Solution 1H NMR Investigation of the Active Site Molecular and Electronic Structures of Substrate-bound, Cyanide-inhibited HmuO, a Bacterial Heme Oxygenase fromCorynebacterium diphtheriae * 210

The molecular structure and dynamic properties of the active site environment of HmuO, a heme oxygenase (HO) from the pathogenic bacterium Corynebacterium diphtheriae, have been investigated by 1H NMR spectroscopy using the human HO (hHO) complex as a homology model. It is demonstrated that not only the spatial contacts among residues and between residues and heme, but the magnetic axes that can be related to the direction and magnitude of the steric tilt of the FeCN unit are strongly conserved in the two HO complexes. The results indicate that very similar contributions of steric blockage of several meso positions and steric tilt of the attacking ligand are operative. A distal H-bond network that involves numerous very strong H-bonds and immobilized water molecules is identified in HmuO that is analogous to that previously identified in hHO (Li, Y., Syvitski, R. T., Auclair, K., Wilks, A., Ortiz de Montellano, P. R., and La Mar, G. N. (2002) J. Biol. Chem. 277, 33018–33031). The NMR results are completely consistent with the very recent crystal structure of the HmuO·substrate complex. The H-bond network/ordered water molecules are proposed to orient the distal water molecule near the catalytically key Asp136 (Asp140 in hHO) that stabilizes the hydroperoxy intermediate. The dynamic stability of this H-bond network in HmuO is significantly greater than in hHO and may account for the slower catalytic rate in bacterial HO compared with mammalian HO.

ferryl intermediate, the reactive form of HO is a ferric hydroperoxy intermediate (2)(3)(4). In mammals, the ϳ300-residue membrane-bound enzyme occurs as an inducible HO-1, whose primary roles are iron homeostasis and heme catabolism (5,6), whereas the constitutive HO-2 has been proposed (7) to generate CO as a neural messenger. In higher plants, algae, and cyanobacteria, HO generates the open tetrapyrroles as lightharvesting pigments (8). HO has also been identified in several pathogenic bacteria, where its role appears to be the essential "mining" of iron from hemes in the host (9,10). Plant and bacterial HOs are soluble and somewhat shorter (ϳ200 residues) (9, 10) than mammalian HO (11). Among the characterized bacterial HOs, sequence homology to the more extensively studied mammalian HO varies from relative high (33% sequence identity/70% similarity) for HmuO from Corynebacterium diphtheriae (10) to low (Ͻ25%) for HemO from Neisseria meningitides (9).
The remarkable recent progress in understanding the functional properties of HO based on mutagenesis and spectroscopic studies (3,4,(12)(13)(14), of a slightly truncated, soluble, and completely active recombinant mammalian HO, has been considerably enhanced by the successful x-ray crystallographic characterization of the substrate complexes of first human HO (hHO), followed by rat HO (15,16). These structures shed light on a key determinant of the ␣-stereoselectivity, in that the distal helix covers the heme so as to sterically completely block access to the ␤and ␦-meso positions and partially block access to the ␥-meso positions (15)(16)(17). Although no distal residue that would stabilize the hydroperoxy unit could be identified, the occurrence in the crystal of a localized water molecule H-bonded to the distal helix Asp 140 carboxylate, together with the observation that mutating Asp 140 to a non-anionic side chain abolishes HO activity (12,14), has led to the proposal that the water molecule may be sufficiently stabilized in its crystallographically defined position to serve as the weak Hbond donor to stabilize the hydroperoxy unit.
Solution 1 H NMR characterization of hHO and its substrate complex has contributed to the understanding of the structure/ function relationship of HO (18 -22). An annoying, but functionally irrelevant property of the mammalian HOs is that binding of the native substrate, protohemin (PH; R ϭ vinyl in Fig. 1), leads to ϳ1:1 orientational isomerism about the ␣/␥axis (18 -20), which leads to spectral congestion and limits both the range and reliability of structural characterization. Nevertheless, the pattern of dipolar shifts for the protons on the proximal helix allowed determination of the orientation of the major magnetic axis, which could be correlated with a ϳ20°tilt of the FeCN in the direction of the ␣-meso position (19,20). The orientation of ligated azide in the rat HO⅐heme⅐N 3 complex confirms such a steric influence (23).
Thus, both the position of the distal helix in blocking access to other meso positions (15,16) and its influence on tilting (19 -21, 24) the axial ligand toward the ␣-meso position contribute to the stereoselectivity of the reaction. Two-dimensional 1 H NMR of a hHO complex with the 2-fold symmetric substrate 2,4-dimethyldeuterohemin (DMDH; R ϭ CH 3 in Fig. 1) (25) allowed sufficiently definitive and extensive assignments to identify (21) an unusual distal H-bond networks involving some extremely strong H-bonds (labile proton shifts between 17 and 10 ppm) whose acceptor could be identified in the hHO⅐PH⅐H 2 O crystal structure (15). Moreover, it was demonstrated that water molecules were in the immediate vicinity (ϳ3 Å) of each of the strong H-bond donors (22). The strongest of these H-bonds is between a conserved Tyr 58 serving as a donor to the catalytically critical Asp 140 . We proposed that this network has, as one of its primary roles, the stabilization of the Asp 140 side chain and the H-bonded water molecules, one of which can interact with the heme ligand (4,21).
We report herein on the extension of our 1 H NMR investigation to HmuO, the 216-residue soluble bacterial HO from C. diphtheriae (10), using hHO as a homology model (21). Functional (26,27) and spectroscopic (27,28) studies, as well as mutagenesis (29,30), have confirmed the same mechanism and stereospecificity as for mammalian HOs, although the turnover rate is slower (27); the enzyme has been crystallized (31), and the structure of the substrate hemin complex has been refined to 1.4-Å resolution. 2 Our interests are to establish the degree to which the available extensive NMR data on hHO⅐DMDH⅐CN (19 -22) and the crystal structure of hHO⅐PH⅐H 2 O (15) can be used to assign the resonances and to structurally interpret the NMR spectral parameters (32) of HmuO⅐PH⅐CN in terms of the orientation of the FeCN vector and the presence or absence of a distal H-bond network similar to that reported for hHO⅐DMDH⅐CN (21,22). This 216-residue soluble HmuO enzyme has His 20 as its axial ligand (30) and exhibits extensive sequence homology to the distal helix and the four fragments of HO shown (21) to participate in the H-bond network in hHO⅐DMDH⅐CN (Fig. 2). To provide a broader comparison with the NMR data on hHO complexes (19 -22), we explore in parallel both the disordered HmuO⅐PH⅐CN complex and the homogeneous HmuO⅐DMDH⅐CN complex to show that this bacterial HO exhibits remarkable conservation of the distal steric effects on the axial ligand and distal H-bond network relative to hHO.

EXPERIMENTAL PROCEDURES
Protein Sample-HmuO was expressed and purified as reported previously (27). PH was purchased from Sigma. 2,4-Dimethyldeuteroporphyrin was purchased from Mid-Century Chemicals, and the iron was incorporated to yield DMDH by standard procedures (25). PH and DMDH were titrated into apo-hHO to a 1:1 stoichiometry in the presence of a 10-fold molar excess of KCN in a 90% H 2 O and 10% 2 H 2 O solution buffered at pH 7.4 with 100 mM phosphate. The final concentrations of the HmuO⅐substrate⅐CN complexes were ϳ1.5 mM.
NMR Spectroscopy-1 H NMR data were collected on a Bruker AVANCE 600 spectrometer operating at 600 MHz. Reference spectra were collected in both 1 H 2 O and 2 H 2 O over a temperature range of 10 -40°C at a repetition rate of 1 s Ϫ1 using a standard one-pulse sequence with saturation of the water solvent signal. Chemical shifts are referenced to 2,2-dimethyl-2-silapentane-5-sulfonate (DSS) through the water resonance calibrated at each temperature. Nonselective T 1 values were determined in both 1 H 2 O and 2 H 2 O at 20, 25, and 30°C from the initial magnetization recovery of a standard inversion-recovery pulse sequence. The distance of proton H i from the iron, R H i , was estimated from the relation R H i ϭ R* Fe (T 1 */T 1i ) 1/6 , using the heme for the ␣-meso-H for H* (R* Fe ϭ 4.6 Å and T 1 * ϭ 50 ms) as reference (20,21,32). Steady-state NOEs from HmuO⅐DMDH⅐CN in 1 H 2 O were recorded with and without saturation of the solvent resonance for 300 ns using 3:9:19 detection (33). NOESY spectra (mixing time of 40 ms, 10 -40°C) (34) and Clean-TOCSY spectra (25, 35°C; spin lock of 15 and 30 ms) (35) using MLEV-17 (36) were recorded over a bandwidth of 14 kHz (or 28 kHz) (NOESY) and 14 kHz (TOCSY) with recycle times of 1 s (or 0.33 s) using 512 t1 blocks of 128 and 250 scans, each consisting of 2048 t2 points. Two-dimensional data sets were processed using Bruker XWIN software on a Silicon Graphics Indigo work station and consisted of 30°sine-squared bell apodization in both dimensions and zero filling to 2048 ϫ 2048 data points prior to Fourier transformation.
Magnetic Axes-The magnetic axes ( Fig. 1) were determined by a least-squares search for the minimum in the error function (21,32,37) (Equation 1).
␣, ␤, and ␥ are the Euler angles that rotate the reference system (xЈ,yЈ,zЈ) into the magnetic coordinate system, x,y,z (R,,⍀), where ␤ reflects the tilt of the major magnetic axis (z) from the heme normal; ␣ defines the projection of z onto the xЈ,yЈ plane (tilt direction); and ϳ ␣ ϩ ␥ locates the rhombic axes, as shown in Fig. 1 The symbols ␦ tetr , ␦ sec , and ␦ rc are the chemical shifts of an unfolded tetrapeptide (38) relative to DSS and the effect of secondary structure (39) and ring currents (40) on the shift, respectively.

RESULTS
The initially assembled HmuO⅐PH⅐CN complex exhibits two sets of hyperfine shifted resonances (data not shown; see Sup- The orientation of the axial His 20 ring plane is shown as a rectangle. The magnetic coordinate system, x,y,z, is related to the iron-centered reference coordinate system, xЈ,yЈ,zЈ, by the Euler angles ⌫(␣,␤,␥), where ␤ is the tilt of the major magnetic axis (z) from the heme normal (zЈ), ␣ describes the direction of the tilt in the angle between the projection of z on the zЈ,yЈ axes and the xЈ axis, and the rhombic axes (x,y) are related to the reference xЈ,yЈ axis by the angle ϳ ␣ ϩ ␥.
plemental Material), one set of which loses intensity over several days to yield an M i :m i or H j :h j isomer equilibrium ratio of ϳ3:1 (M i and H i represent a methyl and single proton of the major or only equilibrium species, respectively; and m i and h j reflect a methyl and hydrogen of the minor equilibrium species, respectively), as illustrated in Fig. 3B. The subscript i refers to heme pyrrole substituent positions 1-8, heme ␣/␦-meso positions, or the residue number and proton position. Hence, the substrate is initially bound disordered about the ␣/␥-meso axis and, like mammalian HO complexes, equilibrates to a ϳ3:1 ratio with the more stable heme orientation depicted in Fig. 1 (18 -20). This heterogeneity is absent in the complex with the symmetric DMDH substrate (Fig. 3C), as found previously for hHO⅐DMDH⅐CN (21). We will concern ourselves further only with the major isomer of HmuO⅐PH⅐CN and the single species HmuO⅐DMDH⅐CN.
The resolved portions of the 600-MHz 1 H NMR spectra of equilibrated hHO⅐PH⅐CN and HmuO⅐PH⅐CN (19,20) are compared in Fig. 3 (A and B, respectively). Similarly, the traces of HmuO⅐DMDH⅐CN and hHO⅐DMDH⅐CN (21) are compared in Fig. 3 (C and D, respectively). The very close similarity of the pattern of resolved resonances in the hHO and HmuO complexes is quite apparent. The homologous assignments are connected by dashed lines between the two hHO and the two HmuO complexes. The crowded region between 10 and 15 ppm for HmuO⅐DMDH⅐CN in 1 H 2 O is expanded in Fig. 4A. The relevant homologous portions of the amino acid sequences for the two HOs are illustrated in Fig. 2. The nonselective T 1 values for well resolved peaks of interest in hHO⅐DMDH⅐CN and HmuO⅐DMDH⅐CN (as well as in the PH complexes) are the same. In particular, the low-field labile proton peaks Gly 139 NH and upfield Ser 138 C ␤1 H and C ␤2 H exhibit T 1 values indistinguishable from those of Gly 143 NH and Ser 142 C ␤1 H and C ␤2 H (ϳ50, 85, and 50 ms, respectively) reported previously (20,21).
Comparison in Fig. 3 of the NMR spectra of the complexes of the two HOs shows that the patterns of shifts are so similar in the two proteins that it is highly advantageous to pursue assignments on the basis of the comprehensive and definitive assignments previously reported for hHO complexes (20 -22). Hence, two-dimensional NMR data are presented only to define an important distal H-bond network/aromatic cluster as just recently characterized in hHO. We initially (and trivially) as-sign the heme, followed by locating hyperfine shifted protons that arise from TOCSY-detected side chains placed on sequentially assigned backbone via the standard , and/or ␣ i -␤ iϩ3 NOESY connectivities characteristic of helices (41) as described for hHO (20,21).
Heme Assignments-Dipolar contacts were observed in a set of pyrrole substituents on both HmuO⅐PH⅐CN and HmuO⅐DMDH⅐CN (data not shown) that are identical to those reported in detail for the analogous hHO complexes (21). The remarkably similar hyperfine shifts for a given substituent in the HmuO and hHO complexes are in evidence in the data provided in Table I. The essentially identical shift pattern for the major isomer of the PH substrate is evidence for the same heme orientation in HmuO and hHO (see below).
Proximal and Distal Helices-Standard backbone NOESY connectivities (data not shown; summarized in Fig. 2) among TOCSY-detected side chains locate two helical fragments (I and II) for which numerous side chains exhibit moderate-to-large hyperfine shifts. Fragment I is four spins; AMX ϭ three-spin system), which is unique for Ala 17 -Glu 24 on the proximal helix. Consistent with the assignments are the large low-field contact shift for AMX iϩ3 of the axial His 20 (His 25 in hHO), the low-field dipolar shifts for Ala 17  hHO (20,21), indicates that Phe 201 is slightly shifted away from Ala 23 . Finally, note that 3-CH 3 in HmuO⅐PH⅐CN exhibits weak NOESY cross-peaks to Phe 208 (which also exhibits strong NOESY cross-peaks to the 2-vinyl group) (data not shown), whereas the small Phe 214 cross-peak to 3-CH 3 is not seen in hHO⅐PH⅐CN (20, 21) (but the strong 2-vinyl cross-peak to Phe 214 is observed). This indicates that the conserved Phe 208 / Phe 214 at the pyrrole A/B junction is slightly closer to position 2 in HmuO than in hHO. Moreover, NOESY cross-peaks of the labile protons for NH 2 of Gln 38 in hHO ( Fig. 5B) (20,21) are absent in the HmuO complex (Fig. 5A), but strong contacts with some aliphatic protons are present as expected because of the Gln 38 3 Leu 33 replacement in HmuO. The chemical shifts of the two HmuO complexes, as well as of the two hHO complexes (20,21), for these assigned residues are compared in Table II, where we also include the predicted dipolar shifts for the residues in the hHO complex. The observed inter-residue and heme-residue dipolar contacts are summarized schematically in Fig. 6.
The NOESY and TOCSY data (data not shown; summarized in Fig. 2) indicate that helical fragment II is represented by (Fig.  2), where AMX iϩ2 is in contact with a two-spin aromatic ring; Val i , Leu iϩ3 , and AMX iϩ7 exhibit moderate-to-large high-field dipolar shifts; and Gly iϩ8 exhibits strong low-field dipolar shifts. Both the sequence and the dipolar shift pattern identify (20,21) this as a key portion of the distal helix Val 131 -Gly 139 (analogous to Thr 135 -Gly 143 in hHO). As shown in the slices through 8-CH 3 in HmuO and hHO (Fig. 7, A and B), the strong contact with C ␣ H of Val 131 is conserved (relative to C ␣ H of Thr 135 ). However, the weak contacts between 8-CH 3 and NH of the adjacent conserved Leu 134 and Gly 135 in HmuO (residues 138 and 139 in hHO) observed in the hHO complex ( Fig. 7B) (20,21) are not detectable in HmuO (Fig. 7A) and indicate a small movement of the distal helix near its kink away from 8-CH 3 in HmuO relative to hHO. Finally, slices through the similarly relaxed (T 1 ϳ 85 ms) Ser 138 C ␤1 in HmuO (Fig. 7C) and Ser 142 in hHO (Fig. 7D) indicate that the Ser is slightly farther from heme 6-H ␣s in HmuO relative to the hHO complex. The chemical shifts for helix II residues, together with data from hHO complexes (20,21), are listed in Table II. The observed inter-residue and heme-residue contacts are illustrated schematically in Fig. 6.
H-bond Network/Aromatic Cluster-The HmuO complexes, like the hHO complexes (21,22), exhibit a set of strongly low-field shifted labile proton peaks (Figs. 3 and 4A), which (with the unique exception of HmuO Gly 139 NH/hHO Gly 143 NH) exhibit negligible paramagnetic relaxation, so their strong low-field bias must be attributed to strong hydrogen bonds (42). Three sequential fragments are easily recognized (summarized in Fig. 2) by their remarkably similar arrangements compared with the three characterized fragments labeled IV-VI in hHO⅐DMDH⅐CN (21). The analogous fragment III could not be recognized as easily in the HmuO complex (but see "Discussion"). The helical fragment IV, Z i -Gly iϩ1 -AMX iϩ2 -AMX iϩ3 , exhibits strong low-field NH shifts for Gly iϩ1 and AMX iϩ2 (Fig. 8, B and D; summarized in Fig. 2), as found for Ala 165 -Phe 166 in hHO (21). In agreement with the assignment of Leu 159 -Tyr 161 to fragment IV, a three-spin TOCSY ring makes contact with AMX iϩ2 (Fig. 9D), and two spins of a three-spin aromatic ring make contact with AMX iϩ3 (Fig. 9C), which must arise from Tyr 161 . The three TOCSY/NOESY peaks of the AMX iϩ3 side chain exhibit the unusual pattern (Fig. 9, B-D) that one crosspeak becomes narrower, whereas the other broadens as the temperature is elevated. This behavior is consistent with a Tyr ring that reorients sufficiently slowly to resolve the individual C ␦ H or C ⑀ H, but leaves the two C ⑀ H shifts averaged. The NOESY cross-peak to its own backbone, as well as the crosspeak to a new low-field shifted labile proton with no TOCSY connectivity, identifies Tyr 161 OH (Fig. 9D). Tyr 161 C ⑀ H exhibits a weak-to-moderate intensity NOESY cross-peak to NH of Gly 139 (data not shown; as also observed for the homologous Phe 166 3 Gly 143 in hHO) (20,21). A strong dipolar contact of Phe 166 NH with a Met 51 methyl in hHO (21) is lost in HmuO, but is replaced by a dipolar contact of Phe 160 NH with an NH 2 group (data not shown), which sequence comparison identifies as Gln 46 . A strong contact between the Tyr 161 ring and a low-field, dipolar-shifted three-spin aromatic group (Fig. 9C) is analogous to the Phe 167 -Phe 47 contact in hHO (21) and is readily rationalized by the homologous Tyr 161 -Phe 42 contact observed here. This assignment is consistent with the observation (data not shown) of a Gln 46 NH 2 contact with Phe 42 . The dipolar contact for this fragment with the distal helix (II) and Phe 42 /Gln 46 are summarized schematically in Fig. 6.
The extreme low-field labile proton exhibits no TOCSY peak, but displays strong NOESY cross-peaks to a two-spin aromatic ring (Fig. 9E) whose AMX backbone is readily identified. N i -N iϩ1 and ␤-N iϩ1 ( Fig. 8A; summarized in Fig. 2) locate the adjacent residue as Thr, which identifies the Tyr 53 -Thr 54 segment with shifts and dipolar contacts analogous to fragment V in hHO (21). The Tyr 53 ring exhibits the NOESY cross-peak to the Phe 160 ring (Fig. 9B) that was observed between the homologous Tyr 58 or fragment V and Phe 166 or fragment VI in hHO (21). The Tyr 53 ring exhibits NOESY cross-peaks to N ⑀ H of Arg 132 and OH of Tyr 133 (data not shown) on the distal helix II in the same fashion as observed for the homologous residues in hHO (schematically shown in Fig. 6) (21). The dipolar con-tacts of fragment IV are summarized in Fig. 6.
Finally, the other two extreme low-field peptide NH groups are part of a 5-residue fragment, Z i -Ala iϩ1 -Z iϩ2 -Val iϩ3 -Z iϩ4 (Figs. 2 and 8, A, B, and D), which the sequence identifies as Arg 79 -Leu 83 , with the low-field NH groups occurring from the homologous residues i (Arg 79 ) and iϩ1 (Ala 80 ) in HmuO (as observed for Arg 85 and Lys 86 in hHO) (21). The contacts between fragment VI and the distal helix II and fragment V, viz. Asn 78 and Arg 79 to Tyr 133 and Tyr 53 (Fig. 6), expected on the basis of the contact in hHO, are clearly observed, as summarized in Fig. 6. Attempts to assign the fragment in HmuO analogous to fragment III on the basis of the NMR studies on hHO⅐DMDH⅐CN (21) failed, although the 1 H NMR data available can be used to infer that the fragment is similarly highly conserved in HmuO relative to hHO (see below). The peak at 14.1 ppm exhibits equally intense cross-peaks to two non-labile protons in the aromatic window indicative of a His ring N ⑀ H, which, by analogy to hHO (19,22), is His 128 ; the expected strong NOESY cross-peak to Ala 200 is observed (data not shown).
The low-field peak at 11.4 ppm exhibits properties consistent with its arising from N ⑀ H of Trp 50 (which replaces Tyr 55 in hHO) in NOESY cross-peaks to a TOCSY-detected (only two cross-peaks are resolved) aromatic ring and a weak NOESY cross-peak of the ring (Trp 50 ring) to the rings of both Phe 160 and Tyr 161 (data not shown). Hence, we tentatively label it N ⑀ of Trp 50 . It should be noted that there remains one strongly low-field shifted NH (12.2 ppm; labeled a in Fig. 4A) that cannot be assigned at this time and that has no analog in hHO (21).
Acceptors for Strong H-bond Donors-Having demonstrated for HmuO a remarkably conserved arrangement on the distal side of the heme for three of the four fragments involved in the H-bond network/aromatic cluster in hHO (21) (21), could not be assigned in HmuO, the sequence homology suggests that the Asp 86 carboxylate (homologous to Asp 92 in hHO) (Fig. 2) is the acceptor for the NH groups of Gly 159 and Phe 160 , as depicted in Fig. 6 Table III.
Labile Proton Exchange-Comparison of the resolved lowfield 1 H NMR trace of HmuO⅐DMDH⅐CN in 1 H 2 O in Fig. 4A with that of the complex 20 min (Fig. 4B) and 4 h (Fig. 4C) 4A) and with (Fig. 4BЈ) saturation of the bulk water resonance shows significant magnetization transfer to the lowfield peaks (22,43), as shown in the difference trace in Fig. 4CЈ. The magnetization transfer to the four labile protons shown to exchange slowly with water must arise from NOEs between these labile protons and "immobilized" water molecules (43), as we previously observed for hHO⅐DMDH⅐CN (22). The magni-  tude of the NOEs is ϳ10% for His 128 N ⑀ H and ϳ25% for Phe 160 NH, Trp 50 N ⑀ H, and peak a. For the other peaks that exhibit magnetization transfer from water, it is not possible at this time to differentiate between chemical exchange and NOEs as the origin of the magnetization transfer (43).

4, for 90% H 2 O and 10% 2 H 2 O (A) and 20 min (B) and 4 h (C) after converting from 90% 1 H 2 O and
Magnetic Axes and Cyanide Tilt-The completely conserved pattern of large dipolar shifts for proximal helix residues (Table II) and conserved contacts with the heme (Fig. 6) in HmuO relative to the hHO complexes can arise only if HmuO⅐PH⅐CN and hHO⅐PH⅐CN possess very similar orientation for the major magnetic axis (32). A direct determination of the magnetic axes for HmuO⅐PH⅐CN using the present 1 H NMR data and the recently available HmuO crystal coordinates 2 leads to ␣ ϭ 234 Ϯ 16, ␤ ϭ 18 Ϯ 2, and ϭ 47 Ϯ 12, which can be compared with reported values of ␣ ϭ 234 Ϯ 12, ␤ ϭ 20 Ϯ 3, and ϭ 25 Ϯ 13 for hHO⅐DMDH⅐CN (21). The   magnetic axes for both complexes yield similarly excellent correlation between the ␦ dip(obs) and ␦ dip(calc) for the input proximal side residues (data not shown; see Supplemental Material). In each case, z is tilted ϳ20°toward the ␥-meso position. The z direction is oriented toward the proximal side (20,32); hence, the FeCN vector (Ϫz direction) is tilted toward the ␣-meso position. Therefore, a direct contribution to stereoselectivity from the tilt due to direct distal steric interactions with the ligand in the direction of the ␣-meso position is operative in both hHO and HmuO.

DISCUSSION
PH Orientation-The orientation of PH in both hHO (19,20) and HmuO complexes is similarly rotationally disordered about the ␣/␥-meso axis in the initially formed complex, with very similar ϳ3:1 ratios at equilibrium and with the same heme orientation dominating in each complex in solution. Notably, the heme orientation found in the HmuO⅐PH⅐H 2 O crystal structure 2 is the same as the dominant isomer in solution, whereas that in the hHO⅐PH⅐H 2 O crystal structure (15) is the minor form in solution (19,20). The resolution upon using DMDH rather than PH for HmuO is less dramatic than for hHO because, in contrast to hHO (20,21), there are no detectable changes in the intrinsic line width of the signals in the DMDH relative to the PH complexes of HmuO, only the loss of the second set of minor compound signals. The narrower lines for the HmuO complex compared with the HO complex are attributed in part to a reduction in size (216 versus 265 residues) relative to hHO. An example of the narrower line widths in the HmuO complex compared with the hHO complex is the detection of the complete Ser 138 TOCSY connections (data not shown; see Supplemental Material), including the NH-C ␣ H correlation missing in hHO⅐DMDH⅐CN (21), despite unchanged paramagnetic relaxation.
Utility of hHO as a Homology Model-The similarity in the  (15).
g Not uniquely assignable due to spectral congestion. 1 H NMR spectra of hHO⅐PH⅐CN and HmuO⅐PH⅐CN in Fig. 3 is completely confirmed by the remarkable similarities in not only the positions of the various secondary structural elements represented by the homologous fragments I, II, and IV-VI, as reflected in dipolar contacts among each other and with the heme (Fig. 6), but also the pattern of paramagnetic relaxation and hyperfine shifts (Table II). The similar heme methyl contact shifts (Table I) reflect a similarly oriented axial His imidazole plane, and the conserved pattern of dipolar shifts for the proximal helix (Table II) confirms an ϳ20°steric tilt of the FeCN toward the ␣-meso position in both mammalian HO (20,21) and this bacterial HO. The close similarity of the environments of the individual heme methyls not only reflects the numerous completely conserved contacts, but allows the ready identification of the HmuO residues whose nature has been dramatically altered compared with hHO residues, i.e. hHO Ile 211 3 HmuO His 205 in Fig. 5 (C and D) and hHO Gln 38 3 HmuO Leu 33 in Fig. 5 (A  and B). Fragment III (Fig. 2) could not be located in analogy with hHO because fragment III possesses three aromatic residues (Phe 95 , Trp 96 , and Tyr 97 ) in hHO that could be easily identified (21) by their contacts with fragment IV, and these residues on fragment III are substituted by aliphatic residues (Lys 89 , Leu 90 , and Asn 91 ) in HmuO. The cluster of aromatic side chains, i.e. those of HmuO/hHO Tyr 53 /Tyr 58 , Phe 160 / Phe 166 , Tyr 161 /Phe 167 , Tyr 130 /Tyr 134 , Tyr 133 /Tyr 137 , and Phe 42 / Phe 47 , are again largely conserved in the two HO proteins (Fig.  6). Finally, the acceptor for all but one (Tyr 161 O H) of the assigned strong H-bond donors could be identified in HmuO complexes solely on the basis of sequence homology. We therefore conclude that a structurally characterized HO complex will serve as a valuable homology model to facilitate the assignment of residues involved in many details of the active site structure in a related HO. The sequence homology between HmuO and hHO is relatively high (33% identity and 70% similarity if conservative substitutions are included) (10). Other bacterial HOs, such as HemO from N. meningitides (44), exhibit less sequence homology to mammalian HOs, but still exhibit a structure (45) that is related to that characterized in the two mammalian HOs (15,16) and one bacterial HO (45) and exhibit different details of the active site. To date, 1 H NMR data on HemO (44,45) have not been reported to allow comparisons.
Comparison with the HmuO⅐PH⅐H 2 O Crystal Structure-The 1 H NMR data are consistent with the crystal structure 2 for HmuO to the same degree previously found for the same hHO complexes (20,21). The proximal helix is strongly conserved, FIG. 6. Schematic representation of the relative positions of the heme and assigned fragments I, II, and IV-VI in HmuO. Fragment III, which could not be assigned, but was concluded to occupy the same position, is shown by the dashed lines. Observed dipolar contacts (dashed lines) among these fragments and with the heme and the locations of strong H-bonds among these fragments (solid arrows from donor to acceptor) are shown. The arrangement is based on the similarly assigned fragments for hHO⅐DMDH⅐CN (20,21) and comparison with the crystal structure of hHO⅐PH⅐H 2 O (15). The positions of two residue side chains, Phe 42 and Gln 46 , which make key contacts with fragment IV, and two conserved aromatic rings, Phe 201 and Phe 208 , which make contact with the heme, are also shown. The recent crystal structure of HmuO⅐PH⅐H 2 O (Footnote 2) confirms all observed contacts except that between the Tyr 161 ring and Gly 139 (labeled with an asterisk, where a distance Ͼ6 Å is predicted) and indicates that the likely acceptor for the strong H-bond by Tyr 161 OH is Gln 46 . but the distal helix exhibits dipolar shifts that deviate from those predicted by the relatively robust magnetic axes in the same fashion as found for hHO (see Supplemental Material) (20,21). The loss in solution of dipolar contacts between the NH groups of Leu 134 and Gly 135 and 3-CH 3 and the weakening of contacts between Ser 138 C ␤ H and 6-H ␣ (Fig. 7B)  predictions based on the crystal structure 2 suggest possibly only a small (0.5-1.0 Å) movement of the distal helix near its kink. However, similar differences in the distal helix position have been observed in the two non-equivalent molecules in the hHO⅐PH⅐H 2 O crystal (15) and may simply represent the intrinsic mobility of the distal helix.
The distal H-bond network in the HmuO complex, 2 as in the case of hHO⅐PH⅐H 2 O, is not readily discerned in the crystal structure (14,15). However, once the donor NH and OH groups have been identified by 1 H NMR, the crystal structure readily identifies the probable acceptors. The proposed acceptors for the strong H-bonds in HmuO are the Glu 57 , Asp 86 , Asp 136 , and Glu 196 carboxylates, based solely on the homology to hHO⅐DMDH⅐CN NMR data (21) and the hHO⅐PH⅐H 2 O crystal structure (15) and completely confirmed in the HmuO crystal structure (Fig. 6). 2 The crystallographic geometry (distance and angle) 2 for these H-bonds in HmuO is summarized in Table  III, where they can be compared with similar data on hHO⅐DMDH⅐CN and hHO⅐PH⅐H 2 O. Their dispositions are far from ideal (42) to allow the strong H-bond so obvious in the 1 H NMR data, as shown in Table III. This non-ideal orientation of the donors and acceptors may be simply the result of the intrinsic uncertainties in the crystallographic positions of the two interacting units. The acceptor for the new strong H-bond from Tyr 161 O H is identified in the crystal 2 as the side chain of Gln 46 (Fig. 6). Similar strong H-bonds, as yet unassigned, appear in the 1 H NMR spectra of apo-HO in both mammals and bacteria, 3 indicating that the H-bond network plays a key role in the structures of both the apo-HO and substrate complexes.
A structural difference of possible significance between the HmuO cyanide-ligated complex and the crystal structure of the aquo-ligated complex 2 is the relative position of fragment IV relative to the distal helix II. This same difference was previously observed in the hHO complex (20,21). Thus, although moderate intensity NOEs are observed between the Tyr 161 ring and Gly 139 NH (i.e. r ij ϳ 4 Å), the crystal structure indicates r ij Ͼ 6 Å. This same difference was previously observed for the analogous hHO complexes involving the homologous Phe 167 ring and Gly 143 NH (20,21). Thus, the aromatic cluster appears to move ϳ2-3 Å closer to the distal helix in the cyanide complex in solution compared with the aquo complex in the crystal. This difference may be due to the different ligands used in the alternate studies in solution (CN Ϫ , a H-bond acceptor) and in the crystal (H 2 O, a H-bond donor).
Ordered Water Molecules-NOEs indicative of nearby immobilized water molecules (Fig. 4, A, BЈ, and CЈ) (22). Hence, both HOs are characterized by ordered water molecules, particularly in the distal pocket. However, even these preliminary data indicate differences between the two HOs in the organization of these water molecules. Thus, the NOE for His N ⑀ H is weaker in HmuO (His 128 ) than in hHO (His 132 ) (21), but the NOEs for the NH groups of Arg 85 and Phe 160 are significantly larger (ϳ25-30%) in the HmuO complex (Fig. 4CЈ) than in the hHO complex (ϳ10 -15%) (22). These differences could be the result of differences in water-NH distances, the number of nearby water molecules, and/or the mobility of the ordered water molecules (43). The crystal structure 2 of HmuO⅐PH⅐H 2 O reveals numerous water molecules in the distal side of the heme at positions similar to those detected by 1 H NMR in the hHO complex (22). The different stages of refinement for the hHO and HmuO crystal structures and the availability of only preliminary NMR data in solution suggest that a discussion of differences in the occupation of water molecules be deferred until water NOEs can be more effectively studied in 15 N-labeled HO.
Comparison of Dynamic Properties of HmuO and hHO-The very close structural homology between HmuO and hHO apparent in both the solution 1 H NMR data and crystal structures is, however, in contrast to the highly differential dynamic properties of the two enzymes. On the one hand, the rate of exchange with 2 (22). Moreover, the Tyr 53 /Tyr 58 OH groups, which exhibit saturation transfer due to exchange, exhibit a much smaller saturation factor (ϳ10%) in HmuO than in hHO (ϳ40%), dictating a much slower exchange rate in HmuO. The ϳ700 factor decrease in the Phe 160 /Phe 166 NH exchange rate indicates that the dynamic stability near fragment IV is ϳ4 kcal greater in HmuO than in hHO (46). Although the extreme low-field shifts for the labile protons of the H-bond networks are very similar in the two HO complexes (Table III), indicating that the individual H-bonds are compa-rably strong, other factors that contribute to the stability of the folding in the environment of the network are clearly much weaker in hHO than in HmuO.
The 1 H NMR data provide other indicators for a dynamically more stable (and hence, less flexible) HmuO than hHO. On the one hand, two new and strong H-bonds are observed, one of which could be uniquely attributed to Tyr 161 O H (which substitutes for Phe 167 in hHO). The likely acceptor, Gln 46 , is suggested by the HmuO crystal structure, 2 although the strong low-field bias due to H-bonding suggests a stronger or more than one acceptor. In fact, the crystal structure 2 of HmuO⅐PH⅐H 2 O places a water molecule within 3 Å of this OH. The sequence origin of the other strong H-bond (peak a at 12.4 ppm in Fig. 4A) is not identified, but has no homolog in hHO. Nevertheless, these two strong H-bonds in HmuO are incremental over those conserved relative to hHO (21,22). The second observation is that the Tyr 161 ring, unlike the Phe 167 ring, exhibits slow ring reorientation about the C ␤ -C ␥ bond, as evidenced by the resolution of two C ⑀ H peaks at low temperature, whereas an averaged C ␦ H peak is observed for Phe 166 at all temperatures for hHO complexes (20,21). The decreased mobility of Tyr 161 in HmuO relative to the Phe 167 ring in hHO (21) in 2-fold reorientation supports a tighter and more constrained distal environment in HmuO than in hHO.
Role of the H-bond Network-The conservation of the H-bond network/aromatic cluster/ordered water molecules in HmuO relative to hHO argues for important functional roles. The existence of a network of water molecules that includes water molecules near the catalytically critical distal helix Asp 136 / Asp 140 (12,14) supports the notion that water provides the stabilizing H-bond to the novel hydroperoxy intermediate. Interaction of Asp 140 with the distal ligand via two water molecules has been recently characterized in the crystal structure of the rat HO⅐PH⅐N 3 complex (23). The presence of a water/Hbond network that extends from the distal pocket through the enzyme to its surface on the opposite side from the substratebinding pocket (22) 2 suggests that the channel may funnel the required nine protons to the active site in a controlled manner.
The greater dynamic stability of the pocket near the catalytically important Asp 136 /Asp 140 is also apparent in two other observations. The greater dynamic stability of the distal pocket in HmuO relative to hHO, witnessed in both slower labile proton exchange and aromatic ring reorientation, may be responsible for the ϳ4 factor slower turnover rate in HmuO (30) than in mammalian HOs (2,11). More extensive NMR studies of both the dynamic properties of the distal side and the distribution of oriented water molecules in HmuO and other HO complexes are in progress.