Myo-inositol Oxygenase (MIOX) Overexpression Drives the Progression of Renal Tubulointerstitial Injury in Diabetes

Diabetes is a well-described metabolic disorder in which hyperglycemia adversely affects the homeostasis of multiple organs in humans, and the commonly encountered lesions include microangiopathy, neuropathy, retinopathy, and nephropathy (1–4). The latter, i.e., diabetic nephropathy (DN), has been exhaustively investigated, in terms of its pathogenetic mechanisms, in various animal model and cell culture systems (5–10). In this regard, the studies carried out over the last few decades suggest that all cells of the kidney, namely, glomerular podocytes, endothelial and mesangial cells, tubular epithelia, interstitial fibroblasts, and vascular endothelia, are known to be variably affected by hyperglycemia (8,11–16). The major focus in these studies had been to explore the pathogenetic mechanisms pertaining to the “glomerular” compartment, while the “tubulointerstitium” received much less attention (14,17–20).

Among the various pathogenetic mechanisms, reactive oxygen species (ROS) have been regarded as common denominators in the signaling cascade relevant to the pathogenesis of DN, and a majority of the publications refer to ROS in the context of glomerular pathology and a few also address the subject matter of “diabetic tubulopathy” (1,14,15,17–22). ROS can be generated as a result of perturbations either in the mitochondrial homeostasis or NADPH oxidase system (1,21–23). In this regard, it should be noted that increased activity of the polyol pathway leads to an altered redox state of pyridine nucleotides (NADPH-to-NADP⁺ and NAD⁺-to-NADH ratios). The key enzymes that are operative in the polyol pathway include aldose reductase and sorbitol dehydrogenase that are expressed in the “tubules” and modulate renal osmoregulation (24) (Supplementary Fig. 1A). Akin to the polyol pathway seems to be another pathway, i.e., the glucuronate-xylulose (G-X) pathway, in which perturbations in the NADPH-to-NADP⁺ and NAD⁺-to-NADH ratios are anticipated with consequential redox imbalance and oxidant stress in the “tubular” compartment (25) (Supplementary Fig. 1B). In this G-X pathway, exogenous or endogenous myo-inositol (MI) is catabolized by MI oxygenase (MIOX) to glucuronic acid, which is channeled subsequently into d-xylulose-5-phosphate (Supplementary Fig. 1B).

MIOX is a 32 kDa cytosolic enzyme expressed in the renal proximal tubules, and it is upregulated in states of hyperglycemia (26–28). Previous studies suggest that phosphorylation of MIOX’s serine/threonine residues enhances its enzymatic activity (27). Interestingly, MIOX promoter includes osmotic-, carbohydrate-, sterol-, oxidant-and antioxidant-response elements, and thus its transcription is modulated by organic osmolytes, high glucose, fatty acids, and oxidant stress (25–29). Since oxidant stress or ROS are regarded as central to the pathogenesis of DN, it is imperative to assess the lineal contribution of MIOX toward ROS-mediated injury in the renal tubulointerstitial compartment. The impetus to evaluate MIOX’s status in tubulointerstitial injury comes from the observations made in animal models of obesity and acute kidney injury (AKI) where MIOX upregulation was noted to be associated with increased cellular oxidant stress, especially confined to the tubular compartment (29,30). Moreover, a clinical study on patients with renovascular complications reported an association of type 1 diabetes with polymorphism of MIOX gene, which underscores its clinical significance in the pathobiology of DN (31), although a comprehensive analysis of the events responsible for the MIOX-ROS–mediated tubulointerstitial injury in DN needs to be included in experimental animal models. Besides ROS, endoplasmic reticulum (ER) stress has also been implicated in the pathogenesis of diabetes, obesity, AKI, and chronic kidney disease (7,9,32–35). Whether MIOX worsens ER stress in DN remains to be explored. Of note, our laboratory has reported that MIOX upregulation accelerates the ER stress in a tunicamycin mice model of chemical-induced acute tubular injury (36).

In view of the above observations, it is conceivable that cellular burden of both the oxidant and ER stress together accelerates the progression of DN in states of high-glucose ambience associated with MIOX upregulation. This contention is the subject matter of this seminal investigation with emphasis on the delineation of mechanisms relevant to the progression of tubulointerstitial injury, i.e., diabetic tubulopathy, using comprehensive in vitro and in vivo model systems; the latter includes MIOX mutant mice.

Generation of mice with overexpression of MIOX (MIOX-transgenic [MIOX-TG]) and MIOX knockout (MIOX-KO) has previously been described (30). Also, heterozygous Ins2^Akita mice (The Jackson Laboratory) were mated with MIOX-KO to generate double mutant mice, MIOX-KO/Ins2^Akita. The reagents were purchased from the vendors listed in Supplementary Table 1.

Eight-week-old male mice of various strains, i.e., wild type (WT) C57BL/6J, MIOX-TG, and MIOX-KO mice, were batched into control and streptozotocin (STZ) groups (n = 6). The latter received an intraperitoneal injection of STZ (150 mg/kg). After 1 week, mice with blood glucose levels >250 mg/dL were considered diabetic, and they were sacrificed after 4 months. Similarly, 8-week-old Ins2^Akita and Ins2^Akita/KO mice were sacrificed after 4 months. At the time of sacrifice, blood and urine samples and kidneys were collected. Serum creatinine and urea levels were estimated using QuantiChrom Creatinine Assay Kit and QuantiChrom Urea Assay Kit. Urinary albumin and creatinine concentrations were measured using Exocell kits to calculate albumin-to-creatinine ratio (ACR). Serum cystatin C and urinary KIM-1 levels were measured using ELISA assay kits. For albuminuria, SDS-PAGE analyses were also performed. The kidney cortices were processed for various other studies. All procedures used in this study were approved by the Institutional Animal Care and Use Committee of Northwestern University.

A full-length MIOX cDNA was generated by RT-PCR using sense (5′-TCGTGATAAGCTTATGAAGGTC GATGTGG-3′) and antisense (5′-ATCACACGGGATCCTCACCAGCTCAGGGT-3′) primers. HindIII and BamH1 sites (underlined) were introduced. Amplified PCR product was digested with respective restriction enzymes and cloned into pcDNA3.1. Transfection of pcDNA into human kidney (HK)-2 cells and generation of stable transfectants were performed as previously described (14). MIOX-overexpressing cells were used for various experiments as follows: ∼2 × 10⁵ cells were seeded onto 55 cm² culture dishes and maintained to achieve ∼80% confluency. Following trypsinization, ∼1 × 10⁵ cells were plated on the 2.2 cm² coverslips for morphological studies. Additional experiments included treatment of cells with high d-glucose (HG) (30 mmol/L) for 24 h or N-acetyl cysteine (NAC) (1 mmol/L) for 6 h. l-glucose was used as an osmotic control. For gene disruption studies, cells were grown in the presence of 50 μmol/L XBP1 siRNA.

For localization of the expression of proteins, immunofluorescence (IMF) studies were carried out on cells and kidney tissues (14,20,29,30). Briefly, HK-2 cells were seeded (1 × 10⁵) on the 2.2 cm² coverslips. They were subjected to various treatments and subsequently processed for IMF using primary and secondary antibodies and TO-PRO-3 iodide (TOPO) red dye as a nuclear marker. For renal expression studies, 4-μm-thick cryostat sections were prepared. They were air dried, washed with PBS, and incubated with primary and secondary antibodies. The cells or tissues were examined using an ultraviolet microscope. The experiments were performed in quadriplicates on kidney tissues harvested from four individual mice/cell culture experiments of each group.

Kidney slices (4–5 mm thick) were prepared. After dehydration, the slices were embedded in paraffin. Sections were prepared (4 μmol/L thick) and mounted on glass slides. Sections were air-dried, deparaffinized, hydrated, subjected to antigen retrieval, and used for detection of XBP1 (14,30). The experiments were performed in quadriplicates on kidney tissues harvested from four individual mice of each group.

The details of immunoblotting procedures are given in previous publications (14,29). Briefly, kidney cortices were diced into 1 mm³ fragments and homogenized in a radioimmunoprecipitation assay buffer. After centrifugation, the supernatants with equal amounts of protein (30 μg) were fractionated by SDS-PAGE and electroblotted onto polyvinylidene fluoride membranes. They were incubated with primary and secondary antibodies, and the autoradiograms were prepared using ECL Western Blotting Detection Kit (Amersham). The immunoblots were performed on the lysates isolated from four different animals individually, and the lysates were not pooled. The blots depicted in various figures are representative of the data of four individual mouse kidneys from each group.

MIOX activity was carried out as described previously (20,26,29). The MI concentration in kidney cortices was measured using a Megazyme assay kit as per manufacturer instructions. Approximately 20 mg kidney tissue was homogenized in 400 μL PBS followed by centrifugation at 10,000g for 5 min at room temperature. Supernatant was collected, and 100 μL sample was incubated in a reaction mixture containing 100 μL ATP, 400 μL H₂O, 100 μL solution 1 (provided in the kit, pH 7.5) and 20 μL hexokinase at 22°C for 15 min. To this supernatant-reaction mixture, 1 mL solution 4 (pH 9.5), 500 μL NAD/ iodonitrotetrazolium chloride, and 20 μL diaphorase were added. After 3 min, first absorbance (A₁) was measured at optical density 492 nm (OD_{492 nm}). Ten minutes later, 20 μL MI dehydrogenase was added, and second absorbance (A₂) was measured at OD_{492 nm}. The concentration of MI in the sample was calculated as follows: C = (2.26 × 180.16/19900 × 1 × 0.1) × A₂ −A₁, where C is concentration. Finally, the concentration in the kidney sample was calculated as follows: C_inositol/weight_sample × 100 and expressed as mg/100 g. Likewise, the MI concentration in 100 μL serum was calculated and expressed as milligrams per deciliter.

For oxidant stress, kidney tissues and HK-2 cells that had undergone various treatments were stained with 10 μmol/L of either 2′-7′-dichlorofluroescein diacetate (DCF-DA) or dihydroethidium (DHE) dye and then photographed and evaluated as previously described (29,30). Glutathione (GSH)-to-glutathione disulfide (GSSG) ratio was measured using GSH assay kit (Cayman Chemical) (30). NAD⁺-to-NADH ratio was determined by using a colorimetric assay kit (Abcam). Twenty milligrams of kidney cortex were homogenized in 400 μL extraction buffer supplied in the kit and then centrifuged at 10,000g for 5 min at 4°C. Supernatant was collected, and it was passed through a 10 kDa spin column (ab93349; Abcam). For total NAD estimation, samples were left at 4°C. For NADH estimation, NAD⁺ was decomposed by incubating 200 μL sample at 60°C for 30 min. The reaction was terminated by placing the samples on ice. Samples were briefly spun to remove precipitates. The reaction was set up by using 50 μL standard or sample, 100 μL NAD cycling mix (98 μL NAD cycling buffer and cycling enzyme) (37). Samples were incubated at 22°C for 5 min for the conversion of NAD to NADH. Ten microliters of developer was added to the samples, and they were left at 22°C. Readings (OD_{450 nm}) were recorded for 1–4 h. Concentration was measured using a standard curve.

HG-induced DNA damage was assessed by TUNEL procedure (14,29). In brief, tissue sections were deparaffinized, hydrated with decreasing concentrations of ethanol, and then digested with proteinase K (240 units/mL; Promega, Madison, WI) for 15 min at 37°C. After washing the sections with PBS twice, nicked DNA was enzymatically labeled with a fluorescent nucleotide probe (Roche). Slides were coverslip mounted with a drop of mounting media and photographed using an ultraviolet microscope.

HK-2 cells transfected with empty vector or MIOX pcDNA were subjected to HG (30 mmol/L) ambience, and in a separate experiment they were concomitantly treated with XBP siRNA. Cells were washed with cold PBS, scraped from the culture dishes, and pelleted by centrifugation at 4°C. They were used for nuclear protein isolation (27,29). Isolated nuclear extracts were used for gel shift assay to determine the binding of XBP1 transcription factor to MIOX promoter. Detailed in silico analysis (TRANSFAC, version 8.3) of MIOX promoter revealed a putative binding site for XBP1. In view of this, a complementary double-stranded oligo having the putative binding site for XBP1 in the MIOX promoter was custom synthesized. The sequence for XBP1 oligo was as follows: 5′-CCTGGCCACGCTCCAAGACG-3′ (−992 to −988). The underlined sequence is the predicted binding site for XBP1. The oligo was labeled with ATP [γ-32]. Binding reaction was performed in a volume of 20 μL containing 1× binding buffer for 15 min at 37°C. The reaction mixture was subjected to nondenaturing 7.5% PAGE to assess DNA-protein interactions. The gels were air-dried and autoradiograms developed.

Details of chromatin immunoprecipitation (ChIP) assay have previously been described (27,29). Briefly, ∼30 mg renal cortical tissue was pulverized in the presence of liquid nitrogen. After cross-linking of DNA and proteins with formaldehyde, the DNA was sonicated to generate 200–1,000 base pair fragments. The sample was centrifuged for 10 min at 10,000g at 4°C, and the supernatant was used for immunoprecipitation with XBP1 antibody. DNA-protein-antibody complex was digested with proteinase K, followed by phenol/chloroform extraction and ethanol precipitation to isolate the pure DNA. The DNA was subjected to quantitative PCR analyses. Primers used for ChIP PCR encompassed the XBP1-binding region of MIOX promoter (−152 to −1). Their sequences were as follows: 5′-AGGAAGGAGCATGGTCACTT (sense) and 5′-CCTGAGGGAGCAGT CACCCG-3′ (antisense). Amplification of target amplicon was reflected as a single primary peak in the melt curve, and secondary peaks of nonspecific PCR products were not discernible. Results of quantitative PCR were analyzed using the 2^−ΔΔCp method, and crossing point (Cp) values were normalized with input samples. The extent of binding of XBP1 with MIOX promoter was expressed as x-fold enrichment compared with the control sample. Mouse IgG served as a negative control.

Statistical analyses were carried out using GraphPad Prism (version 7.01). The significance was determined using one-way ANOVA with Dunn multiple comparisons. Results were expressed as mean ± SD of six samples in each variable.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

MIOX has been reported to be upregulated in diabetic states, and conceivably its overexpression leads to perturbations in the kidney homeostasis and thereby renal injury (14,27,29). Here, we describe key events and conceivable mechanisms that lead to various renal pathophysiologic disturbances in different strains of mice with variable expression of MIOX, i.e., C57BL/6J (WT), MIOX-TG, and MIOX-KO, and in mice involved in rescuing of renal injury, i.e., Ins2^Akita versus Ins2^Akita/KO. The rationale for using an additional well-established genetic mouse model of type 1 diabetes, i.e., Ins2^Akita, was to assess whether or not tubulointerstitial injury can be rescued in these mice when they are crossbred with MIOX-KO mice.

MIOX-TG and Ins2^Akita mice had relatively low body weights, and they were further decreased in mice in a STZ-induced diabetic state (Fig. 1A). The blood glucose levels were high in all strains of mice following induction of diabetes and in Akita mice, and they did not change appreciably in the double mutant (Ins2^Akita/KO) mice (Fig. 1B). Serum creatinine and urea levels remained low with a slight increase in diabetic MIOX-KO mice (Fig. 1C and D). Ins2^Akita/KO mice had overtly decreased levels of creatinine and urea as compared with Ins2^Akita mice (Fig. 1C and D). Parallel changes were observed for the ACR and cystatin C levels in various strains of mice (Fig. 1E and Supplementary Fig. 2). Urinary KIM-1 levels increased in diabetic WT, MIOX-TG, and MIOX-KO mice, with the highest being in the MIOX-TG mice (Fig. 1F). The Ins2^Akita mice also had high levels of KIM-1, and they decreased significantly in Ins2^Akita/KO mice (Fig. 1F). SDS-PAGE analyses of urine samples showed notable excretion of albumin in control MIOX-TG mice, which significantly increased in diabetic MIOX-TG mice (Fig. 1G, arrow and arrowheads). The Ins2^Akita mice had mild excretion of albumin, but it was significantly reduced in Ins2^Akita/KO mice (Fig. 1H, arrowhead). Here, it is worth mentioning that under basal conditions (nondiabetic) no discernible differences in renal morphological features were observed among various strains of mice, i.e., WT, MIOX-TG, and MIOX-KO (Supplementary Fig. 3).

Immunofluorescence microscopy revealed an increased expression of MIOX in diabetic WT mice (Fig. 2B vs. Fig. 2A). Renal cortical tubules had a basal expression of MIOX, which increased in diabetic MIOX-TG mice (Fig. 2D vs. Fig. 2C). Minimal expression of MIOX was observed in MIOX-KO mice (Fig. 2E and F). A marked expression of MIOX was seen in Ins2^Akita mice, which was notably reduced in Ins2^Akita/KO mice (Fig. 2H vs. Fig. 2G). Likewise, immunoblotting experiments revealed similar changes in MIOX expression pattern among different strains of mice (Supplementary Fig. 4). Basal enzymatic activity of MIOX was discernible in WT, MIOX-TG, and Ins2^Akita mice (Fig. 2I). MIOX activity was high in MIOX-TG mice, and it notably increased in a diabetic state. MIOX activity was relatively high in Ins2^Akita mice, and it was reduced in Ins2^Akita/KO mice (Fig. 2I). The MI concentration was very high both in serum and kidney cortices of MIOX-KO mice, and it was reduced to varying degrees in diabetic WT, MIOX-TG, and MIOX-KO mice (Fig. 2J and K). MI concentration was low both in serum and kidney cortices of Ins2^Akita mice, and it notably increased in Ins2^Akita/KO mice (Fig. 2J and K).

Oxidant stress is regarded as one of the major denominators in the pathogenesis of DN (38). In view of this biologic precept and the association of increased MIOX expression with oxidant stress in models of AKI (30), we assessed the status of ROS in various strains of mice. An increased 2’-7’-dichlorofluroescein diacetate (DCF-DA)–associated fluorescence was observed in the tubular cytosolic compartment in diabetic WT mice (Fig. 3B vs. Fig. 3A). In MIOX-TG mice, the fluorescence could be seen in the nondiabetic mice, and it markedly increased in diabetic MIOX-TG mice (Fig. 3D vs. Fig. 3C). MIOX-KO mice had a mild increase of DCF-associated fluorescence in a hyperglycemic state (Fig. 3F vs. Fig. 3E). Ins2^Akita mice had a notably high DCF-associated fluorescence under basal conditions, and it decreased in Ins2^Akita/KO mice (Fig. 3H vs. Fig. 3G). The status of mitochondrial ROS was evaluated using DHE, and the perturbations were similar to those seen for cytosolic ROS. Both WT and MIOX-KO mice had minimal DHE-associated fluorescence (Fig. 3I and M). However, basal DHE reactivity was readily noticeable in MIOX-TG mice and it increased markedly in a hyperglycemic state (Fig. 3L vs. Fig. 3K). Ins2^Akita mice also had a notable DHE-associated fluorescence, and it was reduced in Ins2^Akita/KO mice (Fig. 3P vs. Fig. 3O). These redox perturbations were also reflected in other oxidant stress parameters. NOX4 expression increased in a hyperglycemic state—highest in diabetic MIOX-TG mice while low in MIOX-KO mice (Fig. 3Q, upper panel). The Ins2^Akita mice also had marked NOX4 expression, which was mitigated in Ins2^Akita/KO mice (Fig. 3Q, lower panel). Similarly, under basal control conditions, the NAD⁺-to-NADH and GSH-to-GSSG ratios were quite high in WT and MIOX-KO mice and relatively low in MIOX-TG and Ins2^Akita mice (Fig. 3R and S). The ratios were decreased in a diabetic state and were even lower in diabetic MIOX-TG mice. Both NAD⁺-to-NADH and GSH-to-GSSG ratios were low in Ins2^Akita mice kidneys, and they were increased in Ins2^Akita/KO mice (Fig. 3R and S), suggesting that MIOX-KO mice normally are shielded from oxidant stress and MIOX gene disruption exerts a rescuing effect.

To establish the notion that ROS generation is directly related to MIOX overexpression, we subjected HK-2 cells transfected with MIOX pcDNA to HG (30 mmol/L) ambience. Transfection led to an increased DHE fluorescence under low glucose (LG) (5 mmol/L) ambience, which increased tremendously under HG (Fig. 4A–D). DHE fluorescence was related to ROS generation, since NAC treatment reduced the fluorescence (Fig. 4E and F). A mild fluorescence was observed in cells subjected to LG or HG ambience with l-glucose, and it increased slightly following MIOX pcDNA transfection (Fig. 4G–J). These changes were confirmed by flow cytometric analyses (Fig. 4K–P). The quantitation revealed an increase in the mean fluorescence intensity in cells subjected to HG ambience but more so in cells overexpressing MIOX, while it was notably reduced following NAC treatment (Fig. 4Q).

ROS-induced DNA damage (apoptosis) was assessed by the TUNEL method. An increased apoptosis was observed in diabetic WT mice (Fig. 5B vs. Fig. 5A). It was highly accentuated in diabetic MIOX-TG mice (Fig. 5D vs. Fig. 5C). Minimal apoptosis was observed in diabetic or nondiabetic MIOX-KO mice (Fig. 5E and F). Fulminant apoptosis was observed in Ins2^Akita mice, which was reduced in Ins2^Akita/KO mice (Fig. 5G and H). Parallel changes were observed in the executionary molecules of apoptosis. An increased expression of Bax and cleaved caspase-3 was observed in diabetic WT mice, while Bcl2 was decreased (Fig. 5I). Bax and cleaved caspase-3 basal expression was relatively high in MIOX-TG mice and increased further in a diabetic state, while Bcl2 was reduced. Interestingly, basal expression of Bax and cleaved caspase-3 was high in Ins2^Akita mice and was reduced in Ins2^Akita/KO mice (Fig. 5J).

The foremost upstream kinase of TGF-β signaling would conceivably be phosphoinositide-dependent kinase-1 (PDK-1), a master kinase that regulates the activity of other kinases (39). Its activity increases following phosphorylation under the influence of growth factors and plausibly also by MIOX-induced oxidant stress (40). An increased expression of phosphorylated (p)PDK-1 was observed in kidneys of diabetic WT, MIOX-TG, and MIOX-KO mice, with the highest being in diabetic MIOX-TG mice (Fig. 6A). The pPDK-1 expression was also high in Ins2^Akita mice and low in Ins2^Akita/KO mice (Fig. 6B). PDK-1 phosphorylates a number of kinases, including PKC, and renders them catalytically active (41). The pPKC expression paralleled that of pPDK-1 (Fig. 6A and B). PKC is upregulated in diabetic states, and it modulates the expression of TGF-β (42,43). TGF-β expression was highest in diabetic MIOX-TG and Ins2^Akita mice, and it was reduced in Ins2^Akita/KO mice (Fig. 6B). Both Smad3 and Smad4 exhibited changes similar to those seen in the upstream kinases and TGF-β (Fig. 6A and B). These observations suggested that the entire spectrum of signaling cascade is modulated by the overexpression of MIOX, most likely via augmentation of oxidant stress.

Recent reports indicate a role of ER stress in the pathogenesis of DN (33,44). In this regard, two key ER sensors, i.e., X-Box Binding Protein 1 (XBP1), a transcription factor, and 78 kDa glucose-regulated protein (GRP78), a chaperone, were investigated. An accentuated renal expression of XBP1 (nuclear) and GRP78 in diabetic MIOX-TG mice was observed, suggesting that MIOX augments ER stress (Fig. 7A). The GRP78 expression was high in Ins2^Akita mice, and it was reduced in Ins2^Akita/KO mice (Fig. 7B). Our in silico analysis revealed the presence of XBP1-binding motifs in the MIOX promoter. XBP1 expression paralleled that of GRP78 with MIOX overexpression in the cytosolic fraction in diabetic WT and MIOX-TG mice, but more so in the nuclear fraction, suggesting cytoplasm-to-nuclear translocation (Fig. 7A). XBP1 nuclear translocation was readily seen in Ins2^Akita mice, and it was reduced in Ins2^Akita/KO mice (Fig. 7B). Immunohistochemistry procedures also showed a marked nuclear translocation/localization in renal tubules of diabetic MIOX-TG and Ins2^Akita mice, and it was notably reduced in Ins2^Akita/KO mice (Fig. 7F vs. Fig. 7E and J vs. Fig. 7I, arrows). These observations suggest a possible intertwined and a cyclical-feedback relationship of MIOX and XBP1 in the progression of diabetic tubulopathy. Such an increased DNA-protein interaction in states of MIOX overexpression was also reflected in the ChiP assays where a marked XBP1 enrichment was seen in diabetic MIOX-TG and Ins2 ^Akita mice, which was remarkably reduced in Ins2^Akita/KO mice, suggesting that these in vivo interactions are reduced by MIOX gene disruption (Fig. 7K). Electrophoretic mobility shift assays also revealed increased DNA-protein interaction under HG ambience, which was augmented by MIOX overexpression, as indicated by the increased intensity of the band (Fig. 7L, arrowhead, lane five). XBP1 siRNA reduced these DNA-protein interactions, as reflected by decreased intensity of the band (Fig. 7L, arrowhead, lane six).

To tease out the intricate interrelationship between MIOX and XBP1 relevant to ER stress and TGF-β–initiated fibrogenesis, we performed in vitro experiments. Downstream of TGF-β is the activation of pSmad2/3, which translocates into the nucleus and binds to the promoter of extracellular matrix (ECM) molecules and thereby causing fibrosis (45). Under LG (5 mmol/L) ambience, minimal pSmad2/3 translocation was observed, since 95% nuclei retained TOPO red dye fluorescence (Fig. 8A, red-white arrows, and Fig. 8G). A marked pSmad2/3 translocation was observed under HG (30 mmol/L) ambience, and it was augmented in MIOX-overexpressing cells (Fig. 8C and D, black-yellow arrowheads, and Fig. 8H). The yellow color of nuclei indicates colocalization of dye TOPO (red) and anti-pSmad2/3 antibody (green). Treatment of XBP1 siRNA decreased pSmad2/3 nuclear translocation, as highlighted by reduced population of yellow-colored nuclei (Fig. 8F, black-yellow arrowheads). Parallel results were observed in immunoblotting studies (Fig. 8I). Interestingly, the increased expression of fibronectin (Fn) was notably reduced with XBP1 siRNA, suggesting that ER and oxidant stress together may contribute to the increased de novo synthesis of ECM molecules under HG ambience in MIOX-overexpressing cells (Fig. 8I).

A mild increased renal expression of Fn and Collagen-1 (Coll-1) was observed in diabetic WT mice (Fig. 9B and J), whereas a markedly increased expression was noted in diabetic MIOX-TG mice (Fig. 9D and L). MIOX-KO mice had minimal expression of ECM proteins (Fig. 9E, F, M, and N). Parallel changes were observed in the Fn expression by immunoblot analyses (Fig. 9Q). A high expression of Fn and Coll-1 was observed in the tubulointerstitium of Ins2^Akita mice (Fig. 9G and O), and it was notably reduced in Ins2^Akita/KO mice (Fig. 9H and P). Likewise, parallel changes were observed in the expression of Fn by immunoblotting (Fig. 9R). These observations suggested that MIOX overexpression primes the exacerbation of tubulointerstitial injury, conceivably by induction of oxidant and ER stress, and modulation of signaling events leading to increased synthesis of ECM proteins.

This investigation is based on the notion that MIOX expression modulates outcome of tubulointerstitial injury in diabetic state by regulating various signaling pathways, and our results attest to this contention and yield a glimpse at the relevant mechanisms.

At the outset, a germane question relates to body homeostasis pertinent to renal pathophysiology in various strains of mice. Among them, MIOX-TG mice had the lowest body weight, which decreased further in diabetic state (Fig. 1), which may due to the deficiency of MI. In this regard, MI supplementation has been reported to exert a growth-promoting effect in other systems (46). Such dietary supplemental experiments relating to kidney pathobiology need to be performed, which would be the subject matter of future investigations. In addition, MIOX-TG diabetic mice had greater worsening of renal functions, as reflected in the serum creatinine, cystatin C, and urea levels; ACR and urinary protein excretion; and KIM-1 concentration. Such deterioration of renal functions has been observed in MIOX-TG mice in other forms of chemical injuries (30,36), which was thought to be due to the oxidant stress induced by MIOX overexpression. Since oxidant stress is well-known in the progression of DN, it is likely that excessive generation of ROS played a role in worsening of renal injury. Here, it is reasonable to assume that MIOX gene disruption (MIOX-KO) would ameliorate renal injury. This notion is supported by the results in double mutant (MIOX-KO × Ins2^Akita) mice where renal derangements were notably lessened. Of note, Ins2^Akita/KO mice had minimal expression/activity of MIOX and high tissue and serum MI levels, suggesting that MIOX deficiency rescues from injury, while hyperglycemia-induced increased MIOX expression in diabetic MIOX-TG mice worsens renal homeostasis (Fig. 2). Here, it is conceivable that MI deficiency worsens the renal homeostasis, as elegantly surmised by Chang et al. (28). Nevertheless, the data included in results also suggest that other factors, e.g., oxidant stress, play a significant role in the progression of renal injury, as described in cisplatin-induced AKI models (30,47).

Interestingly, ROS-mediated perturbations were confined to proximal tubules where MIOX is expressed (Fig. 3). Potential sources of ROS generation include mitochondrial respiratory chain, polyol pathway, advanced glycation end products, NADH/NADPH, and possibly the G-X pathway (15,21,22,30,38). ROS are capable of oxidizing proteins, lipids, carbohydrates, and DNA, including mitochondrial DNA, thereby causing cellular dysfunctions. The nephron segment that is most adversely affected is the proximal tubule for many reasons. First, 90% of filtered glucose is taken up by them, which is mediated by transporters like SGLT2 and GLUT2 (48). As a consequence, there is increased glycolysis with excessive generation of ROS via various mechanisms. Secondly, the proximal tubules are enriched with mitochondria that are involved in the generation of ATP via oxidative phosphorylation in the respiratory chain complex. With the increased load on this complex, there will be excessive leakage of electrons at various levels of respiratory chain with excessive generation of superoxide anion (O₂⁻) (15,21,22). Thirdly, NAD(P)H oxidase has been identified in the proximal tubules, i.e., NOX4 (49). NAD(P)H oxidase catalyzes the generation of O₂⁻ by reducing molecular O₂ using either NADPH or NADH. Fourthly, the proximal tubular mitochondria do not synthesize glutathione but derive it with the aid of organic anion carriers (49,50). Lastly, the upregulation of MIOX may lead to changes in NAD⁺-to-NADH ratio and perturbations in cellular redox via the G-X pathway (Supplementary Fig. 1). Since MIOX promoter includes oxidant response elements, oxidant stress would lead to cyclic upregulation of MIOX and a relentless oxidant stress (27). In view of the above, it is likely that proximal tubules have an enhanced vulnerability to oxidant stress. Apparently, tubular expression of MIOX and its cyclical upregulation following hyperglycemia make the matter worse. This contention is supported by our studies describing increased oxidant stress in diabetic MIOX-TG mice, as highlighted by DCF staining that detects H₂O₂ (Fig. 3). Also, a reduction of DCF staining in kidneys of Ins2^Akita/KO mice strengthens our contention. The major source of O₂⁻ is the respiratory chain, and it would reflect mitochondrial status, detected by DHE staining. DHE is seen as nuclear fluorescence, since the oxidized DHE intercalates into the DNA (51,52). Like DCF, the DHE staining was also markedly increased in diabetic MIOX-TG mice, while reduced in Ins2^Akita/KO mice (Fig. 3). Of note, ROS generation by mitochondria is influenced by NADH and FADH₂ in the cytosol, since they serve as electron donors to the mitochondrial respiratory complex (15). On the other hand, the O₂⁻ generated in the mitochondria readily translocates into the cytosol (53), meaning thereby that the biology of ROS generation in these compartments is intertwined. Another potential source of ROS generation identified in our study was the NADPH oxidase system (49). An increased expression of tubular NOX4 in MIOX-TG and Ins2^Akita mice, while notably reduced in Ins2^Akita/KO mice, would be in line with the above assertion (Fig. 3). Other perturbations where MIOX seems to augment ROS-mediated cellular injury include altered ratios of NAD⁺ to NADH and GSH to GSSG (Fig. 3). During glycolysis, glucose is metabolized into pyruvate with the conversion of NAD⁺ to NADH; the latter serves as an electron donor (15,21). Also, glucose is handled via polyol pathway where NAD⁺ is converted to NADH (54,55). With the increased availability of glucose and augmented activity of various pathways, an increased consumption of NAD⁺, its deficiency, and altered NAD⁺-to-NADH ratio would be anticipated. Likewise, altered GSH-to-GSSG ratio, a measure of oxidant stress, has been reported in DN (56,57). Here, it is conceivable that an altered ratio may be because MIOX per se contributes, to a certain extent, to the depletion of GSH via reducing NADPH levels (58). Nevertheless, it is known that GSH protects the cell by neutralizing the ROS, and excessive amounts of ROS in the cellular pool would be anticipated to cause notable deficiency of GSH and thereby an altered ratio (58). Also, hyperglycemia associated with increased activity of glutathione peroxidase and low levels of dicarboxylate and 2-oxoglutarrate would also contribute to GSH deficiency in diabetic tubulopathy (50). To address the issue that MIOX per se can cause generation of ROS, cell culture experiments were performed (Fig. 4). HK-2 cells transfected with MIOX pcDNA had increased DHE fluorescence at LG ambience, which was augmented at HG ambience. It was reduced by N-acetylcysteine treatment, suggesting that DHE fluorescence is related to mitochondrial ROS. These morphologic findings were also supported by flow cytometric analyses (Fig. 4). One of the well-described hyperglycemia-induced ROS-mediated damages is apoptosis, which is associated with changes in the profile of Bax and cleaved caspase-3 (15). ROS facilitate the release of cytochrome C from mitochondria with increased permeabilization of their outer membrane, formation of apoptosomes, caspase 3/8 activation, and tubular apoptosis (49,59). The accentuated expression of Bax and cleaved caspase-3 in MIOX-TG mice indicates that these alterations are related to MIOX-mediated increased oxidant stress. The fact that these perturbations were normalized in Ins2^Akita/KO mice would support the contention that augmented apoptosis is related to the increased ROS generation (Fig. 5).

Besides apoptosis, ROS can initiate the activation of various signaling kinases, e.g., PKC, followed by induction of growth-promoting cytokines, i.e., TGF-β, leading to tubulointerstitial injury/fibrosis (60). Other signaling cascades that are also ROS activated include PKB/Akt and PDK-1, although studies describing their role in ROS-mediated induction following hyperglycemia are limited (40,49,61). The fact that we observed increased expression of pPDK-1 in diabetic MIOX-TG strongly suggests that hyperglycemia/MIOX-induced ROS generation modulates pPDK-1 expression (Fig. 6). Moreover, reduction of pPDK-1 expression in MIOX-KO and Ins2^Akita/KO mice would suggest that the MIOX expression profile is intricately involved in the activation of pPDK and the parallel downstream events, including pPKC, increased TGF-β expression, and Smads’ activation. This indicates that the entire signaling cascade initiated by pPDK-1 is operational, which would lead to tubulointerstitial injury/fibrosis (Fig. 6).

Conceivably, oxidative damage to proteins leads to their inappropriate folding and ER stress; the latter also plays a role in the pathogenesis of DN (33,34). That there was an increased renal expression of GRP78, an important chaperone in unfolded protein response, and an increased nuclear translocation of XBP1, a key transcription factor of unfolded protein response, in the tubular compartment of diabetic MIOX-TG, would suggest that they are afflicted by ER stress (Fig. 7). Moreover, normalization of these changes to baseline in Ins2^Akita/KO mice and minimal nuclear translocation of XBP1 in MIOX-KO mice would suggest that ER stress accentuated by MIOX overexpression plays a role in the pathogenesis of diabetic tubulopathy (Fig. 7). The relevance of MIOX in ER stress is further strengthened by the discovery of XBP1-binding elements in MIOX promoter, suggesting a reciprocal relationship between MIOX and XBP1 in which they can regulate each other. Also, XBP1 can regulate the signaling cascade that leads to tubulointerstitial fibrosis, since MIOX-overexpressing cells had an increased expression of TGF-β, Smad3, and Fn under HG ambience, while significantly decreasing with the XBP1 siRNA treatment (Fig. 8). The observation that XBP1 siRNA also dampened pSmad2/3 nuclear translocation strongly suggests that the molecules that are an integral part of ER stress response also modulate the signaling cascade that downstream regulate the synthesis of ECM proteins like Fn or collagen-1 (Coll-1) (Fig. 8).

The most downstream event of hyperglycemia-induced injury is increased synthesis of ECM proteins, leading to interstitial fibrosis (15,18,20,42,62). Likewise, we observed slightly increased expression of Fn/Coll-1 in diabetic WT mice (Fig. 9). An interesting finding was marked increased expression of Fn/Coll-1 in diabetic MIOX-TG, attributable to MIOX overexpression and a multitude of downstream signaling events. The most important finding of this investigation was the remarkably reduced expression of Fn/Coll-1 in the tubulointerstitial compartment in Ins2^Akita/KO mice compared with Ins2^Akita mice (Fig. 9).

In summary, this investigation highlights that MIOX overexpression worsens the outcome of DN, possibly by accentuating the activity of oxidant and ER stress processes, while its gene ablation ameliorates tubulointerstitial injury. Finally, it is quite conceivable that these processes may mutually boost the actions of each other, which would accelerate the progression of diabetic tubulopathy (Supplementary Fig. 5).

Acknowledgments. The authors thank K.S. Pabhu (The Pennsylvania State University) for allowing them to adapt Supplementary Fig. 1B (25).

Funding. This study was supported by National Institute of Diabetes and Digestive and Kidney Diseases, NIH, grant DK60635.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. I.S. designed and performed the experiments. F.D., Y.L., and Y.S.K. wrote and edited the manuscript. I.S. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.