Allosteric Hemoglobin Assembly: Diversity and Similarity*

From the ‡Department of Biochemistry and Molecular Pharmacology, University of Massachusetts Medical School, Worcester, Massachusetts 01605, ¶Hematology Division, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts 02115, Institute of Veterinary Physiology, Vetsuisse Faculty of the University of Zurich, CH-8057 Zurich, Switzerland, and **Department of Biology, Pennsylvania State University, University Park, Pennsylvania 16802

of oxygen to the heme iron or can increase affinity by providing favorable electrostatic interactions for a bound oxygen molecule. All three of these mechanisms have been found to contribute to the modulation of oxygen affinity in allosteric hemoglobins (13).

Vertebrate Hemoglobins
The most familiar hemoglobins are, of course, those found in mammalian erythrocytes. Mammalian circulating hemoglobins, assembled into tetramers from two copies each of ␣ and ␤ subunits, played a central role in the history of molecular biology in the 20th century. Pioneering crystallographic studies by Max Perutz provided the methods that revolutionized protein crystallography and also revealed, for the first time, structural transitions that underlie allosteric protein behavior, including large quaternary structural changes (14). This work highlighted roles of the stereochemistry on not only the distal side of the heme but also on the proximal side of the heme in regulating oxygen affinity (14 -16). Despite many elegant models and very extensive study, there is no universal agreement concerning the mechanism of cooperativity by human hemoglobin. One intriguing recent finding is that of Ackers and colleagues suggesting cooperativity within each ␣␤ dimer (17). Although controversial, this idea finds support from recent structural results (18) and suggests an intriguing parallel with cooperative invertebrate hemoglobins, which are assembled from dimeric units that likely possess intrinsic cooperativity.

Invertebrate Allosteric Hemoglobins
Unlike the circulating hemoglobins from higher vertebrates, which invariably display the ␣ 2 ␤ 2 tetrameric form observed in human hemoglobin, invertebrate hemoglobins exhibit remarkable variation in their quaternary assembly. Crystal structures are now available of hemoglobin assemblages ranging in size from dimers to assemblies of 180 subunits (Fig. 2). Generally, hemoglobin assembly into oligomers is coupled with cooperative oxygen binding, but a notable exception is the tetrameric hemoglobin found in the "fat innkeeper" worm Urechis caupo (19).
An intriguing finding from a comparison of allosteric invertebrate hemoglobin assemblages is the recurring presence of a similar dimeric unit, termed an "EF dimer" because of extensive interface contacts involving the E and F helices (13). Isolated EF dimeric hemoglobins have been observed in the mollusc, Scapharca inaequivalvis (20), and the echinoderm, Caudina arenicola (21). The Scapharca HbI homodimer shows significant cooperative ligand binding (22), whereas strong cooperativity requires heterodimeric forms of Caudina hemoglobin (21,23). S. inaequivalvis also possesses a cooperative tetrameric hemoglobin, which is assembled from two EF heterodimers (24) (Fig. 2).

Extracellular Annelid Allosteric Hemoglobins
Much larger allosteric hemoglobins are found among the annelids, which show striking variability in form. The crystal structures of two very large annelid hemoglobins have been reported. The extracellular erythrocruorin, also termed hexagonal bilayer hemoglobin, from the common earthworm, Lumbricus terrestris, is assembled from 144 hemoglobin subunits and 36 non-globin "linker" chains using a hierarchy of symmetrical interactions (25). The hemoglobin subunits are organized into 12 dodecamers, each of which binds to a heterotrimer of linker subunits (25). Each dodecamer is a trimer of heterotetramers, with each heterotetramer assembled from two distinct EF heterodimers (Fig. 3). The C1 hemoglobin from the deep sea hydrothermal vent tubeworm Riftia pachyptila is assembled from 24 subunits, each half of which forms a hemoglobin dodecamer structure that is very similar to the dodecamers observed in the L. terrestris erythrocruorin structure (26,27).
The very similar quaternary assembly of dodecamers in 400-kDa vestimentiferan hemoglobin and 3600-kDa erythrocruorins raises interesting issues concerning the evolution of extracellular annelid hemoglobins. The erythrocruorins require assembly of multiple copies of four different types of hemoglobin subunits and at least three distinct linker subunits 1 organized in a complex hierarchy of symmetry (28,29). The 400-kDa vestimentiferan hemoglobins show a much more straightforward arrangement, in which six copies of four distinct, but similar, hemoglobin subunits are arranged with D 3 symmetry (26). Moreover, the hollow spherical assembly of the vestimentiferan geometry provides a structural rationale for the irregular, half-spherical shape of the hemoglobin dodecamers first observed in Lumbricus erythrocruorin (25, 27).

FIG. 1. Variability in the myoglobin fold.
A ribbon diagram is shown for each protein with the heme encapsulating E and F helices shown in cyan and the rest of the polypeptide course shown in gray. Also shown are the heme groups (red), oxygen or water ligand (yellow), proximal His(F8), distal His(E7) or Gln(E7) (blue), and the highly conserved Phe(CD1) (gray). The E and F helices are labeled as are the N and C termini. Note the variability in structure, particularly in the length of the F helices, amino termini, and region just prior to the start of the E helix. A, sperm whale myoglobin (Protein Data Bank (PDB) code 1mbo) (38). B, Scapharca dimeric hemoglobin (PDB code 1hbi) (39). Note the longer F helix characteristic of most invertebrate hemoglobins, including those in EF dimer assemblages. C, Paramecium hemoglobin (PDB code 1dlw) (40). This truncated hemoglobin is about 75% of the length of myoglobin but still provides a site for the reversible binding of oxygen to the heme iron. Dimers and tetramers are depicted as van der Waals spheres for main chain and heme atoms with heme groups shown in red, E and F helices in cyan, and the rest of the main chain in gray. The 24-subunit Riftia C1 hemoglobin is depicted with a main chain trace in color ranging from green to blue according to subunit type and hemes in red, whereas the 180-subunit Lumbricus erythrocruorin is shown in a surface rendition of its 5.5-Å electron density map with hemoglobin subunits in magenta and non-globin linker chains in blue and gold. Of the structures shown here, only Urechis hemoglobin does not exhibit cooperative oxygen binding. Note the similar assembly of subunits in hemoglobins with EF dimers, including echinoderm, mollusc, and subassemblies of annelid (see Fig. 3 These structural considerations suggest that the 400-kDa vestimentiferan hemoglobin assembly may represent a first step in the development of the megadalton erythrocruorins. Such a scenario is consistent with that proposed by Yuasa et al. (30) but has been questioned by Negrisolo et al. (31).

Cooperative Mechanisms within EF Dimers
In only one case, that of the homodimeric hemoglobin from S. inaequivalvis, has the structural basis for cooperativity in an invertebrate hemoglobin been investigated in detail. High resolution crystallographic analyses revealed that the ligand-linked transitions involve only small quaternary subunit movements but striking tertiary rearrangements (20). The observed transitions result in substantial functional changes with the high affinity R state estimated to bind oxygen ϳ300-fold more tightly than the low affinity T state (32). Thus, the EF arrangement of subunits including the heme in the interface permits strong modulation of ligand affinity with limited structural changes. Mutagenesis has confirmed the functional importance of three key aspects of the observed transitions, including residue F4, interface water molecules, and heme group movement. A phenylalanine at position F4 is critical for the functional difference between the low affinity (T) state and high affinity (R) state (33). In the T state, the side chain of Phe(F4) packs in the proximal pocket such that it restricts acquisition of high affinity stereochemistry, but this proximal strain is relieved by its extrusion into the subunit interface in the R state. The ligand-linked movement of Phe(F4) into the interface disrupts a well ordered cluster of interface water molecules that is essential for stabilization of the T state. Stability of the water cluster results from hydrogen bonding by main chain atoms, heme propionates, and Thr(E10). Disruption of the water cluster appears to be at least part of the signal by which one subunit detects the ligand state of its partner subunit (32). Ligand binding also results in a movement of the heme group deeper into each subunit, which is coupled with transitions at the interface involving the heme propionates, water molecules, and Lys(F3). Mutagenesis shows that this heme movement is required for the other ligand-linked transitions suggesting that this movement is the trigger for the allosteric transition (34). Similar ligand-linked heme movements are observed in human hemoglobin ␤-subunits (15) and hexacoordinate neuroglobin (35) indicating that such heme movement may be a widespread response to ligand binding.

Phylogeny of Hemoglobin Assemblages
Why is the EF dimer hemoglobin assemblage so widespread among invertebrates? It has now been observed in three different phyla (and suggested in a fourth (36)), including not only annelids and molluscs, which are thought to be closely related, but also in the deuterostome phylum of echinoderms (Fig. 2). In fact, there is currently no example available of a cooperative invertebrate hemoglobin that does not exhibit the EF dimer assemblage. Despite conservation of quaternary subunit arrangement, residues in the dimeric interface are remarkably variable. This variation becomes even more striking when considering those residues contributing to allosteric behavior. As discussed above, key players in the cooperative ligand binding behavior of Scapharca dimeric hemoglobin include the heme group, ordered interface water molecules, and residues Phe(F4), Lys(F3), and Thr(E10). These residues are conserved among EF dimeric hemoglobins within each phylum but not between different phyla, leaving only the F8-coordinated heme itself as the singular ubiquitous feature of EF dimer and all other hemoglobins (Fig. 4). Position E10 is occupied by an Arg in both annelid and echinoderm EF dimers and

FIG. 4. Comparison of key residues in the EF dimers of Scapharca HbI, Lumbricus erythrocruorin, and Caudina
HbD. The main chain traces for the E and F helices are shown in gray, heme groups are shown in red, and side chains are shown for residues E10, F3, and F4 (yellow) and for the proximal and distal histidines (F8 and E7). The variation in residues at positions demonstrated to contribute to cooperative oxygen binding in Scapharca HbI (E10, F3, and F4) suggests that that diverse mechanisms for cooperativity are operative in these EF dimeric hemoglobins .   FIG. 3. Hierarchical subunit arrangement in extracellular annelid hemoglobins. A, L. terrestris erythrocruorin is assembled from 12 hemoglobin dodecamers and 36 non-globin linker chains, whereas Riftia C1 hemoglobin is assembled from two hemoglobin dodecamers. B, the dodecamers from both hemoglobins have a very similar assembly, with the molecular model of the isolated Lumbricus dodecamer (PDB code 1x9f) (27) shown with main chain traces and red hemes for the 12 hemoglobin subunits. C, each dodecamer is assembled from three identical heterotetramers, with tetramers depicted as described in Fig. 2 for other tetrameric hemoglobins. Note the overall similarity of the tetramer assembly with that of Scapharca HbII. D, each half of the tetramers are assembled as EF heterodimers. In this way, all hemoglobin subunits in these extracellular annelid hemoglobins participate in EF dimers. is likely to be important in the cooperative mechanisms (27). However, the different conformations of this residue in the liganded forms of annelid and echinoderm hemoglobins (Fig. 4) suggest that its contribution to cooperativity might be rather different in the two systems. Thus, despite a similar quaternary arrangement, the allosteric mechanisms used to modulate oxygen affinity are likely to be quite variable in these hemoglobins comprised of EF dimers.
The observed variability of interface residues among EF dimeric hemoglobins raises the issue of the origin of the EF dimer assemblies. Are these similar assemblies the result of divergent evolution, which all derive from some ancestral dimeric hemoglobin, or might the prevalence of this subunit pairing be an indication that dimerization at the EF face provides an efficient means for gaining selective advantage that has originated on multiple occasions?
To investigate this question, we have carried out phylogenetic analyses of a number of invertebrate and vertebrate hemoglobin sequences. Hemoglobin phylogeny has been a popular endeavor to explore the relationships in this omnipresent molecule; however, to our knowledge it has not been carried out previously to specifically address the relationship among the hemoglobins that exhibit EF dimer assemblies. We have used the program PAUP* Version 4.0 Beta (37) to reconstruct phylogenetic tree(s) with maximum parsimony. Twenty-one amino acid sequences from hemoglobins of known structure, including vertebrate tetramers (human, chicken, shark; both ␣ and ␤ chains), EF dimer containing invertebrate hemoglobins (C. arenicola, S. inaequivalvis, L. terrestris, R. pachyptila) and other globins (U. caupo, Aplysia limacina, Ascaris suum, sperm whale myoglobin) were aligned using known helical assignments with minimal gaps introduced in the interhelical regions. Each gap was treated as a missing residue and attached with various degrees of penalty for PAUP analysis. The bootstrap method with the 50% majority rule was used to attach confidence to each node. The entire sample set (21 taxa) or 12 representative sequences were used for heuristic or exhaustive searches.
Extensive searches resulted in no trees that reconstruct the EF dimer sequences onto one common ancestor (data not shown). In addition to the PAUP calculations, direct examination of the sequences that assemble into EF dimers provides no evidence of any uniquely unifying residues among these hemoglobins. Thus, many variations in sequence are used in forming EF dimers. Although one cannot eliminate the possibility of a common ancestor that has diversified greatly, these findings suggest that it is likely, or at least plausible, that the EF dimers found in annelids, molluscs, and echinoderms represent convergent acquisitions of an EF dimer assemblage with cooperative oxygen binding characteristics.

Conclusions
Hemoglobin provides a fascinating example of molecular evolution. Acquisition of regulatory control over ligand binding, using subunit interactions to create allosteric protein molecules, has been important for the efficient transport of oxygen. The dramatically different assemblies of vertebrate and invertebrate allosteric hemoglobins strongly argue for an evolutionarily independent acquisition of cooperativity. The similarity of dimeric pairing in all invertebrate cooperative hemoglobins studied to date suggests the possibility that acquisition of the EF dimer interface in these hemoglobin assemblages could be evolutionarily related. However, the lack of sequence similarity among residues involved in the interface and the calculations described here raise the possibility that development of the EF dimer pairing may have occurred independently among molluscs, annelids, and echinoderms. A remarkable feature of the EF dimers is their ability to exhibit cooperative ligand binding properties in isolation, as in molluscan and echinoderm hemoglobins, or to form building blocks for the assembly of larger and more highly regulated allosteric complexes, as in molluscan and annelid hemoglobins. Thus, the efficient and regulated oxygen transport associated with the EF dimer interface may have been the driving force behind the recurring evolution of this versatile allosteric unit.