Urine mitochondrial DNA and diabetic nephropathy—a new frontier

Diabetic nephropathy is a leading cause of chronic kidney disease (CKD) worldwide [1, 2] and occurs in 35–40% of both type 1 and type 2 patients within 10 years of onset of diabetes [2, 3]. Compared with patients who have diabetes mellitus without kidney involvement, presence of diabetic nephropathy significantly increases the risk of adverse cardiovascular events and mortality [1]. Approximately 40% of patients with diabetic nephropathy eventually progress to end-stage kidney disease [4, 5]. Unfortunately, studies examining urinary and pathologic biomarkers such as neutrophil gelatinase-associated lipocalin (NGAL), kidney injury molecule-1 (KIM-1), proteinuria and extent of interstitial fibrosis on biopsy have been unsuccessful in identifying reliable predictors of disease progression in diabetic nephropathy [6–8].

In recent years there has been growing recognition of the importance of mitochondrial dysfunction in both acute kidney disease and CKD [9, 10]. The kidney has a high metabolic rate and receives 20% of cardiac output, consuming 10% of the body’s oxygen intake [10], and is rich in mitochondria, which is the energy-producing organelle that maintains cellular redox and energy homeostasis. Mitochondria generate adenosine triphosphate (ATP) through oxidative phosphorylation and have essential roles in heme biosynthesis, the Krebs cycle, fatty acid β-oxidation pathways, calcium ion homeostasis, thermogenesis, proliferation and apoptosis [11]. Unlike nuclear DNA, human mitochondrial DNA (mtDNA) is a circular molecule composed of two strands (heavy and light strands) and contains 16 569 base pairs and 37 genes [10]. Since germ cells have few mitochondria and are selectively degraded, most mitochondria are from maternal inheritance [12].

mtDNA is vulnerable to any cellular or oxidative stress due to lack of histone protection and inefficient repair mechanisms in coding regions. Subsequently, mtDNA is susceptible to damage and mutations (10- to 1000-fold greater mutation rate than nuclear DNA) from triggers such as oxidative stress, ultraviolet light, alkylating agents and aging [9]. In diabetes mellitus, activation of the renin–angiotensin system and accumulation of advanced glycation end products [13, 14] incur inflammatory and oxidative stress injury. Upregulation of tumor growth factor-β, reactive oxygen species (ROS) and nicotinamide adenine dinucleotide phosphate activate intracellular signaling via protein kinase C (PKC) and SMAD2/3, resulting in increased levels of the transcription factor nuclear factor-κB [15–19]. Thus, heightened inflammation and ROS in diabetes can incur mtDNA mutations resulting in mitochondrial dysfunction, which encompasses reduced ATP synthesis, accumulation of intracellular calcium (resulting from calcium pump inactivation), decomposition of membrane phospholipids and cell death [9]. This mitochondrial dysfunction in turn promotes end-organ damage; cell culture and animal studies have implicated mitochondrial-mediated overproduction of superoxide and reduced phosphorylation of pyruvate dehydrogenase as mechanisms of kidney damage in diabetic nephropathy [20–22].

The first report of mtDNA as a potential biomarker of diabetic nephropathy was by Sharma et al., who utilized gas chromatography-mass spectrometry to compare urine metabolomics between diabetic CKD patients and non-diabetic controls [23]. Of the 13 urine metabolites that were significantly reduced in the diabetic CKD patients, 12 were linked to mitochondrial metabolism and suggested global suppression of mitochondrial activity in diabetic kidney disease. Consistent with these findings, urinary mtDNA was ‘decreased’ from diabetic CKD patients [23]; however, correlation with kidney biopsy pathology was not available. In contrast, Czajka et al. reported ‘elevated’ blood levels of mtDNA in patients with diabetic nephropathy compared with healthy controls; again, correlation with kidney biopsy pathology was not reported [24].

In this issue of NDT, Wei et al. [25] further examine the prognostic value of baseline urinary mtDNA via correlations with kidney pathology, kidney tissue mtDNA levels and 24 months clinical follow-up in a cohort of 92 patients with diabetic nephropathy at an academic hospital in Hong Kong. mtDNA from urine sediment, urine supernatant and frozen kidney biopsy tissue was isolated and subjected to digital polymerase chain reaction; urinary mtDNA was normalized to urine creatinine level. The investigators found that increased urinary supernatant mtDNA level correlated with decreased intra-renal mtDNA (r = −0.453, P = 0.012) and had a positive correlation with severity of interstitial fibrosis (r = 0.300, P = 0.005). Higher levels of urinary supernatant mtDNA correlated with lower baseline estimated glomerular filtration rate (eGFR) (r = −0.214, P = 0.04). Urinary sediment mtDNA did not correlate with any clinical or pathological parameter. Decreased intra-renal mtDNA correlated with more severe interstitial fibrosis (r = −0.537, P = 0.003).

In terms of clinical outcomes, there was no significant relationship between urinary or intra-renal mtDNA levels and mortality or rate of GFR decline at 24 months follow-up. In subsequent analysis, patient samples were separated into tertiles according to urinary supernatant mtDNA. The lowest concentration of urinary supernatant mtDNA (tertile I, patients deemed to have early stages of diabetic nephropathy and mild mitochondrial dysfunction) had a modest correlation with proteinuria (r = 0.397, P = 0.027) but did not correlate with degree of glomerulosclerosis or tubulointerstitial fibrosis on biopsy. The authors noted a marginal correlation between tertile I of urinary supernatant mtDNA and GFR decline (r = −0.321, P = 0.08).

The major strength of the study by Wei et al. is the correlation with kidney histology and 2-year clinical follow-up. A limitation was the lack of reference controls, i.e. urine samples and kidney tissues from nondiabetic individuals. We can only make conservative conclusions given the small sample size, and lack of serial assessment of changes in mtDNA. Overall, urinary supernatant mtDNA appears to be a promising marker of intra-renal scarring and proteinuria in patients with early diabetic nephropathy. It is unclear at this time whether increased urinary mtDNA is due to ongoing destruction of intra-renal mitochondria or renal clearance of circulating mtDNA; as mentioned above, blood levels of mtDNA are elevated in patients with diabetic nephropathy [24]. The investigators astutely acknowledge that inclusion of only biopsy-confirmed diabetic nephropathy may have introduced sampling bias; patients with presumed diabetic kidney disease are not usually biopsied unless they present with atypical features such as rapid increase in proteinuria, absence of retinopathy or microscopic hematuria. The impact of glycemic control also could not be assessed in this study.

Although the clinical prognostic value of mitochondrial markers has not yet been established, the strong evidence supporting this pathophysiological pathway in diabetic nephropathy has prompted research into novel therapies that rescue mitochondrial function. For example, a molecule of interest downstream of mitochondrial dysfunction is decreased expression of peroxisome proliferator-activated receptor (PPAR)-γ-coactivator-1α (PGC-1α), a master regular of mitochondrial biogenesis. PGC-1α is decreased in the kidney tissues from diabetic nephropathy patients [23] and from mice with streptozotocin-induced type 1 diabetes mellitus [22]. It appears that the protective role of PGC-1α is associated with the inhibition of dynamin related protein 1 (DRP1)-mediated mitochondrial remodeling and ROS production [26]. Along these lines, thiazolidinediones, which are PPAR ligands, are being studied in diabetic nephropathy and other glomerular diseases for their ability to inhibit podocyte and vascular injury via reduction of oxidative stress and recovery of mitochondrial electron transport function [27–31]. Another example is the peptide SS-31, a small cell-permeable molecule that reduces mitochondrial ROS production and prevents mitochondrial permeability and cytochrome C release [32]. In an animal model of diabetic nephropathy, it has been shown to have a protective effect against hyperglycemia-induced kidney injury [33]. Sul-121 and pyruvate kinase M2 activators are other small molecules that have been tested to reverse mitochondrial dysfunction in mouse models of diabetic nephropathy [34, 35].

In summary, the report from Wei et al. adds to our understanding of mitochondrial dysfunction, while more research is needed to fully address the clinical value of mitochondrial markers in diabetic nephropathy. Urinary supernatant mtDNA was increased in tandem with lower levels of baseline eGFR and more severe kidney fibrosis; however, the magnitude of increase is unclear given the lack of healthy reference controls. The associations reported in the paper were moderate (Spearman’s rank correlation coefficient absolute values ranged 0.2–0.5) and baseline urinary mtDNA did not predict clinical outcomes. Given the complex milieu of inflammation, glycation stress, vascular dysfunction and ROS injury in diabetic nephropathy, it seems that a panel of biomarkers (representing various pathologic pathways) may be needed to accurately reflect disease activity, rather than a single biomarker. Regardless, mitochondrial dysfunction being a new frontier in diabetic nephropathy, the pathophysiological role of mitochondria in nephropathy should be further investigated from basic science to clinical application.

CONFLICT OF INTEREST STATEMENT

None declared.

(See related article by Wei et al. Urinary mitochondrial DNA level is an indicator of intra-renal mitochondrial depletion and renal scarring in diabetic nephropathy. Nephrol Dial Transplant 2018; 33: 784--788)

REFERENCES