Molecular identification of adrenal inner zone antigen as a heme-binding protein


M. Okamoto, Department of Biochemistry and Molecular Biology, Graduate School of Medicine (H-1), Osaka University, 2-2 Yamadaoka, Suita, Osaka 565-0871, Japan Fax: +81 6 6879 3289 Tel: +81 6 6879 3280 E-mail:


The adrenal inner zone antigen (IZA), which reacts specifically with a monoclonal antibody raised against the fasciculata and reticularis zones of the rat adrenal, was previously found to be identical with a protein variously named 25-Dx and membrane-associated progesterone receptor. IZA was purified as a glutathione S-transferase-fused or His6-fused protein, and its molecular properties were studied. The UV-visible absorption and EPR spectra of the purified protein showed that IZA bound a heme chromophore in high-spin type. Analysis of the heme indicated that it is of the b type. Site-directed mutagenesis studies were performed to identify the amino-acid residues that bind the heme to the protein. The results suggest that two Tyr residues, Tyr107 and Tyr113, and a peptide stretch, D99–K102, were important for anchoring the heme into a hydrophobic pocket. The effect of IZA on the steroid 21-hydroxylation reaction was investigated in COS-7 cell expression systems. The results suggest that the coexistence of IZA with CYP21 enhances 21-hydroxylase activity.


inner zone antigen


glutathione S-transferase


membrane-associated progesterone receptor

Distinguished histologically, the three zones in the mammalian adrenal cortex have distinct functions. In man, the outermost zona glomerulosa secretes aldosterone, the intermediate zona fasciculata, cortisol, and the innermost zona reticularis is the main site for dehydroepiandrosterone formation, whereas in the rat, corticosterone is the main product of the fasciculata and reticularis, with little if any dehydroepiandrosterone. The molecular mechanisms underlying the functional differentiation of the three zones have been a focus of numerous investigations [1–3]. To facilitate the study of zonal function, Laird et al. [4] produced a monoclonal antibody that recognizes an antigen, named inner zone antigen (IZA), which is present in the zonae fasciculata/reticularis in the rat, but not in the zona glomerulosa. Here we call this antigen, which was originally identified in rat tissue, ‘rIZA1’. The monoclonal antibody was capable of inhibiting dose-dependently adrenal 21-hydroxylation of progesterone and 18-hydroxylation of 11-deoxycorticosterone. When rat adrenal homogenates were subjected to SDS/PAGE followed by immunoblot analysis, two proteins of molecular mass 27–28 kDa and 55–60 kDa reacted with the monoclonal antibody [5]. The larger protein was thought to be a dimer of the smaller protein. rIZA1 appeared to be distributed not only in the adrenal cortex but also in other tissues [6,7].

Using the monoclonal antibody immobilized to Sepharose beads, Raza et al. [8] successfully purified rIZA1 and determined its N-terminal amino-acid sequence. The sequence was found to be consistent with that of a protein reported previously as ‘25-Dx’ (GenBank accession number U63315) [9] or ‘membrane-associated progesterone receptor (MPR)’ (GenBank accession number AJ005837) [10]. In the human genome sequence, two genes encode IZA; one, Hpr6.6 (accession number NM_006667), encodes a protein corresponding to rIZA1, which we name here hIZA1, and the other, Dg6 (accession number NM_006320), encodes a protein similar to, although distinctly different from, the protein named hIZA2 here [11]. Complementary DNA encoding 25-Dx was isolated as one of the dioxin-inducible genes in rat liver [9], whereas MPR had been purified from porcine liver [12] and its cDNA from porcine vascular smooth muscles [10]. The cloned sequence of rIZA1 was identical with that of MPR, although somewhat different from that of 25-Dx at the 3′-terminal. It is possible that a splicing error occurred during the preparation of 25-Dx cDNA. rIZA1 has also been reported as ‘ventral midline antigen’, a protein expressed in the rat central nervous system [13]. A yeast ortholog of IZA was recently reported as ‘Damage response protein related to membrane-associated progesterone receptors I protein’ (Dap1p) [14]. As this brief review indicates, IZA has been studied by many investigators from a variety of viewpoints and a variety of biological functions have been attributed to it. However, its precise physiological role is still unclear. To investigate this point further, IZA was purified to homogeneity to examine its molecular nature. Our preliminary results suggest that IZA contains a heme chromophore [15]. Mallory et al. [16] also reported recently that Dap1p, the yeast homolog of IZA, is a heme-binding protein. Here we report our further characterization of human and rat IZAs.

Results and Discussion

The domain structure of IZA was explored by inputting its amino-acid sequence into a protein domain structure prediction program in the website, The results illustrated in Fig. 1A suggested that IZA contains a heme/ steroid-binding domain similar to a heme-binding domain of cytochrome b5. The 134-amino-acid protein human cytochrome b5 has a heme-binding domain of ≈ 80 residues near its N-terminus in which His44 and His68 act as the sixth-axial and fifth-axial ligands for the heme iron, respectively. The transmembrane region of ≈ 20 amino acids is located near the C-terminus. Conversely, hIZA1, a protein of 195 amino acids, has a transmembrane region at its N-terminal side, and the predicted heme/steroid-binding domain is located in the central portion. The aligned amino-acid sequences of hIZA1, rIZA1, and hIZA2 are shown in Fig. 1B, in which the transmembrane regions and the heme/steroid-binding domains are highlighted in yellow and red, respectively. Amino acids in the heme/steroid-binding domain are well conserved among the three proteins (shown in bold letters). This strongly suggests that this domain plays an important role in the physiological function of IZA. The amino-acid sequence of the heme-binding domain of cytochrome b5 (shown in blue) was aligned with those of the heme/steroid-binding domain of IZA. Surprisingly, the similarity between IZA and cytochrome b5 was rather weak, and only 15 out of 82 residues are identical (highlighted in green; the residues covered with dark green shade are identical residues, whereas those with light green are similar). It should be noted that hIZA1 contains only three His residues, all located outside the heme/steroid-binding domain: one, His23, near the N-terminus and the others, His165 and His166, near the C-terminus (the numbering is that of hIZA1). hIZA2 has only one His near the C-terminus. These findings raised the question whether the heme/steroid-binding domain of IZA actually functions as a specific heme-binding site. We therefore purified IZA and examined its molecular properties.

Figure 1.

Primary structure of IZA reveals a heme/steroid-binding domain. (A) Outline of primary structures of IZA and cytochrome b5. The predicted heme/steroid-binding domain and transmembrane domain are illustrated. (B) Alignment of entire sequences of hIZA1, hIZA2 and rIZA1. A sequence of the heme-binding region of cytochrome b5 was also aligned (blue). Asterisks indicate amino-acid residues that were mutated in this study. (C) Prediction of the 3D structure of hIZA1. The structure of bovine cytochrome b5 (1CYO) is shown in the left panel [34,35]. The numbers of the two axial ligand His residues indicate those predicted from cDNA (M63326). The model of hIZA1 was illustrated by using the loopp program ( As a template for the modeling, the most similar protein, a cytochrome b5 homolog of Ectothiorhodospira vacuolata (1CXY), was used [36]. The model of hIZA1 is shown in the right panel in an orientation similar to that of bovine cytochrome b5. The secondary structures of the two molecules are colored similarly, except the red stretch of β-sheet in hIZA1 highlighted to indicate the region important for heme binding. Two Tyr residues that may play a role in heme binding are shown in yellow.

IZA was expressed as either a His6-fused protein or a glutathione S-transferase (GST)-fused protein in Escherichia coli and purified to homogeneity. The purified protein was tinged with brown, a color clearly distinct from the bright red color of the similarly expressed and purified cytochrome b5 (not shown). The UV and visible light absorption spectra of His6-rIZA1 are shown in Fig. 2A, revealing the oxidized form of the heme chromophore, with a sharp γ-absorption peak at 402 nm and broad absorptions between 497 nm and 616 nm (shown in green). When the sample was treated with sodium dithionite, the spectra were converted into those of the reduced heme chromophore with distinct α and γ peaks at 559 nm, and 426 nm, respectively (shown in red). The addition of CO to the reduced sample changed the spectra into a CO-binding form with α, β and γ peaks at 567 nm, 538 nm and 420 nm respectively (shown in blue). The incubation of the oxidized form with either NADH and NADH-cytochrome b5 reductase or NADPH and NADPH-cytochrome P450 reductase did not influence the absorption spectra. As shown in supplementary material Tables S1 and S2, hIZA1 and rIZA1 had essentially similar spectral properties, no matter whether they were expressed as His6-tagged proteins or GST-tagged proteins.

Figure 2.

IZA1 contains a protoheme. (A) UV and visible light absorption spectra of His6-rIZA1. The oxidized form absorption spectrum (green), the reduced spectrum taken after the addition of sodium dithionite (red) and the CO-bound spectrum (blue) are shown. (B) EPR spectra of the oxidized forms of GST-rIZA1, horse heart myoglobin and human cytochrome b5 are shown. (C) EPR spectrum of the 14NO-bound form of GST-rIZA1 at 35 K. (D) Titration of apo-rIZA1 with hemin. (E) GST-rIZA1 was treated under various conditions, subjected to SDS/PAGE, and stained by the peroxidase reaction (left panel). The right panel shows Coomassie blue staining of the same gel.

The nature of the heme bound to IZA was further studied by measuring EPR spectra (shown in Fig. 2B). The spectra of rat GST-rIZA1 at either 5 K or 15 K showed high-spin type signals with g values near 6.0 and 2.0. Unlike those of oxidized myoglobin, the EPR signals showed strong anisotropy; the signals near g =6.0 appeared to be a mixture of two components. The major component had larger anisotropy (g1 = 6.44 and g2 = 5.57) and the minor, smaller anisotropy (g1 = 6.10 and g2 = 5.90). When 14NO was added to the reduced form (Fig. 2C), the EPR spectra revealed a 14NO-bound penta-co-ordinated heme, indicating that the co-ordination between heme iron and an amino acid was disrupted upon binding of NO. hIZA1 and hIZA2 yielded spectra essentially similar to those of rIZA1 whether purified as (His)6-fused proteins or GST-fused proteins. Taken together these EPR properties suggest that IZA, like myoglobin, contains a high-spin type heme. However, unlike myoglobin which has His as the fifth ligand for heme iron, the ligand of IZA may not be a single amino acid. Rather, it is possible that two amino acids each partially contribute to binding the heme, producing the mixture of two anisotropic EPR signals. The fact that His6-IZA showed essentially the same EPR spectra as GST-IZA excludes the possibility that the heme is nonspecifically bound to an imidazole group contained in the His tag.

Acid/acetone treatment of rIZA1 released heme from the protein. Aliquots of hemin were added to the apoprotein thus prepared, and A402 was monitored (Fig. 2D). This titration revealed a reflection point where 4 µm hemin was added to 5 µm apoprotein, apparently indicating that one molecule of rIZA1 maximally bound 0.8 molecule of heme. However, the absorption coefficient of the rIZA1-bound heme at 402 nm may be different from that of free heme. Therefore, we examined this point further.

The A280 of a protein molecule can be calculated based on the content of aromatic amino acids, and our calculation suggested that 10 µm GST-hIZA1 would have an A280 of 0.357 absorbance unit. On the other hand, when we added small amounts of hemin dropwise to a 20 µm apo-GST-hIZA1 solution and recorded A402, the difference in absorbance between the sample added with 1 µm hemin and that added with 5 µm was 0.263, suggesting that 1 µm bound heme would have an A402 of 0.0658 absorbance unit. Thus, if heme bound to 10 µm GST-hIZA1 stoichiometrically, the A402 of the holoprotein would be 0.658 absorbance unit, and we can determine the value A402/A280 of the holoprotein as 0.658/0.357 = 1.84. In the meantime the maximal value of A402/A280 that we obtained for several purified samples was 1.07. This suggested that one molecule of the purified GST-hIZA1 contained about 0.6 molecule of heme at most. This value was reasonably consistent with the approximate value obtained from the result of Fig. 2D.

Several heme-binding proteins are known to bind heme tightly, so that the proteins can be detected as peroxidase reaction-stained bands even when subjected to electrophoresis in SDS-containing gels. To characterize the heme-binding nature of IZA1, purified GST-rIZA1 was subjected to SDS/PAGE, and then the gel was stained by the peroxidase reaction (Fig. 2E left panel). As shown in the first lane from the left, three bands appeared, with molecular masses of ≈ 50 kDa, 85 kDa and 130 kDa, suggesting that heme was still bound to the monomeric, dimeric, and trimeric forms of GST-rIZA1. (The theoretical molecular mass of GST-rIZA1 is 60.2 Da.) Similar bands appeared in the lane loaded with heat-denatured GST-rIZA1 (the second from the left). Thus, heme bound to GST-rIZA1 seemed not to be released from the protein even when treated in boiling water. In a lane loaded with the sample pretreated with heat in the presence of dithiothreitol, the relevant peroxidase-reaction-stained bands disappeared, suggesting that heme was released from the protein after these treatments, although another interpretation may be that dithiothreitol treatment reduced the heme iron, making it negative to peroxidase activity. In any case, these results indicate that IZA binds heme relatively tightly.

When treated with pyridine under alkaline conditions, the heme molecule produces a pherochrome complex with characteristic absorption spectra. Figure 3A illustrates the redox difference absorption spectra of pyridine pherochromes prepared from rIZA1, myoglobin and cytochrome c oxidase. The spectra of rIZA1-derived pherochrome, like those derived from myoglobin, but unlike those derived from cytochrome c oxidase, had peaks at 419 nm, 525 nm and 556 nm, suggesting that heme bound to rIZA1 is of type b, not of type a. To confirm this point, heme extracted from rIZA1 was subjected to HPLC analysis (Fig. 3B). The results show that heme derived from rIZA1 had the same retention time as that from myoglobin. These results again suggest that rIZA1 contains type b heme, not type a heme.

Figure 3.

IZA1 contains a type b heme. (A) The redox-difference adsorption spectra of pyridine-hemochromogens prepared from His6-rIZA1, GST-rIZA1, myoglobin and cytochrome c oxidase. The absorption spectra of hemochromogens derived from myoglobin and rIZA1s had peaks at 556 nm, 524 nm and 419 nm, whereas that derived from cytochrome c oxidase, at 588 nm, 536 nm and 431 nm. (B) The hemes released from GST-rIZA1, myoglobin and cytochrome c oxidase, and free hemin were subjected to HPLC analysis. The retention times of the hemes extracted from rIZA1 and myoglobin were 28.9 min and 29.2 min, that of hemin 28.4 min, and that of cytochrome c oxidase 38.4 min.

To determine which amino-acid residue interacts with heme in IZA1, a variety of mutant IZA1s were produced in which amino acids thought to bind the heme ligand had been disrupted. The purified mutants were then evaluated for their heme absorption. Because an imidazole group often plays a role in binding heme in many heme proteins, we first introduced mutations into His165 and His166 in hIZA1, even though they are located outside the predicted heme/steroid-binding domain. H165N-hIZA1 and H166N-hIZA1, however, were found capable of binding heme as strongly as the wild-type (not shown). Amino-acid side-chain groups other than imidazole that could interact with heme molecule are thiol and phenol. Noting that Tyr107, Tyr113, Tyr139, and Cys129 are present in the heme/steroid-binding region, and moreover are conserved in hIZA1, hIZA2, and rIZA1, mutants Y107F, Y113F, Y139F, and C129A were produced. The heme absorptions of these mutants, however, again seemed not significantly diminished compared with that of the wild-type. We tested further mutants, such as Y43F, Y164F, Y180F and P109A, but none of these single-amino-acid mutants seemed to lose heme-binding capability completely. When two phenol groups, Tyr107 and Tyr113, were disrupted, the mutant appeared substantially to lose its capacity to bind heme (Fig. 4A). In contrast, another double mutant, Y164F/H166N-hIZA1, retained heme-binding capacity (not shown). When mutations were introduced into a four consecutive amino-acid stretch from Asp99 to Lys102, the mutant bound heme at a level of 10% of the wild-type (Fig. 4A). It should be noted that three amino-acid residues in this tetrapeptide, Asp99, Thr101 and Lys102, are conserved in IZA and cytochrome b5.

Figure 4.

(A) The heme-binding capacities of the wild-type hIZA1 and mutants. The capacities were estimated by measuring the A402/A280 ratios (mean and SD of triplicate expression). The purification of the hIZA1s was given in Experimental procedures. (B) The amounts of protein used for the heme absorbance measurement were shown by immunoblot analysis.

The 3D structure of the heme-binding pocket of bovine cytochrome b5 was adopted from the previously published crystallographic study (left panel in Fig. 1C). Next the 3D structure of the heme/steroid-binding domain of IZA was modeled based on that of the Ectothiorhodospira vacuolata cytochrome b5 homolog (accession number 1CXY) and shown in the same orientation as that of bovine cytochrome b5 (right panel in Fig. 1C). The simulated structure revealed a heme-binding pocket surprisingly similar to that of cytochrome b5, with a space large enough to accommodate a heme molecule. Interestingly, if a heme were inserted into this pocket, those residues mentioned above for their importance in the interaction with heme, i.e. Tyr107, Tyr113 and the tetrapeptide, D99K102, seem to be located at one side of the heme molecule (the upper side in this orientation), constituting the ceiling of the heme-binding pocket.

Determination of the intracellular localization of IZA would provide insights into its physiological function. IZA1 was first reported as MPR and purified from the membrane fractions of rat liver homogenates. Immunohistochemical observations of other investigators revealed that this protein forms vesicle-like structures in cells. Moreover, the predicted domain structure of IZA indicated that it contains a transmembrane region (Fig. 1A). All these previous reports indicate that IZA1 is a membrane-associated protein. However, to which intracellular membrane compartment IZA1 is associated is not clear. To determine the intracellular localization of IZA1 more precisely, we expressed rIZA1, cytochrome b5, an endoplasmic reticulum-associated protein, and CYP11B1, a mitochondrial inner membrane-associated protein, in HeLa cells. The cells were stained with the specific antibodies directed against the respective proteins. As shown in Fig. 5, rIZA1 was distributed diffusely in the cell, forming vesicular structures, suggesting its association with the membrane compartments [8]. In addition, rIZA1 appeared to be somewhat concentrated at a perinuclear region. The intracellular location of rIZA1 was completely consistent with that of coexpressed cytochrome b5, but not with that of coexpressed CYP11B1. These results suggest that IZA1 is associated with the endoplasmic reticulum membrane.

Figure 5.

Intracellular localization of rIZA1, rat CYP11B1 and human cytochrome b5. HeLa cells were cotransformed with rIZA1 and CYP11B1 or cytochrome b5. After 24 h, cells were fixed and subjected to immunocytochemistry by using anti-rIZA1 monoclonal antibody [4], anti-CYP21 polyclonal antibody [32], or anti-(cytochrome b5) Ig [33].

Given that IZA1 is abundantly present in the endoplasmic reticulum of zona fasciculata cells, it would be reasonable to speculate that it is involved in the physiology of the adrenal cortex. We indeed reported previously that the steroid 21-hydroxylation reaction, which is essential for biosynthesis of corticosteroids, was enhanced in the presence of rIZA1 [10]. We re-examined this by expressing CYP21 together with hIZA1 or its mutants in COS-7 cells (Fig. 6A). Secretion of 11-deoxycorticosterone, the CYP21 reaction product from progesterone, was increased about twofold by coexpressing the wild-type hIZA1, whereas it was depressed by ≈ 75% by coexpressing the D99–K102-mutated hIZA1, and increased by ≈ 60% by coexpressing the Y107F/Y113F-hIZA1 (Fig. 6A). When the levels of expressed hIZA1s in cell homogenates were examined, the D99–K102-mutated hIZA1 was expressed at a higher level than the wild-type hIZA1 (Fig. 6A, lower panel), suggesting that this mutant was fairly stable in the cells, although in this experiment we could not confirm the mutant's intracellular localization as the endoplasmic reticulum. On the other hand, the level of expressed CYP21 in these cells seemed to be slightly lower than in the wild-type hIZA1-expressing cells.

Figure 6.

Effects of IZA1 on the CYP21 reactions. (A) Wild-type hIZA1 and mutant hIZA1s were coexpressed with CYP21 in COS-7 cells. The conversion of progesterone into 11-deoxycorticosterone was measured as given in Experimental procedures. Lower panels show the levels of expressed CYP21 and hIZA1s by immunoblot analysis. (B) To test the effect of hIZA1 on the promoter activity of CYP21 plasmid, pSVL-fLuc plasmid was cotransformed with hIZA1s expression vectors and Renilla luciferase vector as an internal standard. The promoter activity of pSVL was normalized to Renilla luciferase activity. (C) Effect of IZA on the CYP21 reaction in the reconstitution system. The reaction mixture contained, in a final volume of 1 mL, various amounts of (His)6-rIZA1, 25 pm CYP21, 50 pm NADPH-P450 reductase, 50 µm progesterone, 0.5 mm NADPH, 8 mm isocitrate, 0.1 U isocitrate dehydrogenase, 50 mm Tris/HCl, pH 7.4, and 10 mm MgCl2. The reactions were carried out at 37 °C for 2 min. (D) Effect of rIZA1 or cytochrome b5 on guinea pig CYP17 reactions in the reconstitution system. The reaction mixture contained, in a final volume of 1 mL, 0.25 µm CYP17, 0.5 µm NADPH-P450 reductase, 50 µm progesterone or 17α-hydroxyprogesterone, 0.5 mm NADPH, 8 mm isocitrate, 0.1 U isocitrate dehydrogenase, 50 mm Tris/HCl, pH 7.4, and 10 mm MgCl2. The reactions were carried out with or without 0.25 µm His6-rIZA1, or 0.25 µm cytochrome b5, at 37 °C for 2 min. After the reactions, steroids were extracted and analyzed as described previously [36].

To exclude the possibility that D99–K102-mutated hIZA1 repressed the expression of CYP21 protein by inhibiting the promoter used for CYP21 expression, firefly luciferase cDNA was introduced into the vector instead of CYP21 cDNA, and the promoter activities were measured in the wild-type hIZA1-expressing cells and the mutant hIZA1-expressing cells. Overexpression of hIZA1, whether wild-type or mutant, did not influence the promoter activity of the expression vector (Fig. 6B), suggesting that the slightly lower concentration of CYP21 protein in the D99–K102-mutated hIZA1-expressing cells may be due to a post-translational event; possibly, CYP21 protein instability is induced by the coexistence of the D99–K102-mutated hIZA1.

Next, we tested the possibility that IZA directly regulates the CYP21-dependent steroid hydroxylation reaction by using the microsomal P450 electron-transport reconstitution system. As shown in Fig. 6C, the addition of rIZA1 failed to stimulate the CYP21-catalyzed 21-hydroxylation of progesterone in the reconstituted system. Rather, the hydroxylation activity seemed to be inhibited in the presence of a large amount of rIZA1. Although we cannot explain this phenomenon beyond doubt, the involvement of a hydrophobic protein such as rIZA1 in the reconstitution system may disturb the smooth conduct of the electron transport to the CYP21 molecule. The effect of rIZA1 on the CYP17-dependent hydroxylation reactions was also tested in comparison with the effect of cytochrome b5, because the latter is well known to regulate the CYP17-mediated 17α-hydroxylation reaction and the consecutively occurring 17,20-lyase reaction [17]. As reported previously, the presence of cytochrome b5 in the CYP17-reconstitution system seemed not to influence the 17-hydroxylation of progesterone, but it indeed activated the lyase reaction of 17α-hydroxyprogesterone (Fig. 6D). In contrast, the presence of rIZA1 seemed to influence neither one of the reactions. We surmised therefore that IZA1 could activate the CYP21-dependent reaction in the transformed cells, but this activation may not be caused by the direct interaction of IZA1 with the microsomal P450 electron-transport components, as seems to be the case for cytochrome b5.

Taken together, the results presented here show that IZA1 is a heme-binding protein present in the endoplasmic reticulum membrane. The primary structure of its heme-binding region looks slightly similar to that of cytochrome b5, presumably forming a hydrophobic pocket. The heme in IZA1 is type b, and binds to the protein in high-spin type. To identify the amino-acid residues involved in binding to the heme, extensive site-directed mutation studies were conducted. However, the results remain somewhat ambiguous. Nevertheless, it is possible to conclude that the heme/steroid-binding region in IZA1 constitutes a hydrophobic pocket that could accommodate a heme molecule, and, in this pocket, two Tyr residues, Tyr107 and Tyr113, and a peptide stretch D99–K102 play important roles in attaching the heme iron to one side of the protoporphyrin ring. Mallory et al. [16] recently reported the nature of Dap1p, the yeast homolog of IZA1, which also seems to bind heme.

As IZA1 cDNA was first isolated as MPR, we tried assaying its progesterone-binding activity under various conditions. For instance, [3H]progesterone was incubated with GST-rIZA1 and the extent of isotope binding to the protein was estimated by GST pull-down assays. The results failed to show specific binding of the isotope to rIZA1 (supplementary Table S3). The addition of an IZA1 antibody to the incubation mixture of the radioactive progesterone and rIZA1 also failed to show specific isotope binding. Therefore our investigation has so far failed to establish that rIZA1 specifically binds progesterone.

Although the results presented here cannot conclusively establish the precise physiological role(s) played by IZA in the adrenal cortex, we surmise that this heme-containing microsomal protein may have a role in supplying heme molecules to cytochrome P450-involved reactions and eventually influence adrenal steroidogenesis. A similar role of Dap1p in the CYP51-catalyzed reaction in yeast has been suggested by Mallory et al. [16].

Experimental procedures


A plasmid containing hIZA1 cDNA (IMAGE clone, No. 5300612) and a transfection reagent, lipofectamine 2000, were purchased from Invitrogen (Carlsbad, CA, USA). pGEX-6p-3 vector was from Amersham Bioscience (Piscataway, NJ, USA). pTargeT vector and pGL3 luciferase reporter assay system were obtained from Promega (Madison, WI, USA). Restriction endonuclease and E. coli strains, JM109 and Bl21, were purchased from Takara (Kyoto, Japan) and Toyobo (Osaka, Japan), respectively. QuikChange XL Site-Directed Mutagenesis Kit was from Stratagene (La Jolla, CA, USA).

Construction of plasmids

A BamHI site (GAATTC) was created at a point before the starting Met codon of hIZA1 cDNA by site-directed mutagenesis. The cDNA containing both coding and 3′ noncoding regions was prepared by BamHI and NotI digestion, and subcloned into the BamHI/NotI site of pGEX-6P-3 or pTargeT. The resultant plasmids were named pGEX-hIZA1 and pTargeT-hIZA1, respectively. Point mutations of hIZA1 were produced by site-directed mutagenesis using pGEX-hIZA1 as a template. To obtain N-terminal His6 rat IZA, cDNA was PCR amplified with 5′ primer containing the NcoI site (TACCATGGCTGCCGAGGATG) and 3′ primer containing the HindIII site (CAAGCTTCAGTCACTCTTCCGAGC). The PCR product was digested with NcoI and HindIII and subcloned into the NcoI/HindIII site of pRSET-B. The resulting construct containing N-terminal His6 was recloned into pCW vector using NdeI and HindIII.

Expression and purification of IZA1

IZA1 was expressed in E. coli JM109 as GST-fused or (His)6-fused protein as reported previously [18,19] with some modifications. For the purification of the GST-fused protein, E. coli JM109 transformed with pGEX-hIZA1 was grown in 1 L 2YT (yeast/tryptone) medium containing 0.2 mmδ-aminolevulinic acid hydrochloride at 30 °C. When culture growth reached D600 ≈ 0.7, 0.1 mm isopropyl thio-β-d-galactopyranoside was added to the medium, and culture was continued for 16 h at 25 °C. E. coli was harvested from the culture solution, and GST-hIZA1 expressed was purified as described previously [19]. (His)6-rIZA1 was coexpressed with glutamyl-tRNA reductase (hemA/gtrA) [20]. E. coli JM109 cotransformed with the pCW-rIZA1 and pHg2 (hemA) was grown in 3000 mL TB (terrific broth) medium containing 100 µg·mL−1 ampicillin, 25 µg·mL−1 chloramphenicol and microelements. The culture was performed at 37 °C until the cell density reached D600 = 0.6–0.8. Then 0.5 mm isopropyl thio-β-d-galactoside, 100 µg·mL−1 ampicillin and 25 µg·mL−1 chloramphenicol were added to induce protein expression. The cells were grown for a further 24 h at 29 °C, harvested, and frozen at −70 °C for later use. The frozen cells were thawed in 100 mL 50 mm Tris/HCl buffer, pH 7.5, containing 20% (v/v) glycerol and 0.3 m NaCl, and sonicated on ice using a Tomy Ultrasonic disruptor UD-200. Proteins associated with the membrane fraction of sonicates were solubilized by the dropwise addition of 10% (w/v) sodium cholate to the final concentration of 1%. To the solution containing the solubilized proteins, imidazole was added to a final concentration of 5 mm. The solution was loaded on to a Ni/nitrilotriacetate/agarose column to absorb the His6-fused proteins. The proteins were eluted with the 50 mm Tris/HCl buffer, pH 7.5, containing 20% (v/v) glycerol, 0.2% (w/v) sodium cholate and 50 mm histidine, and His6-rIZA1 eluted in colored fractions was further purified by hydroxyapatite column chromatography.

Preparation of mutated hIZA1s used for measuring heme-binding capacity

E. coli BL21 was used for expressing GST-hIZA1 and its mutants. The transformed E. coli was cultured in 2YT medium without δ-aminolevulinic acid, and induction of protein expression was initiated as described above. The cells harvested were suspended in buffer A (50 mm Tris/HCl, 1 mm EDTA, 300 mm NaCl, pH 8.0), sonicated on ice, and lysed with 2% (v/v) Triton X-100. The lysate was then centrifuged at 8000 g for 30 min, and the supernatant recovered was applied to a glutathione–Sepharose column (Amersham Bioscience) pre-equilibrated with buffer B [buffer A containing 0.1% (v/v) Triton X-100 and 5% (v/v) glycerol]. The column was washed with 2 column vol. buffer B. GST-hIZA1 protein was retained on the column at this step, although it was uncolored because it had been expressed without δ-aminolevulinic acid. To obtain the protein in a heme-bound form, 0.05 mm hemin chloride dissolved in buffer B containing 1% dimethyl sulfoxide was loaded on to the column. The column was then washed with 5 times the column volume of buffer B and 5 times the column volume of buffer C [50 mm Tris/HCl, 1 mm EDTA, 100 mm NaCl, 5% (v/v) glycerol, and 0.5% (w/v) sodium cholate]. GST-hIZA1 was finally eluted from the column with buffer C containing 10 mm glutathione. The purified protein was dialyzed against buffer C, and its absorption spectrum recorded to evaluate its heme-binding capacity.

Spectrophotometric analysis

UV-visible absorption spectra were measured using a JASCO V-550 UV/VIS spectrophotometer system. Photometric determination of heme type was performed by a pyridine hemochrome method [21]. The heme type was also confirmed by HPLC analysis. Heme bound to IZA1 was extracted by acetone/HCl followed by ethyl acetate treatment, and subjected to HPLC using the Shimadzu CL-10A HPLC system equipped with a reverse-phase column (YMC-Pack, ODS-A303, S-5, 250 × 4.6 mm) as described by Fromwald et al. [22]. Heme a, which was used as the standard, was extracted from bovine heart cytochrome c oxidase purified by the method of Yonetani [23]. Horse skeletal muscle myoglobin and hemin were obtained from Wako Pure Chemical industries, Ltd (Osaka, Japan) and ICN Pharmaceuticals, Inc. (Irvine, CA, USA), respectively.

Preparation of apo-rIZA1

Cold acid/acetone solution [0.2% (v/v) HCl; −20 °C; 10 mL] was added dropwise to 20 nmol rIZA1 dissolved in 0.5 mL potassium phosphate buffer (10 mm, pH 7.4). The mixture was stirred for 1 h at 4 °C, and then centrifuged at 5000 g for 10 min at 4 °C. The precipitate recovered was dried under a stream of N2 and solubilized in 0.5 mL potassium phosphate buffer (100 mm, pH 7.4) containing 0.5% 3-[(3-cholamidopropyl)dimethylammonio]propanesulfonic acid (CHAPS), 1 mm EDTA and 1 mm dithiothreitol. The solution was then dialyzed against three changes of 1 L potassium phosphate buffer (100 mm, pH 7.4), containing 0.1% (v/v) CHAPS, 1 mm EDTA, and 1 mm dithiothreitol. The apo-rIZA1 obtained mobilized as a single band in SDS/PAGE and did not contain any heme absorbance in the Soret region. The binding of heme to apo-rIZA1 was monitored by adding dropwise 1-µL aliquots of hemin (1 mm, dissolved in dimethyl sulfoxide) into the sample cuvette, which contained 5 µm protein in potassium phosphate buffer (50 mm, pH 7.4) and 50 mm NaCl. The reference cuvette contained the same solution without the protein. A402 was monitored after each addition of the aliquot, and plotted against the amounts of hemin added [24].

EPR measurements

EPR measurements were carried out at X-band (9.23 GHz) microwave frequency with a Varian E-12 spectrometer with 100-kHz field modulation. An Oxford flow cryostat (ESR-900) was used for measurements at cryogenic temperatures. The microwave frequency was calibrated with a microwave frequency counter (model TR5212; Takeda Riken Co. Ltd, Osaka, Japan). The strength of the magnetic field was determined with an NMR field meter (model EFM 2000AX; ECHO Electronics Co. Ltd, Hong Kong). Samples were loaded into EPR tubes at 4 °C and frozen immediately in liquid nitrogen. Other conditions were as described previously [25,26].

Cells culture, immunofluorescence microscopy, and steroid secretion

COS-7 and HeLa cells were grown in Dulbecco's modified Eagle's medium (Sigma, St Louis, MO, USA) supplemented with 10% (v/v) fetal bovine serum and antibiotics at 37 °C under an atmosphere of 5% CO2/95% air (v/v). To determine the subcellular localization of IZA1, it was coexpressed with either human cytochrome b5 or rat CYP11B1 in HeLa cells, IZA-nonexpressing cells. The proteins expressed were visualized using fluorophore-labeled antibodies as described previously [27,28]. For measurement of steroid production, COS-7 cells (2 × 105) plated on a 10-cm dish were transfected with 2.0 µg pTargetT-hIZA1 plasmid or its mutants and 1.0 µg pSVL-CYP21 plasmid using lipofection transfection. The cells were incubated in Dulbecco's modified Eagle's medium for 24 h, and then the medium was replaced with fresh medium containing 100 µm progesterone. The incubation was continued for 24 h, and the medium was harvested. Steroid products were extracted from the medium into dichloromethane and analyzed using HPLC with 60% (v/v) ethanol as described previously [18].

Reporter assays

COS-7 cells were transfected with a reporter plasmid pSVL-luc, pTargetT-hIZA1 and its mutants, pRL-TK (Promega) using Escort V (Sigma) reagent. The cells were incubated for 16 h and harvested. The preparation of cell lysates and the assay for luciferase activity using the Dual-Luciferase Reporter Assay System were performed according to the manufacturer's instructions (Promega).

In vitro reconstitution assay

In vitro reconstitution assays for CYP21 and CYP17 activities were as described in [17,28–31].


We thank Dr Shiro Kominami and Dr Takeshi Yamazaki (Hiroshima University, Higashi-Hiroshima, Japan) for providing us with antibodies against cytochrome b5 and CYP21 and IgG against CYP17, and cytochrome b5[32,33]. We also acknowledge that the preliminary work on the progesterone-binding assays of rIZA1 was performed by Dr Nanao Horike at our laboratory. A part of this work was supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Science, Sports and Culture, Japan and Technology and Ministry of Health, Labor and Welfare Japan.