INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

IRON IS ESSENTIAL to the function of many proteins and critical for eukaryotic life. The majority of iron in the body is contained in heme proteins. Although heme iron is ensconced within a porphyrin ring, it can still undergo oxidation/reduction reactions in many cases. This reactive nature of the heme iron is required for the function of enzymes such as the cytochrome P-450 oxygenases. It also creates a hazard when oxidants such as hydrogen peroxide are present, as the ferrous iron in heme can then catalyze free radical production through Fenton reaction chemistry. Thus, although heme iron is vital to eukaryotic life, its presence is perilous to the cell.

Mechanisms that exert tight control over iron probably have been a necessary product of cellular evolution. One means of iron control is the microsomal enzyme heme oxygenase, which catabolizes heme to biliverdin with the release of iron and carbon monoxide (60). The iron is subsequently reused or sequestered in the storage protein ferritin, where it can no longer participate in redox reactions. Both a constitutive form of the heme oxygenase enzyme, heme oxygenase-2, and an inducible form, heme oxygenase-1 (HO-1), are known to exist. The expression of HO-1 is upregulated in response to oxidative stimuli such as hyperoxia (16, 29) and ultraviolet light (34) and in models of oxidative inflammatory processes such as ischemia-reperfusion injury (40, 58) and the acute respiratory distress syndrome (13). This suggests that enhanced heme removal by HO-1 may be a salutary response to inflammatory/oxidative insult, and recent studies show that enhanced HO-1 expression can be protective in models of inflammation (2, 39, 48, 50, 69, 71, 72). Ferritin is probably the final cytoprotectant because of its iron-sequestering function and its ferroxidase activity (6); however, HO-1 acts upstream of ferritin to open the heme ring promoting the transfer of heme iron to ferritin (25, 67). Although HO-1 may be an important protective response against inflammation, little is known about endogenous inflammatory mediators that may act to upregulate this enzyme.

The cytokines interleukin (IL)-1, tumor necrosis factor- (TNF-), and IL-6 are characteristically present during many inflammatory disorders (22, 43, 63). The vascular endothelium is a critical target for TNF- and IL-1, as these agents induce endothelial cells to direct inflammatory cell traffic (9, 10), promote vascular permeability (41), and induce the release of vasoactive substances such as endothelin (31) and prostacyclin (32). IL-1 and TNF- also induce the endothelial release of IL-6 (51), which is an important mediator of the acute phase response (15). Although the ability of endothelial cells to respond to IL-6 has been questioned (51), recently, IL-6 has been reported to inhibit constitutive prostaglandin synthesis (42) and enhance adhesion molecule expression in endothelial cells (70), suggesting that the endothelium may be a target for IL-6 as well.

During inflammation, the vasculature is subject to oxidant exposure from a myriad of sources, such as activated leukocytes (27, 54) and upregulated enzymes like xanthine oxidase and NAD(P)H oxidase, which release hydrogen peroxide (23, 65). Heme, which has been shown to readily incorporate into endothelial cell membranes (8), can increase this oxidant tone by amplifying the production of radical species, exacerbating the damage to cell membranes and other cell constituents (7). Purging iron from endothelial cells affords them protection against later oxidant insult, providing evidence that removal of heme iron may be a beneficial endothelial response to oxidative stress (64). In fact, increased HO-1 activity has been shown to enhance survival of endothelial cells exposed to heme iron (1), and bilirubin, the downstream product of HO-1 metabolism of heme, has recently been shown to be protective against hydrogen peroxide toxicity in a pig aortic endothelial cell line (47).

TNF- and IL-1 have been demonstrated to induce other oxidant-protective mechanisms such as superoxide dismutase (68) and metallothionein (30) in endothelial cells. HO-1 upregulation by these inflammatory mediators may be another protective strategy as well. In vivo studies in mice have shown that the cytokines IL-1, TNF, and to a lesser extent IL-6 induce HO-1 in mouse liver (14, 53). Additionally, lipopolysaccharide (LPS), a causative agent in gram-negative sepsis, induces HO-1, and administration of an IL-1-receptor antagonist partially inhibits the induction (14). This suggests that cytokines participate in LPS induction of HO-1. However, no studies have reported the ability of cytokines to induce HO-1 in nontransformed human cells. Endothelial cells are active participants in inflammation and often undergo oxidant stress during inflammation, and HO-1 activity has been shown to be protective against oxidative stress in endothelial cells. Therefore, this investigation was carried out to determine if the cytokines IL-1, TNF-, and/or IL-6 induce HO-1 in human endothelial cells.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Cell culture. Endothelial cells were isolated from human umbilical veins by collagenase detachment and cultured according to the established method described previously (26). Umbilical cords were obtained from the Labor and Delivery Division of St. Marks Hospital, Salt Lake City, UT. Cells were grown to confluency in 75-cm² plastic flasks in endothelial cell growth media (EGM) (Clonetics, San Diego, CA). Confluent cells were harvested with trypsin and transferred to sterile 1% gelatin-coated tissue culture plasticware. The cells were grown to confluency in 60-mm dishes for RNA extraction and protein extraction and in 24-well plates for prostaglandin analysis experiments. Confluent, first-passage cells were used for all experiments. During agonist treatment, cultures were observed for any sign of cell injury, and, if cell injury was evident, the cultures were excluded from the experiment.

Cell treatment with cytokine. Confluent endothelial cells were incubated at 37°C in 5% CO₂-95% air in serumless Neuman Tytell (GIBCO-BRL, Grand Island, NY) for 3-4 h before agonist addition to allow the cells to recover from washing before treatment. Agonists were then added at the concentrations indicated in each experiment, and incubation was continued for the time periods indicated.

For time and concentration studies, endothelial cells were incubated with either human recombinant IL-1 (1-500 U/ml; Boehringer Mannheim, Indianapolis, IN), human recombinant TNF- (10-1,000 U/ml; Genzyme, Cambridge, MA), or human recombinant IL-6 (10-500 U/ml; Boehringer Mannheim) for 2 to 24 h. For studies involving 12- to 24-h incubations, the cells were incubated with agonist in EGM, whereas studies of <12 h were conducted in serumless Neuman Tytell media.

For studies analyzing the effect of IL-6 on prostacyclin release, the medium from 500 U/ml IL-6-treated cells was sampled at 2, 4, and 6 h and then analyzed for prostacyclin content by radioimmunoassay (RIA).

Actinomycin D and cycloheximide studies. Endothelial cells were incubated with IL-1 or TNF- in the absence or presence of either actinomycin D (0.25 or 0.5 µg/ml; Sigma, St. Louis, MO) or cycloheximide (1-10 µg/ml; Sigma). The actinomycin D and cycloheximide were added 15 min before cytokine addition.

Curcumin studies. Endothelial cells were treated with IL-1 or TNF- in the absence or presence of 20 µM curcumin (Sigma) and incubated for 4 h. Curcumin was added simultaneously with agonist. Cells were then harvested for RNA analysis.

Northern blot analysis. RNA was extracted from cells by using either the acid guanidinium thiocyanate-phenol-chloroform procedure (18) or by using a kit (Purescript; Gentra Systems, Minneapolis, MN). RNA was quantitated by ultraviolet absorbance at 260 nm, and typically 10 µg but occasionally 5 µg of denatured total RNA were separated by electrophoresis on a 1% agarose/formaldehyde denaturing gel. The RNA was transferred to nylon membranes (Hybond N; Amersham, Arlington, IL) and fixed with ultraviolet light cross-linking (Stratalinker; Stratagene, La Jolla, CA). Membrane hybridizations were carried out at 42°C with [³²P]cDNA HO-1 probe synthesized by random priming (Prime-it; Stratagene). The membranes were then washed under stringent conditions, placed between intensifying screens, and exposed to autoradiographic film (Sterling XR-100; Life Sciences, Denver, CO) at 70°C for 4-24 h.

Lane-loading equivalencies were determined by hybridizing the membranes with [³²P]cDNA Chinese hamster ovary-B (CHO-B) probe. CHO-B message expression in endothelial cells has previously been shown to be unaffected by cytokine treatments (61). The autoradiographic images from the HO-1 labels and the CHO-B labels were scanned with a laser densitometer (Ultrascan XL Enhanced Laser Densitometer, LKB Bromma, Pisctaway, NJ), which converted the densities of the bands to relative absorbance units. The densities of the CHO-B bands were used to correct the HO-1 mRNA band density values. These normalized values were plotted in graph form by either comparing mRNA induction with control levels or by setting cytokine induction as 100% and reporting the effect of treatments as a percentage of this maximal induction.

Probes. The human HO-1 cDNA probe was made by polymerase chain reaction (PCR) amplification of a 762-bp fragment of HO-1. The primers for the PCR were based on the human HO-1 cDNA sequence reported by Yoshida et al. (73). The amplification product was ligated into the plasmid, pCR (Invitrogen, San Diego, CA), and was partially sequenced by the Sanger method. The sequence was found to match that expected for the HO-1 sequence. In addition, the amplification product was restriction mapped with Hae II and Ban I restriction enzymes, which produced the expected size fragments according to the reported HO-1 sequence. The probe specifically recognizes an mRNA band of ~1.8 kb from cells exposed to sodium arsenite, a characteristic inducer of HO-1.

The CHO-B cDNA probe was a generous gift from Dr. Bruce Marshall of the Division of Pulmonary Medicine, University of Utah School of Medicine, and recognizes an mRNA species of 1.0 kb in size.

Immunoblotting. HO-1 protein was isolated from endothelial cells following a procedure reported by Kutty et al. (36). Briefly, after experimentation, cells were washed with ice-cold phosphate-buffered saline (PBS) and then scraped into ice-cold freshly prepared sucrose extraction buffer [20 mM tris(hydroxymethyl)aminomethane · HCl, pH 7.4, 0.25 M sucrose, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 50 µg/ml leupeptin, and 100 µg/ml aprotonin]. The cell lysates were frozen in liquid nitrogen and stored at 70°C until further extracted. For protein extraction, the cell lysates were thawed on ice and sonicated for 10 s. The microsomal fraction was then obtained by centrifugation at 10,000 g for 15 min and analyzed by Western blot. The protein concentration of the microsomal fractions was determined by the spectrophotometric bicinchoninic acid assay (Pierce, Rockford, IL).

Western blot analysis of HO-1 was done by fractionating 10 µg of protein on a 12% polyacrylamide gel by denaturing discontinuous gel electrophoresis according to the Laemmli method. The proteins were transferred to a polyvinylidene difluoride (PVDF) membrane (Pall Biodyne, East Hills, NY) by tank transfer (Bio-Rad Laboratories, Hercules, CA). HO-1 protein was detected with a rabbit anti-rat HO-1 antibody (StressGen, Victoria, BC, Canada) using a Western-Light Chemiluminescent detection system (Tropix, Bedford, MA). The antibody specifically recognizes the 32-kDa HO-1 protein and exhibits cross-reactivity with human, mouse, and rat HO-1. As a positive control, protein from endothelial cells exposed to 50 µM sodium arsenite, a characteristic inducer of HO-1, was also analyzed by Western blot and revealed a band at 32 kDa that reacted strongly with the HO-1 antibody.

Prostacyclin measurement. Prostacyclin was measured by RIA of its major metabolite, 6-ketoprostaglandin F₁, as previously described (26).

Statistical analysis. Means and SE for HO-1 message levels were calculated after normalization of the absorbance units derived from scanning densitometry of the autoradiographic images of Northern blots. A pooled t-test analysis of the data was used for determination of statistical significance.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Studies were carried out to characterize time- and concentration-dependent effects of cytokines on HO-1 mRNA expression. Endothelial cells were exposed to either TNF- or IL-1 for 2, 4, or 6 h, and mRNA was isolated and analyzed by Northern blot technique. The results of a representative autoradiograph are shown in Fig. 1A. The HO-1 mRNA reaches maximal levels at ~4 h in both TNF-- and IL-1-treated cells. The membranes were probed for expression of the constitutive message CHO-B to verify equal lane loading. Figure 1B illustrates these data in graph form after correcting for small lane loading differences. The data indicate that induction of HO-1 mRNA by IL-1 or TNF- occurs in a time-dependent manner.

View larger version (23K): [in this window] [in a new window]   Fig. 1.   Tumor necrosis factor (TNF)- and interleukin (IL)-1 induce heme oxygenase (HO)-1 mRNA in human endothelial cells in a time-dependent manner. A: Northern blot analysis of cells exposed to 500 U/ml TNF- or 50 U/ml IL-1 for 2-6 h. Control (CON) lanes contain RNA from cells exposed to media alone. HO-1 mRNA was detected with a radiolabeled human HO-1-specific cDNA probe. B: densitometric analysis of autoradiograph (A) with normalization to constitutive Chinese hamster ovary-B (CHO-B) mRNA as described in MATERIALS AND METHODS.

Next, endothelial cells were exposed to increasing concentrations of either IL-1 or TNF- for 4 h to determine the concentration of each cytokine that is most effective at inducing HO-1 mRNA. Figure 2 illustrates that cytokine induction of HO-1 mRNA is concentration dependent and that maximal induction of HO-1 mRNA occurs between 50 and 100 U/ml IL-1 and between 500 and 1,000 U/ml TNF-. The mean HO-1 induction by 50 U/ml IL-1 was not shown to be significantly different from the mean induction by 100 U/ml IL-1 ( = 0.1, P = 0.6641). The same was true for TNF- concentrations between 500 and 1,000 U/ml ( = 0.1, P = 0.6749). IL-1 appears to be a more effective inducer of HO-1 mRNA than TNF- at this time point. The majority of experiments hereafter were carried out at 50 U/ml IL-1 and 500 U/ml TNF-.

View larger version (16K): [in this window] [in a new window]   Fig. 2.   IL-1 and TNF- induce HO-1 mRNA in a concentration-dependent manner. Cells were exposed to 1-100 U/ml IL-1 or 10-1,000 U/ml TNF- for 4 h and then harvested for RNA analysis. HO-1 mRNA induction is expressed as means + SE; n  3 for all treatments.

IL-1 and TNF- induce production of IL-6 within a few hours of exposure (49). Therefore, induction of HO-1 message by these cytokines could be attributed partly or wholly to an action by IL-6 via an autocrine mechanism. To determine if IL-6 might be involved in IL-1 or TNF- induction of HO-1, the ability of IL-6 to induce HO-1 mRNA was examined. Endothelial cells were exposed to 10, 100, or 500 U/ml IL-6 for 2-24 h. An autoradiograph of a Northern blot containing mRNA from cells exposed to 100 U/ml IL-6 for 2-6 h is shown in Fig. 3. No induction of HO-1 mRNA was observed at any time point or concentration examined (data for 8-24 h exposure or 10 and 500 U/ml IL-6 are not shown). IL-6 can synergize with IL-1 in the induction of gene expression such as with serum amyloid A protein in human hepatoma cells (20). Therefore, endothelial cells were exposed to IL-1 and IL-6 together, and HO-1 mRNA levels were examined. IL-6 had no observable effect on IL-1 stimulation of HO-1 mRNA when examined at 4 h, the time point that IL-1 maximally induces HO-1 (data not shown).

View larger version (36K): [in this window] [in a new window]   Fig. 3.   IL-6 does not induce HO-1 mRNA in human endothelial cells. Cells were treated with media alone () or with 100 U/ml IL-6 (+) for 2-6 h and then harvested for RNA analysis. Northern blot was hybridized with a radiolabeled human HO-1-specific cDNA probe (top) and then radiolabeled with CHO-B probe (bottom) to determine lane loading equivalencies.

To verify that the IL-6 used in these studies was biologically active, an alternative assay of IL-6 effects on endothelial cells was used. IL-6 has been reported to decrease constitutive prostacyclin release in cultured endothelial cells (42). Therefore, prostacyclin concentrations were measured after IL-6 exposure to examine whether the endothelial cells are capable of responding to IL-6. In this study, exposure of endothelial cells to 500 U/ml IL-6 for 2, 4, or 6 h caused a decrease in constitutive prostacyclin release compared with untreated controls [control vs. IL-6 at 2 h (706 vs. 510 pg/well); control vs. IL-6 at 4 h (878 vs. 528 pg/well); control vs. IL-6 at 6 h (938 vs. 498 pg/well), n = 2 for every time point]. Because prostacyclin release was decreased by IL-6, it can be concluded that the IL-6 was active and that the endothelial cells are capable of responding to IL-6. However, IL-6 does not appear to be involved in the induction of HO-1 expression.

Enhancement of HO-1 expression in other cell types has been shown to occur at the level of transcription for a variety of inducers (33). To examine whether cytokine induction of HO-1 message in endothelial cells occurs via transcription, the cells were exposed to cytokine in the presence of the transcription inhibitor actinomycin D. Figure 4 shows a representative autoradiograph of RNA from cells exposed to TNF- in the presence and absence of actinomycin D. Inhibition of transcription with this agent prevented HO-1 mRNA induction by TNF-. In further experiments, actinomycin D (0.25 µg/ml) decreased the transcription of HO-1 mRNA induced by either TNF- or IL-1 by at least 95% (3 experiments with 2 duplicates in each). Thus this concentration of actinomycin D was used in further studies examining the effect that TNF- and IL-1 have on the HO-1 message half-life. Cells were treated with IL-1 for 4 h and then actinomycin D or vehicle (dimethyl sulfoxide) was added. RNA was extracted at 1, 1.5, 2, and 3 h after actinomycin D addition. The mRNA levels for each time point and treatment were then graphed as an exponential plot of HO-1 mRNA levels versus time, as shown in Fig. 5. The slope of the decrease in HO-1 mRNA levels is similar between IL-1 alone and IL-1 plus actinomycin D treatments, indicating that the message decay rate is largely unchanged by IL-1 exposure. Similar results were observed with TNF- in the presence of actinomycin D (data not shown).

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Actinomycin D inhibits TNF- induction of HO-1 mRNA. Northern blot analysis of cells exposed to 500 U/ml TNF- in the presence (+) or absence () of 0.25 or 0.5 µg/ml actinomycin D for 4 h. Control cells were exposed to vehicle [dimethyl sulfoxide (DMSO)] alone. Northern blot that was probed with a radiolabeled human HO-1 specific cDNA probe is shown.

View larger version (12K): [in this window] [in a new window]   Fig. 5.   IL-1 does not affect HO-1 mRNA degradation rates in human endothelial cells. Cells were exposed to 50 U/ml IL-1 for 4 h. Cells were then treated with either actinomycin D (0.25 µg/ml) or vehicle (at 0.005%), and the cells were harvested for RNA analysis at 1, 1.5, 2, and 3 h after actinomycin D or vehicle addition. RNA was analyzed by Northern blot followed by densitometric analysis and normalization. Densitometric value for each treatment was divided by the densitometric value for IL-1 alone and expressed as that percentage. Results shown are from 2 experiments. Data were graphed as an exponential plot of relative HO-1 mRNA levels vs. time.

The HO-1 gene has numerous transcriptional regulatory elements in its 5'-untranslated region. These include consensus recognition sites for the activator protein (AP)-1 transcription factor and sequences that closely resemble the consensus recognition site for nuclear factor-B (NF-B; see Refs. 38 and 59). AP-1 has been shown to be involved in the induction of HO-1 by phorbol myristate acetate and LPS (4, 13). Whether cytokine induction of HO-1 expression involves AP-1 or NF-B has not been studied. Therefore, to further investigate the transcriptional upregulation of HO-1 by IL-1 and TNF-, HO-1 message levels were examined from endothelial cells that were exposed to cytokine in the absence or presence of curcumin. Curcumin has been shown to be an effective pharmacological inhibitor of AP-1 and NF-B activation in endothelial cells (12). Curcumin significantly attenuated both TNF- and IL-1 induction of HO-1 mRNA as shown in Fig. 6. In contrast, an inhibitor of NF-B activation, pyrrolidine dithiocarbamate (PDTC), did not decrease HO-1 induction by cytokine (data not shown).

View larger version (40K): [in this window] [in a new window]   Fig. 6.   Activator protein-1 inhibitor, curcumin, attenuates cytokine induction of HO-1 mRNA. Cells were exposed to 50 U/ml IL-1 or 500 U/ml TNF- in the absence or presence (+Curcumin) of 20 µM curcumin for 4 h. Cells were then harvested for RNA analysis by Northern blot followed by densitometric analysis and normalization. Representative Northern blot is shown in A. In B, induction of HO-1 mRNA in the presence of curcumin is expressed as mean + SE from n  4 for all treatments. Statistical significance: IL-1 + curcumin,  = 0.01, P = 0.007; TNF- + curcumin,  = 0.05, P = 0.0157.

To determine whether de novo protein synthesis is required for cytokine induction of HO-1 mRNA, endothelial cells were exposed to cytokine in the presence of the protein synthesis inhibitor cycloheximide. Cycloheximide completely abrogates TNF- or IL-1 induction of HO-1 message as shown in Fig. 7.

View larger version (12K): [in this window] [in a new window]   Fig. 7.   Protein synthesis inhibition prevents cytokine induction of HO-1 mRNA. Cells were exposed to either 50 U/ml IL-1 or 500 U/ml TNF- in the absence or presence (+Cycloheximide) of 10 µg/ml cycloheximide and incubated for 4 h. Cells were then harvested for RNA analysis followed by densitometric analysis and normalization. Percentage of cytokine induction of HO-1 mRNA is expressed as mean + SE from n  5 for each treatment.

Changes in mRNA levels are not always followed by corresponding increases in protein. Subsequently, studies were carried out using Western blotting techniques to examine whether HO-1 protein is elevated after cytokine exposure as well. Cells were exposed to IL-1 or TNF- for 4, 6, or 8 h, and the protein was isolated and immunoblotted with an antibody against HO-1. Representative Western blots are shown in Fig. 8. Both IL-1 and TNF- caused production of an ~32-kDa protein that was immunoreactive with HO-1-specific antibody. Maximum protein levels occurred by 6 h and began to subside by 8 h after cytokine stimulation.

View larger version (33K): [in this window] [in a new window]   Fig. 8.   IL-1 and TNF- induce HO-1 protein in human endothelial cells. Cells were treated with either 50 U/ml IL-1 or 500 U/ml TNF- (+) or media alone () for 4, 6, or 8 h, and the protein was extracted and analyzed by Western blotting technique with antibody to HO-1.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Heme oxygenase is highly upregulated in response to oxidant stress (5) and has been suggested to play a protective role against the oxidant injury accompanying inflammatory disease processes (2, 39, 48, 50, 52, 66, 69). Previous studies indicated that cytokines can induce HO-1 in rodent cells (14, 24, 28, 35, 53, 62). In contrast, no work had previously examined whether IL-1 or TNF can act to induce HO-1 expression in human cells.

We show in this study that the cytokines IL-1 and TNF- are effective inducers of HO-1 in cultured human endothelial cells. Although the induction of HO-1 by cytokines is less than that observed with sodium arsenite (the prototypical but nonphysiological HO-1 inducer), it is comparable to that seen with hemin (unpublished observation). Thus IL-1 and TNF- join a growing list of HO-1 inducers, including sodium arsenite (34), heavy metals (3), and ultraviolet light (34). However, unlike many of the previously described inducers, cytokines are physiological in vivo signaling mediators.

In endothelial cells, the induction of HO-1 message by both IL-1 and TNF- occurs at the transcriptional level and requires protein synthesis. Although most other agents induce HO-1 at the transcriptional level, the effect of protein translation inhibition on HO-1 induction varies depending on the cell type examined and the agent used. Similar to the present results with cytokines, others have shown that inhibition of protein translation blocks prostaglandin A₂ induction of HO-1 in fibroblasts (17). However, cycloheximide has no effect on the induction of HO-1 by IL-6 in hepatoma cells (46) or on LPS induction of HO-1 in macrophages (13). Conversely, protein translation inhibition causes enhanced induction of HO-1 by cobalt-protoporphyrin (45). These disparate results indicate that different mechanisms exist to increase HO-1 expression, depending on the agent used and cell type examined. The finding that cycloheximide completely abrogates cytokine induction of HO-1 in endothelial cells suggests that a labile protein or de novo protein synthesis is involved in the expression of HO-1.

New protein synthesis may be required to generate or activate a transcription factor involved in HO-1 upregulation. The HO-1 gene contains transcription factor recognition sites for both NF-B and AP-1, and both of these factors have been shown to act in the expression of many oxidative stress genes (55). NF-B binding activity is enhanced by cytokines in endothelial cells, and message and protein levels for components of the AP-1 binding complex are rapidly elevated as well (11, 19). NF-B becomes activated upon dissociation of an inhibitory protein, and thus its binding activity can be increased in the absence of protein synthesis (56). In this study, induction of HO-1 mRNA was decreased when cells were exposed to cytokines in the presence of the AP-1 and NF-B inhibitor, curcumin, suggesting that one of these factors may play a role in cytokine induction of HO-1. To further determine which transcription factor was acting in the HO-1 induction by cytokines, the dithiocarbamate derivative PDTC was used. PDTC inhibits NF-B activation but enhances AP-1 binding activity (44). When endothelial cells were exposed to PDTC and cytokine, enhanced HO-1 expression was observed. In fact, PDTC alone induced HO-1 expression (data not shown). These results together with the curcumin studies suggest that AP-1 may be the transcription factor involved in the expression of HO-1.

Elevated HO-1 levels may be serving another function besides iron conservation in vascular endothelial cells, as this cell type is not typically considered important in in vivo heme catabolism. It is possible that HO-1 induction by IL-1 and TNF- in endothelial cells has evolved in response to another action of these cytokines, that is, recruitment of activated leukocytes to the endothelium. As the presence of free heme amplifies leukocyte-derived oxidant damage, it may be beneficial for the endothelium if HO-1 levels were increased before or early on in leukocyte arrival, allowing for enhanced removal of heme. TNF- and IL-1 induce leukocyte adhesion molecules on endothelial cells between 4 and 12 h (10, 74), and cytokine-induced HO-1 protein expression begins by 2 h, making this a feasible scenario. Another possible benefit of enhanced HO-1 expression is that it results in elevated levels of bilirubin, an effective antioxidant that is a downstream product of heme oxygenase activity (57). Investigations by others showed that bilirubin protects endothelial cells against hydrogen peroxide toxicity (47). Additionally, when HO-1 activity is increased, the levels of ferritin, which scavenges free iron, are also increased. Thus endothelial HO-1 induced by cytokines could be protective by both removing reactive heme, with subsequent iron sequestration in ferritin, and by producing a free radical scavenger in the form of bilirubin. Studies by others support this suggestion. Otterbein et al. (50) showed that preinduction of HO-1 by hemoglobin protected against endotoxemia in rats wherein oxidants play a major role in the pathology of endotoxemia. A hamster cell line that overexpresses HO-1 is resistant to hyperoxia, whereas inhibition of HO-1 expression in that cell line by antisense oligonucleotides increases the susceptibility to hyperoxia (21). Also, overexpression of transfected human HO-1 in rabbit coronary endothelial cells decreases toxicity caused by heme and hemoglobin (1). Induction of HO-1 and ferritin by heme also protects against endothelial cell lysis by both hydrogen peroxide and activated polymorphonuclear cells (6, 8). As TNF and IL-1 induce HO-1 comparable with heme, it is likely that cytokine preinduction of HO-1 would also yield protection against oxidant insult. In preliminary studies, we have observed that cytokine-stimulated endothelial cells are protected against oxidant lysis to a similar degree as heme-stimulated cells. However, it is important to note that IL-1 and TNF induce other antioxidant defense mechanisms, such as superoxide dismutase (68). Thus a comprehensive study is required to accurately assess the potential roles played by HO-1 and other enzymes in the endothelial cell response to oxidant stress.

In our studies, IL-6 does not induce HO-1 mRNA. TNF- and IL-1 stimulate the release of IL-6 from endothelium; however, as IL-6 has no effect on endothelial HO-1 message levels in this study, induction of HO-1 by TNF- and IL-1 cannot be attributed to IL-6 production. These results differ from what has been reported by others in different cell types. For example, IL-6 induces HO-1 in a human Hep3B hepatoma cell line (46) and in mouse liver (53), albeit to minor degrees. In addition, the human HO-1 promotor has been shown to contain IL-6 response elements (46), and promotor activity is upregulated by IL-6 in transfected rabbit coronary microvessel endothelial cells, as measured by chloramphenicol acetyltransferase assays (37). The present study did find that IL-6 reduces endogenous prostacyclin production, demonstrating that these cells are responsive to this cytokine. Thus HO-1 gene induction by IL-6 appears to be cell type dependent.

In summary, the present data show that the cytokines IL-1 and TNF-, but not IL-6, are effective inducers of HO-1 message and protein. As IL-1 and TNF- are important biological participants in inflammation, the results from this study further support a role for this enzyme in the cellular response to inflammatory stress.