Altered Gene Expression Related to Glomerulogenesis and Podocyte Structure in Early Diabetic Nephropathy of db/db Mice and Its Restoration by Pioglitazone

Male diabetic db/db mice and their nondiabetic db/m littermates were used for this study. All mice were purchased from CLEA Japan (Tokyo). These db/db mice began to show hyperglycemia at 5 weeks of age and a significant increase in urinary albumin excretion at 7 weeks of age (Table 1). Mice were killed under pentobarbital anesthesia at 5 and 7 weeks of age to obtain kidney samples for isolation of glomeruli and immunohistochemistry.

To study the role of insulin resistance in the development of diabetic nephropathy, we administered pioglitazone (Takeda Pharmaceutical, Osaka, Japan), a peroxisome proliferator–activated receptor-γ agonist, to two other groups of 5-week-old db/db mice for 2 weeks (n = 12 in each). Pioglitazone was mixed with normal mouse chow and administered at a dose of 3 or 15 mg · kg body wt⁻¹ · day⁻¹ because 15 mg/kg of pioglitazone was reported to improve insulin sensitivity in db/db mice (16).

We obtained 16-h urine specimens from all mice at 5 and 7 weeks of age for the measurement of albumin excretion (17). Urinary albumin excretion was determined by enzyme-linked immunosorbent assay (Albuwell; Exocell, Philadelphia, PA) (17). Urinary creatinine levels were measured by enzymatic method (SRL, Tokyo) (17). For the insulin tolerance test, mice were fasted for 6 h and given 1.25 unit/kg i.p. human regular insulin (Novo Nordisk, Bagsvaerd, Denmark) (18).

We prepared two isolated glomerular samples from each group. An isolated glomerular sample was obtained from the kidneys of six mice by differential sieving method, using mesh diameters of 45, 75, and 150 μm (19). The purity of each sample was confirmed by microscopy. The glomerular samples were ∼80% pure on average, and there was no difference in purity among the samples.

Total RNA was extracted from glomerular samples by the acid guanidine-phenol-chloroform method, using Trizol reagent (Life Technologies) (20). We essentially followed the procedures described in detail in the GeneChip expression analysis manual (Affymetrix, Santa Clara, CA). In brief, 10 μg of total RNA was used for cDNA synthesis (Superscript II kit; Life Technologies, Rockville, MD). Biotin-labeled cRNA was produced through in vitro transcription of cDNA, using an ENZO BioArray high-yield RNA transcript labeling kit (Affymetrix). Fragmented cRNA (15 μg) was hybridized to an Affymetrix Murine Genome U74Av2 GeneChip at 45°C for 16 h. The samples were stained and washed according to the manufacturer’s protocol on a Fluidics Station 400 (Affymetrix) and scanned on a GeneArray scanner (Affymetrix) (8,21).

Primary data extraction was performed with Microarray Suite 5.0 (Affymetrix) because analysis by Microarray Suite 5.0 is more reliable than other methods (22). Microarray Suite 5.0 software normalized the data of each microarray and compared the expression between the two different arrays. Moreover, the software could determine statistically whether each gene was present (reliably detected) or absent (not detected) in one array and whether each gene increased or decreased between two different arrays. Signal normalization across samples was carried out, using all probe sets, with a mean expression value of 500 (8,21). To allow comparisons between any two experiments, pairwise comparisons were made between db/m and db/db mice by Microarray Suite 5.0. Because two arrays were used for each group (db/m 1, db/m 2, db/db 1, and db/db 2), we performed four comparison analyses (i.e., db/m 1 vs. db/db 1, db/m 1 vs. db/db 2, db/m 2 vs. db/db 1, and db/m 2 vs. db/db 2). Genes showing an increased or decreased call in at least three of four comparisons were defined as genes showing a significant change. As an internal control, we chose GAPDH and confirmed that there was no difference in GAPDH expression level between each sample in microarray analysis.

Cultivation of conditionally immortalized mouse podocytes (a gift from Dr. Peter Mundel, Albert Einstein College of Medicine, Bronx, NY) was performed as reported previously (23). Briefly, cells were grown on a type 1 collagen–coated dish (IPC-03; Koken, Tokyo) at 33°C in the presence of 10 units/ml murine γ-interferon (Life Technologies, Gaithersburg, MD) in RPMI 1640 medium (Nihonseiyaku, Tokyo) supplemented with 10% FCS (Cansera International, Etobicoke, ON, Canada) and antibiotics. To induce differentiation, podocytes were maintained at 37°C without interferon. Before the experiment, cells were differentiated for 2 weeks without passage, followed by culture in RPMI 1640 containing 1% FCS supplemented with 5.6 mmol/l glucose (normal glucose) or 25 mmol/l glucose (high glucose) for 14 days.

We reverse transcribed 2.5 μg of total RNA using Ready-To-Go (Amersham Pharmacia Biotech, Piscataway, NJ) (20). TaqMan real-time quantitative PCR was performed and analyzed according to the manufacturer’s instructions (Applied Biosystems, Foster City, CA) (24). Primers and probe sequences were selected by using Primer Express (Applied Biosystems). GAPDH was used for internal control because its expression level did not show a significant difference between db/m and db/db in our microarray analysis.

Kidneys were dissected immediately and fixed in 10% formalin, embedded in paraffin, and sectioned at 4 μm. Hypoxia-inducible factor-1α (HIF-1α) was identified with monoclonal IgG HIF-1α antibody 67 (Novus Biological, Littleton, CO) at a 1:2000 dilution, using the Tyramide signal amplification system (PerkinElmer Life Sciences, Boston, MA) (25). The number of HIF-1α–positive cells was counted in 30 randomly selected glomeruli in the outer cortex.

Data are the means ± SE. Statistical analyses were performed using ANOVA followed by Scheffe’s test. P < 0.05 was considered statistically significant.

At 5 weeks of age, db/db mice already showed significant hyperglycemia, whereas urinary albumin excretion did not increase compared with db/m mice (Table 1). There was no difference in renal histology between db/m and db/db mice at 5 weeks of age under light microscopic observation. At 7 weeks of age, db/db mice showed significant elevation of urinary albumin excretion (Table 1). However, no apparent histological difference was observed between db/db and db/m mice (data not shown). Thus, 5-week-old db/db mice exhibited features similar to the human normoalbuminuric stage, and 7-week-old mice showed features similar to the human early stage of diabetic nephropathy.

Table 2 shows 134 genes with an absolute relative log ratio >0.5 and showing significant change by Microarray Suite 5.0 analysis at 5 and/or 7 weeks of age. At 5 weeks of age, 105 genes were differentially expressed (65 increased, 40 decreased) between db/db and db/m mouse glomeruli. Among them, there were genes related to oxidative stress, glucose metabolism, lipid metabolism, cell growth, fibrosis, apoptosis, vasoactive mediators, calcium-binding proteins, coagulation, cell structure, and extracellular matrix components. We also observed significant differences in the expression of development-related genes, including several genes related to kidney development. At 7 weeks of age, 116 genes expressed differentially (72 increased, 44 decreased) between db/db and db/m mouse glomeruli. In addition to the genes showing differential expression at 5 weeks of age, genes related to cell structure, particularly podocyte structure, and solute carrier family were expressed differentially between db/db and db/m.

We next confirmed the differential expression of genes by quantitative real-time RT-PCR. Because glomerular response to injury is accompanied by activation of the development-related genes (26), we measured mRNA levels of kidney development–related genes, i.e., ephrin B2 (27), Pod-1 (28), glomerular epithelial protein 1 (GLEPP1) (29), and HIF-1α (30), in isolated glomeruli. These four genes are reported to play important roles in glomerulogenesis (27–30). Real-time RT-PCR confirmed significant mRNA elevation of ephrin B2 at 5 and 7 weeks of age (2.2- and 2.9-fold of control at 5 and 7 weeks of age, respectively) (Fig. 1A). Although the upregulation of HIF-1α was not significant by cDNA microarray analysis at 5 weeks of age, real-time RT-PCR revealed significant upregulation of HIF-1α mRNA at both 5 and 7 weeks of age (2.1- and 2.9-fold of control at 5 and 7 weeks of age, respectively) (Fig. 1A). Because HIF-1α is a transcription factor and it is more important to evaluate the expression of HIF-1α protein, we also examined HIF-1α protein expression by immunohistochemistry. HIF-1α protein expression was significantly increased in glomeruli at 7 weeks of age (Fig. 4A, B, and D). GLEPP1 and Pod-1 were significantly upregulated only at 5 weeks of age, and Pod-1 was significantly downregulated at 7 weeks (GLEPP1: 2.3- and 1.1-fold of control at 5 and 7 weeks of age; Pod-1: 4.0- and 0.5-fold, respectively) (Fig. 1A).

We also confirm the differential expression of podocyte structure–related genes, actinin 4α (31) and dystroglycan 1 (DG1) (32), because morphologic changes of podocytes and podocyte injury play key roles in the development of diabetic nephropathy (3–6). Quantitative RT-PCR revealed significant upregulation of actinin 4α and DG1 in isolated glomeruli of 7- but not of 5-week-old db/db mice compared with those of control (actinin 4α: 3.4-fold; DG1: 2.9-fold) (Fig. 1B).

To examine whether the differential expression of kidney development–related genes and podocyte structure–related genes in isolated glomeruli of db/db mice could reflect the alteration in gene expression of podocytes, we next examined the expression of these genes in cultured podocytes by quantitative RT-PCR. Although Pod-1 mRNA expression was not detectable, ephrin B2, HIF-1α, GLEPP1, actinin 4α, and DG1 mRNA were detectable in cultured podocytes under normal glucose conditions. Ephrin B2 and HIF-1α mRNA expression was significantly upregulated under high-glucose conditions (5.7- and 2.3-fold of control, respectively) (Fig. 2A). GLEPP1 mRNA expression also tended to increase (1.6-fold of control) (Fig. 2B). By contrast, actinin 4α and DG1 mRNA did not show a significant change (Fig. 2B).

Insulin resistance is one of the important pathogenic factors for the diabetic nephropathy in type 2 diabetes. Indeed, improvement of insulin resistance by thiazolidinediones resulted in the reduction of albuminuria in diabetic nephropathy (33). Therefore, we examined the effects of pioglitazone on the gene expression profiles in db/db mice. Although administration of pioglitazone at a dose of 3 mg/kg did not affect hyperglycemia, insulin sensitivity, or albuminuria, pioglitazone at a dose of 15 mg/kg significantly reduced but did not normalize the blood glucose level, improved insulin sensitivity, and completely normalized urinary albumin excretion (Tables 3 and 4).

We first examined the effect of pioglitazone using microarray analysis. Table 2 shows genes in db/db mice whose differential expression was restored by pioglitazone treatment. Although the lower dose of pioglitazone (3 mg/kg) restored only a small number of genes (4 of 116), pioglitazone at the higher dose restored the alteration in more than two-thirds of the genes (81 of 116) in db/db mice. Pioglitazone restored most of the genes related to oxidative stress (16 of 17), lipid metabolism (11 of 14), glucose metabolism (4 of 8), development (5 of 7), cell growth (6 of 9), vasoactive mediator (6 of 8), and coagulation (3 of 5). Among these genes, the recovery of oxidative stress–, glucose metabolism–, and lipid metabolism–related gene expression by pioglitazone was compatible with previous reports (34,35). By contrast, only a small number of the genes related to fibrosis (0 of 2), apoptosis (1 of 3), calcium-binding protein (2 of 4), cell structure (1 of 5), and extracellular matrix component (0 of 3) were restored by pioglitazone.

We also examined mRNA expression of kidney development–and podocyte structure–related genes by quantitative RT-PCR. Upregulation of ephrin B2 and downregulation of Pod-1 were blunted by pioglitazone treatment (Fig. 3A). Although suppression of HIF-1α mRNA expression was not observed in microarray analysis and RT-PCR, HIF-1α protein expression was partially attenuated by pioglitazone (Fig. 4). Upregulation of actinin 4α and DG1 genes was significantly attenuated by pioglitazone treatment (Fig. 3B).

Although glomerular injury plays a central role in the development of diabetic nephropathy, most reports using microarray analysis have focused on the gene expression profile of the whole kidney in diabetic animals with nephropathy. In the current study, we examined a gene expression profile of isolated glomeruli from db/db mice, a well-known type 2 diabetes model. To the best of our knowledge, this is the first report on the glomerular gene expression profile in type 2 diabetes models. Our microarray data showed differential expression of genes related to glucose and lipid metabolism, oxidative stress, vasoactive mediators, cell growth, and coagulation in isolated glomeruli between db/db and db/m mice at the normoalbuminuric stage. These results are compatible with previous reports (1–3).

Our first new finding is that the kidney development–related genes were already differentially expressed at the normoalbuminuric stage in the glomeruli of db/db mice. Although the pathophysiological significance of the kidney development–related genes we examined is not fully clarified in diabetic nephropathy, ephrin B2, HIF 1α, Pod-1, and GLEPP1 participate in various stages and aspects of glomerulogenesis and are relevant to some types of glomerular injury, as previously suggested (26). Ephrin B2 is a transmembrane ligand of the ephrin B2 receptor (Eph) and its signaling pathway is required for vascular morphogenesis (36,37) and glomerular microvascular assembly (27). HIF-1α is critical for renal vasculogenesis and glomerulogenesis (30), and nuclear localization of HIF-1α increases in murine adriamycin nephrosis (38). Nyengaad and Rasch (39) reported an increase in glomerular capillary size and number in diabetic nephropathy, suggesting that angiogenesis is associated with glomerular injury. Actually, vascular endothelial growth factor (VEGF) plays a key role in the development of proteinuria and glomerular sclerosis in diabetic nephropathy (14). Although VEGF mRNA were not elevated at the normoalbuminuric stage in this study (data not shown), ephrin B2 and HIF-1α mRNA were already upregulated at the normoalbuminuric stage and remained elevated at an early stage of diabetic nephropathy in isolated glomeruli of diabetic mice (Fig. 1). Because ephrin B2 and HIF-1α relate to angiogenesis, and because HIF-1α induces VEGF (40), elevation of ephrin B2 and HIF-1α may be an important early step for glomerular angiogenic change in diabetic nephropathy. GLEPP1 is related to podocyte differentiation (29), and its expression decreases in dedifferentiated podocytes (41). Pod-1 is one of the transcriptional factors important for glomerulogenesis and podocyte differentiation (24,28). Both ephrin B2 and HIF-1α are abundantly expressed in glomerular podocytes in the developing kidney (27,30). Taken together, podocyte injury may play a pivotal role in diabetic glomerulopathy, including glomerular angiogenic change. Other kidney development–related molecules (e.g., gremlin and transforming growth factor-β) were also suggested to contribute to the pathogenesis of diabetic nephropathy (42,43). Thus, the current study raises the possibility that the alteration of the kidney development–related molecules, particularly glomerulogenesis-related molecules (ephrin B2, HIF-1α, GLEPP1, and Pod-1), is a key mediator for diabetic glomerulopathy. This possibility is strengthened by our finding that high glucose induced a similar pattern of changes in glomerulogenesis-related gene expression in cultured murine podocytes because hyperglycemia is a well-known determinant of diabetic nephropathy.

Another new finding in the current study is that extracellular matrix and cell structure–related genes were differentially expressed at an early stage of diabetic nephropathy. This is consistent with the development of mesangial expansion several weeks later in this model. Among these genes, we focused on genes playing important roles in podocyte structure, i.e., actinin 4α and DG1. Actinin 4α is an actin–cross-linking protein, and mice with mutant actinin 4α revealed foot process fusion and podocyte vacuolization (31). DG1, a heavily glycosylated peripheral membrane protein located in podocytes, is thought to keep foot process shape, and it decreases in proteinuric renal diseases (32,44). Thus, the current study suggests that podocyte structure and function may already alter at an early stage of nephropathy. In contrast to previous reports, actinin 4α and DG1 mRNA expression increased in this study. Induction of these genes might reflect the glomerular repairing process, as reported in a puromycin aminonucleoside nephrosis model (45).

Insulin resistance is a major feature of type 2 diabetes, and it precedes the onset of microalbuminuria. Greater degrees of insulin resistance are evident when urinary albumin excretion is elevated in type 2 diabetes (15), and hyperinsulinemia in the pre-diabetic state may contribute to microalbuminuria in type 2 diabetes (1). In our microarray analysis, alteration in most of the development-related gene expression was restored by pioglitazone treatment with amelioration of albuminuria and hyperglycemia. Among them, the restoration of ephrin B2 and Pod-1 were confirmed by RT-PCR, and nuclear localization of HIF-1α was attenuated by pioglitazone. Although we could not evaluate the effect of pioglitazone on GLEPP1 gene expression because of its transient upregulation in this study, these results suggest that insulin resistance might be important in inducing the alteration in the expression of kidney development–related genes, including glomerulogenesis-related genes at early stages of nephropathy. Similarly, insulin resistance might also induce phenotype alteration of podocytes at an early stage of nephropathy because the upregulation of DG1 and actinin 4α genes was attenuated by pioglitazone treatment. We could not rule out the possibility that hyperglycemia per se directly altered glomerulogenesis-related gene expression because high glucose stimulated expression of glomerulogenesis-related genes in cultured podocytes and because administration of pioglitazone improved insulin resistance as well as hyperglycemia.

In conclusion, we demonstrated that the differential expression of glomerulogenesis-related genes already took place at the normoalbuminuric stage in the isolated glomeruli from db/db mice, whereas the expression of podocyte structure–related genes were altered at an early nephropathy stage with the elevation of microalbuminuria. We also showed that pioglitazone treatment restored most of the differential expression of glomerulogenesis- and podocyte structure–related genes. These findings suggest that the alteration of these genes might be relevant to the pathogenesis of diabetic glomerulopathy in type 2 diabetes with insulin resistance. Pioglitazone treatment even at the normoalbuminuric stage might be useful for the prevention of diabetic nephropathy.

Support for this study was provided in part by the Promotion of Fundamental Studies in Health Science of the Organization for Pharmaceutical Safety and Research of Japan and Grant-in-Aid for Scientific Research 14571044 from the Japan Society for the Promotion of the Science.

We thank Dr. Peter Mundel (Albert Einstein College of Medicine) for providing the mouse podocyte cell line MPC5. We also thank Takeda Pharmaceutical for donating pioglitazone and Dr. Richard J. Johnson (Baylor College of Medicine) for assistance in preparing this manuscript.