The crystal structure of Synechocystis hemoglobin with a covalent heme linkage.

The x-ray crystal structure of Synechocystis hemoglobin has been solved to a resolution of 1.8 A. The conformation of this structure is surprisingly different from that of the previously reported solution structure, probably due in part to a covalent linkage between the heme 2-vinyl and His117 that is present in the crystal structure but not in the structure solved by NMR. Synechocystis hemoglobin is a hexacoordinate hemoglobin in which the heme iron is coordinated by both the proximal and distal histidines. It is also a member of the "truncated hemoglobin" family that is much shorter in primary structure than vertebrate and plant hemoglobins. In contrast to other truncated hemoglobins, the crystal structure of Synechocystis hemoglobin displays no "ligand tunnel" and shows that several important amino acid side chains extrude into the solvent instead of residing inside the heme pocket. The stereochemistry of hexacoordination is compared with other hexacoordinate hemoglobins and cytochromes in an effort to illuminate factors contributing to ligand affinity in hexacoordinate hemoglobins.

The x-ray crystal structure of Synechocystis hemoglobin has been solved to a resolution of 1.8 Å. The conformation of this structure is surprisingly different from that of the previously reported solution structure, probably due in part to a covalent linkage between the heme 2-vinyl and His 117 that is present in the crystal structure but not in the structure solved by NMR. Synechocystis hemoglobin is a hexacoordinate hemoglobin in which the heme iron is coordinated by both the proximal and distal histidines. It is also a member of the "truncated hemoglobin" family that is much shorter in primary structure than vertebrate and plant hemoglobins. In contrast to other truncated hemoglobins, the crystal structure of Synechocystis hemoglobin displays no "ligand tunnel" and shows that several important amino acid side chains extrude into the solvent instead of residing inside the heme pocket. The stereochemistry of hexacoordination is compared with other hexacoordinate hemoglobins and cytochromes in an effort to illuminate factors contributing to ligand affinity in hexacoordinate hemoglobins.
A wide diversity in both form and function has been discovered in the study of hemoglobins (Hbs) 1 from many species, including bacteria (1, 2), plants (3), and humans (4 -7). Al-though the physiological functions of many hemoglobins are still unknown, a number of new functions have recently been described, including the scavenging of nitric oxide and oxygen (2,8), aerotaxis (9), and phototaxis (10). Although functions vary, the tertiary structures of these hemoglobins conserve many of the general features of the globin fold even when truncated to very short primary structures (11)(12)(13). However, like many proteins sharing the same fold but carrying out different functions, the specific amino acid residues surrounding the heme prosthetic group exhibit a large degree of variability across hemoglobins from different species.
An extreme example of this diversity is found in the family of hexacoordinate hemoglobins (hxHbs) in which an endogenous amino acid coordinates the ligand binding site of the heme iron in the absence of exogenous ligands. The fact that hxHbs are capable of reversible exogenous ligand binding distinguishes them from cytochrome b 5 and denatured Hbs that are not capable of ligand binding in the hexacoordinate state (12, 14 -16). This unusual characteristic could be an alternative form of regulating ligand affinity (17,18) or indicate a different functional role for hxHbs compared with the traditional role of oxygen storage and transport (19 -22). Although hxHbs are found in many species, few structures have been reported in the hexacoordinate state. Thus, further investigation of structures across a wide variety of species will aid in understanding reversible hexacoordination and discovering its physiological significance.
The hemoglobin found in the single-celled cyanobacterium Synechocystis sp. PCC 6803 (SynHb) belongs to the truncated hemoglobin (trHb) family found extensively in eubacteria, bacteria, single celled eukaryotes, and plants (2,(23)(24)(25)(26). The trHbs are 20 -40 residues shorter than non-vertebrate hemoglobins and have a 2-on-2 ␣-helical globin fold rather than the typical 3-on-3 fold. They are further characterized by a hydrophobic tunnel connecting the protein surface to the distal heme pocket that could serve as a direct route for ligand entry and/or exit from the heme pocket (2). In addition to being a hexacoordinate member of the trHb family, SynHb is distinguished by the ability to form a covalent bond between the heme 2-vinyl group and the His 117 side chain (27); this is a novel example of covalent attachment in a Hb via the porphyrin ring, and the purpose of this modification is not yet understood.
In the present study, the crystal structure of SynHb has been solved to 1.8-Å resolution. The crystal structure reveals the covalent linkage between the heme 2-vinyl and the N⑀2 atom of His 117 that is not present in the solution NMR structure of this protein (28). The crystal structure is compared with the solution NMR structure and to other trHb and hxHb structures in an effort to understand the role of this unusual covalent mod-ification as well as ligand entry, stabilization, and exit from SynHb.

EXPERIMENTAL PROCEDURES
Protein Production and Crystallization-SynHb was produced as described previously (18,24) with the following exceptions. Expression was performed without induction by isopropyl-1-thio-␤-D-galactopyranoside, and free hemin was not added during fermentation. Following inoculation, cells were grown for 16 h and then harvested by centrifugation. The resulting supernatant was bright red due to soluble hemebound SynHb existing predominately in the ferrous state. Protein purification proceeded as described previously (18), resulting in protein with a Soret/A 280 absorbance ratio greater than 5.0. The purified protein was oxidized with a slight molar excess of potassium ferricyanide followed by desalting on a Sephadex G-25 column in 0.01 M potassium phosphate, pH 7. The resulting ferric SynHb was concentrated to ϳ3 mM and stored at Ϫ80°C until use.
The ferric protein used for crystallization was produced under conditions similar to those in which the His 117 -heme covalent link was observed by NMR experiments in SynHb and the homologue in Synechococcus (28,29). However, the reported method of dithionite treatment leading to the covalent link (27,28) was not used in our treatment of the protein. Crystal growth was achieved by hanging-drop vapor diffusion. Drops were produced by mixing 2 l of 3 mM protein with 2 l of well buffer containing 30 -35% polyethylene glycol monomethylether 5000, 0.2 M ammonium sulfate, 0.01 M cadmium chloride, and 0.1 M MES at pH 6.5. Single crystals grew in 1-3 days at 4°C.
Structure Determination and Refinement-Initially, solution of the crystal structure of SynHb was attempted using molecular replacement starting with the NMR structure of this protein (28). The fact that this method failed is not unusual because the use of NMR structures for molecular replacement starting models is often unsuccessful, even with 100% sequence identity (30,31). Therefore, multiple wavelength anomalous diffraction was employed using anomalous scattering from the heme iron atom. These data were collected at the Stanford Synchrotron Radiation Laboratory (SSRL, Stanford, CA) at 100°K. Diffraction data were processed using the program d*TREK (32) from Rigaku/MSC (peak and inflection data sets required two batches for proper processing of anomalous data). Phases were calculated to 2.0-Å resolution using SOLVE (33), with a starting figure of merit of 0.58. Amino acids were built into the electron density using RESOLVE (34) followed by manual rebuilding in O (35) and refinement using REFMAC5 (36), a program from the CCP4 suite (37). The resolution was extended to 1.8 Å using a native data set collected on a Rigaku/MSC home source generator and further refinement using REFMAC5. The final model contains 123 amino acids and 77 water molecules with R ϭ 21.8%, R free ϭ 22.4%, and acceptably small variations from ideal stereochemistry (Table I). A plot of B factor versus residue number is shown in Fig Calculation of Structural Alignments and Solvent Accessibilities-Structural alignments and r.m.s.d. (root mean square deviation) values were calculated using the Deep View Swiss PDB Viewer (38) and checked with LSQMAN from the Uppsala Software Factory (39). The most representative NMR structure was determined by NMRCLUST (40), and LSQMAN was used to find the average r.m.s.d. among the 20 NMR models relative to the most representative structure. Solvent accessibility was calculated with the program SURFNET (41) using a 1.4-Å probe. All values in Table III and Fig. 6 were calculated with MATLAB (The Math Works, Inc). The least squares plane (42) of the heme was calculated using the 24 atoms of the porphyrin macrocycle, whereas that of the histidine was determined using the 5 atoms of the imidazole ring. The figures were created using the Deep View Swiss PDB Viewer and Igor Pro (Wavemetrics).

RESULTS
The Crystal Structure of SynHb in Comparison to the Solution NMR Structure-The backbone structure and electron density for SynHb are shown in Fig. 1. The overall structure is typical of other trHbs and the NMR structure of SynHb in that the B and E, and G and H helices form the characteristic 2-on-2 helical core of the globin fold (2) (Fig. 1A). The covalent linkage between the N⑀2 atom of His 117 and the heme-2-vinyl is observed as continuous electron density between these atoms (Fig. 1B). There is no evidence for a mixture of states for this side chain, suggesting that the protein is exclusively in the covalently cross-linked conformation. The N⑀2 atom of His 117 is 2.1 Å from the CAB atom of the heme in the crystal structure, whereas these atoms are 5.4 Å apart in the most representative model from NMR. This movement is accompanied by a 136°c hange in the angle of CHI2 and a Ͼ30°increase in the acute angle between the least squares planes of the heme porphyrin and the His 117 imidazole ring.
Another distinct feature of the SynHb crystal structure is the relatively large r.m.s.d. from the NMR structure, as illustrated in Figs. 2 and 3. Two methods were used to align the structures for quantitative comparison. The first method aligns the structures based on the C␣ atoms of 109 residues (89% of the residues in the SynHb structure) with clear backbone density in the crystal structure (residues 5-55, 60 -98, 105, and 107-124). The second method aligns the structures based on the C␣ atoms of 86 residues (70%) involved in clear secondary structure (residues 17-55, 64 -98, and 101-112) and was chosen because it was the method used for aligning the 20 NMR models of SynHb (28). The r.m.s.d. values for the density-based and secondary structure-based alignments were relatively high at 2.18 and 1.6 Å, respectively. The difference in r.m.s.d. values between the two methods arises from the number of residues used for each alignment. Both alignments give an r.m.s.d. of 2.6 Å when all 123 C␣ atoms of the identical sequence are taken into account, rather than only the subset of C␣ atoms used in the alignment. As a frame of reference, the average r.m.s.d. among the 20 NMR models relative to the most representative structure is 1.37 Å for all C␣ atoms. In addition, the comparison between the crystal and NMR structures of SynHb gives a higher r.m.s.d. value than the comparison of the SynHb crystal structure and that of the trHb from Chlamydomonas eugametos (CtrHb). The latter comparison aligns 94 C␣ (76%) with an r.m.s.d. of 1.53 Å, even though CtrHb is in a different liganded state than SynHb and does not share the same primary sequence.
The source of deviation between the crystal and NMR structures is examined on a per residue basis in Fig. 2. Fig. 2A includes a plot of the r.m.s.d. between the SynHb crystal and NMR structures. Fig. 2B is a ribbon representation of the crystal structure backbone onto which r.m.s.d. values have been color-coded for each amino acid (blue is Ͻ1.5 Å, orange 1.5-3 Å, and red Ͼ3 Å). Several regions of increased r.m.s.d. are expected because of increased mobility. These regions include the termini and extended loops, both of which correlate with increased B factors in Fig. 2A. There are also regions described by relatively low B factors that have increased r.m.s.d., including the entire F helix, the first half of the E helix, and a portion of the lower H helix containing the covalently linked His 117 . These differences between the crystal and NMR structures of SynHb are seen clearly in the backbone alignment shown in Fig. 3A.
In the case of hemoglobins, it can be argued that structural alignment based on the heme prosthetic group is the most important functionally, because it gives a direct comparison of the orientation of the amino acid side chains influencing the steric and electrostatic environment immediately surrounding the ligand binding site. Alignment based on heme position rather than the protein backbone increases the deviation between the crystal and NMR structures of SynHb from 2.6 Å to 3.76 Å for all C␣. This indicates that the heme orientations within the crystal and NMR structures of SynHb are different (as can also be seen in Fig. 3A). Both the C␣ and heme-based alignment comparisons indicate significant deviations between the crystal and NMR structures of SynHb, particularly at high resolution.
Inclusion of side chain positions in the backbone comparison of these structures also generates r.m.s.d. values Ͼ 3 Å. Differences occur throughout the backbone but are most pronounced in the heme pocket overlay shown in Fig. 3B, which provides a direct comparison of the crystal structure and the most representative NMR structure. Several side chains, including Tyr 22 and Phe 35 , are shifted upwards, away from the heme in the NMR structure. The H helix is closer to the heme in the crystal structure, which brings His 117 into position to form a covalent bond to the heme 2-vinyl group. Additional high resolution structural differences include the position of the side chains of the proximal and distal histidines, His 70 and His 46 , respectively (Table II), which have different orientations with respect to the heme porphyrins in the two structures. The proximal histidine in the crystal structure is tilted toward the propionates with an acute angle of 79°between the least squares planes of the heme porphyrin and the histidine imid- azole ring, whereas in the NMR structure the proximal histidine is tilted away from the propionates with an acute angle of 77°. The effect of this tilt on the azimuthal angle of the proximal histidine is discussed below. Similarly, the tilts of the distal histidines are nearly opposite; that in the crystal structure tilts toward the 1-and 8-methyl groups with an acute angle of 73°, whereas the one in the NMR structure tilts toward the 4-vinyl and 5-methyl groups with an acute angle of 76°.
Solvent Accessibility in SynHb-The crystal structures of three other group I trHbs have been solved to date, CtrHb, Paramecium caudatum (PtrHb), and Mycobacterium tuberculosis (trHbN) (13,43), all with bound ligands. Each of these structures display a continuous or nearly continuous "ligand tunnel" that connects the distal heme pocket with solvent. The tunnel contains two branches, with openings between the AB and GH inter-helical corners and between the E, G, and H helices (2, 13). Utilizing the same parameters for tunnel calculations in the other trHbs (see "Experimental Procedures"), SynHb does not contain this tunnel. In the SynHb crystal structure, crowding from residues such as Leu 51 , Phe 55 , and Met 98 narrows the space between the AB and GH corners and the distal heme pocket, whereas Phe 50 blocks access to the distal pocket via this path completely. Furthermore, Tyr 53 in SynHb prevents access to the heme from the branch of the tunnel found between the G and H helices. For the NMR structure, it was reported that this tunnel path was detected (28). However, when the same method was again used to check for solvent accessibility in the NMR structure, it was found that, despite the porous surface of the protein, no continuous tunnels extend through the protein matrix to the distal heme pocket.
Both the crystal and NMR structures of SynHb are solventaccessible around the heme propionates. The solvent-accessible cavity of the crystal structure (Fig. 4A) is smaller than that of the NMR structure (Fig. 4B), which is more extensive and reaches into the heme pocket to surround the distal His 46 .  However, the rest of the heme group is blocked from solvent as is the case in other trHbs. For example, in the crystal structures of CtrHb and PtrHb, the side chains of Phe 48 and Trp 59 block the heme from solvent accessibility at the CHD methinic bridge (13). In the crystal structure of SynHb, Tyr 53 and Tyr 61 guard this side of the heme from solvent. In the NMR structure of SynHb, Tyr 53 also serves this function but Tyr 61 does not due to the wide swing of the pre-F loop in this conformation (28).
Comparison of SynHb to the Structures of Other trHbs-SynHb is the first crystal structure of a trHb in the unliganded state. Of the three crystal structures previously solved for other group I trHbs, SynHb shares highest sequence identity with CtrHb (34%) and somewhat lower sequence identities with trHbN (24%) and PtrHb (20%) (25). The overlay of SynHb and CtrHb is shown in Fig. 5A. SynHb shares a 2-on-2 ␣-helical fold more similar to that of CtrHb and PtrHb than trHbN (respective r.m.s.d. values are 1.53 Å over 94 C␣, 1.65 Å over 85 C␣, and 2.16 Å over 51 C␣). All four of these trHbs exhibit the characteristic three glycine motifs, the shortened one-turn A helix (though trHbN also has an N-terminal extension), and significant deletions in the CD-D region.
General differences between the SynHb crystal structure and other trHbs include a shift of the SynHb B helix away from the heme group, and movement of the E helix to a position closer to the heme that facilitates the bond between His 46 and the distal site of the heme iron. The F helices in other trHb crystal structures contain a wide "pre-F" loop preceding a oneturn F helix hosting the proximal His(F8) ("F8" designates the helix position of the proximal His; see Table II). In contrast, the pre-F loop in SynHb is shorter and closer to the heme due to more extensive helical structure in this region. The F helix in the crystal structure of SynHb begins with one and a half turns (residues F1-F6 as opposed to residues F4 -F8 in the others), is interrupted by a three residue loop hosting the His(F8), and ends in a one-turn loop of four residues. This is different from the NMR structure of SynHb, in which the pre-F loop swings out in the opposite direction from the three ligand bound structures, and the F helix is one long, 13-residue helix beginning at F2. Finally, the H helix in SynHb is interrupted and bent by a loop (similar to that in PtrHb) that brings the end of the H helix close to the heme at the His 117 covalent linkage.
Differences in the locations of specific, functionally important heme pocket amino acids in the crystal structure of SynHb compared with those of the other trHb proteins can be summarized as follows. 1) His(F8), Phe(CD1), and Tyr(B10) are conserved in all four structures (Table II). The Tyr(B10) side chain is found in the distal heme pocket and acts to stabilize the ligand in the other trHbs (13,(43)(44)(45) but extrudes out into the solvent in the SynHb structure. 2) Gln(E7) in CtrHb and PtrHb also stabilizes ligand binding (13), but in SynHb, Gln(E7) extrudes out into the solvent, up and away from the heme, and forms hydrogen bonds with two solvent molecules. 3) Thr(E11) in PtrHb and Gln(E11) in both CtrHb and trHbN participate in a network of hydrogen bonds that includes the side-chain hydroxyl of Tyr(B10) in each protein (13,43). Gln(E11) in SynHb is located well above the heme and hydrogen bonds to a solvent molecule. 4) In the other trHb structures, the Lys(E10) side chain extrudes out into the solvent, interacting with the heme propionates in CtrHb and PtrHb (13). The corresponding His(E10) side chain in SynHb binds the distal site of the heme iron. In summary, besides hexacoordination, the major difference among the heme pockets of these structures is that several side chains important for stabilizing the bound ligand in other trHbs are found interacting with solvent in SynHb.
Comparison of SynHb with Other Hexacoordinate Hemoglobins-The structures of hemoglobins in the hemichrome state from five other organisms have been reported. These are the Hb-C chain from the sea cucumber Caudina arenicola (Hb-C) (46), rice non-symbiotic Hb (RiceHb) (12), a tetrameric hemoglobin from the Antarctic fish Trematomus newnesi (HbTn) (47) (with the ␤ chains in a hemichrome state), horse methemoglobin (eHb) at pH 5.4 (with the ␣ chains in a hemichrome state) (48), and human neuroglobin (Ngb) with three Cys mutations (11). Backbone overlays of these hxHbs (none of which are trHbs) do not align well with SynHb. However, SynHb looks most like a truncated version of RiceHb, with an r.m.s.d. of 1.71 Å over 89 C␣ atoms (Fig. 5B).
In the other hxHbs the hexacoordinating side chain is the distal His(E7), but in the two structures of SynHb the hexacoordinating distal residue is His(E10). Heme pocket comparisons reveal further interesting differences in stereochemistry among these hxHbs ( Fig. 6 and Table III). For example, Fig. 6 shows the azimuthal angles for the above hxHbs and the two SynHb structures. Myoglobin (Mb) (49) and leghemoglobin (Lba) (50) are also shown, along with four cytochrome b 5 proteins; bovine (bCYT) and rat (rCYT) cytochrome b 5 (51,52), and the cytochrome b 5 domains from human (HSO) and chicken liver (CSO) sulfite oxidases (53,54).
The azimuthal angle is defined as the intersection of the least squares plane of the proximal histidine with the least squares heme plane, relative to a line connecting diagonally opposed pyrrole nitrogens. The planar intersections are presented as colored lines through a representative heme molecule in Fig. 6A. The intersection for Mb (light cyan) eclipses the pyrrole nitrogens, whereas Hb-C (dark cyan) nearly eclipses the opposite pyrrole nitrogens, and Lba (orange), RiceHb (purple), and Ngb (yellow), are staggered to various degrees. However, the azimuthal angles for eHb (pink), HbTn (gray), the crystal structure of SynHb (red), and the NMR structure (blue) are staggered but do not pass through the iron atom, as shown in Fig. 6B. This is due to the tilt of the proximal histidine imidazole plane. The tilts for all of the proteins in Fig. 6A, as delineated in the first column of Table III, are within about 2°o f perpendicular to the heme plane. However, the proteins in Fig. 6B have tilts of 7°to 14°from the normal to the heme whereas Hb-C (dark cyan) nearly eclipses the opposite pyrrole nitrogens. Lba (orange), RiceHb (purple), and Ngb (yellow) are staggered to various degrees. B, the crystal (red) and NMR (blue) structures of SynHb, eHb (pink), and HbTn (gray) are also staggered, but the intersections of their planes do not pass through the iron atom due to proximal histidine tilts. C, All four cytochrome b 5 proteins are staggered, with CSO (dark green) and bCYT (dark blue) slightly off-center compared with HSO (light green) and rCYT (light blue), also due to slight proximal histidine tilts.
plane. Fig. 6C shows that all four of the cytochrome b 5 proximal histidines are also staggered. Again, plane intersections for CSO (dark green) and bCYT (dark blue) are somewhat off center due to small proximal histidine tilts, as opposed to HSO (light green) and rCYT (light blue).

DISCUSSION
The crystal structure of SynHb does not fall within the deviation of the NMR models and therefore reveals a separate conformation for this protein. Along with the difference in structural methods used, the covalent link between His 117 and the heme prosthetic group is a likely cause for these conformational changes. Although the function of this covalent linkage is not yet understood, its structural implications are evident from a comparison of the crystal and NMR structures. Potential functional significance resulting from the covalent linkage includes inward movement of helices and heme pocket residues toward the heme resulting in decreased solvent accessibility and increased protein stability due to prevention of heme loss (27).
Implications for Ligand Binding in trHbs-SynHb is similar in overall fold to other trHbs with the obvious exception of intramolecular heme iron coordination. This is likely to account for many structural differences with other trHbs, such as the downward shift of the B helix that moves Tyr(B10) into position to stabilize the ligand in other trHbs. Because Tyr(B10) in SynHb is also predicted to stabilize bound ligands (24), a similar downward shift of the B helix is expected upon ligand binding in SynHb. His 46 is predicted to be uninvolved in ligand stabilization (24,25). This suggests that ligand binding in SynHb will cause a shift in the position of the E helix that moves this side chain outwards into the solvent. This shift could also move the Glu(E7) and Glu(E11) residues into position to contribute to a hydrogen bonding network around the bound ligand as seen in the ligand bound structures of other trHbs (2).
Although these structural changes would cause ligand-bound SynHb to resemble other ligand-bound trHbs, the ligand-free structures of CtrHb, PtrHb, and trHbN must be different than that of SynHb. Of these three, only CtrHb is capable of hexacoordination, which occurs only at alkaline pH by the Tyr(B10) side chain (55). This suggests that the B helix rather than the E helix in CtrHb moves closer to the heme in the absence of exogenous ligands. Structures of PtrHb and trHbN in the pentacoordinate state would be different from both SynHb and CtrHb in that neither exhibits hexacoordination.
The conventional "histidine-gate" path for ligand binding is blocked in trHbs (2). It has been proposed that the tunnel found in the previous trHb structures serves as an alternative diffusion path for ligands (43). However, there is no protein matrix tunnel in SynHb. Therefore, three possibilities exist for ligand entry and exit in SynHb. 1) Ligands enter and exit through the solvent face of the heme pocket. 2) Ligands enter through the solvent face of the heme pocket, and then a tunnel forms in the ligand bound state that serves as a route for exit. 3) Ligand entry and exit pathways are not evident in these structures. The solvent accessibility around the propionates would be an appropriate route for ligand access in support of the first two possibilities. In the second case, formation of a tunnel after ligand binding followed by its blockage when the ligand dissociates could serve to trap ligands near the heme pocket, providing a barrier to ligand escape, and enhancing geminate ligand recombination. This possibility is supported by the observation of room temperature CO geminate recombination in SynHb (24,45).
Implications for Ligand Binding in hxHbs-The data in Table III show some of the stereochemical factors likely to contribute to ligand affinity in hxHbs. In general, it is thought that staggering of the azimuthal angle of the proximal histidine away from the heme pyrrole nitrogens increases ligand affinity by allowing the iron to move into the plane of the porphyrin ring (56,57). This holds true for SynHb, RiceHb, and Lba, which are all staggered and which all have relatively high oxygen affinities compared with Mb (17,22,58).
Columns 4 and 5 in Table III indicate differences in Fe-His(N⑀2) bond lengths, with the cytochromes in most cases exhibiting shorter bond lengths between the heme iron and the proximal and distal histidines. The longer bond lengths found in the hxHbs decrease the bond strength and are therefore likely to be a factor enabling distal histidine dissociation and subsequent ligand binding in the hxHbs. Bond strength is also influenced by the degree of histidine tilt (with larger tilts decreasing the strength of the bond) (48). The hxHbs also have generally larger tilt angles for both the proximal and distal histidines compared with cytochrome b 5 , which is likely to be another element contributing to their ability to bind ligands. CONCLUSIONS Gross features of the crystal structure of SynHb are similar to the NMR structure, but many specific differences are also evident. Many of these differences could be due to the covalent linkage between His 117 and the heme-2-vinyl group that is present in the crystal structure but not in the NMR structure. In comparison to other trHbs, hexacoordination in SynHb forces many side chains important for ligand stabilization to reside in contact with solvent in the absence of exogenous a Planes are defined as the least squares plane, determined using the 24-atom porphyrin macrocycle of the heme and the 5-atom imidazole ring of the histidine. The angle given is the acute angle between the two planes.
b Negative distance indicates that the iron is on the proximal side of the heme.
ligands. Hexacoordination in SynHb results from a different tertiary arrangement of the B and E helices compared with other hxHbs, as well as many specific differences in heme pocket stereochemistry, leading to alternative mechanisms for regulating ligand binding. Examination of SynHb in the context of other hxHbs and cytochrome b 5 reveals that destabilization of the distal His-heme iron bond to allow for exogenous ligand binding could be indicated by increased bond lengths and tilt angles characterizing intramolecular coordination in the hxHbs.