Pleiotropic Effects of Common and Rare GCKR Exonic Mutations on Cardiometabolic Traits

Background: The common non-synonymous mutation of the glucokinase regulator (GCKR) gene, namely rs1260326, is widely reported to have pleiotropic effects on cardio-metabolic traits and hematological parameters. Objective: This study aimed to identify whether other GCKR variants may have pleiotropic effects independent of the rs1260326 genotypes. Methods: In total, 81,097 Taiwan Biobank participants were enrolled for the regional plot association studies and candidate variant analysis of the region around the GCKR gene. Results: The initial candidate variant approach showed the significant association of the rs1260326 genotypes with multiple phenotypes. Regional plot association analysis of the GCKR gene region further revealed genome-wide significant associations between GCKR variants and serum total and low-density lipoprotein cholesterol; triglyceride, uric acid, creatinine, aspartate aminotransferase, γ-Glutamyl transferase, albumin, and fasting plasma glucose levels; estimated glomerular filtration rate; leukocyte and platelet counts; microalbuminuria, and metabolic syndrome, with rs1260326 being the most common lead polymorphism. Serial conditional analysis identified genome-wide significant associations of two low-frequency exonic mutations, rs143881585 and rs8179206, with high serum triglyceride and albumin levels. In five rare GCKR exonic non-synonymous or nonsense mutations available for analysis, GCKR rs146175795 showed an independent association with serum triglyceride and albumin levels and rs150673460 showed an independent association with serum triglyceride levels. Weighted genetic risk scores from the combination of GCKR rs143881585 and rs146175795 revealed a significant association with metabolic syndrome. Conclusion: In addition to the rs1260326 variant, low-frequency and rare GCKR exonic mutations exhibit pleiotropic effects on serum triglyceride and albumin levels and the risk of metabolic syndrome. These results provide evidence that both common and rare GCKR variants may play a critical role in predicting the risk of cardiometabolic disorders.


Introduction
The liver is the major organ responsible for both the disposal of oral glucose load (by sensing portal glucose signals, which increase net hepatic glucose uptake) and hypoglycemia (through the initiation of gluconeogenic and glycogenolytic pathways) to maintain healthy blood glucose concentrations [1]. Glucokinase (GCK) is a glucose sensor Figure 1. Study flowchart of inclusion and exclusion criteria used to screen Taiwan Biobank project participants. Other phenotypes include age; body mass index; waist circumference; waist-hip ratio; aspartate aminotransferase, alanine aminotransferase, γ-glutamyl transferase, and serum creatinine levels; estimated glomerular filtration rate; serum albumin; total bilirubin; hemoglobin; hematocrit; red blood cell, leukocyte, and platelet counts; and blood urea nitrogen, albuminuria, microalbuminuria, and metabolic syndrome. Abbreviations: QC, quality control; HL, hyperlipidemia; HTN, hypertension; DM, diabetes mellitus; HbA1C, hemoglobin A1C; SBP, systolic blood pressure; DBP, diastolic blood pressure; MBP, mean blood pressure; LDL-C, low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol; TG, triglyceride; T-CHO, total cholesterol; UA, uric acid.

Clinical Phenotypes and Laboratory Examinations
Demographic data used for the analysis included waist circumference, waist-hip ratio, body mass index (BMI), and systolic, mean, and diastolic blood pressure. Biochemistry data used for the analysis included glucose metabolism parameters such as fasting plasma glucose level and hemoglobin A1C; lipid profiles, namely total, high-density lipoprotein (HDL), and low-density lipoprotein (LDL) cholesterol, and triglyceride levels; and liver Figure 1. Study flowchart of inclusion and exclusion criteria used to screen Taiwan Biobank project participants. Other phenotypes include age; body mass index; waist circumference; waist-hip ratio; aspartate aminotransferase, alanine aminotransferase, γ-glutamyl transferase, and serum creatinine levels; estimated glomerular filtration rate; serum albumin; total bilirubin; hemoglobin; hematocrit; red blood cell, leukocyte, and platelet counts; and blood urea nitrogen, albuminuria, microalbuminuria, and metabolic syndrome. Abbreviations: QC, quality control; HL, hyperlipidemia; HTN, hypertension; DM, diabetes mellitus; HbA1C, hemoglobin A1C; SBP, systolic blood pressure; DBP, diastolic blood pressure; MBP, mean blood pressure; LDL-C, low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol; TG, triglyceride; T-CHO, total cholesterol; UA, uric acid.

Clinical Phenotypes and Laboratory Examinations
Demographic data used for the analysis included waist circumference, waist-hip ratio, body mass index (BMI), and systolic, mean, and diastolic blood pressure. Biochemistry data used for the analysis included glucose metabolism parameters such as fasting plasma glucose level and hemoglobin A1C; lipid profiles, namely total, high-density lipoprotein (HDL), and low-density lipoprotein (LDL) cholesterol, and triglyceride levels; and liver and renal functional test-related parameters, namely creatinine, uric acid, aspartate aminotransferase (AST), alanine aminotransferase (ALT), γ-glutamyl transferase (γGT), albumin, and total bilirubin levels. BMI and estimated glomerular filtration rate (eGFR) were calculated as previously reported [30]. Hematological parameters analyzed included leukocyte, red blood cell, and platelet counts and hematocrit and hemoglobin levels. Because of the absence of urine creatinine level, only spot urine albumin level was used for urine albumin evaluation.

Selection of GCKR Variants and Genotyping
DNA was isolated from blood samples by using a PerkinElmer chemagic 360 instrument following the manufacturer's instructions (PerkinElmer, Waltham, MA, USA). SNP genotyping was conducted using custom TWB chips and performed on the Axiom Genome-Wide Array Plate System (Affymetrix, Santa Clara, CA, USA). The GCKR variant rs1260326 was initially analyzed, followed by other low-frequency and rare exonic mutations selected for further study (Supplementary Table S1). In this paper, we have used a minor allele frequency (MAF) of <0.01 as rare exonic mutations, as suggested by Wang, et al. [31], and MAF between 0.05 and 0.01 as low-frequency exonic mutations.

Regional Plot Association Analysis
To identify the lead single-nucleotide polymorphisms (SNPs) around the GCKR gene region for various studied phenotypes, we performed a regional plot association analysis by using the data of TWB participants enrolled after QC for GWAS and applying the other exclusion criteria ( Figure 1). The Axiom Genome-Wide CHB 1 and 2 Array Plates (Affymetrix, Inc., Santa Clara, CA, USA), comprising 611,656 and 640,160 SNPs, each from 24,927 and 69,529 participants, respectively, were applied for analysis. With 1000 Genomes Project Phase 3 East Asian populations used as a reference panel, genome-wide genotype imputation was performed using SHAPEIT (version 2, Oxford, UK, https://mathgen. stats.ox.ac.uk/genetics_software/shapeit/shapeit.html, accessed on 2 December 2020) and IMPUTE2 (version 2, Oxford, UK, http://mathgen.stats.ox.ac.uk/impute/impute_v2.html, accessed on 2 December 2020). QC was performed after imputation through the filtration of SNPs with IMPUTE2 imputation quality scores of >0.3. Indels were removed using VCFtools (version 0.1, https://vcftools.github.io/index.html, accessed on 2 December 2020). All the samples enrolled for the analysis had a call rate of ≥97%. For SNP QC, the criteria for exclusion from subsequent analyses included an SNP missing rate of <3%, an MAF of <0.01, and a violation of Hardy-Weinberg equilibrium (p < 10−6). Finally, 81,097 participants and 139 SNPs were used for regional plot association analysis of the GCKR gene region on chromosome 2p23.3 at positions ranging between 27.62 and 27.85 Mb.

Selection of Rare Exonic GCKR Mutations from the Pre-QC Imputation Data for Analysis
During QC for regional plot association analysis, rare variants with MAF of less than 0.01 were excluded. To test for the role of rare exonic mutations in genotype-phenotype associations, we selected variants from the pre-QC imputation data for analysis. A total of five rare exonic GCKR mutations were enrolled for the genotype-phenotype association analysis, including four non-synonymous mutations (rs146175795, p.Val103Met; rs150673460, p.Pro132Leu; rs1414321043, Ala314Ser; and rs146285804, p.Trp517Cys) and one nonsense mutation (rs149847328, p.Arg227Ter) (Supplementary Table S1).

Statistical Analysis
Continuous variables were expressed as mean ± standard deviation. When the distribution was strongly skewed, median and interquartile ranges were given, which were tested using a two-sample t test or analysis of variance. Differences in categorical data distribution were examined using a chi-squared test or chi-squared test for trend. Before analysis, all study parameters were logarithmically transformed to adhere to a normality assumption. We assumed the genetic effect to be additive after adjustment for age, sex, BMI, and current smoking status, and a general linear model was used to analyze the studied phenotypes in relation to the predictors of investigated genotypes and confounders. Regional plot association studies with conditional analysis were conducted using the analysis software package PLINK (version 1.07, Shaun Purcell, Cambridge, MA, USA, https://zzz.bwh.harvard.edu/plink/, accessed on 14 August 2021). Conditional analysis is a tool to identify secondary association signals at a candidate locus by adjusting the index variant in the region [32,33]. For the regional plot association study, conditional analysis was conducted by testing the residual association with all remaining SNPs, and another round of conditional analysis can be performed by adjusting the second index variant in this region for independent signals. Genome-wide significance was defined by p < 5 × 10 −8 . For Bonferroni correction of regional plot associational analysis, the significant value was defined as p < 1.0 × 10 −5 , calculated as 0.05/(139 × 33), according to a total of 139 variants and 33 traits analyzed. For Bonferroni correction of each genotypephenotype analysis with rare mutations, we used a more liberal threshold of p < 2.9 × 10 −4 , calculated as 0.05/5 × 33, according to a total of 5 rare variants and 33 traits analyzed. For weighted genetic risk score (WGRS), we weighted the SNPs in each allele score by using the β coefficients from our association analysis, and the risk allele was selected with directionally concordant associations of target parameters. The LDmatrix (https://analysistools.nci.nih.gov/LDlink/?tab=ldmatrix, accessed on 19 April 2021) was used for the analysis of linkage disequilibrium (LD). SPSS (version 22; SPSS, Chicago, IL, USA) was used to perform all calculations.

Association of GCKR rs1260326 Genotypes with Clinical, Metabolic, and Biochemical Phenotypes and Hematological Parameters
The data of the GCKR exonic mutations enrolled for the analysis are presented in Supplementary Table S1. In total, >80,000 volunteers participated in the genotype-phenotype association analysis (Table 1). By using an additive model, after adjustment of age, sex, BMI, and smoking status, genome-wide significant associations were found for rs1260326 genotypes with total and LDL cholesterol and triglyceride levels, fasting plasma glucose, serum uric acid levels, renal functional parameters (serum creatinine and urine albumin levels and eGFR), liver functional parameters (AST, γGT, and serum albumin), and leukocyte and platelet counts, whereas associations with p < 1.0 × 10 −5 were noted for systolic and mean blood pressure. The C allele of the rs1260326 variant is associated with higher systolic and mean blood pressure; serum total and LDL cholesterol, triglyceride, uric acid, AST, γGT, and albumin levels; eGFR; and leukocyte and platelet counts, and lower fasting plasma glucose and serum creatinine levels.

Association of GCKR rs1260326 Genotypes with Risk Factors for Atherosclerosis
We analyzed the association of rs1260326 genotypes with atherosclerotic risk factors. In our study, after adjustment for sex, age, BMI, and current smoking status, hypertension, microalbuminuria, and metabolic syndrome were found to be significantly associated with rs1260326 genotypes ( Table 2). The C allele of the rs1260326 genotype is associated with a higher risk of hypertension, microalbuminuria, and metabolic syndrome.

with Study Phenotypes
Regional plot association analyses were performed to determine the association of genetic variants around the GCKR gene region at positions 27.62-27.85 Mb on chromosome 2p23.3 with study phenotypes. Our data revealed that the lead SNP for each phenotype was situated at or near the GCKR gene region, revealing pleiotropic effects on this gene locus ( Figure 2, Supplementary Figures S1-S20). Regional plot association analysis showed genome-wide significant associations between GCKR variants and serum total and LDL cholesterol; triglyceride, uric acid, creatinine, AST, γ-GT, albumin, and fasting plasma glucose levels; eGFR; urine albumin levels, and leukocyte and platelet counts, microalbuminuria, and metabolic syndrome (Table 3).

Regional Plot Association Studies for Determining the Associations of Genetic Variants at Positions 27.62 to 27.85 Mb on Chromosome 2p23.3 with Study Phenotypes
Regional plot association analyses were performed to determine the association of genetic variants around the GCKR gene region at positions 27.62-27.85 Mb on chromosome 2p23.3 with study phenotypes. Our data revealed that the lead SNP for each phenotype was situated at or near the GCKR gene region, revealing pleiotropic effects on this gene locus ( Figure 2, Supplementary Figures S1-S20). Regional plot association analysis showed genome-wide significant associations between GCKR variants and serum total and LDL cholesterol; triglyceride, uric acid, creatinine, AST, γ-GT, albumin, and fasting plasma glucose levels; eGFR; urine albumin levels, and leukocyte and platelet counts, microalbuminuria, and metabolic syndrome (Table 3).

Linkage Disequilibrium between GCKR Gene Region SNPs
The rs1260326 was the most common lead SNP for lipid profile, blood pressure status, fating plasma glucose, serum uric acid, albumin, and AST levels, urine albumin levels, and metabolic syndrome ( Table 3). Most of the other lead SNPs had a strong LD with the rs1260326 variant (all r 2 > 0.82), whereas for the phenotypes of leukocyte count and serum alanine aminotransferase (ALT) levels, the lead SNPs rs6744393 and rs12989678, respectively, had a moderate LD with the rs1260326 variant (r 2 = 0.537 and r 2 = 0.476, respectively; Figure 3). Further, after serial conditional analysis was conducted, the lead SNPs rs143881585 and rs8179206 for serum triglyceride and albumin levels showed a weak LD with the lead SNPs (all r 2 < 0.015 for the rs143881585 variant and maximal r 2 = 0.029 for the rs8179206 variant). The linkage disequilibrium map of the GCKR variants is presented in Figure 3. Genes 2022, 13, x FOR PEER REVIEW 9 of 19

Association of GCKR rs143881585 and rs8179206 Genotypes with Clinical, Metabolic, and Biochemical Phenotypes, Hematological Parameters, and Risk Factors for Atherosclerosis
We further tested the association of rs143881585 and rs8179206 genotypes with the study phenotypes (Supplementary Tables S2-S5). After adjustment for age, sex, current smoking, and BMI, our data revealed that individuals with the A and G alleles of the rs143881585 and rs8179206 variants, respectively, had significantly higher serum triglyceride and albumin levels (p = 1.17 × 10 −10 and p = 7.00 × 10 −6 , respectively, for the rs143881585 variant and p = 0.0046 and p = 1.19 × 10 −4 , respectively, for the rs8179206 variant; Figure 4). The associations of the rs143881585 genotypes became more significant after further adjustment of the rs1260326 genotypes (p = 3.69 × 10 −22 and p = 1.24 × 10 −10 , respectively, for serum triglyceride and albumin levels; Supplementary Tables S2 and S3 and Figure 4C,D). With further adjustment of both rs1260326 and rs143881585 genotypes for the conditional analysis, genome-wide significant associations were noted between the G allele of the rs8179206 variant and higher serum triglyceride and albumin levels (p = 3.89 × 10 −8 and p = 3.11 × 10 −8 , respectively; Supplementary Tables S4 and S5 and Figure  4E,F). We further tested the association of rs143881585 and rs8179206 genotypes with the study phenotypes (Supplementary Tables S2-S5). After adjustment for age, sex, current smoking, and BMI, our data revealed that individuals with the A and G alleles of the rs143881585 and rs8179206 variants, respectively, had significantly higher serum triglyceride and albumin levels (p = 1.17 × 10 −10 and p = 7.00 × 10 −6 , respectively, for the rs143881585 variant and p = 0.0046 and p = 1.19 × 10 −4 , respectively, for the rs8179206 variant; Figure 4). The associations of the rs143881585 genotypes became more significant after further adjustment of the rs1260326 genotypes (p = 3.69 × 10 −22 and p = 1.24 × 10 −10 , respectively, for serum triglyceride and albumin levels; Supplementary Tables S2 and S3 and Figure 4C,D). With further adjustment of both rs1260326 and rs143881585 genotypes for the conditional analysis, genome-wide significant associations were noted between the G allele of the rs8179206 variant and higher serum triglyceride and albumin levels (p = 3.89 × 10 −8 and p = 3.11 × 10 −8 , respectively; Supplementary Tables S4 and S5 and Figure 4E

Association between Rare GCKR Exonic Mutations and Clinical Phenotypes and Laboratory Parameters
From the pre-QC imputation data, five rare exonic nonsynonymous or nonsense mutations were selected for genotype-phenotype association analysis (Supplementary Tables S6-S15). All the rare GCKR mutations showed a weak LD with the other genotypes (r 2 < 0.01) (Supplementary Figure S21). After adjustment for age, sex, body mass index (BMI), and current smoking, in genotype-phenotype association analysis, rs146175795 genotypes showed significant associations with serum triglyceride and albumin levels (p = 1.50 × 10 -5 and p = 3.50 × 10 -5 , respectively, Supplementary Tables S8 and S9, Figure 4G,H) and rs150673460 genotypes showed significant associations with serum triglyceride levels (p = 2.70 × 10 -5 ; Supplementary Table S10, Figure 4I).

Association between Rare GCKR Exonic Mutations and Clinical Phenotypes and Laboratory Parameters
From the pre-QC imputation data, five rare exonic nonsynonymous or nonsense mutations were selected for genotype-phenotype association analysis (Supplementary Tables S6-S15). All the rare GCKR mutations showed a weak LD with the other genotypes (r 2 < 0.01) (Supplementary Figure S21). After adjustment for age, sex, body mass index (BMI), and current smoking, in genotype-phenotype association analysis, rs146175795 genotypes showed significant associations with serum triglyceride and albumin levels (p = 1.50 × 10 −5 and p = 3.50 × 10 −5 , respectively, Supplementary Tables S8 and S9, Figure 4G,H) and rs150673460 genotypes showed significant associations with serum triglyceride levels (p = 2.70 × 10 −5 ; Supplementary Table S10, Figure 4I).

WGRS from the Combination of GCKR rs143881585 and rs1461755795 Revealed Significant Association with Metabolic Syndrome
In addition to rs1260326, the low-frequency and rare GCKR variants rs143881585 and rs146175795 were linked to metabolic syndrome, but the p values (0.0402 and 0.0005, respectively) fell short of the statistical significance cut-off (1.0 × 10 −5 ). We wanted to see if the combination of these two variations had an influence on metabolic syndrome risk. We discovered that the combined low-frequency and rare variants rs143881585 and rs146175795 in the GCKR gene are highly linked with the risk of metabolic syndrome (p = 1.83 × 10 −6 ) using WGRS from rs143881585 and rs146175795 variants (Supplementary Table S16).

Discussion
In this investigation, the combined approaches of analyzing candidate gene variants and a regional plot association study were used to confirm the pleiotropic effect of the rs1260326 genotypes on multiple quantitative traits and diseases. Through regional plot association analysis, all lead SNPs showed moderate to strong LD with the rs1260326 variant. With serial conditional analysis, we demonstrated novel genome-wide significant associations of GCKR exonic mutations rs143881585 and rs8179206 with serum triglyceride and albumin levels. With further rare GCKR exonic mutation analysis, GCKR rs146175795 also showed a significant association with serum triglyceride and albumin levels and rs150673460 showed an independent association with serum triglyceride levels. In combination, the aforementioned variants contributed to 1.02% and 0.45% of serum triglyceride and albumin levels, respectively. All these results showed the same direction of elevated serum triglyceride and albumin levels with mutation alleles ( Figure 5). We then tested the effect of combination of the low-frequency and rare GCKR variants by using WGRS analysis and found a significant association with metabolic syndrome. To our knowledge, this is the first report to describe novel associations of multiple low-frequency and rare GCKR variants with serum triglyceride and albumin levels and metabolic syndrome independent of the rs1260326 variant. With the combination of candidate variants and regional plot association studies, these results may provide further evidence for understanding the critical role of the GCKR gene in the risk of cardiometabolic disorders.  Figure 1 and Table 2.

Pleiotropic Effect of GCKR Gene Locus
Kanai et al. [16] showed that the GCKR region is one of the most pleiotropic regions and is associated with 18 quantitative traits. The NHGRI-EBI GWAS Catalog is a publicly available resource for GWASs, and it contains useful visualizations of variant-trait associations. All variants are mapped onto chromosomal positions on the human genome (https://www.ebi.ac.uk/gwas). When the data of the GWAS Catalog were reviewed, genome-wide significant associations with the GCKR region variants were noted, with the associations including various demographic factors (body height, BMI, lean body mass, and heart rate), cardiometabolic traits, liver and kidney functional tests and diseases, lifestyle factors (such as alcohol consumption, coffee consumption, and dietary factors), biomarker and hormone levels (such as C reactive protein, leptin, YKL40, testosterone, and estradiol levels), hematological parameters, coagulation factors, electrolyte and metabolite levels, and various diseases (such as gallstone, urolithiasis, and age-related diseases). Moreover, our data revealed the association of GCKR gene variants with multiple quantitative traits and diseases, in which 19 of them have genome-wide significance (p < 5 × 10 −8 ). All these data support the notion that the GCKR gene and its encoded protein GKRP are involved in the functions of multiple organs and systems, revealing its critical role in mediating homeostasis in the human body. In addition to the pleiotropic effect of the GCKR gene locus, our findings confirmed a recent finding in a Korean population that many cardiometabolic traits had a common genetic foundation [34]. The relevance of understanding pleiotropy and its consequences for genetic testing and personal genomics has been highlighted by the expanding use of genetic information in clinical practice.

Bidirectional Effects of GCKR rs1260326 Variant on Associated Phenotypes
With the pleiotropic effect of GCKR variants, the allelic effects of SNPs conferred both favorable and unfavorable outcomes on the human body. As previously reported [17][18][19][20][21][22][23][24][25][26], our data revealed that the C allele carrier of the rs1260326 variant is associated with multiple unfavorable phenotypes, such as higher systolic, diastolic, and mean BP and risk of Figure 5. Genomic structure of GCKR variants and their association with various phenotypes/diseases. Green arrow: association with low-frequency and rare GCKR variants. Blue arrow: association with rs1260326 genotypes. Abbreviations: MetS, metabolic syndrome; MA, microalbuminuria; PLT, platelet. Other abbreviations as in Figure 1 and Table 2.

Pleiotropic Effect of GCKR Gene Locus
Kanai et al. [16] showed that the GCKR region is one of the most pleiotropic regions and is associated with 18 quantitative traits. The NHGRI-EBI GWAS Catalog is a publicly available resource for GWASs, and it contains useful visualizations of variant-trait associations. All variants are mapped onto chromosomal positions on the human genome (https://www.ebi.ac.uk/gwas, accessed on 2 December 2020). When the data of the GWAS Catalog were reviewed, genome-wide significant associations with the GCKR region variants were noted, with the associations including various demographic factors (body height, BMI, lean body mass, and heart rate), cardiometabolic traits, liver and kidney functional tests and diseases, lifestyle factors (such as alcohol consumption, coffee consumption, and dietary factors), biomarker and hormone levels (such as C reactive protein, leptin, YKL40, testosterone, and estradiol levels), hematological parameters, coagulation factors, electrolyte and metabolite levels, and various diseases (such as gallstone, urolithiasis, and age-related diseases). Moreover, our data revealed the association of GCKR gene variants with multiple quantitative traits and diseases, in which 19 of them have genome-wide significance (p < 5 × 10 −8 ). All these data support the notion that the GCKR gene and its encoded protein GKRP are involved in the functions of multiple organs and systems, revealing its critical role in mediating homeostasis in the human body. In addition to the pleiotropic effect of the GCKR gene locus, our findings confirmed a recent finding in a Korean population that many cardiometabolic traits had a common genetic foundation [34]. The relevance of understanding pleiotropy and its consequences for genetic testing and personal genomics has been highlighted by the expanding use of genetic information in clinical practice.

Bidirectional Effects of GCKR rs1260326 Variant on Associated Phenotypes
With the pleiotropic effect of GCKR variants, the allelic effects of SNPs conferred both favorable and unfavorable outcomes on the human body. As previously reported [17][18][19][20][21][22][23][24][25][26], our data revealed that the C allele carrier of the rs1260326 variant is associated with multiple unfavorable phenotypes, such as higher systolic, diastolic, and mean BP and risk of hypertension and microalbuminuria, elevated levels of liver function indicators (such as serum ALT, AST, and γGT levels), higher total and LDL cholesterol and triglyceride and uric acid levels, as well as a higher risk of metabolic syndrome and gout. Furthermore, the C allele carrier of the rs1260326 variant is associated with preferable phenotypes, such as a lower fasting plasma glucose level and higher eGFR and serum albumin levels. These results are consistent with those reported previously [17][18][19][20][21][22][23][24][25] and the diverse and bidirectional effects of the rs1260326 variant reveal that a more cautious approach should be considered when GCKR is used as a target of drug therapy.

Association between the rs143881585 Variant and Serum Triglyceride and Albumin Levels Is Independent of the rs1260326 Variant
The rs143881585 variant is a low-frequency one in the Taiwanese population, with a minor allele frequency (MAF) of 1.19% in TWB project participants. According to the PUBMed.gov website, the allele frequency of rs143881585 was 0% in 1006 participants from the 1000 Genomes project and 0.102% in 60,706 unrelated individuals from The Exome Aggregation Consortium, with exome sequencing data obtained from diverse large-scale sequencing projects. These results suggest that rs143881585 may be a rare variant in other populations. Furthermore, the association of rs143881585 genotypes with serum triglyceride and albumin levels was partially suppressed by rs1260326, with the association becoming stronger after adjustment for rs1260326 genotypes. In addition, rs143881585 is a synonymous variant (c.354 G > A, p.Ser118Ser) that lies at the branch point between exon 4 and intron 5. Whether this variant affects gene splicing is unknown. Thus, a further mechanistic study may be necessary to elucidate the molecular basis of the rs143881585 variant's association with serum triglyceride and albumin levels.

Association of the rs8179206 Variant with Serum Triglyceride and Albumin Levels Is Independent of the rs1260326 Variant
The SNP rs8179206 (c. 296 A to G, p.Glu77Gly), a previously reported, rare GCKR variant, is a non-synonymous variation located on exon 3 of the GCKR gene [28,29]. From the dbSNP database, the MAFs for European Descent participants of 1000 Genomes Phase V1, African, and East Asian populations were 0%, 0%, and 2.3%, respectively, suggesting that this variant is rare in non-East Asian populations. In our study, the frequency of minor allele rs8179206 was 2.78%, similar to those previously reported in East Asian populations, including Chinese [35]. In terms of function, the rs8179206 variant was classified as a putative loss-of-function mutation with reduced nuclear localization of GKRP, GKRP expression, and F6P binding [28,29]. Chinese studies with a relatively small sample size have shown no significant association of SNP rs8179206 with obesity, type 2 diabetes mellitus, and various lipid profiles [35][36][37]. With >80,000 TWB project participants for analysis, we provide the first evidence regarding the genome-wide association of this variant with serum triglyceride and albumin levels.

Role of Rare GCKR Exonic Mutations in Serum Triglyceride Levels
Rare nonsynonymous variants in the GCKR gene have been associated with variations in metabolic traits, particularly serum triglyceride levels [28,29,38]. A recent study highlighted the clinical relevance of the collective burden of rare alleles in GCKR, reporting that non-synonymous variants with an MAF of <0.01 in the GCKR gene are enriched in cases of extreme hypertriglyceridemia [38]. When exonic Sanger sequencing of the GCKR gene in 800 individuals (mostly of non-Hispanic mixed European descent) was applied, 19 nonsynonymous rare GCKR variants were detected, which, in combination, showed significantly higher serum triglyceride levels compared to those without such variants [28]. However, rare loss-of-function GCKR variants do not co-segregate with increased plasma triglyceride levels in families [29]. By contrast, in two South Asian population-specific, functionally disruptive, rare GCKR non-synonymous mutations, rs774930016 (p.Ser105Asn) and rs55537970 (p.Arg553Trp), significantly increased risks of hypertriglyceridemia have been reported [39]. Rees et al. [28] subdivided rare GCKR variants into three classes according to cellular localization, cellular interaction with GCK, and kinetic effects. Several reports have shown GCKR rs146175795 (p.Val103Met) in patients with hypertriglyceridemia [28,[40][41][42]. The GCKR p.Val103Met mutation was predicted to be disease-causing by in silico algorithms (SIFT, Poly-Phen-2, and FATHMM). The rs146175795 variant was also classified as a putative loss-of-function mutation with reduced nuclear localization of GKRP, GKRP expression, and F6P binding [28]. Jin et al. [40] performed genetic analysis for 103 Chinese patients with very high triglyceride levels and found that 46 patients had rare pathogenic/potential pathogenic variants in 15 triglyceride-related genes, in which four of them had the GCKR p.Val103Met mutation. The MAF of the GCKR p.Val103Met mutation has been shown to be <0.001 in European and African populations and 0.0069 in East Asian populations according to the PUBMed.gov website. In our study participants, the MAF was 0.0065. In addition to serum triglyceride levels, our results also showed a genome-wide significant association of GCKR p.Val103Met mutation with serum albumin levels. One non-synonymous exonic mutation, rs150673460, has not been previously reported and was found to be associated with serum triglyceride levels in our study cohort.

GCKR Variants and Serum Albumin Levels
Previous GWAS studies have shown GCKR as a candidate gene locus for serum albumin levels, with rs1260326 as the lead SNP [13,43,44]. Our data also revealed the consistent association of several low-frequency and rare GCKR variants with elevated serum albumin levels. Albumin is a multifaced protein synthesized in the hepatocytes that has many physiological properties, including anti-inflammatory, antioxidant, anticoagulant, and anti-platelet aggregation activities, and serum albumin has been assigned as an indicator of malnutrition [45]. In a meta-analysis of more than one million apparently healthy adult participants, elevated serum albumin levels were associated with a reduced risk of hypertension, adverse vascular events, all-cause mortality, certain cancers, and fracture, revealing serum albumin as a biomarker in determining the risk of adverse cardiometabolic outcomes [46]. Using a GWAS study, Loomis et al. [47] further showed that GCKR rs1260326 genotypes are significantly associated with percent glycated albumin, which is in the inverse direction of the association with serum albumin levels. Glycated albumin, a ketoamine formed by the non-enzymatic glycation of serum albumin, is calculated as the percentage of serum albumin that is in glycated form. A significant and negative correlation was found between glycated albumin and serum albumin levels [48], and the serum glycated albumin value decreased by 0.23% with every 1g/dL increase in serum albumin [49]. Glycated albumin has been suggested to be an alternative to HbA1C in situations when HbA1C is not recommended for plasma glucose monitoring, such as hemoglobinopathies, pregnancy, or chronic kidney disease [50][51][52]. Increase glycated albumin values have been associated with diabetic nephropathy, neuropathy, and cardiovascular complications [53]. With increased serum albumin levels, the GCKR variants may also decrease the glycated albumin values. Thus, the GCKR variants may affect the cardiometabolic outcomes not only by increasing serum albumin levels but also by decreasing glycated albumin values. Further studies may help to elucidate more of the potential roles for the GCKR variants in serum albumin levels, serum glycated albumin values, and cardiometabolic outcomes.

Association between GCKR Variants and Metabolic Syndrome
Previous studies have focused on gene-environmental factor (obesity or diet) interactions explaining 'missing heritability' for phenotypic variance in MetS risk [54,55], accounting for strong but paradoxical relationships of GCKR polymorphisms with lower fasting plasma glucose and higher TG levels among MetS components. This study looked into other possibilities, such as significantly larger numbers of variants with lower effects that have yet to be discovered or rarer variants (perhaps with larger impacts) that are poorly detected by existing genotyping arrays. Because four low-frequency and rare GCKR variants have a strong and persistent connection with serum triglyceride levels, aggregation of their effects may increase the risk of metabolic syndrome independently of the established risk of common polymorphism rs1260326. Our findings, by the combination of GCKR variants rs143881585 and rs1461755795, add to the growing body of evidence that GSKR plays a critical role in the development of metabolic syndrome.

Conclusions
Our investigation confirmed the pleiotropic effect of the GCKR variants, particularly the GCKR rs1260326. Moreover, multiple low-frequency and rare GCKR variants were found to contribute to various aspects of cardiometabolic traits, such as serum triglyceride and albumin levels and metabolic syndrome, in this Taiwanese population, independent of the rs1260326 variant. A larger population study with meta-analysis and trans-ethnic population analysis may help to further elucidate the critical role of GCKR variants in cardiometabolic traits and diseases.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/genes13030491/s1. Method S1: Definitions of hypertension, diabetes mellitus, obesity, current smoking, microalbuminuria and metabolic syndrome. Table S1: GCKR exonic mutations in Taiwan Biobank participants: Data derived from the Axiom Genome-Wide CHB 1 and 2 Array plates with genotype imputation; Table S2: Association of GCKR rs143881585 genotypes with clinical phenotypes and laboratory parameters; Table S3: Association of GCKR rs143881585 genotypes with atherosclerotic risk factors; Table S4: Association of the GCKR rs8179206 genotypes with clinical phenotypes and laboratory parameters; Table S5: Association of the GCKR rs8179206 genotypes with atherosclerotic risk factors; Table S6: Association between GCKR rs1414321043 genotypes and metabolic and hematological phenotypes; Table S7: Association between GCKR rs1414321043 genotypes and atherosclerotic risk factors; Table S8. Association of the GCKR rs146175795 genotypes with metabolic and hematological phenotypes; Table S9. Association of the GCKR rs146175795 genotypes and atherosclerotic risk factors; Table s10: Association of the GCKR rs150673460 genotypes with metabolic and hematological phenotypes; Table S11: Association of the GCKR rs150673460 genotypes and atherosclerotic risk factors; Table S12: Association of the GCKR rs149847328 genotypes with metabolic and hematological phenotypes; Table S13: Association between GCKR rs149847328 genotypes and atherosclerotic risk factors; Table S14: Association of the GCKR rs146285804 genotypes with metabolic and hematological phenotypes; Table S15: Association of the GCKR rs146285804 genotypes and atherosclerotic risk factors; Table S16: Logistic regression analysis for metabolic syndrome, including genotypes, in 81,097 participants. Figure S1