The variant at TGFBRAP1 is significantly associated with type 2 diabetes mellitus and affects diabetes‐related miRNA expression

Abstract While the transforming growth factor‐β1 (TGF‐β1) regulates the growth and proliferation of pancreatic β‐cells, its receptors trigger the activation of Smad network and subsequently induce the insulin resistance. A case‐control was conducted to evaluate the associations of the polymorphisms of TGF‐β1 receptor‐associated protein 1 (TGFBRAP1) and TGF‐β1 receptor 2 (TGFBR2) with type 2 diabetes mellitus (T2DM), and its genetic effects on diabetes‐related miRNA expression. miRNA microarray chip was used to screen T2DM‐related miRNA and 15 differential expressed miRNAs were further validated in 75 T2DM and 75 normal glucose tolerance (NGT). The variation of rs2241797 (T/C) at TGFBRAP1 showed significant association with T2DM in case‐control study, and the OR (95% CI) of dominant model for cumulative effects was 1.204 (1.060‐1.370), Bonferroni corrected P < 0.05. Significant differences in the fast glucose and HOMA‐β indices were observed amongst the genotypes of rs2241797. The expression of has‐miR‐30b‐5p and has‐miR‐93‐5p was linearly increased across TT, TC, and CC genotypes of rs2241797 in NGT, P trend values were 0.024 and 0.016, respectively. Our findings suggest that genetic polymorphisms of TGFBRAP1 may contribute to the genetic susceptibility of T2DM by mediating diabetes‐related miRNA expression.

Transforming growth factor beta 1 (TGF-β1) regulates cellular communications in multiple cell types, including the growth and differentiation of pancreatic β cell which secrets insulin. 13 TGF-β1 also played a role in human neuroendocrine to induce the production of somatostatin (SST), while SST acts as a growth inhibitor. 14 The TGF-SST connection provides control of cell growth and potentially stimulates an autocrine feedback loop in diabetics. 14,15 In addition, insulin resistance (IR), as an important physiological marker of T2DM, is closely related to the impaired endothelium-dependent vasodilation. 16 The TGF-β1/Smad signalling pathway was found to be involved in vascular development and epithelial remodelling. 17 It is reasonable to infer the potential involvement of TGF-β1-related pathway in diabetes.
Indeed, TGF-β interacts with transmembrane receptors such as TGF-β1 receptor 1 (TGFBR1), TGF-β1 receptor 2 (TGFBR2), and TGF-β1 receptor 3 (TGFBR3) to mediate its effects. Amongst these three receptors, only TGFBR2 can bind TGF-β1, and then it recruits and phosphorylates TGFBR1. 18 Animal experiments showed that TGFBR2 facilitated the cell differentiation and proliferation of β-cells through the activation binding of Smad 2/3. 19 TGF-β receptor-associated protein 1 (TGFBRAP1) was recently shown to be the molecular chaperone of Smad 4. It carries Smad 4 to the activated TGFBR2complex and promotes the phosphorylation of Smad 2/3, which subsequently induces the biological functions of the Smad network. 20 Of particular interest to this study is the relevance of TGFBR2 and TGFBRAP1 polymorphisms to the genetic susceptibility of T2DM.
As a kind of noncoding RNA, microRNA (miRNA) is generated from endogenous hairpin structured transcripts throughout the genome and regulates at least 20%-30% of all human genes by epigenetic modification. 21 Specifically, miRNA involves in insulin secretion, β-cell differentiation, glucolipid metabolism, and many other diabetes-related processes 22 and a number of studies have reported that miRNA contributes to the progression of T2DM. 23,24 To date, however, it has not been clarified whether the gene expression of TGFBR2 and TGFBRAP1 involving in the development of T2DM is stimulated or suppressed by miRNA-binding target SNPs, or the variants at TGFBR2 and TGFBRAP1 contribute to its genetic effects on diabetes-related miRNA expression by epigenetic modifications.
The purpose of this study is to investigate the associations of nine single nucleotide polymorphisms (SNPs) at TGFBR2 and three SNPs at TGFBRAP1 with T2DM and to evaluate its genetic effects on diabetes-related miRNA expression. These would provide a novel insight into our better understanding of the TGF-β1 pathway with diabetes.

| Study population
A total of 4222 subjects were recruited from a rural population in Yixing city (Jiangsu province, China), which had been described previously. 25 The individuals were considered to be T2DM cases according to the presence of fasting plasma glucose (FPG) ≥7.0 mmol/L or a self-reported T2DM history. Subjects with FPG between 5.6 and 6.9 mmol/L were defined to have impaired fasting glucose (IFG), and those with FPG <5.6 mmol/L normal glucose tolerance (NGT). After further verification with 3 months, a total of 468 T2DM cases and 899 IFG subjects were selected, excluding individuals with cardiovascular diseases, stroke, and cancer. Two thousand eight hundred and fifty-five of age-grouped (±2 years) and gender-matched healthy individuals were identified as NGT controls.
The study protocol was approved by the Research Ethics Committee of Nanjing Medical University (NMU03307). All subjects were well informed about the current study and provided written consents; all methods were performed in accordance with the relevant guidelines and regulations.

| Questionnaire survey and anthropometric measurement
The investigators were uniformly trained and qualified. All subjects completed a standard questionnaire including demographic characteristics, smoking and drinking habits, medical history, and underwent physical examinations including weight, height, and blood pressure (BP) by trained research staff. Body weight and height were measured twice for each individual without heavy clothes and shoes, and were rounded to the nearest 0.1 kg and 0.1 cm, respectively.
Body mass index (BMI) was then calculated as weight (kg)/height squared (m 2 ).

| Chemical indices detection
Blood samples were collected after 8 hours from the last meal or during an overnight to measure fast glucose (GLU) using the glucose oxidase method. Insulin was detected using chemiluminescence while homeostatic model assessment of IR (HOMA-IR) and HOMA of β-cell functions (HOMA-β) was calculated. HOMA-IR = fasting plasma glucose × fasting plasma insulin/22.5, HOMA-β = 20 × fasting plasma insulin/(fasting plasma glucose-3.5).

| miRNA isolation and detection
At the preliminary screening stage, total RNA was extracted in 600 μL plasma from 24 T2DM cases and 24 NGT respectively using miRNA microarray chip (Human microRNA Panelversion 1.0; Applied Biosystems, Foster City, CA). Specifically, 15 miRNAs were identified with the differential expression more than 2-fold changed between T2DM and NGT. These 15 miRNAs were further validated in 75 T2DM cases and 75 NGT (Table S8). The qPCR reaction was performed in triplicate to evaluate miRNA expression (5 μL reaction) in the plasma using the ABI RT-PCR 7900 system (Applied Biosystems; Thermo Fisher Scientific, Inc.). The qPCR parameters were 10 minutes at 95°C followed by 40 cycles of 15 seconds at 95°C and 1 minute at 60°C. Cel-miR-39 was used as an endogenous control. The relative expression of miRNAs in plasma was calculated with comparative cycle threshold (ΔCT) method. The CT value >35 was considered to be undetectable data, and CT value ≤35 was normalized by the ΔCT method with cel-miR-39, which had a stable CT value in the plasma of two groups.

| SNP selection and genotyping
The  (Table S1). SNP genotyping was performed using TaqMan technology (Applied Biosystems). All the genotype-calling success rates were greater than 99.9%.

| Statistical Analysis
The database was established in Epidata 3.0 (The Epidata Association, Odense, Denmark) and all statistical analyses were performed in SPSS version 15.0 (SPSS Inc., Chicago, IL). Qualitative variables amongst subject groups were compared using the Chi square (χ 2 ) test and a two-tailed P value of 0.05 was defined to be statistically significant.

| Demographic characteristics
In case-control study, the demographic and clinical characteristics of participants were summarized in Table 1. No significant difference in gender ratio was found amongst T2DM, IFG, and NGT groups, with males accounting 36.2%, 41.2%, and 41.1%, respectively (P > 0.05).
The ages of T2DM cases were slightly higher than those of the NGTs (+1.42 years) while BMI in both T2DM and IFG groups were significantly higher than in the NGTs. Thus, demographic characteristics of age, gender, and BMI were adjusted also before the genetic effects of SNPs were evaluated.

| Plasma levels of TGF-β1 amongst subject groups
There was significant difference in Sqrt-TGF-β1 concentration between T2DM, IFG and NGT groups with P trend = 0.004 ( Figure 1).

| Association analysis for T2DM and IFG
The allele frequency distributions of 12 SNPs were complied with Hardy-Weinberg equilibrium. Ordinal logistic regression analyses displayed a significant association of dominant rs749794 at TGFBR2 with T2DM and IFG, the OR (95% CI) was 1.146 (1.007-1.304), P = 0.038 ( Table 2). The other eight SNPs showed no significant associations with T2DM or IFG (Table S2).  A total of 12 miRNAs presented differential expression between T2DM cases and controls (Table S7). The expression of has-miR-30b-5p was found to be significantly different amongst the genotypes of rs749794 in T2DM, P = 0.041; post hoc multiple comparisons showed T2DM subjects with rs749794 CC genotype had a high level of has-miR-30b-5p than CT carriers, P = 0.018. The expression of has-miR-720 was significantly decreased across rs749794 CC, CT and TT genotypes, with a P trend value of 0.038. In controls, the expression of has-miR-139-5p gradually elevated across rs749794 CC, CT and TT carriers, with a P trend value of 0.043. The expression of has-miR-30b-5p and has-miR-93-5p was significantly increased across TT, TC, and CC genotype of rs2241797, P trend values were 0.024 and 0.016, respectively ( Figure 2). These results were also listed in Table S8.

| DISCUSSION
TGF-β1 plays a vital role in regulating the growth and proliferation of pancreatic β cells which are responsible for the insulin secretion. 14 Although the distinct role of TGFBR2 and TGFBRAP1 in the TGF-β1/SMAD signalling pathway had been observed previously, 18,20 no genetic association study was conducted to evaluate the correlation of TGFRB2 and TGFBRAP1 polymorphisms with T2DM. The current study firstly adopted a function candidate strategy to investigate the relevance of TGFBR2 and TGFBRAP1 polymorphisms to the genetic susceptibility to T2DM. (BDP1) motif. 26 As the expression of has-miR-30b-5p and has-miR-93-5p were significantly increased across rs2241797 TT, TC and CC genotype, further functional experiment is warranted to illuminate whether rs2214797 affects the susceptibility to T2DM through the epigenetic mechanism.
The joint effect of rs2241797 and rs749794 on T2DM was identified as an allele number ranked dose-response, which would be helpful to understand the molecular pathogenesis mechanism of T2DM, and to provide a scientific research basis for individualized drug therapy of patients with T2DM. Meanwhile, the plots of rs2241797 and rs749794 observed less medium to high LD (r 2 > 0.6) SNPs, which makes them more available to the selection of biofunctional research and prediction of T2DM.
An A>G mutation of rs2679860 located downstream of the TGFBRAP1 gene would result in a negative influence on the combining functions of the transcriptional factors GCM and GATA-1. The TGFBRAP1 gene expression might therefore be modified and contribute to the risk of T2DM; however, the absence of its association with the quantitative traits of GLU might indicate an actual involvement of its closely related SNPs. Furthermore, we make a regional  In the current study, 12 miRNAs expressions were found to be significantly different in T2DM and NGT subjects. Four miRNAs (has-miR-30b-5p, has-miR-93-5p, has-miR-126-3p, and has-miR-320a) [27][28][29][30] were further validated with previous studies; five miRNAs (has-miR-150-5p, has-miR-328-3p, has-miR-335-5p, has-miR-511-5p, and has-miR-720) were first demonstrated to be differentially expressed in T2DM and NGT, which might become new biomarkers for T2DM diagnosis. Nevertheless, three miRNAs expressions were contradictory with our findings. The level of has-miR-139-5p was found to be significantly higher in T2DM than that in NGT in this study. However, it was also illustrated that no differential expression of has-miR-139-5p in 55 T2DM patients and 80 controls, the conflicting result appeared due to the various tissues (microparticles) for miRNA isolation. 31 Meanwhile, higher levels of has-miR-191-5p and has-miR-574-5p in T2DM cases were observed, which were inconsistent with previous studies. 32,33 This could be linked to the population studied, with ethnicities, age or gender difference considered. 34 However, whether the expression of miRNA is ethnicity related in T2DM is not totally elucidated.
Functional studies have identified that increased miR-30b level contributes to cytokine-mediated β-cell dysfunction occurring during the development and progression of type 1 diabetes. 27 The miR-93 expression was higher in the diabetic retinopathy group than those in the healthy group, and severed as a diagnostic marker for type 2 diabetic retinopathy. The current study showed that has-miR-30b-5p and has-miR-93-5p were elevated in T2DM compared with NGT, which were consistent with previous reports. 27,35 Specifically, we evaluated the effect of rs2241797 on these miR-NAs expression, and has-miR-30b-5p and has-miR-93-5p were significantly increased across rs2241797 genotypes. Besides, the has-miR-30b-5p and has-miR-720 expression were significantly distinct amongst rs749794 variation in T2DM, while has-miR-139-5p expression gradually increased amongst rs749794 variation in NGT. These results indicate rs2241797 and rs749794 may F I G U R E 2 Comparison of miRNA expression amongst different genotypes in T2DM and NGT. Significant different expressions for has-miR-30b-5p and has-miR-720 were observed in T2DM cases with rs749794 CC, CT and TT genotype (A and B); while the has-miR-139-5p had an elevated trend amongst rs749794 CC, CT and TT in NGT (C). The expression of has-miR-30b-5p and has-miR-93-5p was significantly increased across TT, TC and CC genotypes of rs2241797, P trend values were 0.024 and 0.016, respectively (D and E). T2DM, type 2 diabetes mellitus; IFG, impaired fasting glucose; NGT, normal glucose tolerance contribute to the genetic susceptibility to T2DM by mediating diabetes-related miRNA expression. This further favours the potential role of rs2241797 participating in the molecular mechanism of T2DM.
Several limitations in the present study were acknowledged.
First, all participants were from the same area (Han population in south China) so that the subject diversity of varied cultures and lifestyles was limited. Second, serum TGFBRAP1 and TGFBR2 levels were not yet detected in our study. Finally, the association analyses would be better refined in a larger sample size.
To the authors' best knowledge, the effects of genetic variants related to TGF-β1 on the susceptibility of diabetes and related diseases have not been documented. The current study presents the novel and original findings that TGFBRAP1 SNP rs2241797 was significantly associated with T2DM. The mutations were also found to be correlated with the quantitative characters of GLU, insulin, HOMA-IR and HOMA-β in NGT or IFG population. In addition, the expression of has-miR-30b-5p and has-miR-93-5p was significantly different amongst rs2241797 genotypes. This indicated that TGFBRAP1 might participate in the epigenetic mechanism of diabetes; however, a further systemic functional analysis would be warranted and future studies of population diversity are desired.
In conclusion, our findings suggest that genetic polymorphisms of TGFBRAP1 may contribute to the genetic susceptibility to T2DM by mediating diabetes-related miRNA expression.