Polymorphism of the GLIS3 gene in a Caucasian population and among individuals with carbohydrate metabolism disorders in Russia

Earlier, GLIS3 gene polymorphisms have been shown to be associated with the development of maturity onset diabetes of the young (MODY). We screened GLIS3 gene sequences among patients with MODY to identify probably pathogenic variants by whole-exome sequencing. We estimated frequency of rare single-nucleotide variants in the coding region of GLIS3 in a Caucasian population and among individuals with carbohydrate metabolism disorders in Russia. We identified 15 single-nucleotide variants in GLIS3. Three rare variants (minor allele frequency < 1%) rs806052, rs143051164, and rs149840771 were genotyped in 126 cases of MODY, in 188 patients with type 2 diabetes mellitus (DM2), and 564 randomly selected Caucasian individuals in Russia. A heterozygous rs806052 variant was identified in one patient with DM2; c.1270T frequency was 0.003. Prevalence of rs143051164 c.844G was 0.003 in the control population and 0.004 and 0.003 in MODY and DM2 samples, respectively. Prevalence of rs149840771 c.2096A was 0.003 and 0.004 in the control population and among MODY patients, respectively. In DM2 patients, rs149840771 c.2096A was not identified. We did not detect any associations of rs806052, rs143051164, and rs149840771 with carbohydrate metabolism disorders among patients with MODY and DM2 in Russia.


Introduction
Maturity onset diabetes of the young (MODY) is a hereditary form of diabetes with autosomal dominant inheritance and is characterized by onset at a young age and the presence of an initial defect of pancreatic β-cell function. This type of diabetes differs from classic types of diabetes mellitus-type 1 (DM1) and type 2 (DM2)-in disease progression, in treatment strategies, and prognosis [1]. To date, 13 types of MODY (MODY1 through MODY13) have been identified, each associated with mutations in a specific gene: HNF4A, GCK, HNF1A, PDX1, HNF1B, NEUROD1, KLF11, CEL, PAX4, INS, BLK, KCNJ11, and ABCC8 [2]. Dysfunction of other genes causes 11-30% of cases of MODY and are defined as MODY-X [3]. Previously, some of GLIS3 gene polymorphisms have been shown to be associated with the development of MODY [4].
The gene codes for a Krüppel-like zinc finger transcription factor that is located in a nuclear compartment of the cell and in the Golgi complex. GLIS3 expression in the embryo shows spatial and temporal dependence on embryogenesis stages, whereas in adults it is tissue-specific and depends on cell type. GLIS3 expression occurs Open Access BMC Research Notes *Correspondence: 2117409@mail.ru; niitpm.office@gmail.com 1 Research Institute of Internal and Preventive Medicine-Branch of Institute of Cytology and Genetics, Siberian Branch of Russian Academy of Sciences, Bogatkova Str. 175/1, Novosibirsk 630004, Russia Full list of author information is available at the end of the article mainly in kidneys and in the pancreas, thyroid, thymus, uterus, ovaries, brain, and lungs [5].
GLIS3 plays a critical role in the regulation of many physiological processes (insulin gene expression, β-cell formation, thyroid hormone biosynthesis, spermatogenesis, and kidney function) [6]. It is one of the most differentially expressed genes in β-cells between healthy people and patients with DM2 [7]. GLIS3 mRNA expression is decreased by 50% in pancreatic islets of patients with DM2 in comparison with healthy people [8]. It is known that mutations in GLIS3-binding sites of the INS gene promoter cause neonatal diabetes [9]. All singlenucleotide variants (SNVs) of GLIS3 that have been associated with DM1 and DM2 in genome-wide association studies are located in noncoding regions in contrast to the SNVs associated with neonatal diabetes. It is thought that these substitutions may regulate GLIS3 gene expression. Probably, they are associated with β-cells' ability to respond to immune, viral, and metabolic stressors and have made the GLIS3 gene a common member of the genetic risk factors of DM1 and DM2 [10].
The aim of this study was identification of probably pathogenic variants of GLIS3 in patients with MODY using a whole-exome sequencing approach and estimation of the frequency of rs806052, rs143051164, and rs149840771 in a Russian population and DM2 patients.

Methods
The study protocol was approved by the local Ethics Committee of the Institute of Internal and Preventive Medicine (a branch of the Institute of Cytology and Genetics, the Siberian Branch of the Russian Academy of Sciences, Novosibirsk, Russia). Written informed consent to be examined and to participate in the study was obtained from each patient. For individuals younger than 18 years, the informed consent was signed by a parent or legal guardian.
The inclusion criteria were as follows: signing informed consent for participation in the study, a verified diagnosis of diabetes, a debut of the disease for probands at the age of 35 years and earlier, the absence of obesity, absence of pancreas islet cell antibodies, absence of glutamic acid decarboxylase antibodies, sufficient secretion of β-cells, absence of necessity of insulin therapy, and the absence of ketoacidosis at the onset of the disease. The following exclusion criteria were applied: a history of tuberculosis of lungs or other organs or of infection with the human immunodeficiency virus, an infectious disease caused by a hepatitis B virus or a hepatitis C virus that requires antiviral treatment, substance abuse or alcoholism for 2 years before the examination.
In the first phase of the study, whole-exome sequencing was performed on 21 probands with a MODY phenotype and five of their first-degree relatives (three with a MODY phenotype) for analysis of the GLIS3 mutation spectrum. A total of 26 patients underwent a full medical examination; a biochemical blood test; determination of HbA1c, C-peptide, anti-GAD (glutamic acid decarboxylase) antibodies, pancreas islet cell antibodies (ICA), thyroid status, and microalbuminuria; abdominal and renal ultrasonography; real-time continuous glucose monitoring with Medtronic Paradigm MMT-722; and genetic tests. Genomic DNA was isolated from leukocytes of venous blood by phenol-chloroform extraction [11]. Quality of the extracted DNA was assessed on a capillary electrophoresis system (Agilent 2100 Bioanalyzer; Agilent Technologies Inc., USA). Sequencing of the first group of samples (11 patients) was conducted on an Ion Proton instrument (Thermo Fisher Scientific). Whole-exome libraries were prepared with the AmpliSeq Exome Kit (Thermo Fisher Scientific). The data-processing pipeline included mapping the reads to the reference human genome (GRCh37) using the Burrows-Wheeler Aligner (BWA) v.0.7.12 software followed by SNV calling performed by means of GATK. SNV annotation was conducted in the IonReporter 4.0 software. The average depth of coverage was 30×. Sequencing in the second group of patients (15 individuals, respectively) was carried out on an Illumina HiSeq 1500 instrument (Illumina, San Diego, CA, USA). The enrichment and library preparation were performed using the SureSelectXT Human All Exon V5 + UTRs Kit. Reads were mapped to the reference human genome (GRCh37) in the BWA v.0.7.12 software. PCR duplicates were removed by means of the Picard software. A search for polymorphisms was conducted using the Genome Analysis Toolkit v.3.3 package by the procedure for local remapping of short insertions/ deletions and recalibration of the read quality. The depth of coverage was from 34× to 53×.
In sequenced groups, we filtered sequence variants if they were present in 10 or more variant reads with a quality score ≥ 30. In this study, the search for polymorphic sites covered known MODY genes and GLIS3. A comparison of the detected substitutions was performed with a number of specialized databases: the 1000 Genomes Project (http://www.1000g enome s.org/), dbSNP database (https ://www.ncbi.nlm.nih.gov/SNP/), and ClinVar (http://clinv ar.com/). We selected the spectrum of rare sequence variants in MODY genes (nonsynonymous, alternative splice site, and coding indels variants) and only nonsynonymous SNVs for common substitutions. Rare variants were selected if their minor allele frequency (MAF) was ≤ 1% in the 1000 Genomes database and if they were present in the dbSNP database.
In the second phase of analysis, we selected only the functional variants with MAF < 0.5% and a potential pathogenic effect: rs806052, rs143051164, and rs149840771. In most cases, the MODY-causing SNVs are rare variants with low MAFs and cannot be highly distributed in the population; therefore, we focused on these substitutions. Verification for rs806052, rs143051164, and rs149840771 was performed by Sanger sequencing and oligonucleotides primers for SNVs were designed in the Primer-Blast software (https ://www.ncbi.nlm.nih.gov/tools /prime r-blast /). The oligonucleotides are shown in Additional file 1. These SNVs were tested by real-time PCR (Additional file 1) in 564 randomly selected individuals (Caucasians in Russia), in 188 patients with DM2, and in 126 patients with a MODY phenotype ( Table 1).
The population group included in the analysis was selected from a survey of the population interviewed within the framework of the HAPIEE project [12], Novosibirsk, Russia (9360 participants, aged 45-69 years, men 50%, Caucasians 97%). A total of 564 randomly selected patients (some may have diabetes, age 54.2 ± 0.4, mean ± SD) were included.
The DM2 group consisted of 68 women and 87 men (age 59.0 ± 6.7, mean BMI among men: 31, women: 32). The DM2 sample was randomly selected from the HAP-IEE participants [12]. We used the criteria of the American Diabetes Association [13] for the diagnosis of DM2.
A total of 126 patients with a MODY phenotype (age 23.8 ± 2.6) underwent the same medical examination as described above.
Statistical data processing was performed in the IBM SPSS Statistics 23.0 software. Validity of differences in allele prevalence between groups was determined using the χ 2 criterion. Differences were considered statistically significant at p < 0.05.

Results
The spectrum of GLIS3 gene polymorphism in patients with a MODY phenotype was identified by the wholeexome sequencing method. During this analysis, 15 SNV were found ( Table 2). Most of these SNVs are common variants and located in noncoding exon-intron boundaries. Rs806052 is located in a DNA-binding domain, rs143051164 in the region of potential interaction with proteins ISL1, ITCH, Smurf 2, Smurf 1, DNAJA1, and HIPK3 [14], and rs149840771 is located near the transactivation C-terminal domain of the GLIS3 protein.
The GG rs806052 (c.1270T>C) genotype in the homozygous form was identified in our group of Caucasians in Russia and in all patients with MODY (Table 3). This variant is present with a rare allele in one patient with MODY, but after validation it was not confirmed. Nevertheless, we tested this SNV in our groups because it had been found in patients with DM2 and seemed interesting [15]. In the DM2 group, a heterozygous rs806052 variant was identified in one male patient without a family history of diabetes mellitus. Prevalence rates of rs806052, rs143051164, and rs149840771 are presented in Table 3.
Frequencies of rs806052, rs143051164, and rs149840771 in the Russian population are close to the ones described in databases.
Prevalence of rare allele C rs143051164 (c.844C>G) in the Caucasian population was 0.003, in the MODY group 0.004, and in the DM2 group 0.003. In the Caucasian group, a heterozygous genotype was identified in three individuals without carbohydrate metabolism disorders and without a family history of diabetes mellitus.
Prevalence of rare allele T rs149840771 (c.2096G>A) in the Caucasian population was 0.003, and in the MODY group, 0.004. In the DM2 group, the rare allele of rs149840771 was not identified.

Discussion
A number of studies have shown that GLIS3 plays a role in the regulation of at least four different processes associated with development and normal functioning of the pancreas. Regulation of differentiation of pancreatic islets at an early stage of embryogenesis via activation of transcription factor neurogenin 3 (Ngn3) is involved in β-cell development. Impairment of this process leads to neonatal diabetes. Control of β-cell proliferation under stressful conditions in adults by the regulation of cyclin D2 (CCND2) expression is important for a compensatory increase in β-cell mass during obesity. Impairment of this regulatory step causes the development of DM2. Two more processes-βcell apoptosis regulation by activation of SRp55 splicing factor expression and regulation of insulin gene expression-can perform a major function at all stages of organismal development [16]. There are more than 10 genes, and many variants of these genes have been found to be associated with diabetes phenotypes. These data allowed us to suppose that substitutions in exons of GLIS3 are associated with MODY-X or are phenotype modifiers in DM2.
No differences were identified in allele frequencies of the analyzed SNV of GLIS3 in patients with confirmed MODY and the other two groups (patients with DM2 and the Caucasian population). We did not find any novel mutation in coding and splicing site regions of the GLIS3 gene.
Considering the lack of data on a correlation between rare exon SNVs rs806052, rs143051164, and rs149840771 of the GLIS3 gene and the development of MODY and DM2 in this study, we suppose that these variants are not associated with these pathological phenotypes in Russia. It seems reasonable to analyze a noncoding part of this gene by searching for SNVs associated with different forms of diabetes mellitus.

Limitations
The GLIS3 gene mutation spectrum did not show substantial differences among the different tested samples. Whole-exome sequencing did not reveal new SNVs in GLIS3 of patients with MODY in Russia.
The results of genotyping of the Caucasian group and patients with carbohydrate metabolism disorders allow   us to suggest that rare SNVs rs806052, rs143051164, and rs149840771 of the GLIS3 gene do not take part in the pathogenesis of MODY and DM2 in Russia. One of the limitations of our study is the low sample size. Another limitation is analysis of SNVs only in the coding region of GLIS3 because there are some SNVs in the noncoding region that are associated with DM1 and DM2 [10].