Endothelial cells actively participate in inflammatory events by regulating leukocyte recruitment via the expression of inflammatory genes such as E-selectin, VCAM-1, ICAM-1, IL-6, IL-8, and cyclooxygenase (COX)-2. In this study we showed by real-time RT-PCR that activation of human umbilical vein endothelial cells (HUVEC) by TNF-α and IL-1β differentially affected the expression of these inflammatory genes. Combined treatment with TNF-α and IL-1β resulted in nonadditive, additive, and even synergistic induction of expression of VCAM-1, IL-8, and IL-6, respectively. Overexpression of dominant-negative inhibitor κB protein blocking NF-κB signaling confirmed a major role of this pathway in controlling both TNF-α- and IL-1β-induced expression of most of the genes studied. Although dexamethasone exerted limited effects at 1 μM, the thioredoxin inhibitor MOL-294, which regulates the redox state of NF-κB, mainly inhibited adhesion molecule expression. Its most pronounced effect was seen on VCAM-1 mRNA levels, especially in IL-1β-activated endothelium. One micromolar RWJ-67657, an inhibitor of p38 MAPK activity, diminished TNF-α- and IL-1β-induced expression of IL-6, IL-8, and E-selectin but had little effect on VCAM-1 and ICAM-1. Combined treatment of HUVEC with MOL-294 and RWJ-67657 resulted in significant blocking of the expression of E-selectin, IL-6, IL-8, and COX-2. The inhibitory effects were much stronger than those observed with single drug treatment. Application of combinations of drugs that affect multiple targets in activated endothelial cells may therefore be considered as a potential new therapeutic strategy to inhibit inflammatory disease activity.
- inflammatory gene expression
- anti-inflammatory drugs
- combination treatment
endothelial cells form the natural barrier between the blood and surrounding tissue. During inflammation they control leukocyte trafficking and actively participate in angiogenesis through differential expression of inflammation- and angiogenesis-associated genes, including cytokines, chemokines, growth factors, and adhesion molecules (4). Prevention of activation of endothelial cells has been suggested to be beneficial in the treatment of chronic inflammatory diseases like rheumatoid arthritis (36) and inflammatory bowel disease (23). Furthermore, their position in the body makes them an easily accessible target for systemically applied drugs.
The proinflammatory cytokines TNF-α and IL-1β, having a similar but not identical effect on gene expression, are often present simultaneously in chronic inflammatory diseases (29, 31). They exert a prominent effect on the expression of proinflammatory genes in endothelial cells. This effect takes place predominantly through activation of intracellular signaling pathways involving NF-κB and p38 MAPK (13, 39). The transcription factor NF-κB is present in the cytoplasm of unstimulated cells in an inactive form because of its association with the inhibitory protein inhibitor κB (IκB). On cytokine activation, degradation of IκB and subsequent nuclear translocation of active NF-κB takes place (18). The p38 MAPK activation pathway engages diverse upstream kinases responsible for p38 MAPK activation as well as downstream substrates (26). In endothelial cells both NF-κB and p38 MAPK are involved in the regulation of the expression of genes encoding E-selectin, VCAM-1, ICAM-1, IL-6, IL-8, and cyclooxygenase (COX)-2, among others (7, 11, 30). The regulation takes place at transcriptional and posttranscriptional levels (28, 32, 39).
Both activated NF-κB and p38 MAPK have been shown to be present in rheumatoid arthritis and inflammatory bowel disease lesions and are therefore interesting targets for pharmacological intervention (17, 33, 37, 40). However, inhibition of NF-κB or p38 MAPK can have the serious drawback of undesired toxic effects on nondiseased cells (3, 38). Incorporation of such inhibitors in endothelial cell-specific drug targeting systems can theoretically overcome these undesired side effects (8). The antioxidant and metal-chelating compound pyrrolidine dithiocarbamate (PDTC) (21), the glucocorticoid dexamethasone (Dex) (43), the thioredoxin inhibitor methyl-(4R/S)-4-hydroxy-4-[((5S,8S)/(5R,8R))-8-methyl-1,3-dioxo-2-phenyl-2,3,5,8-tetrahydro-1H-[1,2,4]triazolo[1,2-a]pyridazin-5-yl]-2-butynoate (MOL-294) (19), and the p38 MAPK inhibitor 4-(4-(4-fluorophenyl)-1-(3-phenylpropyl)-5-(4-pyridinyl)-1H-imidazol-2-yl)-3-butyn-1-ol (RWJ-67657) (27) are potential candidates for incorporation in drug targeting constructs. However, limited data are available on quantitative comparison of the effects of these anti-inflammatory drugs on endothelial cell gene expression under proinflammatory conditions.
In the present study we investigated the effects of TNF-α, IL-1β, and a combination of TNF-α and IL-1β on the kinetics and levels of expression of the proinflammatory genes E-selectin, VCAM-1, ICAM-1, IL-6, IL-8, and COX-2 by human umbilical vein endothelial cells (HUVEC). The importance of NF-κB signaling in TNF-α- and IL-1β-induced gene expression was investigated by overexpression of an IκB mutant inhibiting NF-κB signal transduction. Furthermore, we analyzed the effects of the above-mentioned drugs on the expression levels of the proinflammatory genes and their capacity to potentiate their inhibitory effects when added simultaneously. To quantitatively compare the effects of activators and drugs, real-time RT-PCR analysis of mRNA levels and, in separate experiments, ELISA and flow cytometric analyses of the proteins produced were performed.
MATERIALS AND METHODS
HUVEC obtained from the Endothelial Cell Facility UMCG (Groningen, The Netherlands) were isolated from two umbilical cords to circumvent donor bias and cultured as previously described (20). In short, the cells were cultured on 1% gelatin-precoated tissue culture flasks (Corning, Costar) at 37°C under 5% CO2-95% air. The endothelial culture medium consisted of RPMI 1640 supplemented with 20% heat-inactivated FCS, 2 mM l-glutamine, 5 U/ml heparin, 100 IU/ml penicillin, 100 μg/ml streptomycin, and 50 μg/ml endothelial cell growth factor supplement extracted from bovine brain. On confluence, cells were detached from the surface by trypsin-EDTA (0.5 and 0.2 mg/ml in PBS) and split at a 1-to-3 ratio into 6- or 12-well tissue culture plates (Corning). In all experiments HUVEC were used up to passage 4. All experiments were performed with confluent HUVEC monolayers, except when adenovirus-encoded dominant-negative (dn)IκB was used, for which confluence was 70%. The experiments were performed with at least two and in most cases four different HUVEC isolates in independent experiments. Data shown are representative of the data from the different experiments.
Activation of HUVEC.
Confluent HUVEC were activated with 1 and 10 ng/ml TNF-α (Boehringer, Ingelheim, Germany) and 1 and 10 ng/ml IL-1β (R&D Systems, Minneapolis, MN), added separately or in combination, for 6 h (early gene expression) and 24 h (late gene expression). After incubation cells were microscopically analyzed with regard to their morphology and consistently were found to be adherent and viable. 3-(4,5-Dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS; Promega Benelux, Leiden, The Netherlands) assays were executed according to the manufacturer's protocol to corroborate the light microscopy analysis.
Recombinant, replication-deficient adenovirus Ad5IκBAA, hereafter referred to as dnIκB, was a gift from Dr. C. Trautwein from the Medical School of Hannover, Germany. Adenovirus contained an IκBα sequence, in which serine at positions 32 and 36 was substituted by alanine and which was fused to an influenza A virus hemagglutinin (HA) tag. The expression was controlled by the cytomegalovirus promoter/enhancer (15). Virus was grown on HEK293 cells and purified from cell lysates by banding twice on CsCl gradients. Virus was desalted with a 10-kDa slide-A-lyzer (Pierce Chemical, Rockford, IL) in HEPES-sucrose buffer, pH 8.0 and stored at −80°C. Viral particles (vp) were determined by UV spectrophotometric analysis at 260 nm. Furthermore, a standard limiting dilution assay was performed to determine the vp-to-plaque-forming unit (pfu) ratio. As a control, adenovirus Ad5LacZ, containing the Escherichia coli β-galactosidase gene (12), was grown and purified as described above.
Virus infection protocol.
For the transduction of HUVEC with dnIκB or the control virus Ad5LacZ, HUVEC were plated at 12,500 cells/cm2 in six-well tissue culture plates (Costar) and cultured overnight before actual transduction. The viral vectors, diluted in DMEM (GIBCO, Paisley, UK) without FCS, were added at 500 pfu/seeded cell (corresponding to 7.5 × 103 vp/cell) and incubated for 90 min at 37°C. The incubation medium was then replaced by endothelial culture medium. Cells were subsequently incubated for 24 h before activation to allow transgene expression.
Western blot analysis of dnIκB expression in HUVEC.
After 24 h of culturing, cells were detached from the surface by trypsin-EDTA treatment, lysed in the cell culture lysis reagent (Promega, Madison, WI), and sonicated twice at 4°C for 5 s. After centrifugation for 10 min at 10,000 g and 4°C, cleared cell lysates were collected, and the protein content was determined with Bradford protein assay reagent (Bio-Rad Laboratories, Hercules, CA), with BSA as the standard. Samples were then mixed with reducing SDS sample buffer and boiled for 5 min, and 30 μg of protein was loaded on SDS-PAGE 10% acrylamide gel. After separation proteins were electrophoretically transferred on a nitrocellulose membrane (Bio-Rad Laboratories). Blots were blocked in blocking buffer containing 5% nonfat dry milk in PBS-0.1% Tween for 2 h. Next, blots were incubated for 1 h with rabbit anti-HA-probe antibody (Santa Cruz Biotechnology, Santa Cruz, CA; dilution 1:200 in blocking buffer) for dnIκB detection and rabbit anti-IκBα antibody (Santa Cruz Biotechnology; dilution 1:20 in blocking buffer) for both endogenous and dnIκB detection. Blots were washed with PBS-0.1% Tween and incubated for 1 h with horseradish peroxidase-conjugated swine anti-rabbit antibody (Dako, Glostrup, Denmark) diluted 1:2,000 in blocking buffer. After washing as described above, detection was performed with enhanced chemiluminescence detection reagents (Amersham, Arlington Heights, IL) according to the manufacturer's protocol.
Incubation of HUVEC with drugs.
The following drugs were used: PDTC (Sigma, Zwijndrecht, The Netherlands), Dex (9α-fluoro-16α-methyl-11β,17α,21-trihydroxy-1,4-pregnadiene-3,20-dione; Genfarma, Maarssen, The Netherlands), MOL-294 (kindly provided by Dr. M. Kahn from Pacific Northwest Research Institute, Seattle, WA), and RWJ-67657 (kindly provided by Johnson & Johnson Pharmaceutical R&D, Raritan, NJ). Stock solutions (10 mM) of PDTC, Dex, MOL-294, and RWJ-67657 were prepared in DMSO (Merck, Darmstadt, Germany). The stock solutions were diluted in endothelial culture medium to final concentrations as indicated in each experiment.
Anti-inflammatory drugs were added to confluent HUVEC 1 h before activation by TNF-α or IL-1β. After 6 and 24 h of stimulation cells were analyzed microscopically with regard to morphology and viability, after which cells or supernatants were subjected to further analysis. The occurrence of toxic effects of drugs to the cells was excluded by MTS assay.
RNA isolation and real-time RT-PCR analysis.
Total RNA was isolated with an Absolutely RNA Microprep Kit (Stratagene, Amsterdam, The Netherlands) according to the protocol of the manufacturer. RNA was analyzed qualitatively by gel electrophoresis and quantitatively with a RiboGreen RNA Quantitation Kit (Molecular Probes Europe, Leiden, The Netherlands). One microgram of total cellular RNA was subsequently used for the synthesis of first-strand cDNA with SuperScript III RNase H minus reverse transcriptase (Invitrogen, Breda, The Netherlands) in a 20-μl final volume containing 250 ng of random hexamers (Promega) and 40 units of RNase OUT inhibitor (Invitrogen). After RT reaction cDNA was diluted with distilled water to 100 μl, and 1 μl cDNA was used for each PCR reaction. Exons overlapping primers and minor groove binder (MGB) probes used for real-time RT-PCR were purchased as Assay-on-Demand from Applied Biosystems (Nieuwekerk a/d IJssel, The Netherlands): housekeeping gene GAPDH (assay ID Hs99999905_m1), endothelial cell marker CD31 (platelet endothelial cell adhesion molecule 1, Hs00169777_m1), E-selectin (Hs00174057_m1), VCAM-1 (Hs00365486_m1), ICAM-1 (Hs00164932_m1), IL-6 (Hs00174131_m1), IL-8 (Hs00174103_m1), and COX-2 (Hs00153133_m1). The final concentration of primers and MGB probes in TaqMan PCR MasterMix (Applied Biosystems, Foster City, CA) for each gene was 900 and 250 nM, respectively. As controls, RNA samples not subjected to reverse transcriptase were analyzed to exclude unspecific signals arising from genomic DNA. Those samples consistently showed no amplification signals.
TaqMan real-time RT-PCR was performed in an ABI PRISM 7900HT Sequence Detector (Applied Biosystems). Amplification was performed with the following cycling conditions: 2 min at 50°C, 10 min at 95°C, and 40–45 two-step cycles of 15 s at 95°C and 60 s at 60°C. Triplicate real-time RT-PCR analyses were executed for each sample, and the obtained threshold cycle values (Ct) were averaged. According to the comparative Ct method described in the ABI manual (http://www.appliedbiosystems.com), gene expression was normalized to the expression of the housekeeping gene GAPDH, yielding the ΔCt value. The average ΔCt value obtained from resting, nontreated HUVEC was then subtracted from the average ΔCt value of each sample subjected to the experimental conditions described, yielding the ΔΔCt value. The gene expression level, normalized to the housekeeping gene and relative to the control sample, was calculated as 2−ΔΔCt.
IL-6 and IL-8 production measured by ELISA.
In the designated drug combination treatment experiments, HUVEC medium was harvested, centrifuged, and stored at −20°C before cytokine quantification by ELISA. Ninety-six-well plates (Costar) were precoated with MoAb.CLB.MIL6/16 (CLB, Amsterdam, The Netherlands) diluted 1:1,000 in PBS for IL-6 and with MoAb.anti-IL-8 (R&D Systems) for IL-8 analysis. After blocking with 2% BSA-0.05% Tween in PBS, samples were incubated for 2 h in incubation buffer containing 0.2% gelatin-0.05% Tween in PBS. After being washed, bound IL-6 or IL-8 was detected with biotinylated polyclonal swine anti-human IL-6 (CLB) or polyclonal swine anti-human IL-8 (R&D Systems), respectively, in combination with streptavidin-E+ (CLB). Peroxidase activity was determined with tetramethylbenzidine (Roth, Karlsruhe, Germany) as substrate. IL-6 and IL-8 levels were calculated in the linear range of the assay from a standard curve (10–1,000 pg/ml) with recombinant human (rh)IL-6 (R&D Systems) and rhIL-8 (R&D Systems).
Flow cytometric analysis of cell adhesion molecule expression.
HUVEC pretreated with MOL-294 and/or RWJ-67657 as indicated were stimulated for 6, 12, and 24 h with TNF-α and subsequently detached from the wells by a short treatment with trypsin. After being washed with PBS-5% FCS, cells were incubated for 45 min on ice with mouse monoclonal antibodies against human E-selectin (H18/7-acb), VCAM-1 (E1/6-aa2), and ICAM-1 (hu5/3–2.1) (all 3 antibodies kindly provided by Dr. M. A. Gimbrone from Harvard Medical School, Boston, MA) and CD31 (DakoCytomation, Glostrup, Denmark). After a PBS-5% FCS wash, detection was performed with FITC-conjugated rabbit anti-mouse antibody for 45 min on ice. Cells were fixed with 0.5% paraformaldehyde-PBS, after which flow cytometric analysis was performed on an Epics-Elite flow cytometer (Coulter Electronics, Mijdrecht, The Netherlands). A total of 5,000 events were analyzed per sample. Nonspecific staining was assessed by incubation with mouse IgG monoclonal antibodies specific for an irrelevant antigen as primary antibody. The mean fluorescence intensity values of these controls were consistently found to be <5.
Statistical significance of differences for experiments with adenovirus and single cytokine or drug treatment was studied by means of the two-sided Student's t-test, assuming equal variances. Differences were considered to be significant when P < 0.05.
Linear mixed models and ANOVA were used for data analysis of combination treatment experiments to address the significance of observed effects. Sidak and Tukey adjustment methods were used to adjust the confidence intervals and significance values to account for multiple comparisons. All analyses were performed with SPSS 12.01. Differences were considered to be significant when P < 0.05.
In endothelial cells different gene expression patterns are induced by TNF-α and IL-1β.
To study gene expression profiles induced by TNF-α and IL-1β, we activated HUVEC with 1 and 10 ng/ml of both cytokines for 6 and 24 h and examined the expression of E-selectin, VCAM-1, ICAM-1, IL-6, IL-8, and COX-2 genes. The 6 -h time point reflects early gene expression profiles, whereas the 24-h time point reflects late gene expression during prolonged activation, possibly comparable with prolonged exposure of the endothelium to activators in chronic inflammatory disorders.
Between experiments, basal Ct values of proinflammatory genes in nonstimulated HUVEC samples ranged between 1 and 3 Ct units, whereas for each gene the Ct value obtained after 6 h of cytokine activation reached levels similar in each experiment (data not shown). These results imply that although the basal expression levels of genes might differ because of the heterogeneity of endothelial cells obtained from different donors, the activation status achieved by cytokine signaling was similar in all experiments. Expression of the endothelial cell marker CD31 remained constant under all conditions studied (Table 1), a result corroborating previously published studies (22). Patterns of proinflammatory gene expression, however, differed markedly with regard to the activator used and the incubation time studied (Fig. 1 and Table 1). At both time points studied TNF-α induced the expression of adhesion molecules VCAM-1 and ICAM-1 to a higher extent than IL-1β, whereas IL-1β more profoundly affected IL-6, IL-8, and COX-2. At 6 h, the level of the gene expression was independent of the concentration of each activator, the only deviation being IL-1β-induced IL-6 and ICAM-1 expression (P > 0.05). Notably, for TNF-α-induced activation at 24 h, a higher concentration of TNF-α induced gene expression to a higher level (only for TNF-α-induced COX-2 and IL-6; P > 0.05). In time, E-selectin, VCAM-1, and ICAM-1 expression levels diminished, whereas IL-8 and COX-2 mRNA levels increased (P < 0.05). IL-6 exhibited a mixed response, as its level of expression induced by TNF-α was significantly higher at 24 h, whereas IL-1β-induced expression was significantly higher at 6 h.
Adhesion molecule and cytokine gene expression are differently affected by TNF-α + IL-1β cotreatment.
Because the proinflammatory cytokines TNF-α and IL-1β can be present simultaneously at sites of inflammation, we investigated the gene expression profile of HUVEC when incubated with both TNF-α and IL-1β. After cytokine cotreatment, HUVEC expressed all genes to the same or a significantly higher extent compared with the levels induced by each cytokine alone. This was observed for all combinations studied except for VCAM-1 expression induced by the mix of 1 ng/ml TNF-α and 1 ng/ml IL-1β. However, a striking difference in the levels of gene expression induction was observed between cell adhesion molecules on one hand and cytokines and COX-2 on the other hand. In Fig. 2 representative examples for the three different types of responses are shown. Whereas the expression of VCAM-1 on coincubation was lower than the expected level calculated by summation of mRNA levels induced by separate cytokine treatment, the mRNA levels of IL-8 showed true additive effects. The definition of “additive effect” was based on absence of statistically significant differences (P > 0.05) between mRNA increase experimentally obtained for the combination of both cytokines and the calculated increase obtained by summation of the mRNA levels induced by either cytokine alone. Only at 1 ng/ml of the combined cytokines was the observed IL-8 mRNA level significantly higher than the level calculated by mere summation of mRNA levels induced by either cytokine. In contrast, mRNA levels of IL-6 were significantly higher than expected from their response to separate TNF-α or IL-1β treatment (P < 0.05). E-selectin and ICAM-1 followed the VCAM-1 response, whereas COX-2 exhibited an additive response to combined treatment similar to that observed with IL-8 (data not shown).
dnIκB mutant inhibits both TNF-α- and IL-1β-driven proinflammatory gene expression.
TNF-α- and IL-1β-induced signal transduction in endothelial cell relays mainly through the NF-κB and p38 MAPK routes. However, the relative contributions of these pathways to the control of the expression of the genes under study are largely unknown. We investigated the functional relationship between NF-κB activity and proinflammatory gene expression in HUVEC by overexpressing dnIκB. HUVEC infected with LacZ adenovirus and HUVEC infected with dnIκB adenovirus were activated with 1 and 10 ng/ml of TNF-α or IL-1β for 6 and 24 h. Approximately 70–80% of cells were infected by adenoviral infection, as established previously (24). Expression of dnIκB transgene in HUVEC was confirmed by Western blot analysis (Fig. 3A). Under the conditions studied, only a minor fraction of cells went into apoptosis on TNF-α activation. This is likely a consequence of NF-κB-independent cell survival signaling controlled by serum-derived growth factors, as the cells were continuously cultured in medium containing FCS.
An almost total inhibition of adhesion molecule expression induced by 10 ng/ml TNF-α or IL-1β was obtained in dnIκB-expressing HUVEC compared with uninfected or LacZ-infected cells at both time points studied (Fig. 3B). A partial inhibition of gene expression was observed for IL-6 and IL-8 at 6 h, whereas at 24 h the inhibition of these genes was almost complete (P < 0.05). The COX-2 gene expression observed at 6 h after start of activation was not NF-κB dependent (P > 0.05). Possibly, p38 MAPK and PKC control COX-2 expression in this short time frame (1, 35, 39). However, after 24 h COX-2 expression was significantly inhibited in dnIκB-expressing cells. It is likely that during this longer time period molecules produced in an NF-κB-independent manner induce NF-κB-dependent expression of COX-2 gene in an autocrine fashion (34, 42). A similar pattern of inhibition of gene expression for all genes was seen with 1 ng/ml activator (data not shown). These data demonstrated in a quantitative manner that the upregulation of proinflammatory genes induced by TNF-α and IL-1β is largely (adhesion molecules) or at least partly (interleukins and COX-2) under the control of NF-κB.
Effects of chemical inhibitors of intracellular signaling pathways on inflammatory gene expression.
From the experiments performed with dnIκB-expressing HUVEC, a possible involvement of (an)other cell activation pathway(s) in regulation of cytokine-induced inflammatory gene expression became apparent. Therefore, we investigated the effects of NF-κB inhibitors with different molecular targets and of one p38 MAPK inhibitor on proinflammatory gene expression. HUVEC were incubated with 1 μM of PDTC (data not shown), Dex, MOL-294, or RWJ-67657 1 h before the addition of 10 ng/ml TNF-α or IL-1β. The choice of this fixed concentration of drugs was based on the experience that 1 μM is a drug concentration that could be achieved with drug targeting constructs (9). The minor modulatory effects of PDTC on gene expression (data not shown) were possibly due to the use of a relatively low concentration of this drug compared with >10 μM concentrations used in other studies (21). As shown in Fig. 4, the drugs affected inflammatory gene expression differently depending on the activator used and the time interval studied. RWJ-67657 and MOL-294 were the most potent inhibitors, showing downregulation of several inflammatory genes. Inhibition of p38 MAPK activity by RWJ-67657 resulted in blocking of gene expression of the interleukins and COX-2 at 6 h and additionally of the adhesion molecules after 24 h. In contrast, MOL-294 treatment resulted in an inhibitory effect on both TNF-α- and IL-1β-induced adhesion molecule expression, with the most pronounced effect on VCAM-1 expression at both 6 and 24 h. We consistently found significantly higher IL-8 mRNA levels at 6 h after TNF-α treatment in combination with this drug. Dex downregulated COX-2 gene expression in all conditions studied for ∼25–50%, TNF-α-induced IL-6 gene expression after 6 h of TNF-α activation, and VCAM-1 mRNA levels after 24 h of IL-1β activation. Moreover, pretreatment of HUVEC with 1 μM Dex resulted in a significant increase in IL-6 mRNA level 24 h after IL-1β stimulation compared with untreated cells.
Combination of drugs enhances inhibitory effects on gene and protein expression in HUVEC.
RWJ-67657 and MOL-294 affect different routes of the main proinflammatory activation pathways. Because they both showed pronounced inhibitory effects on TNF-α- and IL-1β-induced gene expression in HUVEC, we hypothesized that combination treatment using these two drugs might result in enhanced inhibitory effects. Therefore, the effects of simultaneous addition of these drugs in concentrations ranging from 0.1 to 10 μM on TNF-α- and IL-1β-induced gene expression were investigated. These effects were measured at 6 h after induction of activation by 10 ng/ml TNF-α or IL-1β. In all concentrations and combinations, the drugs appeared not to be toxic to the HUVEC (data not shown). In the case of cell adhesion molecules, a synergistic inhibitory effect of RWJ-67657 + MOL-294 cotreatment was observed only on E-selectin gene expression (Fig. 5; VCAM-1 and ICAM-1 data not shown). One micromolar MOL-294 by itself downregulated TNF-α-mediated E-selectin mRNA levels from a 5,725 (±1,583)-fold increase to a 4,340 (±614)-fold increase (±SD; n = 3). Coincubation with 1 and 10 μM RWJ-67657 resulted in a further, statistically significant, decrease of gene expression to 2,943 (±283; P < 0.05 among activated, MOL-294-, RWJ-67657-, and MOL-294+RWJ-67657-cotreated cells)-fold and 2,117 (±223; P < 0.05)-fold, respectively. A similar, yet less pronounced, effect was observed in IL-1β-activated cells with the 10 μM MOL-294 + 1 μM RWJ-67657 combination. TNF-α- and IL-1β-induced IL-6, IL-8, and COX-2 expression was also more strongly inhibited by drug combination treatment, although different concentration combinations were responsible for this enhanced inhibition. Thus, by combining the NF-κB and p38 MAPK inhibitors, synergistic pharmacological effects could be induced.
To confirm this increase in pharmacological potential of RWJ-67657 and MOL-294 combination treatment, expression of membrane-associated E-selectin protein and soluble IL-8 protein were measured by flow cytometry and ELISA, respectively (Fig. 6). These two genes were affected most pronouncedly by drug cotreatment. The trend in the protein data corroborated the gene expression data, with the effects of drug cotreatment on the protein production being as pronounced as the effects on gene expression. Inhibition of the expression of E-selectin protein was synergistic with the combination treatment of 1 μM MOL-294 + 10 μM RWJ-67657 compared with single drug treatment (Fig. 6), whereas no such effect was observed with the 1 μM combination of both drugs (data not shown). TNF-α-induced expression of IL-8 protein was synergistically blocked by all 10 μM MOL-294 combinations, showing an even more pronounced effect of the NF-κB-p38 MAPK coinhibition on the protein level than on the gene expression level.
A consistently found deviation from the gene expression data was IL-8 protein production in samples treated with 10 μM MOL-294. Total blocking of the protein expression was paralleled by upregulated mRNA levels.
In the current study we showed that TNF-α and IL-1β differentially activated inflammatory gene expression in endothelial cells in vitro. Of the four different inhibitors of intracellular signaling cascades studied, the p38 MAPK inhibitor RWJ-67657 and the small redox-active protein thioredoxin inhibitor MOL-294 were identified as the most potent drugs, showing downregulation of a number of different inflammatory genes at relatively low concentrations. Cotreatment with both drugs resulted in an enhancement of the inhibitory effect on proinflammatory gene and protein expression. This observation has important implications for (targeted) pharmacological intervention to downregulate endothelial cell activation in order to reduce leukocyte infiltration in the diseased tissue. The drugs could either be administered in a combination treatment protocol or both included in immunoliposomes harnessed with antibodies specifically recognizing activated endothelial cells (9). Immunoliposome-based drug delivery systems can theoretically deliver low micromolar concentrations of drugs. Besides the pharmacological profile demonstrated here, 1 μM RWJ-67657 was shown to completely inhibit MAPKAPK-2 phosphorylation in HUVEC (41). A low micromolar concentration of MOL-294 was furthermore able to block the DNA binding ability of NF-κB in HUVEC, with an IC50 value for VCAM-1 expression inhibition of 2.5 μM (19). Future studies on incorporation of these drugs in the carrier systems developed in our laboratory will allow us to further investigate endothelial cell response to these targeted drugs in vitro and in vivo.
We found that the pattern of proinflammatory gene expression markedly differed depending on the activator used. IL-1β mainly induced IL-6, IL-8, and COX-2 gene expression in HUVEC, an effect possibly due to more efficient utilization of p38 MAPK or other routes that control NF-κB-dependent expression (32, 39, 45). On the other hand, TNF-α more profoundly affected the expression of cell adhesion molecules. These data are in line with those reported by others (39, 45), although the latter studies were not performed in a quantitative manner and did not include a direct comparison of all genes as demonstrated in our study.
TNF-α and IL-1β can be present simultaneously in proinflammatory diseases (29, 31). Because limited data are available on endothelial cell proinflammatory gene expression after simultaneous cytokine treatment (5, 6), we studied the effects of TNF-α + IL-1β cotreatment on gene expression by HUVEC. As long as a combinatory activation does not saturate common cofactors, TNF-α and IL-1β partly utilize different signaling pathways (2). Therefore, one could expect that cytokine cotreatment would induce gene expression to a level that would be the sum of the levels obtained by either cytokine treatment alone. Interestingly, less than additive mRNA levels of adhesion molecules and additive or even synergistic levels of interleukins and COX-2 mRNA were observed. The less than additive effects on cell adhesion molecules might be explained by saturation of the molecules that control gene expression at low concentrations of either TNF-α or IL-1β (see also Fig. 1); other phenomena likely underlie the effects observed for the cytokines and COX-2. In dendritic cells it was shown that lipopolysaccharide was able to strongly activate p38 MAPK, an event instrumental for efficient phosphorylation and phosphoacetylation of histone H3. This marked IL-6 and IL-8 promoters for increased NF-κB recruitment, thereby enhancing the gene expression induction of these cytokines. Direct activation of p38 MAPK by TNF-α receptor signaling was weak, resulting in minor recruitment of p65 to IL-6 and IL-8 promoters (32). We showed that IL-1β-induced IL-6, IL-8, and COX-2 gene expression was stronger than their TNF-α-induced expression and that IL-6, IL-8, and COX-2 are under p38 MAPK control to a significant extent (Figs. 4 and 5). Possibly, also in endothelial cells IL-1β- and TNF-α-induced p38 MAPK and NF-κB collaborate in a more pronounced transcriptional induction of the cytokine genes. Conversely, ICAM-1 expression is almost completely p38 MAPK independent (Fig. 4 and Ref. 10); thus its expression is likely not affected by histone phosphorylation/acetylation-dependent NF-κB recruitment. The observation that on TNF-α + IL-1β cotreatment ICAM-1 mRNA levels were less than additive fits this model of facilitated transcription. Because the molecular mechanisms of the observed attenuation/increases in gene expression induced by proinflammatory activator cotreatment are not completely clear at present, we decided to perform our subsequent pharmacological studies while using TNF-α and IL-1β separately. Still, this observation justifies further research, as it has important implications for the choice of the experimental conditions to study the pharmacological potency of new chemical entities in the drug development pipeline.
Both the dnIκB adenovirus and drug treatment experiments confirmed a major role of the NF-κB and p38 MAPK pathways in regulation of gene expression in HUVEC as part of the inflammatory response induced by TNF-α and IL-1β (7, 11, 30, 39). An almost total inhibition of adhesion molecule expression in dnIκB-expressing HUVEC was observed at both early and late time points of activation. The induction of IL-6 and IL-8 expression was also inhibited by dnIκB, but to a lower extent than the effects on the adhesion molecules, indicating considerable involvement of (an)other pathway(s) regulating IL-6 and IL-8. Ridley et al. (30) reported that the expression of IL-6 and IL-8 induced by IL-1β in HUVEC was p38 MAPK dependent. In another study, TNF-α- and IL-1β-induced IL-8 gene and protein expression were shown to be partly dependent on reactive oxygen species generation and activator protein 1 activation (14, 44). Also, from our pharmacological experiments, the conclusion seems justified that p38 MAPK is, at least partly, controlling expression of these genes.
An interesting example of uncoupling of gene and protein expression was noted at the early time point in MOL-294-pretreated cells, where a consistent increase in IL-8 mRNA together with a significant block of IL-8 protein production was observed. Possibly, inhibition of thioredoxin resulted in an increased production of reactive oxygen species, thereby increasing IL-8 mRNA expression (14). The inhibition of IL-8 protein production induced by MOL-294, on the other hand, may be due to the fact that thioredoxin is required for efficient proteolysis catalyzed by thiol-dependent Cys proteases such as cathepsin (16). Cathepsins are known to be essential in processing of mature IL-8 protein at inflammatory sites (25). In Fig. 7 a schematic representation of the pathways studied and their possible interaction is given.
In summary, we performed a quantitative study on proinflammatory gene expression in HUVEC in which we showed that TNF-α and IL-1β differentially induced cell adhesion molecule and cytokine gene expression when added alone. Combination treatment with both cytokines resulted in deviations from the expected induction of mRNA levels of the genes depending on the gene of study. This observation may be relevant for endothelial cell activation and its complex control mechanisms in inflammatory conditions in vivo. The demonstrated synergistic effects of combinations of anti-inflammatory drugs that inhibit NF-κB and p38 MAPK signal transduction form the basis for further research to investigate whether combination treatment can improve the efficacy of (targeted) drugs to inhibit inflammatory disease activity.
We thank Henk Moorlag for technical support with endothelial cell culture and Naomi Werner for excellent technical assistance in RNA isolation.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2005 the American Physiological Society