The recognition domain of the BpuJI restriction endonuclease in complex with cognate DNA at 1.3-A resolution

Type IIS restriction endonucleases recognize asymmetric DNA sequences and cleave both DNA strands at fixed positions downstream of the recognition site. The restriction endonuclease BpuJI recognizes the asymmetric sequence 5′-CCCGT; however, it cuts at multiple sites in the vicinity of the target sequence. BpuJI consists of two physically separate domains, with catalytic and dimerization functions in the C-terminal domain and DNA recognition functions in the N-terminal domain. Here we report the crystal structure of the BpuJI recognition domain bound to cognate DNA at 1.3-Å resolution. This region folds into two winged-helix subdomains, D1 and D2, interspaced by the DL subdomain. The D1 and D2 subdomains of BpuJI share structural similarity with the similar subdomains of the FokI DNA-binding domain; however, their orientations in protein-DNA complexes are different. Recognition of the 5′-CCCGT target sequence is achieved by BpuJI through the major groove contacts of amino acid residues located on both the helix-turn-helix motifs and the N-terminal arm. The role of these interactions in DNA recognition is also corroborated by mutational analysis.     

Crystal structures of type II restriction endonuclease EcoO109I and its complex with cognate DNA

EcoO109I is a type II restriction endonuclease that recognizes the DNA sequence of RGGNCCY. Here we describe the crystal structures of EcoO109I and its complex with DNA. A comparison of the two structures shows that the catalytic domain moves drastically to capture the DNA. One metal ion and two water molecules are observed near the active site of the DNA complex. The metal ion is a Lewis acid that stabilizes the pentavalent phosphorus atom in the transition state. One water molecule, activated by Lys-126, attacks the phosphorus atom in an S(N)2 mechanism, whereas the other water interacts with the 3'-leaving oxygen to donate a proton to the oxygen. EcoO109I is similar to EcoRI family enzymes in terms of its DNA cleavage pattern and folding topology of the common motif in the catalytic domain, but it differs in the manner of DNA recognition. Our findings propose a novel classification of the type II restriction endonucleases and lead to the suggestion that EcoO109I represents a new subclass of the EcoRI family.     

Crystal structure of spinach plastocyanin at 1.7 A resolution.

The crystal structure of plastocyanin from spinach has been determined using molecular replacement, with the structure of plastocyanin from poplar as a search model. Successful crystallization was facilitated by site-directed mutagenesis in which residue Gly8 was substituted with Asp. The region around residue 8 was believed to be too mobile for the wild-type protein to form crystals despite extensive screening. The current structure represents the oxidized plastocyanin, copper (II), at low pH (approximately 4.4). In contrast to the similarity in the core region as compared to its poplar counterpart, the structure shows some significant differences in loop regions. The most notable is the large shift of the 59-61 loop where the largest shift is 3.0 A for the C(alpha) atom of Glu59. This results in different patterns of electrostatic potential around the acidic patches for the two proteins.     

Crystal structure of human alpha-lactalbumin at 1.7 A resolution

The three-dimensional X-ray structure of human alpha-lactalbumin, an important component of milk, has been determined at 1.7 A (0.17 nm) resolution by the method of molecular replacement, using the refined structure of baboon alpha-lactalbumin as the model structure. The two proteins are known to have more than 90% amino acid sequence identity and crystallize in the same orthorhombic space group, P2(1)2(1)2. The crystallographic refinement of the structure using the simulated annealing method, resulted in a crystallographic R-factor of 0.209 for the 11,373 observed reflections (F greater than or equal to 2 sigma (F)) between 8 and 1.7 A resolution. The model comprises 983 protein atoms, 90 solvent atoms and a bound calcium ion. In the final model, the root-mean-square deviations from ideality are 0.013 A for covalent bond distances and 2.9 degrees for bond angles. Superposition of the human and baboon alpha-lactalbumin structures yields a root-mean-square difference of 0.67 A for the 123 structurally equivalent C alpha atoms. The C terminus is flexible in the human alpha-lactalbumin molecule. The striking structural resemblance between alpha-lactalbumins and C-type lysozymes emphasizes the homologous evolutionary relationship between these two classes of proteins.     

Calmodulin structure refined at 1.7 A resolution.

We have determined and refined the crystal structure of a recombinant calmodulin at 1.7 A resolution. The structure was determined by molecular replacement, using the 2.2 A published native bovine brain structure as the starting model. The final crystallographic R-factor, using 14,469 reflections in the 10.0 to 1.7 A range with structure factors exceeding 0.5 sigma, is 0.216. Bond lengths and bond angle distances have root-mean-square deviations from ideal values of 0.009 A and 0.032 A, respectively. The final model consists of 1279 non-hydrogen atoms, including four calcium ions, 1130 protein atoms, including three Asp118 side-chain atoms in double conformation, 139 water molecules and one ethanol molecule. The electron densities for residues 1 to 4 and 148 of calmodulin are poorly defined, and not included in our model, except for main-chain atoms of residue 4. The calmodulin structure from our crystals is very similar to the earlier 2.2 A structure described by Babu and coworkers with a root-mean-square deviation of 0.36 A. Calmodulin remains a dumb-bell-shaped molecule, with similar lobes and connected by a central alpha-helix. Each lobe contains three alpha-helices and two Ca2+ binding EF hand loops, with a short antiparallel beta-sheet between adjacent EF hand loops and one non-EF hand loop. There are some differences in the structure of the central helix. The crystal packing is extensively studied, and facile crystal growth along the z-axis of the triclinic crystals is explained. Herein, we describe hydrogen bonding in the various secondary structure elements and hydration of calmodulin.     

Crystal structure of chondroitinase B from Flavobacterium heparinum and its complex with a disaccharide product at 1.7 A resolution

Glycosaminoglycans (GAGs) are a family of acidic heteropolysaccharides, including such molecules as chondroitin sulfate, dermatan sulfate, heparin and keratan sulfate. Cleavage of the O-glycosidic bond within GAGs can be accomplished by hydrolases as well as lyases, yielding disaccharide and oligosaccharide products. We have determined the crystal structure of chondroitinase B, a glycosaminoglycan lyase from Flavobacterium heparinum, as well as its complex with a dermatan sulfate disaccharide product, both at 1.7 A resolution. Chondroitinase B adopts the right-handed parallel beta-helix fold, found originally in pectate lyase and subsequently in several polysaccharide lyases and hydrolases. Sequence homology between chondroitinase B and a mannuronate lyase from Pseudomonas sp. suggests this protein also adopts the beta-helix fold. Binding of the disaccharide product occurs within a positively charged cleft formed by loops extending from the surface of the beta-helix. Amino acid residues responsible for recognition of the disaccharide, as well as potential catalytic residues, have been identified. Two arginine residues, Arg318 and Arg364, are found to interact with the sulfate group attached to O-4 of N-acetylgalactosamine. Cleavage of dermatan sulfate likely occurs at the reducing end of the disaccharide, with Glu333 possibly acting as the general base.     

Crystal structure of BstYI at 1.85A resolution: a thermophilic restriction endonuclease with overlapping specificities to BamHI and BglII

We report here the structure of BstYI, an "intermediate" type II restriction endonuclease with overlapping sequence specificities to BamHI and BglII. BstYI, a thermophilic endonuclease, recognizes and cleaves the degenerate hexanucleotide sequence 5'-RGATCY-3' (where R=A or G and Y=C or T), cleaving DNA after the 5'-R on each strand to produce four-base (5') staggered ends. The crystal structure of free BstYI was solved at 1.85A resolution by multi-wavelength anomalous dispersion (MAD) phasing. Comparison with BamHI and BglII reveals a strong structural consensus between all three enzymes mapping to the alpha/beta core domain and residues involved in catalysis. Unexpectedly, BstYI also contains an additional "arm" substructure outside of the core protein, which enables the enzyme to adopt a more compact, intertwined dimer structure compared with BamHI and BglII. This arm substructure may underlie the thermostability of BstYI. We identify putative DNA recognition residues and speculate as to how this enzyme achieves a "relaxed" DNA specificity.     

Crystal structure of Escherichia coli CheY refined at 1.7-A resolution

The three-dimensional structure of wild-type CheY from Escherichia coli has been refined by stereochemically restrained least squares minimization to a crystallographic R-factor of 15.1% at 1.7-A resolution. The structure contains 1165 atoms, including all atoms of the protein, 147 water molecules, and three sulfate ions. The final model has root mean square deviations of 0.018 and 0.049 A from idealized bond lengths and angle distances, respectively. Seven amino acid side chains have been modeled in dual conformations. CheY folds as a compact (beta/alpha)5 globular protein, with the phosphorylation region contained in a cavity on one face of the molecule. This active site area is bordered by the carboxyl termini of the three central beta-strands, by alpha 1, and by the loop connecting beta 5 to alpha 5. The Lys-109 side chain of this loop extends into the active site by virtue of its cis peptide bond conformation preceding Pro-110. The epsilon-amino group of Lys-109 is in close bonding contact with the carboxyl group of Asp-57, the residue that is phosphorylated in the activation process of CheY. The details of the hydrogen bonding network in the phosphorylation region indicate that structural rearrangements must accompany the phosphorylation of Asp-57.     

Specific DNA recognition by the type II restriction endonuclease MunI: the effect of pH.

To investigate the effect of pH on sequence-specific binding, a thermodynamic characterization of the interaction of the protein MunI with a specific, and a nonspecific, oligonucleotide was performed. MunI is a type II restriction endonuclease which is able to bind specifically, but loses its enzymatic activity in the absence of magnesium ions. Comparison of the specific and nonspecific interactions at 10 and 25 degrees C shows that the latter is accompanied by a small change in enthalpy, and a negligible change in constant pressure heat capacity. On going through the pH range 5.75-9.0 at 25 degrees C, the affinity of specific complex formation is reduced by 20-fold. The interaction is accompanied by the protonation of groups assumed to be on the protein. Based on the simplest model that will fit the data, two distinct protonation events are observed. At low pH, two groups per protein molecule undergo protonation with a pK(a) of 6.0 and 6.9 in the free and bound forms, respectively. At high pH, a further independent protonation occurs involving two groups with pK(a) values of 8.9 and approximately 10.7 in the free and bound forms, respectively. The change in heat capacity ranges from -2.7 to -1.7 kJ mol(-1) K(-1) in going from pH 6.5 to 8.5. This range of variation of change in heat capacity can be accounted for by the effects of protonation of the interacting molecules. The change in heat capacity, calculated from surface area burial using a previously established relationship (1.15 kJ mol(-1) K(-1)), does not correlate well with the experimentally determined values.     

A homology model of restriction endonuclease SfiI in complex with DNA

Background

Restriction enzymes (REases) are commercial reagents commonly used in recombinant DNA technologies. They are attractive models for studying protein-DNA interactions and valuable targets for protein engineering. They are, however, extremely divergent: the amino acid sequence of a typical REase usually shows no detectable similarities to any other proteins, with rare exceptions of other REases that recognize identical or very similar sequences. From structural analyses and bioinformatics studies it has been learned that some REases belong to at least four unrelated and structurally distinct superfamilies of nucleases, PD-DxK, PLD, HNH, and GIY-YIG. Hence, they are extremely hard targets for structure prediction and homology-based inference of sequence-function relationships and the great majority of REases remain structurally and evolutionarily unclassified.

Results

SfiI is a REase which recognizes the interrupted palindromic sequence 5'GGCCNNNN^NGGCC3' and generates 3 nt long 3' overhangs upon cleavage. SfiI is an archetypal Type IIF enzyme, which functions as a tetramer and cleaves two copies of the recognition site in a concerted manner. Its sequence shows no similarity to other proteins and nothing is known about the localization of its active site or residues important for oligomerization. Using the threading approach for protein fold-recognition, we identified a remote relationship between SfiI and BglI, a dimeric Type IIP restriction enzyme from the PD-DxK superfamily of nucleases, which recognizes the 5'GCCNNNN^NGGC3' sequence and whose structure in complex with the substrate DNA is available. We constructed a homology model of SfiI in complex with its target sequence and used it to predict residues important for dimerization, tetramerization, DNA binding and catalysis.

Conclusions

The bioinformatics analysis suggest that SfiI, a Type IIF enzyme, is more closely related to BglI, an "orthodox" Type IIP restriction enzyme, than to any other REase, including other Type IIF REases with known structures, such as NgoMIV. NgoMIV and BglI belong to two different, very remotely related branches of the PD-DxK superfamily: the α-class (EcoRI-like), and the β-class (EcoRV-like), respectively. Thus, our analysis provides evidence that the ability to tetramerize and cut the two DNA sequences in a concerted manner was developed independently at least two times in the evolution of the PD-DxK superfamily of REases. The model of SfiI will also serve as a convenient platform for further experimental analyses.     

Structure of PvuII endonuclease with cognate DNA.

We have determined the structure of PvuII endonuclease complexed with cognate DNA by X-ray crystallography. The DNA substrate is bound with a single homodimeric protein, each subunit of which reveals three structural regions. The catalytic region strongly resembles structures of other restriction endonucleases, even though these regions have dissimilar primary sequences. Comparison of the active site with those of EcoRV and EcoRI endonucleases reveals a conserved triplet sequence close to the reactive phosphodiester group and a conserved acidic pair that may represent the ligands for the catalytic cofactor Mg2+. The DNA duplex is not significantly bent and maintains a B-DNA-like conformation. The subunit interface region of the homodimeric protein consists of a pseudo-three-helix bundle. Direct contacts between the protein and the base pairs of the PvuII recognition site occur exclusively in the major groove through two antiparallel beta strands from the sequence recognition region of the protein. Water-mediated contacts are made in the minor grooves to central bases of the site. If restriction enzymes do share a common ancestor, as has been proposed, their catalytic regions have been very strongly conserved, while their subunit interfaces and DNA sequence recognition regions have undergone remarkable structural variation.     

Crystal structure of restriction endonuclease BglI bound to its interrupted DNA recognition sequence.

The crystal structure of the type II restriction endonuclease BglI bound to DNA containing its specific recognition sequence has been determined at 2.2 A resolution. This is the first structure of a restriction endonuclease that recognizes and cleaves an interrupted DNA sequence, producing 3' overhanging ends. BglI is a homodimer that binds its specific DNA sequence with the minor groove facing the protein. Parts of the enzyme reach into both the major and minor grooves to contact the edges of the bases within the recognition half-sites. The arrangement of active site residues is strikingly similar to other restriction endonucleases, but the co-ordination of two calcium ions at the active site gives new insight into the catalytic mechanism. Surprisingly, the core of a BglI subunit displays a striking similarity to subunits of EcoRV and PvuII, but the dimer structure is dramatically different. The BglI-DNA complex demonstrates, for the first time, that a conserved subunit fold can dimerize in more than one way, resulting in different DNA cleavage patterns.     

Crystal structure of glucose dehydrogenase from Bacillus megaterium IWG3 at 1.7 A resolution

The crystal structure of glucose dehydrogenase (GlcDH) from Bacillus megaterium IWG3 has been determined to an R-factor of 17.9% at 1.7 A resolution. The enzyme consists of four identical subunits, which are similar to those of other short-chain reductases/dehydrogenases (SDRs) in their overall folding and subunit architecture, although cofactor binding sites and subunit interactions differ. Whereas a pair of basic residues is well conserved among NADP(+)-preferring SDRs, only Arg39 was found around the adenine ribose moiety of GlcDH. This suggests that one basic amino acid is enough to determine the coenzyme specificity. The four subunits are interrelated by three mutually perpendicular diad axes (P, Q, and R). While subunit interactions through the P-axis for GlcDH are not so different from those of the other SDRs, those through the Q-axis differ significantly. GlcDH was found to have weaker hydrophobic interactions in the Q-interface. Moreover, GlcDH lacks the salt bridge that stabilizes the subunit interaction in the Q-interface in the other SDRs. Hydrogen bonds between Q-axis related subunits are also less common than in the other SDRs. The GlcDH tetramer dissociates into inactive monomers at pH 9.0, which can be attributed mainly to the weakness of the Q-axis interface.     

Crystal structure of decameric 2-Cys peroxiredoxin from human erythrocytes at 1.7 A resolution

BACKGROUND: The peroxiredoxins (Prxs) are an emerging family of multifunctional enzymes that exhibit peroxidase activity in vitro, and in vivo participate in a range of cellular processes known to be sensitive to reactive oxygen species. Thioredoxin peroxidase B (TPx-B), a 2-Cys type II Prx from erythrocytes, promotes potassium efflux and down-regulates apoptosis and the recruitment of monocytes by endothelial tissue. RESULTS: The crystal structure of human decameric TPx-B purified from erythrocytes has been determined to 1.7 [corrected)] A resolution. The structure is a toroid comprising five dimers linked end-on through predominantly hydrophobic interactions, and is proposed to represent an intermediate in the in vivo reaction cycle. In the crystal structure, Cys51, the site of peroxide reduction, is oxidised to cysteine sulphinic acid. The residue Cys172, lies approximately 10 A away from Cys51 [corrected]. CONCLUSIONS: The oxidation of Cys51 appears to have trapped the structure into a stable decamer, as confirmed by sedimentation analysis. A comparison with two previously reported dimeric Prx structures reveals that the catalytic cycle of 2-Cys Prx requires significant conformational changes that include the unwinding of the active-site helix and the movement of four loops. It is proposed that the stable decamer forms in vivo under conditions of oxidative stress. Similar decameric structures of TPx-B have been observed by electron microscopy, which show the protein associated with the erythrocyte membrane.     

Crystal structure of yeast cytochrome c peroxidase refined at 1.7-A resolution

The crystal structure of cytochrome c peroxidase (EC 1.11.1.5) has been refined to an R factor of 0.20 computed for all reflections to 1.7 A. The refined molecular model includes 263 bound water molecules and allows for x-ray scattering by amorphous solvent. The mean positional error in atomic coordinates is estimated to lie between 0.12 and 0.21 A. Two factors are identified which may account for the ability of the enzyme to stabilize high-oxidation states of the heme iron during catalysis: 1) the proximal histidine forms a hydrogen bond with a buried aspartic acid side chain, Asp-235; and 2) the heme environment is more polar than in the cytochromes c or globins, owing to the presence of the partially buried side-chain of Arg-48 and five water molecules bound in close proximity to the heme. Two of these occupy the presumed peroxide-binding site. Two candidates are likely for the side chain that is oxidized to a free radical during formation of Compound I: 1) Trp-51, which rests 3.3 A above the heme plane in close proximity (2.7 A) to the sixth coordination position; and 2) Met-172, which is 3.7 A from the heme. Nucleophilic stabilization of the methionyl cation radical may be possible via Asp-235. His-181 is found to lie coplanar with the heme in a niche between the two propionates near the suspected cytochrome c-binding site. A network of hydrogen bonds involving this histidine may provide a preferred pathway for electron transfer between hemes.     

Crystal structure of the RIM1alpha C2B domain at 1.7 A resolution

RIM proteins play critical roles in synaptic vesicle priming and diverse forms of presynaptic plasticity. The C-terminal C2B domain is the only module that is common to all RIMs but is only distantly related to well-studied C2 domains, and its three-dimensional structure and interactions have not been characterized in detail. Using NMR spectroscopy, we now show that N- and C-terminal extensions beyond the predicted C2B domain core sequence are necessary to form a folded, stable RIM1alpha C2B domain. We also find that the isolated RIM1alpha C2B domain is not sufficient for previously described protein-protein interactions involving the RIM1alpha C-terminus, suggesting that additional sequences adjacent to the C2B domain might be required for these interactions. However, analytical ultracentrifugation shows that the RIM1alpha C2B domain forms weak dimers in solution. The crystal structure of the RIM1alpha C2B domain dimer at 1.7 A resolution reveals that it forms a beta-sandwich characteristic of C2 domains and that the unique N- and C-terminal extensions form a small subdomain that packs against the beta-sandwich and mediates dimerization. Our results provide a structural basis to understand the function of RIM C2B domains and suggest that dimerization may be a crucial aspect of RIM function.     

Refined structure of baboon alpha-lactalbumin at 1.7 A resolution. Comparison with C-type lysozyme

The solution of the structure of alpha-lactalbumin from baboon milk (Papio cynocephalus) at 4.5 A resolution using the isomorphous replacement method has been reported previously. Initial refinement on the basis of these low-resolution studies was not successful because of the poor isomorphism of the best heavy-atom derivative. Because of the striking similarity between the structure of lysozyme and alpha-lactalbumin, a more cautious molecular replacement approach was tried to refine the model. Using hen egg-white lysozyme as the starting model, preliminary refinement was performed using heavily constrained least-squares minimization in reciprocal space. The model was further refined using stereochemical restraints at 1.7 A resolution to a conventional crystallographic residual of 0.22 for 1141 protein atoms. In the final model, the root-mean-square deviation from ideality for bond distances is 0.015 A, and for angle distances it is 0.027 A. The refinement was carried out using the human alpha-lactalbumin sequence and "omit maps" calculated during the course of refinement indicated eight possible sequence changes in the baboon alpha-lactalbumin X-ray sequence. During the refinement, a tightly bound calcium ion and 150 water molecules, of which four are internal, have been located. Some of the water molecules were modelled for disordered side-chains. The co-ordination around the calcium is a slightly distorted pentagonal bipyramid. The Ca-O distances vary from 2.2 A to 2.6 A, representing a tight calcium-binding loop in the structure. The calcium-binding fold only superficially resembles the "EF-hand" and presumably has no evolutionary relationship with other EF-hand structures. The overall structure of alpha-lactalbumin is very similar to that of lysozyme. All large deviations occur in the loops where all sequence deletions and insertions are found. The C terminus appears to be rather flexible in alpha-lactalbumin compared to lysozyme. The experimental evidence supports the earlier predictions for the alpha-lactalbumin structure that were based upon the assumption that alpha-lactalbumin and lysozyme have similar three-dimensional structures, with minimal deletions and insertions. A detailed comparison of the two structures shows striking features as well as throwing some light on the evolution of these two proteins from a common precursor.     

Crystal structure of Escherichia coli ketopantoate reductase at 1.7 A resolution and insight into the enzyme mechanism.

Ketopantoate reductase (KPR, EC 1.1.1.169) catalyzes the NADPH-dependent reduction of ketopantoate to pantoate on the pantothenate (vitamin B(5)) biosynthetic pathway. The Escherichia coli panE gene encoding KPR was cloned and expressed at high levels as the native and selenomethionine-substituted (SeMet) proteins. Both native and SeMet recombinant proteins were purified by three chromatographic steps, to yield pure proteins. The wild-type enzyme was found to have a K(M)(NADPH) of 20 microM, a K(M)(ketopantoate) of 60 microM, and a k(cat) of 40 s(-1). Regular prismatic KPR crystals were prepared using the hanging drop technique. They belonged to the tetragonal space group P4(2)2(1)2, with cell parameters: a = b = 103.7 A and c = 55.7 A, accommodating one enzyme molecule per asymmetric unit. The structure of KPR was determined by the multiwavelength anomalous dispersion method using the SeMet protein, for which data were collected to 2.3 A resolution. The native data were collected to 1.7 A resolution and used to refine the final structure. The secondary structure comprises 12 alpha-helices, three 3(10)-helices, and 11 beta-strands. The enzyme is monomeric and has two domains separated by a cleft. The N-terminal domain has an alphabeta-fold of the Rossmann type. The C-terminal domain (residues 170-291) is composed of eight alpha-helices. KPR is shown to be a member of the 6-phosphogluconate dehydrogenase C-terminal domain-like superfamily. A model for the ternary enzyme-NADPH-ketopantoate ternary complex provides a rationale for kinetic data reported for specific site-directed mutants.     

A third quaternary structure of human hemoglobin A at 1.7-A resolution.

Previous crystallographic studies have shown that human hemoglobin A can adopt two stable quaternary structures, one for deoxyhemoglobin (the T-state) and one for liganded hemoglobin (the R-state). In this paper we report our finding of a second quaternary structure (the R2-state) for liganded hemoglobin A. The magnitudes of the spatial differences between the R- and R2-states are as large as those between the R- and T-states. Of particular interest are the structural changes that occur as a result of R-T and R-R2 transitions at the so-called "switch" region of the critical alpha 1 beta 2 interface. In the R-state, His-97 beta 2 is positioned between Thr-38 alpha 1 and Thr-41 alpha 1, whereas in transition to the T-state His 97 beta 2 must "jump" a turn in the alpha 1 C helix to form nonpolar contacts with Thr-41 alpha 1 and Pro-44 alpha 1. This facet of the R-T transition presents a major steric barrier to the quaternary structure change. In the R2-state, His-97 beta 2 simply rotates away from threonines 38 alpha 1 and 41 alpha 1, breaking contact with these residues and allowing water access to the center of the alpha 1 beta 2 interface. With the switch region in an open position in the R2-state, His-97 beta 2 should be able to move by Thr-41 alpha 1 and make the transition to the T-state with a steric barrier that is less than that for the R-T transition. Thus the R2-state may function as a stable intermediate along a R-R2-T pathway. The T-, R-, and R2-states must coexist in solution. That is, the fact that these states can be crystallized implies that they are all energetically accessible structures. What remains to be determined are the T-to-R, T-to-R2, and R-to-R2 equilibrium constants for hemoglobin under various solution conditions and ligation states. Although this may prove to be difficult, we discuss previously published results which indicate that low concentrations of inorganic anions or low pH may favor the R2-state and at least one alpha 1 beta 2 interface mutation stabilizes a quaternary structure that is very similar to the R2-state.