If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Laboratory of Molecular Life Science, Institute of Biomedical Research and Innovation (IBRI), Kobe, JapanClinical Research Department, Institute of Biomedical Research and Innovation (IBRI), Kobe, Japan
Diabetic nephropathy (DN), once manifested, is unlikely to completely recover. Factors that influence DN progression were explored by investigating the process of glomerulosclerosis and interstitial fibrosis and chronological changes in glucose, albuminuria, hyperfiltration, and expressions of sodium-glucose cotransporter 2 (SGLT2) and hypoxia-inducible factors (HIFs) up to 50 weeks in inducible cAMP early repressor transgenic mice, a model of severe DN. Long-term intervention with the SGLT2 inhibitor canagliflozin or islet transplantation or heminephrectomy was used. Inducible cAMP early repressor transgenic mice exhibited progressive diabetic glomerulosclerosis and mild interstitial fibrosis, and expressed extensive HIF-1α and HIF-2α in glomerulus and tubules, with sustained hyperfiltration up to 50 weeks. Canagliflozin ameliorated glomerulosclerosis/interstitial fibrosis gradually and reduced HIF overexpression. Islet-transplanted mice exhibited no amelioration. None of the heminephrectomized diabetic mice survived the hyperfiltration overload, but all of the canagliflozin-treated mice survived with re-expressions of HIF-1α and HIF-2α. These results suggest that persistent glomerular hyperfiltration might initiate glomerular injury, and persistent overexpression of HIFs could promote the development of glomerulosclerosis and interstitial fibrosis. Canagliflozin attenuated both changes. Oxidative stress or hypoxia was undetectable in this model. The abnormal expression of HIF-1α and HIF-2α may be a potential therapeutic target for preventing glomerulosclerosis and interstitial fibrosis.
The number of patients with diabetes and diabetic nephropathy (DN) is increasing worldwide.
According to the International Diabetes Federation, as many as one in 10 people are living with diabetes. Once DN is overt, complete recovery is unlikely. DN is the primary cause of end-stage renal disease requiring renal replacement therapy, but the underlying mechanisms involved in the development of DN in severe diabetes remain to be fully elucidated.
Sodium-glucose cotransporter 2 (SGLT2) inhibitors are a new class of glucose-normalization drugs that inhibit urinary glucose and sodium reabsorption by blocking SGLT2 in renal proximal tubules. Canagliflozin potently inhibits SGLT2 and also partially inhibits sodium-glucose cotransporter 1, in contrast to other SGLT2 inhibitors.
It dose dependently blocked up to 94% of glucose reabsorption in a diabetic rat model. Recently, the global CANVAS (Canagliflozin Cardiovascular Assessment Study) and CREDENCE (Canagliflozin and Renal Events in Diabetes with Established Nephropathy Clinical Evaluation) clinical trials have shown that treatment with canagliflozin reduced the risk of renal failure in patients with type 2 diabetes and DN compared with placebo during a median 2.62-year follow-up period.
SGLT2 inhibitors have a renoprotective effect, independently of their glucose-normalizing effect, presumably by activating tubuloglomerular feedback to increase the delivery of sodium and chloride to the macula densa and reduce glomerular hyperfiltration.
However, the underlying mechanisms involved in the effects of SGLT2 inhibitors remain unknown. Because administration of SGLT2 inhibitors increases β-cell glucose sensitivity in the pancreas of patients with type 2 diabetes,
Male mice display many of the features of progressive diabetic renal pathology arising from sustained hyperglycemia. However, other diabetic mouse models, including the most commonly used STZ model, are often associated with drug effects such as tubular damage, and the results can vary widely depending on the method of administration. Furthermore, db/db mice is an obese background model of type 2 diabetes in which obesity develops from 3 to 4 weeks of age, and these mice have a short life expectancy of only about 10 months. Therefore, we believe that the ICER-Tg mouse is the best model of human diabetes for the present purpose because of the ability to examine the long-term effects of hyperglycemia alone.
Hypoxia-inducible factors (HIFs) such as HIF-1α and HIF-2α protect against hypoxic injury when their expression is transiently regulated.
but in marked contrast, few studies have examined HIF-1α expression in DN, with no studies on HIF-2α thus far. Furthermore, the effect of long-term HIF expression in glomeruli and tubules on the development of DN (glomerulosclerosis and interstitial fibrosis) has also not been elucidated. However, the effects of HIF blockade have been reported; 4-week HIF-1 inhibitor treatment of type 1 diabetic OVE26 mice,
inhibit the development of tubulointerstitial fibrosis and mesangial expansion. Sustained nuclear hyperexpression of HIF-1α in tubular and mesangial cells for a long period under hyperglycemic conditions might cause fibrosis and sclerosis. Thus, it seems plausible that dysregulated production of HIF-1α in the kidneys of hyperglycemic diabetes patients, leading to sustained hyperexpression, might have a fundamentally different effect from the protective action against acute ischemic kidney injury, and may be harmful and possibly fibrogenic.
We hypothesized that the development of DN is due to hyperfiltration and HIF overexpression, as well as a persistent hyperglycemia. Therefore, the aim of the present work was to elucidate the factors influencing the progression of DN and the involvement of SGLT2 and HIFs in this extended process by means of a long-term (50-week) administration study in a mouse model of severe DN. The study especially focused on the process of glomerulosclerosis and interstitial fibrosis, as well as the chronological changes of glucose, hyperfiltration, and HIF overexpression. Their relations were assessed over 50 weeks in light of the effect of long-term intervention with SGLT2 inhibitors canagliflozin or islet transplantation (glucose normalization) or heminephrectomy (additional hyperfiltration load). The results suggest that persistent glomerular hyperfiltration might be an initial driving force of glomerular injury, and the persistent abnormal overexpression of HIFs could promote the development of glomerulosclerosis and interstitial fibrosis, which do not appear to be associated with detectable levels of oxidative stress/hypoxia.
Materials and Methods
ICER-Tg mice (C57BL/6 background) were generated as described previously.
Mice were housed in the RIKEN (Kobe, Hyogo, Japan) and Kyushu University Animal Facilities (Fukuoka, Japan) on a 12-hour light–dark cycle, with water and a standard rodent diet (CE-2; CLEA Japan, Tokyo, Japan) ad libitum. The standard rodent diet CE-2 contains 8.9% water, 24.9% crude protein, 4.6% crude fat, 4.1% crude fiber, 6.6% crude ash, and 51% nitrogen-free extract; it includes 344.9 kcal per 100 g, which is suitable to maintain normal blood glucose levels for long-term research.
Body weight and morning-fed blood glucose levels were measured every other week. Blood glucose levels and glycosylated hemoglobin values were measured in blood from a tail snip using a OneTouch UltraVue device (Johnson & Johnson K.K., Tokyo, Japan) and a DCA 2000 analyzer (Siemens Healthineers, Erlangen, Germany), respectively. Animal care and experimental procedures were reviewed and approved by the Animal Care and Experimentation Committees of the Institute of Biomedical Research and Innovation (Kobe, Hyogo, Japan), RIKEN, and Kyushu University.
Male ICER-Tg mice with severe diabetes were divided into four groups: untreated diabetic control (Tg-control), SGLT2 inhibitor treatment (Tg-SGLT2i), SGLT2 inhibitor treatment plus heminephrectomy (Tg-SGLT2i + nephrectomy), and islet transplantation (Tg-IT) (Figure 1). Long-term intervention was initiated after the onset of DN at 13 to 14 weeks of age, when the animals exhibited persistent hyperglycemia and albuminuria. The mice were euthanized at 50 weeks of age.
To determine the effect of the SGLT2 inhibitor canagliflozin on the development of diabetic renal injury, Tg-control mice (n = 6), Tg-SGLT2i mice at 33 weeks old (n = 3) and 50 weeks old (n = 5), and wild-type (WT) mice (n = 12) were analyzed. For SGLT2 inhibitor treatment (Tg-SGLT2i mice), Tg mice were fed a standard diet (CE-2) containing 0.02% (w/w%) canagliflozin (40 mg/kg per day). This dose was well tolerated in Zucker diabetic fatty rats and significantly decreased blood glucose levels.
Canagliflozin was provided by Tanabe Mitsubishi Pharma Corporation (Osaka, Japan). Other Tg-control mice and littermate WT mice were fed CE-2 throughout the study.
Pancreatic Islet Transplantation
To evaluate the effect of glucose normalization alone, Tg mice were given islet cell transplantation at 13 weeks (Tg-IT mice, n = 8) and compared with Tg-SGLT2i mice (n = 5). For islet transplantation (Tg-IT mice), 500 handpicked islets from 12- to 14-week–old C57BL/6 WT mice (n = 10 per experiment) were transplanted under the kidney capsule of age-matched Tg mice. The experiments with the four groups of mice were conducted simultaneously.
To assess the effect of canagliflozin under conditions of increased hyperfiltration overload, heminephrectomy was applied to Tg-control (n = 3), Tg-SGLT2i (n = 3), and WT (n = 3) mice at 33 weeks. The right kidney was removed. All Tg-control mice died within 1 week after heminephrectomy. Mice were examined at 50 weeks (Supplemental Figure S1).
Measurement of Serum and Urinary Variables
Urine collection and measurement of urine volume, urinary parameters, and food and water consumption were performed at 11, 13 to 15, 19 to 20, 24, and 48 to 50 weeks of age by using 24-hour samples from mice housed in individual cages (n = 5 to 12 per group). During sample collection, mice were allowed free access to food and water. Measurements of urine samples were repeated three times for each group at the indicated times and at 48 to 50 weeks. It took several days to collect urine due to the limited number of urinary collection cages. Collection of small amounts of blood from the tail was performed at 11, 13 to 15, and 19 to 20 weeks of age, and a large amount of blood was collected from the heart with a heparin-coated syringe immediately before isolation of the organs, after mice had been anesthetized with a combination of intraperitoneal anesthesia (sodium pentobarbital, 50 to 60 mg/kg intraperitoneally) and inhalation anesthesia (isoflurane). For glucagon measurement, blood was collected in a tube containing aprotinin (final concentration, 500 kIU/mL; Wako, Osaka, Japan) and EDTA-2Na (final concentration, 0.75 mg/mL). The serum parameters, including albumin, creatinine (Cre), total cholesterol, blood urea nitrogen, triglyceride, amylase, aspartate aminotransferase, and alanine aminotransferase, were measured by using a FUJI DRI-CHEM NX700 analyzer (Fujifilm, Tokyo, Japan). Plasma insulin and glucagon were assayed with ELISA kits (Morinaga Institute of Biological Science Inc., Yokohama, Japan; Cosmic Corporation Co., Ltd. Tokyo, Japan); the intra-assay and interassay CVs were 10% and 15%, respectively, for both kits. Urinary Cre levels were measured with a CicaLiquid-N CRE kit (Kanto Chemical Co., Inc. Tokyo, Japan) using an autoanalyzer (Hitachi Ltd., Tokyo, Japan), which showed a high correlation with the values determined by high-performance liquid chromatography (r = 0.999); the intra-assay and interassay CVs were <5% and 0.2% to 0.3%, respectively. Body weight–adjusted Cre clearance (Ccr) was calculated by use of the following equation: Ccr = urine Cre (mg/dL) × urine volume (μL/minute)/serum Cre (mg/dL)/body weight (g). Urinary albumin was assayed with an ELISA kit (Fujifilm Wako Shibayagi Corporation, Gunma, Japan). The intra-assay and the interassay CVs were both <10%.
Isolation of Pancreatic Islets and Islet Transplantation
Pancreatic islets of 12- to 14-week–old male C57BL/6 WT mice (n = 10 per experiment) were isolated via four steps as follows: pancreatic perfusion (inserting a cannula into the common bile duct and perfusing the pancreas with cold Hanks’ balanced salt solution containing collagenase (Sigma Aldrich, St. Louis, MO), digestion of the pancreas, separation on Histopaque-1077 and 1119 (Sigma Aldrich) gradients, and washing. Then, 500 high-quality islets were handpicked by using a Stemi 2000 stereomicroscope (Carl Zeiss, Oberkochen, Germany) and immediately transplanted under the kidney capsule of male hyperglycemic Tg-control mice under a stereomicroscope (Olympus, Tokyo, Japan). The transplantation status of each mouse was confirmed at the time of transplantation. Four independent isolation and transplantation experiments were performed.
Kidney halves were fixed in cold 10% buffered formalin and embedded in paraffin, and 1- or 2-μm-thick sections were stained with periodic acid-methenamine silver (PASM), hematoxylin and eosin, or periodic acid-Schiff using standard histologic procedures. All staining procedures were performed at the same time. Multiple hematoxylin and eosin images (16 to 36 images) were captured and merged to form a single image of the whole kidney by using Image-Merging software on a BZ-9000 microscope (KEYENCE Corporation, Osaka, Japan). Glomeruli were counted per mid-transverse section of the kidney for each mouse by using the Dynamic cell-count BZ-HIC software package (KEYENCE Corporation).
Measurements of Glomerular Surface Area, Type 4 Collagen–Positive Area, and PASM-Positive Area of the Glomerulus or Interstitium
The severity of mesangial matrix expansion (glomerular sclerosis) or interstitial fibrosis or glomerular type 4 collagen (ColIV) expression was quantified by measuring the mesangial PASM-positive area in the glomeruli or interstitium or glomerular ColIV-positive area as previously reported.
Briefly, >50 glomeruli present in each section were photographed for each mouse. Noise was removed by using the eliminate-particle technique, and the area of interest was measured within the specified range using the Dynamic Cell-Count BZ-HIC software package. The mesangial matrix area fraction was determined as the percentage of the mesangial PASM-positive area to the glomerular surface area. In the interstitium, PASM-positive areas between the tubules excluding tubular basement membrane were taken as interstitial fibrosis and analyzed. The artery and vein wall and surrounding fibers were not quantified in this experiment; these areas could be easily excluded. To avoid bias, all images were blinded before quantification.
Immunohistochemistry and Antibodies
Paraffin-embedded sections of mouse pancreas and kidney halves (2 to 4 μm thick) were immunostained as described previously.
Antigen retrieval was performed in 0.01 mol/L citrate buffer (pH.6.0) using a pressure cooker or microwave oven. The endogenous peroxidase activity was blocked by 0.3% H2O2 in phosphate-buffered saline. Nonspecific staining caused by endogenous biotin was blocked via incubation with unconjugated avidin using a biotin-blocking system (Dako North America, Inc., Carpinteria, CA; X0590). Mouse on Mouse (M.O.M.) blocking reagent (Vector Laboratories, Burlingame, CA; MKB-2213-1) was used with mouse primary antibody to block endogenous mouse immunoglobulins in mouse tissue sections. The following primary antibodies were used: guinea pig anti-insulin (Dako; N1542); rabbit anti-glucagon (Linco Research, St. Charles, MO; 4030-01F); rabbit anti-SGLT2 (Abcam, Cambridge, UK; ab85626); rat anti-ColIV (Shigei Medical Research Institute, Okayama, Japan; clone-H22), mouse anti–8-hydroxy-2′deoxyguanosine (8-OHdG; Japan Institute for the Control of Aging, Shizuoka, Japan; clone-N45.1); mouse anti–HIF-1α (Novus Biologicals, Littleton, CO; NB100-105; α67); and rabbit anti–HIF-2α (Abcam; Ab109616, knockout verified). Primary antibodies were detected with the following secondary antibodies: Alexa Fluor 488– or Alexa Fluor 594–conjugated IgG, anti–guinea pig (Invitrogen, Carlsbad, CA; A11073) or anti-rabbit (A11037); biotinylated IgG, anti-rabbit (Jackson ImmunoResearch Laboratories Inc., West Grove, PA; 111-065-144) or anti-rat (Vector Laboratories; BA9400) or anti-mouse (BA9200); and Alexa Fluor 488 or 594 streptavidin-conjugates (Invitrogen; S11223, S11227). For detection of HIF isoforms, a fluorescein-tyramide signal amplification system (PerkinElmer, Waltham, MA; NEL750001EA) was used; this provides high fluorescence signal amplification and improves sensitivity by up to 100-fold.
In this method, no significant staining was observed in the absence of the primary or the secondary antibody, whereas distinct staining patterns were observed with antibodies, confirming the specificity of the reaction. All slides for immunochemical comparisons among groups were stained simultaneously to minimize variability. The quantification was not based on the intensity of fluorescence but on the presence/absence and extent of localization. DAPI was used to identify nuclei. Images were captured in the confocal mode on an LSM 710 META confocal microscope (Carl Zeiss) or on a BZ9000 microscope (KEYENCE Corporation).
The results are expressed as means ± SEM. Statistical significance was assessed by one-way analysis of variance with a post hoc Tukey-Kramer test for multiple comparison analysis. P < 0.05 was considered to be statistically significant. Statistical analyses were performed by using BellCurve for Excel software version 3.22 (Social Survey Research Information Co., Ltd., Tokyo, Japan).
The overall experimental protocol is summarized in Figure 1. Severely diabetic male ICER-Tg mice were divided into four groups: untreated diabetic control (Tg-ctrl), SGLT2 inhibitor-treated (Tg-SGLT2i), SGLT2i-treated plus heminephrectomized (Tg-SGLT2i + nephrectomy), and islet-transplanted (Tg-IT). Mice were treated with SGLT2i (canagliflozin) or islet transplantation from 13 to 14 weeks of age, when DN appeared. At the time of starting SGLT2i treatment at 13 weeks after onset, no pathologic glomerular or tubular damage or inflammatory changes such as cellular infiltration were seen. Heminephrectomy was applied at 33 weeks. Littermate normoglycemic WT-control mice were also analyzed as controls. All mice were euthanized at 50 weeks of age.
The effect of canagliflozin on the development of diabetic renal injury was evaluated by comparing Tg-control mice, 33-week–old Tg-SGLT2i mice, and 50-week–old Tg-SGLT2i mice. The effect of glucose normalization alone was analyzed by comparing Tg-IT and Tg-SGLT2i mice. The effect of canagliflozin in mice subjected to additional hyperfiltration overload by heminephrectomy was evaluated by comparing Tg-SGLT2i + nephrectomy mice and Tg-SGLT2i mice.
Effect of Long-Term Canagliflozin Intervention
Whether long-term administration of canagliflozin would achieve strict glycemic control, alter the histology of pancreatic islets, and block the development of renal injury in severely diabetic male ICER-Tg mice was examined first (Figure 2A).
In male Tg-control mice, blood glucose levels were severely elevated by 8 weeks of age, and they remained high throughout life because of reduced insulin secretion caused by severe depletion of pancreatic β cells (Figure 2B). In Tg-SGLT2i mice, blood glucose declined rapidly within 1 week, followed by a gradual further decline to the WT level: blood glucose levels remained stable at the normal level for the remainder of the study.
Tg-SGLT2i mice exhibited only a slight gain in weight after 6 weeks, but subsequent growth was similar to that of Tg-control mice until 48 weeks of age (Figure 2C). Food intake was similar in Tg-control and Tg-SGLT2i mice (Figure 2C).
Histology of Pancreatic Islets
The typical islet morphology exhibits a core of insulin-positive cells with a mantle of glucagon-positive cells, as was seen in WT-control mice (Figure 3A). In Tg-control mice, however, islets were severely disorganized with markedly reduced β cells and an increased proportion of α cells. Tg-SGLT2i mice showed a similar pattern, except that a slight increase in β cells was observed in some islets. Consistent with this finding, the plasma insulin level was significantly increased (Figure 3B) and the plasma glucagon level was decreased in Tg-SGLT2i mice (Figure 3C). Glucose transporter 2 (GLUT2), which is a glucose sensor that transports glucose into the cell in response to subtle changes in blood glucose concentration, was highly expressed in the transmembrane region of β cells in WT-control mice (Figure 3D). In diabetic Tg-control mice, however, GLUT2 was broadly expressed in the cytoplasm. In Tg-SGLT2i mice, many β cells expressed GLUT2 in the transmembrane region, suggesting that β-cell function was improved.
Albuminuria, Ccr, and Morphology
Male Tg-control mice develop severe diabetic renal injury due to sustained hyperglycemia. Typical progression of diabetic renal injury was seen with age: albuminuria appeared at around 10 to 11 weeks and continued to increase (Figure 4A), and Ccr (both body weight–adjusted and unadjusted Ccr) peaked at 20 weeks and then declined in association with worsening glomerulosclerosis but remained higher than that of WT mice at 48 to 50 weeks (Figure 4B).
Therapeutic effect of canagliflozin on histology and renal function in this model was analyzed. Glomerulosclerosis was evaluated at two different time points by measuring PASM-positive areas. In hyperglycemic Tg-control mice, mesangial matrix expansion (glomerular sclerosis) was prominent (Figure 4C). However, in Tg-SGLT2i mice, the mesangial matrix fraction was significantly decreased at 33 weeks of age. It was further decreased at 48 to 50 weeks of age (Figure 4D), suggesting that amelioration of already-developed sclerosis is a slow process. Glomerular deposition of extracellular matrix ColIV, a major component of the expanded extracellular matrix, was next examined. In hyperglycemic male Tg-control mice, ColIV protein expression was significantly increased, mainly in the mesangial areas (Figure 4E). In contrast, mesangial ColIV protein localization was reduced at 33 weeks and further decreased at 50 weeks in canagliflozin-treated mice. Thickening of the tubular basement membrane is a typical sign of DN; at 50 weeks, tubular basement membrane was thicker in Tg-control mice than that in WT or Tg-SGLT2i mice (Figure 4E).
The expression of SGLT2 protein, which promotes glucose transport, was examined. Marked expression of SGLT2 was seen at the apical membrane of renal proximal tubules and Bowman’s capsules in hyperglycemic Tg-control mice (Figure 5A). In Tg-SGLT2i mice, however, the SGLT2 expression was reduced.
Urinary Volume, Glucose Excretion, and Kidney Morphology
Hyperglycemic Tg-control mice exhibited high levels of water intake, urinary volume, and urinary glucose; high urinary glucose excretion; and increased kidney weight with age (Figure 5, B-F). These findings were associated with pelvic expansion (hydronephrosis) and tubular dilation (Figure 5G). In Tg-SGLT2i mice, water intake, urinary volume, and kidney weight per body weight were significantly decreased (Figure 5, B, C, F). Urinary volume and urinary glucose excretion, as well as water intake, of Tg-SGLT2i mice increased within the first 1 to 3 days (at 13 weeks), and then decreased. The decrease may be related to the improvement in blood glucose levels (Figure 5, C and E). Glomerular size in Tg-SGLT2i mice was smaller than in Tg-control mice (Figure 5, H and I). There was no significant difference in the number of glomeruli in mid-transverse sections of the kidneys (Figure 5J).
The change in renal function was examined next. In hyperglycemic male Tg-control mice, Ccr peaked at 20 weeks of age and declined with progression of sclerosis but remained high at 48 weeks of age with increased urinary volume (Figure 5K). However, this glomerular hyperfiltration was gradually reduced in Tg-SGLT2i mice concomitantly with reduced daily urinary volume and water intake but remained higher compared with that in WT mice at 50 weeks. Albuminuria was quickly decreased (Figure 5L). These results show that long-term intervention with canagliflozin improved both histology and renal function, and inhibited the progression of glomerulosclerosis.
Glucose-Normalization Effect of Islet Transplantation
Whether the morphologic change in the SGLT2i group is glucose dependent or independent (ie, whether glucose-normalization alone is as effective as canagliflozin treatment) was investigated next. Five hundred islets obtained from C57BL/6 WT mice were implanted under the kidney capsule of male Tg mice (Tg-IT) (Figure 6A) at 12 weeks; the presence of these islets in the kidney graft was clearly visualized by insulin staining. The blood glucose levels of the islet-transplanted Tg-IT mice decreased rapidly to the normal range within 1 week and remained at this level throughout the study (Figure 6B).
The pancreatic islets of Tg-IT showed a pattern similar to Tg-SGLT2i; that is, disorganized islets with reduced β cells (Figure 6C). GLUT2 was expressed in the β-cell membrane (Figure 6D), suggesting that β-cell function was improved.
Serum parameters such as triglyceride level, and parameters such as water intake, food, and urinary volume, were all markedly improved by islet transplantation (Table 1). The values of Ccr (normalized by body weight) in the diabetic Tg-control group, islet-transplanted group, SGLT2i-treated group, and WT group at 50 weeks of age were 8.7, 4.9, 3.9, 2.6 (μL/minute per gram), respectively (islet transplantation group versus SGLT2i-treated group, P = 0.01).
Table 1The Physical Profile and Glomerular Pathology
P < 0.05. AST/ALT, aspartate aminotransferase/alanine aminotransferase; ColIV, type 4 collagen; HbA1c, glycosylated hemoglobin; TCHO, total cholesterol.
1226 ± 71
The physical profile and glomerular pathology of male transgenic (Tg) mice treated with islet transplantation (IT) or canagliflozin and age-matched diabetic Tg-control (Tg-ctrl) mice at 50 weeks. Data are expressed as means ± SEM [wild-type (WT; n = 12), Tg-ctrl (n = 6), Tg–sodium-glucose cotransporter 2 inhibitor (SGLT2i) (n = 5), and Tg-IT (n = 8)]. Significance was determined by one-way analysis of variance with a post hoc Tukey-Kramer test.
∗ P < 0.05.AST/ALT, aspartate aminotransferase/alanine aminotransferase; ColIV, type 4 collagen; HbA1c, glycosylated hemoglobin; TCHO, total cholesterol.
SGLT2 expression was decreased in the kidneys of Tg-IT mice compared with those of Tg-control mice (Figure 6E), but not as much as in those of Tg-SGLT2i mice. The expansion of the renal pelvis and tubules appeared to be considerably reduced by islet transplantation (Figure 7A). However, the glomerular sclerosis (Figure 7, B and C) and deposition of extracellular matrix in the glomeruli (Figure 7D) were not fully reversed at 50 weeks, compared with those seen in Tg-SGLT2i mice. These findings suggest that canagliflozin treatment is more effective for the amelioration of glomerulosclerosis than glucose normalization alone.
In our model, glomerulosclerosis progressed concomitantly with diabetes, and mild interstitial fibrosis appeared as a widening of the space between tubular cells (Figure 8A). The PASM-positive fibrotic area was analyzed quantitatively (Figure 8B). In hyperglycemic Tg-control mice, this fibrotic area had progressed at 50 weeks, which in contrast was decreased in SGLT2i-treated mice at 33 weeks of age and further reduced at 48 to 50 weeks of age. In Tg-IT mice, the fibrotic area of the interstitium was significantly reduced compared with Tg-control mice but not as much as SGLT2i-treated mice at 50 weeks (Figure 8B).
Effect of Canagliflozin on Progression of Renal Damage Induced by Heminephrectomy
In Tg-control mice, glomerulosclerosis appeared, but Ccr was still at a high level at 33 weeks. Therefore, an additional hyperfiltration overload was applied by means of heminephrectomy at 33 weeks (Figure 9A). All diabetic Tg-ctrl mice subjected to heminephrectomy died within 1 week and thus could not be used for subsequent comparison as controls. However, the Tg-SGLT2i mice survived, and changes in the remnant kidney were analyzed at 48 to 50 weeks (Tg-SGLT2i + nephrectomy). The effect of heminephrectomy in WT mice was examined. Heminephrectomy was performed at 33 weeks in WT mice, and analysis was done at 50 weeks (WT + nephrectomy) (Supplemental Figure 1A). There was no change in SGLT2 expression or in the renal pelvic expansion (Supplemental Figure S1, B and C), but tubules were expanded, and the glomerular surface area was significantly increased (Supplemental Figure S1, D and E).
In Tg-SGLT2i mice, the glomeruli were relatively enlarged and the sclerotic area was increased compared with that in WT mice (Figure 9D). In Tg-SGLT2i + nephrectomy mice, urinary glucose excretion was reduced to one-half of that in Tg-SGLT2i mice (Figure 9B). Blood glucose levels remained in the normal range in Tg-SGLT2i + nephrectomy mice (Figure 9C). Food intake was unchanged.
In the remaining kidney of Tg-SGLT2i + nephrectomy mice, renal lesion progressed rapidly. Mesangial matrix expansion (glomerular sclerosis) was prominent, and the surface area was markedly increased (glomerular hypertrophy) (Figure 9, D and E). Deposition of the extracellular matrix was considerably increased (Figure 9F). Further expansion of the renal pelvis and tubules was observed (Figure 10, A and B ). However, albuminuria was not increased (Figure 10C), and Ccr was significantly decreased (Figure 10D). SGLT2 expression was similar in Tg-SGLT2i with or without heminephrectomy (Figure 10E).
Changes in Expression and Localization of HIF-1α and HIF-2α
Expression and localization of HIF-1α and HIF-2α was examined in renal tubular cells, glomerular cells, and pancreas.
In WT mice, HIF-1α was not expressed in tubules or glomeruli at 50 weeks, whereas in Tg-control mice, it was strongly and diffusely expressed in the nuclei of tubular cells and in some peritubular capillaries throughout the experimental period (Figure 11, A and B ). However, this nuclear expression was reduced in Tg-SGLT2i mice at 33 weeks and was completely lost at 50 weeks, becoming comparable to that of WT mice. In Tg-IT mice, tubular nuclear expression was reduced but apparently remained in some regions at 50 weeks. Interestingly, when SGLT2i-treated mice were heminephrectomized, tubular nuclear expression reappeared.
A similar change in HIF-1α expression was also observed in glomerular cells (Figure 11C). Glomerular expression of HIF-1α was prominent mainly in nuclei of Tg-control mice; it was only segmentally expressed in Tg-SGLT2i mice at 33 weeks and completely disappeared at 50 weeks.
Thus, both tubular and glomerular expressions of HIF-1α were persistently up-regulated in Tg-control mice but were partially attenuated by islet transplantation (correction of hyperglycemia), and completely disappeared with canagliflozin intervention.
The pattern of HIF-2α expression was somewhat different from that of HIF-1α. In WT mice at 50 weeks, HIF-2α was weakly and diffusely expressed in tubular cytoplasm, and was more prominently expressed in the deeper cortex, whereas it was weakly expressed in the glomerulus, most likely in the endothelium (Figure 12A). No significant staining was observed in the absence of the primary or the secondary antibody, confirming that the staining was specific. In Tg-control mice, HIF-2α was diffusely expressed in tubular nuclei and cytoplasm, as well as in glomerular cell nuclei (Figure 12, A and B). In Tg-SGLT2i mice at 33 weeks, tubular nuclear expression partially remained in some regions, and at 50 weeks, nuclear expression had almost disappeared, with only weak residual expression in the cytoplasm. In Tg-IT mice, nuclear HIF-2α in tubules was similarly lost at 50 weeks, indicating that glucose normalization is sufficient to attenuate the tubular nuclear expression of HIF-2α. However, after heminephrectomy, tubular nuclear expression reappeared, as did glomerular nuclear expression. Thus, sudden hyperfiltration overload might induce the expression of HIF-2α.
In glomerular cells of Tg-control diabetic mice, HIF-2α was continuously expressed in nuclei (Figure 12, A–C). Canagliflozin treatment reduced the expression of HIF-2α at 33 weeks, and the expression was mostly lost at 50 weeks. Attenuation of HIF-2α by islet transplantation differs between renal tubules and glomeruli: HIF-2α was still expressed in some glomerular cells in Tg-IT mice at 50 weeks. Thus, the expression of HIF-1α and HIF-2α was not synchronized, and HIF-1α and HIF-2α appear to be regulated differently.
The localization and expression level of HIF-1α in the pancreas were also examined by immunostaining. In hyperglycemic Tg-control mice, HIF-1α was expressed in the nuclei of islet cells and duct cells diffusely throughout the experimental period. In Tg-SGLT2i mice, HIF-1α expression was reduced in islet cells at 48 to 50 weeks, becoming comparable to that of WT mice (Figure 12D). In Tg-IT mice, expression of HIF-1α was reduced at 50 weeks.
Detection of Cellular Oxidative Stress
Oxidative stress induced by hypoxia was further investigated using 8-OHdG, an oxidative DNA damage marker. 8-OHdG was not expressed in diabetic, SGLT2i-treated, or WT kidneys (Figure 13). 8-OHdG was detected only in nuclei of renal tubules after nephrectomy (application of acute hyperfiltration overload). Therefore, the dynamic changes in HIF expression do not seem to be related to oxidative stress, at least at a level detectable by 8-OHdG immunohistochemistry.
In the present work, to explore the factors that influence the progression of DN, whether the development and amelioration of DN (the severity of glomerulosclerosis and interstitial fibrosis) are related to changes in hyperglycemia, hyperfiltration, and HIF expression, was investigated. In this extended process, DN in mice was compared after a long-term intervention (50 weeks) with canagliflozin and glucose normalization alone by islet transplantation. The overall results are summarized in Figure 14. The model mice developed severe diabetes early in life due to profound depletion of insulin by pancreatic β-cell–directed ICER-Iγ overexpression.
In this model, proteinuria appeared at 13 to 14 weeks of age (initiation of DN), whereas Ccr peaked at 20 weeks and then declined, but still remained higher than in WT mice at 50 weeks. Albuminuria increased continuously. Glomerulosclerosis was marked by 33 weeks and progressed at 50 weeks. Similarly, tubulointerstitial fibrosis appeared at 33 weeks and progressed at 50 weeks, as indicated by the increase of the space between tubular cells.
With regard to these changes, this model mimics human DN more closely than other experimental models,
such as the frequently used STZ model. In this context, it should be noted that the Animal Models of Diabetic Complication Consortium recommendations state that the effect of STZ is not consistent, but depends on the method of administration. Furthermore, STZ is toxic to kidney cells and causes acute kidney injury, especially when administered intraperitoneally in a large dose as a single shot.
In contrast to human clinical trials, this mouse model provides a uniform renal pathology that is free from the influence of genetic background, environment, diet, exercise, and life cycle. Therefore, it should be suitable to evaluate the real therapeutic effect of an SGLT2 inhibitor on the progression of DN.
When canagliflozin was started after appearance of albuminuria at 13 to 14 weeks, blood glucose levels were rapidly reduced within 1 week and thereafter remained stable (Figure 14). Urine volume increased immediately after SGLT2i administration within the first 1 to 3 days, and then decreased, probably because of the improvement in blood glucose levels. Albuminuria quickly decreased nearly to the control level. Ccr was reduced gradually but remained higher than in WT mice at 50 weeks, suggesting that hyperfiltration was still present. Glomerulosclerosis seen at 33 weeks was further ameliorated at 50 weeks by canagliflozin intervention, showing a slow recovery. Islet transplantation quickly normalized the glucose level and reduced Ccr, but Ccr remained higher than in the Tg-SGLT2 group at 50 weeks. Furthermore, glomerulosclerosis and tubular fibrosis were observed at 33 weeks, and were not improved at 50 weeks. Hyperfiltration was suppressed in the order of Tg-SGLT2i < Tg-IT < Tg-ctrl; thus, islet transplantation suppressed hyperfiltration to a lesser extent than SGLT2i treatment. Both Tg-IT and Tg-SGLT2i improved albuminuria quickly, but it took a long time for hyperfiltration to improve. This delay was probably caused because the hyperexpression of SGLT2 in the tubules was reduced gradually by the administration of SGLT2i.
Once diabetic glomerulosclerosis appears, it is difficult to reverse solely by normalizing glycemic control.
also showed, in a human renal biopsy study, that glomerulosclerosis was not improved 5 years after pancreas transplantation but was improved at 10 years. In accordance with this important result, the current study found that glomerulosclerosis observed at 33 weeks was improved at 50 weeks by canagliflozin but not by glucose normalization alone. Similar changes were seen in interstitial fibrosis, although fibrosis remained mild at 50 weeks. One of the reasons for this improvement may be the suppression of hyperfiltration by canagliflozin, because Ccr at 50 weeks was decreased more by canagliflozin treatment than by islet transplantation. Glucose normalization and reduction of albuminuria improved much earlier. Li et al
have reported improvements in glomerular expansion and interstitial fibrosis in an STZ-induced diabetic model; however, they used a high dose of STZ (200 mg/kg), which might induce acute tubular necrosis. They also used periodic acid-Schiff staining for the assessment of glomerular expansion, and periodic acid-Schiff stains not only sclerotic areas but also mesangial cells. In addition, blood glucose was not properly controlled by insulin or SGLT2 inhibitor in their experiment; neither empagliflozin nor insulin treatment normalized the STZ-induced high blood glucose levels, so that the degree of recovery from glomerular injury and fibrosis was evaluated under conditions of maintained hyperglycemia. In contrast, in the current model, DN progressed slowly, and interstitial fibrosis appeared at 33 weeks as a slight enlargement of the spaces between PAM-stained tubules. Therefore, this model is very different from STZ-induced DN, making it a better representative of the human condition. The current findings are the first demonstration of amelioration of already-initiated diabetic glomerulosclerosis by canagliflozin.
SGLT2 in the kidney affects hyperfiltration. The current study found that Tg-control mice exhibited increased expression of SLGLT2 in the proximal tubules, in agreement with the finding by Wang et al
that SGLT2 expression is increased in human DN and db/db diabetic mice. The enhanced SGLT2 expression was decreased to some extent by islet transplantation. Although it was further decreased by canagliflozin treatment, it did not reach the level of WT mice at 50 weeks. These results suggest that high SGLT2 levels in the tubules may lead to prolonged hyperfiltration, even if glucose and albuminuria levels are improved.
Glomerular and tubular expression of HIF-1α and HIF-2α was also examined, as it might influence glomerulosclerosis and interstitial fibrosis. Thus far, localization studies of HIF have primarily been performed by using immunohistopathologic methods in short-term experiments,
Thus, the patterns of long-term regulation and the effects of HIF-1α and HIF-2α on a long-term chronic hyperglycemic state have not been explored. This study found that in WT mice (50 weeks old), HIF-1α was not expressed throughout the experimental period in glomerulus or in tubular cells. In contrast, HIF-2α was expressed weakly in glomerular endothelial cells and more markedly in the tubular cytoplasm in WT mice. Rosenberger et al
similarly observed weak cytoplasmic HIF-2α staining in renal tubular epithelial cells in ischemic mouse kidney, although they did not comment on this finding. Therefore, the current study carefully investigated the specificity of the immunohistochemical technique and concluded that the cytoplasmic expression of HIF-2α in tubular cells was specific. However, cytoplasmic HIF-2α was inactive, because HIF-2α must be translocated to the nucleus to activate gene transcription. Thus, cytoplasmic expression and nuclear expression should be differentiated. Hanna et al
described the expression of HIF-2α in an Adriamycin model and proposed a possible contribution of HIF-2α to glomerulosclerosis. Clear nuclear HIF-2α staining was seen in von Hippel Lindau knockout mice, and continuous HIF-2α expression in tubular cells led to renal fibrosis and multiple cysts only in aged mice.
Overall, it appears that continuous abnormal expression of HIF-2α could be pathologic and might contribute to renal fibrosis.
Here, long-term changes in nuclear HIF-1α and HIF-2α expression both in glomerular and tubular cells in the kidneys of diabetic, canagliflozin-treated, and Tg-IT mice were shown for the first time. Under sustained hyperglycemia, HIF-1α was persistently overexpressed in the nuclei of glomerular and tubular cell nuclei throughout the experimental period of 50 weeks. However, strikingly, nuclear HIF-1α overexpression was remarkably decreased at 33 weeks by canagliflozin intervention and completely disappeared at 50 weeks (Figure 14). Although nuclear HIF-1α overexpression decreased in Tg-IT mice, it was still detectable even at 50 weeks. HIF-2α was weakly expressed in tubular cytoplasm. In Tg-control mice, HIF-2α was strongly expressed in glomerular cell and tubular cell nuclei, as well as tubular cytoplasm. HIF-2α was expressed to some extent in glomerular cell nuclei of Tg-SGLT2i mice at 33 weeks but was decreased in tubular cell nuclei. At 50 weeks, glomerular nuclear expression had almost disappeared. In contrast, Tg-IT mice exhibited reduced but still distinct expression of HIF-2α in glomerular cell nuclei. Thus, suppression of HIF-1α and -2α appears to be a slow process.
A recent review proposed that HIF-2α is down-regulated in diabetic kidney but increased by SGLT2 inhibitor and acts to improve DN
Role of impaired nutrient and oxygen deprivation signaling and deficient autophagic flux in diabetic CKD development: implications for understanding the effects of sodium-glucose cotransporter 2-inhibitors.
; however, the proposal was based on fragmentary findings and has not been experimentally established. Contrary to this, in the present study, HIF-2α induction was not observed in response to canagliflozin treatment; instead, overexpression of HIF-2α was definitely observed in hyperglycemic diabetes and a reduction by canagliflozin. HIF-2α expression may be different depending on conditions such as the duration of hyperglycemia and location in the kidney.
Although HIF-1α reportedly plays a protective role in hypoxia,
Due to the long-term nature of the in vivo study (50 weeks), it was not possible to focus on specific mechanisms or ascertain the direct evidence of the involvement of HIFs in glomerulosclerosis/interstitial fibrosis. Instead, the chronological consistency of persistent hyperfiltration, the increase and suppression of HIF expression, and the development and amelioration of glomerulosclerosis/interstitial fibrosis over a 50-week process were noted. The following facts should be considered: i) HIFs were persistently and extensively expressed in glomerulus and tubules, which should be different from the regulated transient expression induced by hypoxic injury; ii) as mentioned in the previous paragraph, there was a chronological consistency between overexpression of HIFs and progression of glomerulosclerosis and interstitial sclerosis, and between suppression of HIFs by SGLT2i and amelioration of glomerulosclerosis and interstitial sclerosis; and iii) HIFs have been reported in the literature to be involved in fibrosis not only in the kidney but also in other organs. It was therefore suggested that abnormal expression of HIFs is involved in the process of glomerulosclerosis and interstitial fibrosis.
The effects of oxidative stress were also examined. To detect oxidative stress specifically in the kidney, sections were immunostained for 8-OHdG, but staining was negative in all cases except in the kidney after heminephrectomy, which presumably reflects acute tubular damage. There have been many reports on the excretion of 8-OHdG in urine and/or blood or tissue of diabetic animals, mostly in STZ rats, and these findings may reflect acute tubular damage. However, in the current experiment, the oxidative stress specifically occurring in the kidney was considered to be important, and immunohistochemistry was applied. In this Tg mouse model, 8-OHdG expression was not detected in the kidney, except in the kidney after heminephrectomy. In the case of slow progression over a very long period, as in this Tg model (and in human DN, which takes 15 to 20 years from the onset of diabetes to the onset of DN), 8-OHdG may have been present but at levels too low to be detected, in contrast to STZ and db/db mice (which have a short life span), where oxidative stress increases rapidly. The possibility that persistent low-level oxidative stress and hypoxia could cause HIF expression can not be ruled out.
An anti-inflammatory effect of SGLT2i has been reported in many studies.
However, because SGLT2 is exclusively expressed locally in the tubules, it has not yet been proven whether SGLT2i directly suppresses inflammation. Ischemia and oxidative stress are the most common causes of inflammation. Although there was no detectable increase in 8-OHdG in the present experiment, except in the heminephrectomized group, it is possible that a low (undetectable) level of chronic ischemia and/or oxidative stress was present, and SGLT2i may improve relative ischemia and oxidation in part by inhibiting hyperfiltration.
For hypoxia, pimonidazole adduct immunohistochemistry was performed, but HIF expression was observed only in the medulla of diabetic mice. Because the localizations of HIF expression and the hypoxia probe were completely different, no relation between HIF expression and the levels of hypoxia measured with the hydroxy probe (A.I. and A.F., unpublished data) was observed. These results suggest that neither oxidative stress nor hypoxia plays a major role in the development of DN, except in mice subjected to heminephrectomy. Notably, Tg-control mice did not survive this hyperfiltration overload and oxidative stress, but SGLT2i-treated mice did. Hyperfiltration caused by the increased expression of SGLT2, which resulted in increased energy consumption, may have induced the overexpression of HIFs.
In pancreatic islets, β cells were depleted in Tg mice, but in mice treated with canagliflozin, some β cells were observed and membrane GLUT2 expression was increased, indicating the functional recovery of β cells. SGLT2 was not detected immunohistochemically. Insulin secretion was also increased. Similar findings were observed after islet transplantation, indicating this improvement can be attributed to glucose normalization. This is consistent with the report by Dai et al
that dapagliflozin does not directly affect human α or β cells.
HIF stabilizer has been proven to be effective in renal anemia, but thus far no clinical trials have investigated the effect on fibrosis. Currently, HIF stabilizers are used cautiously in retinopathy, malignancy, and cystic kidney disease, as they increase vascular endothelial growth factor. However, it is now known whether fibrosis in various organs is associated with HIF, and the possibility of fibrosis in the kidney and other organs should be considered. These results are consistent with this possibility. However, involvement of HIFs in the induction of kidney fibrosis remains controversial. Contradictory reports have appeared. Among them, Yu et al
reported that the timing of HIF expression during the process of kidney injury is important for renal fibrogenesis. Physiologically, HIF is activated in the nucleus and acts to promote transcription of various genes. Its up-regulation is very quick (within 45 minutes), and the signal disappears within 8 hours.
The current results suggest that sustained overexpression of HIFs may play a crucial role.
In conclusion, the Tg-ICER-Iγ mouse model shows progressive DN, including glomerulosclerosis and mild interstitial fibrosis, with persistent overexpression of HIF-1α and HIF-2α in glomerulus and tubules. Canagliflozin treatment was effective in improving hyperglycemia and albuminuria and, in the long-term, it reduced HIF expression and hyperfiltration presumably through suppression of SGLT2, and ameliorated glomerulosclerosis and interstitial fibrosis even after the establishment of DN. Thus, these findings suggest that persistent glomerular hyperfiltration and persistent HIF overexpression contribute to the development of glomerulosclerosis and interstitial fibrosis (Figure 15), and that the abnormal expression of HIF-1α and HIF-2α is a potential therapeutic target for preventing glomerulosclerosis and interstitial fibrosis.
We thank Hiroshi Fujii (Kyushu University) for providing tissue sections, Kei Fujishima (Kyushu University) and Yumi Maruyama (Institute of Biomedical Research and Innovation) for technical assistance, and the RIKEN BioResource Research Center (Tsukuba) for technical support with urinary albumin and Cre measurements.
Effect of heminephrectomy in wild-type (WT) mice. A: Heminephrectomy was performed at 33 weeks in WT mice, and subsequent changes in remnant kidney were analyzed at 48 to 50 weeks (WT + neph). B: No change in sodium-glucose cotransporter 2 (SGLT2) expression was observed. C: Representative low-power micrographs of kidneys. D: Magnified view of the renal structure. Renal damage with the expanded renal pelvis and tubules. Periodic acid-Schiff staining. E: Quantification of the mesangial matrix area and glomerular surface area. Mesangial matrix expansion and glomerular surface area are significantly increased, confirming overload of renal damage by heminephrectomy. Significance was determined by means of a two-tailed t-test (unpaired). ∗P < 0.05. Scale bars: 50 μm (A and D). Original magnification: ×10 (C).
A.I. designed the study, performed the experiments, collected and interpreted data, obtained funding, and wrote the manuscript; Y.Y. developed the protocols for the islet isolation and transplantation and provided critical advice; O.I. performed islet isolation and transplantation with A.I., interpreted data and revised the manuscript; K.A. provided canagliflozin and participated in discussions; Y.N. contributed to the launch of this project with A.I.; and A.F. participated in critical discussions and revised the article for intellectual content. A.I. is the guarantor of this work, and as such, had full access to all of the data in the study, and takes responsibility for the integrity of the data and the accuracy of the data analysis. All authors discussed the results, commented on the manuscript, and approved the final version.
Diabetic nephropathy: worldwide epidemic and effects of current treatment on natural history.
Role of impaired nutrient and oxygen deprivation signaling and deficient autophagic flux in diabetic CKD development: implications for understanding the effects of sodium-glucose cotransporter 2-inhibitors.
Supported by Tanabe Mitsubishi Pharma Corporation (collaborative study) and Novartis Foundation (A.I.).
Disclosures: A.I. and A.F. are listed as inventors in patent applications JP4576514 and US7745689B2, both relating to the diabetic mouse model. K.A. is an employee of Tanabe Mitsubishi Pharma Corporation.
Current address of O.I., unaffiliated, Kyoto, Japan.