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

Rare variant analysis 

Based on 4,879 GCKD participants with available exome chip data, we additionally conducted aggregated rare variant testing (S1 Fig). Variants with a MAF <1% and having a major effect on the gene product (nonsynonymous, stop gain/loss, splicing; "qualifying variants") as annotated by dbNSFP v.2.0 were aggregated (Material and methods, [24,25]). While there was no significant association result when the burden test was used, we found a significant association using the sequence kernel association test (SKAT) for the SPP1 gene (Table 3, p-value = 2.5E-08), which remained significant after adjustment for the two replicated SNPs from the initial GWAS (p-value = 9.4E-08). Seven variants were aggregated for the analysis of the SPP1 gene. In the single variant analysis, effect estimates of the seven variants ranged from -1.20 to 0.24 with rs139555315 (4,88901197), a splice-site variant (CADD score: 23.7) showing the most significant association (effect estimate = -1.20, SE = 0.19, p-value = 2.5E-10) and thus likely driving the association. This is supported by the non-significant result (p-valueSKAT = 3.5E-01) when rs139555315 was excluded from the variant set. Other genes implicated by GWAS of common variants reached only nominal statistical significance (p<0.05; MEPE: p-valueSKAT = 4.6E-03; KLKB1: p-valueBurden = 5.6E-04, p-valueSKAT = 2.8E-03; F12: p-valueBurden = 3.4E-02, p-valueSKAT = 3.0E-02).

ORGANIC HERBS FOR KIDNEY DISEASE

Discussion

In this study, we focused on characterizing the genetics of OPN within a CKD cohort, because OPN levels are known to be associated with adverse kidney outcomes, but the genetic underpinnings of this kidney-enhanced protein are not fully understood. The use of a CKD patient cohort might present an advantageous setting in which the transcription of kidney-specific genes may be altered in comparison to the general population, making identification of specific SNPs possibly easier. We identified three loci, two on chromosome 4 (rs10011284, rs4253311), that could be replicated in an external population-based cohort, and rs2731673 on chromosome 5, which could not be replicated. When aggregating rare variants, the SPP1 gene encoding OPN was detected.

To our knowledge, this is the first GWAS of serum OPN levels quantified via ELISA. Other studies like Sun et al. conducted GWAS of plasma proteins (pGWAS) including OPN. Here proteins were quantified differently via an aptamer-based technology (SOMAscan, [22]). While this study was likely too small (N = 3,301) to detect any genetic signals for OPN, a study by Pietzner et al. (N up to 10,708) provided signals on chromosomes 3, 4, and 10 associated with SOMAscan-measured OPN [26]. The detected signal on chromosome 4 (rs5860110, 4:88897106, MAF = 0.30) is a common, intronic indel located within SPP1 not in linkage disequilibrium (LD) with common variants identified in our project. In the next step, Pietzner et al. reported association statistics for the SOMAscan detected loci using a different technique to measure proteins (Olink, antibody-based protein panels). Olink measurements were, however, only available for a small fraction of the study population and none of the three SOMAscanmeasured OPN signals could be confirmed. The systematic comparison of protein levels quantified by these two techniques revealed varying correlations (median 0.38, IQR: 0.08–0.64). For OPN, a correlation coefficient of 0.51 was reported. A similar comparison of even more proteomics platforms also reported a wide range of correlations among measurements [27]. Differences in the detection of genetic signals could thus not only be by differences in power but also by technical, protein, and variant characteristics. Across platforms, any comparison results are thus difficult and meta-analyses could lead to wrong inferences.

rs10011284; 4:88833389 (SPP1/MEPE intergenic) 

The index SNP rs10011284 (MAF = 0.43) of the first locus on chromosome 4 maps between the SPP1 and MEPE genes. Whether this variant or another is the causal variant underlying the observed association remains unclear as results from statistical fine-mapping are inconclusive. Nevertheless, when aggregating rare variants, SPP1, the gene encoding the protein OPN, showed a significant association driven by a splice-site variant. Any errors occurring during the splicing process can lead to false intron removal causing alterations to the open reading frame. In turn, this may either lead to the formation of a premature stop codon and a shortened protein or more likely to a faster mRNA degradation called nonsense-mediated decay [28]. The variant driving this association at SPP1, a splice-site variant (rs139555315, 4:88901197, MAF = 1.54E- 03), was reported to be associated with pediatric systemic lupus erythematosus [29].

SPP1 is made up of 7 exons containing 942 transcribed nucleotides from the start codon in exon 2 to the stop codon (within exon 7, [30]). OPN belongs to the SIBLING glycoprotein family of secreted phosphoproteins; other members of this family comprise dentin matrix protein 1, dentin-sialophosphoprotein, Catherine, bone sialoprotein, and matrix extracellular phospho-glycoprotein (MEPE). Results from our colocalization analysis using GeneAtlas are pointing toward a connection of SPP1 with bone disorders. Fitting with these results, one of OPN's main physiological functions in the body is the regulation of the biomineralization processes [31]. Conflicting results have been reported on OPN as well as SPP1 polymorphisms and susceptibility to nephrolithiasis in the past [32–36]. From in vitro studies it may be inferred, that OPN inhibits nucleation, growth, and aggregation of calcium oxalate crystals [37], but clinical studies draw a more unclear picture. Some researchers report on the protective role of OPN, whereas others do not [38,39]. Nonetheless, a recent study from South Asia found a significant association of the SPP1 rs2853744:G>T polymorphism with urolithiasis [40].

OPN is mostly secreted, but an intracellular form has also been reported [41]. Using reverse-transcription-PCR OPN was found to be expressed in normal human adult kidneys, further immunohistochemical analyses and in situ hybridization revealed OPN expression to be restricted to the distal convoluted and straight tubules in kidney cortex and medulla in monkey kidneys [42]. Looking at GTEx tissue expression data, a positive colocalization of the OPN association signal at SPP1 with pancreas tissue was detected. This is in line with findings in the literature where OPN has been suggested to have a role in type 2 diabetes. One study performed by Cai et al. investigated a diabetic mouse model SUR1-E1506K+/+ and islets from human donors and was able to demonstrate that in islets from human cadaver donors, OPN gene expression was elevated in diabetic islets, and externally added OPN significantly increased glucose-stimulated insulin secretion from diabetic but not normal glycemic donors [43]. Many other studies have also investigated OPN's role in pancreatic cancer, here, OPN was found to be a prognostic marker associating higher levels with poor overall prognosis in patients [44].

MEPE (Matric Extracellular PhosphoglycoprotEin) is the gene encoding the secreted calcium-binding phosphoprotein MEPE. A common feature of SIBLING proteins is the Acidic Serine Aspartate Rich MEPE-associated motif (ASARM), which is involved in the regulation of mineralization, bone turnover, mechanotransduction, phosphate, and energy metabolism [45]. The ASARM motif is also the connecting link between SIBLINGs and FGF23 thereby being part of the physiological bone-kidney link [45]. MEPE is involved in the regulation of phosphate homeostasis by the kidney and intestine [46,47]. Colocalization of the OPN associatedition signal at MEPE leads to the detection of an association with lung tissue. So far a connection between several members of the SIBLING family with lung cancer has been reported, but a definite connection between MEPE and lung has not been made before [48]. Since there are multiple transcript variants known due to alternative splicing a connection between MEPE and lung cannot be ruled out and could offer possible research areas in the future. Results from our colocalization analysis using GeneAtlas are pointing towards a connection of MEPE with bone disorder. Diseases associated with MEPE are osteomalacia and autosomal-dominant hypophosphatemic rickets supporting this connection between MEPE and bone disorders [49]. Another phenotype seemingly related to rs10011284 is gout. This association is most likely driven by ABCG2, which is located near SPP1 and MEPE and is one of the best-described uric acid transporter genes to date [19,50].