region SNPs. We previously showed that heritable differences in Mina protein level are transcriptionally regulated. Methods Accordingly, we decided to test the hypothesis that at least one of the promoter/intron 1‐region SNPs perturbs a Mina cis‐regulatory element (CRE). Here, we have comprehensively scanned for CREs across a Mina locus‐spanning 26‐kilobase genomic interval. Results We discovered 8 potential CREs and functionally validated 4 of these, the strongest of which (E2), residing in intron 1, contained a SNP whose BALB/c—but not C57Bl/6 allele—abolished both Smad3 binding and transforming growth factor beta (TGFβ) responsiveness. Conclusions Our results demonstrate the TGFβ signaling pathway plays a critical role in regulating Mina expression and SNP rs4191790 controls heritable variation in Mina expression level, raising important questions regarding the evolution of an allele that uncouples Mina expression from the TGFβ signaling pathway.


Introduction
Mina is a widely expressed pleiotropic protein with known oncogenic and immunoregulatory roles [1][2][3][4][5][6][7]. Originally discovered as a Myc-induced nuclear antigen of 53 kDa with pro-proliferative activity in promyelocytic leukemia HL60 cells [1], it has subsequently been shown to be overexpressed in a wide variety of human cancers, in some cases providing prognostic value [2]. Independently, its encoding gene was mapped to a locus regulating Th2-bias [3,8,9], a genetic trait defined as the propensity of na€ve T helper cells to develop in vitro into IL4-producing Th2 cells [9,10] and it was shown to act as a dose-dependent transcriptional corepressor of the gene encoding interleukin-4 (IL4) [3], a key regulator of Th2 development [9][10][11][12]. Later work with Mina KO mice, however, revealed the dispensability of Mina for normal Th2 development, perhaps due to functional redundancy with its close evolutionary and structural paralog No66 [13]. More recently, Mina was found to possess a non-redundant role in promoting Th17 differentiation [4]. Th17 cells are known to be important in pulmonary inflammatory disease [14]. Consistent with this, Mina KO mice exhibited protection from silica-induced lung fibrosis, associated with impaired Th17 and elevated iTreg cell infiltration in the lung [7]. Further, in a house dust mite model of allergic asthma, Mina KO mice exhibited attenuated airway disease [6]. And finally, genetic variation at the MINA locus was found in a Han Chinese population to be associated with the development of childhood atopic asthma [15]. Most recently, Mina was found to possess a non-redundant intestinal epithelial cell-intrinsic role in constraining a latent anthelmintic pathway associated with down-regulation of Th1 responses and upregulation of a family of small cationic anti-microbial peptides called a-defensins (manuscript submitted).
While evidence supporting a key role for Mina in a variety of important physiological and cellular contexts has mounted, understanding of Mina gene expression regulation has lagged. To begin addressing this gap, we previously reported the molecular characterization of the Mina promoter region and its trans-acting factors in murine T cells [16]. This study defined a 144 bp minimal Mina promoter region that contained two promoters (P1 and P2), the stronger of which (P1) comprised four functional Sp1/3 binding sites. Consistent with this, chromatin immunoprecipitation (ChIP) assays in primary T helper cells revealed the Mina promoter region to be enriched in bound Sp1/3 as well as lysine 4 trimethylated histone H3 (H3K4me3), a marker of transcriptionally active chromatin [16]. Pharmacological inhibition of Sp1 binding activity and RNAi knockdown of Sp1 protein both substantially diminished Mina mRNA expression, indicating a requirement of P1 and Sp1 for Mina transcription in primary T helper cells.
Here, we extend these results by exploring the distal regulatory landscape of the Mina locus. Using DNAseI hypersensitivity and reporter assays, we identified 4 new Mina CREs. The most powerful of these (E2) comprised a Smad3-binding, TGFb responsive element whose activity was regulated by SNP rs4191790 located in the middle of a Smad3 binding element. We found that the BALB/c rs4191790 A allele dramatically attenuated Smad3 binding, TGFb responsiveness and Mina transcription, while the C57BL/6 rs4191790 G allele was permissive for all three. Our results demonstrate that the TGFb signaling pathway can promote Mina transcription. Further, the ability of rs4191790 to uncouple Mina expression from TGFb signaling provides an explanation for heritable variation in Mina expression level. We discuss the implications of these findings from physiological and evolutionary perspectives.

Results
Although Mina expression is ubiquitous, its quantitative level varies across different cell types, developmental stages, and activation states [2,3]. For example, its expression is elevated in parts of the CNS, the testes, the eye, lymph nodes, the spleen, T helper cells and plasmacytoid dendritic cells. Developmentally, in the lung, Mina expression is low in neonatal mice and high in adults (Stephania Courmier, personal communication). In na€ve T helper cells activated by PMA/ionomycin, Mina protein level peaked in the cytosol and the nucleus, respectively, at around 72 h (eightfold above basal) and 18 h (threefold above basal) [3]. At the transcriptional level, TCR crosslinking of na€ve CD4 T cells induced Mina mRNA level to rise approximately threefold over 24 h [3]. In order to build a framework for understanding how Mina gene expression is regulated we set out to explore the cis regulatory landscape of the Mina gene locus.

The Mina Promoter Contains an Enhancer E1
Previously, we showed that a Mina promoter regionspanning DNA fragment À1588/þ354 drove strong luciferase reporter activity in EL4 thymoma cells ( Fig. 1 and [16]; Numbering throughout is with respect to the transcriptional start site in promoter P1, marked ''0'' in Fig. 1B). Analysis of a set of 5 0 nested deletions of this fragment identified two Mina promoters P1 (À64/þ19) and P2 (þ150/þ280), each capable of driving low-level reporter activity, explaining part but not all of the reporter activity in fragment À1588/þ354. To explain the missing reporter activity, we inferred the existence of an enhancer (E1) in region þ80/þ354 downstream of P1 [16]. In order to verify this, we focused on fragment À64/þ354 that exhibited similarly strong reporter activity as the parental À1588/þ354 fragment (Fig. 1). While 3 0 nested deletions of fragment À64/þ354 to þ262 and þ151 had no effect on reporter activity, further deletion to þ80 dramatically reduced but did not abolish it, verifying the existence of E1 and locating it to region þ80/þ151. A 5 0 deletion fragment þ150/þ354 lacking P1 and E1 still exhibited low but significant reporter activity, corroborating previous evidence for the existence of a second Mina promoter P2. Extension of this P2-containing fragment to include E1 (but not P1) on fragment þ19/þ354 did not result in an increase in reporter activity, verifying our previous conclusion that E1 is P1-specific.

Eight DNAseI Hypersensitive (DHS) Sites Across the Mina Locus
Next, to screen the remaining Mina gene locus for additional CREs, we employed DNAseI hypersensitivity (DHS) mapping, a well-established approach that exploits relative nucleosome depletion (and hence susceptibility to nuclease digestion) arising from transcription factor binding to DNA [17]. We performed DHS analysis on two adjacent Kpn1 fragments that together spanned a 26-kb interval encompassing the Mina locus ( Fig. 2A). This analysis identified eight DHS sites (Fig. 2B). DHS sites 1-6 were constitutive and located in the Mina promoter/intron 1 region where functional CREs often reside [18,19]. DHS sites 7 and 8, located, respectively in introns 2 and 4, were PMA/ionomycin-dependent and likely bind factors involved in modulating Mina expression upon cellular activation. DHS sites 2 and 3 flanking the Mina promoter region likely correspond, respectively, to the locations of the Sp1-binding Mina P1 promoter and the downstream E1/P2 element.

Mina Intron 1 Contains Three CREs (E2, E3, and E4)
In order to begin exploring the allelic regulation of Mina expression, we decided to focus our analysis on DHS sites 4-6 that occurred in a region where we had previously identified a 21 SNP biallelic haplotype block (Fig. 3A). We used dual luciferase reporter assays to screen C57BL/6-derived DHSspanning DNA fragments for CRE activity. Overlapping DNA fragments þ354/þ1201 and þ904/þ3755 from Mina intron 1 each drove strong reporter activity, respectively five and twofold greater than the E1-containing promoter fragment À1588/þ354 (Fig. 3B). These results suggested that regions þ354/þ1201 and þ904/þ3755 each contained at least one Mina CRE. We inferred fragment þ354/þ1201 with its single DHS site (4) contained a single CRE and named this enhancer 2 (E2). By contrast fragment þ904/þ3755 contained two DHS sites (5 and 6) leading us to hypothesize it contained two additional CREs. To map their locations, we generated and functionally screened three sub-fragments of fragment þ904/ þ3755: þ904/þ3181, þ1092/þ1410, and þ1411/þ3755. Overlapping fragments þ1092/þ1410 and þ904/þ3181 exhibited similar moderate reporter activity, $2-fold greater than the promoter region alone (Fig. 3B). As both fragments contained DHS5, we named their resident CRE enhancer 3 (E3). This result also indicated that the non-overlapping region þ1411/þ3181 containing SNPs 6-10 harbored no CRE activity. Finally, fragment þ1411/þ3755 containing DHS6 was found to exhibit $3-fold greater reporter activity than the promoter region alone, indicating the presence of another CRE we named E4. Fragment þ904/þ3755 combining both E3 and E4 exhibited greater reporter activity than fragments that contained only E3 or E4, indicating the additivity of their respective activities. Nevertheless, the activity of fragment þ904/þ3755 containing both E3 and E4 was still $2-fold lower than that of the E2-containg fragment þ354/þ1201, highlighting the relative strength of E2. In summary, we have identified three additional CREs in Mina intron 1 corresponding to the locations of DHS sites 4-6: E2 on fragment þ354/þ1201, E3 on fragment þ1092/þ1410, and E4 on fragment þ3181/þ3755.
Next, we examined the E2-containing fragment þ354/ þ1210 (Fig. 5A). Compared to C57BL/6, the BALB/c version of this fragment exhibited a 50% reduction in activity (Fig. 5B). To determine which of the three SNPs (16)(17) in fragment þ354/þ1210 contributed to the reduction in E2 reporter activity, we performed site directed mutagenesis to convert each SNP from its C57BL/6 to its BALB/c allele. Conversion of SNPs 15 and 16 to their respective BALB/c alleles had an insignificant effect on E2's activity (not shown). By contrast, conversion of SNP17 (rs4191790) from its C57BL/6 (G) to BALB/c (A) allele resulted in a 50% reduction in E2 activity (Fig. 5C), demonstrating that allelic regulation of E2 activity depended on SNP rs4191790.

SNP RS4191790 Controls Smad3 Binding to Mina E2
We hypothesized that SNP rs4191790 modulated E2 activity by influencing transcription factor binding. Transfac analysis revealed rs4191790 to reside within a predicted SMAD binding element (SBE) (Fig. 6A). SMADs are a family of transcription factors that mediate the gene regulatory activities of TGFb signaling [20]. To test whether the predicted SBE was a functional SMAD binding site, we generated a series of five overlapping, C57BL/6-derived, rs4191790-containing double stranded DNA probes (p16-p20; Supplemental Table S1) for use in EMSA (Fig. 6A). Only one of the 5 probes (p18) formed a complex with EL4 nuclear lysate, suggesting that, in addition to the central SBE, p18-flanking sequences are critical for nucleoprotein complex formation (Fig. 6B). Strikingly, conversion of rs4191790 in p18 from the C57BL/6 (G) to the BALB/c (A) allele abolished nucleoprotein complex formation (Fig. 6B), correlating with the elevated reporter activity of the E2 fragment containing the C57BL/6 versus the BALB/c rs4191790 allele (Fig. 5C). Together, these results suggest that rs4191790 exerts its E2 regulatory activity by controlling the formation of a nucleoprotein complex.
Next to explore whether the rs4191790-spanning nucleoprotein complex contained a SMAD transcription factor, we performed EMSA supershift assays with transcription factorspecific antibodies. Neither an IgG isotype control antibody nor an Sp3-specific antibody influenced formation of the p18 nucleoprotein complex (Fig. 7A). By contrast, an antibody against Smad3 completely abolished it, demonstrating that the p18 nucleoprotein complex contains Smad3. To determine whether Smad3 binds the E2 chromatin region in living cells, we performed ChIP assays on activated EL4 cells ( Fig. 7B and Supplemental Table S1). The results revealed enrichment of Smad3 in chromatin mapping to the Mina E2 region but not a neighboring control region (Mina intron 2), demonstrating Smad3 binding to E2 at the Mina locus in living cells.

E2 Is a Smad3-dependent, TGFb Responsive Element
Binding of Smad3 to the SBE in Mina E2 suggested that E2 may confer rs4191790-regulated TGFb responsiveness. To test this, we performed dual luciferase reporter assays with rs4191790 G (C57BL/6) and rs4191790 A (BALB/c) allelic versions of E2 fragment þ354/þ1201 (Fig. 8). Transfected EL4 cells were treated or not with TGFb in the presence or absence of SIS3, a specific inhibitor of Smad3 phosphorylation and signaling [21]. In untreated cells, þ354/þ1201 rs4191790A exhibited half the activity of þ354/þ1201 rs4191790G , while TGFb treatment enhanced the activity of þ354/þ1201 rs4191790G but not þ354/þ1201 rs4191790A . The enhanced activity of þ354/ þ1201 rs4191790G in response to TGFb exhibited dose responsive ablation by SIS3 treatment. Together, these results show that E2 acts as an rs4191790-regulated, Smad3-dependent, TGFb responsive element.

Role of E2 and RS4191790 on Endogenous Mina Transcription
Finally, we asked whether in its endogenous genomic context E2 confers rs4191790-regulated, TGFb-dependent expression on the Mina gene. Previously, we had shown that Mina is required for normal in vitro differentiation of T helper 17 (Th17) cells [4], an inflammatory CD4 T helper subtype with host protective roles in fungal and certain mucosal bacterial infections as well as pathological roles in pulmonary inflammation and autoimmune diseases including multiple sclerosis and rheumatoid arthritis [14,22]. The development of Th17 cells can be driven in vitro by a combination of interleukin-6 (IL6) and TGFb [23]. To explore whether E2 contributed to Mina transcription during in vitro differentiation of Th17 cells, we used deep sequencing to enumerate Mina mRNA molecules transcribed from each parental allele in developing Th17 cells generated from [BALB/c Â C57BL/ 6]F1 (CB6F1) mice. We reasoned that if rs4191790 in E2 regulates TGFb signaling-dependent Mina transcription, preferential use would be made of the C57BL/6 allele whose E2 SBE, unlike its BALB/c counterpart, is responsive to TGFb. Using this approach, we found a small but clear bias toward transcription of the C57BL/6 over the BALB/c Mina allele (60% vs. 40%) (Fig. 9), consistent with a role for E2 in conferring TGFb responsiveness upon Mina transcription in T helper cells.

Discussion
We have performed a comprehensive survey of the Mina genomic locus to identify cis regulatory elements (CREs) using DNAseI hypersensitivity site (DHS) analysis to identify candidate regions followed by dual luciferase reporter assays for functional assessment. Across a 26 kb interval spanning the Mina gene locus including upstream and downstream sequences we identified 8 DHSs (two in the 5 0 promoter region, 4 in intron 1 and one each in introns 2 and 4). We extended our previously published work on the Mina promoter by locating a CRE we named enhancer E1 to a region between promoters P1 and P2 and confirming its specificity for P1 versus P2 [16]. DHSs 2 and 3 likely correspond, respectively, to the locations of promoter P1 and the bipartite element comprising enhancer E1 and promoter P2. DNA fragments individually containing DHSs 4, 5, or 6 were each found to possess reporter activity and their corresponding CREs were named, respectively, E2, E3, and E4. E3 and E4 activities were additive but even when combined were weaker than that of E2 alone. DHSs 1, 7, and 8 were not explored further in the current study, but likely correspond to three additional novel CREs.
Mina is a widely but non-uniformly expressed gene with pleiotropic functions [1][2][3][4][5][6][7]. It is likely that the four validated CREs (E1-E4) as well as the additional three likely to reside near DHSs 1, 7, and 8 coordinate to confer tissue-, developmental stage-, and signaling-specificity upon Mina transcription. We recently found that Mina KO mice harboring a deletion spanning exons 3 and 4 (encoding the catalytic JmjC domain) were developmentally normal and fertile but exhibited a defect in Th17 development [3] and expelled parasitic nematodes more efficiently than their WT counterparts (due to an intestinal epithelial cell-intrinsic defect) (manuscript submitted). An independently-derived Mina KO mouse strain in which exon 2 was replaced with a Neo cassette was also viable and fertile; furthermore, in an experimental model of house dust mite allergic airway inflammation it exhibited disease resistance that the authors suggested may arise from an observed defect in IL4 regulation [6]. However, it is possible that ameliorated airway disease in the exon2/Neo replacement strain resulted instead from a defective Th17 response (not explored in their report). Support for this idea comes from a third independent Mina KO strain in which exons 2-8 were replaced with a Neo cassette [7]. Unlike the exon3-4 deletion and the exon2/Neo replacement strains, the exon2-8/Neo replacement strain was homozygous lethal. Nevertheless, in a model of silicainduced lung fibrosis in comparison to WT controls heterozygous mice exhibited protection that was associated with an impaired Th17 and an elevated Treg response. The homozygous lethality of the exon2-8/Neo replacement strain is curious and likely to be independent of the loss of Mina enzymatic activity (data not shown) which occurred without lethality in our exon3-4 deletion strain.
Six distinct Mina transcripts are documented in AceView, one of them un-spliced and non-coding [24]. It is possible that the three mutant Mina alleles differentially impact the expression of alternative Mina transcripts in different tissues, leading to distinct biological effects, including the homozygous lethality of the exon2-8/Neo replacement strain. The  exon2-8/Neo replacement allele lacks $84 base pairs from the 3 0 end of intron 1 (which could perturb E4 function) and also lacks DHSs 7 and 8 (in introns 2 and 4, respectively). The exon3-4 deletion allele preserves E4 but lacks 131 bp from the 3 0 end of intron 2 and 101 bp from the 5 0 end of intron 4, potentially impacting the functions of DHSs 7 and 8. The exon2/Neo replacement allele preserves E4 as well as DHSs 7 and 8. Thus, it is possible that E4 and/or DHSs 7 and 8 may control essential Mina functions or act on a surrounding gene(s) essential for viability. Detailed knowledge will require systematic genetic dissection of each CRE and DHS.
Previously, we described 2 Mina promoters (P1 and P2) and showed that P1, the stronger of the two in EL4-based reporter assays, bound and required Sp1/3 transcription factors to function [16]. Three of four AceView proteinencoding Mina transcripts map to Mina promoter P1 while one (the RefSeq annotated transcript) maps to P2 [24]. The RefSeq Mina transcript (encompassing all 10 Mina exons) is supported by only three accessions. The only other transcript encompassing all 10 Mina exons originates from P1 and is supported by 56 accessions, consistent with the relative strength of P1 over P2. However, as both transcripts encode identical proteins the functional significance of P1 versus P2 usage in the Figure 8. Effect of rs4191790, TGFb, and the Smad3 inhibitor SIS3 on Mina E2 reporter activity. TGFb enhances the reporter activity of the rs4191790 G but not the rs4191790 A allelic version of E2 and this enhancement is abolished by the Smad3 inhibitor SIS3. PGL3 basic vector containing the Mina promoter and the E2 fragment þ354/þ1201 was transfected into EL4 cells for 24 h and then treated with 5 ng/ml TGFb alone or together with 0.5 SIS3 for 24 h before harvesting for luciferase analysis. FL/RL, the ratio of firefly over renilla luciferase activity. Data are representative of two independent experiments with similar results. generation of these transcripts is unclear, although it may pertain to differential expression level in different tissues. The one other major Mina transcript also originates from P1 (supported by 56 accessions), contains an alternative exon 8 and lacks coding exons 9 and 10. Detailed knowledge of the tissue-specific distribution of alternative Mina transcripts may help to explain the lethality of mice homozygous for the exon2-8/Neo replacement allele.
Previous work showed that the promoter/intron 1 region of the Mina locus segregated in two major haplotypic blocks (comprising 21 bialleleic SNPs) that correlated with the bimodal distribution of Mina expression level across inbred mouse strains, as typified by C57BL/6 (high Mina expression) and BALB/c (low Mina expression) [3]. Thus, we hypothesized that at least one of the 21 haplotypic SNPs resided in and perturbed the function of a Mina-regulatory CRE. Reporter assays revealed differential activity exhibited by C57BL/6 and BALB/c allelic versions of DNA fragments containing E2 (spanning SNPs 15-17) but not E3 or E4 (collectively spanning SNPs 1-15). Closer analysis revealed that allelic control of E2 was specified by SNP17 (rs4191790) with greater activity conferred by the C57Bl/6 (G) versus the BALB/c (A) allele. Sequence analysis revealed that rs4191790 lies within a SMAD-binding element (SBE) located within Mina intron 1. EMSA and EMSA supershift analyses showed that a DNA fragment (p18) spanning the rs4191790-containing SBE formed a Smad3-containing nucleoprotein complex with EL4 nuclear extract. Strikingly, nucleoprotein complex formation was controlled by allelic variation at rs4191790, forming only with the rs4191790 G (C57BL/6) and not the rs4191790 A (BALB/c) allele, consistent with the relative strength of the former versus the latter in reporter assays. Further, nucleoprotein complexes did not form with SBE-spanning fragments that overlapped p18 on either flank, suggesting that-in addition to the SBE-nucleoprotein complex formation required factors recruited to sequences flanking the SBE, consistent with the structure and function of other known SBEs [20]. ChIP assays in EL4 cells demonstrated that Smad3 was enriched in the genomic E2 region but not a neighboring control region of the Mina locus. Consistent with the structural interaction of Smad3 with E2, reporter assays revealed that E2 functioned as a TGFb-responsive, Smad3-dependent, rs4191790-regulated CRE. Finally, CB6F1 Th17 cells generated by activation in the presence of IL6 and TGFb, made preferential use of the C57BL/6 versus the BALB/c Mina allele.
Fetal bovine serum and mouse blood are each known to contain biologically active levels of TGFb [25][26][27], raising the possibility that TGFb in FBS-supplemented culture media and in the blood may be sufficient to tonically induce E2 activity. In support of this, we found a similar allelic bias toward usage of the C57BL/6 allele in CB6F1 Th0 cells (not explicitly cultured with TGFb) (data not shown). And there was a 2-3-fold elevation in basal Mina level in na€ve CD4 T helper cells ex vivo purified from C57BL/6 versus BALB/c mice [3]. Thus, differential E2 responsiveness to tonic TGFb exposure due to genetic variation at rs4191790 may explain the variation in Mina expression level across different inbred mouse strains [3].
Recently, we showed that Mina acts in intestinal epithelial cells to suppress a latent anthelmintic pathway associated with the regulation of anti-microbial a-defensin genes (manuscript submitted). This work raised the question why a host pathway would evolve that acts to suppress parasite expulsion. As TGFb is known to play a critical role in promoting chronic nematode infections [28], it is possible that activation of the Mina pathway by TGFb (or a nematode TGFb mimic [29]) is an evolved parasite immune evasion mechanism. Self-regenerating 3D intestinal organoid cultures are a dynamic in vitro model of functional intact intestinal epithelium featuring villus-and crypt-like structures comprising the major intestinal epithelial cell lineages [30]. TGFb treatment of C57BL/6 small intestinal organoid cultures induced Mina gene expression (personal communication with Y. Eriguchi and A. Ouellette), supporting a role for TGFb in promoting Mina expression in intestinal epithelial cells. Thus, emergence and fixation of the rs4191790 A allele that uncouples Mina from the TGFb signaling pathway may reflect evolutionary selection to counteract a nematode immune evasion mechanism that would otherwise opportunistically activate Mina to shut down a latent anthelmintic pathway. Differential activation of the Mina pathway in intestinal epithelial cells by TGFb may also contribute to the well-established elevated susceptibility to gastrointestinal nematode infection of C57BL/6 versus BALB/c strain mice [31].
In summary, our results elucidate the cis regulatory landscape of the Mina gene locus by identifying 4 novel enhancer elements and pointing out the locations of 3 potential additional ones. Further, we show that TGFb plays a critical role in regulating Mina expression through mobilizing Smad3 binding to an SBE in enhancer E2. Allelic variation at SNP rs4191790 in the E2 SBE is found to modulate Smad3 binding and E2 TGFb responsiveness, providing an explanation for genetic variation in Mina expression level across inbred mouse lines. Finally, we propose that emergence of an rs4191790 allele that uncouples Mina expression from the TGFb signaling pathway represents an evolutionary response to counteract a parasitic gastrointestinal nematode immune evasion strategy based upon opportunistic activation of the Mina pathway by TGFb mobilization.

Mice
Mice were bred and maintained in specific pathogen-free conditions in accordance with the ethical guidelines of both the Institutional Animal Care and Use Committee of St. Jude Children's Research Hospital, USA; and the Institutional Animal Care and Use Committee of Chiba University, Japan. BALB/c and C57BL/6 mice and CB6F1 were purchased from Jackson Lab (Bar Harbor, ME, USA).

EMSA
EL4 cell nuclear extracts were prepared with the NER-PER extraction kit according to manufacturer's directions (Pierce, 78833). Protein concentrations were determined using the Bradford Assay (Thermo Scientific, 1856209) using bovine serum albumin as a standard (Sigma A9418). Probes were generated by annealing 5 0 biotin-labeled oligonucleotides at 958C in annealing buffer (100 mM Tris pH 7.5, 10 mM EDTA, 2M NaCl, 50 mM MgCl2) followed by slow cooling to 228C ($3 h). Probe sequences are given in Supplemental Table S1. Nuclear extracts (10 mg), Poly(dA: dT) (1 mg), and biotinylated probes (200 fmol) were incubated together at 228C for 30 min in 20 ml binding buffer (10 mM Tris pH7.5, 60 mM KCl, 2 mM MgCl2, 0.15 mM dithiothreitol). For competition, 100-fold molar excess of unlabeled probe (20 pmol) was added to the reaction before adding biotinylated probe. For supershift, antibody (2 mg) was included in the reaction for 15 min before biotinylated probe was added. Binding reactions were resolved in 0.5X TBE 4% polyacrylamide gels at 150V for 2-3 h, transferred to nitrocellulose for 30 min at 48C, and then UV crosslinked for 45-60 second using the auto crosslink function of the UV-light crosslinking instrument (Stratagene-Agilent, Santa Clara, CA, USA). Probe signals were detected using the Chemiluminescent Nucleic Acid Detection Module kit (Cat # 89880, Pierce).

SUPPORTING INFORMATION
Additional supporting information may be found in the online version of this article at the publisher's web-site. Table S1. EMSA probes and ChIP Primers.