According to the Chinese Guidelines for the Prevention and Treatment of Type 2 Diabetes (2020 Edition), the diagnostic criteria for diabetes are typical diabetes symptoms plus random blood glucose ≥ 11.1 mmol/L; Or add fasting blood glucose ≥ 7.0mmol/L; Or add OGTT 2h blood sugar ≥ 11.1mmol/L; Or add HbA1C≥6.5%. Typical diabetes symptoms include polydipsia, polyuria, polyphagia and unexplained weight loss. Excluding T1DM and special types of diabetes, the patients with T2DM were included in the study. At the early stage of disease, it is sometimes difficult to determine the type of diabetes. If the classification cannot be determined immediately, a temporary classification can be carried out to guide the treatment. Then, according to the patients' initial response to treatment and the clinical manifestations during follow-up, the patients were re-evaluated and classified (20).

Coronary atherosclerotic heart disease refers to heart disease caused by myocardial ischemia, hypoxia or necrosis due to stenosis or occlusion of lumen caused by coronary atherosclerosis, which is referred to as coronary heart disease (CHD) for short (21). The diagnostic criteria of patients with CHD in this study were implemented in accordance with Chinese Guidelines for Clinical Diagnosis and Treatment of Coronary Heart Disease (2010 Edition) (22).

According to the Guidelines for Primary Diagnosis and Treatment of Hypertension in China (2019 Edition), without using antihypertensive drugs, the blood pressure should be measured three times on different days, with systolic blood pressure (SBP) ≥140 mmHg and/or diastolic blood pressure (DBP) ≥90 mmHg. SBP ≥ 140 mmHg and DBP <90 mmHg are simple systolic hypertension. The patient has a history of hypertension, and is currently taking antihypertensive drugs. Although the blood pressure is lower than 140/90 mmHg, he or she is still diagnosed as hypertension (23).

This was a retrospective study involving patients with T2DM. The experimental design was approved by the ethics committee of the Institute of Clinical Pharmacology, Central South University and was registered at http://www.chictr.org.cn (Registration number: ChiCTR1800015661). Participants in this study were randomly enrolled from inpatients at the Second Xiangya Hospital of Central South University between December 2017 and December 2019. Our study included patients who were diagnosed with T2DM on admission and aged between 18 and 80 years. We collected clinical data for all participants, including gender, age, height, weight, body mass index (BMI), waist circumference, hip circumference, waist-to-hip ratio, systolic blood pressure (SBP), diastolic blood pressure (DBP), cardiovascular history, cerebrovascular history, smoking history, drinking history, hypertension, hyperglycemia, family diabetes history, triglyceride (TG) level, cholesterol (CHOL) level, high density lipoprotein-cholesterol (HDL-CH) level, low density lipoprotein-cholesterol (LDL-CH) level, and fasting blood glucose level. We also recorded the levels of GLU-60, GLU-120, glycosylated hemoglobin (HbA1C), 25-hydroxyvitamin D, Cpst-0, Cpst-60, Cpst-120, along with the glomerular filtration rate (eGFR) and medication status (hypoglycemic drugs, lipid-lowering drugs, antihypertensive drugs). The clinical endpoint events included type 2 diabetes vascular complications, blood glucose, blood pressure, TG, CHOL, HDL-CH, and LDL-CH. We excluded patients whose discharge diagnosis was not T2DM and those from which the blood samples were not obtained. We also excluded patients who did not provide informed consent and those did not have a complete set of clinical data.

We identified two gene chip datasets (GSE22255 and GSE58294) relating to cardioembolic stroke patients in the GEO database (Supplementary Table 1). We merged the two raw datasets and applied RMA standardization. Next, we used the R-Limma package to process differential gene analysis, |log₂FC| > 1.5 and p < 0.05 were considered to be statistically significant. Then, we used the ENCODE database to assess the potential function of SNPs within or near candidate genes. We included candidate SNPs with a minor allele frequency ≥ 5% and a pairwise linkage disequilibrium (r²) < 0.30 within the same and adjacent genes [1,000 Genomes phase 3 Han Chinese in Beijing (CHB)].

Peripheral venous blood was collected from all patients who met the inclusion criteria. DNA was extracted from peripheral venous blood using the E.Z.N.A. SQ blood DNA Kit II (Omega Bio-Tek Company, USA). The extracted DNAs were then stored at −80°C wait for analysis. The Sequenom MassARRAY SNP system was used to screen the genotypes of all candidate SNPs (Bioyong Technologies Inc.). Information relating to the probe is shown in Supplementary Table 2. Finally, 5% of the participants were randomly selected for verification by Sanger sequencing.

An immortalized hepatocyte cell line, MIHA, was cultured in Roswell Park Memorial Institute (RPMI) 1640 Medium supplemented with 10% fetal bovine serum (FBS). Cells were then cultured in a standard humidified incubator at 37°C in a 5% CO₂ atmosphere.

Plasmids containing an empty vector, and the GRB10 mutation were subcloned into a lentivirus vector (pHBLV-CMV-MCS-3FLAG-EF1-ZsGreen-T2A-PURO) to construct the recombinant plasmid (Hanheng Biotechnology, Shanghai, China). The GTB10-Mut plasmid featured an rs1800504 mutation (allele C to T). The detailed information of the construction of GRB10 mutant lentiviral vector is shown in the Supplementary Files 2. MIHA cells were transfected with Lentiviral vector using Polybrene (Hanheng Biotechnology,Shanghai, China) according to the manufacturer's instructions.When the expression of GRB10 reached its peak, the cell lines with stable expression were screened by Puromycin.

The protocol used for western blotting was described in previous report (24). The membranes were first probed with GRB10 (#3702) and GAPDH (#5174S) primary antibodies purchased from Cell Signaling Technology (CST). Following incubation, membranes were probed with a horseradish peroxidase (HRP)-labeled anti-rabbit secondary antibody from CST (#2708) (diluted with 5% BSA to 1:1,000). Antibody binding was subsequently detected by an enhanced chemiluminescence detection kit (ECL) (Biosharp Biotechnology, BL520B).

Total RNA was extracted from MIHA cells using an RNA-Quick Purification Kit (ES Science). Complementary DNA (cDNA) was synthesized using ReverTra Ace reverse transcriptase (Novoprotein, E04710A) in accordance with the manufacturer's instructions. Real-time RT-PCR was performed with a SYBR qPCR SuperMix Plus Kit (Novoprotein, E096) on a 7300 plus/Bio-RAD iCycler in accordance with the manufacturer's instructions. The primer sequences were as follows: GRB10, forward, 5′-CGAAACTCACCCGTCCAG-3′; reverse, 5′-GGATTCACAGTGTCGGTTGG-3; β-actin: forward, 5′-CATGTACGTTGCTATCCAGGC-3′; reverse, 5′-CTCCTTAATGTCACGCACGAT-3′. The gene expression levels for each amplicon were calculated by the ΔΔCT method and normalized against β-actin mRNA.

According to the Trinder reaction principle, cholesterol makes 4-aminoantipyrine react with phenol (PAP) to produce a red quinoneimine pigment. Using this strategy, we used a TC kit operating standard (BB-47435) and a Microplate Reader to determine the absorbance of samples at a wavelength of 510 nm.

Oil red O can be used to stain lipids in cells. Positive staining changes the color of fat from orange to red. After 24 h, the cells were washed twice with phosphate buffered saline (PBS) and fixed with 4% paraformaldehyde at room temperature for 30 min. After removing the fixative, the cells were washed twice with distilled water and soaked in 60% isopropanol for 5 min. After discarding the isopropanol, we added freshly prepared ORO stain (ORO stain:distilled water; 3:2) for 10–20 min. The staining solution was then discarded, and the cells were washed with water twice. We then added Mayer hematoxylin staining solution to stain the nuclei for cycler 1–2 min, the staining solution was then washed away and replaced with water so that we could monitor the developing staining effect. Finally, we used distilled water to cover the cells and observed the cells with an EVOS microscope (Life Technologies/Thermo Fisher Scientific, US).

SPSS Statistical software (version 22.0 for windows, Chicago, IL) was used for statistical analysis and GraphPad Prism version 5 software (GraphPad Prism Software Inc., La Jolla, CA) was used to create figures. Power and sample size calculations software (Version 3.0.43) was used to calculate the sample size required for the study. The T-test or the Mann–Whitney U-test were used for the statistical analysis of measurement data (mean ± standard deviation), as appropriate. The chi-squared or Fisher's exact test was used to determine whether the gene distribution conformed to the Hardy–Weinberg equilibrium. Logistic regression analysis was used to calculate the odds ratio (OR) and 95% confidence interval (CI) between the GRB10 single-nucleotide polymorphisms (SNPs) and diabetic vascular complications after adjusting for age, BMI, gender, hypertension, smoking and drinking. Comparison of biochemical indicators variance between the GRB10 genotype was performed with Univariate Analysis of Variance (ANOVA) after adjusting for potential confounders, such as age, BMI and gender. Linkage disequilibrium (LD) and haplotypes analyses were performed using SHEsis online software (http://analysis.bio-x.cn/myAnalysis.php). p < 0.05 (2-tailed) was considered to indicate statistical significance.

We collected clinical data and matched DNA samples from 1,026 Chinese patients with an initial diagnosis of T2DM. Following final diagnosis, 70 cases were classified as non-T2DM, and 22 DNA samples were not genotyped. Finally, 934 patients were included in the study. According to our research scheme, eight candidate SNPs in GRB10 were screened, including rs1800504, rs2237460, rs17133917, rs55834323, rs4947710, rs4245555, rs9791817, and rs9791887 (The SNP genotyping data has been uploaded to the DRYAD, https://datadryad.org/stash/share/pDGln0dQK5R8YEMJKAZLJEFs6xWtE7xOi8rD9frUSYA). In total, 919 patients were successfully genotyped with all SNPs (Figure 1). The detailed results related to the loci of these SNPs are shown in Supplementary Table 3. Analysis showed that rs9791887, rs9791817, and rs424555, were not consist with the Hardy-Weinberg Equilibrium. Next, we analyzed the SNP linkage disequilibrium of GRB10. Results showed that there was strong linkage among rs9791887-rs4245555, rs9791887-rs9791817, rs55834323-rs4947710, and rs4245555-rs9791817 (Supplementary Figure 1). The baseline clinical characteristics of participants with rs1800504 are shown in Table 1. There were significant differences in BMI; p = 0.02.

We analyzed the relationships between the eight candidate SNPs in GRB10 and T2DM cardiovascular complications. We found that rs1800504 genetic variation was significantly related to the occurrence of CHD in T2DM patients; p = 0.011. However, there was no significant association with other cardiovascular complications (Table 2). In this study, 930 patients were successfully genotyped for rs1800504. In the recessive model, the CC+CT genotypes were associated with a significantly increased risk of CHD compared with the TT genotype (OR: 2.24; 95% CI: 1.36–3.70; p = 0.002; Table 3). Other SNPs had no significant effect on the CHD of T2DM patients (Supplementary Table 4).

Moreover, we analyzed the relationship between GRB10 rs1800504 genetic variation and biochemical indicators. We found that rs1800504 variation was associated with plasma lipids level differences in T2DM patients. There were significant differences in the levels of CHOL and LDL-CH when compared between different rs1800504 genotypes. The levels of CHOL in the TT and CC+CT genotypes were 4.10 ± 1.00 mmol/L and 4.44 ± 1.25 mmol/L, respectively; p = 0.009. The levels of LDL-CH for the TT and CC+CT genotypes were 2.53 ± 0.82 mmol/L and 2.81 ± 1.07 mmol/L, respectively; p = 0.01 (Figure 2 and Table 4). However, none of significant effect was observed as for GRB10 rs1800504 mutation on blood glucose or blood pressure. Haplotype analysis also showed that the haplotypes of these SNPs did not have a significant impact on the relevant clinical endpoint events (Supplementary Table 5).

To further determine the role of GRB10 rs1800504 in liver lipid metabolism, we performed verification experiments in vitro with normal hepatocyte MIHA cells. We constructed a GRB10 site-directed mutagenesis lentiviral vector and transfected the empty vector, GRB10-WT, and GRB10-Mut into MIHA cells (Figure 3A). We found that the expression of GRB10 in MIHA cells was significantly increased after transfection, and that the expression levels of cells transfected with GRB10-Mut were significantly higher than the levels of cells transfected with GRB10-WT (Figures 3B,C). Next, we measured the levels of TC in each group. Following the transfection of GRB10, the TC level in MIHA cells was significantly reduced, the levels of TC in the cells transfected with GRB10-Mut were 2–3 times lower than those in cells transfected with GRB10-WT (Figure 3D). Previous studies have shown that palmitic acid (PA) can significantly induce lipid accumulation in hepatocytes. We found that levels of the GRB10 protein were significantly decreased in MIHA cells that had been treated with PA (100 μM) (Figure 3E). After adding PA (100 μM) to MIHA cells transfected with GRB10-WT, GRB10-Mut, and the empty vector, we found that the overexpression of GRB10 reversed the formation of lipid droplets induced by PA, and we also discovered that the inhibitory effect of the GRB10-Mut vector had the most significant effect (Figure 3F).