Solution of 1H NMR structure of the heme cavity in the oxygen-avid myoglobin from the trematode Paramphistomum epiclitum.

A two-dimensional 1H NMR study has been carried out on the heme cavity of the extreme oxygen-avid and autoxidation-resistant oxy-myoglobin complex from the trematode Paramphistomum epiclitum, and the residues were identified which potentially provide hydrogen bond stabilization for the bound oxygen. Complete assignment of the heme core resonances allows the identification of 10 key heme pocket residues, 4 Phe, 4 Tyr, and 2 upfield ring current aliphatic side chains. Based solely on the conserved myoglobin folding topology that places the E helix-heme crossover and the completely conserved Phe(CD1)-heme contact at opposing meso positions, the heme orientation in the cavity and the E helix alignment were unambiguously established that place Tyr66 at position E7. Moreover, all eight aromatic and the two aliphatic side chains were shown to occupy the positions in the heme cavity predicted by amino acid sequence alignment with globins of known tertiary structure. The dipolar contacts for the Tyr32(B10) and Tyr66(E7) rings indicate that both residues are oriented into the heme cavity, which is unprecedented in globins. The ring hydroxyl protons for both Tyr are close to each other and in a position to provide hydrogen bonds to the coordinated oxygen, as supported by strong retardation of their exchange rate with bulk solvent. A more crowded and compact structure increases the dynamic stability of the distal pocket and may contribute to the autoxidation resistance of this myoglobin.

Myoglobin and hemoglobin are oxygen binding proteins that reflect extraordinary structural homology in spite of often very limited sequence homology (1,2). Both are composed of ϳ150residue globular proteins containing eight helices (labeled A-H) and an iron-protoporphyrin-IX (heme) bound to the completely conserved His(F8); the only other completely conserved residue is Phe(CD1). Among the vertebrate Hb 1 s/Mbs, the strongly conserved His, but also Gln, provide the crucial hydrogen bond to stabilize the bound O 2 (3)(4)(5)(6). Identified natural human Hb mutants with Tyr at position E7 do not bind oxygen but convert to the Met or ferric form under physiological conditions (7). Hbs and Mbs from invertebrates exhibit a much broader range of functionality and show key substitutions at position E7 that are unknown among vertebrates and provide hydrogen bonds to the bound O 2 via residues at positions other than E7 (2). Prominent among these globins are the Mb from the sea hare Aplysia limacina, which, with a Val(E7), provides the key hydrogen bond via Arg(E10) (8), and the parasitic trematode, Ascaris suum, HbI, which provides two hydrogen bonds to the bound O 2 via a relatively common Gln(E7) and an uncommon Tyr(B10) (9,10). Tyr(B10) is present in several other trematode Hbs/Mbs (11,12), among others (13)(14)(15).
The Tyr(B10) hydrogen bond has been identified as the source of the extra stabilization that leads to extreme O 2 avidity in A. suum HbI through primarily an extremely slow O 2 off-rate (9,12); the autoxidizability is similar to that of common Mbs/Hbs. The Tyr(B10) is present in Lucina pectinata Hbs (13,14), A. suum Mb (12), and legume Hb (16,17) but does not result in similarly slow O 2 off-rates and likely does not provide a hydrogen bond to the O 2 . 1 H NMR studies of the nematode Dicrocoelium dendriticum Hb had identified a Tyr hydrogenbonded to the bound ligand (18,19). Hence it is clear that Tyr at position B10, and possibly E7, can strongly stabilize the Fe-O 2 bond in some cases but not in others. Engineering a Tyr(B10) and Gln(E7) into sperm whale Mb to mimic A. suum HbI failed to reproduce the high oxygen affinity for the latter Hb but demonstrated that Tyr(B10) provides H bond stabilization at the bound O 2 (20).
Alignment of the amino acid sequence of the globin from the trematode Paramphistomum epiclitum clearly reveals Tyr at position B10 and places, unprecedented, Tyr 66 rather than His 65 at position E7 (21), as shown in Fig. 1. P. epiclitum Mb has a high oxygen affinity created by a high "on-rate" and is rather resistant to autoxidation. 2 In order to shed light on the nature of the active site in this unusual P. epiclitum MbO 2 complex, unequivocally establish the alignment of the E helix, and determine the potential role of Tyr(E7) and Tyr(B10) interacting with bound oxygen, we report herein on a solution 1 H NMR study of its heme cavity. Our interests are to assess the degree to which homologous residues on the various helices make contact with the heme when compared with A. suum HbI (9, 10) or sperm whale Mb crystal structures (22) and as suggested by the molecular modeling (21). The limited availability of the protein dictated that a choice had to be made between sample concentration and sample purity for the desired twodimensional 1 H NMR study. It was concluded that the needed ϳ1.5 mM concentration was possible only for a sample ϳ60% pure, with the remainder one or more of the minor Mb isoforms. This sample heterogeneity limits the scope of the study but not the details to which the positions of the identified heme cavity residues can be placed relative to each other and to the heme.

EXPERIMENTAL PROCEDURES
Protein Purification-P. epiclitum (Platyhel-minthes, Trematoda, Paramphistomatidae) Mb was prepared by ammonium sulfate precipitation and gel filtration chromatography as described elsewhere (21). The resulting preparation is homogeneous in M r as evidenced by SDSpolyacrylamide gel electrophoresis and equilibrium centrifugation. Separation by isoelectric focusing under native conditions reveals two major isoforms differing in less than 0.1 pH unit which could not be purified further. Isoelectric focusing under denaturing conditions reveals four different globin isoforms. It is not known if these are real isoforms, coded by different genes, or if they are the result of posttranslational modifications or artificial modifications during the isolation process. Due to its high oxygen avidity, the purified protein remains in the oxy-Mb form as shown by spectral analysis (21). The final oxy-Mb sample contained ϳ1.5 mM total heme concentration in 50 mM NaCl and 50 mM phosphate buffer at pH 7.1. The 1 H 2 O solution was subsequently converted to 2 H 2 O solution using an Amicon ultrafiltration cell. Solution pH was adjusted with NaO 1 H (NaO 2 H) or 1 HCl ( 2 HCl) solution.
NMR Spectra-1 H NMR data were collected on a GE Omega-500 spectrometer operating at 500 MHz for protein samples in both 1 H 2 O and 2 H 2 O. The observed chemical shifts are referenced to 2,2Ј-dimethyl-2-silapentane-5-sulfonate (DSS) via the solvent signal. Solvent suppression, when necessary, was achieved by direct saturation during the relaxation delay. Phase-sensitive TOCSY (23,24) and NOESY (25) were collected (512 t 1 ϫ 2048 t 2 ) at 30°C in order to identify scalar and dipolar connectivities among heme and amino acid residues. Spectral widths of 8.0 kHz and mixing times of 35 ms for TOCSY and 150 ms for NOESY were used. The 90°pulse width was ϳ8 s (ϳ25 s for TOCSY). 128 scans were collected for each block with a repetition time of 1.2 s. Two-dimensional data sets were processed by FELIX software on a Silicon Graphics work station. Both NOESY and TOCSY spectra were processed by a 15°-shifted sine bell-squared apodization and zero-filled to 2048 ϫ 2048 points prior to Fourier transformation.  Fig. 2. Inspection of both the low field and high field resolved portions of the spectrum reveals a molecular heterogeneity indicative of two obvious compounds, in relative ratio ϳ2:3. In the low field spectrum, the three likely meso-H candidates differ in intensity with the extreme low field peak both weaker and significantly broader than the other two equal intensity resolved low field peaks. Similarly, the high field portion of the spectrum reveals several ring-current shifted peaks with intensity for methyl groups for both the major and minor components. Contributions from signals from both components in the composite peak centered at Ϫ0.8 ppm are partially resolved at 38°C (Fig. 2B). Our interest is in the peak assignment and structure elucidation of solely the major component whose residue origin in Fig.   2 are labeled by the one-letter amino acid code with the deduced helical positions given in parentheses. Resonances for any minor component are simply labeled X 1 ; the possible relationship of the minor to the major component will be addressed briefly below. The determination of the alignment of the E helix and nature of the interaction of distal residues with the bound ligand are pursued by identifying contacts to the heme, with emphasis on resolved aliphatic residue peaks in the ring-current upfield spectral window and the aromatic side chains, each of which can be readily identified by highly characteristic TOCSY cross-peak patterns even in an impure sample. To guarantee that all assignments relate solely to the major isomer of interest, we report assignments only when we observe multiple dipolar contacts between pairs of residues or between a residue and two or more heme substituents.

Sample
Heme Assignment-The use of the heme as a "template" on which to place the upfield aliphatic and low field aromatic side chains first requires unambiguous assignment of the heme resonances. The low field 8 -11 ppm portion of the NOESY spectrum in Fig. 3 reveals four nonlabile proton peaks which fail to exhibit TOCSY peaks and hence must arise from meso-Hs; two are resolved (the heme positions are labeled in Fig. 4). The four meso-Hs exhibit strong NOESY cross-peaks to two, one, one, and no likely heme-methyls in the 2.5-3.8 ppm window (Fig. 3B) which also fail to exhibit TOCSY cross-peaks. These dipolar contacts therefore differentiate the ␦-meso-H (nuclear Overhauser effects to 1-CH 3  of the 1-CH 3 , 8-CH 3 pair identify 1-CH 3 and the 2-vinyl group and, by inference, 8-CH 3 and the 4-vinyl group (Fig. 5B). The NOESY cross-peaks of the 2-vinyl group to a meso-H differentiate ␣-meso-H from ␤-meso-H and identify both the 3-CH 3 and 5-CH 3 (Fig. 5B) The remaining ␥-meso-H exhibits NOESY cross-peaks to two sets of spin-coupled protons near 3.5 ppm (Fig. 5B) which exhibit NOESY cross-peaks to 5-CH 3 or 8-CH 3 . These identify the H ␣ s for the two propionate groups and complete the heme ring core substituent assignment in a unique and self-consistent manner (26,27) that guarantees that the major component alone is being characterized. The chemical shifts for assigned heme resonances are listed in Table I.
Heme Pocket Residues-Assuming the basic Mb fold and a heme orientation about the ␣,␥-meso-H axis for P. epiclitum Mb that is the same as in A. suum HbI (9, 28) and sperm whale Mb (22), the sequence alignment and homology model (21)  Phe 36 (B14) are expected to contact Phe 46 (CD1). Lastly, the upfield shifted E helix position residues Leu 70 (E11) and Ala 73 (E14) are expected to make contact to the pyrrole A/B junction. The expected dispositions relative to the heme for the eight aromatic side chains, Leu 70 (E11), and Ala 73 (E14) is shown schematically in Fig. 4. It is to be noted that, although the assignments for amino acid residues and their placement near the heme and each other are presented for simplicity as confirmations of aspects of the homology model (21), the conclusion on the E helix alignment and conserved folding topology are, in fact, determined independently by 1 H NMR data.
The TOCSY spectrum (Fig. 6A) for the upfield methyl peak at Ϫ1.1 ppm reveals its origin as a Leu, which has dipolar contacts to the heme 1-CH 3 (Fig. 6B), 2-vinyl (Fig. 5D), and ␦-meso-H (Figs. 3C). The TOCSY map (Fig. 6A) also identifies a complete Ala spin system with similarly strong NOESY crosspeaks to the heme 1-CH 3 , 8-CH 3 (Fig. 6B), and ␦-meso-H (not shown). The upfield ring current shifts and contacts to the heme 1-CH 3 and 8-CH 3 are characteristic of Leu 70 (E11) and Ala 73 (E14) in the normal Mb fold if the heme is oriented like in sperm whale Mb (22) or A. suum Hb (9,28). While the peptide NHs for both residues could be located, the small chemical shift dispersion and the resultant spectral congestion precluded tracing the backbone to uniquely elucidate the sequence origin of the two residues; however, the expected NOESY cross-peaks from Ala(E14) C ␤ H 3 to Leu(E11) C ␣ H and C ␦ ЈH 3 (Fig. 6B) confirms their i, i ϩ 3 positions on a helix.
Comparison of NOESY maps in 1 H 2 O and 2 H 2 O reveals the presence of two labile protons with a weak NOESY cross-peak  between them (Fig. 5E), for which the absence of any TOCSY cross-peaks indicates that they likely arise from two Tyr ring hydroxyls (labeled OH1 and OH2 in Fig. 4). OH1 exhibits moderate intensity NOESY cross-peaks to the rings of Tyr 32 (B10) and Tyr 66 (E7), as well as cross-peaks to the ring (C H) of Phe 46 (CD1) (Fig. 5B), and to a C ␦ H 3 of Leu 70 (E11) (Fig.  5D), as shown schematically in Fig. 4. OH2 exhibits moderate intensity NOESY cross-peaks to both the ring of Tyr 66 (E7) (Fig.  5E) and the C ␦ H 3 s of Leu 70 (E11) (Fig. 5E). These NOESY patterns identify OH1 and OH2 as the side chain hydroxyl protons of Tyr 32 (B10) and Tyr 66 (E7), respectively, and places both labile protons well within the heme cavity and in the vicinity of the ligand. The hydroxyl protons of Tyr 94 (F4) and Tyr 42 (C4) could not be located, either because of spectral congestion and/or rapid exchange with solvent. The absence of detectable magnetization transfer to the Tyr 32 (B10) and Tyr 66 (E7) hydroxyl protons upon saturating the solvent resonance dictates that the labile proton lifetime, OH , are much longer (by Ͼ10) than the expected T 1 for the resonances (ϳ250 ms), leading to OH Ն 2 s at pH 7.0 (29). The positions of the 10 identified heme pocket residues relative to the heme and to each other and the observed NOESY cross-peak patterns are depicted schematically in Fig. 4, and the chemical shifts are listed in Table II.
The Minor Component-Sufficient assignments are pursued in order to establish whether the minor component (ϳ40%) arises from the heme reoriented 180°about the ␣,␥-meso axis (30,31), from another polypeptide chain (isoform), or from damage or modification during isolation. The upfield TOCSY map reveals three minor component methyl peaks (labeled X 2 , X 3 , and X 4 in Fig. 2A) which exhibit spin topology indicative of three CH 2 -CH 3 fragments (Fig. 6A) of three different Ile, two of which exhibit NOESY cross-peaks between each other (Fig.  6B), to the low-field resolved meso-H (Fig. 3C) and its adjacent methyl, as well as to an aromatic side chain (not shown). The presence of upfield Leu signals for the major component and Ile signals for the minor component in solution indicates that the heterogeneity is due to the presence of two different polypeptide chains and not due to either heme orientational isomerisms or damage to the major component. Lastly, the significantly broader resolved meso-H signal and strong intensity of NOESY cross-peaks for the minor relative to the major component suggests that the minor component may be oligomeric. Assignments were not pursued further for the minor component.

DISCUSSION
The unique assignment of the heme for the major isoform of P. epiclitum MbO 2 , together with the observation of all aromatic ring contacts to the heme and among each other, as expected by amino acid sequence homology with globins of known tertiary structure, provide direct evidence that the heme pocket architecture is very similar to that of other typical Mbs/Hbs. These data also establish both the alignment of the E  32 (B10) over the heme is supported by both its upfield ring current shifted side chain protons and its NOESY pattern to surrounding residues. The Tyr 32 (B10) ring labile proton also exhibits NOESY cross-peaks to a series of peripheral residues (Phe 46 (CD1), Leu 70 (E11)) that both confirm its position and support the presence of a hydrogen bond between Tyr 32 (B10) and the bound oxygen. The present failure to detect the Tyr 94 (F4) hydroxyl proton is consistent with the predicted orientation (21) of this ring toward the surface rather than in the pocket and parallel to the His 98 (F8) as found for Phe 94 (F4) in Aplysia limacina Mb (8).
The computer model suggests two alternate orientations of Tyr 66 (E7), one turned out and another turned into the heme pocket (21) within hydrogen-bonding distance of the bound oxygen and in contact with both Phe 115 (G12) and Leu 70 (E11). The proposed sequence alignment for the E helix and the present confirmation by 1 H NMR place an unprecedented Tyr at position E7 but do not answer whether it is oriented into the pocket or into solution. The monomeric Chironomus Hbs possess a His(E7), but it is oriented into solution in all functional forms (32). Strong and moderate intensity NOESY cross-peaks from the Tyr 66 (E7) ring to the Leu 70 (E11) methyls (Fig. 5D), one of which is close to the heme 1-CH 3 and both of which exhibit strong NOESY cross-peaks to the Phe 115 (G12) ring over the pyrrole A/B junction (Fig. 5D), clearly place the Tyr 66 (E7) ring well within the distal heme pocket. The NOESY crosspeaks from the Tyr 66 (E7) hydroxyl proton to the Leu 70 (E11) methyls and the Tyr 32 (B10) hydroxyl proton place it in a posi-tion where it could provide the second hydrogen bond to the bound oxygen, as well as form a hydrogen bond to the Tyr(B10) hydroxyl oxygen to stabilize the optimal orientations of both Tyr 32 (B10) and Tyr 66 (E7) for hydrogen bonding to O 2 . The role of both Tyr 32 (B10) and Tyr 66 (E7) hydroxyl protons in forming hydrogen bonds to the ligand is independently supported by the very slow exchange rate with bulk water, with estimated lifetime ϳ2 s at pH 7.1, some 10 4 slower than for a free Tyr at this pH (29). This is the first case where a Tyr is demonstrated not only to occupy position E7 but to orient into the heme pocket of a functional globin in a manner common to His(E7) in mammalian Mbs and Hbs.
The present study, however, does not shed much light on the extraordinary resistance of P. epiclitum MbO 2 to autoxidation when compared with Hbs of other nematodes such as Isoparorchis hypsolbargi, (21) and nematodes such as A. suum (9). Autoxidation of Mb is strongly dependent on hydrogen bonding to the bound ligand by the distal residue and the size and hydrophobicity of the distal side of the heme pocket (5). NMR data on P. epiclitum MbO 2 indicate the formation of two hydrogen bonds by Tyr 32 (B10) and Tyr 66 (E7) to the bound oxygen and establish the presence of an unprecedented number of bulky hydrophobic residues in the crowded distal heme pocket (21). A more crowded distal pocket for P. epiclitum Mb than other Mbs, Hbs is supported by the observed line broadening of the Phe 46 (CD1) ring protons. In spite of its close proximity to the heme, 1 H NMR has shown that this highly conserved aromatic ring undergoes rapid 180°ring reorientation that completely averages the environments of the two H ⑀ (and two H ␦ ) ring protons in common Mbs and Hbs (27,33,34). The significant broadening of the Phe 46 (CD1) H ⑀ resonance (Fig. 2,A and B) (and cross-peaks; Fig. 5B) in P. epiclitum MbO 2 is indicative of slow ring reorientation and suggests a more compact and dynamically more stable distal pocket than in other NMR characterized Hbs and Mbs, which likely contribute to the resistance to autoxidation.
It is clear, however, that the presence of certain residues in the heme pocket (i.e. Tyr(B10)) is insufficient information for predicting oxygen binding and autoxidation resistance. This is illustrated by the A. suum Mb and A. suum HbI, both with Tyr(B10) and Glu(E7), but with differences of a factor 50 in k off for oxygen (9,10,12), and by the P. epiclitum and I. hypselobargi Mbs, which exhibit very different autoxidation rates in spite of both possessing Tyr(B10) and Tyr(E7) (21). Indeed, the complete tertiary structure is involved in fine tuning the orientations of residues to their critical positions. Crystallization experiments with different trematode Mbs are in progress.