Identification of Crucial Histidines for Heme Binding in the N-terminal Domain of the Heme-regulated eIF2 (cid:1) Kinase*

The heme-regulated eukaryotic initiation factor-2 (cid:1) (eIF2 (cid:1) ) kinase (HRI) regulates the initiation of protein synthesis in reticulocytes. The binding of NO to the N-terminal heme-binding domain (NTD) of HRI positively modulates its kinase activity. By utilizing UV-visible absorption, resonance Raman, EPR and CD spectroscopies, two histidine residues have been identified that are crucial for the binding of heme to the NTD. The UV-visible absorption and resonance Raman spectra of all the histidine to alanine mutants constructed were similar to those of the unmutated NTD. However, the change in the CD spectra of the NTD construct containing mutation of His 78 to Ala (H78A) indicated loss of the specific binding of heme. The EPR spectrum for the ferric H78A mutant was also substantially perturbed. Thus, His 78 is one of the axial ligands for the NTD of HRI. Significant changes in the EPR spectrum of the H123A mutant were also observed, and heme readily dissociated from both the H123A

The heme-regulated eukaryotic initiation factor-2␣ (eIF2␣) 1 kinase (HRI) is a member of a family of kinases that regulate initiation of protein synthesis in eukaryotic cells. HRI func-tions, in part, to coordinate the synthesis of globin chains in reticulocytes with heme availability (1). A deficiency of heme, which is required for the assembly of ␣and ␤-globin chains into hemoglobin, induces the activation of HRI, which subsequently phosphorylates the ␣-subunit of eIF2, leading to the inhibition of polypeptide chain initiation and the arrest of protein synthesis.
HRI is a multidomain hemoprotein that contains two distinct heme-binding sites (2). One site, which has been tentatively assigned as being located in the "kinase insertion domain" of HRI, appears to bind heme "reversibly" and regulates HRI activity in response to changes in heme concentration (3). The other heme-binding site is located in the N-terminal domain (NTD) of HRI, which consists of ϳ165 amino acids (4). This domain is responsible for the stable "constitutive" binding of heme to HRI (4) and appears to be the active center for nitric oxide (NO)-induced activation of HRI (5). The isolated NTD stably binds heme but shows no kinase activity (4). The kinase activity of HRI is coded by the C-terminal region (2). The isolated C-terminal region has the "reversible" heme-binding site and phosphorylation activity, but it does not respond to NO. Although axial ligands for the heme-binding sites have not yet been determined, sequence alignment of the NTD to globins (5) and preliminary spectroscopic analyses of the NTD (6) suggested that histidine was the axial ligand of the heme-binding site in the NTD.
In the presence of NO, heme in the NTD forms a six-coordinate NO complex (5,6). The formation of this six-coordinate NO-Fe-His complex is of interest, because it differs from what occurs in the prototypical NO-responsive sensor protein, soluble guanylate cyclase. In soluble guanylate cyclase, the binding of NO to the heme iron induces cleavage of the iron-histidyl bond to form a five-coordinate NO complex, which is postulated to be the primary trigger for the activation of its cyclase activity (7). However, we have shown that the iron-histidyl bond is retained in the NO adduct of the NTD (6). This observation has led us to propose that changes in the conformation of amino acid residues adjacent to the NO binding site (the "distal" heme ligand), not cleavage of the axial histidine ligand at the "proximal" side of the heme-binding site, would be crucial for the NO-induced activation mechanism in HRI (6).
To determine the axial ligands in the NTD, we prepared point mutants substituting Ala for His at each of the seven histidine residues present in the NTD of rabbit HRI (His 78 , His 81 , His 83 , His 122 , His 123 , His 129 , and His 148 ) (5). Two of the residues (His 78 and His 123 ) are invariant amino acids from zebrafish to human and the other four histidines (His 81 , His 83 , His 122 , and His 129 ) are conserved in mammalian HRIs (Table  I). In addition, we prepared another mutant in which Gln 58 , which molecular modeling suggested might be equivalent to * This work was supported by the Oklahoma Agricultural Experiment Station Project No. 1975 (to R. L. M.) and Grants-in-aid 12002008 (to I. M.) and 14658217 and 15350101 (to K. I.) from the Ministry of Education, Culture, Sports, Science, and Technology in Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ the distal histidine ligand of globins (4), was replaced with Ala. Using a combination of these mutations in the NTD with UVvisible absorption, resonance Raman, EPR and CD spectroscopies, His 78 and His 123 were identified as being two crucial residues for the ligation of heme within the NTD of HRI.

Expression and Purification of Recombinant NTD and Its Mutants-
Ala to His mutants within the NTD were constructed as described previously (4). All of the mutants were expressed in Escherichia coli and isolated as holoproteins. The unmutated and mutated NTDs of HRI were purified as described previously (6) with some modifications. To avoid nonspecific heme binding, the (His) 5 -tag was cleaved by enterokinase (4), and the cleavage of the (His) 5 -tag was confirmed by the N-terminal amino acid analysis and SDS-PAGE. The spectroscopic data obtained with the proteins from which the (His) 5 -tag was cleaved were identical to the data obtained from (His) 5 -tag fused proteins, as reported previously (5).
The heme concentration was determined by the pyridine hemochromogen assay (8), and the heme content was estimated by the absorption ratio of the Soret peak to 280 nm (A Soret /A 280 nm ). We observed that the A Soret /A 280 nm for the recombinant NTD is about 4.5. However, heme dissociated from the His 78 , His 122 , and His 123 mutants during the purification. The A Soret /A 280 nm for these mutant proteins were at least 1.5, indicating that more than 30% of the heme binding sites were occupied by heme. The destabilized protein structure and weak heme affinity of these mutants prevented further purification.
Spectroscopy-The UV-visible absorption and CD spectra of the purified NTD and its mutants were obtained using a Perkin-Elmer lambda 19 and a Jasco model 720 spectrometers, respectively, at room temperature. Proteins at a concentration of 5-10 M in 50 mM Tris-HCl (pH 7.5) were placed in a cuvette (1-cm cell length). The ellipticity in the CD spectra was normalized by the heme concentration, and the presence of the apoprotein did not affect the CD spectra for the Soret region. Resonance Raman spectra were measured using the system and the method reported previously (6). The EPR spectra of the NTD and its mutants were measured at 15 K with a Varian E-12 spectrometer (X-band; 9.22 GHz) equipped with an Oxford ESR-900 liquid helium flow cryostat. The sample concentration for the EPR measurements was 200 M in 50 mM Tris-HCl at pH 7.5.

RESULTS AND DISCUSSION
As reported previously (5), the ferric NTD exhibited a UVvisible absorption spectrum with a Soret peak at 415 nm and two bands at 534 and 565 nm in the visible region, which is characteristic of the low spin state in which two axial ligands are ligated to the heme iron. A similar UV-visible spectrum was also encountered for ferric cytochrome b 5 , a typical bishistidine-ligated hemoprotein (9) (Table II). While the resonance Raman spectrum of the ferric NTD exhibited a weak 3 line at 1473 cm Ϫ1 , implying the presence of a five-coordinate heme species, an intense 3 line characteristic of a six-coordinate heme species was observed at 1506 cm Ϫ1 (Table III). Thus, the ferric state of the recombinant NTD has a six-coordinate heme species as its major component.
Upon the reduction of the heme, the Soret peak was redshifted to 429 nm, and addition of CO shifted the peak to 424 nm. Although the Soret peak of the ferrous recombinant NTD was red-shifted, compared with that of ferrous cytochrome b 5 , and was rather close to that of five-coordinate species, such as deoxymyoglobin (534 nm), the clear presence of two peaks in the visible region of the UV-visible absorption spectrum (Table  II) indicated that the ferrous recombinant NTD was predominantly in the six-coordinate state, as we reported previously (6). The intense 3 line at 1492 cm Ϫ1 in the resonance Raman spectrum (Table III) supports the conclusion that the six-coordinate state was the major species present in the ferrous state of the recombinant NTD, while a very weak 3 line at 1468 cm Ϫ1 suggested that a minor amount of five-coordinate heme was also present. For the CO adduct of the recombinant NTD, the correlation between the stretching lines of Fe-C and FeC-O showed that the ligand trans to CO was neutral (6,10). Although the UV-visible absorption and resonance Raman spectra cannot unambiguously identify the neutral axial ligands in the recombinant NTD, additional data presented below suggest that the most probable amino acid residue for the axial ligands is histidine.
Unexpectedly, the UV-visible absorption spectra of the mutated NTDs, whose maxima are summarized in Table II, indicate that the His to Ala mutations did not highly perturb the spectral pattern. However, detailed inspection revealed that the His 78 mutant has a red-shifted Soret peak in the ferric state, while the Soret peak in the ferrous and ferrous-CO states was blue-shifted. The His 123 mutant also showed enhanced spectral shifts. While the UV-visible absorption spectrum of the ferric His 123 mutant was quite similar to that of the unmutated NTD, significant peak shifts, as found for the His 78 mutant, were observed for the ferrous and ferrous-CO states.
The spectral differences of the His 78 and His 123 mutants from the unmutated NTD were more evident in the CD spectra. Fig. 1A shows the CD spectra of the Soret region for the ferric NTDs. Except for the His 78 and His 123 mutants, the mutants showed CD spectra similar to the unmutated NTD, exhibiting a positive peak around 400 nm and a negative peak at 417 nm. In sharp contrast to the spectrum of the unmutated NTD, the CD spectrum of the His 78 mutant was almost featureless. The disappearance of the ellipticity corresponds to the loss of the optical asymmetry of the heme, implying nonspecific binding of heme to the His 78 mutant (11). His 78 is, therefore, crucial for the specific binding of heme, and it may be an axial ligand for the ferric heme iron in the NTD.
On the other hand, the His 123 mutant gave an intense positive peak at 412 nm and no negative peak in the Soret region. Although some other mutants showed deviations of the peak positions and intensities probably due to the slight structural perturbation around the heme and/or uncertainty of the heme concentrations, the spectral changes in the His 123 mutant were distinct and the mutation at His 123 significantly perturbed the heme environment of the ferric NTD. However, the effects of the His 123 mutation were different from those of His 78 . The intense peak in the CD spectrum of the His 123 mutant indicated that the mutation did not abolish the specific binding of heme. Thus, His 123 might be the other axial ligand to the heme iron with the contribution of His 123 to the specific binding of heme to the NTD being less significant than that of His 78 .
The differences between the effects of the His 78 and His 123 mutations were more enhanced in the ferrous state of the NTD. In the ferrous state, the recombinant NTD gave a positive peak at 418 nm and a negative peak at 427 nm (Fig. 1B). The CD spectrum of the His 78 mutant also had no intense signals in the Soret region, as observed for the ferric state, indicating that His 78 was ligated to the ferrous heme iron in the NTD. However, the spectral changes in the ferrous His 123 mutant were less drastic than those in the ferric state. The spectral pattern of the ferrous His 123 mutant is essentially the same as that of the unmutated NTD. The structural changes induced by the mutation at His 123 appear to be relieved by the reduction of the heme. Thus, the ligation of His 123 to the ferrous heme iron has only limited contribution to the structural environment of the bound heme, or the reduction of the heme iron might cause His 123 to dissociate from the heme iron.
The positive and negative peaks in the CD spectrum of the recombinant NTD were blue-shifted to 413 and 422 nm, respectively, upon addition of CO to the ferrous state of the NTD (Fig.   1C). The His 78 mutant still showed very weak ellipticity and no clear peaks in the spectrum. Such drastic changes in the CD spectra of His 78 mutant lead us to conclude that His 78 is the essential axial ligand, which is required for the specific binding of heme to the NTD, corresponding to the proximal histidine in globins. On the other hand, the similarities between the CD spectra of the ferrous-CO state of the unmutated and the His 123 -mutated NTD (Fig. 1C) indicated that specific binding of heme was maintained without the ligation of His 123 . However, during the purification of the His 123 mutant, we noted that dissociation of heme from its heme-binding site was enhanced, indicating that the affinity of the His 123 mutant for heme was decreased to the same extent as was found for the His 78 mutant. Thus, the ligation of heme by His 123 appears to be required for stabilizing the specific binding of heme to the NTD.
Such nonequivalent contributions of two axial ligands to the structural environment of the heme-binding site are also encountered in cytochrome b 5 (12). Mutation of one axial ligand, His 39 , resulted in minor spectral changes and formation of a rather stable hemoprotein, while the replacement of the other axial ligand, His 63 , induced severe destabilization of the protein structure and decreased the heme binding affinity of the protein (12). Based on the heme binding and spectroscopic data for the two axial mutants of cytochrome b 5 , we had proposed that His 63 plays a primary role for the binding of heme to cytochrome b 5 and acts in a manner analogous to the proximal histidine in globin, while His 39 would correspond to the "distal  histidine", with ligation of His 39 to the heme iron being critical for the electron transfer reaction catalyzed by cytochrome b 5 and not for heme binding (12).
To further examine the roles of the axial histidines in determining the structural environmental of the heme-binding site, the resonance Raman spectra were measured. Unexpectedly, most of the stretching modes of the porphyrin ring in the spectra of the His 78 and His 123 mutants could be superimposed on those of the unmutated NTD (Table III). Compared with the spectral changes reported for the axial ligand mutants in other hemoproteins (13)(14)(15), one might conclude that such small spectral changes would not reflect the substitution of the axial ligands for the heme iron of the NTD. However, it should be noted here that the NTD has histidine residues adjacent to the positions of His 78 and His 123 : His 81 and His 83 are located near His 78 , and His 122 is located next to His 123 . These histidines may coordinate to the heme iron instead of His 78 or His 123 , resulting in the slight changes in the resonance Raman spectra. With respect to the His 122 mutant, heme was also observed to dissociate during its purification, indicating that this residue plays some role in maintaining the integrity of the hemebinding site. For the His 78 mutant, loss of the intrinsic axial ligand would severely perturb the structural environment of the heme-binding site. Ligation of an alternate axial ligand may allow variations in the orientation of the bound heme, leading to the loss of the "specific binding" of heme and no clear peaks in the CD spectrum of mutant. Mutation of His 83 to Gln has previously demonstrated that this residue plays a crucial role in maintaining the integrity of the heme binding site (4). This observation further supports the hypothesis that the region of the NTD containing His 78 corresponds to the proximal side of the heme-binding site.
The EPR spectrum for the ferric state of the mutants also supports the notion that His 78 and His 123 are ligated to the heme iron. As illustrated in Fig. 2A, the EPR spectrum for the ferric state of the unmutated NTD showed g values at 3.07, 2.20, and 1.46, which is typical of the low spin bishistidineligated heme (16), although a low intensity signal at approxi- mately g ϭ 6 assignable to the high spin species was detected. In the spectrum of the His 78 mutant (Fig. 2B), the low spin g y and g z signals were shifted to g ϭ 2.27 and g ϭ 2.97, respectively, and a broad g x signal appeared around g ϭ 1.5. These g shifts of the low spin signals were also encountered for the His 123 mutant (Fig. 2F). Some additional signals around g ϭ 2.4 and g ϭ 1.9 were detected for both of the mutants, but these signals were also found in the spectrum of the partially purified recombinant protein (figure not shown). These signals may originate from a partially denatured form of the protein, with these signals being enhanced in the His 78 and His 123 mutants due to their destabilized structures.
The low spin EPR signals were slightly shifted in the His 81 and His 83 mutants as well. However, distinct changes in the EPR spectra were only observed for the His 78 and His 123 mutants. Thus, the His 81 or His 83 mutations likely caused minor changes in the coordination structure of His 78 , leading to the subtle changes in the EPR spectra. Although the structural perturbations responsible for the changes in the EPR spectra of the His 78 and His 123 mutants are not yet fully understood, the g shifts of the low spin signal at g z ϭ 3.07 to g z ϭ 2.97 and g z ϭ 2.95 in the His 78 and His 123 mutants, respectively, imply that the these mutants still have a bishistidine-ligated heme, but the rhombicity of the electronic state of the ferric heme iron is increased. Such alteration in the rhombicity corresponds to the displacement of His 78 or His 123 and the ligation of another histidine nearby, which is also suggested by the resonance Raman spectra.
Despite the clear structural significance of His 78 as an axial heme ligand, mutation of His 78 was found to have limited effects on the NO-induced activation of HRI. The kinase activity of the His 78 mutant of HRI was somewhat less sensitive to activation by NO, particularly at low NO concentrations (70% less active in the presence of 0.01 mM NOC-9, a NO generator), but the His 78 mutant still showed NO-dependent phosphorylation activity. 2 The low sensitivity to NO in the His 78 mutant would reflect some structural changes induced by the replacement of His 78 with another histidine, indicating that the proper orientation of the heme moiety, which requires His 78 , might be also important in the mechanism by which NO activates HRI. Such limited effects of the mutation at His 78 correspond to our previous suggestion (6) that the NO-induced activation of HRI is primarily based on the conformational changes in the ligand binding site, not in the cleavage of the axial ligand from the protein.
Previous results (6) have led us to propose that changes in the conformations of amino acid residues on the distal side of the heme-binding site, which are induced by the binding of NO to the ferrous heme iron, trigger the NO-induced activation of HRI. Despite the unambiguous identification of His 78 as the proximal ligand of the NTD, the identity of the amino acid residue at the NO/CO binding site, which can be referred to as the distal ligand, in the ferrous state is still uncertain. Molecular modeling based on the sequence alignment of the NTD with ␣-globin suggested that Gln 58 might act as the distal heme-binding ligand (4). However, mutation of Gln 58 to Ala did not induce significant spectral changes in the UV-visible absorption, CD, and EPR spectra. Comparison of the EPR spectra of the ferric unmutated NTD and the His 123 mutant (Fig. 2) implies the ligation of His 123 to heme in the ferric state, suggesting that the distal axial ligand in the ferrous state is His 123 , not Gln 58 . Systematic mutational studies to identify critical amino acid residues on the distal side of the heme binding site in the NTD are now in progress.