ASK1 Inhibitor Halts Progression of Diabetic Nephropathy in Nos3-Deficient Mice

Diabetic nephropathy is the most common single cause of end-stage renal failure in many countries. Current therapies of controlling blood pressure and blood glucose levels have only a limited benefit in slowing progression to end-stage disease (1), indicating a major gap in our treatment options. The diabetic kidney is stressed by multiple factors including hyperglycemia, reactive oxygen species, advanced glycation end products, proinflammatory cytokines, and angiotensin II—all of which can induce signaling via the p38 mitogen-activated protein kinase (MAPK) (2–5). Increased activation of p38 MAPK in glomeruli and the tubulointerstitial compartment has been described in animal models of diabetic nephropathy as well as in patients with diabetic nephropathy (6–9). Inhibition of p38 MAPK activation via pharmacologic and genetic approaches can suppress the induction of albuminuria, glomerular matrix expansion, and inflammation in models of type 1 and type 2 diabetic nephropathy (9,10), although intervention studies in established disease are lacking. Importantly, clinical trials of p38 inhibitor compounds in rheumatoid arthritis have failed to deliver the promise of animal studies owing to toxicity issues, which have led to investigation of the upstream kinases involved in context-dependent p38 MAPK activation (11).

Apoptosis signal-regulating kinase 1 (ASK1/MAP3K5) is a member of the large family of MAPK kinase kinase enzymes, many of which have the potential to activate the downstream p38 MAPK via phosphorylation of the MKK3 and MKK6 enzymes (12,13). p38 MAPK and, to a lesser extent, c-Jun terminal kinase are the only known downstream targets of ASK1 signaling (14). ASK1 is activated in response to oxidative stress; specifically, ASK1 exists as an inactive dimer coupled to thioredoxin and undergoes autoactivation after the oxidation and dissociation of thioredoxin (13). ASK1 is widely expressed in diverse tissues and is most highly expressed in the kidney (15). Mice lacking the Ask1 gene (Ask1^−/−) have a normal phenotype, including normal kidney structure and function, which contrasts with the fetal lethality of mice lacking the p38α/Mapk14 gene (16,17). In addition, Ask1^−/− mice show a dramatic inhibition of p38 MAPK activation and significant protection from kidney injury in models of acute tubular necrosis and renal interstitial fibrosis, providing results consistent with administration of p38 inhibitors (4,18–20). Therefore, inhibition of ASK1/p38 signaling may have therapeutic potential for diabetic glomerulosclerosis.

This aim of this study was to investigate the therapeutic potential of a highly selective small molecule ASK1 inhibitor (GS-444217) in experimental diabetic nephropathy. We investigated streptozotocin (STZ)-induced type 1 diabetes in hypertensive mice lacking the gene for endothelial nitric oxide synthase (Nos3), as this model exhibits robust development of diabetic glomerulosclerosis (21).

GS-444217 was synthesized by Gilead Sciences in Foster City, CA (manuscript in preparation). In a competitive, time-resolved fluorescence resonance energy-transfer immunoassay, GS-444217 directly inhibited ASK1 kinase activity in vitro with an IC₅₀ of 2.9 ± 0.8 nmol/L. In a kinase selectivity panel that included 451 kinases (KINOMEscan; DiscoverRX Corporation, Fremont, CA), GS-444217 exhibited >50-fold greater affinity for ASK1 compared with all other kinases measured. For delivery in mice, GS-444217 was incorporated into standard mouse chow at 0.1% (w/w). The concentration of GS-444217 in mouse plasma was measured at the end of each study and found to be 20.2 ± 8.2 μmol/L for the early-intervention study and 21.3 ± 7.4 μmol/L for the late-intervention study. These exposures of GS-444217 in mice are expected to result in complete and selective inhibition of ASK1 (data not shown).

Nos3^−/− mice on the C57BL/6 background were obtained from The Jackson Laboratory (Bar Harbor, ME) and bred under pathogen-free conditions at the Monash Medical Centre Animal Facility (Clayton, Australia). At 8 weeks of age, male Nos3^−/− mice were given injections of STZ (5 × 55 mg/kg/day i.p.; Sigma, St Louis, MO). Groups of mice (n = 10) with diabetes (fasting blood glucose >16 mmol/L at 2 weeks after the last STZ injection) were followed for 8 or 15 weeks. In the early-intervention study, diabetic mice were fed either normal chow or chow containing 0.1% GS-444217 between weeks 2 and 8 and then killed. In the late-intervention study, diabetic mice were fed either normal chow or chow with 0.1% GS-444217 over weeks 8–15 and then killed. Age-matched nondiabetic Nos3^−/− mice (n = 10) were used as controls. In an additional study, a group of nondiabetic Nos3^−/− mice (n = 9) was assessed for systolic blood pressure before and after a 6-week period on chow containing 0.1% GS-444217. GS-444217 treatment was well tolerated by the mice in all studies, and no drug-related clinical observations were noted in any of the studies. Blood glucose (measured by a tail vein sample) and body weight were measured weekly after a 3-h fast (0800–1100 h), and mice with fasting glucose levels >30 mmol/L were given 0.5 units s.c. protophane insulin (Novo Nordisk, Sydney, Australia) three times a week to maintain body weight. Urine was collected pre-experiment and at weeks 5, 8, 12, and 15 after STZ injection to assess urine albumin excretion. Glycated hemoglobin (HbA_1c) was measured from blood samples taken at weeks 8 and 15. Tissues were collected at week 8 or 15 and were fixed in 4% (v/v) formaldehyde or 2% (w/v) paraformaldehyde-lysine-periodate or snap-frozen and stored at −80°C. These animal studies were approved by the Monash Medical Centre Animal Ethics Committee in accordance with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes, 7th edition (2004).

Fasting blood glucose was measured from tail blood by glucometer (Medisense, Abbott Laboratories, Bedford, MA). Urine was collected from mice housed in metabolic cages for 8 h. Heparinized whole blood was collected from tail veins for analysis of HbA_1c. At the end of experimentation, whole blood was collected by cardiac puncture in anesthetized animals and stored as serum or heparinized plasma. Urine creatinine levels were determined by the Jaffe rate reaction method. ELISA kits were used to assess levels of urine albumin (Bethyl Laboratories, Montgomery, TX), plasma insulin (Mercodia, Uppsala, Sweden), and serum cystatin-C (Enzo Life Sciences, Farmingdale, NY). HbA_1c was measured by DCA Vantage Analyzer (Siemens, Camberley, U.K.).

Systolic blood pressure was measured in conscious mice by tail-cuff plethysmography (IITC Life Science, Woodland Hills, CA). Mice were trained twice weekly for 3 weeks prior to experimental readings. At each recording, mice were acclimatized to a preheated chamber (29°C) for 15 min and the pressure readings were recorded over three consecutive manual inflation-deflation cycles to obtain an average.

The following primary antibodies were used in this study: mouse anti–phospho-p38α (Upstate Biotechnology, Lake Placid, NY), rat anti-CD68 (FA-11; Serotec, Oxford, U.K.), goat anti–collagen IV (Southern Biotechnology, Birmingham, AL), and rabbit anti–Wilms tumor 1 (WT1) antigen (Santa Cruz Biotechnology, Santa Cruz, CA).

Formalin-fixed sections (2 μm) were stained with periodic acid-Schiff (PAS) to assess structure and counterstained with hematoxylin to identify nuclei. Immunostaining for phospho-p38, WT1, and collagen IV was performed on 4 μm paraffin-embedded sections, which were fixed in 4% formalin or methylcarn solution. Immunostaining for CD68 was performed on 5 μm paraformaldehyde-lysine-periodate–fixed cryostat sections. For antigen retrieval of phospho-p38, dewaxed paraffin sections were heated in a microwave oven (800 watts for 12 min) in 10 mmol/L sodium citrate buffer (pH 6.0) (22). For immunostaining, sections were treated with 20% rabbit serum or 20% goat serum for 30 min and then incubated with primary antibody in 3% BSA overnight at 4°C. Sections were then placed in 0.6% hydrogen peroxide in methanol for 20 min to inactivate endogenous peroxidase. Bound primary antibodies were detected using a standard ABC-peroxidase system: avidin-biotin block, biotinylated antibodies (rabbit anti-goat IgG, rabbit anti-rat IgG, or goat anti-mouse IgG), and ABC-peroxidase (Vector Laboratories, Burlingame, CA). Sections were developed with 3,3-diaminobenzidine (Sigma) to produce a brown color. CD68 sections were counterstained with hematoxylin to assist cell counting. Normal goat serum or isotype-matched irrelevant IgG was used as negative control.

The number of CD68⁺ cells and WT1⁺ podocytes was counted in 30 hilar glomerular cross-sections (gcs) per animal (×400). Glomerular collagen IV staining was quantitated by computer image analysis (Image-Pro Plus, Media Cybernetics, Silver Spring, MD) in 50 hilar glomerular cross-sections (×400) and expressed as the percentage of glomerular area stained. All scoring was performed on blinded slides.

Total RNA was extracted from whole kidney using Trizol (Invitrogen) and reverse transcribed with random primers using the Superscript First-Strand Synthesis kit (Invitrogen). Real-time PCR analysis was performed as previously described (5,23). The Taqman probe and primer sequences are listed in Supplementary Table 1.

Statistical differences were analyzed by one-way ANOVA with Tukey multiple comparison posttest. Data were recorded as mean ± SEM with P < 0.05 considered significant. All analyses were performed using GraphPad Prism 6.0 (GraphPad Software, San Diego, CA).

Consistent with previous studies (6,9,24), basal p38 phosphorylation (activation) was detected in nondiabetic kidney by Western blot, while immunostaining identified phosphorylated p38 in glomeruli, tubules, and the interstitium (Fig. 1A and E). A significant increase in p38 activation in the kidney was evident at both weeks 8 and 15 (Fig. 1). In contrast, treatment with GS-444217 over weeks 2–8 and over weeks 8–15 of diabetes caused a dramatic inhibition of p38 activation in all cell types (Fig. 1).

Since diabetes does not affect hypertension in Nos3^−/− mice (25,26), we measured the effect of GS-444217 on blood pressure in nondiabetic Nos3^−/− mice. Systolic blood pressure in Nos3^−/− mice was 133 ± 5 mmHg before treatment and was unchanged (134 ± 5 mmHg) after 6 weeks of GS-444217 administration, demonstrating that GS-444217 does not affect hypertension in these mice.

Groups of diabetic Nos3^−/− mice were fed standard chow or chow with 0.1% GS-444217 from week 2 after STZ injection until being killed on week 8. GS-444217 had no effect upon fasting body weight, fasting blood glucose levels, or plasma insulin levels, although a partial reduction in HbA_1c levels was evident in the drug-treated group (Fig. 2). A substantial increase in urinary albumin excretion was evident on weeks 5 and 8 after STZ injection, which was not affected by GS-444217 treatment (Fig. 3A). There was also a reduction in the number of glomerular WT1⁺ podocytes at week 8 after STZ, which was unaffected by drug treatment (Fig. 3B). However, the decline in renal function at week 8 after STZ, as assessed by increased serum cystatin C levels, was significantly improved by drug treatment (Fig. 3C).

Nos3^−/− mice with no treatment developed mild-to-moderate glomerulosclerosis by week 8 of diabetes, as shown by increased PAS-stained glomerular deposits and an increase in glomerular staining for collagen IV (Fig. 4). In addition, a significant glomerular infiltrate of CD68⁺ macrophages was evident (Fig. 4). GS-444217 significantly reduced both glomerulosclerosis and glomerular macrophage infiltration (Fig. 4).

Groups of diabetic Nos3^−/− mice were fed standard chow or chow with 0.1% GS-444217 from week 8 after STZ injection until being killed on week 15. In animals fed the standard diet, disease severity was maintained or increased between weeks 8 and 15 (Figs. 5 and 6). GS-444217 had no effect upon fasting body weight, fasting blood glucose levels, or HbA_1c levels (Fig. 5). The diabetic mice had established albuminuria and raised serum cystatin-C levels at week 8, when GS-444217 treatment commenced (Fig. 6). Drug treatment had no impact upon established albuminuria or podocyte loss, but there was a significant reduction in the levels of serum cystatin-C, suggesting that GS-444217 treatment improved renal function (Fig. 6).

There was a clear increase in diabetic glomerulosclerosis between weeks 8 and 15 in untreated diabetic Nos3^−/− mice as shown by glomerular collagen IV staining (Fig. 7A–C and E). This was associated with a sustained increase in renal mRNA levels for profibrotic molecules (collagens 1 and 4, fibronectin, transforming growth factor-β1, and plasminogen activator inhibitor-I) (Fig. 7F–J) and a sustained glomerular CD68⁺ macrophage infiltrate and increases in renal mRNA levels for proinflammatory molecules (chemokine CC motif ligand 2 [CCL2] and tumor necrosis factor-α) (Fig. 8A–F). In addition, expression of the tubular damage marker kidney injury molecule-1 (KIM-1) was elevated at week 8 of diabetes and remained elevated through to week 15 (Fig. 8G).

GS-444217 treatment over weeks 8–15 prevented the increase in glomerular deposition of collagen IV and downregulated renal mRNA levels for all of the profibrotic molecules analyzed (Fig. 7D–J). In addition, GS-444217 treatment significantly reduced the glomerular macrophage infiltrate and reduced the renal mRNA expression of proinflammatory markers and tubular damage marker KIM-1 (Fig. 8).

Administration of an ASK1 inhibitor, GS-444217, was shown to be effective in a mouse model of established diabetic nephropathy on the basis of halting progressive glomerulosclerosis, reducing inflammation, and improving renal function despite ongoing hyperglycemia, hypertension, and albuminuria.

The protective effect of ASK1 inhibitor treatment in diabetic nephropathy was attributed to the highly effective blockade of p38 activation—the major downstream target for ASK1 (14). GS-444217 is highly selective for ASK1, although as in all pharmacology studies, we cannot formally rule out unanticipated off-target effects. Increased p38 activation has been described in human and experimental diabetic nephropathy (6–9), which was confirmed in the current study. Administration of a p38 MAPK inhibitor has been shown to reduce albuminuria and prevent the development of mild glomerular fibrosis in nonhypertensive diabetic rats (10). In addition, deletion of the Mkk3 gene in db/db mice with type 2 diabetes significantly reduced p38 activation, albuminuria, podocyte loss, and glomerular collagen deposition and improved renal function (9). The current study significantly extends these findings in two ways. First, diabetic Nos3^−/− mice develop a much more severe form of glomerulosclerosis compared with the above two models (21), thereby providing a much sterner test of the ability of ASK1/p38 blockade to prevent diabetic glomerulosclerosis. Second, we performed an intervention study and demonstrated that GS-444217 could inhibit p38 activation, halt glomerulosclerosis, reduce glomerular inflammation, and improve renal function in established diabetic nephropathy. Indeed, the results of the current study are an important component of the preclinical data that support a current phase 2 trial evaluating an ASK1 inhibitor, GS-4997, in patients with stage 3/4 diabetic kidney disease (27).

The ability of GS-444217 to reduce collagen deposition in the diabetic glomerulus is consistent with previous studies in which p38 inhibitors have been shown to reduce renal fibrosis in nondiabetic models of kidney disease (19,28,29). In addition, Ask1^−/− mice show a profound suppression of p38 activation and tissue fibrosis in models of both renal and cardiac disease (4,30,31).

A second mechanism whereby ASK1 inhibition reduced diabetic glomerular damage is through the inhibition of macrophage infiltration and inhibition of mRNA levels of proinflammatory mediators. Previous studies have shown that macrophages cause glomerular injury in experiment models of diabetic nephropathy (32,33). This is consistent with the reduction in CCL2 mRNA levels and macrophage infiltration seen in the tubulointerstitium in the obstructed kidney in Ask1^−/− mice (4).

The aggressive nature of disease development in diabetic Nos3^−/− mice is attributed to the presence of hypertension. However, the ASK1 inhibitor protected against diabetic glomerulosclerosis despite having no effect on hypertension in Nos3^−/− mice. This is consistent with previous studies on Ask1^−/− mice, which are protected from end organ damage in the setting of hypertension. For example, in models of angiotensin II and mineralocorticoid-driven hypertension, Ask1^−/− mice had reduced cardiac fibrosis despite unaltered hypertension (30,31). In the current study, ASK1 inhibition with GS-444217 appears to reduce renal injury and improve renal function by directly decreasing inflammation and fibrosis in the diabetic kidney.

The loss of podocytes and development of albuminuria in diabetic Nos3^−/− mice were unaffected by ASK1 inhibitor treatment in our early- and late-intervention studies. This suggests that activation of ASK1/p38 signaling is not involved in the mechanism of diabetic albuminuria. However, there are some limitations in this interpretation. Recent studies of Nos3^−/− mice has shown that they are highly susceptible to glucose-induced podocyte damage and show marked albuminuria 2 weeks after STZ administration (34), which was the time at which we began ASK1 inhibitor treatment in the early-intervention study. Therefore, it is possible that ASK1 inhibition may have been commenced too late to have an effect on the induction of podocyte damage/loss and albuminuria. Previous studies in both diabetic and nondiabetic kidney disease have shown that p38 inhibitors can reduce albuminuria (9,10,28,35), arguing that ASK1/p38 signaling may contribute to albuminuria in some contexts.

Early intervention with the ASK1 inhibitor saw a reduction in HbA_1c levels in diabetic Nos3^−/− mice despite no detectable difference in fasting blood glucose or plasma insulin levels. The reason for this discrepancy is unclear. However, no change in HbA_1c levels or other parameters of diabetes was seen in the late-intervention study, demonstrating that halting established diabetic glomerulosclerosis with GS-444217 was not due to modification of the diabetic state.

While there are several members of the MAP3K family that can induce p38 activation, we have shown that ASK1 plays a nonredundant and essential role in activating p38 in the diabetic kidney. This is consistent with a role for ASK1 in high glucose–induced p38 activation in cultured glomerular mesangial cells (36) and the various stresses present in diabetes (e.g., oxidative stress, endoplasmic reticular stress, hyperglycemia, and angiotensin II) that are known to activate ASK1/p38 signaling (37). However, activation of p38 MAPK signaling can operate independent of ASK1 in kidney cells as shown by the unaltered lipopolysaccharide and interleukin-1–induced p38 activation and biological responses seen in Ask1^−/− tubular epithelial cells (4). Thus, ASK1 is a context-dependent regulator of p38 activation, which is particularly important in settings of oxidative stress owing to the regulation of ASK1 activation by antioxidant proteins such as thioredoxin. One limitation of the current study is that currently available commercial antibodies are not sensitive enough to detect ASK1 in tissue sections or for detection of ASK1 phosphorylation in kidney tissues.

ASK1/p38 signaling has been implicated in other aspects of diabetes. ASK1/p38 signaling is involved in stress-induced death of pancreatic β-cells and in the induction of endothelial cell senescence induced by high glucose (38,39). Ask1^−/− mice have reduced levels of maternal diabetes-induced endoplasmic reticulum stress in the developing embryo (40). Indirect blockade of ASK1 signaling can protect against diabetic cardiomyopathy (41,42), although transgenic mice overexpressing a kinase dead form of ASK1 are not protected from diabetic neuropathy (43). Therefore, pathological signaling by ASK1 may be a common contributor to diabetic end organ damage in the pancreas, heart, and kidney.

In conclusion, treatment with an ASK1 inhibitor halted the progression of glomerulosclerosis and renal inflammation and improved renal function in an aggressive model of diabetic nephropathy. This protective effect was attributed to blockade of ASK1-dependent p38 signaling. This study identifies ASK1 as a new therapeutic target in diabetic nephropathy, and a related ASK1 inhibitor is now being tested in a phase 2 clinical trial (NCT02177786).

Funding. This study was partly funded by the National Health and Medical Research Council of Australia (APP1044289).

Duality of Interest. This study was also funded by Gilead Sciences. G.H.T. was partly supported by Gilead Sciences. J.T.L. and D.G.B. are employees of Gilead Sciences. D.J.N.-P. acts as a consultant for Gilead Sciences. Gilead Sciences is currently running a clinical trial with an ASK1 inhibitor in patients with diabetic kidney disease (NCT02177786). No other potential conflicts of interest relevant to this article were reported.

Author Contributions. G.H.T., F.Y.M., and Y.H. performed the study. G.H.T., F.Y.M., Y.H., and D.J.N.-P. analyzed data. G.H.T. and D.J.N.-P. designed the study. J.T.L. and D.G.B. provided the ASK1 inhibitor and pharmacokinetic data. G.H.T. and D.J.N.-P. wrote the manuscript, which was approved by all authors. G.H.T. 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.