The MR analysis of mitochondrial genome-wide cis-eQTL and CKD.Through SMR analysis, this study identified a unique genetic locus associated with CKD from mitochondrial-related genes, as well as 6 distinct genetic loci associated with eGFRcrea (Fig. 2). It’s important to note the slight differences in how these results were presented-CKD risk was typically represented in OR format, while the risk for eGFRcrea was directly presented in the form of βmitodys−outcome. These variations stem from the distinct data types and analytical methodologies employed. These findings garnered consistent support from HEIDI tests, high-strength evidence of colocalization analysis, heterogeneity analysis, pleiotropy analysis, and the utilization of five additional MR analysis methods (Supplementary Table S3, S4). Their support by various methods and the consistency across these results enhance the credibility of these findings.Figure 2MR results for the association between the expression of mitochondrial-related genes and CKD progression outcomes.For CKD, an increase of one SD in gene expression of rs7164602-glycine amidinotransferase (rs7164602-GATM) is associated with a 12.1% decrease in the risk of CKD (OR = 0.879, 95% CI 0.848, 0.910, FDR = 5.25E−10, PHEIDI = 6.07E−02, PP.H4 = 0.815). The inverse variance weighted analysis yielded the same results (OR = 0.895, 95% CI 0.822, 0.975, P = 1.10E−02). Meanwhile, there was no heterogeneity (PCochran’s Q = 3.67E−01) or pleiotropy (PIntercept = 8.27E−01).Moreover, gene expressions of rs12706818-SND1, rs3752493-MCRIP2, rs12983363-ATP5F1D, rs479572-ATP5IF1, rs3864285-HINT1, and rs3769698-COA5 were also causally associated with eGFRcrea, but their contribution were relatively small (the absolute value of β < 0.03).The MR analysis of mitochondrial genome-wide cis-mQTL and CKDFor the causal relationship between DNA methylation of mitochondrial-related genes and CKD, we discovered a total of 4 association signals across 3 unique genetic loci associated with CKD and 15 association signals across 13 unique genetic loci for eGFRcrea (Fig. 3, Supplementary Table S5). The sensitivity analysis corroborated these associations (Supplementary Table S6).Figure 3MR results for the association between the DNA methylation of mitochondrial-related genes and CKD progression outcomes.Here, we identified a total of 2 unique loci that regulate the methylation level of 3 different CpG sites in GATM, positively associated with CKD risk. A one SD increase in methylation at rs56850226-GATM by cg00767496 was linked to a 20.1% higher risk of CKD (OR = 1.201, 95% CI 1.132, 1.275, FDR = 3.51E−07, PHEIDI = 1.25E−01, PP.H4 = 0.815) and a 2.4% higher risk of decreased eGFRcrea (β = − 0.024, 95% CI − 0.028, − 0.02, FDR = 3.22E−24, PHEIDI = 6.82E−02, PP.H4 = 0.814). Additionally, a one SD increase in methylation at rs56850226-GATM by cg14910265 corresponded to an 18.9% higher risk of CKD (OR = 1.189, 95% CI 1.125, 1.256, FDR = 2.15E−07, PHEIDI = 9.86E−02, PP.H4 = 0.815). Furthermore, a one SD increase in methylation at rs12593371-GATM by cg24328539 was associated with an 11.3% higher risk of CKD (OR = 1.113, 95% CI 1.079, 1.148, FDR = 8.84E−09, PHEIDI = 5.80E−02, PP.H4 = 0.815). The inverse variance weighted analysis yielded the same results (OR = 1.124, 95% CI 1.076, 1.175, P = 1.97E−07). Meanwhile, there was no heterogeneity (PCochran’s Q = 1.36E-01) or pleiotropy (PIntercept = 8.06E−01). Lastly, a one SD increase in methylation at rs2233956-MCCD1 by cg04048339 was linked to a 25.8% higher risk of CKD (OR: 1.258, 95% CI 1.114, 1.420, FDR = 3.30E−02, PHEIDI = 8.24E−01, PP.H4 = 0.978). The wald ratio analysis yielded the same results (OR = 1.258, 95% CI 1.140, 1.388, P = 4.64E−06).The methylation at various loci of genes including rs2058389-STYXL1, rs13246286-SND1, rs322817-SND1, rs140291167-SND1, rs35660964-SND1, rs2242517-GCDH, rs75461554-CASP9, rs2276731-ALDH1L1, rs34293138-GPX1, rs55663021-SLC25A29, as well as rs11235548-PDE2A and rs148038373-PDE2A, were also found to be causally associated with eGFRcrea, but their contributions were relatively small (the absolute value of β < 0.01).The MR analysis of mitochondrial genome-wide cis-pQTL and CKDConcerning protein expression of mitochondrial-related genes, we identified a unique genetic locus significantly associated with CKD and 3 unique genetic loci associated with eGFRcrea (Fig. 4, Supplementary Table S7, S8). Notably, an increase of one standard deviation in protein expression of rs1153858-GATM reduced the risk of CKD occurrence by 33.9% (OR = 0.661, 95%CI 0.587, 0.744, FDR = 9.26E−10, PHEIDI = 8.91E−02, PP.H4 = 0.027). The inverse variance weighted analysis yielded the same results (OR = 0.733, 95% CI 0.640, 0.840, P = 7.12E−06). Meanwhile, there was no pleiotropy (PIntercept = 5.89E−01), and the only heterogeneity had little impact on the results (PCochran’s Q = 3.60E−02). Furthermore, protein expression of rs1486308-NIT2, rs4699179-PPA2, and rs10887871-PRXL2A genes exhibited causative relationships with eGFRcrea. Unfortunately, their PP.H4 were all less than 0.8.Figure 4MR results for the association between the protein expression of mitochondrial-related genes and CKD progression outcomes.Bi-directional MR analysis of mitochondrial dysfunction and CKDThe mtDNA copy number variation is considered a surrogate biomarker for mitochondrial dysfunction. Currently, a GWAS dataset, based on exome sequencing results from 415,422 individuals of European descent within the UK Biobank, has identified genetic variations associated with mtDNA copy27. We utilized this dataset to investigate whether mitochondrial dysfunction contributes to CKD. Excitingly, our results suggested that CKD or eGFRcrea didn’t impact changes in mtDNA copy number (Supplementary Table S9). The direction of causality suggests a specificity of mitochondrial dysfunction predisposing towards CKD susceptibility.External dataset validation of the GATM expression and CKD.Additionally, we validated our findings across multiple external RNA sequencing datasets. The results demonstrated that the expression levels of GATM were significantly reduced in nine CKD conditions, including diabetic nephropathy, vasculitis, focal segmental glomerulosclerosis, membranous glomerulonephritis and so on (Fig. 5). Further analysis revealed a significant positive correlation between the expression levels of GATM and GFR (calculate using MDRD method) across multiple datasets, as indicated by the fitted curve equation and associated P-values in Fig. 6.Figure 5Comparison of GATM expression levels between various types of CKD and healthy control patients. **, P < 0.01; ****, P < 0.0001.Figure 6Correlation analysis between GFR and GATM expression levels in different datasets.Ethics approval and consent to participateThis study was conducted following the Strengthening the Reporting of Observational studies in Epidemiology—Molecular Epidemiology (STROBE-ME): An extension of the STROBE statement38. The summary statistics data utilized in the MR analysis were obtained from publicly available databases or previous studies, all of which had received individual ethical approvals.
