Molecular Mechanisms of Tumor Necrosis Factor α Gene Expression in Monocytic Cells via Hyperglycemia-induced Oxidant Stress-dependent and -independent Pathways*

Increased oxidative stress has been reported in vivo in the diabetic state via the production of reactive oxygen species (ROS). Such stress is bound to play a key role on activation of circulating monocytes, leading to the accelerated atherosclerosis observed in diabetics. However the exact molecular mechanisms of monocyte activation by high glucose is currently unclear. Here, we demonstrate that chronic high glucose (CHG) causes a dramatic increase in the release of the inflammatory cytokine tumor necrosis factor α (TNFα), at least in part through enhanced TNFα mRNA transcription, mediated by ROS via activation of transcription factors nuclear factor κB (NF-κB) and activating protein-1 (AP-1). TNFα accumulation in the conditioned media was increased 10-fold and mRNA levels were increased 11.5-fold by CHG. The following observations supported that both NF-κB and AP-1 mediated enhanced TNFα transcription by CHG: 1) A 295-base pair fragment of the proximal TNFα promoter containing NF-κB and AP-1 sites reproduced the effects of CHG on TNFα transcription in a luciferase reporter assay, 2) mutational analyses of both NF-κB and the AP-1 sites abrogated 90% of the luciferase activity, 3) gel-shift analysis using the binding sites showed activation of NF-κB and AP-1 in CHG nuclear extracts, and 4) Western blot analyses demonstrated elevated nuclear levels of p65 and p50 and decreased cytosolic levels of IκBα in CHG-treated monocytes. That ROS acted as a key intermediate in the CHG pathway was supported by the following evidence: 1) increased superoxide levels similar to those observed with PMA or TNFα, 2) increased phosphorylation of stress-responsive mitogen-activated protein kinases p38 and JNK-1, 3) counteraction of the effects of CHG on TNFα production, the 295TNFluc reporter activity, activation of NF-κB, and repression of IκBα by antioxidants and p38 mitogen-activated protein kinase inhibitors. The study suggests that ROS function as key components in the regulatory pathway progressing from elevated glucose to monocyte activation.

Cellular redox state has been shown to play an important role in the pathogenesis of cardiovascular disease including atherosclerosis, the rate of which is higher in diabetics (1)(2)(3). Hyperglycemia in the blood stream could generate free radicals and peroxide species by slow "autooxidation" of glucose, causing oxidative stress to circulating monocytes (4,5). Furthermore, glycosylation of low density lipoprotein increases its susceptibility to oxidation, generating byproducts in circulation that preferentially accumulate in foam cell-generating monocytes/macrophages (6,7). Additionally, soluble advanced glycated end products (AGEs) 1 present in the blood stream could also generate reactive oxygen species (ROS) (8 -10). AGEs deposited in the arterial walls generate free radicals capable of oxidizing vascular lipids and accelerating atherogenesis in hyperglycemia (9,11).
As peripheral blood glucose levels increase in hyperglycemia, there is simultaneous rise in intracellular glucose levels, utilizing the sorbitol pathway and altering the redox balance inside the cells. Hyperglycemia also leads to increased NADH/ NAD ϩ ratio, thereby decreasing the availability of NAD ϩ as a co-factor for other metabolic events (12)(13)(14). The redox changes induced by hyperglycemia, AGEs, and lipid peroxidation have been shown to alter cellular functions via activation of key signal transduction pathways involving MAPKs such as ERK 1/2, JNKs, and p38 (15)(16)(17)(18). High glucose and diabetes have been shown to specifically activate p38 MAPK via ROS intermediates in smooth muscle cells (19 -20), and oxidant stress has been shown to incite macrophage spreading via the p38 MAPK pathway (21). In addition, production of inflammatory cytokines such as TNF␣ and interleukin-6 by activated rat smooth muscle cells was regulated by the p38 MAPK pathway (22). Activation of the p38 MAPK has been observed in a number of physiological responses such as apoptosis of myocardial cells (25) and adipogenesis in 3T3-L1 cells (23). Altered NADH/NAD ϩ ratio caused by hyperglycemia results in de novo synthesis of diacylglycerol and activation of various protein kinase C (PKC) isoforms in cell/tissue-type and stimulus-specific manner (7,25,44). That hyperglycemia induced ROS may function as a key intermediate leading to the activation of PKC has been shown in many cell types of human and porcine origin * The work was supported in part by National Institutes of Health Grant POI HL55798 and a grant from the Juvenile Diabetes Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The inflammatory cytokine human tumor necrosis factor ␣ (hTNF␣) is produced by activated monocytes in response to a variety of signals including stress response, phorbol esters, cytokines, endotoxin, and substrate adherence (27)(28)(29)(30)(31). TNF␣ gene expression is regulated both at the levels of transcription and post-transcription. Elevated levels of TNF␣ and other inflammatory cytokines have been detected in atherosclerotic plaques of diabetic and nondiabetic patients (32).
The role of monocytes in increased foam cell formation in diabetic patients is well established (25). Hyperglycemia-induced oxidative stress, further accentuated by the inactivation of superoxide dismutase (12), along with soluble AGEs and products of lipid peroxidation possibly serve as key activators of circulating monocytes via the activation of upstream kinases, leading to induction of inflammatory gene expression. However the signaling kinases or transcription factors specifically involved in high glucose-induced monocyte activation leading to the production of the inflammatory cytokine TNF␣ are still unclear and are the focus of the present study.
Activation of genes in response to inflammatory stimuli has been shown to involve coordinated participation of transcription factors NF-B and AP-1. Regulation of many inflammatory cyokines, tissue factor, and matrix metalloproteinases involve dual transcriptional regulation by NF-B and AP-1 (33)(34)(35)(36). NF-B/Rel proteins are heterodimeric transcription factors retained in the cytoplasm of unstimulated cells by the inhibitory subunit IB, the NF-B/IB forming an inactive ternary complex (37). Stimulation with stress-inducing agents or other proinflammatory mediators causes rapid phosphorylation, ubiquitination, and degradation of IB-subunit, allowing translocation of NF-B to the nucleus (37). NF-B then induces transcription of several genes, including that of its inhibitor IB (38). Another transcription factor regulated by cellular stress is AP-1, a transacting molecule consisting mainly of homodimers of Jun or heterodimers of Fos and Jun (39). The hTNF␣ gene has in its promoter region canonical binding sites for transcription factors NF-B and AP-1 (40).
Since the exact mechanism for the activation of monocytes leading to inflammatory cytokine production by hyperglycemia is currently unclear, we evaluated some of the key molecular and cellular events leading to TNF␣ secretion by high glucose. Our study suggests that CHG-induced monocyte activation, as evidenced by increased TNF␣ expression, was regulated at least in part through increased TNF␣ mRNA transcription. The process involved ROS-dependent and -independent pathways, requiring coordinate activation of both p38 MAPK and PKC as upstream kinases and NF-B and AP-1 as downstream transcription factors.

MATERIALS AND METHODS
Cell Culture-U937 (monoblastoid) cells and THP-1 (histiocytic) cells were obtained from ATCC and maintained in RPMI 1640 medium containing 7% heat-inactivated fetal calf serum, ␤-mercaptoethanol (50 M), HEPES (10 mM), glutamine (2 mM), streptomycin (50 g/ml), penicillin (50 units/ml), and 5.5 mM glucose (NG). For chronic high glucose (CHG) conditions, cells were cultured in 12.5, 15, or 25 mM glucose for 2 days before being depleted of serum and treated with various agents as indicated under "Results." High glucose (HG) culturing was done in 15 mM glucose for 18 h. This HG condition was used to culture peripheral human monocytes for TNF␣ production and for 295TNFluc luciferase activity studies. CHG-or HG-treated cells were washed and resuspended in depletion medium containing 0.5% bovine serum albumin for 18 h or for other periods as indicated prior to stimulation.
Preparation of Human Monocytes-Fresh human monocytes were obtained from healthy donors using an approved Institutional Review Board protocol and isolated as described previously (41). Autologous serum was used for the attachment purification and culture of the monocytes.
Lucigenin Chemiluminescence Assay (LCA)-This assay was performed as described earlier (6) to measure superoxide anion (O 2 Ϫ ). Briefly, U937 or THP-1 cells were cultured in NG versus CHG, depleted of serum using depletion media, and prepared for the assay. Cells were treated with TNF␣ (5 ng/ml) or PMA (10 ng/ml; positive control) for 60 min, washed in a balanced salt solution, and resuspended at 2 ϫ 10 6 cells/ml in aerated balanced salt solution. 1 ϫ 10 6 cells/ml was used for the assay. O 2 Ϫ was measured in intact cells by LCA. Cells from various treatments were added to a scintillation vial containing lucigenin (500 M) in the aerated balanced salt solution. Photon emission was measured for 10 min using the Beckman LS6500 Multipurpose Scintillation counter measuring single photon emission. First, photon emission was measured using a buffer blank and dark-adapted lucigenin, and the blank reading was subtracted from the sample reading. A standard curve was generated using xanthine and xanthine oxidase. Superoxide dismutase (100 units/ml) was used as an inhibitor for superoxide production. H 2 O 2 was added to the buffer blanks or to the NG control cells to determine if the photon emission in the lucigenin assay was induced by O 2 Ϫ or peroxide species. Inhibitors and Reagents-N-Acetyl-L-cysteine (NAC; 100 M) was purchased from Calbiochem. It was dissolved in phosphate-buffered saline, and the pH was adjusted to 7.2. Pyrrolidine dithiocarbamate (PDTC; Sigma; 50 -100 M) was dissolved in water. Me 2 SO was used to resuspend the rest of the inhibitors and was added to the control plates. p38 MAPK inhibitor SB202190 (SB, 10 M) was purchased from Upstate Biotechnology. Mannitol, 3-O-methyl glucose, and 2-deoxyglucose were purchased from Sigma.
Detection of Secreted TNF␣ in the Culture Supernatant by Specific hTNF␣ Enzyme-linked Immunosorbent Assay-U937 or THP-1 cells were cultured in NG or CHG, depleted of serum in depletion media, and cultured in NG, CHG, or NG ϩ PMA for another 24 h. Quantitative detection of hTNF␣ in conditioned media was performed using a specific antibody sandwich Cytoscreen TM enzyme-linked immunosorbent assay assay from BIOSOURCE International using the manufacturer's suggested directions. Known concentrations of hTNF␣ were used to generate the standard curves. Supernatant from NG-cultured cells treated with PMA (10 ng/ml) for 24 h served as the positive control. Streptavidin-horseradish peroxidase served as the detection system. For each experiment, duplicate samples were measured. Data were represented as the mean (pg/10 6 cells) Ϯ S.E. The assay was linear between 15.6 pg/ml and 1000 pg/ml. Detection of TNF␣ Message by Competitive RT-PCR Assay-TNF␣ message was measured in U937 cells using the Quantitative RT-PCR Cytoexpress detection kit from BIOSOURCE International. Cells were cultured and depleted as in the preceding paragraph above, and RNA was extracted from CHG or NG cells stimulated for 4 h with various concentrations of TNF␣ (0.25, 0.5, 1, or 5 ng/ml) or PMA (10 or 50 ng/ml). RNA was reverse-transcribed to cDNA using murine leukemia virus reverse transcriptase. A known copy number of exogenously synthesized DNA, known as the internal control standard (ICS) was mixed with sample cDNA before PCR amplification. The ICS contained PCR primer binding sites similar to the TNF␣ cDNA and a unique capturebinding site to distinguish the ICS amplicon from the TNF␣ amplicon. In samples containing ICS, two amplicon bands were visible following PCR amplification, the 382-base pair TNF␣ band and the 432-base pair ICS band. After amplification, the amplicons were hybridized to the ICS or TNF␣-specific oligonucleotide-coated wells. Biotinylation of an original primer allowed streptavidin coupled to biotin to be used as the detection system. The signal generated in the hybridization reaction was proportional to the number of amplicons present in the starting cDNA. Since the ICS (known copy number) was amplified at the same frequency as the TNF␣ cDNA, it served to determine the copy number of TNF␣ cDNA in each sample (see Equation 1).
Copy number of TNF␣ message ϭ 2 ϫ known ICS copy number ϫ A in TNF␣ wells A in ICS wells (Eq. 1) Plasmids and Luciferase Reporter Gene Assays-The 295TNF␣ luciferase (295TNFluc) deletion construct from the human TNF␣ promoter was a kind gift from Dr. James S. Economou, UCLA (44). 5 ϫ 10 6 U937 or THP-1 cells were depleted of serum 2 h before transfection. The cells were transfected with 2 g of the 295TNFluc deletion construct and co-transfected with 0.2 g of reporter plasmid (RSV␤Gal) using the Effectene reagent (Qiagen Inc.) in 2% serum-containing media overnight. The cells were washed and depleted of serum for 4 h before stimulation with glucose (NG versus HG) and/or TNF␣ for 18 h. Alternatively, cells were cultured in NG or CHG for 24 h, washed, depleted for 4 h, and transfected with 295TNFluc plasmid overnight. On day 3, cells were washed, depleted of serum, and stimulated with NG or CHG plus/minus TNF␣ for 18 h, and stimulated luciferase activity was measured. Treatment with PMA served as the positive control for the stimulated luciferase activity. Inhibitors were added at the indicated concentrations 1 h before CHG culturing or before treatment with TNF␣ or PMA. The inhibitors were present throughout the culturing and the stimulation phase. The luciferase assay was performed using the firefly luciferase kit (Promega) using a Turner TD-20e luminometer measuring light intensity over a 5-log range. Results from the luciferase assay were normalized to ␤-galactosidase levels, and relative luciferase units were determined. The results were reported as fold stimulation over XP-1 control plasmids. Concentrations of inhibitors used for the study are listed under "Inhibitors and Reagents" of this section.
Mutational Analysis of the NF-B3 and AP-1 Sites in the TNF␣ Promoter-Site-directed mutagenesis was performed at the NF-B3 and the AP-1 sites using the Quick-change TM site-directed mutagenesis kit from Stratagene using the manufacturer's suggested directions. The NF-B3 (GGGTTTCTCC) and AP-1 (TGAATGA) sequences were mutated to taGTTTCTCC (mNF-B3) and gtAATGA (mAP-1), respectively, using primers as suggested in published protocols (47).
Preparation of Nuclear and Cytosolic Extract for Gel-shift (Electromobility Shift Assay (EMSA)) and Western Analyses-Cells were cultured in NG or CHG, depleted of serum in depletion media, and treated with TNF␣ or preincubated with the inhibitors at concentrations for the time indicated under "Results." Nuclear and cytosolic extracts were made by a modification of the protocol of Lin et al. (42). The cytosolic extracts (Western blot analysis) and nuclear extracts (gel-shift assay and Western blot analysis) were divided into aliquot and frozen at Ϫ70°C for future use. The concentrations of inhibitors used for the study are listed above.
Binding Reaction and EMSA-Oligonucleotide probes for EMSA were synthesized in the City of Hope National Medical Center DNA synthesis facility. The proximal NF-B (NF-B3) and AP-1 sites from the TNF␣ promoter were used for the EMSA. The sequence for the NF-B3 site was: GCTCATGGGTTTCTCCACCAAG. The sequence for the AP-1 site was: CCAGATGAGCTCATGGG. EMSA was performed according to published protocol (43). The probes were labeled with [␥-32 P]ATP using T4 kinase (Stratagene). Antibodies used for supershifting experiments, anti-p50 (NLS), anti-p65 (c-20), anti-C rel, anti-Rel B (C- 19), and anti-p52 (K-27), were purchased from Santa Cruz Biotechnology. The gels were dried and visualized using a PhosphorImager screen (Molecular Dynamics, San Jose, CA) and quantified using the Imagequant software (NIH, Bethesda, MD).
Western Analysis-Nuclear extracts (5 g) were mixed with equal volume of 2ϫ sample buffer (4% SDS, 10% glycerol, 0.006% bromphenol blue, and 2% ␤-mercaptoethanol), boiled for 5 min, and run on SDSpolyacrylamide reducing gel (20). Antibodies for Western analyses were purchased from Santa Cruz Biotechnology. Western blot analysis was done using anti-p50, anti-p65, anti-IB␣, anti-IB␤, and anti-Ib⑀ antibodies at a dilution of 1:1000. Anti-histone H1 (C-17), used at 1:10000 dilution, served as the loading control for nuclear extracts. Cytosolic extracts were treated the same way as the nuclear extracts, and Western blot analysis was done using anti-IB␣ and anti-IB⑀ antibodies. Anti ␣-actin antibody (1:20000 dilution), purchased from Sigma, served as the loading control for the cytosolic extracts. Phosphorylated ERK 1/2, JNK-1, and p38 were detected by Western blot analysis using phosphorylated ERK 1/2 (pERK), JNK (pJNK), and p38 (pp38) mitogenactivated protein kinase antibodies purchased from New England Biolabs. Nonphosphorylated ERK 1/2, JNK-1, and p38 served as the loading controls. For these experiments, cells lysates were made using lysis buffer described earlier (20). All Western blots were developed using the ECL detection system (Amersham Pharmacia Biotech), following the protocol suggested by the manufacturer. Anti-rabbit horseradish peroxidase-linked secondary antibody served as the detecting antibody. Stripping and reprobing of the membranes for Western blot analysis was performed according to stringent conditions suggested in the ECL handbook. Western blots were quantified with the Alpha Imager documentation and analysis system using the Alpha Imager 3.24 software.
Statistical Analysis-Results are expressed as the mean Ϯ S.E. of the average responses in multiple experiments. Data were analyzed by analysis of variance followed by Tukey's test or by Student's t tests for paired components.

CHG Induces Oxidant Stress in Monocytes CHG Increased Reactive Oxygen Species in Monocytic Cells
Detected by LCA-To investigate whether CHG increased ROS in monocytic cell lines (U937 and THP-1), we examined the effect of CHG culturing alone on superoxide (O 2 Ϫ ) production.

CHG increased O 2
Ϫ levels significantly (p Ͻ 0.005) over NGcultured U937 cells (2.3 Ϯ 0.87-fold; Fig. 1A). O 2 Ϫ levels generated by CHG were comparable with that observed in NG cells following treatment with the inflammatory cytokine TNF␣ or phorbol ester PMA (positive control) for 1 h. Treatment of cells with superoxide dismutase before measuring O 2 Ϫ levels by LCA quenched any detectable O 2 Ϫ produced in CHG-, TNF␣-, or PMA-treated cells. Adding H 2 O 2 to NG cells did not increase chemiluminescence counts, indicating that O 2 Ϫ but not peroxide was the contributor for the photon emission. THP-1 cells generated similar trends but had higher levels of O 2 Ϫ than U937 cells.
Elevated Phosphorylation of Stress-responsive MAPKs by CHG, Confirming Oxidant Stress-An important measure of oxidative stress is the activation of upstream stress-responsive MAPKs. We therefore examined the effects of CHG on phosphorylation of p38 (pp38), JNK-1 (pJNK-1), and ERK 1/2 MAPKs using Western blot analysis. Fig. 1B, top panel 1, shows a representative immunoblot probed with antibodies to pp38. In the middle panel, the same blot was stripped and re-probed with pJNK-1 antibody, and in the bottom panel, the blot was stripped further and probed with nonphosphorylated p38 antibody serving as loading control. THP-1 cells cultured in CHG showed a striking increase in basal levels of pp38 ( Fig.  1C, 2.6-fold) and pJNK-1 (Fig. 1D, 2.3-fold) over NG controls. Treatment of cells with TNF␣ for 5 or 10 min (Fig. 1C) showed a stronger increase in pp38 levels in NG (2.1-fold) compared with CHG (0.6-fold). On the contrary, the pJNK-1 levels ( Fig.  1D) following TNF␣ treatment showed a much stronger and faster increase (2.5-fold) in CHG compared with NG (1.7-fold). The activation profile of the two stress-responsive kinases in CHG showed positive correlation with O 2 Ϫ data. TNF␣-induced elevated pp38 and pJNK-1 levels in CHG were additive. In contrast, the third member of the MAPK family, ERK 1/2, which generally is responsive to stress and mitogenic stimuli, showed no activation by either high glucose or TNF␣ in these monocytic cells. CHG or TNF␣ showed similar trends in U937 cells as THP-1. Increased superoxide levels and elevated phosphorylation of stress-responsive MAPKs, p38, and JNK-1 by CHG demonstrate that ROS induced by high glucose could potentially contribute to downstream signaling via activation of MAPKs.

Elevated Levels of TNF␣ Accumulate in Conditioned Media of U937, THP-1, or Normal Human Monocytes Cultured in CHG
TNF␣ is a potent cytokine involved in inflammation, and elevated levels of TNF␣ are seen in atherosclerotic plaques of diabetics. We therefore evaluated if culturing monocytic cells in CHG could lead to increased TNF␣ accumulation in conditioned medium. CHG alone induced a dramatic increase in TNF␣ accumulation in conditioned media of U937 ( Fig To determine whether the effects of CHG (15 mM)-induced TNF␣ accumulation was due to increased osmolality of CHG, mannitol (9.5 mM) in NG was used as a control. To evaluate if glucose metabolism was required for the increased TNF␣ secretion by CHG, 3-O-methyl glucose (9.5 mM) in NG was tested. Culturing cells in mannitol, or 3-O-methyl glucose showed comparable levels of TNF␣ in the conditioned media as NG controls. These results confirmed that the effect of CHG on elevated TNF␣ accumulation was not due to hyperosmolality of CHG and that glucose metabolism was essential for the elevated levels of TNF␣.

TNF␣ Is Transcriptionally Regulated in U937 Cells Cultured in CHG
To evaluate if CHG-induced TNF␣ is regulated transcriptionally in monocytic cells, we used competitive RT-PCR to monitor the levels of TNF␣ message. Since TNF␣ is regulated both by autocrine and paracrine pathways, we compared the levels of TNF␣ message induced by the cytokine to that induced by CHG. Ethidium bromide staining of products from RT-PCR analysis (Fig. 3A) demonstrated that low levels of TNF␣ (0.25, 0.5, or 1.0 ng/ml), similar to that secreted in conditioned media by CHG culturing, could stimulate TNF␣ message in an autocrine fashion (Fig. 3A, top panel, lanes 5-7). The peak of this stimulation was at 1 ng/ml. The addition of higher levels of TNF␣ (5 ng/ml) did not further increase the levels of the cytokine message (Fig. 3A, top panel, lane 8). Interestingly, CHG induced TNF␣ message at levels similar to that induced by 0.25-0.5 ng/ml TNF␣ (Fig. 3A, top panel, lanes 11 and 12).
PMA at 10 ng/ml induced about 2.4-fold higher message compared with CHG (Fig. 3A, top panel, lane 4) and served as the positive control for TNF␣ message induction. Glyceraldehyde-3-phosphate dehydrogenase was used to check the integrity of RNA (Fig. 3A, bottom panel). TNF␣ cDNA was also amplified using competitive PCR in tubes containing known copy numbers of ICS and hybridized to wells containing either the ICS oligonucleotide or the TNF␣-specific oligonucleotide to determine copy number of TNF␣ message in the different cDNA samples (Fig. 3B). At low levels (0.25-1 ng/ml), TNF␣ showed a dose-dependent increase in the copy number of specific TNF␣ message. Copy number of TNF␣ message induced by CHG was comparable with that induced by TNF␣ (0.5 ng/ml) (Fig. 3B). The competitive RT-PCR data in U937 cells suggest that secreted TNF␣, at levels induced by CHG, could act through an autocrine loop to transcriptionally regulate further TNF␣ message in monocytic cell lines. THP-1 cells showed a similar trend as U937 cells but gave a higher copy number of TNF␣ message following different stimulations (data not shown). Data from RT-PCR studies indicate that the regulation of TNF␣ by CHG in U937 and THP-1 cells is controlled, at least in part, transcriptionally, although post-transcriptional control may also play a significant role in TNF␣ regulation.

FIG. 1. Increased oxidant stress induced in monocytic cells by CHG.
A, higher levels of superoxide production by U937 cells cultured in CHG. Cells were cultured in NG or CHG for 48 h, then in serum-depleted medium with 0.5% bovine serum albumin for 18 h (maintaining the appropriate glucose concentration) and, finally, stimulated with TNF␣ (5 ng/ml) or PMA (10 ng/ml; positive control) for 1 h. LCA was performed as explained under "Materials and Methods" ("Lucigenin Chemiluminescence Assay"). Data are the means Ϯ S.E. from 4 separate experiments and are represented as cpm/10 6 cells. Data were analyzed using analysis of variance followed by Tukey's test. Compared with NG control, superoxide production in CHG, NG/CHG plus TNF␣, or NG plus PMA were significantly different (p Ͻ 0.005). SOD, superoxide dismutase. B, Western blot analyses showing increased phosphorylation of p38 (pp38) and JNK-1 (pJNK-1) stress-responsive kinases by CHG or TNF␣ in THP-1 cells. Cells were cultured in NG versus CHG for 48 h, then in serum-depleted medium as in A and, finally, stimulated with TNF␣ for 5 or 10 min. Western blots were probed with anti-pp38 antibody (top panel), stripped and reprobed with anti-pJNK-1 antibody (middle panel), and finally, stripped and reprobed with nonphosphorylated p38 antibody (bottom panel) to show equal loading. C, the bar graph represents density of pp38 from B. D, the bar graph represents density of pJNK-1 band from B.

Identification of Cis-elements Involved in CHG Mediated
Increased Transcriptional Regulation of TNF␣ TNF␣ is regulated both transcriptionally and post-transcriptionally in response to various stimuli. In our study, the competitive RT-PCR data demonstrated that TNF␣ expression in monocytic cells has a transcriptional component in response to CHG, and therefore, some of the cis-elements in the TNF␣ promoter involved in the induction of this inflammatory cytokine by CHG was evaluated. The 295TNFluc promoter construct selected for our study has been used to study the ciselements involved in TNF␣ promoter regulation following induction by various stimuli such as TNF␣, PMA, LPS, and cyclosporin A (29,30,39). U937 cells cultured in NG or CHG were transfected with the 295TNFluc plasmid and cotransfected with the ␤-galactosidase as internal control, and normalized luciferase activity was measured following stimulation with or without TNF␣ for 18 h. CHG stimulated similar lucif-erase activity as TNF␣-treated NG cells (Table I). However, the effect of CHG plus TNF␣ on luciferase activity was additive over CHG alone (Table I), possibly suggesting the involvement of more than one pathway in transcriptional regulation of TNF␣ by CHG and TNF␣. Specificity for CHG-stimulated luciferase activity was determined by using mannitol (control for osmolality), 3-O-methyl glucose and 2-deoxyglucose (control for glucose metabolism). All of the controls demonstrated near NG levels of luciferase activity (Table I). In some experiments luciferase activity was measured in transfected NG cells stim- , or NG ϩ 10 ng/ml PMA in depletion medium for an additional 24 h. Conditioned medium was collected as described under "Materials and Methods" ("Detection of Secreted TNF␣ in the Culture Supernatant By Specific hTNF␣ Enzyme-linked Immunosorbent Assay"), and TNF␣ was quantified using enzyme-linked immunosorbent assay assay. Data represent the mean Ϯ S.E. of four separate experiments, each sample run in triplicate. C, primary human monocytes (C) were isolated and cultured as outlined under "Materials and Methods" ("Preparation of Human Monocytes"), and TNF␣ levels were quantified in the conditioned media. The data represent the average Ϯ range of two separate donor samples, each sample run in triplicate. TNF␣ secretion by normal human monocytes in HG was significantly higher compared with NG.

Mutational Analysis of the NF-B and AP-1 Sites in the Proximal TNF␣ Promoter
The 295TNFluc plasmid has, in its proximal promoter, consensus binding sites for NF-B and AP-1 transcription factors (Fig. 3C). Site-directed mutagenesis was used to introduce two point mutations each, at the NF-B3 (GGGTTTCTCC mutated to taGTTTCTCC) and AP-1 sites (TGAATGA mutated to gtA-ATGA). Mutation of these sites completely abrogated binding of NF-B or AP-1 to their respective sites using CHG-or TNF␣stimulated nuclear extracts in EMSA (data not shown). Mutations at the NF-B3 and the AP-1 sites reduced CHG-stimulated 295TNFluc activity by 71.8 Ϯ 3.73 and 31.5 Ϯ 4.7%, respectively (Table II). Interestingly, double mutations of both the sites abrogated CHG-stimulated luciferase activity by 89.3 Ϯ 4.31%, confirming the critical contributions of NF-B (major) and AP-1 (minor) in CHG-mediated induction of the TNF␣ promoter.

EMSA to Study the Effect of CHG Culturing on NF-B Activation
The pleiotropic transcription factor NF-B has been shown to be responsive to oxidant stress in endothelial cells, smooth muscle cells, and mesangial cells, all of which play a critical role in atherosclerosis (1,4,11). Mutational analyses demonstrated NF-B as a major transcription factor in CHG-stimulated luciferase activity. Therefore CHG-and/or TNF␣-mediated activation of NF-B in monocytic cell lines was confirmed using EMSA, performed with a 22-base pair oligonucleotide containing the NF-B3 site from the hTNF␣ promoter. Nuclear extracts from U937, THP-1, and fresh human monocytes (from healthy donors) were used for EMSA (Fig. 4, A-D). Culturing U937cells in 5.5 mM (NG), 15 mM (CHG), or 25 mM (VHG) glucose alone for 3 days showed a dose-dependent increase in NF-B binding (Fig. 4A, first, third, and fifth lanes). Phospho-rImager quantitation of data from several experiments showed that NF-B binding was significantly increased (p Ͻ 0.05) in 15 mM (2.1 Ϯ 0.83-fold) and 25 mM glucose (3.2 Ϯ 0.68-fold) over NG (Fig. 4C). Representative gels showing the effect of CHG on NF-B binding are shown in Fig. 4, A and B. Culturing primary human monocytes overnight in 15 mM glucose (HG) also caused a marked increase in NF-B binding over NG controls (Fig. 4D,  first and second lanes). Since CHG leads to accumulation of TNF␣ in the conditioned media, we evaluated the NF-B bind-ing response to TNF␣ in CHG versus NG cells. After 1 h of continuous treatment with TNF␣, NF-B binding in NG and CHG (15 mM) cells were not significantly different (representative gels shown in Fig. 4, A and B, second and fourth lanes). CHG at 25 mM, however, showed an additive increase in NF-B binding with TNF␣ (13.8 Ϯ 1.77-fold) compared with NG ϩ TNF␣ (Fig. 4A, second lane versus sixth lane) in U937 cells, possibly due to significantly higher osmolality of glucose at such high concentration. The time course of NF-B activation following continuous treatment with TNF␣ (5-60 min) showed early activation (5-30 min) in CHG versus NG (Fig. 5, A and B). PhosphorImager quantitation of NF-B binding using nuclear extracts from NG-or CHG-cultured U937 cells treated with TNF␣ for 30 min was performed, and the results are presented graphically in Fig. 4C. Stimulation with TNF␣ for 30 min in CHG showed a significant increase in NF-B binding ( Fig. 4C; 2.01 Ϯ 0.87, p Ͻ 0.05) over CHG alone, suggesting an autocrine role of this cytokine in CHG-stimulated TNF␣ gene regulation. The stimulatory effect of CHG and TNF␣ were specific to NF-B, because the ubiquitous transcription factor Oct-1 was not affected (panels 2 of Figs. 4, A and B).
Specificity for CHG-stimulated NF-B binding was determined by culturing U937 cells in NG plus 9.5 mM mannitol, as a control for osmolality (Fig. 4E, third lane). To determine if glucose metabolism was essential for NF-B activation, nuclear extracts were made from U937 cells cultured in NG plus 9.5 mM 3-O-methyl glucose (Fig. 4E, panel 1, second lane). These results confirm that effects of CHG on NF-B activation were not due to increased osmolality of CHG and that metabolism of glucose was necessary for NF-B activation by CHG.
The subunits of NF-B stimulated by CHG were compared with that stimulated by NG plus TNF␣ in U937 by supershift analysis (EMSA). Both TNF␣ (Fig. 4F) and CHG (Fig. 4G) demonstrated the same supershift profile of the stimulated NF-B complex, showing p50 and p65 as the major subunits. THP-1 cells showed similar shifts as U937 cells (data not shown). Primary human monocytes cultured in HG (15 mM) overnight also demonstrated p50 and p65 as the major components of the activated NF-B complex (Fig. 4D, fourth and fifth  lanes). The basal NF-B complex in NG showed weak supershifting only with the anti-p50 antibody but not with the anti-p65 antibody (data not shown). The supershift data confirm that p50/p65 heterodimers and p65 homodimers are the major transcriptionally active NF-B complexes stimulated by CHG, important in regulating TNF␣ gene expression.

Elements of NF-B Complex Affected by CHG Culturing and/or TNF␣ Treatment
To identify the elements of NF-B complex affected by CHG or TNF␣ stimulation, a time course study was performed in U937 cells cultured in NG or CHG following TNF␣ treatment.  The regulation of the NF-B complex over time was monitored using EMSA and Western blot analysis. Western blots of nuclear extracts were sequentially probed with anti-p65, anti-p50, anti-IB␣, and anti-Histone H1 (loading control) antibodies, and those of cytosolic extracts were sequentially probed with anti-IB␣, anti-IB⑀, and anti-actin (loading control) antibodies with stripping of the blots following each detection. Representative gels from the time course study are shown in Fig. 5, A, C, and E (NG) and B, D, and F (CHG), and some of the key data from the same study are graphically summarized in Fig. 5, G, I, and K (NG) and H, J, and L (CHG). The time course of NF-B activation in NG showed a gradual increase in NF-B binding, with peak binding at 60 min after TNF␣ treatment. The increased binding persisted for 2 h post-TNF␣ stimulation (Fig. 5A, panel 1, fifth and sixth lanes) and fell to basal NG levels by 4 h (data not shown). In CHG, NF-B activation following TNF␣ treatment was strongly detectable as early as 5 min (Fig. 5B, panel 1, second lane). There was a gradual increase in NF-B binding, which peaked around 15-30 min following TNF␣ treatment (Fig. 5B, panel 1, third and fourth lanes) and remained sustained even at 18 h (data not shown). Oct-1 (Fig. 5, A and B, panels 2) served as a control for nuclear extract preparation and was not affected by CHG or TNF␣ at any time points. Under CHG conditions, basal levels of NF-B binding were about 2-fold higher compared with basal NG levels (Fig. 5, A versus B, panel 1, first lane). The NF-B binding data suggest that CHG-cultured U937 cells are primed for a faster responsiveness to TNF␣.
To explore the mechanism for this observed increase in NF-B binding in CHG, we examined the levels of the p50 and p65 subunits in nuclear extracts, since they were the only subunits identified by supershifting. The Western blots from nuclear extracts demonstrated higher p65 levels in CHG (2.67 Ϯ 0.69-fold) compared with the barely detectable levels in NG (Fig. 5, C and D, panel 1, first lane). Basal p50 levels in CHG were also higher (3.25 Ϯ 0.73-fold) compared with NG (Fig. 5, C and D, panel 2, first lane). Mimicking the EMSA (Fig.  5, A versus B), the graphical results of the Western analysis (Fig. 5, I versus J) showed significantly higher levels of p65 in CHG compared with NG (p Ͻ 0.05) at the early time points (5-15 min) of TNF␣ treatment. However at the later time points (60 -120 min) of continuous TNF␣ treatment, p65 levels  Fig. 1A and then treated with TNF␣ at 5 ng/ml for 60 min. Nuclear and cytosolic extracts were made as outlined under "Materials and Methods." EMSAs were performed using 5 g of total protein from the nuclear extract as described under "Materials and Methods." A representative gel is shown in A. B, EMSA for THP-1. Cell treatments were same as for U937 cells. C, graphical demonstration of NF-B following CHG culturing or treatment with TNF␣ for 30 min in NG or CHG conditions. All bars represent fold stimulation over NG controls. Data represent the averages Ϯ S.E. of eight different experiments. NF-B binding in CHG and NG plus TNF␣ was significantly higher than NG controls. NF-B binding in CHG plus TNF␣ was significantly higher than CHG alone. D, EMSA and supershift for peripheral blood human monocytes (normal donors) cultured in NG versus CHG. E, effects of glucose versus glucose analogues on NF-B binding. U937 cells were cultured in NG, NG ϩ 9.5 mM 3-O-methyl glucose, NG ϩ 9.5 mM mannitol or CHG, nuclear extracts were made, and EMSA was run as in A. In A, B, and D, the upper panels represent NF-B complexes, and the lower panels represent Oct-1 complexes. F and G, identification of NF-B binding activities in U937 cells. Supershift assays were performed following treatment with TNF␣ in NG (F) or with CHG (G) with the antibodies indicated.
in NG and CHG were equal, drawing similarity to the EMSA binding data. These results confirm that CHG-cultured U937 cells are primed to respond faster to the inflammatory cytokine TNF␣ compared with NG cultured cells.
To further elucidate the mechanism of NF-B activation, the levels of the inhibitory IB subunits (IB␣, and IB⑀) were checked in cytosolic and nuclear extracts under basal conditions and following TNF␣ challenge in NG and CHG. TNF␣ mediates degradation of pre-existing IB␣, releasing the transacting NF-B complex, allowing its translocation to the nucleus (37). In addition, the re-synthesized IB␣ translocates to the nucleus, binding to the p50⅐p65 complex, thereby exposing the nuclear export signal for the removal of the latter out of the nucleus (37). We therefore examined re-synthesized levels of IB␣ in the nuclear extracts of NG-versus CHG-treated U937 cells following 60 -120 min of continuous TNF␣ treatment. The re-synthesized levels of IB␣ in NG were significantly higher (2.87 Ϯ 0.78-fold) compared with its CHG counterpart (Fig. 5, C  versus D, panel 3, fifth and sixth lanes). The Western blots of the cytosolic extracts from the time course study were also sequentially probed with anti-IB␣ and anti-IB⑀ antibodies with intermediate stripping. We report here for the first time that the level of IB␣ in the cytosol of NG-cultured cells was significantly higher (2.65 Ϯ 0.53-fold) compared with that observed in CHG (Fig. 5, E versus F, panel 1, first lane). Resynthesized levels of IB␣ in the cytosol following 60 or 120 min of TNF␣ treatment were also higher in NG versus CHG cells (1.85 Ϯ 0.71-fold, Fig. 5, E versus F, panel 1, sixth lane), lending additional support to the higher basal p65 levels in CHG-cultured cells. In contrast to IB␣, levels of IB␤ (data not shown) and IB⑀ were not affected at any of the time points evaluated (Fig. 5, E versus F, panel 2).
Data from the time course study indicated that U937 cells cultured in CHG are primed for faster response to TNF␣ and show a stronger rapid activation of NF-B compared with NG cells. The levels of IB␣ were higher in the cytosol of NG versus CHG cells, suggesting a possible mechanism for the higher basal levels of p65 in CHG, which could further explain the increased DNA binding and luciferase activity.

Effects of Antioxidants on CHG-induced NF-B Activation
U937 cells were incubated with antioxidants NAC (100 M) and PDTC (50 M) for 1 h before culturing them in CHG for 24 h in the presence of the inhibitors, following which nuclear extracts were made for EMSA. The antioxidants showed striking inhibition of CHG-induced NF-B binding by blocking the translocation of the p65 subunit to the nucleus (Fig. 6, A and B,  third lane versus eighth lane). In addition, the antioxidants also partially blocked TNF␣-induced NF-B activation in NG and CHG (Fig. 6, A and B, second and fourth lanes versus fifth  and sixth lanes). PhosphorImager data from multiple EMSA are tabulated in Table III.

Activation of AP-1 by High Glucose but Not TNF␣
Mutational analysis of the TNF␣ promoter confirmed a cooperative role of AP-1 with NF-B in CHG-mediated 295TN-Fluc activity. The U937 nuclear extracts from the time course study (Fig. 5, A and B) were used to perform EMSA with a 17-base pair oligonucleotide encompassing the AP-1 site from the proximal TNF␣ promoter ("Materials and Methods"). AP-1 activation was dramatically stimulated by CHG over NG (Fig.  7A, first lane versus seventh lane). TNF␣ treatment showed a rapid increase in AP-1 activation in NG (Fig. 7A, first lane  versus second through sixth lanes). However, in CHG cells TNF␣ down-regulated AP-1 activation compared with that observed in CHG alone (Fig. 7A, seventh lane versus  tenth lane). Activation of AP-1 was also confirmed in normal human monocytes cultured overnight in HG (Fig. 7B). Supershift analysis confirmed c-Fos and c-Jun to be a part of the CHG-activated AP-1 complex (Fig. 7B, fourth and fifth lanes). Specificity of the AP-1 complex was confirmed by competition of the specific complex with cold AP-1 probe (Fig. 7B, third lane), and the complex was supershifted with c-Fos and c-Jun antibodies (Fig. 7B, fourth through sixth lanes). The PKC inhibitor GF109203X blocked CHG-induced AP-1 binding by 68% and reversed the down-regulation observed in CHG following TNF␣ greater than 75%, further suggesting involvement of similar PKC isoforms in CHG and TNF␣-mediated activation of AP-1 (data not shown).

Mechanism of CHG-induced Monocyte Activation
Pharmacological inhibitors were used to further evaluate the mechanism of CHG-induced monocyte activation. Antioxidants, a p38 MAPK inhibitor, and PKC inhibitors were used to block each of the activated pathways so far identified to be important in monocyte activation, and the counter effects of each inhibitor were evaluated separately or in combination. In our study, CHG induced oxidant stress (higher O 2 Ϫ ) in monocytes and increased phosphorylation of oxidant stress-sensitive MAPKs (p38 and JNK-1). The antioxidants NAC and PDTC had similar counter effects on CHG-or TNF␣-induced NF-B binding (Table III) and 295TNFluc activity (Table IV) or CHGinduced TNF␣ secretion (Table V). The inhibitory effects of PDTC, known to function as an antioxidant and a NF-B inhibitor, were slightly greater than that seen with NAC. These results suggest that CHG-and TNF␣-induced monocyte activation could be mediated at least in part by ROS. Since the antioxidants were 60 -70% effective, ROS-independent pathways must also contribute to monocyte activation by CHG and TNF␣.
We show here for the first time the inhibitory effects of the   7. A, time course of AP-1 activation in U937 cells following treatment with TNF␣. EMSA was run using samples from the time course experiment shown in Fig. 5, A and B, using AP-1 probe as explained under "Materials and Methods" ("Mutational Analysis of the NF-B3 and AP-1 Sites in the TNF␣ Promoter"). B, EMSA and supershift analysis of peripheral blood human monocyte (normal donors) samples from Fig. 4D with AP-1 probe. Ab, antibody.
p38 MAPK inhibitor, SB202190, on CHG-induced monocyte activation (Tables III-V). The partial blocking effect of the inhibitor on NF-B activation, induced by CHG or TNF␣ in monocytes, is interesting because the effect of p38 MAPK on NF-B activation through a TRAF2-dependent, but NIK-independent, pathway was recently reported (46). Inhibition of monocyte activation by the p38 inhibitor was partial. We therefore evaluated the role of PKC inhibitors in counteracting monocyte activation by CHG or TNF␣ (data not shown). Interestingly, the PKC inhibitors, calphostin C or GF109203X, had similar counter effects on monocyte activation stimulated by CHG and TNF␣ separately or in combination (data not shown). The p38 MAPK inhibition data suggest that downstream of p38, CHG or TNF␣ possibly use similar signals for monocyte activation. The combination of PDTC and GFX blocked 89%, and combination of p38 MAPK inhibitor and GFX blocked more than 92% of all aspects of monocyte activation (data not shown), suggesting the importance of both oxidant stress and PKC pathways in monocyte activation by CHG and/or TNF␣. The inhibitor data suggest that ROS-dependent pathways are important in monocyte activation by CHG or TNF␣. In addition, p38 MAPK and PKC also appear to be crucial upstream kinases for CHG-and/or TNF␣-induced increased TNF␣ expression in monocytes, possibly via ROS-dependent and -independent mechanisms. DISCUSSION In this study we demonstrated that CHG induced increased oxidant stress in monocytic cells by stimulating elevated levels of O 2 Ϫ (ROS) and increased phosphorylation of stress-responsive MAPKs, p38, and JNK-1. Such oxidant stress incited monocyte activation, measured by elevated TNF␣ secretion. We have presented evidence that CHG culturing of monocytic cell lines or normal human monocytes caused a dramatic increase in the release of the inflammatory cytokine TNF␣, due in part to enhanced TNF␣ mRNA transcription. Furthermore, the transcriptional activation was mediated in part by ROS via activation of NF-B and AP-1, as suggested by a number of observations. A 295-base pair fragment immediately upstream of the proximal TNF␣ promoter containing canonical NF-B and AP-1 binding sites reproduced the effect of CHG on TNF␣ transcription in a luciferase reporter assay. Mutational analy-ses of the NF-B and AP-1 sites showed cooperativity of the two transcription factors in regulating 295TNFLuc activity. Gel shift and supershift analysis using these binding sites showed activation of both the p65 and p50 subunits of NF-B and the c-Fos and c-Jun subunits of AP-1 in nuclear extracts of CHG lysates. Western blot analyses demonstrated elevated nuclear levels of p65 and p50 and decreased cytosolic levels of IB␣ in the CHG-treated monocytes. The contention that ROS act as a key intermediate in this pathway was also supported by several lines of evidence. CHG increased O 2 Ϫ levels comparable with that found with TNF␣ or PMA. We show evidence here for the first time that CHG stimulated phosphorylation of oxidant stress-sensitive p38 and JNK-1 MAPKs in monocytes. More importantly, antioxidants and p38 inhibitors partially counteracted the effects of CHG on TNF␣ production, the 295TNFluc reporter activity, activation of NF-B, and repression of IB␣. The p38 inhibitor blocked monocyte activation by 60%. A combination of the p38 and PKC inhibitors or the PDTC and PKC inhibitors blocked monocyte activation more than 90%. Our study, using pharmacological inhibitors, further suggested that CHG activated monocytic cells by ROS-dependent and -independent pathways. CHG-mediated monocyte activation involved upstream kinases such as PKC and MAPKs, affecting increased transcription of TNF␣ via coordinate activation of downstream transcription factors NF-B and AP-1.
We used U937 and THP-1 cells in our study, since these cells have been routinely used to replace human monocytes, the latter of which is difficult to obtain in highly pure and unactivated state. Monocyte activation was not affected by hyperosmolality of CHG over NG, since adding mannitol to NG culture media did not cause monocyte activation. Our results also suggested that glucose metabolism was essential for monocyte activation, since analogues of glucose, which are not metabolized, such as 3-O-methyl glucose (nonphosphorylated but transported) or 2-deoxyglucose (phosphorylated but not metabolized or transported) did not activate monocytes over NG controls.
Hyperglycemia causes altered redox changes by altering the NADH/NAD ϩ ratio in the cells, leading to the activation of key inflammatory signals and gene expression (1,4,12). Studies in several diabetic microvascular and macrovascular models demonstrate the role of oxidant stress leading to diabetic complications (3). Monocytic cells are capable of O 2 Ϫ production in response to a variety of stimuli (6). Our data using monocytic cell lines demonstrated that CHG-mediated O 2 Ϫ release was equivalent to TNF␣-or PMA-stimulated O 2 Ϫ release in NG, indicating that CHG generated ROS as potently as the inflammatory cytokine or the phorbol ester. The role of oxygen radicals as second messengers for inducing expression of various inflammatory genes in monocyte (macrophage chemotactic protein 1 (MCP-1) and colony stimulating factor-1 (CSF-1)) was reported recently (45). Naturally occurring cellular factors such as glutathione (GSH) are important in maintaining the redox potential in cells and in mounting a defense against oxidant stress in a thiol-sensitive fashion (4). We therefore used two antioxidants particularly effective against intracellular thiol levels, NAC and PDTC, to block monocyte activation. In addition to its role as an antioxidant, PDTC is also an inhibitor of NF-B activation. The antioxidants, when present before and during culturing of monocytes in CHG demonstrated significant but partial counter effect on NF-B activation, 295TNFluc activity, and TNF␣ secretion. PDTC showed additional blockade of all aspects of monocyte activation, strengthening the notion that NF-B is important in monocyte activation by CHG. Our results therefore suggest that both ROS-dependent and -independent pathways are important in monocyte activation by CHG. This study provides evidence for the first time for increased phosphorylation of p38 and JNK-1 stress-responsive MAPKs by CHG in monocytes. Basal levels of pp38 and pJNK-1 were higher in CHG-cultured THP-1 and U937 cells. Since CHG stimulated release of low levels of TNF␣ into the conditioned media, we evaluated if the monocytic cells cultured in CHG were responsive to further stimulation with TNF␣. There was a strong overall increase in the levels of pp38 and pJNK-1 in CHG-cultured THP-1 cells, confirming that CHG did not blunt the TNF␣ responsiveness for MAPK activation in these cells. The overall levels of pp38 or pJNK-1 were higher in CHG over NG following TNF␣ treatment. CHG and TNF␣ could potentially use more than one pathway for activation of these MAPKs. Both p38 and JNK-1 are activated by oxidative stress (21), and since CHG culturing significantly increased higher O 2 Ϫ levels over NG, it may be logical to suggest that oxidative stress (ROS) generated by CHG could at least in part account for the observed activation of these two MAPKs in the monocytic cells. Glucose-and TNF␣-induced p38 activation was recently reported in rat aortic smooth muscle cells (19,20,22). In our study, TNF␣ stimulated similar levels of O 2 Ϫ in NG or CHG cells, but activation of MAPKs was additive following TNF␣ treatment in CHG. Therefore increased phosphorylation of p38 or JNK-1 could only be partially accounted for by CHG-stimulated ROS-dependent pathways. The antioxidant PDTC showed a stronger inhibitory effect on monocyte activation in CHG plus TNF␣ compared with CHG alone or NG plus TNF␣. The p38 inhibitor showed similar counter effects on monocyte activation in CHG alone, NG plus TNF␣, or CHG plus TNF␣. Therefore, it may be reasonable to conclude that more than one pathway is involved in CHG-or TNF␣-mediated p38 activation. However the counter effect of p38 inhibitor on monocyte activation suggested that downstream of p38, CHG or TNF␣ possibly use similar signals for monocyte activation.
CHG-stimulated ROS and MAPKs were only partially responsible for monocyte activation (inhibitor data), and activation of PKC by high glucose is well established (25,44), PKC being a known activator of NF-B (37). We evaluated the importance of the PKC pathway in CHG-and/or TNF␣-mediated monocyte activation. PKC activation has been documented both by ROS-dependent and -independent pathways (7,13,26). We evaluated the counter effect of PKC inhibitor GF109203X on monocyte activation. The PKC inhibitor blocked different aspects of monocyte activation between 60 -65%, similar to the p38 inhibitor, and the inhibition profile was very similar in CHG or with TNF␣. Interestingly, GFX in combination with p38 MAPK inhibitor or GFX in combination with PDTC greatly blocked monocyte activation (ϳ90%), confirming the multilevel regulation of the activation process by oxidant stress and/or inflammatory cytokine. Further evidence in support of our finding was recently reported in smooth muscle cells (19,20) and kidney mesangial cells (24), showing glucose-mediated activation of PKC and p38.
The pleiotropic NF-B/Rel family of transcription factor has been implicated in a wide variety of inflammatory response (34,35,38). The transacting NF-B complex, stimulated by high glucose or TNF␣, was similar, composed of p65/p50 heterodimer. TNF␣ stimulation of monocytic cells showed an early activation phase (5-15 min) and a late activation phase (Ͼ30 min), which was possibly PKC-dependent. Furthermore, the roles of p38 and PKC in activation of NF-B has been well documented (19,22). In U937 cells, the late phase (30 -120 min) of NF-B activation following TNF␣ treatment was similar in NG and CHG, suggesting that CHG culturing possibly depleted PKC pools, commonly used by both CHG and TNF␣ for the late phase of NF-B activation. Nevertheless, the early phase of NF-B activation was stronger; the 295TNFluc activity as well as TNF␣ accumulation in the conditioned media were all greater in CHG cells following treatment with TNF␣. These results suggest that monocytic cells cultured in CHG were responsive to further challenge with inflammatory cytokines, supporting the vicious inflammatory potential of CHG.
A new finding of our study was the potential mechanism for the CHG-stimulated NF-B activation. The basal level of the inhibitory subunit IB␣ was reduced by 2-fold in the cytosol of the CHG cells compared with NG. We can therefore rationalize that the higher monocyte activation in CHG versus NG could be attributed to the overall lower levels of the inhibitor subunit (IB␣), resulting in failure to sequester all the transacting NF-B complex in the cytosol. The conclusion was further strengthened by the fact that CHG cells had 2-fold lower levels of resynthesized IB␣ in the cytosol following TNF␣ treatment, as was the level of IB␣ in the nuclear extracts, responsible for the export of the transacting complex. This could explain the persistent NF-B levels in the nucleus, the elevated luciferase activity, and the increased levels of TNF␣ mRNA and protein observed in CHG-cultured cells. The results suggested a possible impairment in the overall synthesis, specifically of IB␣, since IB⑀ was not affected.
Another important finding of this study was that CHG-cultured cells were resistant to further AP-1 stimulation by TNF␣. Since CHG and AP-1 are both well known activators of PKC, the results indicate that CHG and TNF␣ may stimulate the same PKC isoforms needed for AP-1 stimulation. TNF␣-mediated AP-1 activation could be restored by pretreating the cells with PKC inhibitors but not with the p38 inhibitor SB202190. 2 Examples of such resistance in the activation of kinases or transcription factors were seen in secondary challenges with LPS (34) or in long term challenges with phorbol esters (30). The results from the mutational analysis of the NF-B3 and AP-1 sites in the 295TNFluc promoter confirmed that NF-B is the major transcription factor involved in regulating TNF␣ promoter activity, but maximal transcriptional regulation is achieved by the concerted effort of NF-B and AP-1 in monocytes.
Using a combination of cellular, biochemical, and molecular assays, we have demonstrated the involvement of ROS and MAPKs in monocyte activation and transcriptional regulation of inflammatory gene expression by CHG, simulating a diabetic state. We have also elucidated a possible mechanism for constitutive NF-B activation in CHG by demonstrating for the first time lower basal levels of inhibitor IB␣ in CHG-cultured U937. The levels of TNF␣ secreted in CHG could stimulate further production of this cytokine by autocrine control. It is therefore tempting to propose a simple model (Fig. 8) for monocyte activation in hyperglycemia.