|Home | About | Journals | Submit | Contact Us | Français|
Conformational flexibility is emerging as a central theme in enzyme catalysis. Thus, identifying and characterizing enzyme dynamics is critical for understanding catalytic mechanisms. Herein, coupling analysis, which uses thermodynamic analysis to assess cooperativity/coupling between distal regions on an enzyme, is used to interrogate substrate specificity among fructose-1,6-(bis)phosphate aldolase (aldolase) isozymes. Aldolase exists as three isozymes, A, B, and C distinguishable by their unique substrate preferences despite the fact that the structures of the active sites of the three isozymes are nearly identical. While conformational flexibility has been observed in aldolase A, its function in the catalytic reaction of aldolase has not been demonstrated. To explore the role of conformational dynamics in substrate specificity, those residues associated with isozyme specificity (ISRs) were swapped and the resulting chimeras were subjected to steady-state kinetics. Thermodynamic analyses suggest cooperativity between a terminal surface patch (TSP) and a distal surface patch (DSP) of ISRs that are separated by >8.9Å. Notably, the coupling energy (ΔGI) is anti-correlated with respect to the two substrates, fructose 1,6-bisphosphate and fructose 1-phosphate. The difference in coupling energy with respect to these two substrates accounts for about 70% of the energy difference for the ratio of kcat/Km for the two substrates between aldolase A and aldolase B. These non-additive mutational effects between the TSP and DSP provide functional evidence that coupling interactions arising from conformational flexibility during catalysis are a major determinant of substrate specificity.
Recent evidence shows dynamic fluctuations of structure are essential components of enzyme catalysis (1). While it is likely that most enzymes exhibit flexibility as part of the catalytic process, the role of this movement is enzyme specific. For example, in the case of dihydrofolate reductase (DHFR), backbone and side-chain motions are essential for cofactor binding, substrate binding, and catalysis (2). Thus, identifying and characterizing these motions is paramount to understanding enzyme mechanisms.
Recently, the importance of conformational flexibility has been highlighted in the study of ligand binding and catalysis, as inferred from X-ray crystallographic and NMR structures as well as computational analysis. The term “allostery” has been applied to the transmission of energy resulting from structural rearrangement and/or dynamic changes within a single-domain protein (3–5). In addition to structural and computational data, evolutionary data for a protein family has been used to show that conserved sets of interacting residues form connected pathways through the protein to transmit changes upon ligand binding (6). Not only ligand binding but catalysis is dependent upon inherent mobility of a protein during the catalytic cycle with one or more of the catalytic steps being reflected in an intrinsic motion of the enzyme (for review see (1) and for a recent example (7)). Thus, the innate mobility of the protein scaffold is manifested in the various steps of the catalytic cycle. Indeed, extensive studies of hydrogen transfer in proteins are consistent with the concept that protein dynamics create transient heavy-atom configurations with a favorable electrostatic environment for proton and hydride transfer (1, 8–12). In comparison, determining how conformational flexibility affords the remarkable selectivity of enzymes has received less effort. However, recent work on T7-DNA polymerase has invoked a role for enzyme dynamics in distinguishing correct nucleotide incorporation from misincorporation (13).
Here, multiple mutational analyses in fructose-1,6-(bis)phosphate aldolase (aldolase, EC 126.96.36.199) are used to identify and characterize conformational coupling as a potential mechanism conferring substrate specificity among aldolase isozymes. Aldolase catalyzes the reversible C-C bond cleavage of fructose 1,6-(bis)phosphate (Fru 1,6-P2) and fructose 1-phosphate (Fru 1-P) in its roles in glycolysis, gluconeogenesis, and fructose metabolism (14). Aldolase isozymes are distinguished by their unique expression profiles and substrate specificities (15). Aldolase A is expressed primarily in skeletal muscle and has the highest turnover toward Fru 1,6-P2 of the three isozymes (15–17). Aldolase B is expressed primarily in liver and kidney where most Fru-1-P degradation takes place (14, 18). While the catalytic mechanism of aldolase is well characterized (19–21), and crystal structures of all three human isozymes have been determined (22–24), it is still unclear how aldolase catalyzes the cleavage of the two substrates with distinct and physiologically relevant efficiencies. For example, consistent with its role in gluconeogenesis, aldolase B has equal turnover numbers for the two hexose substrates and has a lower Km value for glyceraldehyde 3-phosphate and dihydroxyacetone phosphate than aldolase A (17). Correspondingly, aldolase A has evolved to cleave Fru 1,6-P2 more efficiently than Fru 1-P (as evidenced by the kcat/Km ratio for Fru 1,6-P2:Fru 1-P of 70,000). This ratio for aldolase B is ~900, therefore the evolution of these parameters for aldolase A compared to that of aldolase B has resulted in a ~80-fold (70,000/900) difference in selectivity between the two hexoses.
Previous work attempting to decipher the root of isozyme specificity has identified a set of residues conserved among, but not between each isozyme, termed isozyme-specific residues (ISRs). Surprisingly, ISRs are not located at or near the active site, but mostly (24 out of 27) fall on the surface of the protein (25). Moreover, a large proportion of ISRs are found in the C-terminal region (CTR) of the enzyme. The role of ISRs was previously examined by swapping ISRs from aldolase B into aldolase A demonstrating that the ISRs are necessary and sufficient to confer kinetic parameters (kcatand Km) of aldolase B onto aldolase A (25). Because of their conservation and involvement in conferring substrate specificity, it is hypothesized that interactions between ISRs are responsible for altering the structure of the active site from a distance.
One means by which enzymes may confer substrate specificity at a distance would involve dynamic cooperation/coupling among sets of ISRs. Herein, multiple mutational analyses were employed as a novel method to attribute the role of conformational coupling to substrate specificity among aldolase isozymes. Traditionally, this analysis is performed by making two separate, single-residue mutations and the corresponding double mutation, calculating the change in free energy (ΔΔG) associated with these changes, and comparing the change in free energy with that of the double mutant (9, 26, 27). If the changes in free energy of the two single mutants do not sum to the change in free energy for the double mutant, then a cooperative/coupling mechanism between the residues is invoked. To perform the analogous experiment for aldolase ISRs, groups of mutations were analyzed rather than single amino-acid substitutions. Multiple mutational analyses revealed that substrate specificity of aldolase A was conferred by cooperative effects between two distantly located surface patches, one of which includes the CTR. This analysis exposes a means by which these cooperative effects among ISRs affect substrate specificity in enzymes with seemingly identical active sites.
Restriction endonucleases, T4 DNA ligase, VentR® DNA polymerase and ThermoPol buffer were from New England Biolabs. AmpliTaq® DNA polymerase was from Applied Biosystems. Glycerol-3-phosphate dehydrogenase/triose phosphate isomerase were from Roche Applied Science. Deoxynucleoside triphosphates and CM-Sepharose® CL-6B were from Amersham Biosciences. Oligonucleotides used for site-directed mutagenesis and sequencing were from Midland Certified Reagent Company, Inc. All other chemicals were from Sigma-Aldrich Chemical Co.
All mutant aldolases were generated via multiple site-directed mutagenesis using overlapping oligonucleotides (28). Unless otherwise stated, all oligonucleotides complement pPB14, which expresses rabbit aldolase A (29). Oligonucleotide sequences are provided in supplemental material.
Full-length cassettes were subcloned into EcoR I and Hind III sites in pPB1 (30). The different expression constructs are described in Table 1. The expression plasmid, pAB_DSP, was constructed with the following oligonucleotides: geneU, AB_DSP1, AB_DSP2, AB_DSP3, AB_DSP4, AB_DSP5, AB_DSP6, and geneL. The expression plasmid, pAB_TSP was generated in two steps. First, the plasmid pAB_Cterm was constructed with the following oligonucleotides: geneU, ABC1, ABC2, and geneL. Secondly, pAB_Cterm was used as a template for the following oligonucleotides to generate pAB_TSP: geneU, ABD1, ABD2, AB_TSP3, AB_TSP4, AB_TSP5, AB_TSP6, and gene L. The expression plasmid, pAB_P, was also generated in two steps. First, the plasmid pInt was created by splicing together the 474 bp fragment generated by cutting pAB_TSP with Nco I and Hind III into the 3,402 bp fragment of pAB_DSP, cut with the same enzymes. Secondly, pInt was used as a template for the following oligonucleotides to generate pAB_P: gene U, ABD1, ABD2, AB_DSP5, AB_DSP6, and geneL. The expression plasmid, pAB_NP, was constructed with the following oligonucleotides: geneU, AB_NP1, AB_NP2, AB_NP3, AB_NP4, AB_All5, AB_All6, 5′--3ime; (forward); AB_All7, AB_All8, AB_NP5, AB_NP6, ABD9, ABD10, AB_NP7, AB_NP8, and geneL. The expression plasmid, pAB_NPS, was constructed with the following oligonucleotides: geneU, AB_NP3, AB_NP4, AB_All5, AB_All6, AB_NP5, AB_NP6, ABD9, ABD10, and AB_NP7. The pAB_NPS full-length PCR product was subcloned into pPB14 at the EcoR I and Eag I sites. The expression plasmid, pAB_NPB, was constructed with the following oligonucleotides: geneU, AB_NP1, AB_NP2, AB_All7, AB_All8, AB_NP7, AB_NP8, and geneL. The expression plasmid, pAld337, was constructed in two steps. First, the oligonucleotides U1nde and L337sap, complementing pPB14, were used to generate a PCR fragment corresponding to the first 337 amino acids of rabbit aldolase A. The engineered restriction sites Nde I and Sap I were used for in-frame subcloning in the Sce VMA intein/chitin binding domain from pTYB1 (New England Biolabs). The construct encoding an aldolase with only two Cys residues, gBi, used a 207 bp BstE II - Eag I fragment from pGTET, which has all surface Cys (72, 239, 289, and 338) substituted by Ala (31), ligated pPB14. The 207 bp fragment encodes the C239A and C289A substitutions of pGTET. All fragments were amplified as previously described (25) and resulting clones were screened by DNA sequence determination (Boston University Core Sequencing Facility).
Expression and purification of substituted aldolases was performed as described previously (29). Briefly, purification used affinity elution from CM-Sepharose CL-6B with 2.5 mM Fru 1,6-P2 at pH 8.3 for AB_NP, AB_P, AB_NPS, AB_NPB, AB_TSP, AB_DSP, and gBi. The truncated form of aldolase, Ald337, was purified using the IMPACT™-CN protein purification system as previously described (32).
Aldolase activity towards Fru 1,6-P2 and Fru 1-P was measured as described previously (29, 33). Briefly, a decrease in A340 was measured from an assay coupled to β-NADH oxidation by glycerol-3-phosphate dehydrogenase. Enzymes were diluted in 50 mM TEA-HCl, pH 7.4, 10 mM EDTA, 20 μg/ml glycerol-3-phosphate dehydrogenase/triose phosphate isomerase and 0.2 mM β-NADH. For the chimera AB_P, Fru 1,6-P2 concentration ranged from 0.6 to 6.0 μM with 2 μg/ml enzyme and Fru 1-P concentration ranged from 1,100 to 8,000 μM with 4 μg/ml enzyme. For chimeras AB_NP, AB_NPS and AB_NPB, Fru 1,6-P2 concentration ranged from 3.0 to 100 μM with 0.5 μg/ml enzyme and Fru 1-P concentration ranged from 2,500 to 80,000 μM with 100 μg/ml enzyme. For the chimera AB_TSP, Fru 1,6-P2 concentration ranged from 4.5 to 35 μM with 1.1 μg/ml enzyme and Fru 1-P concentration ranged from 5,000 to 60,000 μM with 10 μg/ml enzyme. For the chimera AB_DSP, Fru 1,6-P2 concentration ranged from 5.0 to 80 μM with 0.5 μg/ml enzyme and Fru 1-P concentration ranged from 5,000 to 60,000 μM with 10 μg/ml enzyme. For the chimera Ald337, Fru 1,6-P2 concentration ranged from 20 to 600 μM with 18 μg/ml enzyme and Fru 1-P ranged from 2,500 to 80,000 μM with 70 μg/ml enzyme. For gBi, Fru 1,6-P2 concentration ranged from 6 to 200 μM with 1.3 μg/ml enzyme. Assays were performed at 30°C in triplicate using a SpectroMAX 190 spectrophotometer (Molecular Devices, Inc.) in a final volume of 350 μl.
Oxidation was performed as described previously (34) except for changes in concentrations. The oxidants were o-phenanthroline (60 μM) and cupric sulfate (20 μM). Concentrations of Fru 1,6-P2 and Fru 1-P were 10Km (0.5 mM and 200 mM, respectively). Briefly, aldolases (1 mg/ml) in 100 mM Tris-HCl, pH 8.4, were incubated with oxidants in the presence or absence of substrate. At various times, aliquots (10 μg) were removed, diluted 100-fold, and assayed for activity at 25 °C. The presence of un-oxidized thiols was determined by reaction of 90 μg aliquots of aldolase with 3 mM 5,5′dithiobis(2-nitrobenzoic acid) (DTNB) in 100 mM Tris, pH 9.4 (34). Absorbance was measured at 412 nm after five minutes at 25 °C.
Isozyme-specific residues (ISRs) that confer the kinetic properties of the aldolase isozymes have previously been identified (25). Here, ISRs were further analyzed based on their location in the three-dimensional structure (Fig 1). Of the 23 ISRs in aldolase A and aldolase B, 16 cluster into two surface patches, the terminal surface patch (TSP) and the distal surface patch (DSP), hereafter referred to as “patch” ISRs and the remaining seven residues are referred to as “non-patch” ISRs. Non-patch ISRs were categorized based on their exposure to solvent. Four of the seven non-patch ISRs were solvent accessible as determined from a Connolly surface map (35), while the remaining three residues were buried (residues 58, 182, and 327).
To determine if potential coupling interactions among these conserved groups of residues confers the differential kinetics between isozymes, chimeric enzymes were created by swapping each subset of ISRs (Table 1). Swapping ISRs was performed by incorporating aldolase B ISRs into aldolase A while replacing aldolase A ISRs with the corresponding aldolase B residue. The location of these substitutions in the primary sequence is shown in Fig 2. The resulting chimeric proteins were expressed in E. coli and purified to >95% as assayed by SDS-PAGE.
All chimeras were characterized via steady-state kinetics with the substrates Fru 1,6-P2 and Fru 1-P (Table 2). It is known that the rate-limiting step for wild-type aldolases is product release (20, 36), and since the chimeras are fully active enzymes, the rate-limiting step is not likely changed for the chimeras. If so, Km would be defined identically for all enzymes. The chimera in which all ISRs had been swapped from aldolase B into aldolase A (AB_All) exhibits kcat and Km parameters indistinguishable from aldolase B (25). Compared to AB_All, the loss of non-patch residues in AB_P resulted in a five-fold increase in Km towards Fru 1,6-P2, making it more similar to aldolase A than aldolase B (Table 2). However, the turnover number was very similar to that of aldolase B. Towards the substrate Fru 1-P, AB_P displayed Km and kcat values more similar to aldolase B than aldolase A. In a complementary experiment, the contribution of non-patch ISRs alone was analyzed with the construct AB_NP, which showed kinetic parameters toward Fru 1,6-P2 and Fru 1-P more similar to aldolase A than aldolase B. The possible exception was the kcat toward Fru 1-P, which was similar to that of aldolase B. Overall, the results are consistent with an effect on Km toward Fru 1,6-P2 associated with the combined presence of patch and non-patch residues.
Similarly, the involvement of the two patches of AB_P in substrate specificity was analyzed using the chimeras AB_TSP and AB_DSP. The kinetic profile of AB_TSP towards Fru 1-P showed that the kcat was similar to that of aldolase B (Table 2). The Km value was decreased by 5-fold compared to aldolase A, making it intermediate between the values for A and B. Remarkably, towards Fru 1,6-P2, AB_TSP had a Km value more similar to that of aldolase A, yet attained a kcat indistinguishable from aldolase B. As for AB_DSP, steady-state kinetic analysis towards Fru 1-P revealed activities similar to those of AB_TSP, whereas the kinetic parameters towards Fru 1,6-P2 were indistinguishable from those of aldolase A.
Similar analysis of the chimeras that separate the surface and buried non-patch ISRs using AB_NPS and AB_NPB showed that they were similar to each other and each demonstrated the same kinetic differences from wild-type aldolase A as did the combination of the two (AB_NP) (Table 2). Thus, dividing the non-patch residues into buried and surface residues did not dissect function. In addition, comparison of all non-patch chimeras (AB_NP, AB_NPS, and AB_NPB) relative to aldolase A and B showed that the role of these residues in distinguishing kinetic parameters was relatively minor. The fact that the kcat value toward Fru 1-P increased >2-fold for most chimeras relative to aldolase A demonstrated that all these enzymes were not simply impaired in function.
Interaction energies between two substitution sites can be assessed by summing the change in energy associated with each substitution and comparing this sum to the change in energy associated with the double substitution (9, 26, 27). This approach was applied to the ISR subsets; the subsets of ISRs represent a single substitution and a combination of these subsets represents the double substitution. Three thermodynamic cycles among such aldolase chimeras were analyzed (Fig 3). First, the change in free energy associated with the separate patches, AB_TSP and AB_DSP was compared to the swapping of both patches (AB_P) (Fig 3A). Second, the change in free energy associated with separating patch (AB_P) and non-patch (AB_NP) residues was compared to those produced by swapping all ISRs (AB_All) (Fig 3B). Finally, the free energy associated with separating the non-patch ISRs into surface (AB_NPS) and buried (AB_NPB) residues was compared to the swapping of all non-patch (AB_NP) residues (Fig 3C).
where ΔΔGT‡ is the transition-state stabilization energy or change in free energy to reach the transition-state complex from free enzyme and substrate, R is the gas constant, and T is the absolute temperature. The sum of the changes in free energy for the two single sets of substitutions is related to the change in free energy of the double sets of substitutions by equation 2:
where ΔΔG (X,Y)‡ is the difference in free energy between wild type and the chimera with the double set of substitutions, ΔΔG(X)‡ and ΔΔG(Y)‡ are the differences in free energy between wild type and the two chimeras with a single set of substitutions, and ΔGI, or coupling energy, is the interaction energy between substitution sites (38, 39). If ΔGI ≈ 0, then the interactions between the sets of ISRs are additive, thus indicating that the ISRs have independent effects. If ΔGI ≈ 0, then the effects are non-additive, indicating that one set of ISRs is affected by the other set of ISRs. Negative ΔGI values represent a positive coupling interaction that enhances the kinetic parameter measured, while positive ΔGI values represent coupling interaction that reduce this parameter (40). This analysis was repeated with kcat, and again with Km, substituting each for kcat/Km in equation 1.
The ΔGI calculated for both Fru 1,6-P2 and Fru 1-P using the three thermodynamic cycles are shown in Tables 3 and and4,4, respectively. For the thermodynamic cycle “Patch”, there was a non-zero ΔGI for the kinetic parameter kcat/Km toward Fru 1,6-P2 (−0.7 kcal/mol). As described above, the negative ΔGI indicated coupling energy that enhances the kinetic parameter kcat/Km. This cooperativity was largely associated with the rate-limiting step, kcat (−0.6 kcal/mol). The same analysis using Fru 1-P demonstrated a similar cooperative effect associated with kcat/Km (ΔGI = 1.0 kcal/mol). However, the change in sign indicated a negative cooperativity between these two patches for this substrate. Again, as it was for Fru 1,6-P2, this cooperativity was largely due to an effect on kcat (0.7 kcal/mol).
On the other hand, for the thermodynamic cycles “All” and “Non-patch”, the change in free energies associated with each kinetic parameter for Fru 1,6-P2 was additive (ΔGI ≈ 0). However, the coupling energy for Fru 1-P indicated a small but significant coupling energy between non-patch surface and non-patch buried residues associated with kcat/Km (ΔGI = 0.3 kcal/mol). This was the net effect of two larger but opposing changes on kcat (ΔGI = 0.7 kcal/mol) and Km (ΔGI = 0.45 kcal/mol). Of the seven non-patch surface and non-patch buried residues, five are aldolase B ISRs suggesting that the coupling interaction between these residues is conserved mainly in aldolase B. It must be noted that this analysis is based on the assumption that the rate-limiting step (product release) for the chimeras has not changed such that Km has the same definition for all enzymes. Moreover, the pseudo second-order rate constants, kcat/Km, are comparable and it is in this constant that the important distinctions lie.
Coupling interactions between sets of ISRs acting at a distance affects substrate specificity. As details of these interactions are unknown, we hypothesized that the flexible C-terminus of aldolase, which is part of the TSP, is a potential “bridge” for mediating these interactions. This is based on X-ray crystallographic data in which disorder (19, 24, 41) or multiple conformations (23, 42–44) of the 18–20 C-terminal residues (CTR) indicate flexibility. Moreover, removal of residues in the CTR of aldolase A disrupts Fru-1,6-P2 turnover with limited effects on Fru-1-P turnover (21, 45, 46).
To test this model, a C-terminal mutant of aldolase was generated and the steady-state kinetic parameters were determined. For this variant, the entire CTR was deleted, to make Ald337 (Fig 2), which exhibited a 30–100-fold decrease in kcat compared to wild type for both substrates (Table 2). However, the Km values were affected in opposite directions; toward Fru 1,6-P2, the Km value increased 20-fold, whereas toward Fru 1-P Km decreased 20-fold. Thus, this distinction highlights the importance of the CTR in differential effects on substrate binding and intermediate steps in the catalytic cycle (Km), and in substrate-specific rate-limiting steps (kcat), likely involving conformational changes. This conclusion was reinforced by comparison to an enzyme that added back the CTR, but included all the ISRs from the TSP of aldolase B (AB_TSP). The turnover number for both Fru-1,6-P2 and Fru-1-P cleavage was restored to the values for aldolase B (Table 2). Yet, the values for Km remained more like aldolase A. Thus, the inclusion of the CTR alone is not sufficient to confer the kcat values of aldolase B, and the CTR, while required for full activity toward both substrates, may behave differently upon the binding of each.
The thermodynamic analyses of the kinetic behavior pinpointed the involvement of both the TSP and DSP regions of aldolase. ISRs that comprise the TSP and DSP are generally located on opposite faces of a previously unappreciated cluster of α-helices (Fig 4). The first and last helices off the α/β-barrel (α2 and α13) along with three other non-barrel helices (α3, α4, and α14) form the five-helix cluster. One mechanism explaining the cooperativity in determination of aldolase substrate specificity may involve conformational changes of these helices during catalysis.
Two analyses were preformed to test the model that the α-helical cluster exhibits substrate-dependent conformational flexibility. The first was an analysis of B-factors in the region of the α-helical cluster (Fig 5). Indeed, parts of helices α2, α13, and α14 show elevated values relative to the average B-factor of the structure. However, interpretation of mobility from X-ray crystallographic data is not definitive. A direct analysis was performed that derives from the well-known effect of oxidation on aldolase activity (34, 47). Oxidation of surface cysteine residues on aldolase can be protected against loss of activity to different extents by different substrates (34). It is thought that the site of oxidation is probably due to the proximal residues Cys72 and Cys338, which are in helices α4 and α14 of the α-helical cluster (see Fig 4). A construct was made that removed the two other reactive cysteine groups and left only Cys72 and Cys338. This construct, gBi, was purified, characterized, and was shown to possess kinetic values similar to wild-type aldolase (kcat was 14 s−1 and Km toward Fru 1,6-P2 was 8.6 μM). This aldolase variant was used to test if oxidation of these Cys residues in the α-helical cluster would show different rates of activity loss due to oxidation in the presence of Fru 1,6-P2 versus Fru 1-P. When substrates were used at the same concentrations (relative to Km), Fru 1,6-P2 showed much slower loss of activity compared to Fru 1-P, which was itself slower relative to enzyme without any substrate (Fig 6). That this effect was due to oxidation at these two Cys residues was shown by the lack of any substrate-dependent loss of activity in the aldolase variant, gTet, which does not possess any surface Cys residues (31), and in the loss of DTNB reactivity at the end of the oxidation reaction, which confirmed loss of available thiols in gBi (data not shown). This result is a clear and direct demonstration that the conformation of the helixes containing these two proximal Cys residues in the α-helical cluster is different depending on the substrate present.
Recently, it has been proposed that enzyme dynamics is a necessary feature of enzyme catalysis and that structural flexibility dictates the rate of enzyme catalyzed reactions (1, 12). Thus, identifying what part of an enzyme moves and the consequences of these movements is critical for understanding enzyme kinetics and function. Crystal structures at various stages of catalysis are often used to identify the result of structural movements. For example, crystal structures of DHFR reveal a flexible loop that lies directly over the active site. Movement of the loop causes alterations of the architecture of the active site (8). Alternatively, direct visualization of structural movements is possible via NMR (48, 49). Recently, direct observation of motion has been made for T7 DNA polymerase using a conformation sensitive fluorophore to show that discrimination between correct and incorrect nucleotides depends on substrate-induced structural alignment or misalignment of catalytic residues (13).
The first evidence for conformational flexibility in aldolase came from studies measuring the accessibility of a surface cysteine in aldolase A, which links conformational flexibility with substrate turnover (47). In these studies, the degree of modification of Cys239 (a residue outside the active site) varied depending on the different substrates present; however, these studies did not identify the elements that were moving, or how the movements might differ for different substrates. In attempts to define the important regions as the source of substrate specificity between aldolase isozymes, others have swapped relatively large sections of the primary sequence between isozymes (50–54). Not surprisingly, the results suggested a role for the CTR in catalysis; however, the minimal set of residues involved in conferring substrate preferences was not identified. Using a more comprehensive approach, a set of residues (ISRs) conserved within and unique to each isozyme that play a role in specificity among isozymes was defined (25). Given the clustering of these residues into discrete patches outside the active site (>8 Å from any catalytic residue), and knowing that they must be involved in distinguishing isozyme specificity, the question of how or if they cooperate in this task was asked herein using thermodynamic cycles.
Cooperation was directly assessed by interchanging ISR patches between aldolase A and B. Thermodynamic analysis of the kinetic parameters assessed coupling interactions between the various patches of ISRs (Fig 3). The coupling energy between the TSP and DSP revealed that they function cooperatively in analysis of kcat/Km. The positive cooperative effect towards Fru 1,6-P2 (negative ΔGI) was opposite to the effect observed for cleavage of Fru 1-P (positive ΔGI). This distinction is illustrated by plotting the coupling energy associated with kcat/Km for all three thermodynamic cycles towards both substrates (Fig 5). This difference in coupling energy between the two substrates for the two conserved surface patches (TSP and DSP) is consistent with a model that defines the roles of these patches in conferring substrate specificity of aldolase A. These residues work in a cooperative way to enhance kcat for Fru-1,6-P2 cleavage and work in a negative fashion for Fru-1-P cleavage.
The energetic difference between the two wild-type isozymes for the two substrates can be calculated by substituting the kcat/Km ratio of Fru 1,6-P2:Fru 1-P for each isozyme into equation 1. The resulting value of 2.5 kcal/mol is the quantitative description of the substrate specificity difference between isozymes. Thus, the 1.7 kcal/mol measured for the difference in the cooperative effect (ΔGI) between the two ISR patches (Fig 5) accounts for about 70% (1.7/2.5) of the total energy difference between the ratio of specificities toward the substrates between aldolase A and B and is therefore likely the major source of the distinction between the two substrates. These energy values are similar in magnitude to those previously determined for the contribution of conformational coupling to single steps in catalysis. For example, ΔGI from non-additivity among sets of single, double and quadruple substitution mutants of DHFR have been measured for the hydride transfer rates (1.7 to 2.1 kcal/mol) and cofactor binding (0.6 to 1.7 kcal/mol) (9, 55).
Mapping the TSP and DSP on the structure of aldolase A revealed that the closest residues were over 8 Å apart (see Fig 4). The thermodynamic analyses and substrate protection against oxidation support a model that involves differential conformational changes of these regions. In this model, movement of the flexible CTR is coupled to transient structural changes in the ISRs of the TSP and the DSP. Structural changes induced by each substrate would influence the location of the flexible CTR, which is structurally connected to this cluster at α14. Importantly, α2 includes Lys41 and Arg42, which comprise part of the C6-phosphate binding site (56, 57). Thus, a structural connection exists between a substrate binding site and the ISRs involved in determining substrate specificity.
Such a model for conformational flexibility not only fits the data presented herein, but also provides an explanation for previously unexplained observations for aldolase. The previous observation is that Cys72 and Cys338, which are located 8.4 Å apart (42), can form an inactivating disulfide bond (34). Due to the distance separating them, conformational changes would be necessary to allow bond formation. Our analysis reveals that these two cysteines are part of the five-helix cluster. Cys338 is located at the end of α14, and Cys72 located after the end of α4 (see Fig 4). The formation of a disulfide bond between Cys72 and Cys338 could be the result of a trapped transient conformational change in the five-helix cluster. In the model, rigidification of this region by disulfide-bond formation would severely impact catalysis. Secondly, studies measuring the accessibility of surface cysteines in aldolase A have linked conformational changes to catalysis (47). In these studies, the accessibility of Cys239 was monitored by chemical modification in the presence of the substrates. Cys239 was less accessible to protein modification in the presence of Fru 1,6-P2 than in the presence of Fru 1-P. The distance of Cys239 from the active site indicated that long-range conformational changes are coupled to cleavage of specific substrates. Thus, the authors conclude “substrate induced conformational changes in aldolase are syncatalytic in nature” (47).
The aldolase isozymes have evolved to perform functions that are physiologically relevant to the tissue in which they are expressed. Results herein define the differences between substrates in terms of conformational coupling between residues in the TSP and those distant from the active site in the DSP as well as the CTR. The two patches comprise a newly identified structural element composed of five helices. These results explain kinetic distinctions among isozymes that have identical active sites. Thus, this study shows the value of thermodynamic analysis of modified enzymes for revealing cooperative interactions linked to substrate specificity. Use of this method is the first of what will likely be many demonstrating the role of enzyme dynamics in determining substrate specificity.
We thank Dr. Nicholas Silvaggi for critical review of the manuscript.
†This work was supported by Grants GM60616 (to D.R.T. and K.N.A.), DK065089 (to D.R.T), and Training Grant HL07291 (to J.A.P.) from the National Institutes of Health.
1ISR, isozyme-specific residue; Fru 1,6-P2, fructose 1,6-bisphosphate; Fru 1-P, fructose 1-phosphate; CTR, C-terminal region; TSP, terminal surface patch; DSP, distal surface patch; DTNB, 5,5′dithiobis(2-nitrobenzoic acid).