Delineating a role for the mitochondrial permeability transition pore in diabetic kidney disease by targeting cyclophilin D

Chronic kidney disease (CKD) is recognised as a leading cause of morbidity and mortality amongst the 425 million people estimated to have diabetes globally [1]. This represents a significant burden to healthcare systems as the number of people with all types of diabetes is projected to rise over the next few decades [1]. Currently, best practice clinical management involves strict glucose control and blood pressure reduction to alleviate risk for progression to end stage renal disease. Although these approaches can successfully stabilise the rate of progression in many patients, they do not entirely eliminate disease or risk for renal replacement therapy [2,3]. There remains an urgent need for new therapeutic approaches in targeting diabetic kidney disease (DKD) and a requirement to deepen our understanding of the aetiology and pathogenesis to improve patient outcomes.

The therapeutic potential of maintaining and/or restoring mitochondrial ‘health’ is currently being investigated in DKD, since a decline in mitochondrial function is associated with renal functional decline and injury [4]. Mitochondria are considered vital for the function of nephrons which consume high levels of oxygen and adenosine triphosphate (ATP) to support the metabolic demands of selective reabsorption in the proximal and distal tubules [5,6]. Furthermore, mitochondria play an important role in adaptation to the metabolic changes and hypoxia associated with diabetic renal pathophysiology [6,7]. Thus, a suite of mitochondria-targeted compounds are currently being developed and tested in pre-clinical models to determine their efficacy in the treatment of DKD. One potential target for therapy is the mitochondrial permeability transition pore (mPTP), which is a stress-sensitive pore that traverses both the inner and outer mitochondrial membranes. Transient mPTP opening is thought to dissipate chemical and electrical ion build up from the mitochondrial matrix to the cytosol to restore ionic balance [8,9]. However, persistent opening of the mPTP results in mitochondrial swelling and subsequent outer membrane rupture, along with a concomitant decline in capacity for oxidative phosphorylation (OXPHOS) [10]. Inappropriate mPTP opening is associated with necrotic cell death as the maintenance of ATP production is necessary for the energy demanding process of apoptosis [11,12]. Previous preclinical studies of DKD have shown increased mPTP opening events occur in kidney mitochondria, which coincided with the development of renal morphological injury and a decline in renal cortical mitochondrial respiratory chain function in a rat model of streptozotocin (STZ)-induced diabetes [4].

Although the exact mechanism that enables trafficking through the mPTP is still under debate, newer models suggest several key membrane proteins are involved in pore opening. This includes the outer membrane pore-forming proteins Bax and Bak, the F1alpha unit of ATP synthase, and cyclophilin D (CypD) [13,14]. CypD is considered to regulate pore opening and it therefore remains as a unique target which can be directly inhibited by broad spectrum immunosuppressants such as cyclosporine A (CsA) and its newer non-immunosuppressive analogue Alisporivir (Debio025) [15,16]. In the kidney, CypD deletion in mice is protective against the development of renal fibrosis and proximal tubular epithelial cell (PTEC) damage following unilateral ureteral obstruction, and ischemia reperfusion injury [17–19]. However, the effects of targeting CypD on DKD progression have not been investigated. The present study aims to investigate the role of CypD in DKD by utilising two disparate yet complementary approaches. Firstly, mice with a global genetic knock out (KO) of the peptidylprolyl isomerase F gene, Ppif, (which is the gene encoding for CypD), and secondly using pharmacological inhibition by Alisporivir in a murine model of experimental diabetes.

All animal studies were approved by the Alfred Medical Research and Education Precinct Animal Ethics Committee and the National Health and Medical Research Council of Australia. All procedures were performed in accordance with the Australian code of conduct for ethical research involving animals. All animals were housed at the Alfred Medical Research and Education Precinct Animal Centre, Melbourne, Australia, and were maintained following established ethical procedures, in a temperature controlled environment at 22°C with a 12 h light/dark cycle with ad libitum access to rodent chow (standard mouse chow, Specialty Feeds, Perth, WA, Australia) and water.

Six week old Male C57BL/6 mice with a global deletion of Cyp D (Ppif^−/−) were purchased from the Jackson Laboratory (Bar Harbor, ME, U.S.A.). Diabetes was induced at 6 weeks of age (n=15 mice per group, Ppif^WT/WT STZ or Ppif^−/− STZ) by five daily intraperitoneal injections of low dose STZ (55 mg/kg, Sigma-Aldrich, St Louis, MO), with control mice receiving the vehicle, sodium citrate buffer (CIT) alone (n=15 mice per group, Ppif^WT/WT or Ppif^−/−). After 10 days, plasma glucose concentrations were determined to confirm diabetes development and mice with blood glucose not exceeding >15 mmol/l glucose (observed in >98% of mice) were excluded from the study. None of the animals with diabetes required exogenous insulin supplementation to maintain body weight or to prevent ketosis.

At 20 weeks after induction of diabetes, mice were placed individually into metabolic cages (Iffa Credo, L’Arbresele, France) for 24 h. Urine and plasma from submandibular collections were stored at −80°C for subsequent analysis. Blood glucose was monitored using a glucometer (Accutrend; Boehringer Manheim Biochemica, Manheim, Germany). Glycated haemoglobin (GHb) was determined by turbidimetric inhibition immunoassay using a Cobas Integra 400 autoanalyzer (Roche Diagnostics Corporation, U.S.A.) in lysates of erythrocytes separated from whole blood. After 22 weeks, animals were killed by sodium pentobarbitone intraperitoneally (100 mg/kg body weight) (Euthatal; Sigma-Aldrich, Castle Hill, NSW, Australia) followed by cardiac exsanguination. The kidneys were rapidly dissected, weighed, and snap-frozen or placed in 10% (v/v) neutral buffered formalin for fixation before paraffin embedding. In a subset of mice (n=5 per group), mitochondria were isolated from renal cortices as later described.

Male db/db mice (C57BKS.Cg-Dock7^m+/+Lepr^db/J) and their db/m littermate controls were obtained from colonies maintained at the Alfred Medical Research and Education Precinct Animal Services (Melbourne, Australia). Mice were randomly assigned at 7 weeks of age to either vehicle (10% Cremophor EL in water) or Alisporivir (Novartis, Basel, Switzerland) at 5 mg/kg/day by oral gavage. Animals were treated for 16 weeks with blood glucose and body weight assessed weekly. Any animal which lost 20% of body weight or developed severe ketoacidosis during the study was excluded before endpoint in accordance with ethical guidelines. HbA_1c was determined using the Cobas b 101 system (Roche diagnostics, IN, U.S.A.). At week 15, body composition was determined using an EchoMRI (EchoMRI™, Houston, TX, U.S.A.) and mice were placed into individual Comprehensive Laboratory Animal Monitoring System (CLAMS, Columbus Instruments, Columbus, Ohio, U.S.A.) chambers to determine physical activity and respiration by indirect calorimetry as previously described in full [20]. Metabolic caging to collect 24 h urine samples was performed at 8 weeks (midpoint) and 15 weeks (endpoint). Mice were killed and kidneys were harvested and processed as described above. Additionally, mitochondria were isolated from the whole cortex of the left kidney for all animals (described below). A small section of renal cortex was also reserved for transmission electron micrography (described below).

The 24-h urinary albumin excretion rates were determined using a mouse-specific sandwich ELISA in timed urine collections (Bethyl Laboratories, Montgomery, TX, U.S.A.) according to the manufacturer’s instructions. Creatinine was assessed in both plasma and urine samples using commercially available kits for the Cobas Integra 400 autoanalyzer (Roche Diagnostics, IN, U.S.A.). Plasma cystatin C was measured using ELISA (R&D Systems Inc., Minneapolis MN, U.S.A.). Urinary content of kidney injury molecule (KIM)-1 was measured using a commercially available sandwich ELISA (USCN Life Sciences, Wuhan, China) according to the kit protocol. Glomerular sclerosis index (GSI) was assessed in 3-μm paraffin embedded periodic acid Schiff (PAS) stained sections by a semi-quantitative method as previously described [21]. Paraffin sections of kidneys were used to stain for collagen type IV (Col IV) using a goat polyclonal collagen IV antibody (Southern Biotech, Birmingham, AL, U.S.A.) [22].

Freshly collected kidney cortex was minced on ice and placed into 1 ml of ice-cold Isolation Buffer (IsoB) (70 mM sucrose; 210 mM d-mannitol; 5 mM HEPES; 1 mM EGTA; 0.5% (w/v) fatty acid free bovine serum albumin [Sigma-Aldrich, NSW, Australia]) followed by 6–8 careful strokes in a Potter–Elvehjem homogenizer set in an ice-bath. Lysate was then collected and any remaining tissue homogenised with an additional 1 ml IsoB for 2–3 more strokes. The lysate underwent differential centrifugation at 4°C as previously described [23]. Briefly, ∼2 ml total volume was centrifuged for 5 min at 800 g. Supernatant was collected and centrifuged at 8000 g ×10 min. Mitochondrial pellet was washed in 500 μl IsoB and centrifuged again at 8000 g × 10 min. Final pellet was resuspended in 115 μl IsoB and stored on ice for use in functional assays (below). Total protein was determined by the bicinchoninic acid method and corrected for background protein in IsoB (Pierce-Thermo Fisher Scientific, Melbourne, Australia).

Mitochondria were resuspended in mitochondrial assay buffer (MAS) (70 mM sucrose; 220 mM d-mannitol; 10 mM KH₂PO4; 5 mM MgCl₂·6H₂O; 2 mM HEPES; 1 mM EGTA; 0.2% (w/v) fatty acid free bovine serum albumin, pH 7.2) and loaded into a XFe96well Seahorse Bioanalyzer plate (Seahorse Bioscience, Agilent, Santa Clara CA, U.S.A.) with a minimum of five replicate wells per mouse. For complex I respiration 10 μg of mitochondria was loaded per well with 0.5 mM glutamate and 0.5 mM malate. Plates were centrifuged at 2000g for 20 min at 4°C before adding substrates and then immediately loaded into the XFe96 Seahorse machine. Basal respiration with substrates was measured twice for 3 min each, followed by subsequent injections of (A) 0.5 mM ADP (State 3o), (B) 2.5 μg/μl oligomycin (State 4), (C) 1 μM carbonyl cyanide trifluoromethoxyphenylhydrazone (FCCP) (State 3u) and (D) 4 μM antimycin-A. Data were analysed per individual mouse with each injection point assessed for successful response via pre-defined criteria. Injection measurements were excluded where oxygen consumption was reduced to 0 pmol/min, or where the compounds did not perform as expected for that individual replicate (e.g. no response to oligomycin). The average value of remaining replicates was used. Individual animals were excluded where ADP injection did not increase oxygen consumption rate (OCR).

Hydrogen peroxide (H₂O₂) production in renal cortical mitochondrial preparations was measured by fluorescence using the Amplex Red reagent (Molecular Probes, Invitrogen). In addition, H₂O₂ production was assessed in the presence of 10 mM glutamate and 10 mM malate for complex I, or, 10 mM succinate for complex II, plus 5 μM rotenone (a complex I inhibitor). Results were then normalised to no substrates group to determine effects of complex I and II stimulation.

The induction of mitochondrial permeability transition (mPT) in mouse kidney mitochondria was monitored by absorbance changes at 540 nm based on a microtiter plate procedure developed by Waldmeier et al. [15], with the following modifications. Mitochondria, isolated by differential centrifugation as described above were washed twice and resuspended in energised respiratory buffer (ERB) (210 mM mannitol, 70 mM sucrose, 5mM HEPES, 5 mM glutamate, 5 mM malate, 0.05 M Tris, pH 7.4). Mitochondrial protein, 100 μg, was added to each well in 110 μl ERB, then incubated with 10 μM CsA or ERB control where appropriate for 5 min before the addition of 5 μl of calcium solution (2.5 mM CaCl₂·2H₂O, 1 mM K₂HPO₄) or ERB. Total volume = 125 μl per well. Δabsorbance was determined over 10 min with a microtiter plate reader (Ppif study) or 15 min (db/db study) using an EnSpire multimode plate reader (Perkin Elmer, Waltham, MA, U.S.A.).

RNA was isolated from kidney cortex (20–30mg) using TRIzol Reagent (Life Technologies). Contaminating DNA was removed after treatment with DNA-free DNAse according to the manufacturer’s specifications (Ambion Inc, Austin, U.S.A.). DNA-free RNA was reverse transcribed into cDNA using the Superscript First Strand Synthesis System according to the manufacturer’s specifications (Life Technologies BRL, Grand Island, NY). Real-time PCR was performed using SYBR green PCR mix (Applied Biosystems, primer concentration of 500 nM) or Taqman PCR kit (18S rRNA TaqMan Control Reagent kit, ABI Prism 7500; Perkin-Elmer) using a 7500 Fast Real-time PCR System (Applied Biosystem, VIC, Australia), and normalized relative to 18S ribosomal RNA. All values are presented as fold change relative to Ppif^WT/WT CIT control or db/m vehicle control. The nucleotide sequences for primers examined are shown in Supplementary Table S1.

Citrate synthase activity was assayed as previously described via colorimetric release of acetyl coenzyme A in the presence of DTNB 5,5-dithiobis-2-nitrobenzoate [24]. Calculated activity levels were then normalised to total mitochondrial isolation protein concentrations.

Mitochondrial morphology (length and width) was determined in electron micrographs of proximal tubule epithelial cells. Immediately following exsanguination, renal cortices were fixed in 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer and tissue was processed for standard transmission electron microscopy using post-fixation in 1% OsO₄ and embedded in Epon resin. Ultrathin sections were cut and stained with lead citrate and uranyl acetate and imaged on a Hitachi H-7500 transmission electron microscope equipped with a Gatan Multiscan 791 CCD camera. Up to 10 images of proximal tubule epithelial cells per kidney section were randomly collected for each mouse at ×8000 magnification. Image field size was 17.44 × 17.44 µm each. Individual complete mitochondria were circled by hand for a subset of n=3 mice per group. Mitochondria in each image per mouse were hand circled and measured with (Fiji Is Just) ImageJ software and values used to calculate Form Factor (length/width) as previously described [25]. In addition, the 2D volume of mitochondria within each field per section were measured and expressed as mitochondrial percentage area per mm² cytosol. For purpose of cytosolic area determination nuclei, brush borders, and non-proximal tubular structures were excluded. Cumulative results for each mouse were used to express mean volume and aspect ratio per group.

Immunoblotting was performed as previously described with minor modifications [26]. Briefly, 8 μg of isolated mitochondria were lysed by three freeze-thaw cycles and separated by electrophoresis on 4–20% Mini-Protean TGX Stain-Free Gel (Bio-Rad laboratories, Gladesville, NSW, Australia), transferred to polyvinylidene fluoride (PVDF) membranes, before being blocked for 1 h in 5% skim milk in Tris buffered saline with 0.1% Tween-20 (TBS-T). Membranes were incubated overnight with gentle agitation at 4°C with 6.2 μg/μl total OXPHOS rodent antibody (ab110413, Abcam, Melbourne, Australia), before being incubated at room temperature in an IRDye® 800CW goat anti-mouse IgG secondary antibody (LI-COR Biosciences, Lincoln, NE, U.S.A.) for 90 mins. Membranes were visualised using an Odyssey Infra-REd imaginf platform (LI-COR Biosciences, Lincoln, NE, U.S.A.) and were quantified using Image Studio Lite v5.2. An internal standard was loaded on each gel to account for gel-to-gel variability. Whole lane stain-free quantification (Bio-Rad Chemi-Doc imaging system, and Bio-Rad Image Lab 5.2.1 imaging software; Bio-Rad laboratories, Gladesville, NSW, Australia) was used to normalise total protein per well, and qualitatively verify correct loading and equal transfer between lanes. Additionally, an internal control sample was applied across all gels and used to normalise data between gels.

All statistical computations were performed using GraphPad Prism version 7.0 (GraphPad Software, San Diego, California, U.S.A.). Values of experimental groups are given as mean, with error bars showing SD, unless otherwise stated. Outliers were only excluded where a valid biological or technical issue indicated exclusion. Two-way ANOVA was performed with unweighted means. Bonferroni post-hoc test was used to determine statistical significance and correct for multiple comparisons between each factor. P<0.05 was considered to be statistically significant. Additionally, significance in graphs is indicated as: *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001, where required a two-way ANOVA with repeated measures was performed on data representing different time points with ^§P<0.05.

As expected, hyperglycaemia and decreased body weight occurred with diabetes (Table 1). Additionally, diabetes-associated renal hypertrophy developed, shown as an increase in total kidney weight and kidney to body weight ratio. Diabetic mice had increased daily water and food consumption and developed polyuria. Genetic deletion of CypD did not alter fasting blood glucose concentrations in non-diabetic mice nor in the setting of diabetes (Table 1). There was a significant decrease in body weight in non-diabetic Ppif^−/− mice. CypD (Ppif) deletion also decreased polyuria at both 12 and 20 weeks of diabetes. Water consumption and food intake remained similar between diabetic Ppif^WT/WT and Ppif^−/− groups during the 24 h period.

Albuminuria (urinary albumin excretion) (Figure 1A) and urinary albumin to creatinine ratio were significantly increased in all STZ treated mice (Figure 1B). In addition, urinary excretion of the tubular damage-associated marker KIM-1 was increased in diabetic animals but was not different Ppif^−/− mice (Figure 1C). Plasma cystatin C, a surrogate marker of glomerular filtration rate, was reduced in the STZ model indicative of hyperfiltration (Figure 1D). Genetic deletion of CypD in the setting of diabetes led to a prevention of hyperfiltration, as shown by no decrease in plasma cystatin c in the Ppif^−/− STZ group. Conversely, deletion of CypD decreased GFR, measured as elevated cystatin C in both control and diabetic Ppif^−/− mice.

GSI was assessed in PAS-stained glomeruli (Figure 1E) and mean histology scores were significantly increased in all STZ treated animals compared with controls. Ppif^−/− STZ mice exhibited higher GSI histology scores compared with wild-type STZ indicating that the absence of CypD exacerbated glomerular injury in a diabetic setting. This was reiterated by an increase in diabetes-associated glomerular collagen IV deposition (Figure 1F) in the Ppif^−/− STZ mice. Thus, loss of CypD via deletion of the Ppif gene exacerbated glomerular structural injury in our model of DKD.

To confirm that the renal phenotype was indeed related to a loss of CypD, deletion of the gene encoding for CypD, Ppif was confirmed by quantitative reverse transcription-polymerase chain reaction (RT-qPCR). No expression of the Ppif gene was detected in the Ppif^−/− mice (Figure 2A). No change in Ppif expression was observed between WT control and STZ diabetic animals. To ascertain that deletion of Ppif did functionally reduce mitochondrial swelling, a calcium overload assay was performed in isolated mitochondria from Ppif^WT/WT and Ppif^−/− animals to induce mPTP opening (Figure 2B). Diabetic Ppif^−/− mice had decreased calcium-induced swelling as compared with WT diabetic mice. Since CypD binds to F₁/F₀ of the ATP synthase complex, we tested the gene expression of the F1-α component. Ppif^−/− mice had a marked down-regulation of this gene, in addition to a 50% reduction seen in the Ppif^WT/WT STZ treated mice (Figure 2C). Another mitochondrial gene, cytochrome c, was also down-regulated by both STZ treatment and in Ppif^−/− mice (Figure 2D). Gene expression of the fibrosis-associated fibronectin protein (Figure 2E) was increased 5-fold in Ppif^WT/WT STZ mice. Genetic deletion of Ppif^−/− in diabetic mice decreased the fibronectin gene expression similar to control levels. Collagen IV gene expression however was not increased with diabetes, and was down-regulated in Ppif^−/− mice (Figure 2F).

Db/db mice had increased body weight compared with db/m mice consistently over the course of the study (Figure 3A). A decline in body weight towards end point of the study was noted in some diabetic animals, which is considered to reflect the development of pancreatic β-cell failure. Alisporivir did not reduce diabetes-associated hyperglycaemia nor glycated haemoglobin levels in db/db diabetic mice (Table 2). Assessment of body composition by EchoMRI (Table 2) showed that the increase in body mass was attributable to a two-fold increase in fat mass in all db/db mice irrespective of therapy (Figure 3B). Endpoint body mass of db/db mice also did not differ between vehicle and drug treated animals at week 16 of treatment (Figure 3C). As expected, diabetic db/db animals developed renal hypertrophy with increased total kidney weight (Table 2). confirmed by an increase in the kidney to body weight ratio (Figure 3D) which was not affected by Alisporivir.

Plasma cystatin C as a marker of glomerular filtration rate, was unchanged for all db/db and db/m mice in the present study (midpoint—Figure 3E, endpoint—Figure 3F). Both urinary albumin excretion rate (midpoint—Figure 4A, endpoint—Figure 4B) and albumin to creatinine ratios (Figure 4C) were increased in all db/db animals consistent with DKD progression and was not attenuated by Alisporivir. Urinary KIM-1 was also increased in both db/db groups and unaffected by Alisporivir (Figure 4D). Db/db mice had increased Glomerular Sclerosis Index compared to db/m controls, however, Alisporivir had no impact on glomerular injury in db/db mice (Figure 4E). No increase in glomerular Col IV deposition between vehicle db/m and db/db was seen in the present study (Figure 4F), however, Db/m animals receiving Alisporivir had decreased Col IV deposition compared with vehicle controls. A comparison of these results and those of the Ppif study for the effects of diabetes and targeting Ppif (CypD) is provided in Table 3.

Daily physical activity was assessed by ‘beam breaks’ in X+Y+Z directions in the CLAMS system over 24 h. These data are shown as discrete measurements across the light (inactive/sleep) and dark (active) cycles (Figure 5A) and also as cumulative totals (Figure 5B). Db/m mice performed double the physical activity during the dark cycle (active period) compared to the light cycle. Db/db animals remained sedentary during dark cycle, with a significant reduction in physical activity compared to db/m controls.

Respiration was analysed via changes in oxygen consumption and carbon dioxide production. All individual mice had increased (^§P<0.05) respiration during the dark period compared to the light for both O₂ (Figure 5C) and CO₂ data (Figure 5D). No significant differences were evident between db/db versus db/m mice for either oxygen (Figure 5Ci) or carbon dioxide respiration (Figure 6Ci) measured in ml/h. There were, however, slight trends toward an increase in O₂ (Figure 5Ci) and a decrease in CO₂ (Figure 5Di) for db/db mice on vehicle but not Alisporivir. When normalised to body weight (ml/kg/h) this trend disappeared and both db/db groups were significantly decreased compared with their db/m controls (Figures 5Cii and 6Cii). Energy expenditure (heat production) was unchanged between all groups (Figure 5E) respiratory exchange ratio (RER), which is reflective of metabolic substrate preference, was calculated from the ratio of Volume of CO₂ to Volume O₂. Once again, values were higher in db/m mice during the dark cycle compared to the light, indicating increased carbohydrate metabolism during active hours. Interestingly, however, no such difference was observed in db/db mice. Additionally, the slight trends in O₂ and CO₂ respiration observed in Figure 5C,D synergistically resulted in a statistically significant decrease in RER in db/db animals with no change with Alisporivir (Figure 5F). This decrease in RER is reflective of a preference for fat metabolism in db/db mice. Overall, pharmacological inhibition of CypD had little effect on any parameter observed with the CLAMS system.

Next, markers of mitochondrial respiratory function and mPTP were assessed. Citrate synthase activity, an overall marker of mitochondrial quality, was determined and revealed no significant difference in enzyme activity between groups per μg of isolated mitochondria (Figure 6A) indicating that any differences observed in mitochondrial activity are attributable to functional measures rather than changes to overall mitochondria quality. The mitochondrial swelling response induced by calcium challenge showed no statistically significant differences for the rate of Δabsorbance at 540 nm (Figure 6B). The isolated mitochondrial preparations were also pretreated with the mPT inhibitor CsA, prior to receiving the calcium challenge. Interestingly, this pretreatment differentially affected the total duration of swelling inhibition between db/m and db/db mice, whereby mitochondria from the db/db mice underwent a rapid drop in absorbance (i.e. no swelling inhibition). Treatment with Alisporivir partially restored CsA inhibition of the mPTP in db/db mice, however, the mean group average was not significantly different from the vehicle db/db mice.

To assess mitochondrial respiration in the kidney, a mitochondrial stress test was performed in the Seahorse Bioanalyser using glutamate/malate to test complex I-linked respiration in isolated mitochondria (Figure 6C). A significant increase in OCR following ATP and FCCP injections was observed for vehicle treated db/db mice, but this was lessened in Alisporivir treated db/db mice. RCR (State 3/State 4o) and ATP-linked respiration of isolated kidney mitochondria were increased in vehicle db/db mice compared with db/m controls (Figure 6D). Although there was a trend towards a reduction in complex I-stimulated respiration in Alisporivir treated db/db mice this was not statistically significant. No differences in proton leak or FCCP stimulated respiration were observed.

H₂O₂ generated by isolated mitochondria, and measured by Amplex red fluorescence intensity was halved in the db/db mice compared to db/m controls with no difference between vehicle and Alisporivir treated db/db mice (Figure 6E). Percentage H₂O₂ generation with substrates was normalised to no substrates, and was found to be unchanged between groups (Figure 6F), therefore indicating that the decrease in total H₂O₂ was not substrate-dependent.

Renal cortical gene expression of several cell death and mitochondrial-associated proteins were assessed by RT-qPCR in the db/db model to determine the effect of diabetes and treatment with Alisporivir on these mitochondrial centred pathways (Figure 7A). Treatment with the CypD inhibitor Alisporivir did not alter the gene expression of its associated gene (Ppif) in db/m mice, however, in the context of diabetes Ppif expression was significantly increased. Interestingly there was an up-regulation of Cytc observed in the db/m mice treated with Alisporivir, however no change was present with diabetes in the db/db mice in contrast with the STZ model described above. Multiple other genes in both cell death and mitochondrial function were also assessed. Firstly, inflammation-associated TGF-β and macrophage-associated F4/80 gene expressions were unchanged. Secondly, expression of several cell death activator caspases were also unchanged, although there was a significant increase in the inflammation-associated caspase 12 (Human equivalent of CASP 4) in vehicle treated db/db diabetic mice only. Additionally, some fibrosis related genes, including collagen I and fibronectin, were up-regulated in vehicle db/db mice and partially attenuated in those treated with Alisporivir. Smooth muscle actin-alpha (α-SMA) and connective tissue growth factor (CTGF) did not change with diabetes.

Changes in genes associated with mitochondrial dynamics were minimal with the exception of a modest increase in Mief1 expression in db/db mice treated with vehicle which was not seen with Alisporivir. Finally, a fold change increase in the mitochondrial uncoupling protein-2 (Ucp-2) gene expression was statistically significant for db/db mice treated with and without Alisporivir.

Mitochondrial electron transport system (ETS) complex protein expression was not different between db/m and db/db mice with no change in complex I to V protein expression detected by Western blot (Figure 7B).

Transmission electron microscopy of dissected renal cortices of db/m and db/db mice at endpoint showed that mitochondrial dynamics were not altered in db/db mice compared with db/m mice as assessed by the mitochondrial shape descriptors for Form Factor (Supplementary Figure S1A). However, db/db mice exhibited an increase in 2D volume of mitochondria from approximately 30–35%/μm² of cytosolic area in the proximal tubule cells (Figure 7C). This may be reflected in the shift in the frequency distribution of mitochondrial area (Supplementary Figure S1B).

The mPTP has long been considered to play an important role in mediating the demise of mitochondria in response to overwhelming stress. Although the exact mechanism of pore activation remains under investigation, the protein CypD continues to be regarded as an important regulator of mitochondrial swelling and function [10,14,27–29]. In the present study we demonstrated that global deletion of the CypD gene, Ppif, in C57BL6 mice was not associated with the development of any overt systemic pathology. This is consistent with other reports using Ppif^−/− mice where phenotypes were limited to minor metabolic alterations, for example calcium handling and body fat metabolism [30–32]. Indeed, our study showed some significant differences between Ppif^−/− and Ppif^WT/WT in non-diabetic mice with changes limited to slightly reduced body weight and lower collagen IV deposition. The increase in cystatin C in Ppif^−/− mice indicated that deletion of CypD was protective against the development of diabetes-associated hyperfiltration. However, this elevated cystatin C could also be interpreted as a decline kidney function, even in non-diabetic mice, considering that the albumin to creatinine ratio was unchanged between Ppif^WT/WT and Ppif^−/−. Additionally, lack of CypD was associated with an increase in the structural damage in the kidney in our model of STZ induced DKD. This increase in renal damage, following deletion of a key mPTP sensitising protein, provides supporting evidence for the complex biological role of CypD beyond providing a failsafe to initiating mitochondrial cell death [33].

Data from our study also revealed that CypD (Ppif) deletion was associated with a decrease in the relative fold change expression of multiple genes tested by qPCR (plus additional genes not presented) compared to the WT controls. This reduction in gene expression may be explained by the ability of cyclophilins to regulate gene expression processes [34]. Although the qPCR data for fibrosis genes appear to suggest an improvement in Ppif^−/− STZ diabetic mice, the histology data may suggest that remodelling has already occurred.

It is generally well established that mPTP activity affects multiple organ systems, including liver function and skeletal muscle [30,35–37]. This is in part due to its role in regulating crucial mitochondrial functions and temporary mPTP depolarisation for physiological purposes [9,33,35]. Further, CypD has been shown to negatively regulate F₁/F₀ ATP synthase activity, by binding to the Oligomycin-Sensitivity Conferring Protein (OSCP) (which links the F1 portion with the lateral stalk) thereby limiting ATP coupling and energy production [38–40]. ATP synthase remains a reasonable candidate for the main component of the mPTP channel, making this interaction with CypD imperative [41,42]. The requirement of these processes in the diabetic kidney is highlighted by the apparent increase in glomerular structural damage observed in the Ppif^−/− mice in our study. Interestingly, the mechanism for why CypD deletion attenuates other diabetes-related complications, such as pancreatic function and cognitive decline, has been demonstrated previously to be modulated by inhibiting β-oxidation and increasing substrate metabolism through glycolysis [36,43]. However, metabolic alterations of CypD deletion have demonstrated variability in different organs for different genetic models, e.g. liver, skeletal muscle [34–36]. In the kidney, CypD has also been previously demonstrated to be a key regulator of cellular metabolism with deletion of the Ppif gene (CypD) leading to an increase in glycolysis at the expense of fatty acid oxidation (FAO) [44]. These changes in glycolytic metabolism are pertinent in the renal cortex where β-oxidation provides a crucial fuel source to PTECs and this switch in fuel generation is likely to be detrimental, as reduced capacity of FAO has been associated with increased renal fibrosis as a consequence of acute stress [45]. Further, disruption to FAO pathways, as observed in patients with CKD, is associated with renal fibrosis [46]. Diabetes is also independently associated with changes in renal cellular metabolism including reduced FAO alongside a total increase in lipid accumulation [47]. The combined effect of both loss of CypD and the chronic metabolic stress induced by diabetes could be expected to lead to an increase in renal fibrosis. Indeed, loss of CypD has been previously hypothesised to prime PTECs to stress [44]. Thus, a switch to glycolysis in the diabetic kidney, at the expense of β-oxidation, could explain why renal injury is recapitulated in our model.

Although complete loss of CypD, via Ppif deletion, was associated with negative consequences with respect to our STZ model of DKD, pharmacological blockade of CypD was still a pertinent option to explore as partial inhibition may allow maintenance of some physiological functions. The drug Alisporivir is a cyclosporine analogue which has demonstrated ability to bind CypD and prevent its participation in mPT pore formation, without conferring the immunosuppressive effects of earlier cyclosporines [48–50]. Although Alisporivir was initially developed by Debiopharm as an antiviral drug to target HIV and hepatitis C infection, it has demonstrated efficacy in improving mitochondrial function and associated pathological outcomes in a range of acute disease settings by preventing cell death and rescuing mitochondrial function [51–54]. The drug profile includes a demonstrated ability to reduce lesion size following cardiac infarction, and potentially as a prophylactic measure to protect kidney cells from renotoxic substances used during cardiac surgery [55,56]. This is a promising result since a wide range of commonly used drugs induce renal toxicity and require monitoring of kidney function. Some prominent clinical studies, however, question the efficacy of acute CypD targeted therapy (cyclosporines) in preventing cardiac pathologies [57,58]. Nonetheless, research into Alisporivir’s efficacy in treating chronic diseases is now being considered, with promising studies in mouse models of muscular dystrophy revealing the potential for a long-term therapeutic strategy to rescue mitochondrial function in certain chronic diseases [59,60].

Here, we reveal that Alisporivir at a dose of 5 mg/kg/day did not protect the kidney from the development of clinical nephropathy markers in DKD. Indeed, we may consider that higher therapeutic doses could have increased efficacy, however in our study, a dose of 10 mg/kg was associated with an attenuation of hyperglycaemia in db/db mice to db/m control levels (data not shown). This effect is a confounding variable as high blood glucose is a well-established risk factor for diabetic complications. Strict blood glucose control is the gold standard for clinical diabetes care; thus, any renal protection mechanism of the drug alone at such glucose lowering doses cannot be independently assessed. This result may be worth exploring in future studies to further elucidate alterations in glucose metabolism and glycaemic control with such an agent. Our data revealed that treatment with 5 mg/kg/day Alisporivir was not associated with an exacerbation of renal damage, in either control or diabetic animals. This is important as cyclosporines can induce significant nephrotoxicity in a dose dependent manner; mediated through autophagy, oxidative stress and ER stress [61–63]. Indeed, dose management concerns are often raised in patients prescribed analogues for immunosuppressant purposes [64–65]. Currently there are limited data available on the pharmacodynamics of Alisporivir in renal tissues. Our data indicates that 5 mg/kg/day Alisporivir was well tolerated in male db/db mice, providing evidence that this modified analogue may have reduced renal toxicity over traditional cyclosporine analogues.

Interestingly, although Alisporivir had minimal effect on renal injury, there were a number of changes observed in the renal cortical mitochondria of db/db mice compared with db/m controls. Some results were unexpected, but may align with previous models of mitochondrial protection. These include changes to fuel substrates, reactive oxygen species (ROS) generation and oxygen availability per milligram of mitochondria. It is already hypothesised that metabolic changes in diabetes affect fuel preferences in renal proximal tubule cells [66]. Indeed, studies of PTEC metabolism indicate a switch to a reliance on fatty acid synthesis is a protective measure against the high glucose present in the proximal tubule cells under normal conditions [67]. With insulin resistance a decreased availability of intracellular glucose occurs in peripheral tissue, however, as renal proximal tubules are responsible for both gluconeogenesis and glucose recycling from the urinary ultrafiltrate, glucose uptake and release is increased in these cells under diabetic conditions [68–70]. Although CypD deletion induced changes in renal function in our first study, we did not find any statistically significant change in any renal functional parameter with Alisporivir treatment in our db/db model, although some trends were evident. However, the trend in body weight data for example is likely to be an artefact of randomisation and stochastic variation in db/db body size given the body composition and percentage fat mass remained consistent between treatment groups indicating mice were just smaller, not less obese. This is also demonstrated with the results of the CLAMS. Indeed, body size has been shown to be directly correlated with respiration, thus no real difference can be attributed to Alisporivir [20].

Some diabetes-associated mitochondrial changes, however, were altered with Alisporivir, indicating that some pharmacological activity was occurring in the renal cortex. Our data reveal that both mitochondrial function, and mitochondrial content, were altered in the diabetic kidney. All functional data were assessed per microgram of mitochondrial protein to confirm that these changes were a constitutive adaptation and did not simply reflect the increased mitochondrial content observed in renal tissue. Mitochondrial ETS components (citrate synthase activity, protein expression of ETS complexes) were also unchanged per microgram of protein between all groups. Calcium challenge induced rapid mitochondrial swelling in both db/m and db/db renal cortical mitochondria. This illustrates that some level of mitochondrial impairment was present in the renal mitochondria as a slower rate of swelling is generally observed in liver and cardiac mitochondria in other models [4,71,72]. However, CsA was able to effectively inhibit swelling in mitochondria isolated from db/m but not db/db mice. This abnormality in db/db mice was able to be rescued, with Alisporivir treatment able to partially restore mPTP inhibition. These data indicate that the drug was present in the renal cortical mitochondria, even though it did not alter OCR nor H₂O₂ production. CsA has been shown to reduce maximum respiration rates in a dose-dependent manner mediated through calcium sequestration, yet we did not see any respiration changes with Alisporivir at our dose [73].

It is possible that our data reveal that renal mitochondria are still able to adapt to chronic diabetes stressors via endogenous protective mechanisms. For example, previously elucidated adaptations in db/db mice include increased ROS generation mediated by ETS uncoupling [74]. One adaptation to this involves the up-regulation of Ucp-2 which is directly activated by the presence of ROS, and has been previously shown to act as a positive feedback brake on damaging ROS by uncoupling ETS activity [75]. In our study we observed an increase in OCR, paired with a concomitant decrease in radical sequestration in the production of H₂O₂. This indicates that mitochondrial uncoupling was occurring in the db/db mice. As we also observed an up-regulation of Ucp-2 gene expression, our data are consistent with this model and other studies demonstrating the protective nature of Ucp-2 from oxidative stress [76,77].

Finally, hypoxia is known to be a stressor of PTECs with decreased intrarenal oxygen levels previously demonstrated under diabetic conditions [78,79]. Although this was beyond the scope of our study, it is possible that adaptations to hypoxia may also be responsible for changes in ETS efficiency and ATP generation in addition to those shown here. Many factors influence the behaviour of CypD and the formation of the mPTP in a complex interactive network, however, targeting CypD with Alisporivir was unable to prevent the development of DKD in our db/db mice.

The present study has demonstrated that neither genetic deletion, nor partial blockade of CypD with Alisporivir, was able to rescue the common pathologies associated with progression of DKD. We conclude that CypD is an unsuitable target for rescuing mitochondrial heath in DKD. Although targeting mitochondrial function via inhibition of the mPTP has been demonstrated as an effective intervention in the pathophysiology of diseases in other organs, renal proximal tubular cells have a unique requirement for CypD under diabetic conditions and are unable to circumvent its physiological function to prevent mPTP-associated pathology.

• Mitochondrial dysfunction, including enhanced susceptibility to mPTP, is an early event in the pathogenesis of DKD, however, the role of the mPTP and its activator, CypD, in DKD has not been fully explored.

• Genetic loss of CypD exacerbated diabetic renal injury, and treatment with a pharmacological inhibitor did not rescue mitochondrial function.

• CypD has a complex physiological and pathological role in the diabetic kidney which should be considered when treating DKD patients with CypD inhibitors.

The authors would like to thank Sih Min Tan, Sally Penfold, Brooke Harcourt, Amy Morley and Karly Sourris for their technical assistance and also thank the Cellular and Molecular Metabolism Laboratory of the Baker Heart and Diabetes Institute, Melbourne, AU, for use of the CLAMs and EchoMRI equipment. We thank Carlos Rosado for assistance with generating PCR primers specifically for this project. We also acknowledge Georg Ramm, Joan Clark and Adam Costin of the Ramacciotti Centre for Cryo-electron Microscopy, at Monash University for technical assistance; and Monash Micro Imaging, for use of their equipment. We thank Novartis for providing the Alisporivir.

The authors declare that there are no competing interests associated with the manuscript.

This work was completed with support from: the Australian government Research Training Program [Scholarship (to R.S.J.L.)]; the Australian Diabetes Society [Skip Martin Early Career Fellowship (to M.T.C.)], M.T.C. is currently the recipient of a Career Development Award from JDRF Australia, Australian Research Council Special Research Initiative in Type 1 Juvenile Diabetes [grant number 4-CDA-2016-229-M-B]; the postdoctoral fellowship from JDRF [grant number 1-PDF-2017-421-A-N (to G.C.H.)]; and the National Health and Medical Research Council Research Fellows [grant number 1102935 (to J.M.F.) and grant number 1175760 (to M.E.C.)].

Open access for this article was enabled by the participation of Monash University in an all-inclusive Read & Publish pilot with Portland Press and the Biochemical Society under a transformative agreement with CAUL.

R.S.J.L. ran the animal studies, performed experiments, researched the data and co-wrote the paper. M.T.C. conceived the studies, ran the animal studies, researched the data and co-wrote the paper. G.C.H. ran the animal studies, performed experiments and provided intellectual input on the data. T.V.N., M.A., C.G., V.T.B., M.S., and J.M.F. performed experiments and assisted in researching and interpreting the data. D.C.H. and M.E.C. provided intellectual input on the data. All authors reviewed and edited the paper.

A.

Alisporivir treated mice

ATP

adenosine triphosphate

CIT

citrate (vehicle control)

CKD

chronic kidney disease

Col IV

collagen type IV

CsA

cyclosporine A

CypD

cyclophilin D

db/db

diabetic, homozygous leptin receptor deficient (Lepr^db/Lepr^db)

db/m

lean, heterozygous leptin receptor deficient (Lepr^db/Lepr⁺)

DKD

diabetic kidney disease

ETS

electron transport system

FAO

fatty acid oxidation

FCCP

carbonyl cyanide trifluoromethoxyphenylhydrazone

GSI

glomerular sclerosis index

H₂O₂

hydrogen peroxide

KIM-1

kidney injury molecule -1

mPT

mitochondrial permeability transition

mPTP

mitochondrial permeability transition pore

OCR

oxygen consumption rate

OXPHOS

oxidative phosphorylation

Ppif

peptidylprolyl isomerase F

PTEC

proximal tubular epithelial cell

RCR

respiratory control ratio

RER

respiratory exchange ratio

ROS

reactive oxygen species

RT-qPCR

quantitative reverse transcription-polymerase chain reaction

STZ

streptozotocin

V.

vehicle control treated (10% Cremophor EL)

WT

wild type