INTRODUCTION

A great deal remains to be learned about the biochemical and physiological functions of heme oxygenase (HO)¹ isozymes, HO-1 and HO-2 (1, 2), which have been traditionally viewed only in terms of heme catabolism. And some confusion persists about why there are two forms of an enzyme that, by all appearances, have the same catalytic activity and substrate specificity. More puzzling is the high level expression of the second isozyme, HO-2, in tissues and cells that have no role in hemoglobin heme turnover (3-6). The aim of this study was to learn more about HO-2.

HO-1 and HO-2 catalyze the conversion of heme (Fe-protoporphyrin IX) to biliverdin and CO and release Fe in a reaction that utilizes 3 mol each of oxygen and NADPH (7). All products of HO activity are suspected to be physiologically active (4-6). The constitutive form, HO-2, and the inducible form, HO-1, are different gene products (8-10). Except for catalyzing heme and sharing a stretch of amino acids known as the "HO signature" (GenBank^TM), they have little resemblance to one another. The "HO signature" is part of a 24-residue domain, which forms the heme catalytic pocket and that, except for one residue, is conserved (11) among all HO-1s and HO-2s characterized to date. The HO signature motif has a conserved histidine residue, His-151 in HO-2, that is essential for its activity (12). In HO-1 the conserved histidine stabilizes a distal water ligand, and based on experimental findings (13) it is placed in a position close to the ligand binding site and plays a role in oxygen binding/activation in the distal heme pocket. At the primary amino acid level, the similarity between rat HO-1 and HO-2 is a mere 43% (14, 15). HO-1, also known as HSP32, is responsive to an extensive array of chemical agents and stimuli (reviewed in Ref. 16), while HO-2 is constitutively expressed in all cell types, and the only inducers of the enzyme identified to date are the adrenal glucocorticoids (10, 17). Glucocorticoids cause increased association of the protein with the nuclear envelope as visualized by immunostaining (18). HO-2 is a single copy gene with multiple transcripts (3, 19, 20) ranging in size between ~1.3 and 2.1 kilobase pairs that differ in use of three different 5-untranslated regions and two poly(A) signals (19, 20). Between the two poly(A) signals a consensus sequence of 5-TTTTTGCA-3 is found, which is 100% identical to the oxygen/nitrogen-sensing sequence (21-23) and is found in the erythroprotein gene (21, 22).

Aside from the overall differences in amino acid composition of HO-1 and HO-2, a major difference is the presence of two cysteines in all HO-2s and the absence of this residue in all HO-1s (11, 14, 15, 19, 24-28). This residue is the axial ligand for the heme prosthetic moiety in various hemoproteins, including all cytochrome P450s and nitric-oxide synthase isozymes (29-33). In HO-2 this residue is flanked downstream by a proline residue followed by phenylalanine (Cys-Pro-Phe); two copies of such arrangements are present in the predicted sequence of the protein. The Cys-Pro dipeptide, often flanked downstream by a hydrophobic residue, phenylalanine, is the absolutely conserved core of the recently identified motif called the heme regulatory element (HRM) (34). There is a tendency for a positively charged residue (arginine or lysine) to flank the core upstream. The HO-2 core and the surrounding residues are as follows: Val²⁶¹-Arg-Lys-Cys-Pro-Phe-Tyr-Ala-Ala-Gln and Gly²⁷⁸-Ser-Asp-Cys-Pro-Phe-Arg-Thr-Ala-Met.

In the second (Cys²⁸¹-Pro) dipeptide, the upstream residues (Gly-Ser-Asp) are polar; polar residues (glycine and serine) also flank two copies of heme lyase HRM (34). HO-2 is among only six proteins identified to have this core motif and the characteristic flanking residues (34). The others are Saccharomyces cerevisiae heme lyase (35), human erythroid  aminolevulinate synthase (36), Escherichia coli catalase (37), rabbit heme-regulated initiator factor  kinase (38), and S. cerevisiae HAP1, a transcriptional activator that responds to oxygen/heme (34, 39, 40). The HRMs bind heme and confer regulation by heme to proteins; in fact, only one copy of HRM is adequate for such activity (34).

Given the fact that in HO-2, two copies of the core HRM are present, we undertook the present study to examine whether, in intact HO-2 protein, the HRMs are involved in heme binding and to investigate their function in heme catalysis. In the course of this investigation we have found that HO-2 is a hemoprotein and provide good evidence to suggest that HRMs are involved in binding of heme but not in heme catalysis. Furthermore, we confirm that the His¹⁵¹ in the 24-residue heme pocket is essential for HO-2 activity.

Experimental Procedures

Oligonucleotides for sequencing and mutagenesis were obtained from Midland Certified Reagent Co. (Midland, TX) or Life Technologies, Inc. Two oligonucleotides 10 residues long encompassing the HRMs of HO-2 (Val²⁶¹-Arg-Lys-Cys-Pro-Phe-Tyr-Ala-Ala-Gln) and (Gly²⁷⁸-Ser-Asn-Cys-Pro-Phe-Arg-Thr-Ala-Met) were synthesized and purified by high pressure liquid chromatography (95%) by Primm Laboratories (Cambridge, MA). These peptides herein are referred to as the Cys²⁶⁴ and Cys²⁸¹ peptides, respectively. Nitrocellulose for Western blot analysis and Nytran for Northern hybridization were from Schleicher and Schuell (Keene, NH). Sequenase, version 2.0, random primer labeling system, restriction enzymes, and other DNA modification enzymes were purchased from U.S. Biochemical Corp. [-³²P]dCTP and [-³⁵S] dATP were obtained from U.S. Biochemical Corp. or DuPont NEN. Reagents for protein determination were obtained from Bio-Rad. All other reagents were of the highest quality commercially available. Adult male Harlan Sprague Dawley rats and New Zealand rabbits were obtained from Harlan Industries (Madison, WI). Rat biliverdin reductase was purified essentially by the method described previously (41). NADPH-cytochrome P450 reductase was purified as described by Yasokuchi and Masters (42).

E. coli Inv  F (F end A1 rec A1 hsd R17 (r_k, m_k⁺) sup E44 thi-1 gyrA96 relA1 80 lac ZM15 (lac ZYA-arg F) U169 ) carrying HO-2 plasmids were grown to saturation overnight at 37 °C in 2 × YT medium containing 50 µg/ml ampicillin. Cultures were diluted 1:100 in the same medium and grown to an A₆₀₀ of ~1.0. One milliliter was removed from the culture and prepared for SDS-polyacrylamide gel electrophoresis as previously detailed (19). To assay HO-2 expression, 30-µl aliquots of bacterial lysates were examined on a 12.5% polyacrylamide gel. The gel was electroblotted onto nitrocellulose and probed with HO-2 polyclonal antibody as described previously (8). Cell lysates were prepared essentially by the method of Scopes (43) in a buffer containing 20 mM Tris-HCl, pH 7.5, 1 mM EDTA, 20% (v/v) glycerol, and 0.4% (v/v) Triton X-100. Total extracted protein in bacterial lysates was quantitated by the method of Bradford (44) using bovine serum albumin as the standard. HO activity was measured as previously detailed (15) in the presence of added purified rat liver biliverdin reductase and NADPH-cytochrome P450 reductase. One unit of activity was defined as producing 1 nmol of bilirubin/h.

Plasmid DNA from the rat HO-2 expression clone, pRHOP (12, 15) was utilized as the substrate to carry out site-directed mutagenesis of cysteine residues using the mutagenic primers indicated below as detailed previously (12), and the products were transformed into E. coli XL-1 Blue (recA1, lac(F, proAB, lacI^qZM15, Tn10(tetR))) cells. The HO-2 mutagenesis primers were 5-AGCATAAAAGGGGGCTTTACGTACATC-3, complimentary to nucleotides 778-804 (15) with a GC mismatch (boldface type) for CA to convert Cys²⁶⁴ into alanine; 5-CCGGAAGGGGGCGTTGCTGCCTCC-3, complimentary to nucleotides 829-852 also with a GC for CA mismatch to convert Cys²⁸¹ into alanine; and 5-CGAGTGTAAGCCGCGGCCACCAG-3, complementary to nucleotides 442-464 with mismatches to convert His¹⁵¹ to alanine. Transformants were screened initially for the loss of the unique NdeI site and the gain of a single NcoI site and were subsequently sequenced by the method of Chen and Seeburg (45) to identify mutants. DNA from mutant clones and the parental plasmid were separately transformed into InvF cells (Invitrogen, San Diego, CA) which were assayed for immunoreactive protein and heme oxygenase activity as described above. The double mutant Cys²⁶⁴/Cys²⁸¹  Ala/Ala, referred to as "HO-2 mut," was generated from the Cys²⁶⁴  Ala mutant using the Cys²⁸¹  Ala mutagenesis primer. The same method was also used to generate Pro²⁶⁵ and Pro²⁸²  Ala mutants using oligonucleotide primers complimentary to nucleotides 778-810 (5-CTGAGCAGCATAAAATGCGCATTTACGTACATC-3) and nucleotides 832-850 (5-GGCTGTCCGGAATGCGCAGTTGCTGGC-3), respectively (mismatched nucleotides are in boldface type). Both mutagenic primers introduce an FspI site, and transformants were initially screened for the presence of the restriction site prior to sequencing.

The wild-type and mutant HO-2 constructs were subjected to an additional round of polymerase chain reaction-mediated mutagenesis. DNA from the carboxyl-terminal substitution mutant, pRHOP (12), and the double Cys  Ala mutant above were utilized as template for the polymerase chain reaction utilizing the primers 5-CGCAATTAACCCTCACTAAAG-3, which represents the sequence upstream of the polylinker of the vector, pBS+, and 5-GTCGACTAATGATGATGATGATGATGCTTATCGTCATCGTCCTGCAGGCTAGGCTTCCTG-3, which represents the reverse complement of HO-2 nucleotides 870-888 plus a histidine "tag" of six codons (boldface type) and an enterokinase cleavage site (underlined) to facilitate removal of the tag from the fusion protein. The primer also introduces a stop codon (double underline) and a SalI restriction site (italics) following the histidine tag (5 to it in the primer). Polymerase chain reaction products were cloned into the vector pCRII (Invitrogen Corporation, San Diego, CA). Transformants were screened for the presence of a 934-base pair SalI fragment representing the tagged HO-2 coding region. The inserts were sequenced to confirm their identity and subcloned into the SalI site of pBS+ (Stratagene, La Jolla, CA), and the resultant plasmids were transformed into Inv  F. The orientation of the insert was determined by restriction analysis and confirmed by sequencing.

A similar construct was generated for HO-1. In this case first strand cDNA was generated from 1 µg of testis poly(A) RNA using the cDNA cycle kit (Invitrogen) priming with oligo(dT) and was used as a template for polymerase chain reaction using the primers 5-GGGAAGCTTGGAGCGCCCACAGCTCG 3, representing nucleotides 2-18 of HO-1 (14) and a HindIII restriction site (italics) and 5-AAGCTTATGATGATGATGATGATGCTTATCGTCATCGTCCATGGCATAAATTCCCACTG-3, which contains the reverse complements of nucleotides 848867, an enterokinase cleavage site (underlined), histidine tag (boldface type), stop codon (double underline), and HindI III site (italics). The fragment was cloned into the HindIII site of the pBS+.

Fusion proteins were purified from overnight bacterial cultures utilizing ProBond^TM (Invitrogen) columns in accordance with the manufacturer's instructions. Pooled peak fractions from the ProBond^TM column elution were buffer-exchanged into 50 mM Tris-HCl, pH 8.0, containing 10% glycerol, 1 mM CaCl₂, and 0.1% Tween 20 and concentrated to approximately 1 mg/ml. When necessary, the histidine tag was removed using enterokinase (EKMax^TM, Invitrogen) utilizing 1 unit/mg protein and digesting for 16 h at 14 °C. Removal of enterokinase was accomplished using soybean trypsin inhibitor-agarose (Sigma) as described by the manufacturer. Eluted proteins were then buffer-exchanged into a buffer appropriate to the final application (20 mM Tris-HCl, pH 7.5, containing 1 mM EDTA and 20% glycerol for storage at 80 °C) using Centricon 10 microconcentrators. Preparations were judged to be >90% homogeneous as assessed by SDS-polyacrylamide gel electrophoresis followed by staining with Coomassie Brilliant Blue. The SDS gel profile of final preparations is shown in Fig. 1. (The lower molecular weight band in HO-2 mut is often observed in SDS gels of HOs and is due to the cleavage of the proteins.)

The hemoprotein nature of HO-2 was established by examining spectral properties of the purified protein preparations over the range of 350-650 nm at 2 nm/s. The reference was 0.1 M Tris-HCl, pH 7.5, with 0.01% Tween 20, while the test cuvette contained purified protein in the same buffer at the concentrations indicated in the figure legends. Heme was quantitated by a pyridine hemochromogen assay (46). The change in OD between 557 and 575 nm was used to determine the heme concentration using an extinction coefficient of 32.4 mM¹ cm¹. For heme binding studies, heme was prepared fresh by dissolving in a 1:1 (v/v) mixture of 1 M NH₄OH/methanol, and volume was adjusted by the addition of 0.1 M Tris-HCl (pH 7.5) containing 0.01% Tween 20. Heme binding was determined by absolute absorption spectroscopy using buffer solution as the reference. Reconstitution of HO-2 with heme was carried out by incubating purified HO-2 with a 5-fold molar excess concentration of heme. After incubation at 4 °C for 1 h, excess heme was removed by chromatography through a G25 column. 0.5-ml fractions were collected, and heme and protein concentrations were measured in each fraction.

The heme binding of the purified wild-type HO-2 and HO-2 mut proteins was also examined by UV fluorescence quenching (47, 48), and for comparison that of rat HO-1 was also examined. The fluorescence of a solution of protein (0.5 µM) in 0.1 M Tris-HCl, pH 7.5, was measured using 280 nm as the excitation wavelength and scanning emissions from 300 to 450 nm. The emission spectrum had a maximum at ~330 nm. Subsequently, incremental additions of heme were made to the cuvette, and the fluorescence was measured following each addition.

A full-length (1300-base pair) HO-2 cDNA isolated from a rat testis DNA library (15) was used as HO-2 hybridization probe. Mouse -actin and HO-2 cDNA probes were labeled with [³²P]dCTP according to the manufacturer's instructions, using the random primer DNA labeling system, and further purified by spin column chromatography.

ptk cells were maintained at 37 °C under 5% CO₂ in Dulbecco's modified Eagle's medium containing 10% bovine calf serum supplemented with 4.1 mM L-glutamine and 100 units/ml penicillin and 100 µg/ml streptomycin sulfate. For RNA analysis, cells were grown to 60-70% confluence. The medium was replaced with serum-free medium supplemented with GMS-X (Life Technologies, Inc.) for 2.5 h and then subsequently replaced with fresh serum-free medium or the same medium containing 25 µM heme. After 1 or 4 h, total RNA was prepared from control or heme-treated cells. Poly(A) RNA was isolated by oligo(dT)-cellulose chromatography, fractionated on a 1.2% (w/v) agarose gel, and transferred to Nytran. Prehybridization, hybridization of the appropriate ³²P-labeled cDNA, and posthybridization treatment of the blots were performed essentially as described earlier (3).

RESULTS

The involvement of HRMs and HO signature domains of HO-2 in heme catalysis and binding was examined using wild-type and mutant HO-2 proteins. The oligonucleotide primers indicated under "Experimental Procedures" were utilized to substitute alanine for Cys²⁶⁴ and Cys²⁸¹ of the HRM sequences or His¹⁵¹ in the conserved 24-residue domain in bacterial plasmids, and HO-2 was expressed as a LacZ fusion protein in E. coli. Mutations were confirmed by sequence analysis, and expressed proteins were analyzed for HO activity; data are presented in Fig. 2. As shown in panel a, the bacterial strain expressing the His¹⁵¹ mutant does not have detectable activity, confirming the previous report (12). Neither the mutation of Cys²⁶⁴ nor that of Cys²⁸¹, however, caused a notable effect on activity. Also, substitution of alanine for Pro²⁶⁵ or Pro²⁸² had no discernible effect on enzyme activity, and essentially the same results were obtained as with Cys²⁶⁴  Ala and Cys²⁸¹  Ala mutants (data not shown). To address the possibility that only a single copy of the HRM sequence is required for activity, a construct was generated combining both cysteine mutations. As is shown in the figure, the double mutant also did not display a considerable difference in activity from either of the single mutant or the wild-type constructs. To investigate whether the absence of activity in the His¹⁵¹ mutant was due to decreased expression of the mutated protein as well as whether there was an overexpression of Cys  Ala mutants, Western blot analysis of the same E. coli expression cultures used for activity analysis was carried out using equal amounts of bacterial cell lysate. Fig. 2b shows that the lack of a detectable activity of His¹⁵¹  Ala clearly was not due to the absence of the expressed protein (lane 3). Indeed, we consistently observe a higher expression of His¹⁵¹  Ala mutant than of the wild-type protein (lane 2). Another important observation is that an overexpression of Cys mutants, particularly the Cys double mutant (lane 6) was not the reason for the unaffected heme-degrading activity of the expressed protein.

Heme oxygenase activity and HO-2 immunoreactive protein in E. coli expressing wild-type or different Cys  Ala mutants as a fusion protein.

The finding that HRMs are not involved in heme degradation encouraged us to question whether HO-2 is a hemoprotein. The inserts from the wild-type and the Cys^264/281  Ala double mutant plasmids were utilized to generate plasmids expressing the same proteins with a histidine tag (His₆), which could be removed by enterokinase digestion, at their carboxyl termini. Wild-type HO-2 and HO-2 mut (double cysteine mutant) were expressed in E. coli and purified. The purified proteins, which were >90% homogenous, as assessed by SDS-polyacrylamide gel electrophoresis (see "Experimental Procedures"), were used to assess heme content and for spectral analysis; results are presented in Fig. 3. The purified wild-type protein has an intense Soret band at 406 nm; upon reduction with dithionite, the maximum shifts to 424 nm, and the peak is at 421 nm for the ferrous CO complex. The absorption in the visible region (500-700 nm) of the ferrous heme (inset a) shows a 632-nm absorption band. The 630-nm band is typical of high spin hemoproteins. The values obtained for absorbance maxima for the oxidized form at 406 nm are close to that of hemoglobin (403-406), which, at neutral pH, exists predominately as a six-coordinate form with water at the sixth position. In contrast, the double mutant, in which the Cys residues of both HRMs have been replaced by alanine, does not have a discernible heme spectrum, suggesting that, while wild-type HO-2 is a hemoprotein, this property is dependent on the presence of HRMs. The hemoprotein nature of HO-2 was confirmed using the pyridine hemochromogen assay (Fig. 3, inset b). While the HO-2 mut had negligible levels of heme, the wild-type protein contained 0.66 nmol of heme/mg of protein or an approximately 1:50 molar ratio of heme:HO-2. Clearly the ratio of heme to protein is low, but it is important to remember that the bacterial system is expressing the enzyme that degrades heme synthesized by the bacteria and that the expression system was designed to express HO-2, not heme biosynthesis enzymes. In addition, the purification procedure required prolonged exposure to an acidic pH of 6.0 and 0.5 M imidazole; at acidic pH the propionate side chains of heme are protonated.

Next, we determined if the purified protein could bind additional heme. For this, heme in molar excess was added to the purified HO-2 protein (5:1). The solution was incubated 1 h at 4 °C with gentle mixing, and then the nonspecific heme bound and free heme were removed by gel filtration chromatography. The molar ratio of protein to heme (pyridine hemochromogen) in the fractions was determined, and a value of 1:3 was obtained. In a previous study (49), using a conventional method of protein purification (all steps carried out at pH 7.5) and utilizing spectrophotometric titration (50) when plotting increments of the increase in absorbance at the Soret band against the molar ratio of exogenously added heme, the increase in absorbance reached a maximum at a ratio of 1:1. The difference between the present results and the previous one likely reflects the duration of interaction between heme and HO-2 protein, which was immediate in the previous study as well as the purification procedures; the association of heme with apoprotein can be time-dependent (51). Also, at that time we were unaware that the enzyme is a constitutive hemoprotein.

To further test heme binding to HRMs, the following experiments were conducted. In the first series, the visible spectrum of wild-type HO-2-heme and HO-2 mut-heme complexes using a fixed protein concentration and increasing heme concentrations was examined (Fig. 4). As noted, at a protein:heme ratio up to 1:1 a shift of the Soret band of heme from 389 nm (inset) to 406 nm was recorded for both the wild-type HO-2 and the HO-2 mutant. At a 1:1 ratio, the wild-type HO-2-heme complex had an extinction coefficient of 85 mM¹ cm¹, whereas the HO-2 mut-heme complex had an extinction coefficient of 65 mM¹ cm¹ for the Soret band. As seen, up to a 1:2 protein:heme ratio, the absorbance for the wild-type HO-2-heme complex was consistently higher than the HO-2 mut. Above 1:1 there is a blue shift for the HO-2 mut-heme complex as the heme to protein ratio increases, contrasting with the absence of such a shift for the wild-type protein-heme protein complex up to a 1:3 ratio. The Soret band of the wild-type protein-heme complex shifts toward blue when heme is present at or above a 4-fold molar excess. We infer from these observations that the HRMs present in the wild-type protein bind heme and are responsible for the difference in spectral behavior of the two heme-protein complexes at higher concentrations. Thus, the addition of heme at concentrations exceeding the potential specific heme binding sites of the proteins appears to result in a shift toward the 389-nm maximum of heme and may be due to nonspecific interactions. From the differences in the amount of heme required to cause this shift, the presence of two additional binding sites in the wild-type protein compared with the mutant may be inferred.

Additional evidence suggesting binding of heme to HRMs is provided by analysis of UV fluorescence quenching (Fig. 5) and the visible hemochrome (Fig. 6). Although fluorescence of the wild-type HO-2 is not completely quenched until a 4-fold molar excess of heme is added, only a 2-fold molar excess completely quenches fluorescence of the HO-2 mut. For comparison, the fluorescence quenching of HO-1, which binds heme with 1:1 stoichiometry (50) is included. Further evidence for the interaction of HRMs with heme was obtained from a visible hemochromogen. Hemochromogens have a different color than heme, which is visible to the naked eye. The results when 10-residue Cys²⁶⁴ or Cys²⁸¹ peptides in water were added at a 2:1 molar ratio to a solution of heme are shown in Fig. 6. As noted, in the presence of Cys²⁶⁴, the brown color of the heme solution shows a dramatic change in color and becomes red-brown; the color change in the presence of Cys²⁸¹ is more subdued. A visible change in color was also observed at a 1:1 molar ratio.

A possible consequence of heme binding by HO-2 could be autoregulation of HO-2 transcription. To address this possibility, ptk cells were examined by Northern hybridization following exposure to 25 µM heme. As shown in Fig. 7a, there was no detectable change in the level of the single ~1.3-kilobase pair HO-2 homologous transcript in this cell line 1 h after the addition of heme (lane 1) compared with untreated cells (lane 2). Equal loading was confirmed by probing the same blot with an actin probe (panel b). The same result was also obtained when RNA was prepared 4 h after exposure to heme (data not shown).

DISCUSSION

Presently we report that HO-2 is a constitutive hemoprotein, and, as indicated by site-directed mutagenesis experiments, the hemoprotein nature appears linked to the HRMs of the protein; these motifs are not involved in the catalytic activity of the protein. At this time, we cannot assuredly predict whether, in the natural state of the enzyme, each HRM binds one heme or a single heme forms a "bridge" between the two HRMs either by coordination or electron interactions or, for that matter, whether HO-2 is associated with heme under all conditions. Such analyses await more detailed structural characterization of the HO-2-heme complex. Nonetheless, theoretically speaking, if each HRM were to bind one heme molecule, then, when fully occupied, HO-2 protein would have three bound heme molecules, the third site being the 24-residue conserved "heme pocket" domain. The findings of the heme binding/heme reconstitution analysis experiments showing that 3 mol of heme appear to specifically bind to 1 mol of protein would be consistent with the possibility of each HRM binding one heme. Indeed, the involvement of HRMs in heme binding is supported by UV quenching analysis (Fig. 5) in that the HO-2 mutant profile closely resembles that of HO-1, which has a single binding site, while that of wild-type HO-2 is multiphasic, suggesting multiple sites with differential affinity for heme. Because of this apparent differential affinity of the sites for heme, prediction of the number of binding sites based on the initial slope of the titration curve (52, 53) is not possible. Interaction with the heme would be anticipated to take place with the catalytic site. This is suggested by the fact that heme quenches HO-1. HO-2 and HO-1 have a conserved tryptophan upstream of the conserved domain, Trp¹⁰¹ in HO-1 and Trp¹²⁰ in HO-2. There are also two tyrosines in the conserved domain, Tyr¹⁵³ and Tyr¹⁵⁶ in HO-2 and Tyr¹³⁴ and Tyr¹³⁶ in HO-1. As has been reported for various hemoproteins, interaction of the side chains of nearby aromatic amino acids with heme can influence protein fluorescence (52-54).

While the HRMs do not appear to have a role in catalysis of heme by HO-2, revealed by the observation that Cys²⁶⁴  Ala, Cys²⁸¹  Ala, Pro²⁶⁵  Ala, Pro²⁸²  Ala, or double Cys mutations have little effect on heme oxidation, the single histidine (His¹⁵¹) in the conserved domain of HOs is clearly indispensable for HO-2 activity. As noted earlier, HO-2 shares this domain of 24 amino acids with HO-1, which also has no HRMs and is not known to be a hemoprotein (50, 55); therefore, it is not surprising that HO-2 mut functions effectively as a catalytic enzyme but does not display hemoprotein characteristics. His¹⁵¹ of HO-2 is equivalent to His 132 of HO-1. In HO-1 mutation of His¹³² results in up to 80% loss of activity and is essential for H₂O₂-supported oxidation of heme by the enzyme (13). As suggested by studies with HO-1 mutagenesis, other histidines also seem to be important for heme catalysis activity; for instance, His²⁵, which corresponds to His⁴⁴ in HO-2, is the proximal heme iron ligand in HO-1 (56, 57).

It seems reasonable to suspect that HO-2, like the other five HRM-containing proteins (34-40), has a regulatory role, linked to heme binding, in the cell. HO-2 is unique among HRM-containing proteins by having what appears to be two distinctly different kinds of relation with the heme molecule; it binds heme at the HRMs, without degrading the metalloporphyrin, and it interacts with heme at the "heme pocket" for catalysis. In fact, HO-2 is the only protein described to date for which such dual function can be ascribed. The possibility that two different sites of HO-2 may have different functions is not, however, unprecedented; for example, the FixL protein of Rhizobium meliloti has a separate heme-binding/oxygen-sensing domain and a functional kinase domain (58).

The regulatory function of HO-2 can be envisioned to ultimately result in controlling cellular heme concentration, with HO-2 functioning as a "heme sensor." In a capacity as a heme sensor, HO-2 could also be visualized to serve as a transcriptional regulator by fine tuning the activity of transcriptional factors and genes that are heme-responsive, including HO-1 (7, 59). One may reason that in such a capacity, heme bound by the HO-2 HRMs would be protected from degradation and that when the concentration of heme exceeds the binding capacity of HRMs then excess heme would become available to serve as substrate for catalysis through interactions with the "catalytic pocket." As such, HO-2 could control expression of genes that are responsive to heme, although the enzyme itself is not regulated by heme (Fig. 7). The unresponsiveness of HO-2 gene regulation to heme would be an important aspect of its regulatory activity. Another heme-related regulatory function of HO-2 could involve the ability of heme to activate molecular oxygen and form reactive oxygen radicals; catalyzing oxygen radical formation is an inherent activity of the heme molecule that is greatly potentiated when it is bound to proteins (60). If HO-2 were involved in oxygen radical generation, then, aside from regulatory mechanisms for control of stress protein gene expression, another system that likely could utilize the reactive oxygen-generating function of HO-2 would be the male reproductive system, specifically the function of the sperm cells. These cells depend on hydroxyl radicals for function and are also damaged by excess levels of the radicals (61). Sperms, as well as their progenitor cells, have high levels of HO-2 (5), and the protein is present in the mature sperm acrosomes and the flagella. These segments of the sperm are essential for its capacitation. And capacitation of sperm is mediated by hydroxyl radicals (61). Given the fact that mature sperm neither have detectable biliverdin reductase nor NADPH-cytochrome P450 reductase, then generation of oxygen radicals would appear to be a plausible function of HO-2.