KCNJ11, ABCC8 and TCF7L2 polymorphisms and the response to sulfonylurea treatment in patients with type 2 diabetes: a bioinformatics assessment

Background Type 2 diabetes (T2D) is a worldwide epidemic with considerable health and economic consequences. Sulfonylureas are widely used drugs for the treatment of patients with T2D. KCNJ11 and ABCC8 encode the Kir6.2 (pore-forming subunit) and SUR1 (regulatory subunit that binds to sulfonylurea) of pancreatic β cell KATP channel respectively with a critical role in insulin secretion and glucose homeostasis. TCF7L2 encodes a transcription factor expressed in pancreatic β cells that regulates insulin production and processing. Because mutations of these genes could affect insulin secretion stimulated by sulfonylureas, the aim of this study is to assess associations between molecular variants of KCNJ11, ABCC8 and TCF7L2 genes and response to sulfonylurea treatment and to predict their potential functional effects. Methods Based on a comprehensive literature search, we found 13 pharmacogenetic studies showing that single nucleotide polymorphisms (SNPs) located in KCNJ11: rs5219 (E23K), ABCC8: rs757110 (A1369S), rs1799854 (intron 15, exon 16 -3C/T), rs1799859 (R1273R), and TCF7L2: rs7903146 (intron 4) were significantly associated with responses to sulfonylureas. For in silico bioinformatics analysis, SIFT, PolyPhen-2, PANTHER, MutPred, and SNPs3D were applied for functional predictions of 36 coding (KCNJ11: 10, ABCC8: 24, and TCF7L2: 2; all are missense), and HaploReg v4.1, RegulomeDB, and Ensembl’s VEP were used to predict functions of 7 non-coding (KCNJ11: 1, ABCC8: 1, and TCF7L2: 5) SNPs, respectively. Results Based on various in silico tools, 8 KCNJ11 missense SNPs, 23 ABCC8 missense SNPs, and 2 TCF7L2 missense SNPs could affect protein functions. Of them, previous studies showed that mutant alleles of 4 KCNJ11 missense SNPs and 5 ABCC8 missense SNPs can be successfully rescued by sulfonylurea treatments. Further, 3 TCF7L2 non-coding SNPs (rs7903146, rs11196205 and rs12255372), can change motif(s) based on HaploReg v4.1 and are predicted as risk factors by Ensembl’s VEP. Conclusions Our study indicates that a personalized medicine approach by tailoring sulfonylurea therapy of T2D patients according to their genotypes of KCNJ11, ABCC8, and TCF7L2 could attain an optimal treatment efficacy.


Background
The prevalence of diabetes is increasing at a fast rate, which was 6.4% (285 million) among adults aged 20-79 years in 2010, and will increase to 7.7% (438 million) by 2030 [1]. Among all diabetic cases, approximately 90% are patients with type 2 diabetes (T2D), which is associated with a number of microvascular complications including retinopathy, nephropathy, neuropathy, as well as macrovascular complications [2]. T2D is caused by a plethora of lifestyle and genetic factors [3,4]. Current therapies for T2D include life-style modifications and use of oral antidiabetic drugs, with sulfonylurea being one of the most frequently used one [5]. There are a number of different sulfonylurea treatments for T2D patients, among which the commonly used ones are gliclazide, glibenclamide, glimepiride and glipizide [6].
Sulfonylurea promotes insulin secretion from the pancreatic β cells of the pancreas in a glucose-independent manner by binding to ATP-sensitive K + (K ATP ) channel on the cell membrane of pancreatic β cells. K ATP channel is a heterooctamer comprising the inward-rectifier potassium ion channels K ir 6.x (i.e., K ir 6.1 and K ir 6.2) that form the pore, and sulfonylurea receptors (SUR; i.e., SUR1, SUR2A, and SUR2B) that regulate the opening and closing of its associated K ir 6.x potassium channel, as SUR is sensitive to ATP and ADP levels. The binding of sulfonylureas to the corresponding receptors could lead to an efflux of intracellular potassium, hyperpolarization of the β cell membrane, and the opening of voltage-gated calcium channels, which result in an increased secretion of insulin to circulation (Fig. 1).
The pancreatic β cell K ATP channel consists of four poreforming subunits of the inwardly rectifying potassium channel Kir6.2 and four regulatory subunits of the SUR1 [7][8][9]. When blood glucose concentrations rise, an increase in glucose metabolism results in a change of ADP/ATP ratio, which leads to a closing of K ATP channel. The respective genes encoding K ir 6.2 and SUR1, i.e., KCNJ11 and ABCC8, are located next to each other on human chromosome 11p15. 15. Mutations in KCNJ11 or ABCC8 genes could decrease or abolish the metabolic sensitivity of β cell K ATP channel function, leading to a constant depolarization of the cell membrane and a persistent insulin secretion even at very low plasma glucose concentrations [10]. E.g., single nucleotide polymorphism (SNP) E23K (i.e., rs5219) of KCNJ11 gene is associated with T2D risk (reviewed in [11]), is shown to result in a decrease or loss of sensitivity of K ATP channel to the inhibitory effect of ATP [12] and/or an enhancement of activation by free fatty acids [13]. Further, mutations in ABCC8 gene could cause hyperinsulinemic hypoglycemia [10]. The β cell K ATP channel can be pharmacologically regulated by sulfonylureas, which function by binding to and closing the K ATP channel [14] that leads to membrane depolarization, which subsequently results in an activation of voltage-dependent calcium channels causing an influx of calcium, which then triggers insulin granule exocytosis.
TCF7L2 encodes a member of the T-cell factor (TCF) transcription factor that plays a critical role in Wnt signaling pathway [15], which is shown to be involved in β cell dysfunction in T2D [16]. TCF7L2 is a member of the TCFlymphocyte enhancer factor (LEF) protein family [17], and the bipartite transcription factor β-catenin/TCF-LEF serves as an effector of cAMP-dependent protein kinase A (PKA) signaling to mediate the physiological effects of peptide hormones including glucagon-like peptide-1 (GLP-1), which utilizes cAMP as a second messenger [18,19]. TCF7L2 gene SNPs are strongly associated with a higher risk of T2D development [15], which could be mediated by their influences on blood glucose homeostasis [20]. Sulfonylureas show considerable inter-individual variations in the hypoglycemic response, with approximately 10-20% of patients having a less than 20 mg/dl reduction in fasting plasma glucose (FPG) following the initiation of sulfonylurea therapy (called primary sulfonylurea failure) [21]. Further, about 50-60% of patients will initially have a greater than 30 mg/dl reduction in FPG, but will fail to reach the desired glycemic treatment goals [21]. In contrast, some T2D patients could have higher risks of mild or severe hypoglycemia in response to sulfonylurea treatment [22][23][24]. Molecular variants of sulfonylurea drug target genes KCNJ11, ABCC8, and TCF7L2 could lead to different responses to sulfonylurea therapy in T2D patients. Therefore, their impacts need to be carefully evaluated. The primary objective of this study is to predict functional effects of 36 coding (KCNJ11: 10, ABCC8: 24, and TCF7L2: 2, and all missense) and 7 non-coding (KCNJ11: 1, ABCC8: 1, and TCF7L2: 5) SNPs that were identified from published literatures and MutDB database (http://www.mutdb.org/) by applying a spectrum of in silico bioinformatics tools. Each Kir6.2 subunit has two transmembrane domains called M1 and M2, and the pore-forming domain is located between them [25]. The locations of 10 missense SNPs (including the well-studied E23K) in the KCNJ11 protein that comprises 390 amino acids [26] are shown in Fig. 2, respectively. Each SUR1 subunit has three transmembrane domains, i.e., TMD0, TMD1, and TMD3, and two nucleotide binding domains, i.e., NBD1 and NBD2. Between TMD0 and TMD1, there is a cytosolic loop called CL3 [27]. The locations of 24 missense SNPs (including the well-studied A1369S) in the ABCC8 protein that comprises 1581 amino acids [28] are shown in Fig. 3. The human TCF7L2 gene consists of 17 exons, five of which are alternatively spliced (i.e., exons 4, 13, 14, 15, and 16) and exhibits tissue-specific expression [29]. The differential splicing of TCF7L2 potentially gives rise to three groups of protein isoforms (i.e., short-, medium-, and large-length isoforms) with highly differential functional properties. These three groups depend on the predicted stop codon usages, which are located in exons 15, 16, 17 [30]. To date, TCF7L2 intronic SNP, rs7903146, represents the most significant risk variant for T2D [31]. However, four other non-coding SNPs, i.e., rs7901695, rs7895340, rs11196205 and rs12255372, have also been significantly associated with an increased risk of T2D [32] and have been widely studied. The locations of these 5 non-coding SNPs in the gene structure of TCF7L2 (including the well-studied intronic SNP rs7903146) are illustrated in Fig. 4.

Literature search strategy
Comprehensive electronic literature searches of databases including PubMed, Google Scholar, Cochrane Library, Excerpta Medica Database (EMBASE) were performed up to June 1, 2016 using the following keywords: sulfonylurea, type 2 diabetes, KCNJ11, ABCC8, and TCF7L2. A manual search of the references cited in initially identified articles was also performed. Furthermore, we searched all relevant references of three comprehensive review articles [5,33,34]. The search was restricted to English language articles.

Inclusion and exclusion criteria
Randomized controlled trials and observational studies were eligible for inclusion in the current study. In vitro studies, animal studies, letters, reviews, and unrelated articles and duplicates were excluded from this study.

Data extraction
From each included study, the following data were extracted: first author, publication year, SNP name, gene name, National Center for Biotechnology Information (NCBI) dbSNP (http://www.ncbi.nlm.nih.gov/snp/) ID, study design, study subjects, control source, length of follow-up, and results.
In silico bioinformatics analysis Computational predictions of functional impacts of non-synonymous SNPs (nsSNPs) Five in silico tools were applied: (i) SIFT [35]

KCNJ11
The most widely studied genetic polymorphism of KCNJ11 for sulfonylurea response is E23K (i.e., rs5219) located in exon 1 [33]. However, functional effects of KCNJ11 E23K polymorphism on the secretion and sensitivity of insulin in humans remain contentious [5]. Recent larger studies demonstrated that a significant reduction of insulin secretion, lower levels of insulin, and an improvement of insulin sensitivity were related to E23K variant in KCNJ11 gene [61]. Moreover, E23K variant was associated with T2D development, which means that the K allele carriers had an increased risk of T2D [44,62,63]. Furthermore, some studies also found that the K allele carriers had better therapeutic response to gliclazide in comparison with the EE homozygous wild-type group [50], as well as an increased risk of sulfonylurea treatment failure [45,49]. In addition, E23K variant was significantly associated with an increase of glycated hemoglobin A1c (HbA1c) level [47] and fasting glucose level that patients with the KK homozygous variant genotype had lower fasting glucose levels than those with the EE/EK heterozygous genotype [52]. Importantly, recent evidence demonstrated that patients with KCNJ11 variants responded more efficiently to sulfonylurea than insulin [64][65][66]. Another KCNJ11 polymorphism that was associated with sulfonylurea treatment responses is rs5210 which is located in 3'-untranslated region (UTR). A study conducted in two independent cohorts of Chinese T2D patients (cohort 1: n = 661, cohort 2: n = 607) treated with gliclazide demonstrated that KCNJ11 rs5210 was positively associated with gliclazide response in cohort 1 study [46].

ABCC8
The most widely studied genetic polymorphism of ABCC8 for sulfonylurea response is S1369A (i.e., rs757110) located in exon 33 [67]. This genetic variant was demonstrated to influence antidiabetic efficacy of sulfonylurea treatment in Chinese [46,56], as well as an increased sensitivity to gliclazide [56]. More importantly, KCNJ11 E23K and ABCC8 S1369A, two common K ATP channel mutations that were in strong linkage disequilibrium, form a haplotype that appears to be associated with an increased T2D risk [68]. Additional ABCC8 gene polymorphisms including rs1799854 (intron 15, exon 16 -3C/T) and rs1799859 (exon 31) had been shown to be associated with sulfonylurea treatment efficacy in Caucasians [48,55].

TCF7L2
Previous studies have shown that several non-coding genetic variants of TCF7L2 are associated with T2D risk in populations of diverse ancestries from countries encompassing United Kingdom [69], the Netherlands [70], Finland [32], Sweden [71], France [72], United States [73], India [74], and Japan [75] populations. Among these T2Dassociated TCF7L2 variants, rs7903146 (intron 4) showed the strongest association with T2D [76]. Significant reductions in HbA1c and fasting plasma glucose levels following a combined sulfonylurea and metformin treatment between T2D patients with CC genotype and those with CT/TT genotype were associated with TCF7L2 rs7903146 variant allele [59]. Moreover, the rs12255372 variant, together with the rs7903146 variant, was shown to be associated with a significantly more frequent treatment failure [58][59][60]. It shall be noted that although in previous literatures, e.g., as in [32,77], TCF7L2 rs7901695 and rs7903146 are indicated to be in intron 3, and rs7895340, rs11196205 and rs12255372 are indicated to be in intron 4, this is because exon 4, which is a variable exon, is often named as "3a" [78]. Because of a high incorporation in pancreatic β cells [79], exon 4 shall be included in the gene structure, such that rs7901695 and rs7903146 shall be indicated as located in intron 4, and rs7895340, rs11196205, and rs12255372 in intron 5, respectively, e.g., as in [80]. For the linear ordering of these 5 non-coding SNPs, according to the most updated (i.e., as of April 18, 2017) NCBI dbSNP, the chromosomal coordinates for rs7901695, rs7903146, rs7895340, rs11196205 and rs12255372 are 112994329, 112998590, 113041766, 113047288, and 113 049143, respectively, on human chromosome 10 based on GRCh38.p7 assembly. Therefore, the linear ordering shall be rs7901695-rs7903146-rs7895340-rs11196205-rs1225537 2, as shown in Fig. 4 (all drawings in Figs. 1, 2, 3, and 4 are not to their exact scales and are for illustration purposes), which is agreement with that of [77].
Analysis of functional consequences of SNPs by HaploReg v4.1 HaploReg v4.1 is an online software for exploring annotations of the non-coding genome among those results of published genome-wide association studies or new sets of genetic variants, which help researchers to integrate DNA regulatory elements data with genetic variants to quickly formulate novel biological hypotheses [40,41]. As predicted by HaploReg v4.1, rs1799854, rs7895340, rs7903146, rs11196205 and rs12255372 could change 4, 2 (i.e., Irf and PRDM1), 7, 1 (i.e., SMC3), and 5 DNA motifs   Table 3).

Analysis of functional consequences of SNPs by
RegulomeDB RegulomeDB is a database that annotates SNPs with known and predicted regulatory elements in the intergenic regions of the human genome. Of the 7 noncoding SNPs, rs5210, rs1799854, rs7901695, rs7903146, and rs11196205 had RegulomeDB scores of 4, 5, 5, 5, and 5, respectively, which were all classified as having minimal binding evidence. Predictions were not available for either rs7895340 or rs12255372 ( Table 3).

Analysis of functional consequences of SNPs by
Ensembl's VEP The Ensembl's VEP determines the effects of genetic variants on genes, transcripts, and protein sequences, as well as regulatory regions. Three noncoding SNPs of TCF7L2 gene, i.e., rs7903146, rs11196205 and rs12255372, were predicted as risk factors (Table 3).

Discussion
Sulfonylureas are a class of drugs that stimulates insulin secretion by closing K ATP channels in pancreatic β cells. It has been estimated that 10-20% of individuals treated do not attain adequate glycemic control, and 5-10% initially responding to sulfonylurea subsequently lose the ability to maintain near-normal glycemic level [86]. This implies that genetic factors are linked with treatment efficacy of sulfonylureas. In our study, that includes 17 studies, two KCNJ11 SNPs -rs5219 (E23K) (exon 1) and rs5210 (

Conclusion
The ultimate goal of pharmacogenetics is the development of personalized medicine through individual genetic profiles which would accurately predict which individuals with a specific medical condition would respond to a specific medical therapy. Traditional medicine refers to the broad application of "standard of care" or "one-size-fits-all" treatments to all patients with a given diagnosis. In contrast, personalized medicine, often described as providing "the right drug for the right patient at the right dose and time" [87], tailors medical treatment according to each patient's personal history, genetic profile and/or specific biomarkers [88,89], Therefore, the full application of personalized medicine in health care will require significant changes in regulatory and reimbursement policies as well as legislative protections for privacy. The U.S. Food and Drug Administration has updated the labels of more than 120 drugs with recommendations for genetic testing prior to their use [90]. Currently, most genetic testing is based genotypic effects. Haplotypes of multiple linked genetic variants provide more precise information of their functional impacts than individual genetic markers [91,92], which could also be potentially important for diagnosis and prognosis [93]. In future, regulatory authorities shall formulate clear guidelines for evaluating and approving personalized diagnostics and therapeutics and identify patients who can benefit from them.