Evidence for a Novel TGF-β1−Independent Mechanism of Fibronectin Production in Mesangial Cells Overexpressing Glucose Transporters

Oligonucleotides were synthesized by Life Technologies (Karlsruhe, Germany). Cell culture media, supplements, Ultroser, and FCS were obtained from Gibco (Eggenstein, Germany); minocyclin, from Pan Systems (Aidenbach, Germany); Klenow enzyme and poly[d(I-C)], from Boehringer (Mannheim, Germany); bisindolylmaleimide I, phorbol myristoyl acid (PMA), anisomycin, and curcumin, from Sigma (Munich, Germany); calphostin, from Calbiochem (Schwalbach, Germany); PKC-peptide inhibitor, from Promega (Madison, WI); [α-³²P]dATP, from Hartmann (Braunschweig, Germany); antibodies against c-Jun (1694X), JunB (73X), JunD (74X), c-Fos (253X), stimulating protein 1 (Sp-1; 59X), and PKC-α (C-20) and -β1 (C-16), from Santa Cruz Technologies (Santa Cruz, CA); antibodies against phospho-p42/p44 MAPK thr-202/tyr-204, p42/p44 MAPK, phospho-activating transcription factor-2 thr-71, and phospho-p38 MAPK thr-180/tyr-182, from New England Biolabs (Schwalbach, Germany); the neutralizing TGF-β antibody (against TGF-β1, -β2, and -β5), from R&D Systems (Minneapolis, MN); the antibody against fibronectin, from Biomol (Hamburg, Germany); and the RL-2 antibody, from Dianova (Hamburg, Germany). The antibody against glutamine:fructose-6-phosphate amidotransferase (GFAT) has been previously described (39). The fluorescent probe CM-H₂DCFDA was obtained from Molecular Probes (Eugene, OR).

Rat mesangial cells stably transfected with the human GLUT1 transporter gene (GT1 cells) or bacterial β-galactosidase (as controls; LacZ cells) were cultured as previously described (18). For experimental purposes, RPMI 1640 contained 8 mmol/l glucose (corresponding to normal glucose levels in rats), 2% Ultroser, 2 mmol/l glutamine, 25 mmol/l HEPES, 100 units/ml penicillin, and 100 μg/ml streptomycin. For determination of TGF-β1 and fibronectin concentration in supernatants, medium without Ultroser or FCS was used.

In supernatants, glucose and lactate were measured with Ektachem Vitros systems (Ortho-clinical-diagnostics, Neckargmünd, Germany). DNA in total cell extracts was measured by fluorometry with bisbenzimidazol (24).

RNA was prepared from LacZ and GT1 cells with the RNeasy Kit (Qiagen, Hilden, Germany). Total RNA (25 μg) was separated on a formaldehyde-containing agarose gel and transferred to a nylon membrane. Preparation and labeling of RNA probes for TGF-β1 and glyceraldehyde-3-phosphate dehydrogenase and hybridizations were performed as previously described (40).

For quantification of the proteins, cell culture supernatants were collected after the indicated time points and stored at −80°C. Determination of total TGF-β1 protein was performed with the Quantikine Immunoassay (R&D Systems, Minneapolis, MN). Fibronectin was measured by the Quantimatrix human fibronectin enzyme-linked immunosorbent assay kit (ELISA) from Chemicon International (Temecula, CA).

For Western blotting studies, 6.0 × 10⁶ cells were seeded onto 15-cm culture dishes with 15 ml experimental medium and harvested after 12, 24, or 48 h. Nuclear proteins were prepared as recently described (41). Preparation of total, cytosolic, and membrane fraction and detection of PKC isoforms was recently described (40). Western blotting was performed as previously described (29).

The synthetic oligonucleotides phorbol myristic acid responsive element:5′-GATCTGTGACTCAGCGCGAG-3′ and Sp-1:5′-GATCTCGGGGAAAGGGGGCGGGCTCCCCAG-3′ were end-labeled with [α-³²P]dATP (3,000 Ci · mmol⁻¹ · l⁻¹) and Klenow enzyme and were incubated with up to 5 μg nuclear protein in 20 μl of 7 mmol/l Hepes-KOH (pH 7.9), 100 mmol/l KCl, 3.6 mmol/l MgCl₂, and 10% glycerol on ice for 20 min. Next 0.05 mg/ml poly[d(I-C)] was added as an unspecific competitor. The samples were run on a 5% nondenaturating polyacrylamide gel in a buffer containing 25 mmol/l Tris-HCl (pH 8.0), 190 mmol/l glycine, and 1 mmol/l EDTA. Gels were dried and analyzed by autoradiography.

Cell extracts were prepared and determination of uridine-5′-diphosphate-N-acetylglucosamine (UDP-GlcNAc) by capillary electrophoresis was performed as previously described (42).

Cells were cultured with 8 or 30 mmol/l glucose in experimental medium for 48 h, loaded with 10 μmol/l CM-H₂DCFDA, incubated for 30 min at 37°C, and analyzed by fluorescence-activated cell sorting using FACScalibur (Becton-Dickinson, San Jose, CA).

Results presented are derived from at least three independent experiments. Means ± SE were calculated, and groups of data were compared using Student’s t test. Statistical significance was set at P < 0.05.

The increased flux through glycolysis in the GT1 cells was confirmed by a 3.2-fold increase in glucose consumption and a 3.6-fold higher lactate production after 72 h compared to LacZ cells, thereby indicating that glucose was almost completely (99%) metabolized to lactate. Furthermore, through Western blotting, we obtained a time-dependent accumulation of fibronectin protein in the supernatant of the GT1 cells, but observed only negligible fibronectin production in LacZ cells (Fig. 1). The results indicated that our cell culture conditions were suitable for investigating the mechanism of fibronectin induction by overexpression of GLUT1 in mesangial cells.

To study the hypothesis that the increased fibronectin production in GT1 cells is mediated by an enhanced expression of TGF-β1, the TGF-β1 mRNA levels in GT1 and LacZ cells were determined by Northern blotting. As shown in Fig. 2A, over a time course of 96 h, the amount of TGF-β1 mRNA was comparable in GT1 and LacZ cells, with similar increases after 48, 72, and 96 h. Because the secretion of the TGF-β1 protein is necessary for autocrine/paracrine bioactivity, the concentration of total TGF-β1 protein in the supernatant of the cells was measured by ELISA. We found no increased TGF-β1 protein production in GT1 cells (Fig. 2B), a finding that supports the observed TGF-β1 mRNA levels (Fig. 2A). In fact, the TGF-β1 protein production of GT1 cells appeared to be lower. The same result was obtained when TGF-β1 was related to cellular DNA (data not shown). These results indicated that overexpression of GLUT1, which leads to increased flux through glycolysis, does not induce TGF-β1 overexpression.

To test whether increased TGF-β1 bioactivity might account for the increased fibronectin synthesis, we studied the fibronectin production in the presence of neutralizing TGF-β antibodies. As shown in Fig. 2C, inhibition of TGF-β1 bioactivity did not prevent the increase in fibronectin protein in the supernatant of GT1 cells compared to LacZ cells, indicating that the increased fibronectin production was mediated by an TGF-β1−independent pathway.

Because the rat fibronectin promoter contains a regulatory active AP-1 binding site (43), and the expression and activity of several AP-1 proteins were increased by high glucose, we studied the nuclear concentration of the AP-1 proteins c-Fos and c-Jun, JunB, and JunD in GT1 and LacZ cells by Western blotting. The amount of all three Jun proteins was clearly increased in GT1 cells compared to LacZ cells after 12, 24, and 48 h (Fig. 3A– C), except that no difference could be detected after 48 h in JunB proteins (Fig. 3B). The nuclear content of all three Jun proteins was lower in the GT1 cells cultured for 48 h compared to those cultured for 24 h. We found no increase of c-Fos in the nuclear extracts of GT1 cells after 12, 24, and 48 h compared to control cells (Fig. 3D). The 62-kDa band that was detected by the anti−c-Fos antibody was verified as c-Fos when after cells were stimulated with PMA, a strong activator of c-Fos expression, the 62 kDa band was clearly enhanced. Together, these data demonstrated that increases of glucose transport and metabolism by overexpression of GLUT1 in mesangial cells led to a strong elevation of the nuclear level of Jun proteins while the amount of c-Fos remained unchanged.

The increased amounts of Jun proteins in the nuclear extracts of GT1 cells resulted in enhanced DNA binding to a PMA responsive element (TRE) containing consensus AP-1 binding site. Through electrophoretic mobility shift assays (EMSAs), we found a strongly time-dependent increase in DNA-binding activity of nuclear proteins from GT1 cells compared to LacZ cells (Fig. 3E). Although the difference between GT1 and LacZ cells was excessive at 12 and 24 h, the difference was less pronounced after 48 h, a finding that supports the lower nuclear amount of Jun proteins at this time point (Fig. 3A– C). Using specific antibodies against AP-1 proteins, we characterized the DNA-binding protein complex in GT1 cells as a heterodimer consisting of c-Fos and the Jun proteins c-Jun, Jun B, and Jun D (Fig. 3F).

The activation of AP-1 in the GT1 cells was specific, as we found no enhanced nuclear content or DNA-binding activity of other transcription factors that have been linked to high glucose−induced promoter activation, such as Sp-1 (25,31) or cAMP-responsive element binding protein (CREB)/activating transcription factor (ATF) 1 (34). Determination of Sp-1 in the same nuclear extracts that were used for studies on AP-1 revealed changes in neither the amount of Sp-1 protein (Fig. 4A) nor in DNA-binding activity to the Sp-1 site (Fig. 4B). Specificity of Sp-1 DNA binding was determined by mutated Sp-1 site and supershift analysis (data not shown). Similarly, we found no changes in the nuclear amounts of the AP-1−related proteins ATF-1, ATF-2, or CREB by Western blotting or the binding of nuclear proteins to a cAMP-responsive element (CRE), which contains the consensus binding sequence for ATF and CREB proteins (data not shown). These data indicated the specificity of the AP-1 activation in GT1 cells.

Our studies revealed that TGF-β1 is not responsible for the enhanced expression of fibronectin in the GT1 cells. Furthermore, we observed a strong activation of AP-1. Therefore, we tested whether the enhanced expression and DNA-binding activity of AP-1 would mediate the accumulation of fibronectin in the supernatant of the GT1 cells by the addition of curcumin. Curcumin is known to specifically inhibit the activation and DNA binding of AP-1 (44,45). The strong increase of fibronectin produced by the GT1 cells compared to LacZ cells after 48 h (41.9 ± 7.1 vs. 5.4 ± 1.1 ng/μg DNA) was reduced by curcumin in a dosage-dependent manner, whereas LacZ cells were not affected (Fig. 5). At a concentration of 20 μmol/l curcumin, the fibronectin production was decreased by 66% compared to GT1 cells cultured without the AP-1 inhibitor. These data indicated that activation of AP-1 is necessary for the enhanced expression of fibronectin observed in the GT1 cells.

Because it has been reported that the hyperglycemia-induced activation of PKC isoforms is involved in the increased production of fibronectin, we studied the effect of two inhibitors of PKC, calphostin and a peptide inhibitor specific for PKC isoforms α and β, on the induction of AP-1 and the fibronectin production in the GT1 cells. The addition of 1,000 nmol/l calphostin or 10 μmol/l peptide inhibitor to the GT1 cells for 24 h reduced the nuclear amounts of c-Jun and JunD (Fig. 6A) and consequently the DNA-binding activity of the nuclear extracts to the AP-1 consensus sequence (Fig. 6B). Furthermore, the fivefold increased production of fibronectin in GT1 cells was reduced to the level of LacZ cells (Fig. 6C). Treatment of the cells with 100 nmol/l calphostin showed a minor reduction of c-Jun and JunD protein content, DNA-binding activity of an AP-1 complex, and fibronectin production of the GT1 cells. In LacZ cells, the addition of calphostin or the peptide inhibitor also decreased the amount of Jun proteins and the DNA-binding activity of AP-1, but had no effect on fibronectin production (Fig. 6A–C). The effect on the nuclear amount of JunB was not tested, as the JunB protein content in nuclear extracts was very low and the increase in GT1 cells was much less pronounced when compared to c-Jun or JunD.

Because the peptide inhibitor is specific for PKC isoforms α and β, we studied the amount and translocation of PKC-α and -β1 (PKC-β2 was not expressed in GT1 and LacZ cells; data not shown) by Western blotting. PKC-α and -β1 were increased in the total, cytosolic, and membrane extracts of the GT1 cells compared to LacZ cells (Fig. 6D). The enhanced translocation of PKC-α and -β1 indicated a participation of these PKC isoforms in AP-1 activation and subsequent enhanced fibronectin production in the GT1 cells.

In addition to the PKC-dependent pathways, the hexosamine pathway and the ERK-1 and -2, and p38 MAPK pathways were activated by ambient high glucose and were linked to the development of late diabetic complications, particularly to the high glucose−enhanced expression of TGF-β1 in mesangial cells. Therefore, we asked whether a missing activation of one or more of these pathways in the GT1 cells might explain the missing increase in TGF-β1 synthesis.

We studied the flux through the hexosamine pathway by measuring its end product UDP-GlcNAc, and by determining the endogenous levels of O-GlcNAc−modified nuclear proteins and the cellular levels of the key enzyme GFAT by Western blotting (Fig. 7). Assessment of intracellular UDP-GlcNAc by capillary electrophoresis revealed that the concentration of the hexosamine pathway end product was unchanged or even reduced in the GT1 cells compared to LacZ cells after 24, 48, 72, and 96 h when related to the DNA content (Fig. 7A) or cell number (data not shown). This observation was supported by the reduced amount of O-GlcNAc−modified proteins in the nuclear extracts of the GT1 cells after 24 and 48 h (Fig. 7B). Furthermore, we found reduced endogenous levels of GFAT protein in the cytosolic extracts of GT1 cells harvested after 24 and 48 h compared to LacZ cells (Fig. 7C), which may have contributed to the low activity of the hexosamine pathway in GT1 cells.

We then studied the possible activation of the p38 MAPK and ERK-1 and -2 pathways in GT1 cells by using specific phosphopeptide antibodies. The intensities of the bands detected by the phosphospecific antibodies against p38 MAPK and ATF-2 were not different in LacZ and GT1 cells (Fig. 8A). Stimulation of the cells with anisomycin, an activator of the p38 MAPK pathway, resulted in a strongly increased intensity of the respective bands. Furthermore, we studied the phosphorylation of the MAPK ERK-1 and -2, which contribute to the activation of this MAPK pathway. No enhanced levels of phosphorylated ERK-1 or -2 could be detected in GT1 cells compared to LacZ cells, whereas stimulation of the cells with epidermal growth factor (EGF) resulted in increased intensity of the bands, detected by the phosphospecific antibodies against ERK-1 and -2 (Fig. 8B).

ROS as signaling molecules have been implicated in the activation of the hexosamine and p38 MAPK pathways. Therefore we studied the production of ROS in the GT1 and LacZ cells by fluorescence activated cell sorting. After a 48-h cell culture with ambient 8 mmol/l glucose, no increased fluorescence intensity signal could be detected in the GT1 cells compared to LacZ cells, indicating no increased generation of ROS by overexpressing GLUT1 (Fig. 8C). As expected, exposure of LacZ cells to 30 mmol/l glucose for 48 h resulted in a 1.6-fold increase in fluorescence signal compared to exposure to 8 mmol/l glucose, indicating that oxidative stress is caused by ambient high glucose. In the GT1 cells, ambient 30 mmol/l glucose failed to generate a fluorescence signal above that of the GT1 cells cultured in 8 mmol/l glucose.

This is the first report demonstrating that in mesangial cells, excessive transport of glucose by overexpressing GLUT1 in the absence of ambient high glucose activates synthesis of fibronectin independent of TGF-β1. Three distinct lines of evidence support this conclusion: 1) TGF-β1 mRNA levels were unchanged in mesangial cells overexpressing GLUT1; 2) the concentration of the secreted TGF-β1 protein in the supernatant of the GT1 cells was not increased compared to control cells; and 3) the addition of neutralizing antibodies against bioactive TGF-β1 did not reduce the strongly increased concentration of fibronectin in the supernatant of the GT1 cells. Moreover, our results indicated that fibronectin production in the GT1 cells was mediated through activation of PKC and AP-1. Overexpression of GLUT1 increases the expression and content in the membrane fraction of PKC isoforms α (46) and β1, indicating an activation of these isoforms. Inhibition of PKC with two different inhibitors, one of them specific for isoforms α and β1, blocked the excessive activation of AP1 in the GT1 cells. This activation was demonstrated by the strong increase of c-Jun, JunD, and JunB proteins, subsequently leading to increased DNA-binding activity of the AP-1 complex and to enhanced fibronectin production. The finding that fibronectin accumulation is prevented by inhibition of PKC provides very clear evidence for the following sequential signaling pathway: excessive glucose uptake and metabolism → activation of PKC → activation of AP-1 → fibronectin accumulation.

The second main observation from this study is the missing upregulation of TGF-β1, despite the increases in glucose uptake and metabolism and the activation of PKC and AP-1. We could clearly demonstrate that the participation of other high glucose−inducible pathways is necessary for the induction of TGF-β1 expression by ambient high glucose as well as by increased glucose uptake. Specifically, the activation of the hexosamine (24,25) and the p38 (28) and ERK-1 and -2 MAPK pathways (26,27) by high glucose, which has been demonstrated by several groups in vivo and in vitro and has been linked to the induction of TGF-β1 expression (24,27,29), is completely missing in the GT1 cells. In addition, after comparing the activation of AP-1 proteins in the GT1 and mesangial cells exposed to ambient high glucose, we found no increased nuclear levels of c-Fos, whereas the expression of c-Fos is induced by high extracellular glucose conditions (29,37) and is implicated in enhanced TGF-β1 gene activation (29). Therefore, our results strongly suggest that overexpression of GLUT1 in mesangial cells exposed to physiologically occurring glucose concentrations and exposure of mesangial cells to high extracellular glucose do not induce the same biochemical pathways and that activation of classical PKC and Jun proteins alone is not sufficient to induce TGF-β1 synthesis.

This observation is supported by the lack of increased ROS production in GT1 cells under normal and high glucose conditions, compared to the generation of ROS in control cells under high glucose conditions. Apparently, the chronically excessive glucose uptake in GT1 cells did not enhance the flux of glucose metabolites into the electron transport chain, as almost all glucose was metabolized to lactate. This prevented increased ROS production in the mitochondria, a result that was recently linked to the development of diabetic late complications in a study of endothelial cells (30,31). The absence of oxidative stress can explain the missing upregulation of the redox-sensitive hexosamine and the p38 MAPK pathways in the GT1 cells. Moreover, this finding uncovers a glucose-dependent, oxidative stress−independent mechanism of fibronectin production in mesangial cells.

Our findings in GLUT1-overexpressing mesangial cells are relevant for the in vivo situation in two ways: first, in contrast to other cells (e.g., endothelial cells) (47), mesangial cells respond to high glucose concentrations with increased GLUT1 levels (12), a finding also seen in mesangial cells obtained from diabetic patients (16). Second, they reflect the activated signaling pathways in situations where blood glucose concentration are normal or near normal, but glucose uptake is still increased because of diabetes-induced cytokine levels (e.g., ANG, insulin-like growth factor 1, pseudohypoxia).

Various in vitro and in vivo studies have shown that inhibition of TGF-β1 by, for example, treatment with monoclonal anti−TGF-β1 antibody (22) prevents the increase in glomerular matrix proteins, indicating that TGF-β1 is a major causative factor in diabetic nephropathy. Therefore, it may well be that blockade of TGF-β leads to a reduction of other, TGF-β1−induced events occurring in diabetic nephropathy, such as cellular hyperplasia or expression of other matrix proteins, and possibly leads to the inhibition of hyperglycemia- (12,13) or ANG-II−induced (14) GLUT1 overproduction, thus interrupting the previously proposed TGF-β1−GLUT1 axis (11). Our finding that the excessively increased flux through glycolysis stimulates the production of extracellular matrix proteins via AP-1 activation may be relevant for other pathological situations with massive overexpression of GLUT1, such as in tumor cells (48) or hypoxic tissues (49).

In summary, the mesangial cells overexpressing GLUT1 may serve as a good model for distinguishing between the effects of ambient high glucose and increased glucose uptake and subsequent enhanced metabolism in physiological glucose concentrations. This might be important because both acute hyperglycemia and normal or mild glycemia with increased GLUT1 caused by, for example, TGF-β1, ANG-II, hypoxia, or pseudohypoxia, may be present in diabetic patients (50). Therefore, both, TGF-β1−dependent as well as TGF-β1−independent pathways may contribute to mesangial matrix accumulation, one major hallmark in the development of diabetic nephropathy. Furthermore, from this and other studies it appears that the efficiency of inhibitors of tissue fibrosis needs to be evaluated on both pathways.

This study was supported by a grant from the Deutsche Forschungsgemeinschaft (Schl 239/6 to E.S.) and grants from the National Institutes of Health (DK-54507), American Diabetes Association, American Heart Association, and National Kidney Foundation of Maryland to C.W.H.