Genetics Of Osteopontin in Patients With Chronic Kidney Disease: The German Chronic Kidney Disease Study Ⅳ

Baseline variables and measurements

A standardized set of biomarkers was measured in a central certified laboratory using standardized protocols [81]. Among others, creatinine and albumin from serum and urine were quantified using an IDMS traceable methodology (Creatinine plus, Roche, Germany) and a turbidimetric method (Tina-quant, Roche, Germany) Roche/Hitachi MODULAR P, respectively. Glomerular filtration rate (GFR) was estimated using the creatinine-based CKD-EPI formula (unit: mL/min/1.73m2 , [82]). UACR was calculated as measured urinary albumin/urinary creatinine (mg/g, [83]). Age and sex were self-reported at the baseline visit. In 2015, OPN was measured from baseline serum samples of the complete GCKD study cohort using a quantitative sandwich enzyme immunoassay technique (solid-phase ELISA; Quantikine Human OPN Immunoassay DOST00 from R&D Systems (R&D Systems Europe, Abingdon, UK)). Quantification was carried out at the Institute of Clinical Chemistry and Laboratory Medicine, Greifswald, Germany. Coefficients of variation (intra-assay) were 4.5%, 5.3% and 3.5% for low, median and high levels, respectively. The inter-assay coefficient of variation was 6.4%. Reagents and secondary standards were used as recommended by the manufacturer.

ORGANIC TRADITIONAL HERB FOR KIDNEY HEALTH

Genotyping, quality control and imputation 

Detailed information on genotyping and data cleaning in the GCKD study has been described previously [18]. Briefly, DNA was isolated from whole blood and genotyped at 2,612,357 variants for 5,123 GCKD participants using the Illumina HumanOmni2.5 Exome BeadChip array (Illumina, GenomeStudio, Genotyping Module Version 1.9.4) at the Helmholtz Center Munich. Data cleaning was carried out separately for the Omni2.5 content and the exome chip content of the array. Based on standardized protocols [84], custom written scripts (R, Perl) and Plink1.9 [85] software was used for quality control (QC) of the Omni2.5 content. Sample-based QC steps included checks of call rate, sex, heterozygosity, genetic ancestry and relatedness, leading to the exclusion of 89 samples. On the variant level, single nucleotide polymorphisms (SNPs) were excluded if the call rate was <0.96, and whenever the assumption of the Hardy-Weinberg equilibrium was violated (p-value <1.0E-05). After removing SNPs on duplicate positions, the cleaned dataset contained 5,034 individuals and 2,337,794 SNPs (S1 Fig). Genotypes were then imputed using minimac3 v2.0.1 at the Michigan Imputation Server [86]. The Haplotype Reference Consortium (HRC) haplotypes version r1.1 were used as the reference panel, and Eagle 2.3 was used for phasing. The final dataset contains data of 5,034 participants and 7,750,367 high-quality autosomal bi-allelic variants (imputation quality of R2 �0.3, MAF �1%). For the exome chip content, QC was similarly conducted [18]. In addition, checks specific for exome variants were added [87]. In brief, 96 individuals and 3,818 SNPs were removed, the latter of which had a call rate <0.95 and a Hardy-Weinberg equilibrium p-value <1.0E−05. The final exome chip dataset contains 5,027 participants with 226,233 variants (S1 Fig). For the exome chip association analysis, the genotypes were post-processed using zCall with a zscore threshold of six [88]. Genomic positions base on human genome build GRCh37.

Genome-wide association study of common variants 

As previously reported [17,18], GWAS was conducted for GCKD participants with complete genotyping (Omni2.5), eGFR, UACR and log2(OPN) measurement (N = 4,897) data using linear regression of log2(OPN) on SNPs (additive genetic model) with a MAF �1%, adjusted for age, sex, log(eGFR), and log(UACR) (S1 Fig). Association analysis was performed using SNPTEST v2.5 [89]. Summary statistics were checked for quality using GWAtoolbox [90] and for inflation using genomic control [91]. A genomic control correction, however, was not requested (λ = 1.01). Associations with a p-value <5.0E-08 were considered significant. Per chromosome, an index SNP was defined as the SNP with the lowest genome-wide p-value with a 1-Mb interval centered around this SNP. This approach was repeated until no further SNP outside the interval(s) was available passing the genome-wide significance threshold. In order to discover further independent signals, we repeated GWAS analysis for chromosomes with significant results by conditioning on the genotype of the SNP with the lowest association pvalue of the respective chromosome. This procedure was repeated until no further genomewide signal was observed.

Functional annotation of variants was conducted using ANNOVAR[92], SNiPA [93], Open Targets Genetics [94], FAVOR [95], and RegulomeDB [96]. Regional association plots were plotted using LocusZoom v1.3 [97].

Fine-Mapping

Statistical fine-mapping [21] was carried out as previously described [17] for the two replicated SNPs within a region ±500kb. Approximate Bayes factors (ABFs) were then derived from the original GWAS statistics estimates. The SD prior was chosen as 0.61 because 95% of the effect size estimates fell within the −1.2 to 1.2 interval [21]. The ABF of the SNPs were used to calculate the posterior probability for each variant driving the association signal (PPA, 'causal variant'). Credible sets were determined by summing up PPA-ranked variants until the cumulative PPA was >99%.

Colocalization analyses

In order to further understand the molecular mechanisms and associated phenotypes underlying the associations, we performed colocalization analyses of the OPN GWAS summary statistics related to the two replicated OPN loci with GWAS summary statistics from three other sources as outlined below. For all colocalization analyses, we used the 'coloc.fast' function from the R package gtx with default parameters and prior definitions (https://github.com/ tobyjohnson/gtx), an implementation of an adapted version of the colocalization method introduced by Giambartolomei et al. [98]. We consider a positive colocalization when the posterior probability of a shared causal variant at the association locus for both traits (H4, p12) was > 0.8.

Gene expression. 

First, we used GWAS summary statistics of gene expression data from the GTEx project [99] and the NEPTUNE study [100]. The eQTL data from the GTEx V8 (49 tissues) and the NEPTUNE study (NephQTL from glomerulus and tubulointerstitial kidney portions) were downloaded from the GTEx Portal (https://www.gtexportal.org/home/) and NephQTL web site (http://nephqtl.org/), respectively. The analysis steps of colocalization have been described in detail elsewhere [17]. Firstly, GWAS summaries of GTEx and NephQTL in genomic regions ±100kb of the two OPN SNPs were extracted. The genes in the extracted GWAS are identified and for each gene, a cis window of 500kb flanking the start and end of the gene are defined. Then, for every such cis gene window, with at least one SNP having an association p-value < 0.001, the GWAS summaries of GTEx and NephQTL tissue as well as the OPN GWAS were extracted and used as input for colocalization analysis.