1H NMR Investigation of the Distal Hydrogen Bonding Network and Ligand Tilt in the Cyanomet Complex of Oxygen-avidAscaris suum Hemoglobin*

The O2-avid hemoglobin from the parasitic nematode Ascaris suum exhibits one of the slowest known O2 off rates. Solution 1H NMR has been used to investigate the electronic and molecular structural properties of the active site for the cyano-met derivative of the recombinant first domain of this protein. Assignment of the heme, axial His, and majority of the residues in contact with the heme reveals a molecular structure that is the same as reported in the A. suumHbO2 crystal structure (Yang, J., Kloek, A., Goldberg, D. E., and Mathews, F. S. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 4224–4228) with the exception that the heme in solution is rotated by 180 ° about the α,γ-meso axis relative to that in the crystal. The observed dipolar shifts, together with the crystal coordinates of HbO2, provide the orientation of the magnetic axes in the molecular framework. The major magnetic axis, which correlates with the Fe-CN vector, is found oriented ∼30 ° away from the heme normal and indicates significant steric tilt because of interaction with Tyr30(B10). The three side chain labile protons for the distal residues Tyr30(B10) and Gln64(E7) were identified, and their relaxation, dipolar shifts, and nuclear Overhauser effects to adjacent residues used to place them in the distal pocket. It is shown that these two distal residues exhibit the same orientations ideal for H bonding to the ligand and to each other, as found in the A. suumHbO2 crystal. It is concluded that the ligated cyanide participates in the same distal H bonding network as ligated O2. The combination of the strong steric tilt of the bound cyanide and slow ring reorientation of the Tyr30(B10) side chain supports a crowded and constrained distal pocket.

The O 2 binding globins, myoglobin (Mb) 1 and hemoglobin (Hb), despite highly varied sequences throughout phylogeny, possess a highly conserved folding topology of 7-8 helices (A-H), with the heme wedged in between the E and F helices and ligated by one, His F8 (proximal), of only two (the other is Phe CD1) completely conserved residues (1)(2)(3). Despite this strong structural homology, the O 2 ligation rates and O 2 affinities vary over a remarkably wide range (by ϳ10 5 ), depending on the exact nature of several distal residues at the key positions B10, E7, and E10 (4 -8). The most important distal interaction for stabilizing bound O 2 is hydrogen bonding to the ligand, for which the donor is generally His E7 (1,7,9,10) and is Gln E7 in a few cases (2). In several invertebrates, such as Aplysia and Dolbella Mbs that possess a Val E7, the distal H bond to the ligand is provided by an Arg at position E10 (11)(12)(13).
A particularly noteworthy class of globins is that of parasitic nematodes that possess, in addition to a H bond donor at position E7, a Tyr at position B10 that is also capable of H bonding to the ligand (4, 5, 8, 14 -17). In the case of the Hb from Ascaris suum, the extraordinarily high O 2 affinity and extremely low O 2 off-rates have been attributed to a distal H bonding interactions for the Tyr 30 (B10) and Gln 64 (E7) side chains with bound O 2 . Stabilizing H bond interactions of Gln 64 (E7) and Tyr 30 (B10) with ligated O 2 are supported by the observations of enhanced O 2 off-rates upon mutating either residue (5,15). The positions of these two key residues are clearly defined in the crystal structure of A. suum HbO 2 , in which the two residues are appropriately poised to serve as H bond donors to bound O 2 , with the Gln 64 (E7) additionally providing an H bond to the Tyr 30 (B10) side chain O that stabilizes the optimal dispositions of these two residues (8). Resonance Raman spectroscopy has confirmed the role of Tyr 30 (B10) as a H bond donor (18,19), and flash photolysis experiments (19) have indicated that A. suum Hb possesses a very compact and constrained distal pocket when compared with other globins.
Resonance Raman spectroscopy has shown that A. suum HbCO exhibits the lowest reported CO stretching frequency (19). The strong modulation of the CO stretching frequency by globin distal environment has been discussed in the context of both steric tilt/bending of the Fe-CO unit from the heme normal and pocket dielectric effects (20 -23), and the currently accepted interpretive basis is that the latter effect is the major determinant of CO (7,24). Nevertheless, crystal structures of myoglobins invariably find the carbonyl oxygen placed off-axis from the heme normal, indicating that the Fe-CO unit is bent/ tilted from the heme normal (7,(25)(26)(27).
The cyanomet derivatives of globins can serve as valuable structural (but not functional) (28,29) models for both O 2 and CO binding, in that Fe III CN, like Fe II O 2 , is polar and is a good H bond acceptor (30) and, like Fe II CO, prefers to bind normal to the heme in the absence of distal steric interactions (31). In the one case where the crystal structures of both the carbonyl and cyanomet globins have been reported, there is a good correlation in the degree and direction of the off-axis placement of the terminal atoms (26). Theoretical considerations have indicated that distal ligand tilt could be modulated by tilt of the proximal His (32). In the crystallographically and NMR characterized globins to date, the axial His is essentially normal to the heme. The crystal structures of A. suum HbO 2 , on the other hand, show that the axial His imidazole plane is tilted some ϳ8°in the direction of pyrrole C with respect to the heme plane (8). Thus a determination of orientation of the Fe III CN or Fe II CO units relative to the heme in A. suum Hb would indicate whether the axial His could contribute to distal ligand tilt and provide some insight as to whether there is likely to be a large Fe-CO tilt that could contribute to the reduced value for co .
Solution 1 H NMR of the paramagnetic cyanomet Hb derivatives can provide significant structural details on the distal pocket in relation to both stabilizing H bonding and destabilizing steric interactions with the bound ligand (30,33,34). On the one hand, the dipolar shifts and moderate relaxation imparted to distal residues and their labile protons facilitate their detection, identification, and detailed placement relative to the bound ligand (35,36). On the other hand, the sizable dipolar shifts for active site residues allow the quantitative determination of the orientation of the paramagnetic susceptibility tensor, for which the major magnetic axis can be correlated to the degree of Fe-CN tilt from the heme normal (34). There is generally very good agreement in the magnitude and direction of Fe-CN tilt observed in crystal structure and the orientation of the major magnetic axes determined by solution 1 H NMR (26, 27, 38 -41). Lastly, the expanded chemical shift scale for heme pocket residues because of the hyperfine interaction increases the prospect for measuring rapid dynamic processes, such as ring orientation, that can constitute probes for the constraints in the heme pocket (42).
We report herein on the solution 1 H NMR characterization of the cyanomet complex of the D1 domain of A. suum metHbCN, which demonstrates that the distal H bonding network is essentially identical to that in HbO 2 , that the Fe-CN unit appears tilted strongly away from the heme normal in the direction of the observed terminal oxygen in HbO 2 , and that the distal pocket is sufficiently crowded to strongly tilt the Fe-CN vector and to impede the reorientation of the Tyr 30 (B10) ring.

EXPERIMENTAL PROCEDURES
Protein Preparation-Native protein samples were prepared as described previously (15,43). The cyanide complexes were prepared by adding KCN to the protein solution approximately in a molar ratio of 10:1 buffered with 50 mM phosphate, 200 mM NaCl at pH 7.2. 2 H 2 O sample was prepared by repeatedly washing protein with 2 H 2 O in the same buffer with a Centricon (Amicon Inc.), and pH was read directly from the pH meter without the isotope effect correction; the final protein concentration was ϳ2 mM.
NMR Spectra-All 1 H NMR spectra were recorded on a GE ⍀ 500 spectrometer operating at 500 MHz. Chemical shift values were referenced to 2,2-dimethyl-2-pentane-5-sulfonate (DSS) through the residual water signal. Reference spectra were collected with 1 H 2 O saturation. Steady-state NOE and inversion-recovery spectra were collected at a repetition rate of 3 s Ϫ1 (34). The residual water signals were removed from the free induction decay by convolution difference. The nonselective spin lattice paramagnetic relaxation times for the resolved peaks were derived from two-parameter exponential least square fits using only short (Յ50 ms) delays. Estimates for distance to the iron for proton i, R Fe-i were obtained from the nonselective paramagnetically dominated T 1 values using the following relation.
where the reference T 1j ϭ 150 ms (R Fe-j ϭ 6.1 Å) for a heme methyl, or T 1j ϭ 30 ms (R Fe ϭ 5.1 Å) for the His(F8) N ␦ H, provided upper and lower limits, respectively (44). Interproton distances, r ij , were estimated from steady-state NOEs, i-j to protons with paramagnetically dominated (nonselective) T 1 values, via the following two equations.
NOESY (45) and TOCSY (46 -48) spectra were collected over a temperature range of 20 -35°C in 1 H 2 O. Two different spectral windows and mixing times were used for NOESY 25.0 KHz using 2048 complex points at 3 scans/s with a mixing time of 35 ms to optimize the observation of the hyperfine shifted signals and 10.0 KHz at 1 scan/s with a mixing time of 100 ms to cover the diamagnetic window at optimal digital resolution. The clean TOCSY spectra were collected over 12.0 KHz at 2 scans/s with a spin locking time of 35 ms using the MLEV-17 mixing scheme (47). All the two-dimensional data sets were processed on a Silicon Graphics workstation either using the software package FELIX from Biosym/MSI (San Diego, CA) or AZARA generously provided by Wayne Boucher (Department of Biochemistry, University of Cambridge). To increase the resolution, data sets were apodized with a sine-bell-squared function shifted by 20 -40°in both dimensions and zero-filled once in t 1 dimension. The spectral assignments were largely facilitated with the aid of ANSIG package (49).
with ␦ dip (calc) given by the following equation.
where ⌬ ax ϭ zz Ϫ 1 ⁄2( xx ϩ yy ) and ⌬ rh ϭ xx Ϫ yy are the axial and rhombic anisotropies of the diagonal paramagnetic susceptibility tensor, . The tilt of the major magnetic axis, z, from the heme normal is given by ␤, the projection of this tilt on the heme plane relative to the xЈ axis is given by ␣, and the location of the rhombic axes projected on the heme plane is approximated by ϭ ␣ ϩ ␥, as labeled in Fig. 1C. ␦ dip (obs) is given by the following equation.
␦ DSS (obs) is the observed chemical shift referenced to DSS. ␦ DSS (dia) is the isostructural diamagnetic shift, which, in this case, was calculated by using: where ␦ tetr is the shift in an unfolded tetra peptide (54), ␦ sec is the shift of an amino acid proton resulted from the protein secondary structure (55), and ␦ rc is the hemeinduced ring current shift (56). Minimizing the error function, F/n, in Equation 4 was performed over five parameters, ⌬ ax , ⌬ rh , ␣, ␤, and ␥, using the HbO 2 crystal coordinates. For the iron ligated porphyrin and axial His, the hyperfine shift is obtained via the following equation.
which yields the contact shift via the following equation.
Molecular Modeling-Protons were added to the crystal coordinates of A. suum HbO 2 using the program INSIGHT II (MSI). This provided unique coordinates for all protons of interest except the Tyr 30 (B10) hydroxyl proton; hence its position was determined from the 1 H NMR spectral parameters.

RESULTS
A schematic representation for selected heme cavity proximal (squares) and distal (circles) residues and their disposition relative to the heme is shown in Fig. 1. The heme is shown in . The region downfield of ϳ9 ppm resolves four three-proton (methyls) and four single-proton signals as well as one two-proton peak at 11 ppm that overlaps a methyl peak. Ten single-proton peaks and four methyl peaks can be resolved at some temperature in the upfield portion of the spectra. The comparison between 1 H 2 O and 2 H 2 O reveals that three of the most strongly relaxed resolved protons are labile.
Although the present 1 H NMR data on metHbCN confirm a highly conserved arrangement relative to each other, of both proximal and distal residues (see below), the data also demonstrate that the heme is oriented differently in the cavity from that originally reported in the crystal structure (8). Thus as-signments for the heme and axial His are pursued first, followed by the key residues (i.e. Phe 44 (CD1) and Met 103 (FG5)) that invariably place strongly relaxed protons in resolved spectral windows and hence can be unambiguously assigned based solely on the conserved globin fold. These two residues are then used to uniquely orient the heme in the cavity. The remainder of the assignments are then presented on the basis of characteristic TOCSY-detected spin systems, with heme contacts expected on the basis of the heme orientation deduced above. Because the protocol for heme and resolved residue assignments in similar cyanomet globins has been described in detail (13), NMR data are illustrated only for the key heme-Phe 44 (CD1) contact that determines the heme orientation and for the distal H bond donors, Tyr 30 (B10) and Gln 64 (E7).
Heme Assignments-The heme substituents could be unambiguously assigned as described in detail elsewhere (13). Two TOCSY-detected vinyl and one propionate groups exhibit NOESY cross-peaks to low field resolved methyls that pair 1-CH 3 , 2 vinyl and 3-CH 3 , 4 vinyl; NOESY cross-peaks between two of the heme methyls (Fig. 3D) assign the 1-CH 3 and 8-CH 3 , and a NOESY cross-peak between the remaining methyls and the one detected propionate uniquely assigns the pyrrole substituents. Common NOESY cross-peaks for the two substituents flanking a meso position, together with large low field intercepts in a Curie plot (not shown), identify the meso-Hs (34). The heme assignments, chemical shifts, T 1 values, and slopes in a Curie plot, are listed in Table I.
Assignment of Key Resolved Resonances-The 19.0 ppm labile proton, when saturated (not shown), exhibits NOEs to a labile proton at 12 ppm, which is part of a nonlabile proton NMR spin system diagnostic of the axial His 97 (F8) C ␤ H 2 C ␣ H-NH fragment. A very broad and strongly relaxed (T 1 ϭ ϳ3 ms) upfield, nonlabile proton peak must arise from the axial His ring (34), and a NOE to the C ␤ H upon saturating this The heme is labeled with the Fisher notation, and the substituents are labeled M (methyl), V (vinyl), and P (propionate). The expected (on the basis of the crystal structure with the rotated heme) and observed inter-residue and residue-heme dipolar contacts are shown in B by double-sided arrows. The iron-centered, crystal structure-based coordinate system (xЈ, yЈ, zЈ) is shown in C, as is the magnetic coordinate system (x, y, z), where is diagonal. The two systems are related by the Euler rotation, ⌫ (␣, ␤, ␥), [x, y, z] ϭ [xЈ, yЈ, zЈ] ⌫ (␣, ␤, ␥), where ␤ is the tilt of the major magnetic axes from the heme normal, ␣ is the angle between the projection of the tilt on the heme plane and the xЈ axis, and ϭ ϳ␣ ϩ ␥ define the projection of the rhombic axes on the xЈ, yЈ plane. is the orientation of the proximal His imidazole ring plane relative to the N A -Fe-N C vector (xЈ axis). It is noted that the convention for xЈ, yЈ, zЈ differs from that used previously by a 45° (34, 39 -41, 44, 50) rotation in the heme plane and referencing ␣ to the ϩxЈ rather than ϪxЈ axis, so that The assigned resolved signals are labeled by the Fisher notation for the heme and by the one-letter code for the residue and sequence position. Also shown are steady-state NOE difference spectra upon saturating the low field Tyr 30 (B10) OH signal (C) and upon saturating the upfield Gln(E7) N ⑀ H 2 (D). The intensity of the saturated signals in traces C and D are identical. The detected NOEs are assigned as presented in the text. An asterisk indicates off-resonance saturation. peak (not shown) establishes that it is the His 97 (F8) C ␦ H. The TOCSY-detected C ␣ HN p H backbone and several C ␤ Hs of the members of the F-helix, Asp 94 (F6) through Arg 98 (F10), could be located via the standard expected helical N i -N iϩ1 , ␣ i -N iϩ1 contacts (not shown) (54). Although these backbone assignments could not be extended to F4 because of spectral congestion, the expected strong NOE between His 97 (F8) N ␦ H and Leu 92 (F4) locates the latter residue C ␣ H.
A resolved (15 ppm), strongly relaxed, nonlabile single proton peak exhibits the TOCSY connectivity (Fig. 3, A and B) and variable temperature slope and intercepts of a rapidly reorienting aromatic ring. The relaxation properties (T 1 ϭ ϳ20 ms) alone dictate that it must arise from the completely conserved Phe 44 (CD1) (42). TOCSY connections (not shown), moreover, involving two sets of upfield hyperfine shifted residues identify AMXPT and AM(X 3 )(Y) 3 spin systems, which, moreover, ex-hibit several NOESY cross-peaks to each other (shown schematically in Fig. 1). The spin topology, their significant dipolar shifts, and their inter-residue contacts uniquely identify these two residues as Met 103 (FG5) and Val 101 (FG3). An upfield, resolved, strongly relaxed methyl peak with no TOCSY connectivities exhibits strong NOESY cross-peaks to the terminus of the AMXPT spin system, locating the C ⑀ H 3 , which, together with the NOESY cross-peak to His 97 (F8) (shown schematically in Fig. 1), confirms the assignment of Met 103 (FG5). A Tyr ring with contacts to Val 101 (FG3) identifies Tyr 43 (C7).
Orientation of the Heme-The Phe 44 (CD1) ring exhibits NOESY cross-peaks to both the 1-CH 3 and 8-CH 3 of the heme (Fig. 3, C and D) and not to the 5-CH 3 and 4-vinyl, as predicted by the crystal structure (Fig. 1A), clearly establishing that the heme is oriented differently from that in the crystal by a 180°r otation about the ␣,␥-meso-axis (Fig. 1B). This conclusion is further confirmed by detecting the characteristic dipolar contact between Met 103 (FG5) and the 1-CH 3 (rather than the 4-vinyl predicted by the crystal structure), as shown schematically in Fig. 1, and between Val 101 (FG3) and 8-CH 3 (rather than the 5-CH 3 , predicted by the crystal structure). Hence all subsequent assignments are determined by using the crystal structure as a guide but with the heme rotated by 180°about the ␣,␥-axis, as shown in Fig. 1B.
Distal Pocket Residues-The extreme low field, strongly relaxed (T 1 ϭ ϳ11 ms) labile proton peak at 22 ppm does not participate in a NOESY map, but when it is saturated (Fig. 2C), it exhibits a strong NOESY cross-peak to a two-proton signal under the 8-CH 3 . This signal under the 8-CH 3 in turn exhibits a TOCSY cross-peak to 9.0 ppm. The strong relaxation of the labile proton and the intercept in Curie plots for the TOCSY detected fragment uniquely identify the complete ring of Tyr 30 (B10). A strong NOE to the Phe 44 (CD1) C H, together with the latter T 1 pf ϳ20 ms, yields, with Equation 2, a 2.8 Ϯ 0.3 Å estimate for the Tyr 30 (B10) OH to Phe 44 (CD1) C H distance. Similarly, the relaxation properties (T 1 ϭ ϳ24 ms) of an upfield shifted labile proton dictate that it must arise from the only other residue that can place labile protons so close to the iron, the terminal N ⑀2 H of Gln 64 (E7). 3 Saturation of this peak results in a very strong (ϳ15%) NOE to a proton near 4 ppm. The large NOE from this upfield labile proton to 4 ppm identifies the latter as the geminal partner of the saturated peak. The strong NOE from Tyr 30 (B10) OH to the 4 ppm Gln N ⑀1 H confirms the assignment and argues for the assignment of the 4 and Ϫ6 ppm peaks to the N ⑀1 H and N ⑀2 H (see below). The ratio of the steady-state NOEs to the 4 ppm Gln 64 (E7) N ⑀ H 1 peak upon saturating the Gln 64 (E7) N ⑀ H 2 and Tyr 30 (B10) OH (ϳ0.5), together with the fixed ϳ1.9 Å distance between N ⑀1 H and N ⑀2 H, leads to a 2.0 Ϯ 0.2 Å estimate for the Tyr 30 (B10) OH to Gln 64 (E7) N ⑀1 H distance.
Additional assignments (data not shown), include two upfield resolved methyls (one strongly relaxed) that are part of a five-spin system diagnostic of a Ile, with a strongly relaxed C ␣ H that exhibits the NOESY cross-peaks to 5-CH 3 and 4-vinyl (as predicted by the 180°rotated heme) for Ile 68 (E11); this residue exhibits the expected NOESY cross-peaks to the Tyr 30 (B10) ring (shown schematically in Fig. 1). Common NOESY contacts to 4-vinyl and 5-CH 3 for a TOCSY-detected Ala uniquely identify Ala 71 (E14). TOCSY spectra detect three Phe rings with weak hyperfine shifts. They are assigned to Phe 34 (B14), Phe 60 (E3), and Phe 140 (H15) based on their predicted dipolar contacts to Phe 44 (CD1) and Tyr 30 (B10), only to Phe 44 (CD1), and to 3-CH 3 , respectively. Two low field shifted TOCSY-detected fragments with slopes and intercepts indicative of aromatic pro- 3 Denotation of N⑀Hs of Gln 64 was based on the x-ray structure (8). C and D, portions of the NOESY spectrum that illustrates the dipolar contact between the two heme methyls that uniquely assigns 1-CH 3 and 8-CH 3 and the dipolar contact of the Phe 44 (CD1) ring with both the 1-CH 3 and 8-CH 3 of the heme, which uniquely characterize the orientation of the heme in the pocket as rotated by 180°about the ␣,␥-meso axis relative to that found in the HbO 2 crystal structure. tons, together with NOESY cross-peaks to Met 103 (FG5) identify the C ␦1 H-C ⑀1 H and C 2 -C 2 H portions of Trp 108 (G5); the remainder of the ring protons could not be located because of likely strong relaxation and near degeneracy with other protons and in position under the residual solvent peak. The observed interresidue and residue-heme dipolar contacts are summarized in Fig. 1B. Spectral congestion precluded further assignments. The assignments, chemical shifts, and T 1 values for the residues described in Fig. 1 are listed in Table II.
Magnetic Axes-The orientation of the magnetic axes 2 was found to be essentially independent of the selection of input data or whether the anisotropies were also determined or held constant at the values determined for sperm whale metMbCN (50,51). The resulting orientation of is defined by ␤ ϭ 29.5°Ϯ 1.0°(tilt from the heme normal), ␣ ϭ 159 Ϯ 10°(direction of tilt projected on the heme plane), and ϭ ␣ ϩ ␥ ϭ 59 Ϯ 10°( rhombic axes projected on the heme plane). The residual error function, F/n, is small in all cases (ϳ 0.05 ppm 2 ), and the resulting correlation between observed and calculated dipolar shifts is very good, as illustrated in Fig. 4. The magnitude of the tilt of the major magnetic axis, z, from the heme normal (zЈ axis), ␤ ϭ ϳ30°, is nearly twice as much as that observed previously in cyanomet globins (34,50,51). The large magnitude of the tilt of the major magnetic axis from the heme normal indicated by the complete magnetic axes determination also reveals itself clearly in the analysis of the dipolar shift pattern for individual residues. Thus the nodal surface for the axial dipolar shift can be mapped by considering the magnitude and direction of dipolar shifts of residues near the nodal surface. The plots in Fig. 5 for the protons whose shift direction/ magnitude reflect primarily the axial geometric factor node are shown as a function of tilt angle ␤. The agreement with the experimental shifts is acceptable within 30 Ϯ 10°.
The Orientation of Tyr 30 (B10) and Gln 64 (E7) -Predicted ␦ dip values for the Tyr 30 (B10) ring and Gln 64 (E7) N ⑀ Hs, based on the HbO 2 crystal coordinates (8), are included in Fig. 4 as filled circles and filled squares, respectively. It is observed that the uniquely placed protons on the Gln(E7) N ⑀2 H result in shifts that are well predicted, indicating that this residue in met HbCN maintains the same orientation relative to the iron as in HbO 2 . The R Fe ϭ 4.5 Ϯ 0.4 Å estimated from the T 1 ϭ 25 ms for the Gln 64 (E7) N ⑀2 H is consistent with the crystallographic R Fe ϭ 4.1Å. Moreover, the predicted dipolar shifts for the Gln 64 (E7) nonlabile side chain protons are small and are consistent with the likely appearance of protons in the poorly resolved and very crowded aliphatic envelope. In the case of Tyr 30 (B10), the dipolar shifts are very well predicted for the ring, which is consistent with conserved 1 , 2 angles with respect to HbO 2 . The placement of the proton on the hydroxyl oxygen crystal coordinates, however, unlike the N ⑀ Hs of Gln 64 (E7), is not unique. Hence the ␦ dip (calc) (Fig. 6A), distance to Phe 44 (CD1) C H (Fig.  6B) , distance to Gln 64 (E7) N ⑀1 H (Fig. 6C), and distance to the iron, R Fe (T 1 ϭ 10 ms, R Fe ϭ 4.0 Ϯ 0.4 Å) (Fig. 6D) for the Tyr 30 (B10) OH are calculated as a function of the dihedral angle between the H-O-C and ring planes, 3 , as illustrated in Fig. 6; the observed values are shown by shaded regions. It is clear that each of the four observable values are optimally predicted for the angle ϳ20°in Fig. 6, which leads us to conclude that we have uniquely spatially located the labile proton for the distal Tyr 30 (B10).
Mobility of the Tyr 30 (B10) Ring-The Tyr 30 (B10) C ⑀ H resonance overlaps, at least in part, the 8-CH 3 peak ( Fig. 2A) over the accessible temperature range but appears to broaden selectively as the temperature is lowered. The line broadening, however, can be quantitated by observing only the steady-state  NOE for the averaged C ⑀ H peak upon saturating the Tyr 30 (B10) OH, as shown in Fig. 7. Plotting the ln (linewidth) versus reciprocal temperature shows a plot with selective increase in slope at low temperature for the C ⑀ Hs peak, which yields an estimated exchange contribution of 20 Hz at 30°C. This value, together with the ␦ dip (calc) for the individual Tyr 30 (B10) C ⑀ Hs, results in an estimated shift difference of 4.4 ppm, which at 500 MHz, results in a reorientation rate of 5 ϫ 10 6 s Ϫ1 using the standard equation for chemical exchange in the first exchange limit (57).

Heme Pocket Molecular and Electronic
Structure-The pattern of the heme methyl contact shifts has been proposed to largely reflect the orientation of the axial His imidazole ring relative to a heme pyrrole-Fe-pyrrole axis (12, 58 -61). For an axial His oriented along such an N-Fe-N axis, large contact shifts are predicted and observed primarily for the pyrroles normal to the His plane. Thus the contact shift patterns among different globins are modulated separately by the orientation of the His relative to the heme and the orientation of the heme about the ␣,␥-meso axis. In cases where the axial His is oriented close to meso-Fe-meso vectors (11,62), the four pyrroles exhibit comparable contact shifts (13,63). A. suum metHbCN, like mammalian globins, exhibits large contact shifts for 1-CH 3 and 5-CH 3 , arguing for orientation of axial His along the N B -Fe-N D vector of the heme if the heme and the axial His 97 (F8) were orientated similarly. However, the heme methyl contact shift pattern in A. suum metHbCN is achieved by completely different means than in sperm whale metMbCN. Thus, as shown in Fig. 1, the axial His ring (as viewed from the proximal side) is rotated by ϳ60°( ϭ Ϫ65°) relative to that in sperm whale Mb ( ϭ Ϫ6°), which should result in larger 3-CH 3 , 8-CH 3 than 1-CH 3 , 5-CH 3 contact shifts, if the heme were seated in the pocket the same as in sperm whale Mb. However, the rotation of the heme by 180°about the ␣,␥-meso axis, when compared with sperm whale Mb, reverses this pattern and leads to larger 1-CH 3 , 5-CH 3 contact shifts than 3-CH 3 , 8-CH 3 contact shifts. Thus the fortuitous similarity in the heme contact shift pattern in A. suum metHbCN and mammalian glo-bins is due to off-setting influences of the differences in the orientation of both the axial His and the heme.
The determined heme methyl contact shifts, together with the x-ray determined axial His orientation, thus independently confirm that the heme in A. suum metHbCN in solution is rotated by 180°about the ␣,␥-meso axis relative to that reported in the HbO 2 crystal structure. These results also suggest that caution should be exercised in assigning a heme orientation based on heme methyl contact shifts in a cyanomet globin unless the orientation of the axial His is known. The re-evaluation of the x-ray diffraction data to reconcile the alternate heme orientation in the crystal and solution has shown that the heme in the crystal is, in fact, rotated by 180°about the ␣,␥-meso axis from that originally reported (8) 4 and the same as found by 1

H NMR in solution.
Theoretical considerations (61,64) confirmed in model compounds (65,66) dictate that if the orbital ground state is determined by the axial His(F8) bonding, the rhombic axes, , and the angle between the heme N-Fe-N and imidazole plane, (Fig. 1C), obey the counter-rotation rule where ϭ Ϫ. The present results conform quite well to these predictions, as shown in Fig. 8. The temperature dependence of the heme methyl shifts reveals that the 1-CH 3 , 5-CH 3 exhibit positive slopes that are steeper than Curie (T Ϫ1 ) behavior, whereas the 3-CH 3 , 5-CH 3 exhibit slopes that are negative or exhibits anti-Curie behavior. This effect is expected on the basis of thermal population of the excited orbital state, where the lone spin on the iron becomes delocalized into pyrroles B and D (60,(67)(68)(69). Lastly, the magnetic axes reported above allow the determination of ␦ dip for the axial His, which, in turn, provides ␦ con for each of the positions, as shown in Table I. Thus only the C ⑀ H exhibits large contact shifts that are very similar to those reported for sperm whale metMbCN (70) and confirms an essentially conserved axial His-Fe bond in A. suum relative to sperm whale Mb.
Distal Hydrogen Bonding Network-The excellent correlation between the observed and crystal-structure predicted values for ␦ dip and T 1 for the Tyr 30 (B10) ring and Gln 64 (E7) N ⑀2 H side chain shows that their dispositions in metHbCN are essentially quantitatively conserved relative to those in the HbO 2 crystal structure (8). The position of the Tyr 30 (B10) hydroxyl proton, deduced from its relaxation, NOESY, and dipolar shift constraints, with an 3 value of ϳ20°, is precisely in the position to make an ideal H bond to the strongly tilted cyanide ligand. The short interproton distance between the Tyr 30 (19). Significant crowding in the distal pocket is evident in two 1 H NMR spectral parameters. The major magnetic axis (Fe-CN tilt) is tilted from the heme normal by ϳ30°, nearly twice as much as in other globins (34, 39 -41, 51, 63). This can be rationalized by the disposition of the Tyr 30 (B10) ring, which provides a steric barrier to ligation along the heme normal. The orientations of the Tyr 30 (B10) ring and the Fe-CN tilt (if only tilted and not primarily bent) determined herein place the two residues in van der Waals' contact between the Tyr O and the N of the bound cyanide. However, the tilt of the major magnetic axis (ϳ30°) is in the same direction (toward pyrrole C), as is the tilt of the proximal His 97 (F8) imidazole plane (by ϳ10°) observed for HbO 2 (8), so that the large tilt in the major magnetic axis, and hence the Fe-CN tilt, could have a significant contribution (to ϳ10°) from the proximal His tilt (32). The present results suggest that the crystal structure of A. suum HbCO would find the CO off axis to a degree that is much larger than found in other carbonyl globins. The role of the tilt for the axial His(F8) in contributing to either Fe-CO (32) or Fe-CN tilt could be addressed by either the crystal structure of the carbonyl complex or the solution 1 H NMR determination of the magnetic axes of the cyanomet complex, for the A. suum Hb mutant where the covalent connection between the axial imidazole and the F-helix backbone is severed in the His(F8) 3 Gly mutant (71,72), allowing an exogenous imidazole to bind in the preferred normal to the heme.
Aromatic rings in the heme pocket of globins are generally found with sufficient local flexibility to yield only rotationally averaged 1 H NMR signals (37,73), despite the apparent close packing suggested by the crystal structures. Thus, Phe(CD1) is generally found packed tightly against the heme surface but nevertheless exhibits an 1 H NMR spectrum that is rapidly averaged by the 180°ring flips. The Tyr 30 (B10) ring exhibits an averaged NMR spectrum, but the rotation contributes significantly to the linewidth, and standard analysis in the fast exchange limit (57) using the ␦ dip (calc) for the individual C ⑀ Hs results in a rotation rate of ϳ1 MHz. A comparison can be made to globins with Phe rather than Tyr(B10) and with a Gln(E7), i.e. elephant Mb and the sperm whale Leu 29 (B10) 3 Phe/ His 64 (E7) 3 Gln and Leu 29 (B10) 3 Phe/His 64 (E7) 3 Gln/ Val 68 (E11) 3 Phe Mb mutants, for which the B10 ring exhibited "normal" linewidth indicative of much faster reorientation (39,51). Whether the constraints on the Tyr 30 (B10) ring in A. suum Hb result from "pinning down" the extremity via the H bond to the ligand or from the tight van der Waals' contacts with the aromatic ring is not known but could be elucidated in a comparison of the solution 1 H NMR spectra of WT and Tyr(B10) 3 Phe A. suum mutant Hb.
Conclusions-The present NMR data provide support that the heme pocket of A. suum Hb is highly constrained, as evidenced by larger tilts from the heme normal for Fe-CN than previously observed and slow reorientation of the Tyr 33 (B10) ring. The heme is shown to be rotated by 180°about the ␣,␥-meso axis relative to that originally reported in the crystal (8), and the pattern of heme methyl contact shifts is shown to be consistent with the deduced heme orientation. The Tyr 33 (B10) and Gln 64 (E7) side chain labile protons in met HbCN are located at essentially the same positions as found in the HbO 2 crystal and hence provide H bonds to the bound cyanide and establish that the metHbCN is a valuable structural model for aspects of both HbO 2 and HbCO. However, although the Fe ϩ3 -CN unit can serve as limited structural 4 F. S. Mathews, personal communication.  OE), and A. suum hemoglobin (q). The line represents a perfect counter-rotation model. models for the Fe ϩ2 -CO and Fe ϩ2 -O 2 units in globins, cyanide ligation rates unfortunately are not functionally relevant to O 2 or CO binding. This is due to the fact that free cyanide at physiologic pH range is protonated, so that both the on-and off-rates involve protonation/deprotonation steps that are strongly influenced by local pocket polarity that modulates the cyanide pK. Thus the cyanide on-and off-rates directly relate to neither distal steric nor H bonding effects (28,29).