Introduction
Diabetes mellitus (DM) is a chronic metabolic disease caused by abnormal glucose metabolism, which is mainly characterized by hyperglycaemia. There are approximately 382 million people affected with DM worldwide, and type 2 diabetes mellitus (T2DM) accounts for 90% of DM patients [1, 2]. Prediabetes refers to blood glucose levels above normal but below diabetes thresholds. The prevalence of prediabetes is increasing worldwide, and it is estimated that 470 million people will suffer from prediabetes in 2030 [3]. In China, it is reported that the overall prevalence of diabetes is 10.9% and that for prediabetes it is 35.7% [4]. It has been demonstrated that prediabetes is a high-risk state of diabetes, and it also increases the risk of myocardial infarction, stroke, and cardiovascular death [5]. There is accumulating evidence to demonstrate that prediabetes could cause damage to the kidneys and the nervous system [6, 7]. Additionally, prediabetes imposes a huge economic burden on individuals and society [8]. Therefore, effective prevention strategies for prediabetes are increasingly important.
Proinsulin (PI), the precursor form of insulin (I), is synthesized and secreted in pancreatic b-cells. PI only accounts for 10–20% of fasting insulin (FI) under physiological conditions. However, some research indicates that the level of PI was highly expressed in glucose-intolerant and insulin-resistant individuals [9, 10]. Also, fasting proinsulin (FPI) has been demonstrated to be associated with impaired b-cell function, T2DM and insulin resistance, and it could be used as a specific predictor of T2DM [10, 11]. Insulin is a well-known hormone to reduce the level of blood glucose via the stimulation of glucose uptake into muscle cells and adipocytes, etc. by binding to its receptor in the target cells. It has been shown that elevated fasting insulin (FI) is a hallmark of T2DM [12]. Our previous study demonstrated that FPI, 2-hour postprandial proinsulin (2hPI), FI, and 2-hour postprandial insulin (2hI) were associated with an increased risk of prediabetes [13]. Despite these findings, it is still unclear how these common phenotypes affect T2DM.
In this study, we performed a genome-wide association study (GWAS) of FPI, 2hPI, FI, and 2hI in 451 prediabetes subjects from the Chinese Han population. The BiosystemsTM AxiomTM Precision Medicine Diversity Array (PMDA) was used to identify single nucleotide polymorphisms (SNPs) associated with FPI, 2hPI, FI, and 2hI. Our study will provide an effective diagnostic method for early screening of people who are susceptible to T2DM, and for controlling and preventing the development of prediabetes to T2DM.
Material and methods
Participants
In this study, we recruited 451 prediabetes subjects aged ≥ 18 years from the Hainan Affiliated Hospital of Hainan Medical University. Participants with 100 mg/dL (5.6 mmol/L) ≤ fasting plasma glucose < 125 mg/dL (6.9 mmol/L) or 5.7% ≤ glycated haemoglobin (HbA1c) < 6.4% were defined as prediabetes [14]. Individuals without a history of diabetes and malignant tumours, or severe liver and kidney diseases were included in this research. This study was conducted with ethical approval from the Hainan Affiliated Hospital of Hainan Medical University Ethics Committees, and was performed in line with the Declaration of Helsinki. We also obtained consent forms signed by each participant.
Metabolic variables
Fasting blood samples were collected from all subjects after an overnight fast. The levels of fasting plasma glucose (FPG), FPI, FI, 2hPI, and 2hI were examined by an automatic biochemical analyser.
Genotyping and imputation
Genomic DNA was isolated from a whole blood sample using a DNA Extraction Kit (GoldMag Co. Ltd., Xi’an, China), as described previously [15]. The Applied BiosystemsTM Axiom TM Precision Medicine Diversity Array (PMDA, Thermo Scientific, USA), the Gene Titan Multi-Channel instrument, and Axiom Analysis Suite 6.0 Software were used for genotyping.
Genotype data in subjects was cleaned using standard thresholds (HWE p > 5 × 10-6, call rate > 95%). Imputation for chromosomes 1 to 22 was performed with IMPUTE 2.0 software from HapMap 1000 Genomes Phase 3 as a reference panel.
Statistical analyses
The association analysis was conducted using Gold Helix SNP and Variation Suite 8.7 software. The association between SNPs and FPI, 2hPI, FI, and 2hI was evaluated using linear regression assuming an additive genetic model. The 4 traits were analysed with adjustments for age and sex. A p < 5.0 × 10-6 was used as the threshold of genome-wide significance.
Results
A total of 451 prediabetes individuals aged ≥ 18 years (216 men and 235 women) were included and genotyped in the present study. The average age of the subjects was 51.78 ± 14.49 years. The clinical parameters of participants are summarized in Table 1.
Variable |
Subjects |
Number of individuals |
451 |
Age (years, mean ± SD) |
51.78 ± 14.49 |
Gender |
|
Male |
216 (47.9%) |
Female |
235 (52.1%) |
FPG [mmol/L] |
5.88 ± 1.42 |
FPI [mU/L] |
15.74 ± 12.18 |
2hPI [mU/L] |
63.42 ± 44.10 |
FI [mU/L] |
72.10 ± 43.08 |
2hI [mU/L] |
578.22 ± 435.40 |
As presented in Table 2, we found that 6 loci in 3 genes (DLG1-AS1, SORCS1, CTAGE11P) reached genome-wide significance associated with FPI, and 27 SNPs in 3 genes (ZNF718, MARCHF2, and HNRNPM) were associated with 2hPI. In addition, the correlation of 6 SNPs in the KRT71 gene with FI reached genome-wide significance. Also, 14 SNPs in 3 genes (UBE2U, ABO, and GRID1-AS1) were genome-wide significant in their relationship with 2hI. The distributions of association p-values for FPI, 2hPI, FI, and 2hI are presented in Figure 1 (the quantile-quantile plots and Locus zoom are shown in Fig. S1 and Fig. S2). Among these, the genetic loci of CTAGE11P, MARCHF2, KRT71, and ABO have the strongest association with FPI, 2hPI, FI, and 2hI, respectively.
Table 2. Significant loci associated with fasting proinsulin (FPI), 2h proinsulin (2hPI), fasting insulin (FI), and 2h insulin (2hI) in study populations |
|||||||||
Gene |
Traits |
Description |
SNP |
Chr |
Position |
Allele |
Minor allele |
MAF |
p |
DLG1-AS1;LINC02012 |
FPI |
DLG1 antisense RNA1 DLG1 antisense RNA1 |
rs78022276 |
3 |
197482798 |
T/C |
T |
0.029 |
2.88E-06 |
DLG1-AS1;LINC02012 |
FPI |
rs78750477 |
3 |
197483030 |
C/G |
C |
0.029 |
2.88E-06 |
|
SORCS1 |
FPI |
Sortilin related VPS10 domain containing receptor |
rs58879794 |
10 |
106889263 |
C/A |
C |
0.150 |
3.94E-06 |
CTAGE11P |
FPI |
CTAGE family member 11, preudogene |
rs9600432 |
13 |
75107716 |
A/C |
A |
0.453 |
1.42E-06 |
CTAGE11P |
FPI |
rs9565135 |
13 |
75122830 |
A/G |
A |
0.454 |
3.75E-06 |
|
CTAGE11P |
FPI |
rs9543852 |
13 |
75124005 |
T/G |
T |
0.454 |
2.51E-06 |
|
ZNF718 |
2hPI |
Zinc finger protein 718 |
rs56128594 |
4 |
188333 |
T/C |
T |
0.446 |
5.88E-08 |
ZNF718 |
2hPI |
rs4690234 |
4 |
192200 |
T/C |
T |
0.446 |
2.41E-08 |
|
MARCHF2 |
2hPI |
Membrane associated ring-CH-type finger 2 |
rs12979798 |
19 |
8419748 |
G/A |
G |
0.395 |
4.05E-07 |
MARCHF2 |
2hPI |
rs12978137 |
19 |
8420144 |
C/T |
C |
0.405 |
2.02E-08 |
|
MARCHF2 |
2hPI |
rs62117527 |
19 |
8421448 |
C/T |
C |
0.405 |
2.02E-08 |
|
MARCHF2 |
2hPI |
rs11259979 |
19 |
8435167 |
C/T |
C |
0.444 |
8.99E-07 |
|
MARCHF2 |
2hPI |
rs12975669 |
19 |
8435935 |
T/G |
T |
0.445 |
1.20E-06 |
|
MARCHF2 |
2hPI |
rs35562870 |
19 |
8436208 |
C/T |
C |
0.445 |
8.84E-07 |
|
HNRNPM |
2hPI |
Heterogeneous nuclear ribonuclaoprotein M |
rs17160491 |
19 |
8448056 |
T/G |
T |
0.469 |
9.47E-07 |
HNRNPM |
2hPI |
rs2081197 |
19 |
8448452 |
A/C |
A |
0.439 |
7.83E-07 |
|
HNRNPM |
2hPI |
rs11666117 |
19 |
8449010 |
A/C |
A |
0.441 |
6.15E-07 |
|
HNRNPM |
2hPI |
rs11259983 |
19 |
8450491 |
A/C |
A |
0.447 |
4.20E-07 |
|
HNRNPM |
2hPI |
rs868781681 |
19 |
8450732 |
T/A |
T |
0.460 |
2.17E-06 |
|
HNRNPM |
2hPI |
rs200358539 |
19 |
8450735 |
T/A |
T |
0.459 |
1.97E-06 |
|
HNRNPM |
2hPI |
rs17160495 |
19 |
8451394 |
A/T |
A |
0.447 |
4.20E-07 |
|
HNRNPM |
2hPI |
rs11259985 |
19 |
8451793 |
A/T |
A |
0.447 |
4.20E-07 |
|
HNRNPM |
2hPI |
rs34337793 |
19 |
8454339 |
A/G |
A |
0.447 |
3.61E-07 |
|
HNRNPM |
2hPI |
rs34244685 |
19 |
8458122 |
T/C |
T |
0.441 |
6.40E-07 |
|
HNRNPM |
2hPI |
rs3764570 |
19 |
8463393 |
A/G |
A |
0.446 |
3.38E-07 |
|
HNRNPM |
2hPI |
rs3794997 |
19 |
8465003 |
A/T |
A |
0.446 |
3.38E-07 |
|
HNRNPM |
2hPI |
rs34445564 |
19 |
8468214 |
A/T |
A |
0.439 |
6.01E-07 |
|
HNRNPM |
2hPI |
Heterogeneous nuclear ribonuclaoprotein M |
rs17159302 |
19 |
8469632 |
A/C |
A |
0.439 |
6.01E-07 |
HNRNPM |
2hPI |
rs17159303 |
19 |
8469677 |
G/A |
G |
0.446 |
3.38E-07 |
|
HNRNPM |
2hPI |
rs74180130 |
19 |
8479935 |
C/T |
C |
0.452 |
1.19E-06 |
|
HNRNPM |
2hPI |
rs17160520 |
19 |
8483705 |
G/A |
G |
0.453 |
1.01E-06 |
|
HNRNPM |
2hPI |
rs2277987 |
19 |
8487389 |
A/G |
A |
0.440 |
5.76E-08 |
|
HNRNPM |
2hPI |
rs1599870 |
19 |
8488516 |
G/A |
G |
0.440 |
6.62E-08 |
|
KRT71 |
FI |
Keratin 71 |
rs12308719 |
12 |
52548451 |
G/T |
G |
0.482 |
2.31E-06 |
KRT71 |
FI |
rs10876309 |
12 |
52548517 |
C/T |
C |
0.482 |
2.31E-06 |
|
KRT71 |
FI |
rs3803084 |
12 |
52548843 |
A/G |
A |
0.495 |
1.02E-06 |
|
KRT71 |
FI |
rs3803085 |
12 |
52548910 |
C/T |
C |
0.483 |
1.08E-06 |
|
KRT71 |
FI |
rs4761930 |
12 |
52549360 |
G/A |
G |
0.487 |
1.52E-06 |
|
KRT71 |
FI |
rs4761933 |
12 |
52555091 |
C/T |
C |
0.491 |
2.12E-06 |
|
UBE2U |
2hI |
Ubiquitin conjugating enzyme E2 U |
rs11585260 |
1 |
64315831 |
G/C |
G |
0.024 |
4.94E-06 |
UBE2U |
2hI |
|
rs11577590 |
1 |
64315842 |
C/G |
C |
0.024 |
4.94E-06 |
ABO |
2hI |
Alpha 1-3-N-acetylgalactosaminyltransferase and alpha 1-3-galactosaminyltransferase |
rs9411372 |
9 |
133258677 |
A/G |
A |
0.138 |
9.59E-07 |
ABO |
2hI |
rs977371848 |
9 |
133266456 |
T/C |
T |
0.163 |
1.56E-07 |
|
ABO |
2hI |
rs879055593 |
9 |
133271182 |
T/C |
T |
0.163 |
1.56E-07 |
|
ABO |
2hI |
rs992108547 |
9 |
133273983 |
A/G |
A |
0.163 |
1.56E-07 |
|
ABO |
2hI |
rs947073006 |
9 |
133274414 |
A/G |
A |
0.163 |
1.56E-07 |
|
ABO |
2hI |
rs600038 |
9 |
133276354 |
C/T |
C |
0.159 |
7.95E-07 |
|
ABO |
2hI |
rs651007 |
9 |
133278431 |
T/C |
T |
0.159 |
3.84E-07 |
|
ABO |
2hI |
rs579459 |
9 |
133278724 |
C/T |
C |
0.159 |
3.84E-07 |
|
ABO |
2hI |
rs495828 |
9 |
133279294 |
T/G |
T |
0.159 |
3.84E-07 |
|
ABO |
2hI |
rs635634 |
9 |
133279427 |
T/C |
T |
0.159 |
3.84E-07 |
|
LINC01520;GRID1-AS1 |
2hI |
GRID1 antisense RNA1 |
rs375709957 |
10 |
85558056 |
T/A |
T |
0.176 |
1.87E-06 |
LINC01520;GRID1-AS1 |
2hI |
rs77136415 |
10 |
85558059 |
T/C |
T |
0.176 |
1.87E-06 |
Discussion
The current study illustrated that the genetic variants of CTAGE11P, MARCHF2, KRT71, and ABO were significantly correlated with FPI, 2hPI, FI, and 2hI in Chinese Han people, respectively. Our research will provide scientific methods and ideas for the prevention and diagnosis of prediabetes, and it will contribute to controlling and reducing the progression of prediabetes to T2DM.
Recently, GWAS was performed by Strawbridge et al., which found that 9 SNPs in 8 genes were associated with FPI levels in the European population [16]. Subsequently, Huyghe et al. also identified low-frequency coding variants associated with FPI at SGM2 and MADD gene in Finnish males [17]. Moreover, it is suggested that IGF-1 genetic variants were associated with FI in European ancestry [18]. Manning et al. also observed that 6 SNPs in COBLL1-GRB14, IRS1, PPP1R3B, PDGFC, UHRF1BP1, and LYPLAL1 are correlated with the FI level [19]. However, those SNPs explained only a small percentage of the total variation in FPI and FI. In the present study, we found 6 SNPs in DLG1-AS1, SORCS1, and CTAGE11P for FPI, 27 SNPs in ZNF718, MARCHF2, and HNRNPM for 2hPI, 6 SNPs in KRT71 for FI, and 14 SNPs in UBE2U, ABO, and GRID1-AS1 for 2hI. Among these, the genetic variants of CTAGE11P, MARCHF2, KRT71, and ABO have the strongest association with FPI, 2hPI, FI, and 2hI.
The E3 ubiquitin ligase membrane-associated ring-CH-type finger 2 (MARCHF2) is a member of the membrane-associated RING-CH E3 ubiquitin ligase family (MARCH) and localizes to the endoplasmic reticulum and Golgi [20]. The known substrate of MARCHF2 includes cystic fibrosis transmembrane conductance regulator (CFTR) [21]. Some studies have indicated that patients with CFTR gene variants show an insufficiency of insulin secretion, leading to the development of DM [22, 23]. Moreover, Khan et al. found that inhibition of CFTR decreased the concentrations of plasma insulin and pancreatic insulin in CFTR-inhibited animals [24]. Another study demonstrated that the mutation of CFTR is associated with insulin resistance and decreased b-cell mass in mice [25]. This evidence led us to believe that MARCHF2 is involved in the development of pancreas and DM by interacting with CFTR.
Keratin 71 (KRT71) is a member of the keratin family and is located on chromosome 12q13.13. Keratin constitutes the intermediate filament proteins of epithelial cells. It is documented that the loss of keratin 8 decreased fasting blood glucose levels, and increased glucose uptake and glycogen synthesis [26, 27]. The abnormal expression of keratin 1 and 10 reduced insulin secretion, thus leading to the development of DM [28].
The ABO gene encodes glycosyltransferases that catalyse the transfer of nucleotide donor sugars to the H antigen to form the A and B antigens. Variation in the ABO gene is the basis of the ABO blood group. Meo et al. found that blood group “B” is associated with a higher risk of T2DM, while blood group “O” has a weak correlation with T2DM [29]. Also, a GWAS reported that ABO variants are associated with increased levels of plasma lipid and soluble intercellular adhesion molecule 1 and tumour necrosis factor 2 (TNF-2). These molecules could affect insulin and its receptors and contribute to the development of DM [30].
CTAGE family member 11 pseudogene (CTAGE11P) belongs to the cutaneous T-cell lymphoma-associated antigen (CTAGE) family and is located on 13q22.2. It is reported that the mutation of family members reduces cholesterol and triglyceride levels in mice [31]. Another family member can regulate the plasma low-density lipoprotein-cholesterol concentration and is associated with coronary artery disease [32]. Our study found for the first time that CTAGE11P genetic variants are associated with FPI in the Chinese people.
Conclusions
We found that the genetic variants of CTAGE11P, MARCHF2, KRT71, and ABO are significantly correlated with FPI, 2hPI, FI, and 2hI in Chinese Han people, respectively. These genetic variants may serve as new biomarkers for the prevention of prediabetes.
Conflict of interest
All authors declare that they have no competing interests.
Funding
This study received the support of the major research and development program of Hainan Province (no. ZDYF2021SHFZ078), and received the support of project supported by Hainan Province Clinical Medical Center.
Acknowledgments
The authors thank all participants and volunteers in this study. We also thank the Hainan Affiliated Hospital of Hainan Medical University for their help with sample collections.
Data availability statement
The data that support the findings of this study are available from the supporting information files of this manuscript.
Ethical approval
This study was conducted with ethical approval from the Hainan Affiliated Hospital of Hainan Medical University Ethics Committees, and it was performed in line with the Declaration of Helsinki. We also obtained consent forms signed by each participant.
Consent to participate
Not applicable.
Code availability
Not applicable.
Authors’ contributions
Le.L. and H.Q. designed this study protocol and drafted the manuscript; T.F. and Lu.L. performed the DNA extraction and genotyping; Q.O. performed the data analysis; H.Z. performed the sample collection and information recording; K.C. and Z.Z. revised the manuscript; H.Q. conceived and supervised the study. All authors read and approved the final manuscript.