Regulation of intracellular heme levels by HMX1, a homologue of heme oxygenase, in Saccharomyces cerevisiae.

Saccharomyces cerevisiae responds to iron deprivation by increasing the transcription of genes involved in the uptake of environmental iron and in the mobilization of vacuolar iron stores. HMX1 is also transcribed under conditions of iron deprivation and is under the control of the major iron-dependent transcription factor, Aft1p. Although Hmx1p exhibits limited homology to heme oxygenases, it has not been shown to be enzymatically active. We find that Hmx1p is a resident protein of the endoplasmic reticulum and that isolated yeast membranes contain a heme degradation activity that is dependent on HMX1. Hmx1p facilitates the capacity of cells to use heme as a nutritional iron source. Deletion of HMX1 leads to defects in iron accumulation and to expansion of intracellular heme pools. These alterations in the regulatory pools of iron lead to activation of Aft1p and inappropriate activation of heme-dependent transcription factors. Expression of HmuO, the heme oxygenase from Corynebacterium diphtheriae, restores iron and heme levels, as well as Aft1p- and heme-dependent transcriptional activities, to those of wild type cells, indicating that the heme degradation activity associated with Hmx1p is important in mediating iron and heme homeostasis. Hmx1p promotes both the reutilization of heme iron and the regulation of heme-dependent transcription during periods of iron scarcity.

Iron is an essential nutrient for virtually all organisms, yet many organisms thrive in environments where the availability of iron is extremely low. Saccharomyces cerevisiae can survive tremendous changes in the abundance of environmental iron, and the strategies that allow cells to adapt to these changes are only partially understood. Under conditions of iron deprivation, S. cerevisiae expresses a set of genes involved in the acquisition of iron from the environment and the mobilization of iron from intracellular stores. These genes are under the control of the major iron-dependent transcriptional activator, Aft1p (1). A related transcription factor, Aft2p, also regulates a subset of the Aft1p regulon, but the role of Aft2p in iron homeostasis is not yet clear (2,3). The genes most strongly induced by Aft1p during iron deprivation are the cell wall mannoproteins encoded by FIT1, FIT2, and FIT3, which facilitate iron uptake by increasing the association of iron chelates with the cell wall (4). Uptake of iron compounds at the cell surface can occur through a reductive mechanism that employs members of the FRE family of plasma membrane metalloreductases (5,6). Fre1p and Fre2p reduce ferric iron salts and iron chelates to the ferrous form, whereas Fre3p and Fre4p catalyze the reduction of ferric siderophores (7). Siderophores are low molecular mass compounds secreted by bacteria and fungi that bind ferric iron with exceptionally high affinity. Fre5p and Fre6p have not been functionally characterized, whereas Fre7p is copper-regulated and not an Aft1p target gene (8). Reduced iron is taken up through a high affinity transport complex consisting of the Fet3p multicopper oxidase (9) and the Ftr1p iron permease (10). Copper loading of Fet3p occurs in a post-Golgi compartment and requires the activities of the Atx1p copper chaperone (11) and the Ccc2p copper transporter (12). A second, nonreductive system of high affinity iron uptake exhibits specificity for siderophore-iron chelates. S. cerevisiae takes up intact siderophore-iron chelates through the transporters encoded by ARN1, ARN2/TAF1, ARN3/SIT1, and ARN4/ENB1 (13)(14)(15)(16)(17)(18).
The Aft1p regulon includes a set of genes involved in the transport of metals into and out of the vacuole, a site of metal storage. 1 The Ftr1p homologue, Fth1p, associates with a Fet3p homologue, Fet5p, and facilitates the mobilization of iron stores from the vacuole (19). Smf3p, which exhibits homology to mammalian iron transporters, is also localized to the vacuole and may participate in the mobilization of stored iron (20). Cot1p is a vacuolar transporter involved in zinc homeostasis (21) and cobalt resistance (22) and may serve to sequester other metals taken up under conditions of iron deprivation (23). The plasma membrane biotin transporter encoded by VHT1 is an Aft1p target, and biotin uptake and synthesis are reciprocally regulated by iron. 1 Additional Aft1p targets are two genes of unknown function: TIS11, a member of the CCCH family of zinc finger proteins, and HMX1, which exhibits homology to heme oxygenases.
Heme oxygenases are found in a wide variety of organisms including bacteria, fungi, plants, and animals, where they play important roles in cellular metabolism. Heme oxygenase catalyzes the oxidative breakdown of heme to biliverdin, carbon monoxide, and iron in a reaction that requires three oxygen molecules and seven electrons (24). In pathogenic bacteria, heme oxygenase activity is important for iron acquisition from host heme and heme proteins (25,26). In higher plants and cyanobacteria, heme oxygenase is required for the synthesis of the light-harvesting pigments, which contain tetrapyrrols derived from biliverdin (27,28). Candida albicans expresses a heme oxygenase that is required for the utilization of heme as a nutritional iron source (29). Mammals express three isoforms of heme oxygenase and mice deficient in HO-1 are anemic and accumulate iron in liver and renal cells because of a defect in heme iron reutilization (30). Carbon monoxide produced by mammalian heme oxygenases can act as a signaling molecule (24).
Unlike C. albicans, S. cerevisiae lacks a high affinity system for heme uptake and is inefficient in the use of heme as a nutritional iron source (31). Previous investigators have reported that S. cerevisiae lacks heme oxygenase activity, and a recent report indicates that when Hmx1p is overexpressed and purified, it has no heme oxygenase activity, although it does bind heme and exhibits weak peroxidase activity (32). Although HMX1 is transcriptionally activated by iron deprivation via Aft1p, the role of Hmx1p in the response to iron deprivation remains unclear. Here we report that deletion of HMX1 resulted in a decrease in heme degradation activity and that Hmx1p was important for heme iron reutilization and the homeostasis of regulatory pools of iron and heme.

EXPERIMENTAL PROCEDURES
Yeast Strains, Media, and Plasmids-Strains were constructed in YPH499 (MATa ura3-52 lys2-801(amber) ade2-101(ochre) trp1-63 his3-200 leu2-1) (Yeast Genetic Stock Center, Berkeley, CA). PCR-mediated gene disruption was used to generate deletions of HMX1 (33). For deletion of HMX1, the YPH499 strain was transformed with a PCR product amplified from genomic DNA isolated from the strain RG4154 (MATa ura3 met-15 his3-1 leu2-1 ylr205c⌬::KANR) (Research Genetics) using the following primers 5Ј-CCGTGGAATTTGAGCCATGTAAC-3Ј and 5Ј-GCCTCCCATGCAAGAGGGC-3Ј. Geneticin-resistant clones were selected on YPD plates containing 80 mg/liter G418 (Invitrogen), and transformants were screened by PCR to isolate the hmx1⌬ strain. Deletion of FET3 was performed as described (34). The HMX1-HA strain, which expresses a triple copy of the hemagglutinin (HA) 2 epitope at the carboxyl terminus, was constructed by PCR epitope tagging as described using the plasmid pMPY-3ϫHA (35) and the following primers: 5Ј-GCGATCTGGGTTCTTTACTTCTTGGTAAAGAGTTTTCTTAG-CATAGTAAGGGAACAAAAGCTGGAG-3Ј and 5Ј-TTTTTCTCTTGCT-GTTTTTCCTTCCCTATTCTTCATATTTTGATATTACTCTAGGGCGA-ATTGGGT-3Ј. Integration of the HA epitope was confirmed by PCR and by Western blotting. HA-tagged Hmx1p was confirmed to be functional by observing that the ␤-galactosidase activity of a FET3-lacZ reporter plasmid was identical in the HMX1-HA and the congenic HMX1ϩ strains. Rich medium (YPD) and synthetic defined medium (SD) were prepared as described (36). Defined iron media were prepared as described (37) using yeast nitrogen base without iron and 1 mM ferrozine, an iron chelator, in addition to the indicated iron supplement. Hemin was added as a complex with bovine serum albumin. The plasmid pHmuO, which carries the HmuO gene of Corynebacterium diphtheriae under the control of the FET3 promoter, was used for expression of HmuO in yeast. Part of the FET3 promoter was amplified by PCR using the following primers 5Ј-CACACACTCGAGCTTCAAAAGTGCACCCATTT-GCAG-3Ј and 5Ј-ACACACGGATCCCATCTAGTTCTAATTTTTTGCTA-C-3Ј and then digested with XhoI and BamHI. A 0.65-kb DNA fragment carrying the HmuO gene was generated by PCR from plasmid pTH793 (gift of Dr. M. P. Schmitt) (25) using the following primers: 5Ј-CACA-CAGGATCCATGACCACTGCAACCGCAGGC-3Ј and 5Ј-CACACAGAG-CTCCCATGGTTACAGGCCTTTACCCAAATC-3Ј and was digested with BamHI and SstI. Both products were ligated into the XhoI and SstI sites of the plasmid pRS404 to yield pHmuO. The plasmid was linearized with BstXI and used to transform the hmx1⌬ and YPH499 strains to tryptophan prototrophy. The sequences were aligned using MacVector (Oxford Molecular Group).
Immunofluorescence and Western Blotting-The strains were grown to mid-log phase in defined iron medium containing 10 M ferrous ammonium sulfate to induce the expression of Hmx1p. The cells were prepared for immunofluorescence microscopy as described (18). For Western blotting, the cells were disrupted with glass beads, and unbroken cells were removed by centrifugation at 500 ϫ g for 2 min. The membrane fraction was isolated by centrifugation at 18,000 ϫ g for 30 min. Western blotting was performed using a 1:5000 dilution of HA.11 as the primary antibody followed by 1:5000 dilution of horseradish peroxidase-conjugated sheep anti-mouse antibody (Amersham Biosciences). Antibody was detected using enhanced chemiluminescence (Amersham Biosciences).
Assays-Heme degradation activity was measured by release of 55 Fe from 55 Fe-hemin as described (38) with the following modifications. The membranes were isolated as for Western blotting, and the protein concentrations were determined in samples solubilized with 3% N-lauroylsarcosine. The 100-l reaction mixture contained 0.1 mg of membrane protein, 50 mM HEPES, pH 7.4, 1 mM EDTA, 5 mM sodium ascorbate, and 0.4 M 55 Fe-hemin in complex with bovine serum albumin (39). The reactions were incubated 15 min at 30°C. 10 l of 1 mM unlabeled hemin/bovine serum albumin was added to stop the reaction. 40 l of 100% trichloroacetic acid was added to the reaction to precipitate intact 55 Fe-hemin and proteins. The samples were incubated on ice for 30 min and centrifuged at 18,000 ϫ g for 15 min, and 50 l of supernatant was used for liquid scintillation counting. Boiled samples (10 min at 100°C) of membranes were used as a control for nonenzymatic degradation of 55 Fe-hemin. Heme degradation activity was determined after subtraction of nonenzymatic controls and expressed as a percentage of the wild type activity. 55 Fe-heme was synthesized as described (40) from protoporphyrin IX (Porphyrin Products, Logan, UT) and 55 FeCl 3 (New England Nuclear). Autoradiographs of thin layer chromatography on silica plates (EM Science) confirmed a single band identical to the unlabeled hemin standard.
To measure heme and iron content, the cells were grown overnight in SD medium supplemented with 0.4 M 55 Fe-ferrichrome. The cells were washed three times with medium without 55 Fe-ferrichrome and two times in 50 mM HEPES buffer, pH 7.4. The cells were disrupted with glass beads in the same buffer, and aliquots of cell lysate were used to determine iron content by scintillation counting. 55 Fe-heme was extracted with n-butyl-acetate as described (41). For ␤-galactosidase assays, the cells transformed with either a FET3-lacZ (gift of D. Winge) or a CYC1-lacZ (gift of A. Hinnebusch) reporter plasmid were grown to mid-log phase in defined iron medium containing the indicated amounts of ferrous ammonium sulfate. ␤-Galactosidase assays were performed as described (42).

RESULTS
Sequence Analysis of Hmx1p-Hmx1p is 21% identical at the amino acid level to human HO-1 and 19% identical to HmuO, the heme oxygenase from C. diphtheriae. HmuO was selected for analysis because it is a well characterized prokaryotic heme oxygenase, and we examined a sequence alignment between Hmx1p, HO-1, and HmuO (Fig. 1). The crystal structure of human HO-1 has been reported, and the amino acid residues involved in heme binding and interaction with cytochrome P-450 reductase have been identified (43,44). Although the amino acid residues involved in binding heme and cytochrome P-450 reductase are conserved between HO-1 and HmuO, most of these residues are not conserved in Hmx1p. Specifically, the proximal heme ligand, His-25 in HO-1, is not conserved in Hmx1p, although this protein has been shown to bind heme (32). A somewhat higher level of sequence conservation was observed in the "heme oxygenase motif" that is highly conserved among heme oxygenase family members (24). A potential carboxyl-terminal transmembrane domain was observed in Hmx1p, which suggested that it was similar to mammalian heme oxygenases in its localization to membranes. These observations indicated that it was not possible to determine whether HMX1 encoded a heme oxygenase enzyme solely on the basis of its primary structure.
Localization of Hmx1p to the Endoplasmic Reticulum and Regulation of Protein Levels by Iron-Bacterial heme oxygenases are soluble proteins of the cytosol, whereas mammalian heme oxygenases are integral membrane proteins of the endoplasmic reticulum (24). We constructed a strain containing three copies of the HA epitope fused to the carboxyl terminus of HMX1 and examined the localization of Hmx1-HA by indirect immunofluorescence and by membrane fractionation and Western blotting. By immunofluorescence, Hmx1-HA was detected at the periphery of the cell and in an intracellular ring-like structure ( Fig. 2A). This intracellular structure was confirmed to be the nucleus by staining with DAPI (Fig. 2, B and C). The peripheral and perinuclear distribution of Hmx1-HA in a "signet ring" pattern was identical to the characteristic pattern of proteins located in the endoplasmic reticulum of yeast (45)(46)(47). We confirmed that Hmx1p was localized to membranes by preparing lysates from cells expressing Hmx1-HA, separating the intracellular membranes from the cytosol by centrifugation at 18,000 ϫ g and analyzing the resulting fractions by Western blotting (Fig. 2D). Hmx1-HA was detected predominately in the membrane fraction, and together these data suggested that Hmx1p was an integral membrane protein of the endoplasmic reticulum, a localization identical to that of mammalian heme oxygenases.
HMX1 is transcriptionally activated through Aft1p by iron deprivation, 1 cobalt stress (48), and defective mitochondrial iron-sulfur cluster synthesis (49), and we questioned whether protein levels of Hmx1p would also reflect iron-mediated regulation. Cells of the Hmx1-HA strain were grown in medium containing varying concentrations of iron, the lysates were prepared, and Hmx1p levels were analyzed by Western blotting (Fig. 2E). Hmx1p was abundant in cells grown in low concen-trations of iron, but much less was detected in cells grown in moderate concentrations of iron, and Hmx1p was virtually undetectable at the highest concentration of iron. These data are consistent with a role for Hmx1p in the response to iron deprivation as well as a role during growth in optimal iron conditions.
Heme Degradation Activity of Hmx1p-To address whether Hmx1p was involved directly in the degradation of heme, we constructed an hmx1⌬ strain and measured the capacity of isolated membranes to catalyze the release of 55 Fe from 55 Feheme (Fig. 3). Degradation of heme through the heme oxygenase reaction requires reducing equivalents, and ascorbate can provide these to reactions catalyzed by HO-1 (50) and HmuO (51). The addition of ascorbate stimulated heme degradation in the wild type strain but failed to stimulate degradation in the hmx1⌬ strain. The membranes derived from the hmx1⌬ strain exhibited less than half of the heme degradation activity that was present in membranes from wild type cells. These data indicated that Hmx1p functioned in the degradation of heme but also suggested the presence of an additional mechanism for heme degradation in yeast membranes.
The Role of Hmx1p in the Utilization of Heme as a Nutritional Iron Source-Although S. cerevisiae lacks a high affinity transport system for heme, exogenous heme can rescue the metabolic defects of yeast carrying mutations in the heme biosynthetic pathway, indicating that heme can be taken up by this species (52). We investigated the role of Hmx1p in the utilization of heme as a nutritional iron source by measuring the growth of congenic HMX1ϩ and hmx1⌬ strains in medium containing hemin as the only source of iron. Because hemin samples can contain small amounts of free iron, we inactivated the high affinity ferrous transport system in both strains by deletion of FET3. The addition of hemin to iron-depleted medium stimulated the growth of both the HMX1ϩ and hmx1⌬ strains in a dose-dependent manner (Fig. 4A), indicating that S. cerevisiae can use hemin as an iron source. However, the HMX1ϩ strain consistently grew to a higher density (Fig. 4A) and exhibited a higher growth rate (Fig. 4B) than the hmx1⌬ strain at all concentrations of hemin, indicating that Hmx1p was important for the use of heme as a nutritional source of iron.
The Role of Hmx1p in the Maintenance of Regulatory Pools of Cellular Iron-Although heme could serve as a nutritional source of iron for S. cerevisiae, this species typically relies on FIG. 2. Localization of Hmx1p to the endoplasmic reticulum and regulation of Hmx1p expression by iron. A-C, indirect immunofluorescence of Hmx1-HA. The Hmx1-HA strain was grown in ironpoor medium to induce the expression of Hmx1p. HA-11 was used as the primary antibody; Cy3-conjugated donkey anti-mouse was the secondary antibody. A, anti-HA staining; B, DAPI staining; C, merged image. D, detection of Hmx1p in membranes. The cells expressing Hmx1-HA were lysed with glass beads, and the membrane fraction (P) was separated from the cytosolic fraction (S) by centrifugation prior to Western blotting with HA-11. E, increased expression of Hmx1p in iron-poor medium. The Hmx1-HA and the untagged parent strains were grown in defined iron medium supplemented with the indicated concentrations of ferrous ammonium sulfate. Hmx1p was detected by Western blotting with HA-11. Molecular mass standards in kDa are indicated. (HO-1) and C. diphtheriae (HmuO). The darkly shaded letters indicate residues with sequence identity, and the lightly shaded letters indicate residues with sequence similarity. Diamonds, contact residues near the proximal heme-binding helix of HO-1; circles, contact residues in the distal heme-binding helix; triangles, the hydrophobic posterior wall of the heme pocket; squares, contact residues for cytochrome P-450 reductase; black bar, region of high sequence conservation within heme oxygenase family; white bar, putative transmembrane domain.

FIG. 1. Sequence alignment of Hmx1p with heme oxygenases from Homo sapiens
iron salts or chelates. Therefore, we investigated the role of Hmx1p in the uptake and maintenance of intracellular iron pools when yeast used iron salts or chelates as the nutritional source of iron. We also examined the capacity of the bacterial heme oxygenase, HmuO of C. diphtheriae, to functionally substitute for Hmx1p. We transformed the congenic HMX1ϩ and hmx1⌬ strains with pHmuO or an empty vector and labeled the intracellular iron pools by growth overnight in medium supplemented with 55 Fe-ferrichrome. Although all of the strains grew at similar rates, the hmx1⌬ strain exhibited a lower level of iron accumulation than the HMX1ϩ strain, and expression of HmuO in the hmx1⌬ strain partially restored iron accumulation (Fig. 5A). These results suggested that the heme degradation activity associated with Hmx1p contributed to the accumulation of intracellular iron. To test whether deletion of Hmx1p affected the regulatory pools of intracellular iron and, thus, the activation of Aft1p, we measured Aft1p activity by transforming the HMX1ϩ and the hmx1⌬ strains with a lacZ reporter construct under the control of the FET3 promoter (Fig.  5B). Aft1p-dependent transcriptional activity is low in cells grown in higher concentrations of iron, but Aft1p activity increases as cells are grown in lower concentrations of iron. Both the HMX1ϩ and the hmx1⌬ strains exhibited higher FET3-lacZ activities as the cells were grown in lower concentrations of iron. However, the hmx1⌬ strain consistently exhibited activity more than 2-fold higher than that of the HMX1ϩ strain, indicating that Hmx1p was involved in maintaining regulatory pools of intracellular iron. To confirm that the heme degradation activity of Hmx1p was required to maintain the iron pools, we measured FET3-lacZ activities in the HMX1ϩ and the hmx1⌬ strains transformed with pHmuO or the empty parent vector (Fig. 5C). Again, the hmx1⌬ strain exhibited much higher activity than the HMX1ϩ strain, and expression of HmuO in the hmx1⌬ strain lowered the FET3-lacZ activity to wild type levels, suggesting that the Hmx1p-mediated degradation of heme liberates iron that can be sensed by Aft1p.
Expansion of Intracellular Heme Pools in the hmx1⌬ Strain-If Hmx1p were involved in the degradation of intracellular heme pools, then the pool of heme subject to turnover would be larger in the hmx1⌬ strain than in the HMX1ϩ strain. We tested this hypothesis by labeling the HMX1ϩ and hmx1⌬ strains with 55 Fe-ferrichrome, allowing the 55 Fe to be incorporated into heme over a 16-h period of growth, and then measuring the amount of heme extracted from the cells (Fig.  6A). As predicted, the hmx1⌬ strain contained a larger amount of extractable heme than did the HMX1ϩ strain, and again expression of HmuO reduced the heme content of the hmx1⌬ strain to the level of the HMX1ϩ strain.
In addition to serving as an oxygen-binding prosthetic group, heme is an important regulatory molecule in S. cerevisiae. Heme activates the transcription factors Hap1p and Hap2/3/4/ 5p, which promote the transcription of a number of genes involved in respiration and aerobic growth (53). We tested whether deletion of Hmx1p affected regulatory pools of heme and the activity of the HAP transcription factors by measuring the activity of a HAP target, CYC1, fused to a lacZ reporter construct, in the HMX1ϩ and hmx1⌬ strains (Fig. 6B). In the HMX1ϩ strain, CYC1-lacZ activity was low in cells grown in iron-poor medium but increased as the concentration of iron increased. This result suggested that heme levels and hemedependent transcription increased as the availability of iron increased. In the hmx1⌬ strain, CYC1-lacZ activity also increased as the concentration of iron increased, but the level of activity was 2-3-fold higher than in the HMX1ϩ strain, indicating an increase in heme-dependent transcription in the hmx1⌬ strain. An exception was at the highest iron concentration, where CYC1-lacZ activity was similar in the HMX1ϩ and hmx1⌬ strains, a result likely caused by the very low level of Hmx1p expressed in wild type cells at this iron concentration (Fig. 2E). Again, expression of HmuO in the hmx1⌬ strain reduced CYC1-lacZ activity to wild type levels (Fig. 6C). Together, these data indicated that Hmx1p was required for the homeostasis of regulatory pools of intracellular heme. DISCUSSION Here we have presented data indicating that Hmx1p is involved in the degradation of heme under conditions of iron deprivation. Nevertheless, it remains unclear whether Hmx1p is a heme oxygenase enzyme. Previous investigators have reported that S. cerevisiae contains no heme oxygenase activity, that amino acid residues considered essential for heme binding are not conserved in Hmx1p, and that overexpressed, purified Hmx1p has no heme oxygenase activity (32). We report here, however, that membranes from yeast contained a heme degradation activity that was dependent on ascorbate and that this ascorbate-dependent activity was lost when HMX1 was deleted. Furthermore, disruptions of intracellular iron and heme homeostasis observed in the hmx1⌬ strain were corrected by expression of the prokaryotic heme oxygenase, HmuO. Previ- and fet3⌬ hmx1⌬ strains were inoculated at an optical density of 0.02 into defined iron medium supplemented with the indicated concentrations of hemin and grown for 16 h. The cell density was measured by determination of A 600 . B, cells from A were reinoculated at an optical density of 0.1 into medium containing 25 or 50 M hemin, and the growth rate was monitored by measurement of the A 600 . The experiment was replicated twice with the same result, and data from a single experiment are presented. ous attempts to measure Hmx1p enzymatic activity in yeast employed detergent-solubilized protein, which may have altered activity. Although Hmx1p lacks the conserved histidine that serves as the proximal heme ligand, Hmx1p binds heme, and, in HmuO, mutation of this histidine residue to alanine resulted in only partial loss of heme oxygenase activity (54). The heme oxygenase from Neisseria meningitidis displays relatively low sequence conservation with HO-1, and there are significant differences in their crystal structures; however, both enzymes catalyze the same reaction with the same regiospecificity (55,56). If Hmx1p were not a bona fide heme oxygenase, it could be acting as a heme chaperone, facilitating degradation by binding heme, and either presenting it to the actual heme oxygenase or making it available for nonenzymatic degradation.
In the transcriptional response to iron deprivation, S. cerevisiae expresses a number of genes involved in the mobilization of intracellular stores of iron, and we propose that Hmx1p functions in the reutilization of intracellular heme iron. Exogenous heme was more efficiently utilized as a nutritional source of iron in the strain that expressed Hmx1p than in the strain that did not. Deletion of HMX1 resulted in a higher level of Aft1p activation in iron-depleted cells, indicating that deletion led to greater iron deficiency. These data suggested that heme degradation increased when cells were grown under conditions of iron deprivation, that iron released from degraded heme was reused to meet metabolic needs, and that Hmx1p facilitated these processes. Deletion of HMX1 also resulted in a depletion of intracellular iron and an accumulation of intracellular heme in iron-replete cells. Heme may have a direct effect on the activity of Aft1p, 3 and therefore deletion of HMX1 could alter the "set point" for iron accumulation by raising heme levels.
In addition to the recycling of heme iron, Hmx1p may have an important role in lowering the regulatory pools of heme when cells are starved for iron. Two lines of evidence suggest that, under conditions of iron deficiency, respiration is impaired. First, strains carrying a deletion of AFT1 or inactivation of the high affinity ferrous transport complex grow poorly on medium containing nonfermentable carbon sources. Second, wild type strains grow poorly on iron-poor medium containing nonfermentable carbon sources. 4 Respiration requires the expression of iron-rich respiratory complexes involved in mitochondrial oxidative phosphorylation, and the cell may gain a survival advantage by diverting this iron to other processes during periods of iron deprivation. Hmx1p may facilitate this metabolic shift by degrading heme and thereby lowering the activity of the heme-dependent transcriptional activators, Hap1p and Hap2/3/4/5p, which activate the genes involved in aerobic growth, including the respiratory cytochromes (53). 3  Fe-ferrichrome. The total 55 Fe content of the cells was determined as described. B, congenic HMX1ϩ and hmx1⌬ strains were transformed with pFET3-lacZ and grown in defined iron medium containing 30 -200 M iron prior to determination of ␤-galactosidase activity. C, congenic HMX1ϩ and hmx1⌬ strains with an integrated copy of pHmuO or the empty parent vector were transformed with pFET3-lacZ and grown overnight in defined iron medium containing 30 M iron prior to determination of ␤-galactosidase activity. The experiments were replicated three times, and data from a representative experiment are shown. The error bars indicate the standard deviations.
FIG. 6. Increased heme content and increased Hap1p activity in the hmx1⌬ strain and complementation by HmuO. A, congenic HMX1ϩ and hmx1⌬ strains with an integrated copy of pHmuO or the empty parent vector were grown as described in the legend to Fig. 5A. The 55 Fe-heme content of the cells was determined as described. B, congenic HMX1ϩ and hmx1⌬ strains were transformed with pCYC1-lacZ and grown in defined iron medium containing 30 -200 M iron prior to determination of ␤-galactosidase activity. C, congenic HMX1ϩ and hmx1⌬ strains with an integrated copy of pHmuO or the empty parent vector were transformed with pCYC1-lacZ and grown overnight in defined iron medium containing 30 M iron prior to determination of ␤-galactosidase activity. The experiments were replicated three times, and data from a representative experiment are shown. The error bars indicate the standard deviations.
Thus, Hmx1p serves the dual purposes of recycling iron and remodeling transcription during periods of limited iron availability.