Propagation of allosteric changes through the catalytic-regulatory interface of Escherichia coli aspartate transcarbamylase

Each of two previously isolated strains of Escherichia coli containing a single nonsense codon within the pyrB gene was suppressed with four different nonsense suppressors. The kinetic analysis using crude extracts of these nonsense-suppressed strains indicated that the mutant aspartate transcarbamylases had altered cooperativity and affinity for aspartate as judged by the substrate concentration at half of the maximal velocity. Both pyrB genes were cloned and then sequenced. In both cases, a single base change was identified which converted a glutamine GAC codon into a TAC nonsense codon. Both mutations occurred in the catalytic chain of aspartate transcarbamylase and were identified at positions 108 and 246. The glutamine at position 108 in the wild-type structure is located at the interface between the catalytic and regulatory chains and is involved in a number of interactions with backbone and side chains of the regulatory chain. The glutamine at position 246 in the wild-type structure is located in the 240s loop of the enzyme. Two additional mutant versions of aspartate transcarbamylase were created by site-directed mutagenesis to further investigate the 108-position in the structure, a glutamine to tyrosine substitution at position 108 of the catalytic chain, and an asparagine to glycine change at position 113 of the regulatory chain, a residue which interacts directly with glutamine-108 in the wild-type structure. Both mutant enzymes have reduced affinity for aspartate. However, the Tyr-108 mutant enzyme exhibits a reduced Hill coefficient while the Gly-113 enzyme exhibits an increased Hill coefficient. The response to the allosteric effectors ATP and CTP is also changed for both the mutant enzymes.(ABSTRACT TRUNCATED AT 250 WORDS)     

Influence of nucleotide effectors on the kinetics of the quaternary structure transition of allosteric aspartate transcarbamylase

We report the effects of allosteric effectors, ATP, CTP and UTP on the kinetics of the quaternary structure change of Escherichia coli ATCase during the enzyme reaction with physiological substrates. Time-resolved, small-angle, X-ray scattering of solutions allows direct observation of structural transitions over the entire time-course of the enzyme reaction initiated by fast mixing of the enzyme and substrates. In the absence of effectors, all scattering patterns recorded during the reaction are consistent with a two-state, concerted transition model, involving no detectable intermediate conformation that differs from the less active, unliganded T-state and the more active, substrate-bound R-state. The latter predominates during the steady-state phase of enzyme catalysis, while the initial T-state is recovered after substrate consumption. The concerted character of the structural transition is preserved in the presence of all effectors. CTP slightly shifts the dynamical equilibrium during a shortened steady state toward T while the additional presence of UTP makes the steady state vanishingly short. The return transition to the T conformation is slowed significantly in the presence of inhibitors, the effect being most severe in the presence of UTP. While ATP increases the apparent T to R rate, it also increases the duration of the steady-state phase, an apparently paradoxical observation. This observation can be accounted for by the greater increase in the association rate constant of aspartate, promoted by ATP, while the nucleotide produces a lesser degree of increase in the dissociation rate constant. Under our experimental conditions, using high concentrations of both enzyme and substrate, it appears that this very mechanism of activation turns the activator into an efficient inhibitor. The scattering patterns recorded in the presence of ATP support the view that ATP alters the quaternary structure of the substrate-bound enzyme, an effect reminiscent of the reported modification of PALA-bound R-state by Mg-ATP.     

Stabilization of the R allosteric structure of Escherichia coli aspartate transcarbamoylase by disulfide bond formation

Here we report the first use of disulfide bond formation to stabilize the R allosteric structure of Escherichia coli aspartate transcarbamoylase. In the R allosteric state, residues in the 240s loop from two catalytic chains of different subunits are close together, whereas in the T allosteric state they are far apart. By substitution of Ala-241 in the 240s loop of the catalytic chain with cysteine, a disulfide bond was formed between two catalytic chains of different subunits. The cross-linked enzyme did not exhibit cooperativity for aspartate. The maximal velocity was increased, and the concentration of aspartate required to obtain one-half the maximal velocity, [Asp](0.5), was reduced substantially. Furthermore, the allosteric effectors ATP and CTP did not alter the activity of the cross-linked enzyme. When the disulfide bonds were reduced by the addition of 1,4-dithio-dl-threitol the resulting enzyme had kinetic parameters very similar to those observed for the wild-type enzyme and regained the ability to be activated by ATP and inhibited by CTP. Small-angle x-ray scattering was used to verify that the cross-linked enzyme was structurally locked in the R state and that this enzyme after reduction with 1,4-dithio-dl-threitol could undergo an allosteric transition similar to that of the wild-type enzyme. The complete abolition of homotropic and heterotropic regulation from stabilizing the 240s loop in its closed position in the R state, which forms the catalytically competent active site, demonstrates the significance that the quaternary structural change and closure of the 240s loop has in the functional mechanism of aspartate transcarbamoylase.     

Spectroscopic markers of the T<-->R quaternary transition in human hemoglobin

In this work, we use a sol-gel protocol to trap and compare the R and T quaternary states of both the deoxygenated (deoxyHb) and carbonmonoxide (HbCO) derivatives of human hemoglobin. The near infrared optical absorption band III and the infrared CO stretching band are used to detect the effect of quaternary structure on the spectral properties of deoxyHb and HbCO; comparison with myoglobin allows for an assessment of tertiary and quaternary contributions to the measured band shifts. The R<-->T transition is shown to cause a blue shift of the band III by approximately 35 cm(-1) for deoxyHb and a red shift of the CO stretching band by only approximately 0.3 cm(-1) for HbCO. This clearly shows that quaternary structure changes are transmitted to the heme pocket and that effects on deoxyHb are much larger than on HbCO, at least as far as the band energies are concerned. Experiments performed in the ample temperature interval of 300-10K show that the above quaternary structure effects are "static" and do not influence the dynamic properties of the heme pocket, at least as probed by the temperature dependence of band III and of the CO stretching band. The availability of quaternary structure sensitive spectroscopic markers and the quantitative measurement of the quaternary structure contribution to band shifts will be of considerable help in the analysis of flash-photolysis experiments on hemoglobin. Moreover, it will enable one to characterize the dynamic properties of functionally relevant hemoglobin intermediates and to study the kinetics of both the T-->R and R-->T quaternary transitions through time-resolved spectroscopy.     

Large differences are observed between the crystal and solution quaternary structures of allosteric aspartate transcarbamylase in the R state

Solution scattering curves evaluated from the crystal structures of the T and R states of the allosteric enzyme aspartate transcarbamylase from Escherichia coli were compared with the experimental x-ray scattering patterns. Whereas the scattering from the crystal structure of the T state agrees with the experiment, large deviations reflecting a significant difference between the quaternary structures in the crystal and in solution are observed for the R state. The experimental curve of the R state was fitted by rigid body movements of the subunits in the crystal R structure which displace the latter further away from the T structure along the reaction coordinates of the T→R transition observed in the crystals. Taking the crystal R structure as a reference, it was found that in solution the distance between the catalytic trimers along the threefold axis is 0.34 nm larger and the trimers are rotated by 11° in opposite directions around the same axis; each of the three regulatory dimers is rotated by 9° around the corresponding twofold axis and displaced by 0.14 nm away from the molecular center along this axis. Proteins 27:110–117 © 1997 Wiley-Liss, Inc.     

Electrostatic interactions in the assembly of Escherichia coli aspartate transcarbamylase

Although ionizable groups are known to play important roles in the assembly, catalytic, and regulatory mechanisms of Escherichia coli aspartate transcarbamylase, these groups have not been characterized in detail. We report the application of static accessibility modified Tanford-Kirkwood theory to model electrostatic effects associated with the assembly of pairs of chains, subunits, and the holoenzyme. All of the interchain interfaces except R1-R6 are stabilized by electrostatic interactions by -2 to -4 kcal-m-1 at pH 8. The pH dependence of the electrostatic component of the free energy of stabilization of intrasubunit contacts (C1-C2 and R1-R6) is qualitatively different from that of intersubunit contacts (C1-C4, C1-R1, and C1-R4). This difference may allow the transmission of information across subunit interfaces to be selectively regulated. Groups whose calculated pK or charge changes as a result of protein-protein interactions have been identified and the results correlated with available information about their function. Both the 240s loop of the c chain and the region near the Zn(II) ion of the r chain contain clusters of ionizable groups whose calculated pK values change by relatively large amounts upon assembly. These pK changes in turn extend to regions of the protein remote from the interface. The possibility that networks of ionizable groups are involved in transmitting information between binding sites is suggested.     

Involvement of tryptophan 209 in the allosteric interactions of Escherichia coli aspartate transcarbamylase using single amino acid substitution mutants

Five mutant versions of aspartate transcarbamylase have been isolated, all with single amino acid substitutions in the catalytic chain of the enzyme. A previously isolated pyrB nonsense mutant was suppressed with supB, supC, supD and supG to create enzymes with glutamine, tyrosine, serine or lysine, respectively, inserted at the position of the nonsense codon. Each of these enzymes was purified to homogeneity and kinetically characterized. The approximate location of the substitution was determined by using tryptic fingerprints of the wild-type enzyme and the enzyme obtained with a tyrosine residue inserted at the position of the nonsense codon. By first cloning the pyrBI operon, from the original pyrB nonsense strain, followed by sequencing of the appropriate portion of the gene, the exact location of the mutation was determined to be at position 209 of the catalytic chain. Site-directed mutagenesis was used to generate versions of aspartate transcarbamylase with tyrosine and glutamic acid at this position. The Tyr209 enzyme is identical with that obtained by suppression of the original nonsense mutation with supC. The two enzymes produced by site-directed mutagenesis were purified using a newly created overproducing strain. Kinetic analysis revealed that each mutant has an altered affinity for aspartate, as judged by variations in the substrate concentration at one-half maximal activity. In addition, the mutants exhibit altered Hill coefficients and maximal activities. In the wild-type enzyme, position 209 is a tryptophan residue that is involved in the stabilization of a bend in the molecule near the subunit interface region. The alteration in homotropic cooperativity seems to be due to changes induced in this bend in the molecule, which stabilizes alternate conformational states of the enzyme.     

Kinetics of structure and activity changes during the allosteric transition of aspartate transcarbamylase

For the first time, the structural change associated with an allosteric transition has been monitored by X-ray solution scattering. The kinetics of the quaternary structure change of aspartate transcarbamylase were first slowed by using acetyl phosphate instead of carbamyl phosphate, and by the presence of 10% or 30% ethylene glycol. At 6.5 degrees C, the quaternary structure change was found to have a time constant of about 11 seconds. This appears to be larger than that obtained for the switching of homotropic co-operativity, measured by chemical quench under the same conditions.     

A possible model for the concerted allosteric transition in Escherichia coli aspartate transcarbamylase as deduced from site-directed mutagenesis studies

Aspartate transcarbamylase is stabilized in a low-affinity-low-activity state exhibiting no cooperativity by selective perturbation of the Glu-50-Arg-167 and Glu-50-Arg-234 interdomain salt bridges. Similarly, a high-affinity-high-activity state of the enzyme, retaining a significant amount of cooperativity, is obtained by perturbation of the interaction between Tyr-240 and Asp-271. In this work, we show that the rupture of the link between Tyr-240 and Asp-271 in the enzyme already lacking the interdomain salt bridges regenerates the homotropic cooperative interactions between the catalytic sites and substantially increases the activity and affinity of the enzyme for aspartate. These results suggest a possible relationship between these two sets of interactions for the establishment of the cooperative behavior of the enzyme. Another mutation, Glu-239 to Gln, introduced to perturb the Glu-239-Lys-164 and Glu-239-Tyr-165 interactions between the two catalytic subunits, is sufficient to "lock" the enzyme in the R state. These observations emphasize the importance of the interactions at the interface between the catalytic trimers in maintaining the T state of the enzyme and shed light on the role played by this pathway in the communication of homotropic cooperativity between the different sites. A model including all these findings, as well as the interactions stabilizing the T state or the R state in the presence of the natural substrates, is proposed.(ABSTRACT TRUNCATED AT 250 WORDS)     

13C isotope effects as a probe of the kinetic mechanism and allosteric properties of Escherichia coli aspartate transcarbamylase.

13C kinetic isotope effects have been measured in carbamyl phosphate for the reaction catalyzed by aspartate transcarbamylase. For the holoenzyme, the value was 1.0217 at zero aspartate, but unity at infinite aspartate, with 4.8 mM aspartate eliminating half of the isotope effect. This pattern proves an ordered kinetic mechanism, with carbamyl phosphate adding before aspartate. The same parameters were observed in the presence of ATP or CTP, showing that there is only one form of active enzyme present, regardless of the presence or absence of allosteric modifiers. These data support the Monod model of allosteric behavior in which the equilibrium between fully active and inactive enzyme is perturbed by selective binding interactions of substrates and modifiers, and there are no enzyme forms having partial activity. Isolated catalytic subunits of the enzyme showed similar 13C isotope effects (1.0240 at zero aspartate, 1.0039 at infinite aspartate, 3.8 mM aspartate causing half of the change from one value to the other), but the finite isotope effect at infinite aspartate shows that the kinetic mechanism is now partly random. With the very slow and poorly bound aspartate analog cysteine sulfinate, the 13C isotope effects were 1.039 for both holoenzyme and catalytic subunits and were not decreased significantly by high levels of cysteine sulfinate. The value of 1.039 is probably close to the intrinsic isotope effect on the chemical reaction, while the kinetic mechanism with this substrate is now fully random because the chemistry is so much slower than release of either reactant from the enzyme.     

The effect of pH on the cooperative behavior of aspartate transcarbamylase from Escherichia coli.

Saturation curves of activity versus concentration were determined for aspartate transcarbamylase from Escherichia coli (EC 2.1.3.2) for the substrate L-aspartate at saturating carbamyl phosphate (4.8 mM) in buffered solution at pH values from 6.0 to 12.0. Hill coefficients were obtained from the sigmoidal curves. At pH values from 7.8 to 9.1, where substrate inhibition causes difficulties in the Hill approximation, our kinetic scheme includes substrate inhibition and residual activity in the abortive enzyme-substrate complex. The plot of Hill coefficient versus pH has pKalpha values of 7.4 and 9.8 at the half-maximum positions of the curve which has a plateau from pH 8.1 to 9.1. These pKalpha values may be associated with functional groups involved in the allosteric transition which activates the enzyme. A plot of [S]0.5 versus pH shows a pKalpha of 8.5, which may belong to a residue either at or near the aspartate binding site. At 50 mM aspartate concentration the pH-rate profile shows maxima at pH values of 8.8 and 10.0 (cf. Weitzman, P.D.J., and Wilson, I.B.(1966)J. Biol. Chem. 2418 5481-5488, who used 100 mM aspartate). However, when the pH-dependent substrate inhibition is included, the calculated Vmax--H curve is bell-shaped like that of the isolated catalytic subunit.     

The pAR5 mutation and the allosteric mechanism of Escherichia coli aspartate carbamoyltransferase.

Mutation pAR5 replaces residues 145'-153' at the C terminus of the regulatory (r) chains of Escherichia coli ATCase by a new sequence of six residues. The mutated enzyme has been shown to lack substrate cooperativity and inhibition by CTP. Solution X-ray scattering curves demonstrate that, in the absence of ligands, its structure is intermediate between the T form and the R form. In the presence of N-phosphonacetyl-L-aspartate, the mutant is similar to the wild type. An examination of the crystal structure of unligated ATCase reveals that the mutated site is at an interface between r and catalytic (c) chains, which exists only in the T allosteric form. A computer simulation by energy minimization suggests that the pAR5 mutation destabilizes this interface and induces minor changes in the tertiary structure of r chains. The resulting lower stability of the T form explains the loss of substrate cooperativity. The lack of allosteric inhibition may be related to a new electrostatic interaction made in mutant r chains between the C-terminal carboxylate and a lysine residue of the allosteric domain.     

The allosteric activator ATP induces a substrate-dependent alteration of the quaternary structure of a mutant aspartate transcarbamoylase impaired in active site closure.

Aspartate transcarbamoylase from Escherichia coli shows homotropic cooperativity for aspartate as well as heterotropic regulation by nucleotides. Structurally, it consists of two trimeric catalytic subunits and three dimeric regulatory subunits, each chain being comprised of two domains. Glu-50 and Ser-171 are involved in stabilizing the closed conformation of the catalytic chain. Replacement of Glu-50 or Ser-171 by Ala in the holoenzyme has been shown previously to result in marked decreases in the maximal observed specific activity, homotropic cooperativity, and affinity for aspartate (Dembowski NJ, Newton CJ, Kantrowitz ER, 1990, Biochemistry 29:3716-3723; Newton CJ, Kantrowitz ER, 1990, Biochemistry 29:1444-1451). We have constructed a double mutant enzyme combining both mutations. The resulting Glu-50/ser-171-->Ala enzyme is 9-fold less active than the Ser-171-->Ala enzyme, 69-fold less active than the Glu-50-->Ala enzyme, and shows 1.3-fold and 1.6-fold increases in the [S]0.5Asp as compared to the Ser-171-->Ala and Glu-50-->Ala enzymes, respectively. However, the double mutant enzyme exhibits some enhancement of homotropic cooperativity with respect to aspartate, relative to the single mutant enzymes. At subsaturating concentrations of aspartate, the Glu-50/Ser-171 -->Ala enzyme is activated less by ATP than either the Glu-50-->Ala or Ser-171-->Ala enzyme, whereas CTP inhibition is intermediate between that of the two single mutants. As opposed to the wild-type enzyme, the Glu-50/Ser-171 -->Ala enzyme is activated by ATP and inhibited by CTP at saturating concentrations of aspartate. Structural analysis of the Ser-171-->Ala and Glu-50/Ser-171-->Ala enzymes by solution X-ray scattering indicates that both mutants exist in the same T quaternary structure as the wild-type enzyme in the absence of ligands, and in the same R quaternary structure in the presence of saturating N-(phosphonoacetyl)-L-aspartate. However, saturating concentrations of carbamoyl phosphate and succinate are unable to convert a significant fraction of either mutant enzyme population to the R quaternary structure, as has been observed previously for the Glu-50-->Ala enzyme. The curves for both the Ser-171-->Ala and Glu-50/Ser-171-->Ala enzymes obtained in the presence of substoichiometric amounts of PALA are linear combinations of the two extreme T and R states. The structural consequences of nucleotide binding to these two enzymes were also investigated. Most surprisingly, the direction and amplitude of the effect of ATP upon the double mutant enzyme were shown to vary depending upon the substrate analogue used.     

A loop involving catalytic chain residues 230-245 is essential for the stabilization of both allosteric forms of Escherichia coli aspartate transcarbamylase

The allosteric transition of Escherichia coli aspartate transcarbamylase involves significant alterations in structure at both the quaternary and tertiary levels. On the tertiary level, the 240s loop (residues 230-245 of the catalytic chain) repositions, influencing the conformation of Arg-229, a residue near the aspartate binding site. In the T state, Arg-229 is bent out of the active site and may be stabilized in this position by an interaction with Glu-272. In the R state, the conformation of Arg-229 changes, allowing it to interact with the beta-carboxylate of aspartate, and is stabilized in this position by a specific interaction with Glu-233. In order to ascertain the function of Arg-229, Glu-233, and Glu-272 in the catalytic and cooperative interactions of the enzyme, three mutant enzymes were created by site-specific mutagenesis. Arg-229 was replaced by Ala, while both Glu-233 and Glu-272 were replaced by Ser. The Arg-229----Ala and Glu-233----Ser enzymes exhibit 10,000-fold and 80-fold decreases in maximal activity, respectively, and they both exhibit a 2-fold increase in the aspartate concentration at half the maximal observed velocity, [S]0.5. The Arg-229----Ala enzyme still exhibits substantial homotropic cooperativity, but all cooperativity is lost in the Glu-233----Ser enzyme. The Glu-233----Ser enzyme also shows a 4-fold decrease in the carbamyl phosphate [S]0.5, while the Arg-229----Ala enzyme shows no change in the carbamyl phosphate [S]0.5 compared to the wild-type enzyme. The Glu-272 to Ser mutation results in a slight reduction in maximal activity, an increase in [S]0.5 for both aspartate and carbamyl phosphate, and reduced cooperativity. Analysis of the isolated catalytic subunits from these three mutant enzymes reveals that in each case the changes in the kinetic properties of the isolated catalytic subunit are similar to the changes caused by the mutation in the holoenzyme. PALA was able to activate the Glu-233----Ser enzyme, at low aspartate concentrations, even though the mutant holoenzyme did not exhibit any cooperativity, indicating that cooperative interactions still exist between the active sites in this enzyme. It is proposed that Glu-233 of the 240s loop helps create the high-activity-high-affinity R state by positioning the side chain of Arg-229 for aspartate binding while Glu-272 helps stabilize the low-activity-low-affinity T state by positioning the side chain of Arg-229 so that it cannot interact with aspartate.(ABSTRACT TRUNCATED AT 400 WORDS)     

The quaternary structure of RNase G from Escherichia coli

RNase G is the endoribonuclease responsible for forming the mature 5' end of 16S rRNA. This enzyme shares 35% identity with and 50% similarity to the N-terminal 470 amino acids encompassing the catalytic domain of RNase E, the major endonuclease in Escherichia coli. In this study, we developed non-denaturing purifications for overexpressed RNase G. Using mass spectrometry and N-terminal sequencing, we unambiguously identified the N-terminal sequence of the protein and found that translation is initiated at the second of two potential start sites. Using velocity sedimentation and oxidative cross-linking, we determined that RNase G exists largely as a dimer in equilibrium with monomers and higher multimers. Moreover, dimerization is required for activity. Four of the six cysteine residues of RNase G were mutated to serine. No single cysteine to serine mutation resulted in a complete loss of cross-linking, dimerization or activity. However, multiple mutations in a highly conserved cluster of cysteines, including C405 and C408, resulted in a partial loss of activity and a shift in the distribution of RNase G multimers towards monomers. We propose that many of the cysteines in RNase G lie on its surface and define, in part, the subunit-subunit interface.     

Glu-50 in the catalytic chain of Escherichia coli aspartate transcarbamoylase plays a crucial role in the stability of the R quaternary structure.

Glu-50 of aspartate transcarbamoylase from Escherichia coli forms a set of interdomain bridging interactions between the 2 domains of the catalytic chain; these interactions are critical for stabilization of the high-activity high-affinity form of the enzyme. The mutant enzyme with an alanine substituted for Glu-50 (Glu-50-->Ala) exhibits significantly reduced activity, little cooperativity, and altered regulatory behavior (Newton CJ, Kantrowitz ER, 1990, Biochemistry 29:1444-1451). A study of the structural consequences of replacing Glu-50 by alanine using solution X-ray scattering is reported here. Correspondingly, in the absence of substrates, the mutant enzyme is in the same, so-called T quaternary conformation as is the wild-type enzyme. In the presence of a saturating concentration of the bisubstrate analog N-phosphonacetyl-L-aspartate (PALA), the mutant enzyme is in the same, so-called R quaternary conformation as the wild-type enzyme. However, the Glu-50-->Ala enzyme differs from the wild-type enzyme, in that its scattering pattern is hardly altered by a combination of carbamoyl phosphate and succinate. Addition of ATP under these conditions does result in a slight shift toward the R structure. Steady-state kinetic studies indicate that, in contrast to the wild-type enzyme, the Glu-50-->Ala enzyme is activated by PALA at saturating concentrations of carbamoyl phosphate and aspartate, and that PALA increases the affinity of the mutant enzyme for aspartate. These data suggest that the enzyme does not undergo the normal T to R transition upon binding of the physiological substrates and verifies the previous suggestion that the interdomain bridging interactions involving Glu-50 are critical for the creation of the high-activity, high-affinity R state of the enzyme.     

Effects of the T→R transition on the electrostatic properties of E. coli aspartate transcarbamylase

Aspartate transcarbamylase is a large (310 kD), multisubunit protein that binds substrates cooperatively and undergoes a large change in quaternary structure when substrates bind. The forces that drive this transition are poorly understood. We evaluated the electrostatic component of these forces by using finite difference and multigrid methods to solve the nonlinear Poisson-Boltzmann equation for complexes of the enzyme with several substrates and substrate analogs. The results have been compared with calculations for the unliganded protein. While pK_½ values of most ionizable residues fall within 3 pH units of values for model compounds, 31 have pK_½ values that fall outside the range 0–17. Many of these residues are at the active site, where they interact with the highly charged substrate, in the 80s loop or 240s loop or interact with these loops. The pK_½ values of eight ionizable residues related by the twofold molecular axes differ by more than 3 pH units, providing additional evidence for asymmetry within the crystal. As in the unliganded structure, a set of residues forms a network in which ionizable groups with W_ij values greater than 2 kcal-m^-1 are separated by distances greater than 5 Å. Some residues participate in this network in both the unliganded and N-phosphonacetyl-L-aspartate (PALA)-liganded structure, while others are found in only one structure. The network is more extensive in the PALA-liganded structure than in the unliganded structure, but consists of two separate networks in the two halves of the molecule. Proteins 32:200–210, 1998. © 1998 Wiley-Liss, Inc.