Flux Through the Hexosamine Pathway Is a Determinant of Nuclear Factor κB– Dependent Promoter Activation

Dulbecco’s modified Eagle’s medium (DMEM), fetal bovine serum (FBS), and Trizol reagent were purchased from Gibco BRL (Life Technologies, Grand Island, NY). Reporter cell lysis buffer and NF-κB consensus oligonucleotides (CGGACTTTC) were obtained from Promega (Madison, WI), and Effectene transfection reagent was from QIAGEN (Mississauga, Ontario, Canada). T4 polynucleotide kinase and T4 polynucleotide kinase 10× buffer were obtained from New England Biolabs (Beverly, MA); rabbit polyclonal NF-κB p65/Rel A antibody (sc-109X) and NF-κB p50 antibody (sc-114X) were from Santa Cruz Biotechnology (Santa Cruz, CA); and [γ-³²P]ATP (3,000 Ci/ml) was procured from DuPont (Boston, MA). Monoclonal (mouse) anti–O-linked N-acetylglucosamine (Clone RL2) was purchased from Affinity Bioreagents (Golden, CO). Poly deoxyinosine triphosphate-deoxycytosine triphosphate (poly dI-dC) and protein A Sepharose CL-4B beads (cat # 17-0780-01) were obtained from Amersham-Pharmacia Biotech (Québec, Canada). All other reagents were of the highest grade available from Sigma-Aldrich (Mississauga, Ontario, Canada).

MCs were obtained from male Sprague-Dawley rats as described (1,23,24). The cells were cultured (37°C, 5% CO₂) in DMEM supplemented with FBS (20%), penicillin (100 units/ml), streptomycin (100 μg/ml), and glutamine (2 mmol/l). Cells were used between passages 14 and 20.

The plasmid pCIS (empty vector), pCIS-GFP (expresses green fluorescent protein), and pCIS-GFAT (expresses the human GFAT gene) were supplied by Dr. M.J. Quon (Cardiology Branch; National Heart, Lung and Blood Institute, National Institutes of Health; Bethesda, MD) and have been previously described (25). The plasmid pNF-κB-Luc, consisting of an NF-κB enhancer element that contains five NF-κB binding sequences in tandem upstream of a luciferase reporter gene was obtained from Stratagene (La Jolla, CA). The other promoter-luciferase reporters used in this study have been previously described. VCAM-1 promoter luciferase construct (26) was provided by Dr. J.M. Redondo (Centro de Biologia Molecular, Universidad Autonoma, Madrid, Spain); TNF-α promoter-luciferase reporter (27) was from Dr. J. Ye (National Institute for Occupational Safety and Health, Morgantown, WV); and the IL-6 promoter luciferase reporter (28) was provided by Dr. A. Nakamura (Teikyo University, Tokyo). pCMV-βgal (Promega) was used to control for variation in transfection efficiency.

MC (1.5 × 10⁵ cells/well) were plated onto six-well plastic plates (SARSTEDT), and transfection was carried out 24 h later using Effectene (QIAGEN) according to the manufacturer’s specifications as we have reported (1). Briefly, MC (70–80% confluent) were co-transfected with 0.1 μg of promoter-luciferase construct or NF-κB enhancer-luciferase reporter, 0.25 μg of pGFAT, and 0.05 mg of pCMVβgal and then cultured for 18 h in DMEM containing FBS (20%) and 5.6 mmol/l d-glucose. With the use of green fluorescent protein, a transfection efficiency of 35–40% was obtained. Subsequently, the media was changed to DMEM with 0.5% FBS and glucose (5.6–30 mmol/l), glucosamine (0.25–5.0 mmol/l) with and without inhibitors azaserine (40 μmol/l), 6-diazo-5-oxonorleucine (DON; 40 μmol/l), benzyl-2-acetamido-2-deoxy-α-d-galactopyranoside (BADGP; 0.5 mmol/l), or pyrolidine dithiocarbamate (PDTC) (25 μmol/l). At various times, the cells were harvested and used for analyses as described below.

Medium was aspirated, wells were washed twice with PBS, and lysis was performed using 0.2 ml/well Reporter Lysis Buffer. MC were incubated for 15 min at 4°C, then transferred to microcentrifuge tubes using a rubber policeman. Cell debris was pelleted by centrifugation (12,000g, 4°C, 1.0 min), and the supernatant was used to assay for luciferase (0.02 ml) and β-galactosidase (0.05 ml) activities using commercially available reagents. Luciferase activity was assessed as per the manufacturer’s protocol (Promega), and β-galactosidase activity was determined using ONPG reagent (Sigma). Luciferase was measured in a luminometer (EG&G Berthold, Oak Ridge, TN), and β-galactosidase activity was based on the absorbance at 405 nm. Luciferase activity was normalized to the β-galactosidase activity and cell protein. Protein was determined on an aliquot of the supernatant obtained from cell lysis using Bio-Rad protein assay dye reagent (Bio-Rad Laboratories, Hercules, CA).

Total RNA from MCs was isolated by the single-step method of Chomczynski and Sacchi (29), as we have published (1,30). Isolated RNA was stored in diethyl pyrocarbonate–treated water at −80°C The purity and concentration were determined by measuring the optical density at 260 nm and 280 nm before use. The absorbance ratio, A₂₆₀:A₂₈₀, ranged from 1.75 to 1.95. Semiquantitative RT-PCR was performed as previously reported (1,30–32). The specific primer sequences were as follows: β-Actin, 5′ AAC CCT AAG GCC AAC CGT GAA AAG 3′, 3′ TCA TGA GGT AGT CTG TCA GGT C 5′; VCAM-1, 5′ GGA GAC ACT GTC ATT ATC TCC TG 3′, 3′ TCC TTT CAT GTT GGC TTT TCT TGC 5′ .

For amplification, 2.5 μl of the RT product was mixed with 7.5 μl of PCR mix containing 0.1 μmol/l of each of the primer pairs and 2 units of Taq polymerase. The sample was placed onto a Perkin-Elmer DNA thermal cycler (model 480) and heated to 94°C for 4 min before the application of temperature cycles. β-Actin was co-amplified to standardize the amount of RNA subjected to reverse transcription. The temperature cycle for amplification was as follows: 1) denature at 94°C for 30 s, 2) cool-anneal at 60°C for 30 s, and 3) heat-extend at 72°C for 30 s. For the β-actin, GFAT, and TGF-β1 primer pairs, the PCR product plateaued at 28 cycles, and therefore 25 cycles were chosen for the final amplification. For VCAM-1 primer pairs, the PCR product plateaued at 40 cycles, and therefore 35 cycles were chosen for final amplification as the PCR product. PCR products were separated on 1% agarose gel containing ethidium bromide, photographed, and quantified with a transmittance/reflectance scanning densitometer (model GS 300, Hoeffer Scientific Instrument) using a Macintosh class II computer (system 7.0) and Dynamax HPLC Method Management software (version 1.2).

The binding of nuclear protein to NF-κB consensus sequence was determined by electromobility shift assay (EMSA) (23). After washing in cold PBS, nuclear extracts of MCs were prepared by lysis in hypotonic buffer (20 mmol/l HEPES [pH 7.9], 1 mmol/l EDTA, 1 mmol/l EGTA, 20 mmol/l NaF, 1 mmol/l Na₃VO₄, 1 mmol/l Na₄P₂O₇, 1 mmol/l dithiothreitol [DTT], 0.5 mmol/l phenylmethylsulfonyl fluoride [PMSF], 1 μg/ml aprotinin, 1 μg/ml leupeptin, 1 μg/ml pepstatin A, and 0.6% Nonidet P-40), homogenized, and sedimented at 16,000g for 20 min at 4°C. Pelleted nuclei were resuspended in hypotonic buffer with 0.42 mmol/l NaCl and 20% glycerol and rotated for 30 min at 4°C. After centrifugation (12,000g, 20 min), the supernatant was collected and protein concentration was measured with the Bio-Rad assay kit.

NF-κB consensus oligonucleotides were prepared by incubating 2 μl of consensus oligonucleotide (1.75 pmol/μl), 1 μl of T4 polynucleotide kinase 10× buffer, 1 μl of [γ-³²P]ATP (3,000 Ci/ml; DuPont, Boston, MA), and 5 μl of nuclease-free water for 10 min at 37°C. The reaction was terminated by addition of 1 μl of 0.5 mol/l EDTA. Unlabeled ³²P-ATP was removed from the oligonucleotide mixture with D-25 Sephadex columns (Pharmacia, Uppsala, Sweden).

Supernatants were used as nuclear proteins for the binding assay. Nuclear proteins (3 μg) were incubated (30 min, room temperature) with 2 μg of poly(dI-dC) (Pharmacia) in binding buffer (20 mmol/l HEPES [pH 7.9], 1.8 mmol/l MgCl_2, 2 mmol/l DTT, 0.5 EDTA, and 0.5 mg/ml BSA) and then reacted with radiolabeled consensus oligonucleotides at room temperature for 20 min (50,000– 100,000 cpm). For supershift experiments, 1.0 μg of rabbit polyclonal NF-κB p65/RelA antibody or rabbit polyclonal NF-κB p50 antibody was added to the reaction mixture and incubated for 2 h at room temperature before addition of the radiolabeled consensus oligonucleotide. Reaction mixtures were electrophoresed in a 6% polyacrylamide gel and autoradiographed. Competition experiments were performed with 100× excess unlabeled consensus oligonucleotides.

Site-directed mutagenesis of the VCAM-1 promoter was carried out by a pair of 5′ -phosphorylated primers using pfu polymerase (33). The intact plasmid was amplified except for the deletion region that was excluded by the two primers. After PCR amplification, parental DNA template was destroyed by addition of DpnI (10–20 units) directly to the PCR reaction and incubated at 37°C for 1 h. DpnI only cuts methylated DNA. Because parental plasmid was purified from Escherichia coli, it was methylated, whereas PCR-amplified DNA was not methylated. After DpnI, the PCR-amplified DNA (linear) was gel-purified and circularized by ligation and transformed into E. coli. Colonies were then screened for the deletion. The two primers used were as follows: vcamlow1, 5′ GAGCCAGGGAAAAAGTTTAACTGA 3′; and vcamup3, 5′ TCCGCCTCTCTAGCAAGACCCT 3′. The PCR conditions were as follows: 94°C for 3 min, followed by 35 cycles at 94°C for 45 s, 56°C for 45 s, and 72°C for 17 min.

MCs were incubated with glucosamine (5.0 mmol/l) for 8–12 h, medium was removed, and the cells were washed (3×) with ice-cold PBS. Cells were harvested under nondenaturing conditions on ice by incubation for 5 min with ice-cold lysis buffer (20 mmol/l Tris [pH 7.5], 150 mmol/l NaCl, 1 mmol/l EDTA, 1 mmol/l EGTA, 1% Triton X-100, 1 mmol/l β-glycerophosphate, 1 mmol/l sodium orthovanadate, 1.0 μg/ml aprotinin, and 1 mmol/l PMSF). Cells were then scraped into microcentrifuge tubes on ice and centrifuged (12,000g, 4°C, 15 min). The supernatant was transferred to a fresh microcentrifuge tube, and the protein concentration was measured with Bio-Rad protein assay reagent.

Protein (500 μg) was mixed with 1 μg rabbit polyclonal NF-κB p65/Rel A antibody and incubated at 4°C overnight. Subsequently, 25 μl of protein A Sepharose CL-4B slurry (50%) was added and incubated for 90 min (4°C, gentle rocking). The mixture was centrifuged, and the pellet was washed with 2× buffer A (10 mmol/l Tris-HCl [pH 7.5]), 100 mmol/l NaCl, 2 mmol/l EDTA, 0.2% Nonidet P-40, and 0.25 mmol/l PMSF) then with 1× buffer B (10 mmol/l Tris-HCl [pH 7.5], 500 mmol/l NaCl, 2 mmol/l EDTA, 0.2% Nonidet P-40, and 0.25 mmol/l PMSF) and 1× buffer C (10 mmol/l Tris-HCl [pH 7.5] and 0.25 mmol/l PMSF). The beads were suspended in 20 μl of 3× SDS sample buffer (125 mmol/l Tris-HCl [pH 6.8], 4% SDS, 20% glycerol, 100 mmol/l DTT, and 0.2% bromophenol blue), boiled for 2 min, then centrifuged for 1 min (12,000g, 4°C).

The supernatant containing immunoprecipitated protein was electrophoresed in SDS-PAGE gel (8%), and protein was transferred to nitrocellulose membranes (Protran; Schleicher and Schuell, New Haven, CT). The membrane was incubated in 25 ml of blocking buffer (5% skim milk in Tween and Tris-buffered saline (TTBS) for 30 min at room temperature. Subsequently, immunoblotting was performed with RL2 antibody (1:1,000 in blocking buffer, 3 h, room temperature). The membrane was washed with TTBS (three times, 5 min/wash, room temperature) and subsequently incubated with 1:5,000 diluted horseradish peroxidase–conjugated anti-rabbit antibody (Bio-Rad, catalogue no. 170-6515) for 30 min at room temperature. The membrane was washed three times with TTBS (5 min/wash), and protein was detected with Luminol reagent (catalogue no. NN-NEL602001 KT, New England Nuclear) after exposure to x-ray film (catalogue no. NEF596; Kodak, Toronto, Ontario, Canada).

Statistical analyses were performed with the INSTAT statistical package (GraphPad Software, San Diego, CA). The difference between means was analyzed using the Bonferroni multiple comparison test. Significance was defined as P < 0.05.

To ascertain whether exposure to glucose or glucosamine led to any changes in cytokine expression in MCs, we used RT-PCR to determine VCAM-1 mRNA levels. In MCs that were exposed to glucose (30 mmol/l) for 24 h, mRNA levels for VCAM-1 increased by 2.3-fold. Similarly, in MCs that were cultured in medium with physiologic glucose concentration (5.6 mmol/l) supplemented with the hexosamine pathway product glucosamine (1 mmol/l) for 12 and 24 h, we observed a comparable increase in VCAM-1 mRNA levels at 24 h (Fig. 1).

The activity of the genes for several cytokines, including that for VCAM-1, is dependent on the activation of the transcription factor NF-κB (14,16). To determine whether high glucose or glucosamine might activate NF-κB, we studied the activity of an NF-κB–responsive luciferase reporter, pNF-κB-Luc, under various conditions.

Incubation of MCs in medium containing 30 mmol/l glucose for 48 h led to a 1.5-fold increase in luciferase activity (P < 0.05, n = 4) (Fig. 2A). This effect was not due to osmolarity because exposure of MCs to medium containing 5.6 mmol/l glucose plus 24.4 mmol/l mannitol did not reproduce the results seen with high glucose. Importantly, glucosamine (5 mmol/l) also activated the NF-κB enhancer element (Fig. 2B), suggesting that, like high glucose, flux through the hexosamine pathway activates NF-κB.

We have previously shown that GFAT overexpression mimics the action of high glucose and glucosamine in MCs (1,6). Hence, given the above findings that high glucose and glucosamine activated the NF-κB enhancer element, we wished to determine whether GFAT overexpression would recapitulate these observations.

MCs cotransfected with pGFAT and pNF-κB-Luc showed a 2.1-fold increase in luciferase activity compared with those transfected with the empty vector (pCIS) (Fig. 2A), thus complementing our observations with glucosamine (Fig. 2B). The effect seen with GFAT overexpression was enhanced in MCs transfected with pGFAT and cultured in medium with high glucose (30 mmol/l). This maneuver led to a 2.8-fold increase in luciferase activity. Furthermore, activation of the NF-κB enhancer element produced by overexpressing GFAT was prevented by the GFAT inhibitors azaserine and DON or by exposing MCs to PDTC (25 μmol/l), an inhibitor of NF-κB (Fig. 2A). These findings further suggest that hexosamine flux may play a role in NF-κB activation in MCs.

After activation, NF-κB is translocated to the nucleus, where it participates in the activation of NF-κB–dependent gene transcription. The increase of NF-κB levels in the nucleus is reflected by augmented NF-κB binding to consensus sequences (18,20,22). To determine whether high glucose, glucosamine, or GFAT overexpression increases binding of nuclear proteins to NF-κB consensus sequence (GGGATTTCCC), we performed EMSA (23).

The binding of nuclear protein to radiolabeled NF-κB consensus sequence was increased by high glucose (Fig. 3A, B, and E) and glucosamine (Figs. 3E and F). Like high glucose and glucosamine, GFAT overexpression increased the binding of nuclear protein to NF-κB consensus sequence (Figs. 3C and D). This was attenuated in cells that were transfected with pCIS-GFAT and exposed to the GFAT inhibitors azaserine and DON (Figs. 3C and D), suggesting that the increased nuclear binding was related to GFAT activity. Supershift experiments performed with antibodies to the p50 and p65 subunits of NF-κB demonstrated retardation by p65 antibody (Fig. 3E, band 3).

Several proteins, including transcription factors Sp1 and c-myc and RNA polymerase II, are O-glycosylated, and this posttranslational modification is believed to be important in protein stability and function (34– 39). Our finding that flux through the hexosamine pathway may increase NF-κB nuclear activity by increasing p65 protein led us to examine whether p65 is O-glycosylated and how this may be affected by high glucose and glucosamine.

NF-κB p65/RelA subunit was immunoprecipitated with anti–NF-κB p65 antibody and subsequently immunoblotted with RL2 antibody to detect O-glycosylated products. Figure 4 demonstrates that in MCs cultured in physiologic glucose, there is minor O-glycosylation of the p65 subunit of NF-κB. Incubation of MCs in either high glucose or glucosamine significantly increases O-glycosylation of the p65 subunit of NF-κB (Fig. 4).

The transcriptional activities of several inflammatory genes and adhesion molecules are induced by NF-κB. Three of the most studied are TNF-α, IL-6, and VCAM-1 (40–44). In view of our observation that GFAT overexpression activated the NF-κB enhancer element, we sought to determine whether the same would hold for NF-κB–dependent promoter activity.

In MCs that overexpressed GFAT, we observed a 2.25-, 1.70-, and 1.5-fold induction of the VCAM-1, TNF-α, and IL-6 promoter, respectively (Fig. 5). Because the response of the VCAM-1 promoter was most marked, we chose to characterize further the activation of this promoter in MCs.

In these experiments, MCs were transiently transfected with the VCAM-1 promoter-luciferase reporter alone, or the former construct was cotransfected with pCIS-GFAT or pCIS. We found that both glucose and glucosamine caused dose-dependent increases in VCAM-1 promoter activity (Figs. 6A and B). TNF-α, a known activator of VCAM-1 gene expression in several cells, including MCs (16,45,46), also activated the VCAM-1 promoter in MCs (Fig. 7). In MCs overexpressing GFAT, the activity of the VCAM-1 promoter was also increased, and incubation in high glucose for 48 h led to additional enhancement of VCAM-1 promoter activity, above that seen in physiologic (5.6 mmol/l) glucose (P < 0.05, n = 4) (Fig. 7). The effect of high glucose on VCAM-1 promoter activity was not due to increased osmolarity because in MCs cultured in 5.6 mmol/l glucose plus 24.4 mol/l d-mannitol, the VCAM-1 promoter was not activated (Figs. 6A and 7).

The above results suggested that activation was dependent on flux through the hexosamine pathway and that GFAT overexpression sensitizes MCs to the effect of high glucose. This is further supported by the observation that inhibition of GFAT enzymatic activity by incubating cells with DON or azaserine prevented the increase in VCAM-1 promoter activity (Fig. 7). We observed similar inhibition of activation of the VCAM-1 promoter by GFAT overexpression when the experiments were performed in the presence of PDTC (25 μmol/l), a compound that is known to interfere with NF-κB activation.

BADGP has been shown to limit O-glycosylation by preventing addition of N-acetylglucosamine to proteins such as mucin (47,48). Recently, we also demonstrated that BADGP may attenuate the activity of the hexosamine pathway (30). Therefore, to characterize further the activation of the VCAM-1 promoter by GFAT, we assessed promoter activity in the presence and absence of this inhibitor. As shown in Fig. 7, BADGP (0.5 mmol/l) attenuated the activation of the VCAM-1 promoter that occurred with GFAT overexpression.

Two NF-κB interacting sites located at −73 and −57 have been identified in the VCAM-1 promoter (49). We used site-directed mutagenesis to modify these sites (Fig. 8A), then we studied the impact of this modification on the activation of the promoter by high glucose, glucosamine, or GFAT overexpression. The results, shown in Fig. 8B, indicate that the mutated promoter was not activated by high glucose, glucosamine, and GFAT overexpression. This further suggested that activation of the VCAM-1 promoter by increasing hexosamine pathway flux was dependent on NF-κB.

Recent studies have suggested that flux through the hexosamine pathway, which produces the substrates for O- and N-glycosylation of proteins, may allow cells to sense the level of glucose in the extracellular environment (50) and may also be an important determinant of hyperglycemic tissue injury (1,50–54). O-glycosylation of serine and threonine residues is important for posttranslational modification affecting the behavior of many nuclear and cytosolic proteins, including some transcription factors (5,34,35). O-glycosylation of transcription factors is believed to be important in the regulation of gene expression (34–36,38, 39). Indeed, protein O-glycosylation has been likened to protein phosphorylation, and for some signaling molecules, these two processes may be competitive (5,34,36, 55). However, the exact role of the hexosamine pathway and O-glycosylation in gene expression remains unclear.

In humans with type 1 diabetes, kidney sections show increased immunostainable GFAT (3), and GFAT activity is elevated in skeletal muscle from patients with type 2 diabetes (56). Our recent studies indicate that GFAT activity is an important determinant of expression of genes implicated in diabetic glomerular disease. As such, we demonstrated that GFAT overexpression or glucosamine activated the promoter for PAI-1 in an Sp1-dependent manner and increased mRNA levels for TGF-β1 and its receptors (1,6).

Inflammatory cytokines such as VCAM-1 are expressed by MCs in culture (57), and VCAM-1 expression is increased in glomeruli from STZ-induced diabetic rats and mice (11,12). Because expression of the genes for these cytokines is NF-κB–dependent (58) and NF-κB is activated by high glucose (13,14,17), we speculated that flux through the hexosamine pathway might also play a role in the activation of NF-κB–dependent genes. Indeed, our first major observation was that either high glucose or glucosamine increased VCAM-1 mRNA levels (Fig. 1) and activated an NF-κB enhancer element (Fig. 2). The magnitude of the activation was comparable to that reported for high glucose–mediated activation of NF-κB–dependent genes by several other investigators (11,14,16, 59).

GFAT overexpression also activated the NF-κB enhancer element, and synergistic activation was observed with the combination of high glucose and GFAT overexpression. The observed increase in luciferase activity was dependent on GFAT activity because it was prevented when MCs were incubated with the GFAT inhibitors azaserine and DON. Because activation occurred when MCs were cultured in medium that contained physiologic glucose concentrations (5.6 mmol/l), these results indicate GFAT overexpression alone is sufficient to activate the NF-κB enhancer element and are consistent with our findings for the PAI-1 promoter (1). A hyperosmolar stimulus with mannitol did not reproduce the effect of 30 mmol/l glucose (Fig. 2B). Our observation that either glucosamine or GFAT overexpression increases the activity of the NF-κB enhancer element (Fig. 2B) bolsters the argument that NF-κB transcriptional elements may be activated by flux through the hexosamine pathway.

The transcription factor NF-κB serves as a critical regulator of the inducible expression of many genes. NF-κB is activated by multiple stimuli and is a critical component of the signal transduction machinery required for tissue remodeling and development (44,58,60– 65). Accordingly, we sought to characterize the effects of hexosamine flux on NF-κB–dependent gene transcription.

Our second major observation was that high glucose and glucosamine increased binding of nuclear proteins to NF-κB consensus sequences (Fig. 3A, B, E, and F). Similarly, GFAT overexpression increased the binding of nuclear proteins to an NF-κB consensus sequence, which was prevented by GFAT inhibition with azaserine and DON (Figs. 3C and D). The high glucose finding is supported by recent studies in which high glucose was found to induce similar magnitudes of NF-κB activation in vascular smooth muscle cells (16), peripheral blood mononuclear cells from patients with diabetes (17), and cultured endothelial cells (66). Our observation that the hexosamine pathway is a determinant of NF-κB activation is novel and may provide one possible mechanism whereby high glucose leads to activation of NF-κB–dependent genes.

The NF-κB/Rel family of transcription factors consists of five members (p50, p52, p65 [RelA], RelB, and c-Rel) that form homo- and heterodimers that are capable of binding to DNA and subsequently activating NF-κB targeted gene (58,67). p50/p65 heterodimers have been the most extensively studied, and specific DNA binding sites for this dimer have been identified in the promoter region of genes for a diverse array of cytokines, adhesion molecules, enzymes, and receptors (58,67). In the present study, we used antibodies to p50 and p65 to ascertain whether these subunits were responsible for the increased binding of nuclear protein to NF-κB consensus sequences upon exposure of untransfected MCs to high glucose (30 mmol/l) or glucosamine (5 mmol/l). Under our experimental conditions, only nuclear protein extracts that were incubated with the p65 antisera produced a supershift (Fig. 3E, band 3). This suggests that in MCs that were exposed to high glucose or glucosamine, the increase in nuclear protein binding to NF-κB oligonucleotide consensus sequences was due primarily to p65/RelA. This is consistent with the recent observations of increased nuclear p65 translocation in MCs exposed to high glucose (68). Furthermore, several reports support a role for selective activation of Rel/NF-κB family members, and some NF-κB target genes contain binding sites that preferentially bind p65 homodimers (16,69–72). In addition, adenovirus-mediated expression of a dominant-negative p65 was found to inhibit induction of several proinflammatory genes, including ICAM-1, VCAM-1, and IL-8 in porcine aortic endothelial cells (62).

Flux through the hexosamine pathway is believed to exert an effect on gene expression by increasing the intracellular concentration of UDP N-acetylglucosamine, a substrate for the O-glycosylation of proteins (10, 34,35,53,73). Intracellular levels of O-linked glycosylated proteins correlate with GFAT activity, and blockade of GFAT activity or inhibition of GFAT expression with antisense oligonucleotides lowers the intracellular levels of O-GlcNAc–modified proteins (10, 73). The posttranslational modification of serine residues in transcription factors by O-glycosylation can affect the activity of the transcription factors. For instance, O-glycosylation of Sp1 stabilizes the protein and prevents proteasomal degradation (74) and may also play a role in the stability of the tumor suppressor p53 in relation to its ability to regulate transcription (75). In light of this, our third major observation was that p65/RelA is O-glycosylated in untransfected MCs that are exposed to high glucose or glucosamine (Fig. 4).

Our finding that activation of an NF-κB enhancer was dependent on flux through the hexosamine pathway led us to hypothesize that the increased flux through the hexosamine pathway might also activate NF-κB–dependent inflammatory genes. To test this hypothesis, we transiently co-transfected MCs with pCIS or pCIS-GFAT and promoter-luciferase reporter constructs for VCAM-1, TNF-α, and IL-6, respectively. Our fourth major observation was that overexpression of GFAT in MCs activated promoter constructs for VCAM-1, TNF-α, and IL-6 (Fig. 5). To relate these effects of GFAT overexpression to NF-κB– dependent promoter activation, we further characterized the effect of glucose and glucosamine on VCAM-1 promoter activity. We observed that both glucose and glucosamine dose-dependently increased the activity of the VCAM-1 promoter (Figs. 6A and B), whereas mannitol was without effect (Fig. 6A). We further found that activation of the VCAM-1 promoter by overexpression of GFAT could be inhibited by azaserine and DON (Fig. 7), strengthening a role for flux through the hexosamine pathway in activation of VCAM-1 promoter. BADGP and PDTC prevented activation of VCAM-1 promoter in GFAT-overexpressing MCs (Fig. 7), supporting a role for O-glycosylation and NF-κB activation, respectively, in VCAM-1 induction by increasing hexosamine flux.

Recent studies have demonstrated increased inflammatory cytokine expression, including VCAM-1, in the glomeruli of STZ-induced diabetic mice and rats (11,12). On the basis of these findings, it has been suggested that hyperglycemia-mediated increased expression of cytokines such as VCAM-1 may promote infiltration of inflammatory cells in diabetes and that inflammation may be an important component of the response of various tissues to chronic hyperglycemia (12). Our observations suggest that glucose flux through the hexosamine pathway may also be a determinant of VCAM-1 expression in MCs.

In summary, our findings support the general hypothesis that metabolic pathways linked to the development of insulin resistance may also influence glucose-mediated inflammation in vascular smooth muscle–like cells. More specific, our findings suggest that the rate-limiting enzyme in the hexosamine pathway, GFAT, sensitizes MCs to the effects of high glucose by regulating flux through the hexosamine pathway, and through this action, the hexosamine pathway may contribute to injury and inflammation by influencing the expression of NF-κB–dependent genes (Fig. 9).

This work was supported by a grant from the Juvenile Diabetes Foundation and the Canadian Institute of Health Research (CIHR) to J.W.S. L.J. is the recipient of a Biomedical Fellowship from the Kidney Foundation of Canada.

We thank Drs. J.M. Redondo, J. Ye, and A. Nakamura for generously providing the promoter reporter constructs for VCAM-1, TNF-α, and IL-6, respectively, and Dr. M.J. Quon for providing the plasmids pCIS, pCIS-GFP, and pCIS-GFAT.