Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

High inspired concentrations of oxygen are often a necessary and lifesaving therapy for the treatment of critically ill patients. Unfortunately, these high oxygen concentrations are also toxic and add to the overall tissue injury, especially in the lung (1, 2). Hyperoxia-induced lung damage includes atelectasis, interstitial and alveolar edema, alveolar hemorrhage, inflammation, fibrin deposition, and thickening and hyalinization of alveolar membranes. Early evidence of oxygen-induced damage consists of endothelial leakage with increased pulmonary edema and protein leak into the alveoli (2). This is followed by an influx of inflammatory cells, worsening of pulmonary edema, alveolar exudates, and hemorrhage. This damage resulting from hyperoxia has also been observed in vitro in cultured pulmonary endothelial cells (3, 4).

Although the pathogenesis of hyperoxia-induced pulmonary damage is not completely understood, the excess production of toxic reactive oxygen species (ROS), which overwhelms the cell's antioxidant defense mechanisms, is thought to play a critical role in damaging cellular components (5). The cytotoxic actions of oxygen radicals are increased in the presence of a transition metal catalyst, with iron being the most abundant transition metal (6). Free iron promotes the generation of the very reactive hydroxyl radical, which is damaging to the cell. Chelators of free iron have also been demonstrated to provide a protective effect against hyperoxia-induced damage to the lung (7, 8). In addition, a number of antioxidants, both enzymatic and nonenzymatic, have been reported to decrease hyperoxia-induced lung injury (9).

Recently, heme oxygenase (HO)-1 has been shown to be induced by oxidants (10) including hyperoxia (13- 15), and this induction appears to be a protective response. HO is best known for its function as the initial and rate-limiting step in the degradation of heme to bilirubin (16). This is not only true for hemoglobin but for many other ubiquitous heme-containing proteins of which cytochrome P450 is the major nonhemoglobin source. There are two isozymes of heme oxygenase that are products of distinct genes and differ in their tissue distribution and regulation (16). Recently, a third isoenzyme, HO-3, which is similar to HO-2 has been described (17). HO-1 is the inducible enzyme that is induced by oxidants, whereas HO-2 is considered constitutive. Dennery and colleagues (18) have recently shown that mice lacking HO-2 are more sensitive to oxygen toxicity, and that such toxicity is related to iron because iron accumulates in the lungs of these mice. Poss and Tonegawa (19, 20) have also shown that mice lacking HO-1 show evidence of iron accumulation and are more sensitive to oxidant injury. They suggested a role for HO-1 in iron utilization and the tissue response to inflammation and injury.

The cytoprotective effects of HO may be due to several functions. Because increased levels of heme enhance the pro-oxidant state of the cell (21, 22), the degradation of heme by HO may be an adaptive mechanism in response to an oxidant stress. Another mechanism may be the production of bilirubin, which has been shown to be a potent antioxidant, with albumin-bilirubin complexes being very efficient scavengers of free radicals (23, 24). A third potential protective mechanism of HO-1 is its interaction with ferritin. Intracellular free iron is sequestered by ferritin, and the induction of ferritin by oxidants appears to be coupled to induction of HO-1 (25). This increase in ferritin reduces intracellular free iron, thereby decreasing the possibility of iron-catalyzed free radical reactions. The potential protective mechanisms of HO against an oxidant insult may be especially relevant to the lung under a hyperoxic insult. Hyperoxia produces an endothelial leak resulting in an extravasation of hemoglobin/heme into the endothelium, interstitium, and alveolar space, thereby increasing the pro-oxidant state of the lung. In addition, excess heme in the lung potentially increases the amount of iron available for generating the hydroxyl radical. Thus hyperoxia and heme/iron are important in propagating endothelial injury seen in such situations as in the lung.

Previously, HO-1 expression in the rat lung was shown to be increased following 24 to 72 h of a hyperoxic insult (14). This increase could be a direct result of hyperoxia or secondary to the hyperoxic effects such as inflammation, increased heme, or reactions to free iron, as these are also stimuli for HO-1. Our laboratory (13) and others (15, 16) have demonstrated that HO-1 is increased in cell culture in response to high oxygen concentrations, and these studies therefore suggest that HO-1 is directly responsive to hyperoxia. At present, a majority of the studies with hyperoxia have focused on a rodent model and the mouse gene, with limited information available regarding the human HO-1 gene in response to hyperoxia. Therefore, in this study we evaluated the molecular mechanisms of HO-1 induction by a hyperoxic insult in human pulmonary endothelial cells, with special emphasis on the role of iron in regulating HO-1 induction.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Cell Culture

Human pulmonary artery (PA) endothelial cells were isolated from segments of PA obtained from heart transplant donors as previously described (26). The cells were grown and maintained in endothelial basal medium (EBM) (Clonetics, San Diego, CA) supplemented with 10 ng/ml human epithelial growth factor, 1.0 µg/ml hydrocortisone, 10% fetal bovine serum (FBS; Sigma, St. Louis, MO), 0.4% bovine brain extract, 50 mg/ml gentamicin sulfate, and 50 µg/ml amphotericin B (Clonetics). The cells were passed every 4 d at a ratio of 1:3. Human pulmonary microvascular (PMV) endothelial cells were obtained from Clonetics. Rat PMV endothelial cells were obtained from Dr. Una Ryan (T Cell Sciences, Cambridge, MA) (27). The rat cells were grown in M199 media with 10 mM L-glutamine, 10% FBS, and an antibiotic solution (ABAM; Sigma) at 37°C to form confluent monolayers. Cells were passed when confluent every 2 to 3 d, with 1× trypsin and 0.2 mM ethylenediaminetetraacetic acid (EDTA) at a passage ratio of 1:3 or 1:4.

Experimental Conditions

Cells grown to confluence on 100-mm tissue-culture dishes were exposed to either room air or hyperoxia for 2 to 24 h. Control cells were exposed to room air/5% CO₂ in the cell culture incubator, and hyperoxia was achieved by placing the cells in an airtight chamber (Modular Incubator; Billups-Rothenberg, Del Mar, CA) and flushing completely for 10 min. The chambers were clamped and placed at 37°C for 2 to 24 h, and flushed again at 4 to 8 h intervals for the extended treatments. To evaluate the response to an RNA synthesis inhibitor, the cells were treated with 4.0 µM actinomycin D (Sigma). For a positive control for HO-1 expression, cells were also exposed to hemin (5 µM), a known inducer of HO-1. To enhance the oxidant affects of hyperoxia, the cells were preloaded with iron by exposing the cells to 1 µM FeSO₄ and 1 µM 8-hydroxyquinoline in Hanks' balanced salt solution for 30 min prior to treatment as previously described (28). Desferrioxamine (DFO) was used as an iron chelator by exposing the cells overnight to a concentration of 0.5 mM.

Isolation and Characterization of Rat HO-1 cDNA

A probe specific for rat HO-1 cDNA was obtained by employing reverse transcriptase of HO-1 mRNA, followed by polymerase chain reaction (PCR) amplification using homologous primers (18mers) based on the HO-1 cDNA sequence (29). The PCR product was electrophoresed on a 1% agarose gel and migrated to a position consistent with a product of 745 nucleotides. The fragment was electroeluted from the agarose and subcloned into pT7 Blue TA cloning vector (Novagen, Madison, WI), which utilizes the terminal transferase activity of Taq polymerase. The fragment was verified as HO-1 cDNA by restriction and sequence analysis.

RNA Isolation and Northern Analysis

Ten to twenty micrograms of total RNA were isolated by guanidine-acid phenol method (30), and fractionated by size on a 1% agarose, 6% formaldehyde gel buffered with 3-(N-morpholino) propanesulfonic acid. The RNA was electrotransferred to a positively charged nylon membrane (Cuno, Meriden, CT) and covalently cross-linked to the membrane with ultraviolet light. HO-1 mRNA levels were evaluated by hybridization with a ³²P-radiolabeled rat or human HO-1 cDNA via random primer extension (GIBCO-BRL, Grand Island, NY). The human HO-1 cDNA was generously provided by Rex M. Tyrell (Swiss Institute for Experimental Cancer Research, Epilinges, Switzerland). The membrane was rehybridized with a radiolabeled rat cathepsin B cDNA (gift from S. Chan, University of Chicago, Chicago, IL) or human glyceraldehyde-3-phosphate dehydrogenase (G3PDH) cDNA (Clonetics) as an internal control to evaluate mRNA loading consistency. The membranes were subjected to autoradiography using an intensifying screen at 85°C, and densitometry was performed using NIH Image software (Bethesda, MD).

Nuclear Run-On Transcription Assay

Nuclei were isolated from both control and iron-loaded plus hyperoxia-exposed cells for 8 to 12 h, and nascent HO-1 transcripts were evaluated by nuclear run-on assays (31). Following treatment, the cells were washed, collected by trypsinization, and centrifuged at 200 × g for 5 min at 4°C. The cells were resuspended in phosphate-buffered saline (PBS), centrifuged, and resuspended in 5 ml of lysis buffer (10 mM Tris-HCl, pH 7.5; 10 mM NaCl; 3 mM MgCl₂; 0.5% NP40). The cells were dounce-homogenized, incubated on ice, centrifuged at 1,000 × g for 20 min, and resuspended in 4 ml of storage buffer, pH 7.5 (50 mM N-2-hydroxyethylpiperazine-N'-ethanesulfonic acid [Hepes], 4 mM MnCl₂, 1 mM MgCl₂, 0.1 mM EDTA, 50% glycerol, and 5 mM dithiothreitol [DTT]). The nuclei were processed through a freeze-thaw cycle at 70°C, and collected again by centrifugation followed by gentle vortexing. The nuclei were resuspended in 200 µl incubation buffer (150 mM Hepes, 200 mM KCl, 5 mM MgCl₂, 0.1 mM EDTA, 50% glycerol, 5 mM DTT, 0.5 mM of dATP, dCTP, and dGTP, 4 U creatinine kinase, 8.8 mM creatine phosphate, 10 U RNase inhibitor, and 250 µCi ³²P dUTP/ reaction). The radiolabeled RNA was isolated and hybridized for 24 to 48 h with 3 µg each of denatured human HO-1 and G3PDH cDNAs and linearized pBluescript vector dot-blotted on a nylon membrane (Cuno). The membranes were washed at high stringency, subjected to autoradiography using an intensifying screen, and evaluated by densitometry.

Plasmid Constructs

Long-range PCR of human genomic DNA using rTth DNA polymerase (Perkin-Elmer, Foster City, CA) was performed to generate a fragment of the human HO-1 gene, which extends from 4.5 kb to 80 bp position within exon 1 using published sequence of the human HO-1 gene (32, 33). The PCR amplification yielded a 4.5-kb product that was ligated into the TA cloning vector and verified by restriction and sequence analysis. BamHI sites were incorporated into the primers so that the fragment could be subcloned into a human growth hormone (hGH) vector (p0gH) for transient expression assays and promoter deletion analysis.

Transient Transfection and Reporter Gene Assay

Human PA endothelial cells were grown to approximately 70% confluence in a T75 cell culture flask and washed with PBS, pH 7.5. The cells were removed by trypsinization and resuspended in 4 ml of medium supplemented with 10% NuSerum (Collaborative Biomedical Products, Bedford, MA). The cells were pelleted and resuspended in 500 µl of Tris-buffered saline (TBS). Five hundred microliters of the transfection solution consisting of 100 µM of diethylaminoethyl/dextran (Sigma), 8 µg of vector DNA, and 100 µM chloroquine diphosphate (Sigma) in TBS were added. The cells were incubated for 1 h at 37°C, gently mixing every 10 to 15 min. The cells were pelleted and resuspended in 360 µl of TBS and 40 µl of dimethylsulfoxide (Sigma) for 1 min. Ten milliliters of medium with 10% NuSerum were added, and the cells were pelleted and resuspended in 1 ml of media with 10% FBS. The cells were then equally distributed into three to six 100-mm cell culture dishes. One day following transfection, or when 50% confluence was reached, the medium was changed to EBM containing 1% FBS, and several hours later two of the four plates were kept as control plates and two were treated with either hyperoxia or heme. Medium was collected 72 h after treatment and hGH levels were measured using a commercially available radioimmunoassay (Nichols Institute, San Juan Capistrano, CA). In separate experiments, RNA was isolated from transient transfected cells after exposure to the above stimuli, and Northern blot analysis was performed using a hGH probe. We observed similar findings in the level of hGH mRNA levels as seen with secreted levels of this reporter gene.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

The endothelium is exquisitely sensitive to the toxic effects of oxygen, and its ability to respond to a hyperoxic insult in a protective manner would be very beneficial for its survival and in maintaining a functional lung. HO-1 is considered to have protective properties against an oxidant insult and has been shown to be induced in a variety of cell types (13) and in whole lung of both mouse (18) and rat (14) models upon exposure to hyperoxia. Our laboratory previously demonstrated an increase in HO-1 protein levels in response to hyperoxia in rat PMV endothelial cells (13); however, there is no information regarding HO-1 expression in human pulmonary endothelial cells. Therefore, we evaluated the expression of HO-1 mRNA levels in both rat PMV endothelial cells and human PA endothelial cells in response to hyperoxia. Figure 1 illustrates the relative fold induction of human PA and rat PMV endothelial cells exposed to hyperoxia for 2 to 24 h. We observed an increase in HO-1 mRNA levels in both cell types with a maximal induction of ~ 3-fold in human PA endothelial cells compared with ~ 6.5-fold in the rat PMV endothelial cells. The induction was evident after only 2 h in the rat cells and after 8 h in the human cells. We also observed 3- to 4-fold HO-1 mRNA induction in human pulmonary microvascular endothelial cells following 24 h of hyperoxia (data not shown). This demonstrates that hyperoxia induces a relatively early response of HO-1 mRNA in pulmonary endothelial cells in culture. However, the level of induction is much lower than expected on the basis of what was previously demonstrated in the rat lung (14) or in response to other stimuli such as heme, where there is generally a much higher level of expression (11, 34).

View larger version (33K): [in this window] [in a new window]   Figure 1.   Northern blot analysis of rat and human pulmonary endothelial cells exposed to hyperoxia. This graph illustrates the relative fold induction of HO-1 mRNA levels in both rat PMV and human PA endothelial cells exposed to hyperoxia for 2 to 24 h. The error bars represent SEM.

The increase of HO-1 that occurs in vivo may be a direct response of hyperoxia or a secondary effect of the hyperoxic insult. Hyperoxia results in endothelial damage with extravasation of proteins and blood, thereby increasing the presence of heme and availability of free iron, both of which can induce HO-1 expression. Because hyperoxia alone resulted in a relatively small induction of HO-1 expression, we evaluated the expression of HO-1 in response to hyperoxia, free iron, and a combination of free iron plus hyperoxia in human PA endothelial cells (Figure 2). Human PA endothelial cells were treated with iron in the presence of hydroxyquinoline to increase the free iron pool of the cell. In response to free iron, we observed an increase within 2 h, with a maximal induction of ~ 18-fold after 4 h and a return to near basal levels after 16 and 24 h. The combination of iron plus hyperoxia produced an induction of HO-1 mRNA levels throughout the 24 h of exposure with a peak induction of ~ 20-fold after 8 h. Figure 2a is a representative Northern blot of human PA endothelial cells exposed to hyperoxia, iron, or a combination of the two for 2 to 16 h, and Figure 2b illustrates graphically the relative fold induction for four to 15 individual experiments for each condition. In addition to free iron potentiating the hyperoxia-dependent induction of HO-1, we also found that to a lesser extent heme potentiated hyperoxia induction of HO-1 (Figure 3).

View larger version (27K): [in this window] [in a new window]   View larger version (14K): [in this window] [in a new window]   Figure 2.   HO-1 mRNA levels in human PA endothelial cells in response to hyperoxia and iron. (a) Northern blot analysis of human PA endothelial cells exposed to room air (C), hyperoxia (H), preloaded with iron (Fe), or iron plus hyperoxia (H + Fe) for 2 to 16 h using radiolabeled human HO-1 and G3PDH cDNAs. (b) Graphical summary of Northern blot analyses illustrating the average relative fold induction of HO-1 mRNA levels for 4 to 15 independent experiments as compared with control conditions.

View larger version (36K): [in this window] [in a new window]   Figure 3.   HO-1 mRNA levels in response to heme plus hyperoxia. Human PA endothelial cells were exposed to hemin 0.5 and 5.0 µM with and without hyperoxia (H) for 16 h. HO-1 mRNA levels were evaluated by Northern blot analysis. This is a representative Northern blot analysis of two independent experiments.

Increasing the free iron pool of the human PA endothelial cells resulted in an increase in HO-1 gene expression by itself within early time points, and, more importantly, potentiated the response to hyperoxia after 8 to 24 h. To determine whether the induction of HO-1 depended on this free iron pool, cells were treated with the iron chelator DFO under basal, hyperoxia, hemin, and hyperoxia plus iron-loaded conditions (Figure 4). Exposure to DFO resulted not only in a loss of hyperoxia induction under both iron-loaded and nonloaded conditions, but also in HO-1 mRNA levels below what was observed under basal conditions. In addition, a decrease in basal expression of HO-1 mRNA was observed upon exposure to DFO. DFO exposure to hemin-treated cells had little effect on HO-1 mRNA levels as compared with hemin exposure alone (Figure 4a). This indicates that HO-1 expression under basal and hyperoxic condition depends on free iron. In addition, the lack of effects of DFO on hemin-exposed cells suggests that heme and hyperoxia induction of HO-1 are under different regulatory mechanisms.

View larger version (41K): [in this window] [in a new window]   View larger version (20K): [in this window] [in a new window]   Figure 4.   DFO treatment of human PA endothelial cells. (a) Representative Northern blot analysis of three individual experiments of human PA endothelial cells exposed to room air, hyperoxia (Hyp), DFO, DFO plus hyperoxia, DFO plus heme, heme, preloaded with iron plus hyperoxia, or DFO plus iron and hyperoxia for 24 h. Total RNA was isolated and HO-1, cathepsin, and G3PDH mRNA levels were identified. (b) A summary of three independent Northern blot analyses of human PA endothelial cells as described above, illustrating the relative levels of HO-1 mRNA.

Several genes involved in iron homeostasis are regulated by cellular iron levels through posttranscriptional mechanisms using the iron response element-regulatory protein for mRNA stability (6, 35). However, HO-1 induction for several other stimuli has been shown to involve a transcriptionally dependent event (14, 36, 37). To determine whether a transcriptional event was necessary for HO-1 induction in response to iron, cells were exposed to iron with or without actinomycin D, an RNA synthesis inhibitor. The iron-dependent HO-1 mRNA induction was inhibited by actinomycin D (Figure 5a). In addition, the induction by hyperoxia or iron plus hyperoxia was also inhibited by actinomycin D, indicating that a transcriptional event was involved (Figure 5b).

View larger version (34K): [in this window] [in a new window]   View larger version (35K): [in this window] [in a new window]   Figure 5.   Actinomycin D treatment of human PA endothelial cells. (a) Human PA endothelial cells were treated with iron (Fe) with and without actinomycin D (A) for 1, 2, and 4 h, and HO-1 mRNA levels were evaluated by Northern blot analysis. (b) Human PA endothelial cells were exposed to room air (C); hyperoxia (H); iron (Fe); actinomycin D (A); iron plus hyperoxia; iron plus actinomycin; or iron, actinomycin, and hyperoxia for 16 h. HO-1 mRNA levels were evaluated by Northern blot analysis.

To determine whether the increase in HO-1 mRNA in human PA endothelial cells in response was a direct increase in transcription of the HO-1 gene, the relative amounts of nascent HO-1 transcripts were evaluated by nuclear run-on assays. Nuclei were isolated from both control and iron-preloaded cells exposed to hyperoxia. G3PDH was used as an internal control, and the relative values of HO-1 expression were corrected according to the constitutive expression of G3PDH. We observed a 7.3-fold increase of HO-1 transcripts in response to hyperoxia exposure of the iron-preloaded cells based on densitometric evaluation of three independent experiments (Figure 6), thereby indicating that the induction of HO-1 in human PA endothelial cells depends at least in part, if not completely, on a direct transcriptional increase of the HO-1 gene.

View larger version (29K): [in this window] [in a new window]   Figure 6.   Nuclear run-on. Nuclei were isolated from both control and iron-loaded plus hyperoxia-exposed cells for 8 h and nascent HO-1 transcripts were evaluated by nuclear run-on assays. The radiolabeled RNA was hybridized to denatured human HO-1 and G3PDH cDNAs dot-blotted on a nylon membrane and was evaluated by autoradiography and densitometry.

To identify hyperoxia-dependent regulatory regions of the human HO-1 promoter, transient transfection experiments were performed using a plasmid containing 4.5 kb of the human HO-1 promoter region and a reporter gene. Transfection of the human 4.5-kb HO-1/hGH (Figure 7a) constructs resulted in basal expression of hGH in human PA endothelial cells; however, we observed no increase above control levels in cells exposed to hyperoxia, preloaded with iron, or with both iron loading and hyperoxia. Exposure to a known potent inducer of HO-1, hemin (11, 34), resulted in a 4-fold increase with the 4.5-kb HO-1/hGH plasmid in human PA endothelial cells (Figure 7b), indicating that this region does contain promoter activity. A cadmium-responsive region was previously identified between 4.5 and 4.0 kb, generating an ~ 3.0-fold increase of the reporter gene (33). We also observed a 3- to 4-fold increase in response to cadmium in human aortic endothelial cells (data not shown) with this construct.

View larger version (11K): [in this window] [in a new window]   View larger version (17K): [in this window] [in a new window]   Figure 7.   Transfection of HO-1/hGH constructs. (a) A human HO-1/hGH plasmid construct containing 4.5 kb to 80 bp of the human HO-1 gene inserted into the BamH1 site of the promoterless hGH vector. (b) The hGH levels from the media of human PA endothelial cells transfected with the human HO-1/hGH construct and exposed to room air (C), hyperoxia (H), preloaded with iron (Fe), iron plus hyperoxia (Fe/H), or heme; n represents the number of individual experiments.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

In this study, we have demonstrated that HO-1 mRNA levels are induced by hyperoxia and that this response is potentiated in the presence of free iron in human PA endothelial cells. This may be an important cellular response, as HO-1 is thought to play a protective role against oxidant injury, including hyperoxia. Overexpression of HO-1 in pulmonary epithelial-like cells protects against hyperoxia (38). The protective effect of HO-1 may be especially important in the endothelium, considering that pulmonary endothelial cells are very sensitive to the toxic effects of oxygen. Injury to the endothelium results in protein leak and an extravasation of heme products into the endothelium, interstitium, and alveolar space. The presence of even small amounts of heme has the potential to amplify greatly cellular damage produced by ROS (21, 22). Heme is a ubiquitous iron-containing compound, and much of the cellular damage involves the collaboration with intracellular iron (22). This iron is then able to interact with oxidants generating the very toxic hydroxyl radical and possibly damaging ferryl compounds. In the cell, free iron is sequestered by ferritin, thereby storing this potentially toxic compound in a safe, yet available, form. Ferritin is normally increased by the liberation of iron from heme, and it appears that HO-1 is an important mediator for this coupled induction of ferritin (39). Therefore, the induction of HO-1 in response to a hyperoxic insult to the lung may be of great benefit. The induction and activation of HO-1 metabolizes the pro-oxidant heme molecule, leading to a reduction of heme and an increase in nonheme iron. This free iron then increases ferritin levels that sequester the free iron, resulting in an overall reduction of the oxidant state of the cell. In addition, the by-product of heme degradation is bilirubin, which in the past has been viewed as a waste product of cellular metabolism. Recently, it has become evident that bile pigments display potent antioxidant activity (23, 24) and, when bound to albumin, can be transported to regions of increased capillary leak offering antioxidant activity.

Even though we observed an increase of HO-1 expression in both rat and human pulmonary endothelial cells, the induction in both cell types was relatively low as compared with what was previously demonstrated in whole lung from rats exposed to hyperoxia or in response to another stimulus such as hemin. This suggests that the induction in vivo may depend on factors other than oxygen tension alone. As discussed previously, hyperoxia results in an extravasation of heme proteins and increases the availability of iron, both of which induce HO-1 expression at much higher levels than does hyperoxia alone. In our cell system, we observed a potentiation of the heme induction of HO-1 with hyperoxia. In addition, we observed a dramatic potentiation of the HO-1 response by increasing the free iron pool of the cells by preloading them with iron. The stimulus of hyperoxia and free iron may be more indicative of what we observe in vivo, as iron is thought to play a major role in oxidative injury (40) and superoxide has the potential to release iron from hemoglobin (41).

The iron-dependent potentiation of HO-1 induction by hyperoxia could occur from iron potentiating the oxidant potential of hyperoxia or by the interaction of iron on gene expression of HO-1. Cellular iron is mostly stored within ferritin molecules; however, there is a free iron pool that is chelatable by DFO (42). This free iron pool is known to have effects on the gene expression of several proteins involved in iron metabolismtransferrin receptor and ferritin (6, 35, 43)and on enzymes potentially involved in oxidant or inflammatory conditionsinducible nitric oxide synthase (44) and the aconitases (45). We have identified HO-1 as another enzyme whose gene expression depends on the iron state of the cell. Iron loading the cells alone produced a transient increase in HO-1 expression that was near baseline within 24 h, at which time there was a dramatic potentiation of HO-1 mRNA by hyperoxia. When this free iron pool is removed with the use of the chelator DFO, there was not only a loss of iron-potentiated hyperoxia induction, but also a loss of hyperoxia-dependent induction under non-iron-loaded conditions. In addition, DFO results in HO-1 mRNA levels below basal expression in both control and hyperoxia-treated cells. Previous work has demonstrated that, similar to our observations with hyperoxia, oxidized low-density lipoprotein (LDL)- mediated HO-1 induction also depends on iron, as pretreatment with DFO decreases oxidized LDL-mediated HO-1 gene induction (46). This indicates that free iron or chelatable iron is a messenger for HO-1 expression under both control and hyperoxia-stimulated conditions.

Increased HO-1 expression for a variety of stimuli, hemin (47), heat shock (36), and oxidants (37, 47) depends on a transcriptional increase of the HO-1 gene. Previously, the hyperoxia induction of the mouse HO-1 gene was also shown to be transcriptionally dependent and required both a basal promoter region and a distal enhancer (14). We also observed that iron- and hyperoxia-dependent induction of HO-1 mRNA in human PA endothelial cells depended at least in part, if not completely, on de novo transcription of the human HO-1 gene. In the mouse HO-1 gene, hemin, heavy metals, and hyperoxia used the same responsive region, termed distal enhancer 1, located ~ 4.0 kb upstream of the transcription initiation site (14, 48, 49). However, we observed a differential response of HO-1 expression with hemin versus iron or hyperoxia. DFO inhibited the HO-1 mRNA induction in response to iron and hyperoxia, yet hemin exposure still resulted in an increased HO-1 expression in the presence of DFO, suggesting that the human HO-1 gene is differentially regulated in response to these stimuli that are unlike the mouse HO-1 gene. This was further verified using the 4.5-kb human HO-1/hGH plasmid for transient transfection studies in the human PA endothelial cells. Again we observed no promoter activity in response to iron and hyperoxia; however, we observed an increase with heme and cadmium, indicating that this region did have promoter activity. This suggests that the iron- and hyperoxia-responsive regions reside outside 4.5 kb of the human HO-1 promoter region.

We have demonstrated that HO-1 is induced in response to hyperoxia in both rat and human pulmonary endothelial cells, however, as compared with the rat model for hyperoxia the induction is relatively small. The induction to hyperoxia is dramatically potentiated in the presence of free iron, which results in a high level of expression similar to other potent inducers of HO-1. Because hyperoxia in the whole-animal model may encompass more than just increased oxygen tension and includes leakage of heme proteins and the increased availability of iron, the iron plus hyperoxia stimulus in cell culture may be more reflective of the in vivo hyperoxic stimulus. Furthermore, we found that HO-1 induction by heme and iron plus hyperoxia was differentially regulated and that the hyperoxia-dependent expression of HO-1 was inhibited by the free iron chelator DFO. This difference between heme and iron/hyperoxia regulation is quite different than what has been seen in the mouse HO-1 gene. Therefore, the molecular mechanisms involved in hyperoxia-dependent induction for the human HO-1 gene cannot be extrapolated by studies of the mouse HO-1 gene. Further studies to identify the promoter elements that control hyperoxia-mediated human HO-1 gene expression are needed to delineate the molecular mechanism underlying HO-1 gene induction in the human gene.