Diabetic Nephropathy Is Accelerated by Farnesoid X Receptor Deficiency and Inhibited by Farnesoid X Receptor Activation in a Type 1 Diabetes Model

Homozygous male FXR knockout mice (FXR KO) of 6 months of age backcrossed onto the C57BL/6 genetic background for 10 generations (13), sex- and age-matched C57BL/6 wild-type mice, and 8-week-old male DBA2/J mice were all obtained from Jackson Laboratories (Bar Harbor, ME). They were maintained on a 12-h light/12-h dark cycle. The deletion of FXR was confirmed with genotyping and Western blot (supplemental Fig. S1, available in the online appendix at http://diabetes.diabetesjournals.org/cgi/content/full/db10-0019/DC1). Mice were injected with streptozotocin (STZ) (Sigma-Aldrich, St. Louis, MO) intraperitoneally (40 mg/kg for DBA/2J and 50 mg/kg for C57BL/6 strains, freshly made in 50 mmol/l sodium citrate buffer, pH 4.5) for 5 consecutive days, or with 50 mmol/l sodium citrate solution only. Tail vein blood glucose levels were measured 1 week after the last STZ injection, and mice with glucose levels >250 mg/dl were considered diabetic. FXR KO mice and their wild-type counterparts were fed with a high-fat high-cholesterol Western diet (WD, TD88137) obtained from Harlan-Teklad (Madison, WI) after the onset of diabetes and were studied after 12 weeks of diabetes. DBA/2J mice were fed WD after the onset of diabetes in STZ groups and were treated for 8 weeks with: 1) WD only; 2) the semisynthetic FXR agonist 6-α-ethyl-chenodeoxycholic acid (INT-747, Intercept Pharmaceuticals, New York, NY) (14): 30 mg/kg body weight admixed with WD. Animal studies and relative protocols were approved by the Animal Care and Use Committee at the University of Colorado Denver.

Blood glucose levels were measured using a Glucometer Elite XL (Bayer, Tarrytown, NY). Plasma lipid levels was measured with kits from Wako Chemical (Richmond, VA). Urine albumin and creatinine concentrations were determined using kits from Exocell (Philadelphia, PA).

Quantitative PCR and Western blotting were performed as previously described (4,,–7). Primer sequences are available from the authors on request or can be found elsewhere (4,,–7). The antibody against α-smooth muscle actin ([α-SMA], Sigma, St. Louis, MO) was used at 1:1,000 dilution for Western blotting.

Lipids from the kidneys were extracted by the method of Bligh and Dyer, as we have previously described (4,,–7). Triglyceride and cholesterol composition were measured by gas chromatography (Agilent Technologies, Wilmington, DE).

Nuclear protein extracts were prepared from kidney tissue and used for the measurement of NFκB transcriptional activity with a kit from Marligen Biosciences (Rockville, MD) according to the manufacturer's instructions.

The amount of oxidized proteins in kidney homogenates was determined by using an OxyElisa Oxidized Protein Quantitation Kit (Millipore, Billerica, MA) according to the manufacturer's instructions.

Sections (4-μm thick) cut from 10% formalin-fixed, paraffin-embedded kidney samples were used for periodic acid-Schiff (PAS) staining and Masson's trichrome staining. Frozen sections were used for oil red O staining of neutral lipid deposits or for immunostaining for nephrin (a gift from Dr Larry Holzman, University of Michigan, Ann Arbor, MI), synaptopodin (Sigma), fibronectin (Sigma), CD68 (AbD Serotec, Raleigh, NC), and α-SMA (Sigma), and imaged with a laser scanning confocal microscope (LSM 510, Zeiss, Jena, Germany). The expression level was quantified as the sum of pixel values per glomerular area using ImageJ version 1.44 image analysis software. Electron microscopy (EM) was conducted in the Mouse Metabolic Phenotyping Center (University of Washington, Seattle, WA). Samples for EM were fixed in 1/2 x Karnovsky's fixative.

All quantifications were performed in a masked manner. Using coronal sections of the kidney, 30 consecutive glomeruli per mouse, 6 mice per group were examined for evaluation of glomerular mesangial expansion. The index of the mesangial expansion was defined as the ratio of mesangial area/glomerular tuft area. The mesangial area was determined by assessment of the PAS-positive and nucleus-free area in the mesangium using ScanScope image analyzer (Aperio Technologies, Vista, CA).

Results are presented as the means ± SE for at least three independent experiments. Data were analyzed by ANOVA and Student-Newman-Keuls tests for multiple comparisons or by Student t test for unpaired data between two groups. Statistical significance was accepted at the P < 0.05 level.

To investigate whether FXR plays a role in diabetic nephropathy and if FXR deficiency accelerates diabetic kidney injury, we induced insulin deficiency and hyperglycemia with STZ in FXR KO mice with the C57BL/6 genetic background. Progress of renal injury was compared with wild-type C57BL/6 mice injected with STZ, a mouse strain previously shown to be resistant to STZ-induced hyperglycemic kidney disease (15). To accelerate disease progression, we placed all mice on WD after they developed hyperglycemia. As shown in Table 1, STZ injections led to a marked increase in blood glucose level in both wild-type and FXR KO mice. Both wild-type mice and FXR KO mice injected with STZ had markedly low insulin levels. FXR deficiency without diabetes caused a mild increase in plasma triglyceride, total cholesterol, HDL cholesterol, and LDL cholesterol levels, compared with C57BL/6 wild-type mice. In contrast, a significant increase of plasma lipid level was observed in diabetic FXR KO mice on WD, associated with proatherogenic changes, including a decrease in HDL cholesterol, increase in LDL cholesterol, and the dominant LDL cholesterol level in total cholesterol.

Diabetic wild-type C57BL/6 mice did not show increased albuminuria, and there was only moderate albuminuria in nondiabetic FXR KO mice fed a WD. However, urinary albumin/creatinine ratio in FXR KO mice after 12 weeks of diabetes was markedly increased by 11-fold over diabetic wild-type C57BL/6 mice (Table 1). The development of albuminuria in diabetic FXR KO mice was associated with increased diabetes-induced renal hypertrophy compared with wild-type mice, but the difference did not reach statistical significance, as revealed by the kidney-to-body weight ratio (Table 1). However, the glomerular volume was significantly increased in diabetic FXR KO mice (Table 1).

Wild-type C57BL/6 mice or nondiabetic FXR KO mice showed nearly normal glomerular structure with only mild mesangial expansion (Fig. 1,A–C). In contrast, the glomerular tufts in diabetic FXR KO mice exhibited foam cell accumulation with glomerular lobulation enlarging the entire glomerular area and more mesangial matrix expansion (Fig. 1,D–I). Some capillaries were extended and occluded by foam cells or dilated and contained pale-staining material (Fig. 1,E). Some glomeruli displayed prominent mesangiolysis accompanied by ballooning of capillaries (Fig. 1,F and G). In tubulointerstitial areas, wild-type mice or nondiabetic FXR KO mice showed no significant tubular damage, whereas diabetic FXR KO mice had remarkable tubulointerstitial changes. The tubules were dilated, lined by flattened epithelium, and contained proteinaceous casts with large amounts of lipid droplets deposited in the tubular epithelial cells (Fig. 1,H). Interstitial collagen deposition was studied in Masson's trichrome-stained renal sections as an index of interstitial fibrosis. Diabetic FXR KO mice showed more interstitial fibrosis than wild-type mice and nondiabetic FXR KO mice (Fig. 1 J–M).

Electron microscopic examination from diabetic FXR KO mice disclosed increased podocyte and endothelial cell reactivity and local foot process effacement (Fig. 1,N–Q). As shown in Fig. 1 R, prominent mesangiolysis and lipid inclusions in the mesangium were observed, accompanied by dissociation of mesangial matrix and disruption of anchoring of glomerular basement membrane (GBM) to mesangium.

In agreement with lipid deposits and foam cell accumulation, immunofluorescence microscopy revealed an increase in CD68-positive macrophages in the glomeruli and tubulointerstitium in diabetic FXR KO mice compared with wild-type controls (Fig. 2,A). This is associated with increased expression of master inflammatory mediator NFκB subunit p65 and elevated NFκB activation level (Fig. 2 B).

An important mechanism causing albuminuria is podocyte dysfunction (16,17). Immunofluorescence microscopy showed that the distribution of podocyte markers, synaptopodin and nephrin, was changed from uniformly linear pattern in wild-type and nondiabetic FXR KO mice to discontinuous pattern with loss of staining in diabetic FXR KO mice (Fig. 2,C). Nephrin is a key component of the slit diaphragm complex, and synaptopodin, a differentiated podocyte marker, is a regulator of actin dynamics for podocyte foot processes. Both are critically important for the sustained function of glomerular filtration barrier to proteinuria (16,–18). Podocyte loss was also confirmed by staining with WT1, a nuclear podocyte marker, showing a significantly reduced podocyte density in diabetic FXR KO mice (Fig. 2 D). Our results thus indicate podocyte loss, decreased podocyte density, and loss of podocyte differentiation markers in the kidneys of diabetic FXR KO mice.

Glomerulosclerosis and tubulointerstitial fibrosis are pathologic hallmarks of diabetic renal disease, characterized by accumulation in the kidney of extracellular matrix (ECM) proteins and myofibroblasts, the primary matrix-producing cells (19). TGF-β is a cytokine that plays a pivotal role in the profibrotic responses (20). In diabetic FXR KO mice, we found a significant increase of TGF-β mRNA expression in the kidney compared with corresponding wild-type or nondiabetic controls, in parallel with increased expression of microRNA-192 and decreased expression of microRNA-29a (Fig. 2,E). MicroRNAs have been shown to be actively involved in TGF-β signaling. MicroRNA-192 is important for TGF-β–induced ECM production (21). MicroRNA-29 inhibits several mRNAs involved in ECM production and fibrosis (22). Our data suggest that TGF-β-induced increase in the expression of microRNA-192 coordinately downregulates microRNA-29a to control the induction of fibrosis gene expression related to the pathogenesis of diabetic nephropathy. We also examined the expression of ECM protein fibronectin by immunofluorescence, and found increased expression of fibronectin in glomeruli of diabetic FXR KO mice (Fig. 2,F). FXR deficiency also increased the renal expression level of two myofibroblast markers, fibroblast-specific protein-1 (FSP-1) and α-SMA (Fig. 2 G).

We found that FXR deficiency caused increased kidney neutral lipid accumulation in both glomeruli and tubulointerstitium, as determined by oil red O staining, and biochemical lipid analysis revealed increased kidney triglyceride and cholesterol content (Fig. 3,A). To explore the mechanism by which FXR regulates renal lipid accumulation, we investigated pathways regulating renal lipid metabolism. In glomeruli isolated from diabetic FXR KO mice, we found significantly increased expression of SREBP-1c and its target genes fatty acid synthase (FAS), acetyl-CoA carboxylase (ACC), and stearoyl-CoA desaturase-1 (SCD-1), which mediate fatty acid and triglyceride synthesis. In addition, we observed increased expression of LDL receptor and lectin-like oxidized LDL receptor-1 (LOX-1), which mediate cholesterol and oxidized LDL uptake (Fig. 3 B).

To further confirm the role of FXR in diabetic nephropathy, we used the selective FXR agonist INT-747 to test whether the FXR activation can ameliorate diabetic nephropathy in a STZ-induced type 1 diabetes model. Because C57BL/6 mice are resistant to diabetic renal injury, STZ-injected nephropathy-susceptible DBA/2J mice were used (15). As shown in Table 2, STZ injections led to a marked increase in blood glucose level in DBA/2J mice. Diabetic DBA/2J mice developed increased triglyceride and cholesterol levels, with a marked increase in LDL cholesterol levels. Treatment with the selective FXR agonist INT-747 did not decrease the blood glucose level in diabetic DBA/2J mice, but it significantly decreased plasma total cholesterol and LDL cholesterol levels. No changes were observed in the HDL cholesterol and triglyceride levels (Table 2).

Diabetic DBA/2J mice on WD developed severe albuminuria, which was significantly decreased and nearly normalized by INT-747 treatment (Table 2). In diabetic DBA/2J mice, morphometric analysis revealed a moderate but significant increase of mesangial matrix expansion in glomeruli which was blunted by INT-747 treatment (Fig. 4,A–D). Patchy fibrosis was demonstrated in the tubular interstitium of diabetic DBA/2J kidneys, but was nearly absent in the kidneys of mice treated with INT-747 (Fig. 4 E and F).

In diabetic DBA/2J mice, immunofluorescence staining showed in the kidney reduced expression of synaptopodin, which was prevented by INT-747 treatment (Fig. 4,G). Podocyte loss revealed by WT1 staining (Fig. 4,H) in diabetic DBA/2J mice was also rescued by INT-747 treatment. Conversely, INT-747 treatment decreased renal Notch1 mRNA level which was enhanced in diabetic DBA/2J kidneys (Fig. 4 I). As Notch activation in podocytes is involved in the development of proteinuria and podocyte dysfunction (23), these data suggest that INT-747 prevents the development of proteinuria by preventing podocyte dysfunction in DBA/2J mice.

In diabetic DBA/2J mice, INT-747 treatment blocked the increase of TGF-β expression in diabetic kidneys, suggesting that FXR activation may counteract kidney fibrosis induced by TGF-β (Fig. 5,A). In addition, INT-747 treatment markedly inhibited fibronectin expression in glomeruli (Fig. 5,B) and significantly reduced renal α-SMA and FSP-1 expression, which were both increased in diabetic DBA/2J mice (Fig. 5,C). FXR activation markedly decreased the expression of macrophage marker CD68 in the glomeruli of diabetic kidneys (Fig. 5,D), which was consistent with its inhibition in p65 expression and NFκB activity (Fig. 5,E). INT-747 also modulated oxidative stress, as shown by decreased NADPH oxidase Nox-2 and p22-phox expression and total protein carbonylation in diabetic kidneys from INT-747-treated mice (Fig. 5 F).

Diabetic DBA/2J mice had increased kidney neutral lipid accumulation in both glomeruli and tubulointerstitium and enhanced kidney triglyceride and cholesterol content. Treatment with INT-747 significantly attenuated kidney lipid accumulation (Fig. 6,A). Correspondingly, glomerular lipid synthesis genes were downregulated, including: 1) SREBP-1c and its target gene SCD-1, carbohydrate responsive element binding protein (ChREBP) and its target gene liver pyruvate kinase (LPK), which mediate fatty acid and triglyceride synthesis, and 2) SREBP-2 which mediates cholesterol synthesis (Fig. 6 B).

The purpose of this study was to examine the role of the nuclear bile acid receptor FXR in diabetic nephropathy. The present data demonstrate accelerated renal injury in a model of experimental diabetic nephropathy in FXR KO mice on the nephropathy-resistant C57BL/6 background. In a type 1 diabetes model using the nephropathy-susceptible DBA/2J strain, renal injury was attenuated by FXR activation with the selective FXR agonist INT-747, suggesting that FXR plays a pivotal role in diabetic renal injury in these models.

Clinical observations have suggested that hyperlipidemia contributes to the progression of diabetic renal disease (24,25). To accentuate diabetic injury in the kidney, we have combined type 1 diabetes and hyperlipidemia by feeding WD to mice rendered diabetic with STZ. FXR deficiency triggers hypertriglyceridemia and hypercholesterolemia in diabetic C57BL/6 mice on WD, but not in nondiabetic mice. In addition, in diabetic mice FXR deficiency switches the lipid profile to pro-atherogenic by decreasing HDL cholesterol and increasing LDL cholesterol. The same degree of hyperlipidemia and lipid profile change is also observed in diabetic DBA/2J mice fed with WD. Interestingly, treatment with INT-747 decreases plasma cholesterol, but not triglyceride levels, in diabetic DBA/2J mice.

Our observations in the STZ-induced type 1 diabetes model are not consistent with FXR activation decreasing both plasma triglyceride and cholesterol levels (11,12,26), and with the observation that FXR KO mice have increased plasma LDL and HDL levels (13,27). This may reflect lack of insulin-mediated regulation of genes controlling lipid homeostasis in STZ-induced diabetes (28). However, we cannot exclude the possibility that INT-747 mixed in the diet admix is less bioavailable compared with the gavage administration (12), blunting some of its effects.

The renal injury in diabetic FXR KO mice was associated with severe albuminuria as well as a variety of structural changes in both glomeruli and tubulointerstitium not seen in the nondiabetic counterpart or in diabetic wild-type C57BL/6 mice. In this model we observed a large amount of lipid droplets and foam cell formation in the kidney, indicating an altered lipid homeostasis. There is growing evidence for the role of dysregulated lipid metabolism in the pathogenesis of renal disease (3,,,,,–9,29). Previous studies in our laboratory have provided evidence that renal de novo lipid synthesis contributes to renal lipid accumulation in the pathogenesis of nephropathy (4,,,,–9,12). Although circulating lipids can amplify renal injury in diabetes, in a separate study with regular low-fat chow diet, diabetic FXR KO mice still exhibited increased proteinuria, mesangial expansion, and fibrosis, as shown by increased collagen III deposition compared with wild-type control mice (supplemental online Fig. S2). These results suggest that absence of FXR, even in the absence of a Western diet, can cause renal pathologic changes. The deleterious effects of FXR deficiency may result from disruption of FXR-regulated renal lipid metabolism or by other FXR-mediated actions, as the kidney expresses high FXR levels (7,30). We have shown before that FXR activation mediates negative regulation of SREBP-1-mediated renal lipid metabolism (7,12). Consistent with this observation, in the current study we found that FXR activation in DBA/2J mice with STZ-induced diabetes significantly decreases glomerular expression of lipid synthesis genes, oil red O staining, as well as kidney lipid content. In addition, FXR deficient mice express more lipogenesis genes in glomeruli, indicating FXR-mediated regulation of renal lipid metabolism in diabetic FXR KO C57BL/6 mice as well as in diabetic DBA/2J mice.

Lipid metabolites can induce cellular dysfunction, a process known as lipotoxicity, to release profibrotic and proinflammatory cytokines and chemokines, increase the generation of reactive oxygen species, and promote the expression of extracellular matrix, contributing to oxidative stress, inflammation, and fibrosis in diabetic nephropathy (31,–33). All these features were inhibited by INT-747 treatment in diabetic DBA/2J mice. Lipid products include oxidized and glycated LDLs, advanced glycation end products (AGEs), free fatty acids, free cholesterol, excess triacylglycerols, diacylglycerols, and ceramides. Kidney cells can acquire these lipids from atherogenic lipoproteins in circulation and intracellular lipid synthesis. All major cell types in kidney, including endothelial cells, monocyte–macrophages. and mesangial cells, have been shown to cause oxidative modification of LDL (34). Our study demonstrates upregulation of LDL receptor and LOX-1 expression in the glomeruli of diabetic FXR KO C57BL/6 mice, which can facilitate the LDL and oxidized LDL uptake, consistent with a previous report showing FXR-mediated regulation of LOX-1 in another diabetic mouse strain, KKAy (35). In obese rats with uncontrolled diabetes and dyslipidemia, renal LOX-1 upregulation promotes lipid peroxidation stress, inflammation, and fibrosis (36). Excess neutral lipid accumulation in mesangial cells and macrophages forms the lipid-laden foam cells that we have observed in kidneys from diabetic FXR KO mice. Foam cell formation is generally thought of as a protective mechanism whereby the potentially harmful lipids are sequestered in lipid droplets. However, recent evidence shows that genetic ablation of lipid droplets surface protein adipocyte differentiation-related protein (ADFP/adipophilin) expression restricts foam cell formation and protects mice against atherosclerosis development, suggesting that foam cell formation per se is a crucial pathogenic event (37). How foam cell accumulation affects the renal function is not fully understood. Based on our present findings, we speculate that severe foam cell accumulation in the mesangium leads to loss of mesangial support for normal capillary architecture, with consequent pathologic manifestations such as impaired glomerular filtration.

Development and progression of diabetic nephropathy are characterized by inflammatory infiltrates (38). The macrophage is a central mediator of renal vascular inflammation, and its accumulation is a feature of diabetic nephropathy (39,–41). Macrophages mediate diabetic injury through a variety of mechanisms, including production of reactive oxygen species, cytokines, and proteases, which result in tissue damage leading to sclerosis and fibrosis (39,–41). Components of the diabetic milieu, including high glucose, AGEs, and oxidized LDL, promote macrophage accumulation via induction of chemokines and adhesion molecules, and macrophage activation within diabetic kidneys (39,–41). Our study shows that FXR deficiency increases macrophage infiltration in both glomeruli and tubulointerstitium, whereas treatment with the FXR agonist INT-747 reverses this course, both associated with regulation of NFκB activity. NFκB transcriptional activity controls the activation of a number of inflammatory genes. Consistent with our findings, the anti-inflammatory properties of INT-747 have been appreciated in other cell types (42,43). In vascular smooth muscle cells (42), the counterpart of glomerular mesangial cells, and in hepatocytes (43), INT-747 has been shown to inhibit the inflammatory response by antagonizing the NFκB signaling pathway. All together, this may represent a mechanism for FXR activation to inhibit inflammatory properties in macrophages and diabetic kidneys.

Our results show that proteinuria in diabetic mice is remarkably augmented by FXR deficiency, and significantly inhibited by FXR activation with INT-747. Damaging any component of the glomerular filtration barrier, fenestrated endothelium, intervening GBM, or slit diaphragms created by epithelial podocyte foot processes, results in proteinuria (16). Furthermore, mesangial cell injury can lead to foot process fusion and proteinuria (44). Tubules also play an important role in proteinuria by tubular reabsorption of filtered protein (16). FXR deficiency in diabetic mice showed the lytic mesangium with foamy cells, activated glomerular endothelium, and podocyte foot process effacement, indicating the involvement of multiple cell types in the pathogenesis of proteinuria in this model, as predicted by FXR expression in most kidney cell types. Indeed, the crosstalk of cytokines generated by mesangial cells, endothelial cells, and podocytes makes alterations in one cell type reverberate on others (44). Recently, FXR was found to directly enhance transcriptional activation of the endothelial nitric oxide synthase (eNOS) gene promoter, leading to increased nitric oxide production in vascular endothelial cells (45). This mechanism still remains to be ascertained in glomerular endothelial cells, but may represent a very important FXR-mediated effect, as eNOS deficiency results in endothelial dysfunction which favors development of diabetic nephropathy (46,47).

The diabetic FXR KO model does not recapitulate all typical features of human diabetic nephropathy, such as nodular mesangial lesions, although early lobulation was found in some glomeruli. This may relate to diabetes duration, resistant C57BL/6 background, or monogenic changes in gene expression intrinsic to this model, which makes it unable to reproduce full-fledged human diabetic nephropathy (48). However, the increased bile acid concentration in FXR KO mice (13) may lead to activation of the membrane bile acid receptor TGR5, a G-protein coupled receptor (49,50), which may exert compensatory actions in FXR deficiency (51). Establishing FXR deficiency in susceptible strains and exploring the role of TGR5 in diabetic nephropathy represent our future directions to clarify these issues and generate more predictive mouse models for human diabetic nephropathy.

In summary, our findings indicate a critical role for FXR in the development of diabetic nephropathy and show that FXR activation by the selective and potent FXR agonist INT-747 prevents nephropathy in type 1 diabetes by inhibiting diabetes-induced alterations in renal lipid metabolism, fibrosis, inflammation, and oxidative stress. This knowledge has translational potential into effective therapies for type 1 diabetic patients with diabetic renal complications.

This work was supported by grants from the National Institutes of Health (NIH) (U01 DK-076134 and R01AG-026529), Juvenile Diabetes Research Foundation, and the Veterans Affairs Merit Review. Y.C. was supported by a minority graduate student supplement from NIH and H.S. was supported by the LAB COATS Program at the University of Colorado Denver.

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

X.X.W., T.J., Y.S., Y.C., S.M.-A., H.S., P.S., and L.L. researched data. X.X.W. wrote the manuscript. F.J.G. contributed to the discussion. L.A., J.B.K., J.W.V., M.P., and M.L. reviewed/edited the manuscript.

The authors would like to thank Kelly Hudkins, NIH Mouse Metabolic Phenotyping Center, University of Washington in Seattle, for assistance with histology staining and electron microscopy.