Mechanism of the phenylpyruvate tautomerase activity of macrophage migration inhibitory factor: properties of the P1G, P1A, Y95F, and N97A mutants.

Phenylpyruvate tautomerase (PPT) has been studied periodically since its activity was first described over forty years ago. In the last two years, the mechanism of PPT has been investigated more extensively because of the discovery that PPT is the same protein as the immunoregulatory cytokine known as macrophage migration inhibitory factor (MIF). The mechanism of PPT is likely to involve general base-general acid catalysis. While several lines of evidence implicate Pro-1 as the general base, the identity of the general acid remains unknown. Crystal structures of MIF with the competitive inhibitor (E)-2-fluoro-p-hydroxycinnamate bound in the active site and that of the protein complexed with the enol form of a substrate, (p-hydroxyphenyl)pyruvate, suggest that Tyr-95 is the only candidate in the vicinity that can function as a general acid catalyst. Although Tyr-95 is nearby the bound inhibitor and substrate, it is not within hydrogen bonding distance of either ligand. In this study, Tyr-95 was mutated to phenylalanine, and the kinetic and structural properties of the Y95F mutant were determined. This alteration produces a fully active enzyme, which shows no significant structural changes in the active site. The results indicate that Tyr-95 does not function as the general acid catalyst in the reaction catalyzed by wild-type PPT. The mechanism of PPT was studied further by constructing and characterizing the kinetic properties of two mutants of Pro-1 (P1G and P1A) and one mutant of Asn-97 (N97A). The mutation of Asn-97, a residue implicated in the binding of the phenolic hydroxy group of the keto and enol isomers of (p-hydroxyphenyl)pyruvate and of (E)-2-fluoro-p-hydroxycinnamate affects only the binding affinity of the inhibitor. However, the mutations of Pro-1 have a profound effect on the values of k(cat) and k(cat)/K(m) and clearly show that Pro-1 is a critical residue in the reaction. The results are discussed in terms of a mechanism in which Pro-1 functions as both the general acid and the general base catalyst.

Expression, purification, and structural analysis of the trimeric form of the catalytic domain of the Escherichia coli dihydrolipoamide succinyltransferase.

The dihydrolipoamide succinyltransferase (E2o) component of the alpha-ketoglutarate dehydrogenase complex catalyzes the transfer of a succinyl group from the S-succinyldihydrolipoyl moiety to coenzyme A. E2o is normally a 24-mer, but is found as a trimer when E2o is expressed with a C-terminal [His]6 tag. The crystal structure of the trimeric form of the catalytic domain (CD) of the Escherichia coli E2o has been solved to 3.0 A resolution using the Molecular Replacement method. The refined model contains an intact trimer in the asymmetric unit and has an R-factor of 0.257 (Rfree = 0.286) for 18,699 reflections between 10.0 and 3.0 A resolution. The core of tE2oCD (residues 187-396) superimposes onto that of the cubic E2oCD with an RMS difference of 0.4 A for all main-chain atoms. The C-terminal end of tE2oCD (residues 397-404) rotates by an average of 37 degrees compared to cubic E2oCD, disrupting the normal twofold interface. Despite the alteration of quaternary structure, the active site of tE2oCD shows no significant differences from that of the cubic E2oCD, although several side chains in the active site are more ordered in the trimeric form of E2oCD. Analysis of the available sequence data suggests that the majority of E2 components have active sites that resemble that of E. coli E2oCD. The remaining E2 components can be divided into three groups based on active-site sequence similarity. Analysis of the surface properties of both crystal forms of E. coli E2oCD suggests key residues that may be involved in the protein-protein contacts that occur between the catalytic and lipoyl domains of E2o.

Crystal structure of human ornithine decarboxylase at 2.1 A resolution: structural insights to antizyme binding.

The polyamines spermidine and spermine are ubiquitous and required for cell growth and differentiation in eukaryotes. Ornithine decarboxylase (ODC, EC 4.1.1.17) performs the first step in polyamine biosynthesis, the decarboxylation of ornithine to putrescine. Elevated polyamine levels can lead to down-regulation of ODC activity by enhancing the translation of antizyme mRNA, resulting in subsequent binding of antizyme to ODC monomers which targets ODC for proteolysis by the 26S proteasome. The crystal structure of ornithine decarboxylase from human liver has been determined to 2.1 A resolution by molecular replacement using truncated mouse ODC (Delta425-461) as the search model and refined to a crystallographic R-factor of 21.2% and an R-free value of 28.8%. The human ODC model includes several regions that are disordered in the mouse ODC crystal structure, including one of two C-terminal basal degradation elements that have been demonstrated to independently collaborate with antizyme binding to target ODC for degradation by the 26S proteasome. The crystal structure of human ODC suggests that the C terminus, which contains basal degradation elements necessary for antizyme-induced proteolysis, is not buried by the structural core of homodimeric ODC as previously proposed. Analysis of the solvent-accessible surface area, surface electrostatic potential, and the conservation of primary sequence between human ODC and Trypanosoma brucei ODC provides clues to the identity of potential protein-binding-determinants in the putative antizyme binding element in human ODC.

Structure of mammalian ornithine decarboxylase at 1.6 A resolution: stereochemical implications of PLP-dependent amino acid decarboxylases.

BACKGROUND: Pyridoxal-5'-phosphate (PLP) dependent enzymes catalyze a broad range of reactions, resulting in bond cleavage at C alpha, C beta, or C gamma carbons of D and L amino acid substrates. Ornithine decarboxylase (ODC) is a PLP-dependent enzyme that controls a critical step in the biosynthesis of polyamines, small organic polycations whose controlled levels are essential for proper growth. ODC inhibition has applications for the treatment of certain cancers and parasitic ailments such as African sleeping sickness. RESULTS: The structure of truncated mouse ODC (mODC') was determined by multiple isomorphous replacement methods and refined to 1.6 A resolution. This is the first structure of a Group IV decarboxylase. The monomer contains two domains: an alpha/beta barrel that binds the cofactor, and a second domain consisting mostly of beta structure. Only the dimer is catalytically active, as the active sites are constructed of residues from both monomers. The interactions stabilizing the dimer shed light on its regulation by antizyme. The overall structure and the environment of the cofactor are compared with those of alanine racemase. CONCLUSIONS: The analysis of the mODC' structure and its comparison with alanine racemase, together with modeling studies of the external aldimine intermediate, provide insight into the stereochemical characteristics of PLP-dependent decarboxylation. The structure comparison reveals stereochemical differences with other PLP-dependent enzymes and the bacterial ODC. These characteristics may be exploited in the design of new inhibitors specific for eukaryotic and bacterial ODCs, and provide the basis for a detailed understanding of the mechanism by which these enzymes regulate reaction specificity.

Crystal structure of macrophage migration inhibitory factor complexed with (E)-2-fluoro-p-hydroxycinnamate at 1.8 A resolution: implications for enzymatic catalysis and inhibition.

Macrophage migration inhibitory factor (MIF) exhibits dual activities. It acts as an immunoregulatory protein as well as a phenylpyruvate tautomerase. To understand better the relationship between these two activities and to elucidate the structural basis for the enzymatic activity, a crystal structure of a complex between murine MIF and (E)-2-fluoro-p-hydroxycinnamate, a competitive inhibitor of the tautomerase activity, has been determined to 1.8 A resolution. The structure is nearly superimposable on that of the free protein indicating that the presence of the inhibitor does not result in any major structural changes. The inhibitor also confirms the location of the active site in a hydrophobic cavity containing the amino-terminal proline. Within this cavity, the inhibitor interacts with residues from adjacent subunits. At the back of the cavity, the side-chain carbonyl oxygen of Asn-97' interacts with the phenolic hydroxyl group of the inhibitor while at the mouth of the cavity the ammonium group of Lys-32 interacts with a carboxylate oxygen. The other carboxylate oxygen of the inhibitor interacts with Pro-1. The hydroxyl group of Tyr-95' interacts weakly with the fluoro group on the inhibitor. The hydrophobic side chains of five active-site residues (Met-2, Ile-64, Met-101, Val-106, and Phe-113) and the phenyl moiety of Tyr-95' are responsible for the binding of the phenyl group. Further insight into the enzymatic activity of MIF was obtained by carrying out kinetic studies using the enol isomers of phenylpyruvate and (p-hydroxyphenyl)pyruvate. The results demonstrate that MIF processes the enol isomers more efficiently than the keto isomers primarily because of a decrease in Km. On the basis of these results, a mechanism is proposed for the MIF-catalyzed tautomerization reaction.

The structural and functional analysis of the hemoglobin D component from chicken.

Oxygen binding by chicken blood shows enhanced cooperativity at high levels of oxygen saturation. This implies that deoxy hemoglobin tetramers self-associate. The crystal structure of an R-state form of chicken hemoglobin D has been solved to 2.3-A resolution using molecular replacement phases derived from human oxyhemoglobin. The model consists of an alpha2 beta2 tetramer in the asymmetric unit and has been refined to a R-factor of 0.222 (R-free = 0.257) for 29,702 reflections between 10.0- and 2.3-A resolution. Chicken Hb D differs most from human oxyhemoglobin in the AB and GH corners of the alpha subunits and the EF corner of the beta subunits. Reanalysis of published oxygen binding data for chicken Hbs shows that both chicken Hb A and Hb D possess enhanced cooperativity in vitro when inositol hexaphosphate is present. The electrostatic surface potential for a calculated model of chicken deoxy-Hb D tetramers shows a pronounced hydrophobic patch that involves parts of the D and E helices of the beta subunits. This hydrophobic patch is a promising candidate for a tetramer-tetramer interface that could regulate oxygen binding via the distal histidine.

Three-dimensional structure of the Gly121Tyr dimeric form of ornithine decarboxylase from Lactobacillus 30a.

Ornithine decarboxylases catalyze the conversion of ornithine to putrescine at the beginning of the polyamine pathway. Ornithine decarboxylase (ODC) from Lactobacillus 30a is a 990612 Da dodecamer composed of six homodimers. A single point mutation (Gly121Tyr) was found to prevent association of dimers into dodecamers. The dimeric protein has been crystallized at pH 7.0 in the presence of guanosine triphosphate (GTP). Crystals belong to space group P3(2)21, with unit-cell parameters a = 111.8, c = 135.9 A and one monomer in the asymmetric unit. The structure was determined by molecular replacement and refined using simulated annealing to R = 0.211 at 2. 7 A resolution. The GTP-binding site was analyzed in detail. The protein exhibits a novel binding mode for GTP which is different from that seen in most G-proteins or GTPases. Central to this binding scheme appear to be three lysines, Lys190, Lys374 and Lys382, which form salt bridges with the three phosphates, and Thr191, which hydrogen bonds with the guanine base. Furthermore, the structure suggests that there is some flexibility in the wing domain, which can change its orientation as the protein adapts to its environment. The active site is similar to that of the native enzyme, consistent with the observation that the enzyme activity does not depend on its dodecameric state.

Crystal structure of 4-oxalocrotonate tautomerase inactivated by 2-oxo-3-pentynoate at 2.4 A resolution: analysis and implications for the mechanism of inactivation and catalysis.

The crystal structure of 4-oxalocrotonate tautomerase (4-OT) inactivated by the active site-directed irreversible inhibitor 2-oxo-3-pentynoate (2-OP) has been determined to 2.4 A resolution. The structure of the enzyme covalently modified at Pro-1 by the resulting 2-oxo-3-pentenoate adduct is nearly superimposable on that of the free enzyme and confirms that the active site is located in a hydrophobic region surrounding Pro-1. Both structures can be described as a trimer of dimers where each dimer consists of a four-stranded beta-sheet with two antiparallel alpha-helices on one side. Examination of the structure also reveals noncovalent interactions between the adduct and two residues in the active site. The epsilon and eta nitrogens of the guanidinium side chain of Arg-39" from a neighboring dimer interact respectively with the C-2 carbonyl oxygen and one C-1 carboxylate oxygen of the adduct while the side chain of Arg-61' from the same dimer as the modified Pro-1 interacts with the C-1 carboxylate group in a bidentate fashion. An additional interaction to the 2-oxo group of the adduct is provided by one of the two ordered water molecules within the active site region. These interactions coupled with the observation that 2-oxo-3-butynoate is a more potent irreversible inhibitor of 4-oxalocrotonate tautomerase than is 2-OP suggest that Arg-39" and the ordered water molecule polarize the carbonyl group of 2-OP which facilitates a Michael reaction between Pro-1 and the acetylene compound. On the basis of the crystal structure, a mechanism for the enzyme-catalyzed reaction is proposed.

Evolution of cooperativity in hemoglobins: what can invertebrate hemoglobins tell us?

While vertebrate hemoglobins typically are tetrameric and show highly regulated and cooperative ligand binding, little is known of the evolution of these properties. We are studying the structural and functional properties of the hemoglobins from Caudina arenicola, an echinoderm. The echinoderms are in the lineage most closely related to the vertebrates to express hemoglobin. C. arenicola has three sets of red cells, in the water vascular system, the coelomic cavity, and in an intestinal vein. Each of these expresses a distinct array of globins. The hemoglobins are cooperative and exhibit unusual ligand-linked associative properties, being dimeric when oxygenated and forming tetramers and higher aggregates on deoxygenation. The major coelomic hemoglobins have been subjected to a detailed examination by a combination of ligand binding analyses and protein and DNA sequencing, as well as X-ray crystallography. Two typical globin introns were identified, along with a unique intron that bisects an N-terminal extension of the globin from the remainder of the gene. X-ray crystallographic analysis shows that the subunit interfaces of C. arenicola hemoglobins differ radically from those of vertebrate hemoglobins and indeed from some other invertebrate hemoglobins, but closely resemble the packing arrangements found in a clam hemoglobin (Scapharca). However, the residues implicated in cooperativity in these two types of hemoglobins differ substantially.

Crystal structure of the truncated cubic core component of the Escherichia coli 2-oxoglutarate dehydrogenase multienzyme complex.

The dihydrolipoamide succinyltransferase (E2o) component of the 2-oxoglutarate dehydrogenase multienzyme complex is composed of 24 subunits arranged with 432 point group symmetry. The catalytic domain (CD) of the E2o component catalyzes the transfer of a succinyl group from the S-succinyldihydrolipoyl moiety to coenzyme A. The crystal structure of the Escherichia coli E2oCD has been solved to 3.0 A resolution using molecular replacement phases derived from the structure of the catalytic domain from the Azotobacter vinelandii dihydrolipoamide acetyltransferase (E2pCD). The refined model of the E. coli E2oCD consists of residues 172 to 404 and has an R-factor of 0.205 (Rfree=0.249) for 9696 reflections between 20.0 and 3.0 A resolution. Although both E2oCD and E2pCD form 24mers, subtle changes in the orientations of two helices in E2oCD increase the stability of the E2oCD 24mer in comparison to the less stable A. vinelandii E2pCD 24mer. Like E2pCD and chloramphenicol acetyltransferase (CAT), the active site of E2oCD is located in the middle of a channel formed at the interface between two 3-fold related subunits. Two of the active-site residues (His375 and Thr323) have a similar orientation to their counterparts in E2pCD and CAT. A third catalytic residue (Asp379) assumes a conformation similar to the corresponding residue in E2pCD (Asn614), but different from its counterpart in CAT (Asp199). Binding of the substrates to E2oCD is proposed to induce a change in the conformation of Asp379, allowing this residue to form a salt bridge with Arg184 that is analogous to that formed between Asp199 and Arg18 in CAT. Computer models of the active site of E2o complexed with dihydrolipoamide and with coenzyme A led to the identification of the probable succinyl-binding pocket. The residues which form this pocket (Ser330, Ser333, and His348) are probably responsible for E2o's substrate specificity.

The GTP effector site of ornithine decarboxylase from Lactobacillus 30a: kinetic and structural characterization.

A nucleotide effector site of the biodegradative form of ornithine decarboxylase from Lactobacillus 30a (OrnDC L30a) has been identified. OrnDC L30a activity at pH 8.0, where the enzyme is normally inactive, is stimulated by GTP and dGTP and to a lesser extent by GDP but not by ATP, CTP, or UTP. The pH profile indicates that activation by GTP is reflected by an increase in kcat/KM,orn (above pH 6.8), while Vmax remains constant over the pH range 4.0-9. 0. Scatchard plot analysis shows that GTP binds to OrnDC L30a at both pH 5.8 (KD = 0.11 microM) and pH 8.0 (KD = 1.6 microM), but unexpectedly, half-site binding is observed at the higher pH. The OrnDC L30a dodecamer dissociates into dimers at high pH in the presence or absence of GTP. The GTP binding site was located in difference electron density maps using low-resolution X-ray data. This represents a new type of GTP binding site. A model explaining the activation of OrnDC L30a by GTP is presented.

Crystallization of a mammalian ornithine decarboxylase.

Crystals of truncated (delta425-461) pyridoxal-5'-phosphate (PLP)-dependent mouse ornithine decarboxylase (mOrnDC') have been obtained that diffract to 2.2 angstroms resolution (P2(1)2(1)2, a = 119.5 angstroms, b = 74.3 angstroms, c = 46.1 angstroms). OrnDC produces putrescine, which is the precursor for the synthesis of polyamines in eukaryotes. Regulation of activity and understanding of the mechanism of action of this enzyme may aid in the development of compounds against cancer. mOrnDC is a member of group IV PLP-dependent decarboxylases, for which there are no known representative structures.

Crystallographic structure of a PLP-dependent ornithine decarboxylase from Lactobacillus 30a to 3.0 A resolution.

Ornithine decarboxylase from Lactobacillus 30a (L30a OrnDC) is representative of the large, pyridoxal-5'-phosphate-dependent decarboxylases that act on lysine, arginine or ornithine. The crystal structure of the L30a OrnDC has been solved to 3.0 A resolution using MIR phases in combination with density modification (space group P6; a = 195.6 A, c = 97.6 A; dimer of 1460 amino acid residues/asymmetric unit; VM = 3.26 A3/Da). The refined crystallographic R-value was 0.219 (Rfree = 0.268) using 2-fold restraints with a 4 sigma cutoff and 8.0 to 3.0 A resolution data. Six dimers related by C6 symmetry compose the enzymatically active dodecamer (approximately 10(6) Da). Each monomer of L30a OrnDC can be described in terms of five sequential folding domains. The amino-terminal domain, residues 1 to 107, consists of a five-stranded beta-sheet termed the "wing" domain. Two wing domains of each dimer project inward towards the center of the dodecamer and contribute to dodecamer stabilization. The "linker" domain, residues 108 to 160, consists of short alpha-helices separated by a loop that fills in the PLP pocket. The third domain, residues 161 to 413, is an alpha/beta domain containing a seven stranded beta-sheet that resembles the PLP-binding domain of the aspartate aminotransferases. The fourth domain, residues 414 to 569, resembles the "small" domain of the aspartate aminotransferases, but is significantly larger due to insertions. The remaining carboxy-terminal domain, residues 570 to 730, is organized into multiple antiparallel loops and seven alpha-helices that help form a deep channel leading to the PLP-binding site.

Three-dimensional structure of a hemichrome hemoglobin from Caudina arenicola.

The structure of a monomeric hemichrome form of an invertebrate hemoglobin, Hb-C chain, from Caudina arenicola has been refined to an R value of 0.16 using the data from 5.0 to 2.5 A resolution (R = 0.21 from 10.0 to 2.5 A resolution). Hb-C crystallizes in space group P2(l) with cell constants a = 45.74, b = 45.23 and c = 40.92 A and beta = 104.4 degrees with two monomers packed in the unit cell (V(m) = 2.34 A(3) Da(-1)). The phases were determined by the multiple isomorphous replacement method with Hg(2+) the major derivative. The structure consists of 157 amino acids with N- and C-terminal regions and eight alpha-helices forming a heme pocket. The unique feature of this structure is the hemichrome form with the proximal and distal histidines coordinated to the heme Fe atom, which is nearly in the plane of the porphyrin ring. A total of 111 solvent molecules were added to the structure using difference density peaks of at least 3sigma over background. Interestingly, all the heme groups present in the crystal are nearly coplanar.

X-ray structure determination of a dimeric hemoglobin from Caudina arenicola.

The X-ray structure of a dimeric, cyanomet-liganded hemoglobin D-chain (Hb-D) from Caudina arenicola has been determined by the molecular-replacement method. The search model was a concatenated model of three hemoglobin structures superimposed on the backbone of monomeric, hemichrome hemoglobin C-chain (Hb-C) from the same organism. Hb-D crystallizes in space group P4(1)2(1)2 with cell constants a = b = 77.0 and c =61.5 A with one subunit in the asymmetric unit. The dimer twofold axis corresponds to a crystallographic twofold along one of the body diagonals of the unit cell. Rotation and translation searches as well as model refinement were carried out in X-PLOR with the final model having an R value of 0.19 using the data from 5.0 to 2.9 A resolution (R = 0.26 for 10.0 to 2.9 A resolution). The homodimeric structure of Caudina Hb-D features close heme-heme contacts with an Fe-Fe distance of 19.0 A. The subunit-subunit interface involves both the E and F helices from each subunit with the E helices oriented antiparallel at 50 degrees with respect to one another, similar to the quaternary structure observed for the homodimeric hemoglobin from Scapharca inaequivalvis.

Structural analysis of monomeric hemichrome and dimeric cyanomet hemoglobins from Caudina arenicola.

The X-ray structures of two hemoglobins (Hb) from the sea cucumber Caudina arenicola (an echinoderm) have been determined: a low spin, hemichrome, monomeric Hb-C chain, and a cyanomet-liganded dimeric Hb-D chain. Attempts to obtain crystal structures of the deoxy-liganded and hemichrome forms from the same chain type have not been successful. In this work, the Hb-C chain and Hb-D chain structures are compared, and differences observed in tertiary structure related to the different ligand states for hemoglobin chains from this organism. In addition to shifts of the distal histidine and E helix, differences are noted in the position of the heme group within the heme pocket, the hydrogen bonding of the heme group to the protein, and the status of the D helix. These differences are important in understanding the ligand-linked association states of these hemoglobins. The quaternary structure of the Hb-D homodimer is compared with those from two other invertebrate hemoglobins from Scapharca inaequivalvis and Urechis caupo, which also have subunit-subunit interactions that involve the E and E' helices. The dimer interactions of the Caudina and Urechis hemoglobins are quite dissimilar. However, the dimer interface observed in cyanomet Hb-D is strikingly similar to that observed for the carbonmonoxy hemoglobin dimer from the clam, Scapharca, yet many of the key amino acid residues implicated in the cooperative mechanism of the Scapharca hemoglobin are not conserved in the Caudina hemoglobins.

Structural motifs for pyridoxal-5'-phosphate binding in decarboxylases: an analysis based on the crystal structure of the Lactobacillus 30a ornithine decarboxylase.

Two of the five domains in the structure of the ornithine decarboxylase (OrnDC) from Lactobacillus 30a share similar structural folds around the pyridoxal-5'-phosphate (PLP)-binding pocket with the aspartate aminotransferases (AspATs). Sequence comparisons focusing on conserved residues of the aligned structures reveal that this structural motif is also present in a number of other PLP-dependent enzymes including the histidine, dopa, tryptophan, glutamate, and glycine decarboxylases as well as tryptophanase and serine-hydroxymethyl transferase. However, this motif is not present in eukaryotic OrnDCs, the diaminopimelate decarboxylases, nor the Escherichia coli or oat arginine decarboxylases. The identification and comparison of residues involved in defining the different classes are discussed.

Sequence of ornithine decarboxylase from Lactobacillus sp. strain 30a.

A gene encoding biodegradative ornithine decarboxylase from Lactobacillus sp. strain 30a was isolated from a genomic DNA library and sequenced. Primer extension analysis revealed two transcription initiation sites. The deduced amino acid sequence is compared with the amino acid sequences of five previously reported bacterial decarboxylases, and conserved pyridoxal phosphate motif residues are identified.

Crystallization of biosynthetic arginine decarboxylase from Escherichia coli.

Putrescine is the immediate precursor for the synthesis of polyamines and is normally generated by the action of ornithine decarboxylase. However, putrescine can also be produced by the conversion of arginine to agmatine by arginine decarboxylase (bADC) followed by the release of urea by agmatine ureohydrolase. Amino-acid sequence homology with the eukaryotic ornithine decarboxylases suggests that bADC may be a model for this group of decarboxylases. We report here the crystallization of arginine decarboxylase from E. coli. Crystals up to 1 mm in size are grown by vapor equilibration using Li(2)SO(4) and polyethylene glycols as precipitants. The crystals exhibit diffraction maxima beyond 3 A resolution and belong to space group P4(1(3))2(1)2 with a = 192.4 and c = 121.0 A. These unit-cell dimensions together with the estimated density of the crystals suggest the presence of one tetramer of bADC (71 kDa subunit(-1)) per asymmetric unit (V(m) = 2.0 A(3) Da(-1)).

Structural analysis of Urechis caupo hemoglobin.

The structure of Urechis caupo hemoglobin in the cyanomet state has been refined to R = 0.148 at 2.5 A resolution. Although the tertiary structure is similar to that of other vertebrate and invertebrate hemoglobins the quaternary structures of this tetramer is unique as suggested by the earlier determination of the 5.0 A resolution structure. The G and H helices of the hemoglobin are on the outside of the tetramer facing the solvent in contrast to human hemoglobin where the G and H helices form inter-subunit contacts. A substantial number of tightly bound water molecules help mediate interactions between subunits. The unusual arrangement of subunits is consistent with the general lack of co-operativity of oxygen uptake for Urechis caupo hemoglobin.