An analysis of the relationship between ABCC8 and KCNJ11 gene polymorphisms and diabetic retinopathy in Turkish population

ABSTRACT Background Diabetic retinopathy (DR) occurs due to high blood glucose damage to the retina and leads to blindness if left untreated. KATP and related genes (KCNJ11 and ABCC8) play an important role in insulin secretion by glucose-stimulated pancreatic beta cells and the regulation of insulin secretion. KCNJ11 E23K (rs5219), ABCC8–3 C/T (rs1799854), Thr759Thr (rs1801261) and Arg1273Arg (rs1799859) are among the possible related single nucleotide polymorphisms (SNPs). The aim of this study is to find out how DR and these SNPs are associated with one another in the Turkish population. Materials and methods This study included 176 patients with type 2 diabetes mellitus without retinopathy (T2DM-rp), 177 DR patients, and 204 controls. Genomic DNA was extracted from whole blood, and genotypes were determined by the PCR-RFLP method. Results In the present study, a significant difference was not found between all the groups in terms of Arg1273Arg polymorphism located in the ABCC8 gene. The T allele and the TT genotype in the −3 C/T polymorphism in this gene may have a protective effect in the development of DR (p = 0.036 for the TT genotype; p = 0.034 for T allele) and PDR (p = 0.042 and 0.025 for the TT genotype). The AA genotype showed a significant increase in the DR group compared to T2DM-rp in the KCNJ11 E23K polymorphism (p = 0.046). Conclusions Consequently, the T allele and TT genotype in the −3 C/T polymorphism of the ABCC8 gene may have a protective marker on the development of DR and PDR, while the AA genotype in the E23K polymorphism of the KCNJ11 gene may be effective in the development of DR in the Turkish population.


Introduction
Diabetic retinopathy (DR) is a microvascular complication of type 2 diabetes mellitus (T2DM) that occurs due to high blood glucose disorder in the retina and leads to blindness if left untreated (1).DR develops in almost all type 1 diabetes mellitus (T1DM) patients and 60% of T2DM patients 20 years after the onset of diabetes mellitus (DM) (2).The two types of diabetic retinopathy are nonproliferative DR (NPDR) and proliferative DR (PDR) (3,4).While NPDR manifests itself with intraretinal microvascular changes, retinal neovascularization is observed in PDR under low oxygen conditions (5).Individuals of all races and ethnic backgrounds can develop DR.However, some populations develop DR at a higher rate than others (6).Family-focused studies have demonstrated that genetic factors play a significant role in DR.In addition, some potential DR susceptibility genes have been indicated in meta-analyses (7).The most notable are those encoding the adenosine triphosphate-sensitive potassium channel (KATP) protein, which consists of the inwardly-rectifying potassium ion channel (Kir6.2) and sulfonylurea receptor 1 (SUR1) subunits.KATP and these genes play an important role in insulin secretion through glucose-stimulated pancreatic beta cells and the regulation of insulin secretion (8)(9)(10).The Kir6.2 and SUR1 subunits are encoded by the inwardly-rectifying potassium channel, subfamily J, member 11 (KCNJ11) and ATP-binding cassette transporter subfamily C member 8 (ABCC8) genes, respectively (9).KCNJ11 and ABCC8 are adjacently located on the same locus 11p15.1 (11).
KCNJ11 is highly expressed in the pancreas (10).Among the Single Nucleotide Polymorphisms (SNPs) in the KCNJ11 gene, E23K (rs5219) has attracted more attention due to its association with diabetes (12).The E23K variation may cause the ion channel to be less sensitive to ATP and consume more ATP until the channel is closed.As a result, the risk of T2DM and its complications increases with impaired insulin secretion (1,7,10,(13)(14)(15). Similarly, variations in the ABCC8 gene can impair potassium channel and insulin secretion, leading to various types of diabetes and their complications (15)(16)(17).Among these variations, −3 C/T (rs1799854), Arg1273Arg (rs1799859), and Thr759Thr (rs1801261) stand out concerning T2DM and its complications (15,17).
In the present study, we sought to compare the distribution of variants in ABCC8 and KCNJ11 genes in T2DM patients with and without retinopathy and nondiabetic control subjects.
Additionally, we examined the relevance of these polymorphisms to susceptibility to DR and the relationship of ABCC8 gene haplotypes with DR in the Turkish population.

Study population
The study included 353 T2DM patients (177 with DR, 176 without DR) and 204 age-matched individuals without T2DM who were admitted to Giresun University, Faculty of Medicine and Ulucanlar Eye Education and Research Hospital in Ankara.While American Diabetes Association criteria were used to diagnose T2DM, visual acuity, colour fundus photography, and standard retinal vessel analysis were used in the diagnosis of DR.A retina specialist carried out a fundoscopic evaluation.T2DM patients were divided into two groups: those with retinopathy (DR) and those without retinopathy (T2DM-rp).Whether or not those in the patient group had diabetic macular edema (DME) and diabetic maculopathy (DMP) was recorded.DR subgroup exclusion criteria included low picture quality, concomitant ocular disorders, and any media opacities.The severity of DR was divided into NPDR (the presence of microaneurysms, intraretinal haemorrhages and microvascular abnormalities and venous pilling) and PDR (the presence of preretinal haemorrhage/neovascularization or vitreous haemorrhage) following the International Clinical Diabetic Retinopathy Severity Scale (ICDRSS).The ophthalmologists were blinded to the genotyping results.The healthy control group consisted mostly of volunteers and hospital staff who had recently undergone a medical examination.The exclusion criteria included positive family history in first-degree relatives of the volunteers as well as a known history of diabetes, eye disease, or renal disease.Written informed consent was obtained from patients and controls under the Declaration of Helsinki.The research protocol of this study was approved by Clinical Trials Ethics Committee of Ordu University's Faculty of Medicine (approval on 12 February 2016-2016/7).All blood samples were examined at the routine hospital laboratory for fasting plasma glucose (FPG), lipid profile (total cholesterol, triglycerides, low-density lipoprotein-cholesterol, high-density lipoprotein-cholesterol), and glycated haemoglobin (HbA1c).All participants were interviewed, and standard clinical (hypertension status, duration of diabetes, insulin use) and demographic (age, gender, smoking habits, alcohol use, height, and weight) data were recorded.Body mass index (BMI) was calculated as a weight divided by the height squared (kg/m 2 ).

Genomic DNA isolation, polymerase chain reaction, and restriction fragment length polymorphism analysis
We performed genomic DNA isolation by the High pure PCR template preparation kit from whole blood in accordance with manufacturer's instructions (Roche Diagnostics, Mannheim, Germany).Amplification of genomic DNA was performed in a 50 µL reaction mixture including 10X buffer, 2.5 mM of dNTP solution, one unit of Taq DNA polymerase (Fermentas, Vilnius, Lithuania), 50 pmol of primers, and different concentrations of DMSO and MgCl 2 for each SNP (Table S1).Our research group designed the F and R primers for ABCC8 rs1799859 and KCNJ11 rs5219 primers, while previously published ABCC8 rs1799854 and rs1801261 primers were used (17).Supplemental Table S1 shows primer sequences and annealing temperatures.Standard amplification was performed in an automated thermal cycler (Biorad, USA).The cycle consisted of 95°C for 5 min, followed by 30 cycles at 95°C for 1 min, 55-65°C for 1 min, 72°C for one minute, and a final extension at 72°C for 5 min.The PCR products were separated on 2% agarose gel and visualised with a Gel Doc XR System (Biorad, USA).
For the determination of E23K, −3 C/T, Thr759Thr, and Arg1273Arg SNPs, approximately 15 mL of 209 bp, 248 bp, 158 bp, and 233 bp PCR products were digested overnight with 10 U of Eco24I, PstI, BsiEI (Fermentas, Vilnius, Lithuania), and BslI (New England Biolabs, USA) under appropriate buffer conditions and compatible temperatures (according to manufacturer's instructions).The RFLP fragments were separated into 3% or 4% agarose gels (1% standard agarose + 3% NuSieve agarose) for approximately 40 min at 100 V and visualised with a Gel Doc XR System (Biorad, USA).The length of the digested fragments is given in Table S1.

Statistical analysis
The SPSS statistical package for Windows (version 15.0, SPSS, Inc.) was used for statistical analysis First, we assessed the genotype distributions of each polymorphism for deviation from Hardy-Weinberg equilibrium (HWE) using a degree of freedom Chi-square (χ2 tests) test.Then, we compared the distributions of genotypes and alleles in the groups (T2DM, DR, and T2DM-rp) by χ2 test or by Fisher's exact test (if expected values are below 5).Next, we calculated odds ratios (ORs) with a confidence interval of 95% (95% CI) to estimate the relative risk of DR.A value of p < 0.05 was considered statistically significant.Finally, we applied SNPStats so that we could estimate ABCC8 haplotype frequencies using an expectation-maximization algorithm (18) and calculated the statistical power using QUANTO 1.2 software (website: http://hydra.usc.edu/gxe)(19).

Results
Table 1 shows the demographic and clinical characteristics of the groups.The results of the analysis indicated remarkable variations among the groups concerning demographic and clinical characteristics except for gender, age, and BMI (p ≤ 0.05).The differences among the groups' clinical and demographic characteristics were significant (p ≤ 0.05) except for age, gender, and BMI (p > 0.05).
In the controls and DR, T2DM-rp and T2DM with retinopathy cases, the genotype frequencies for the E23K, −3 C/T, and Arg1273Arg SNPs did not deviate from the HWE (p > 0.05).The Thr759Thr genotype frequency also did not deviate from the HWE in the control group (p > 0.05) but deviated in all patient groups (p ≤ 0.05).This deviation was unlikely to be associated with a laboratory error as 50 randomly selected samples were genotyped a second time and yielded stable outcomes.The study had a power greater than 80% for all SNPs except Thr759Thr.
Table 2 shows the genotype and allele distributions of the ABCC8 and KCNJ11 SNPs among all groups.Allele frequencies of Thr759Thr were significantly different between the T2DM-rp and control groups (p = 0.037).Moreover, the differences between the control group and the other groups concerning SNPs' allele and genotype frequencies were not significant.
When we compared T2DM patients with and without retinopathy, we found that the TT genotype (p = 0.036, OR = 0.52, 95 CI% = 0.28-0.96)and the T allele (p = 0.034, OR = 0.72, 95% CI = 0.54-0.98) of the −3 C/T polymorphism were less common in the DR group.On the other hand, when examining the E23K polymorphism, the AA genotype frequency (p = 0.046, OR = 1.99, 95% CI = 1.00-3.94)was higher in the DR group compared to the T2DM-rp group.Therefore, the AA genotype increased the development of DR 1.99 times (95% CI: 1.00-3.94,p = 0.046).The differences between T2DM patients with and without DR concerning frequencies of other polymorphisms were not significant (Table 3).
We also evaluated the severity of the patients' DR and found no significant variation between the proliferative (PDR) and non-proliferative (non-PDR) groups concerning the genotype and allele distributions in ABCC8 and KCNJ11 polymorphisms except for the −3 C/T polymorphism.Similar to the comparison of T2DM groups with and without DR, the TT genotype frequency of the −3 C/T polymorphism (OR = 0.27, 95% CI = 0.07-1.01,p = 0.042) was higher in the non-PDR group than the PDR group (Table 4).
When the demographic and clinical characteristics of the patients and ABCC8 and KCNJ11 genotypes were added as covariates, binary logistic regression revealed that age, duration of diabetes, HbA1C, and insulin therapy were the strongest determinants of DR (Table 5).ABCC8 and KCNJ11 genotypes were not identified as risk factors for DR.Concerning DR severity, logistic regression indicated that BMI and insulin therapy were the strongest predictors of PDR.However, the TT genotype of the ABCC8 −3C/T polymorphism (OR = 0.19, 95% CI = 0.045-0.816,p = 0.025) showed a protective effect on the development of PDR.
Supplemental Table S2 shows the estimated population frequencies for ABCC8 haplotypes in the DR and control groups.We found only six haplotype combinations in the DR and control groups.The finding that the CCG haplotype was the most frequent and the CTG haplotype the least frequent was expected, and the frequency of each haplotype in DR patients did not vary significantly compared to the control group (p > 0.05).

Discussion
Although studies have indicated that the common SNPs in the KATP (ABCC8 and KCNJ11) genes are associated with type 2 diabetes in many populations, the number of studies examining the effects of these polymorphisms on the complications of diabetes is limited (10,20).Among these polymorphisms, only the relationship between the KCNJ11 E23K polymorphism and DR has been investigated (1,7), while to the best of our knowledge, the possible relationship between SNPs in ABCC8 and DR has not been investigated in the literature.Variations in the KCNJ11 and ABCC8 genes can cause disrupted insulin secretion and give rise to a risk of diabetes and its complications, such as DR (15)(16)(17).Therefore, the present study discovered A study conducted in a Han Chinese population reported that the E23K AA genotype and A allele in the KCNJ11 gene were both genetic risk factors for DR in patients with type 2 diabetes (1).In an Iranian study, logistic regression analysis revealed that the E23K AG genotype was not a determinant for the development of retinopathy.However, a genotype-phenotype association study showed that GA genotypes had a significantly higher mean age for developing retinopathy than other genotypes (7).Apart from studies of DR, a study examining the relationship between E23K polymorphism and DPN in the Chinese population reported that E23K was associated with peripheral nerve function (14).Mota-Zamorano et al. reported that these polymorphisms were not related to the risk of DN in the Spanish population (15).The literature shows that these polymorphisms mostly focus on T2DM disease and were a T2DM risk factor among different populations in the Americas, East Asia, the Caucasian, Europe, the Middle East, and North Africa  (10,21).In the present study, we observed that the AA genotype increased in the DR group compared to the T2DM-rp group and increased the development of DR by 1.99 times.Polymorphisms in genes encoding KATP channel subunits are reported to affect the function of the channel.One of these polymorphisms, E23K, is known to cause channel overactivity (22,23).Changes in the KATP channel function affect vascular tone and blood flow (24,25).Therefore, due to their effects on the regulation of blood flow, polymorphisms in the Kir6.2 and SUR genes may also contribute to the disturbances in retinal blood flow that occur in the early stages of DR development.
In light of this information, the results of the present study seem to be related to this mechanism, at least for the E23K polymorphism.
When looking at the −3 C/T SNP located in the intron of the ABCC8 gene, we mostly encountered studies on T2DM.Therefore, the present study is probably the first attempt to investigate how this polymorphism contributes to the development of DR.While seven studies have reported associations between this polymorphism and T2DM risk, 10 studies have reported no association between them (20,26).The results for this polymorphism are contradictory.The findings of the present study indicated that the T allele and the TT genotype were less common in the retinopathy group, suggesting that the T allele may have a protective effect on the development of DR.Moreover, the TT genotype was a protective factor in the development of PDR.ABCC8 is known to modulate ATP-sensitive potassium channels and insulin secretion (8).In addition, recent studies have stated that insulin can contribute to or worsen the development of DR (27,28).Variants in the ABCC8 gene are reported to decrease insulin secretion (16).The results suggest that the −3 C/T polymorphism in the ABCC8 gene may affect the polarization of the cell membrane by changing the structure of the KATP channel and reducing insulin secretion.Therefore, the −3 C/T polymorphism of the ABCC8 gene seems to be a protective marker in the development of DR and PDR in the Turkish population.
Apart from the relationship between DN and the ABCC8 Arg1273Arg (rs1799859) polymorphism (15), we could not find any other study investigating the relationship between DR and other T2DM complications.The present study determined that the Arg1273Arg polymorphism did not make a significant contribution to the development of either T2DM and DR or PDR.Studies have mostly focused on T2DM.Of the 10 studies, three reported no relationship between this SNP and T2DM, while other studies reported a higher frequency of GA genotypes in T2DM patients than in controls (9,20).
As with Arg1273Arg, we encountered no other study investigating the relationship between the Thr759Thr polymorphism and DR or other T2DM complications, and most of the studies have focused on T2DM.In a review by Haghverdizadeh et al. in 2014 and later studies, only three of the eight studies found a relationship between this SNP and T2DM risk in Danish, Canadian, and Chinese populations (16,20,29,30).In the present study, we observed that the Thr759Thr polymorphism was not associated with the development of DR.Moreover, we revealed that the T allele was elevated only in the T2DM-rp group.However, this difference may not be reliable due to the low incidence of the T allele; more than 1500 patients are needed to reach 80% power.Furthermore, its genotype frequency deviated in all patient groups including T2DM-rp.The reason for this deviation may be the low frequency of the TT genotype.Moreover, HWE deviations in patients may primarily result from the association effects (31).
To the best of our knowledge, the relationship between DR or other T2DM complications and the KCNJ11 and ABCC8 polymorphisms in the Turkish population has not been investigated.Only one study investigated the association of the E23K and −3 C/T polymorphisms with T2DM, respectively, and showed that they were associated with the development of type 2 diabetes (32).However, this study did not examine the association of these polymorphisms with DR or other T2DM complications.In contrast to that study, the present study revealed that while these polymorphisms were not associated with T2DM, E23K SNP was associated with PDR and −3 C/T SNP was associated with DR and PDR.The relationship between these polymorphisms and T2DM, the number of cases in both studies, and the location of the volunteers (the volunteers in Gönen et al.'s study were limited to a single centre) may explain the difference in the results.The results of the present study showed that these polymorphisms may be directly related to DR and PDR through different mechanisms rather than T2DM and DR association.Therefore, our study presented new information to the literature concerning E23K and −3 C/T variants associated with DR in the Turkish population.
In conclusion, the present study revealed the relationship between the ABCC8 and KCNJ11 polymorphisms and DR in the Turkish population.Based on the results of statistical analyses, we observed that the −3 C/T polymorphism in the ABCC8 gene may have a protective effect on the development of DR and PDR, while the E23K polymorphism in the KCNJ11 gene could lead to the development of DR in the Turkish population.However, studies with larger sample sizes may provide a better understanding of the role of these SNPs in DR in the Turkish population.

Table 1 .
Baseline and clinical characteristics of the patient and control subjects.

Table 3 .
Genotype and allele frequencies of ABCC8 and KCNJ11 polymorphisms between type 2 diabetic patients without retinopathy (T2DM-rp) and patients with diabetic retinopathy (DR).

Table 5 .
Risk factors for DR and PDR using a binary logistic regression model.