The Yeast Homolog of Heme Oxygenase-1 Affords Cellular Antioxidant Protection via the Transcriptional Regulation of Known Antioxidant Genes*

Heme oxygenase-1 (HO-1) degrades heme and protects cells from oxidative challenge. This antioxidant activity is thought to result from the HO-1 enzymatic activity, manifested by a decrease in the concentration of the pro-oxidant substrate heme, and an increase in the antioxidant product bilirubin. Using a global transcriptional approach, and yeast as a model, we show that HO-1 affords cellular protection via up-regulation of transcripts encoding enzymes involved in cellular antioxidant defense, rather than via its oxygenase activity. Like mammalian cells, yeast responds to oxidative stress by expressing its HO-1 homolog and, compared with the wild type, heme oxygenase-null mutant cells have increased sensitivity toward oxidants that is rescued by overexpression of human HO-1 or its yeast homolog. Increased oxidant sensitivity of heme oxygenase-null mutant cells is explained by a decrease in the expression of the genes encoding γ-glutamylcysteine synthetase, glutathione peroxidase, catalase, and methionine sulfoxide reductase, because overexpression of any of these genes affords partial, and overexpression of all four genes provides complete, protection to the null mutant. Genes encoding antioxidant enzymes represent only a small portion of the 480 differentially expressed transcripts in heme oxygenase-null mutants. Transcriptional regulation may be explained by the nuclear localization of heme oxygenase observed in oxidant-challenged cells. Our results challenge the notion that HO-1 functions simply as a catabolic and antioxidant enzyme. They indicate much broader functions for HO-1, the unraveling of which may help explain the multiple biological responses reported in animals as a result of altered HO-1 expression.

as a positive and negative modulator of the transcription of aerobic and hypoxic genes, respectively (14).
It was recognized only recently that Hmx1p possesses classical heme oxygenase activity (15), raising the possibility that in addition to regulating cellular heme and iron levels, Hmx1p may also share some of the additional activities of mammalian HO-1. Here, we show that Hmx1p indeed is induced in response to different stresses in addition to iron starvation, and that it protects yeast cells against oxidant challenge in a glutathione-dependent manner and via transcriptional regulation of genes encoding known enzymes involved in cellular antioxidant defense. Table  S1 lists the Saccharomyces cerevisiae strains used in this study. The HA-HMX1 wild-type strain, which expresses a triple copy of the hemagglutinin (HA) epitope at the N terminus, was constructed by PCR epitope tagging as described (16) using the plasmid pMPY-3ϫHA (a kind gift from Dr. C. C. Philpott, National Institutes of Health, Bethesda, MD) and the following primers: 5Ј-CAGCACACATACTCACTCACACATA-AAATAACCGCAAAAATAGGGACCAAACGCTGG-3Ј and 5Ј-TAGCTCCTCCATGTCAGTGTGTGAGTGTATGATT-GTATTGCTACTGTCCTTCCTGTAGGGCGAATTGGG-3Ј. Integration of the HA epitope was confirmed by PCR and by Western blotting. Strains were grown in rich YEPD medium (2% w/v glucose, 2% w/v bactopeptone, 1% yeast extract) or minimal synthetic-defined media (0.17% yeast nitrogen base without amino acids, 0.5% ammonium sulfate, 2% w/v glucose) supplemented with appropriate amino acids and bases: 2 mM L-leucine, 4 mM L-isoleucine, 1 mM L-valine, 0.3 mM L-histidine, 0.4 mM L-tryptophan, 1 mM L-lysine, 0.15 mM adenine, 0.2 mM uracil. Media were solidified by the addition of 2% (w/v) agar.
Sensitivity to Oxidants-Cells were grown to exponential phase (A 600 nm ϭ 1) in synthetic medium at 30°C and treated with H 2 O 2 , diamide, or menadione at the concentration and for the time indicated. Aliquots of cells were removed, diluted in fresh YEPD medium, plated in triplicate on YEPD plates, and the number of viable colonies counted after 3 days of culture.
Microarray Hybridization and Data Analysis-Cells were grown in triplicate to exponential phase (A 600 nm ϭ 1) in minimal SD medium. Cells were broken in Trizol reagent (Invitrogen) by three cycles of vigorous mixing in the presence of acid-washed glass beads (45 s) and placed on ice for 30 s. RNA was then extracted according to the manufacturer's instructions, and its quality determined by spectrophotometry (Nanodrop) and by Bioanalyser (Ramaciotti Centre for Gene Function Analysis, University of New South Wales, Australia). Preparation of cRNA, probes, and hybridization to whole yeast genome microarrays (YG-S98, Affymetrix) was performed at the Ramaciotti Centre. Affymetrix Yeast Genome  2.0 Arrays contain probe sets for S. cerevisiae and Schizosaccharomyces pombe. The latter probes were excluded from the analysis and normalization was performed using the robust multi-array average (RMA) (18,19) algorithm implemented in BioConductor. For each individual S. cerevisiae gene (probe set) on the array, fold-change, moderated t-statistics and corresponding p value (20) were calculated. Candidate differentially expressed genes with a significant Bonferroni adjusted p value Ͻ 0.05 and fold-change Ͼ 2.0 (hmx1 versus WT) were identified. Relative mRNA levels of the differentially expressed antioxidant enzymes were determined by RT-PCR. Cells were prepared and RNA extracted as described above. cDNA was synthesized using the SuperScript III First-Strand Synthesis System for RT-PCR (Invitrogen) according to the manufacturer's instructions. The resulting cDNA was then probed for GPX1, GSH1, CTT1, MXR1, GPX1, GPX2, and ACT1 by PCR, using the primers listed in supplemental Table S2. Resulting RT-PCR products were visualized by agarose gel electrophoresis and quantified using ImageJ (NIH).
Biochemical Analyses-Total glutathione was determined based on the reduction of oxidized glutathione (GSSG) by GSSG-reductase and NADPH (21). Cells were harvested at exponential growth phase, washed three times with ice-cold PBS, and resuspended in ice-cold 8 mM HCl/1.3% (w/v) 5-sulfosalicylic acid. Cells were then broken as described above, the resulting mixture clarified by centrifugation (10 min, 13,000 ϫ g, 4°C), and total glutathione determined in the resulting supernatant. Glutathione peroxidase activity was determined using tert-butyl hydroperoxide (Sigma) as substrate. Reactions were started by the addition of cell lysates and followed as the oxidation of NADPH coupled to GSSG reduction by glutathione reductase (22). Catalase activity was determined by the loss of added H 2 O 2 (10 mM) in 50 mM K-phosphate buffer (pH 7.0) containing 0.5 mM EDTA after addition of cell extract (23). Blanks were run in the absence of H 2 O 2 and activity calculated using ⑀ 240 nm ϭ 39.4 M Ϫ1 cm Ϫ1 . Methionine sulfoxide reductase activity was determined as described previously (24). Briefly, the reaction mixture contained 0.4 mM NADPH, 5 mM free Met-R-SO (Sigma), and 5 g of thioredoxin and 0.5 l of thioredoxin reductase (both from Escherichia coli, Sigma). The reaction was started by the addition of cell extract and allowed to take place for 15 min at 37°C. Phosphate-buffered saline (200 l) was then added and loss of NADPH determined immediately in a spectrophotometer using ⑀ 340 nm ϭ 6,220 M Ϫ1 cm Ϫ1 .
Images were captured using a confocal laser microscope (Zeiss LSM Meta 510) with a 100ϫ oil objective, 1.4-numerical-aperture at 12-bit resolution in each channel. Yeast strains, without HA tag were used to control for background staining of the anti-HA Alexa conjugate. All captured images were converted to tagged image files, and Z-stacks collected at Z increments of 0.5 m. Image processing three-dimensional analysis and z projection was performed using Image J software (NIH).
Statistical Analyses-Significant differences between treatments and controls were examined using the Wilcoxon-Mann-Whitney rank sum test. Where appropriate, data were analyzed by one-or two-way ANOVA with post-hoc Bonferroni test as indicated. Significance was accepted at p Ͻ 0.05.

Expression of Hmx1p Is Induced in Response to Different
Oxidants-As HO-1 is induced in response to oxidant stress (4), we tested whether Hmx1p expression is induced similarly in yeast cells, using an HA-tagged HMX1 strain and Western blotting with an anti-HA antibody (9). We noted modest expression of Hmx1p in control, unstressed cells (Fig. 1A, lane 1). Iron starvation increased Hmx1p expression (Fig. 1A, lane 2), confirming a finding reported previously by others (9). What is novel, however, is that Hmx1p was also induced when cells were exposed to H 2 O 2 , diamide (a membrane permeable, thiol-specific oxidant that reacts rapidly with reduced glutathione, GSH) and the redox-cycling drug menadione (that generates superoxide anion radicals) (Fig. 1A, lanes [3][4][5]. Treatment of cells with H 2 O 2 caused a time- (Fig. 1B) and oxidant concentration-dependent ( Fig. 1C) increase in Hmx1p. Immunocytochemistry staining and confocal laser scanning microscopy confirmed low level of Hmx1p expression under standard, iron replete growth conditions ( Fig. 2A, left panel arrows and boxed enlargements), and the increase in Hmx1p expression in cells exposed to iron starvation or H 2 O 2 (Fig. 2, A and B). Yeast strains without HA tag yielded low background staining (Fig. 2C). The extent of H 2 O 2 -mediated increase in Hmx1p expression was comparable for N and C terminus HA-tagged HMX1 strains (Fig. 2B). Whereas both iron depletion and H 2 O 2 increased Hmx1p expression visibly, the expression pattern was heterogeneous, with very high expression in a subset of cells, but low to non-detectable expression in most cells ( Fig. 2A). Hmx1p expression was also increased in response to other known inducers of HO-1, such as rapamycin and heat shock (not shown). The observed similarity between induction of HO-1 and Hmx1p in response to oxidative stress supports the contention that Hmx1p is part of the antioxidant defense in yeast cells.
HMX1 Affects Sensitivity to H 2 O 2 , Diamide, and Menadione-To test this possibility we compared the oxidant sensitivity of wild type and hmx1 mutant cells using concen-tration-response curves to H 2 O 2 , diamide, and menadione. At all concentrations tested, the hmx1 mutant was more sensitive to the oxidants than the wild-type strain ( Fig. 3A and supplemental Fig. S1A). Conversely, overexpression of HMX1 increased the resistance of wild-type and hmx1 mutant strains to H 2 O 2 (Fig. 3B), diamide, and menadione (supplemental Fig.  S1B). For the hmx1 mutant strain, overexpression of HMX1 restored the oxidant resistance to that of wild-type cells. Similarly, overexpression of HMX1-HA also fully rescued the oxidant sensitivity of the hmx1-null mutant ( Fig. 3C and supplemental Fig. S1C), confirming that the HA-tagged Hmx1p used in our studies was functional. In addition, overexpression of Following three-dimensional image stack acquisition, the collected composite files of the z-stacks were analyzed and orthogonal planes were projected with the Image J software. Cells were counted that displayed notable staining with the anti-HA Alexa 488 antibody. Quantification of fluorescence images represents data (mean Ϯ S.E.) from a single experiment with 3-8 independent z-stacks per treatment taken with identical settings. C, images of HA-tagged control cells. Parallel control cultures not expressing any HA-tagged constructs were prepared and microscopy performed with identical imaging settings as in A. Transmitted light (TL), anti-HA Alexa 488 (green), and DAPI (red) images are shown. recombinant human HO-1 similarly rescued the oxidant sensitivity of the hmx1 mutant (Fig. 3D). Together, these data indicate that Hmx1p is induced by, and offers protection against, oxidative stress, similar to the situation with HO-1 in mammalian cells.
The hmx1 Mutant Has Altered Expression of Cellular Antioxidant Enzymes-To determine if the transcriptional response of the hmx1-null mutant is altered compared with that of the wild-type strain, Affymetrix microarray analyses were carried out. Loss of HMX1 significantly affected the transcriptome, with 265 open reading frames up-regulated (supplemental Table S3) and 215 down-regulated (supplemental Table S4) (Fig. 4A). Five gene ontologies were significantly over-represented in the up-regulated transcripts: response to stress, sulfur metabolic process, transcription factor activity, antioxidant activity, and transmembrane transporter activity. Two gene ontologies were significantly over-represented in the down-regulated transcripts: RNA processing and ribosome biogenesis.
We further investigated five transcripts encoding enzymes contributing to cellular antioxidant defense (Fig. 4B), the altered transcription of which was confirmed by RT-PCR (Fig.  5). Of these, GSH1, GPX1, CTT1, and MXR1 were down-regulated, while GPX2 was up-regulated. GSH1 encodes ␥-glutamylcysteine synthetase, which catalyzes the first step in the synthesis of GSH (26). It protects cells by scavenging oxidants and by acting as cofactor for several antioxidant enzymes (27). GPX1 and CTT1, respectively, encode a glutathione peroxidase that acts on phospholipid hydroperoxides and other organic peroxides (22,28), and a cytosolic catalase that forms part of a H 2 O 2 detoxification system and is redundant with the glutathione system (29). MXR1 encodes a peptide methionine sulfoxide reductase, which reduces methionine sulfoxide residues in proteins (30). GPX2 encodes an atypical 2-Cys peroxiredoxin, responsible for the reduction of hydroperox-  ides using thioredoxin rather than GSH as the preferred cofactor (31). This difference in cofactor preference may explain why GPX1 and GSH1 expression were down-regulated, while GPX2 was up-regulated in the hmx1 mutant. In addition to the five genes encoding known antioxidant enzymes, seven other transcripts with potential indirect participation in oxidative stress were also differentially expressed (supplemental Tables S3 and S4). The role of these genes was not investigated further.
Overexpression of Down-regulated Antioxidants Rescues Oxidant Sensitivity of hmx1 Mutant-To define the mechanism of antioxidant protection by Hmx1p, each of the downregulated antioxidant enzymes was overexpressed separately and the effect of this on rescue of hmx1 oxidant sensitivity determined. Overexpressing either GPX1 (Fig. 6A), GSH1 (Fig. 6B), CTT1 (Fig. 6C), or MXR1 (Fig. 6D) increased resis-tance of wild-type cells to diamide, H 2 O 2 , and menadione. More importantly, overexpression of any of these genes also increased the resistance of the hmx1 mutant to each of the three oxidants tested (Fig. 6, A--D). The extent of this increased resistance was less than that seen in the corresponding wild-type cells, indicating that each of the four known antioxidant enzymes alone partially rescued the oxidant sensitivity of the hmx1 mutant strain. In contrast, simultaneous overexpression of all four transcripts encoding the antioxidant genes in the hmx1 mutant (supplemental Fig. S2) completely restored its oxidant resistance to that of wild-type cells at all concentrations of H 2 O 2 tested (Fig. 7).
HMX1 Relates to Cellular Antioxidant Activities-Consistent with the observed oxidant sensitivity, total glutathione concentration in the hmx1 mutant was only ϳ20% of the wild-type strain value, and overexpression of HMX1 in the hmx1 mutant increased total glutathione to above wild-type levels (Table 1). Similarly, in the hmx1 mutant, catalase and glutathione peroxidase activities were decreased compared with wild-type cells, and overexpression of HMX1 increased the activity of both enzymes to above the corresponding wildtype values ( Table 1). The activity of methionine sulfoxide reductase was below the limit of detection in all strains, except wild-type cells overexpressing MXR1. Together, these data show that the extent of HMX1 expression relates to the cellular activities of the differentially expressed antioxidant enzymes identified in the microarray experiments.
We next examined whether in hmx1 cells overexpression of each of the HMX1 related antioxidant enzymes affected the activities of the other down-regulated antioxidants. As shown in Fig. 8, overexpression of GSH1 increased both the content of total glutathione and glutathione peroxidase activity. Also, overexpression of GPX1, but not CTT1 or MXR1, increased glutathione peroxidase activity (Fig. 8A). In the case of catalase, only overexpression of CTT1 increased the activity of this enzyme (Fig. 8B), while overexpression of GSH1 increased the levels of total glutathione (Fig. 8C). These results FIGURE 5. Deletion of HMX1 or overexpression of HMX1 affects the expression of antioxidant genes. Wild type (empty bars) and hmx1 mutant strains (gray bars) without (no stripes) and with overexpression of the multicopy plasmid containing HMX1 (striped) were grown to exponential phase, the RNA extracted, and cDNA generated. The resulting cDNA was then probed for GSH1, GPX1, CTT1, MXR1, GPX2, and ACT1 as described under "Experimental Procedures," and antioxidant gene expression shown relative to that of ACT1. Results represent mean Ϯ S.E. of three separate experiments. *, p Ͻ 0.05 compared with the corresponding HMX1 overexpressing strain (one-way ANOVA with Bonferroni correction). indicate that the differentially expressed antioxidant genes, at least in part, acted independently from each other.
The fact that overexpression of all four transcripts encoding antioxidant genes was required to completely restore the oxidant resistance of the hmx1 mutant (Fig. 7), suggested that the products of the reaction catalyzed by Hmx1p themselves did not provide substantial oxidant protection. Consistent with this interpretation, addition of biliverdin and the COreleasing molecule CORM3, singly or together, failed to offer the hmx1-null mutant protection against 4 mM H 2 O 2 (Fig. 9). Similar results were observed when cells were exposed to a lower H 2 O 2 concentration (supplemental Fig. S3).
Dennery and co-workers (32) recently reported HO-1 to localize to the nucleus of heme-treated mammalian cells and to activate transcription factors important in oxidative stress. In yeast, the nuclear and ER membranes are continuous (33), making it difficult to discriminate perinuclear from nuclear localization. We therefore use the term (peri)nuclear hereafter to refer to perinuclear or nuclear localization. We observed (peri)nuclear localization of Hmx1p in cells exposed to 4 mM H 2 O 2 for 6 h, as assessed by microscopy and biochemical analyses (Fig. 10). Similar results were observed with cells stressed with H 2 O 2 for 1 h (supplemental Fig. S4), indicating that (peri)nuclear localization of Hmx1p occurred even after short periods of oxidant stress. In contrast, (peri)nuclear Hmx1p was not detected in cells in the absence of H 2 O 2 . Following oxidant treatment, (peri)nuclear expression of Hmx1p was observed in only a small subset of cells (Fig. 10, A and B), with Hmx1p expressed more commonly in ER regions associated with membranes other than the nuclear membrane (supplemental Fig. S5). As in stressed mammalian cells nuclear localization has been reported to be preceded by calpain-mediated cleavage of the HO-1 C terminus (32), we compared the extent of (peri)nuclear localization of Hmx1p in N versus C terminus HA-tagged HMX1 strains exposed to H 2 O 2 . We observed (peri)nuclear localization with both strains of yeast as assessed by confocal fluorescence microscopy (Fig. 10A) and biochemical analysis ( Fig. 10C and supplemental Fig. S4). However, the extent was greater for N terminus than C terminus HA-tagged Hmx1p (Fig. 10, B and D and supplemental  Fig. S4). As expected from the close physical association of nuclear and ER membranes, the nuclear fraction contained ER markers (Dpm1 and Kar2, Fig. 10C), disallowing unambiguous localization of Hmx1p to the nucleus.

DISCUSSION
The ability of cells to respond to changes in environmental conditions, such as nutrient availability, determines their sur-  Simultaneous overexpression of GPX1, CTT1, GSH1, and MXR1 completely rescues oxidant sensitivity of hmx1 deletion mutant. Wild type and hmx1 mutant strains were transformed with 4 galactose-inducible multi-copy plasmids containing GSH1, GPX1, CTT1, and MXR1. Cells were grown to exponential phase in raffinose-(E) or galactose-containing medium (F) and treated for 1 h with H 2 O 2 at the indicated concentration. Survival is expressed as percentage of that seen with control cells. Results shown are mean Ϯ S.E. of a single experiment performed in triplicate, with standard error bars smaller than the symbols. *, p Ͻ 0.05 compared with the corresponding non-induced strain (two-way ANOVA with Bonferroni correction). There is no significant difference between wild type and hmx1 mutant strains with all four antioxidant genes induced. FIGURE 8. Altering the levels of antioxidants alters the activity of some but not all antioxidants. hmx1 mutant strains were transformed with a galactose-inducible multi-copy plasmid containing GSH1 (1), GPX1 (2), CTT1 (3), or MXR1 (4). Cells were grown to exponential phase in raffinose-(noninduced, open bars) or galactose-containing medium (induced, closed bars) before the activity of (A) glutathione peroxidase or (B) catalase, and (C) total glutathione was determined as described under "Experimental Procedures." Results shown represent mean Ϯ S.E. of three separate experiments. *, p Ͻ 0.05 compared with the corresponding wild-type strain (one-way ANOVA with Bonferroni correction). There were no differences between non-induced hmx1 mutant and (A) the hmx1 mutant with CTT1 or MXR1 induced, (B) GSH1, GPX1, or MXR1 induced, and (C) GPX1, CTT1, or MXR1 induced. vival. Until now, Hmx1p, the yeast homolog of HO-1, was believed to be involved only in the cellular response to iron limitation (9). Here we provide evidence that an additional role of Hmx1p is in the cellular response to and protection against oxidative stress. Our studies, for the first time, show that this antioxidant activity of heme oxygenase is dependent on the up-regulation of several genes encoding known antioxidant enzymes. Several lines of evidence support the conclusion that Hmx1p affords cellular antioxidant protection, and that this is via the transcriptional regulation of known antioxidant genes rather than its oxygenase activity. First, the expression of Hmx1p, like that of HO-1, is induced in cells exposed to different oxidants. Secondly, deletion of HMX1 renders cells more sensitive to oxidant challenge while overexpression of HMX1, like human HO-1, restores oxidant resistance of the hmx1-null mutant to that of wild-type cells. Thirdly, microarray analyses revealed GSH1, GPX1, CTT1, and MXR1 that encode well-established antioxidant defense enzymes to be down-regulated in the hmx1-null mutant compared with wild-type cells. Fourthly, overexpression of each of the downregulated antioxidant genes partially rescues oxidant sensitivity of the hmx1-null mutant, while overexpression of all four down-regulated antioxidant genes provides complete protection to the null mutant. Furthermore, the levels of HMX1 transcript mirrored (at least in the case of total glutathione, catalase, and glutathione peroxidase) the antioxidant activities of the genes down-regulated in the hmx1-null mutant. Therefore, changes to cellular glutathione and GSH-related antioxidant activities are likely key mechanisms by which Hmx1p protects cells against oxidants. This interpretation is consistent with the fact that in human cells one general mechanism of HO-1 induction is via modulation of cellular glutathione status (34).
It is now well established that HO-1 protects mammals against oxidative stress. For example, mice deficient in HO-1 have increased susceptibility to oxidative stress (35), and their cells are less capable to withstand an oxidative challenge than the corresponding wild-type cells (36), while HO-1 overexpression increases cellular resistance to oxidants (37). However, the mechanism underlying the antioxidant protection provided by HO-1 is not well understood. Early studies ascribed the antioxidant action to the HO-1 ability to simultaneously decrease the concentrations of the pro-oxidant heme and increase the levels of the antioxidant bilirubin (38). Bilirubin is an efficient oxidant scavenger in vitro (39), and when added at micromolar concentrations both bilirubin and the HO-1 substrate hemin protect cells against oxidants (40). However, there is little direct evidence that bilirubin produced from endogenous heme as a consequence of increased HO-1 activity acts as a cellular antioxidant (41), and the amounts and sources of cellular heme available for degradation by HO-1 remain unknown. Indeed, addition of the products of heme oxygenase, biliverdin, and CO, had no measurable effect on the sensitivity of the hmx1-null mutant to H 2 O 2 challenge (Fig. 9). Together, our results indicate that the antioxidant protection afforded by the yeast homolog of HO-1 is via an adaptive response that involves the transcriptional control of antioxidant genes, rather than directly via its oxygenase activity (Fig. 11). Transcriptional regulation by Hmx1p was observed with millimolar concentrations of H 2 O 2 , raising the question of physiological relevance. However, extrapolating oxidant concentrations from yeast to mammalian cells is complicated because of several species differences, including the oxidant resistance of the cell wall compared with plasma membrane (42), and redundancies in H 2 O 2 metabolism (29).
In addition to catalyzing heme degradation and providing antioxidant protection, it is increasingly appreciated that HO-1 participates in the regulation of many biological processes, including inflammation, cell growth, vascular tone, and angiogenesis (4) that can translate into protection against various diseases (7,43,44). This raises the intriguing question of how HO-1 achieves these various activities. Using a global microarray analysis approach and yeast as a model, our studies revealed an unexpectedly large number of heme oxygenase-dependent, differentially expressed genes, the products of which are involved in several previously unrecognized processes, such as RNA processing, ribosome biogenesis, transcriptional regulation, and membrane transport (Fig. 4). In fact, of the differentially expressed genes, only a small number, corresponding to ϳ1%, relate to antioxidant defense. This indicates that at least in yeast, antioxidant protection may represent a relatively minor function of heme oxygenase, and that the enzyme likely participates in many presently unappreciated processes. For example, consistent with a regulatory role of HO-1 in the growth of mammalian cells, loss of HMX1 increased the growth of yeast cells, and this was abrogated by overexpression of HMX1 (not shown). This may relate to the differentially expressed genes involved in RNA processing and ribosome biogenesis (supplemental Tables S3  and S4). The down-regulation of transcripts encoding proteins involved in sulfur metabolism has been linked to increased oxidative stress (45) and hence may help explain the oxidant sensitivity of the hmx1 mutant. Likewise, mammalian HO-1 has been reported to affect the activity of transcription factors (32), so that the down-regulation of transcription factors may help explain the inability of the hmx1 mutant to mount an adequate response to oxidative stress.
Our microarray analyses also raise the question of how an ER protein can participate in transcriptional regulation. One possibility is that diffusible product(s) of heme oxygenase activity are involved. Indeed, a recent report suggested that in cardiomyocytes HO-1-derived CO regulates mitochondrial biogenesis via transcriptional regulation of nuclear respiratory factor-1 (46). Inconsistent with this notion, however, transfection of cells with a mutant HO-1 that lacks enzymatic activity still increased cellular resistance to H 2 O 2 (37), and we ob- served no protective effect of CO on the sensitivity of the hmx1-null mutant to H 2 O 2 challenge (Fig. 9). We note however that our experimental design may not have adequately reflected local production of biliverdin/CO, perhaps close to or within the nucleus. An alternative explanation for the ability of HO-1 to regulate gene transcription has been provided by Dennery and co-workers (32) who reported that a C terminus truncated form of HO-1 migrates to the nucleus in response to hypoxia, hemin, and heme-hemopexin. We observed that in oxidant-stressed yeast cells, Hmx1p localized to the (peri)nuclear region where it could conceivably participate in the transcriptional regulation. We observed greater (peri)nuclear localization with N terminus than C terminus tagged Hmx1p, suggesting that translocation to the perinuclear region may precede cleavage. However, our results do not unambiguously establish nuclear localization or cleavage of Hmx1p. Indeed, integral ER proteins can enter the nucleus without a need for proteolytic cleavage (33). Clearly, additional studies are required to elucidate the mechanism by which Hmx1p affects transcriptional regulation.
In summary, our data show that in yeast the HO-1 homolog provides antioxidant protection to cells indirectly via the differential expression and activities of several known antioxidant enzymes. Our findings challenge the paradigm that the cellular/biological effects of HO-1 are explained solely by its enzymatic activity. Instead, they suggest that at least some of the HO-1 cellular/biological effects are the result of an adaptive response by the cells (Fig. 11). Clearly, HMX1 regulates the expression of a large number of genes involved in numerous functions unrelated to heme oxygenase or antioxidant activities. Elucidating the relationship of heme oxygenase with these differentially expressed genes will likely unravel a multitude of novel functions of HO-1.