Long-range effects and conformational flexibility of aldolase.

The conformational flexibility and long-range interactions in rabbit muscle aldolase induced by active-site ligand binding, cross-linking of the enzyme between Cys72 and Cys338, and removal of the C-terminal tyrosine residue were studied by following the changes in the microenvironments of Cys239 and Cys289 located outside the active site. It was found that substrates induced a conformational change in aldolase, which propagates from the active site to Cys239, which is located close to intersubunit contacts. The response of the enzyme is differential. Ligands having both C-1 and C-6 phosphates or C-1 phosphate only induce the enhancement of Cys239 reactivity, whereas those with C-6 phosphates only decrease Cys239 reactivity. This correlates well with a dramatic difference in kinetic parameters for a cleavage of fructose-1,6-P2 and fructose-1-P. Therefore, these changes can be interpreted as syncatalytic. Cross-linking of the aldolase subunit by an -S-S-bridge between Cys72 and Cys338 inactivates the enzyme, abolishes binding of active-site ligands, and induces a conformational change in the enzyme that can be detected far away (at Cys239 and Cys289) from the site of perturbation. Cys72 and Cys338 are not in the active site. This shows that the region of the active site and the environment of Cys72 and Cys338 are tightly coupled and that residues far away from the active site, through such coupling, can possess properties of active-site residues. Similar, although less dramatic changes are observed upon removal of the C-terminal tyrosine residue. In view of the results obtained in this paper, aldolase seems to be quite a flexible molecule, whose conformation is sensitive to the nature of a substrate bound to the enzyme and is able to transmit the information about a local perturbation over long distances within a molecule.


Long-range Effects and Conformational Flexibility of Aldolase*
Tomasz Heyduk$, Ryszard Michalczykg, and Marian Kochman From the Division of Biochemistry, Institute of Organic and Physical Chemistry, Technical University, . .

PL-50370 Wroclaw,-Poland
The conformational flexibility and long-range interactions in rabbit muscle aldolase induced by active-site ligand binding, cross-linking of the enzyme between Cys" and CYS~~', and removal of the C-terminal tyrosine residue were studied by following the changes in the microenvironments of C~S~~' and Cys2'' located outside the active site. It was found that substrates induced a conformational change in aldolase, which propagates from the active site to CYS~~', which is located close to intersubunit contacts. The response of the enzyme is differential. Ligands having both C-1 and C-6 phosphates or C-1 phosphate only induce the enhancement of C Y S~~' reactivity, whereas those with C-6 phosphates only decrease CysZ3' reactivity. This correlates well with a dramatic difference in kinetic parameters for a cleavage of fructose-1,6-Pz and fructose-1-P. Therefore, these changes can be interpreted as syncatalytic.
Cross-linking of the aldolase subunit by an -S-Sbridge between Cys7' and Cys338 inactivates the enzyme, abolishes binding of active-site ligands, and induces a conformational change in the enzyme that can be detected far away (at CysZ3' and CYS~~') from the site of perturbation. Cys7' and Cys338 are not in the active site. This shows that the region of the active site and the environment of Cys7' and Cys33s are tightly coupled and that residues far away from the active site, through such coupling, can possess properties of activesite residues. Similar, although less dramatic changes are observed upon removal of the C-terminal tyrosine residue.
In view of the results obtained in this paper, aldolase seems to be quite a flexible molecule, whose conformation is sensitive to the nature of a substrate bound to the enzyme and is able to transmit the information about a local perturbation over long distances within a molecule.
The classical approach to identify active-site residues in enzymes is to study the inactivation of the enzyme by chemical modification and protection against inactivation by substrate binding. The modern version of this approach would be to employ site-directed mutagenesis to study the effect of amino acid substitutions on enzyme catalytic properties. Interpretation of the results of these experiments frequently relies on the assumption that long-range effects are unimpor-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. tant, i.e. for example, perturbation of the enzyme caused by substitution or modification of a residue distant from the active site is not transmitted to the active-site region. In many instances, this assumption is correct, but probably not always.
In aldolase A, Cys7' and Cys338, which are close to each other (Kobashi and Horecker, 1967;Sygusch et al., 1987), are protected by substrate against chemical modification (Steinman and Richards, 1970); and it was believed that they are essential for catalysis (Lai et al., 1974). Later, based on chemical modification studies and site-directed mutagenesis Takahashi et al., 1989), these cysteines were shown to be nonessential. They also appear to be quite distant from the active site as evident from the crystallographic structure of the enzyme (Sygusch et al., 1987). Therefore, one may conclude that there is long-range communication between the active-site region and the microenvironment of CysT2 and Cys3".
The molecule of rabbit muscle aldolase consists of four identical or nearly identical subunits with an M , of 40,000 (Kawahara and Tanford, 1966;Penhoet et al., 1967;Susor et al., 1969). Each subunit of the enzyme contains four buried and four exposed sulfhydryl groups (Lai et al., 1974;Steinman and Richards, 1970;Anderson and Perham, 1970). One of the exposed thiol groups (CysZ3') exhibits high reactivity toward various thiol reagents; the remaining three groups react much more slowly (Steinman and Richards, 1970). The difference in the kinetics of the chemical modification of these cysteines is so large that it is easy to monitor their reactivity independently (Eagles et al., 1969;Anderson and Perham, 1970;Heyduk and Kochman, 1985).
Long-range effects and substrate-induced conformational flexibility in aldolase were probed in this paper by measuring the changes in the chemical reactivity of CysZ3' and CYS*~' in response to perturbation of protein by substrate binding, removal of the C-terminal tyrosine, and formation of the intrasubunit -S-Sbond between Cys7* and CYS~~'. The sites of perturbation of the enzyme molecule are at considerable distance from Cys239 and Cys'"; therefore, the changes in their reactivity in response to these perturbations can be interpreted as evidence for long-range effects. The obtained results show that aldolase is quite a flexible molecule, able to transmit the information about local perturbation to distant parts of the molecule. Substrate-induced perturbations are syncatalytic in nature. Substrates having only C-1 phosphate induce a different conformational change compared with those having both C-1 and C-6 phosphates. This may explain the dramatic difference in kinetic properties between fructose-1-P and fructose-1,6-P2.
Rabbit muscle aldolase was prepared by phosphocellulose chromatography (Penhoet et al., 1969); its specific activity was in the range of 15.5-17.0 pmol of Fru-l,6-P2 cleaved per min/mg of protein or 1.2-1.4 pmol of Fru-1-P cleaved per min/mg of protein.
Preparation of Aldolase A-C~S~~'-N~S-TO a cuvette containing aldolase A solution (31.3 p~) in 10 mM Tris, 1 mM EDTA, pH 7.5, containing 60 mM Pi, 50 pl of 4.8 mM Nbs2 were added. The progress of the reaction was monitored spectrophotometrically at 412 nm against the appropriate blank. After -5 min when -0.8 residue/ subunit was modified, the reaction mixture was applied to a Sephadex G-25 column (1.6 X 30 cm) equilibrated with 100 mM Tris, 1 mM EDTA, pH 7.5. The protein-containing fractions were collected. The number of Nbs residues bound to the enzyme was determined as described previously . The kinetics of the modification of cysteine residues in the aldolase A -c y~'~~-N b s derivative showed that only CysZ3' was modified because this preparation exhibited three slowly reacting -SH groups and -0.2 fast reacting groups per subunit (data not shown).
The obtained derivative exhibited 100-105% of the native enzyme activity and was stable for at least 24 h at 20 "C (100 mM Tris, 1 mM EDTA, pH 7.5).
Preparation of Aldolase A-CysZ3'-EDANS-l ml of aldolase solution (27 mg/ml) in 100 mM Tris, 1 mM EDTA, 2 mM hexitol-1,6-P2, pH 7.5, was modified by mixing with 1 ml of I-EDANS solution (0.3 mg/ml). The reaction was carried out in the dark for 3.5 h and was terminated by adding 10 pl of 1 M dithiothreitol. After 5 min of incubation, the reaction mixture was applied to a Sephadex G-25 column (1.6 X 30 cm) equilibrated with 100 mM Tris, 1 mM EDTA, pH 7.5; and protein-containing fractions were collected. The number of EDANS molecules introduced to the enzyme was determined spectrophotometrically (similarly as for the aldolase-CysZ3'-Nbs derivative) from the spectrum of the modified protein . Typically, 0.2 molecule of EDANS/subunit of aldolase was introduced. Kinetic experiments on the modification of thiol groups in the aldolase-Cys2"-EDANS derivative with Nbs2 showed that within the precision of this estimation, only fast-reacting CysZ3' residues were modified. The derivative obtained with 0.2 molecule of EDANS/subunit exhibited 107-115% of the initial activity, and no decrease of activity could be detected after 24 h of incubation at 20 "C.
Preparation of De~-Tyr~~-aldolase--Digestion of aldolase with carboxypeptidase A was performed at 25 "C in 50 mM Tris, pH 7.5, containing 1% LiCl with a carboxypeptidase A:aldolase ratio of 1:2000. Immediately after aldolase activity had dropped to -4% of the initial value, the reaction was stopped by adding one-twentieth volume of 0.1 M triethanolamine, 0.2 M EDTA, pH 7.5, The reaction mixture was applied to a Sephadex G-75 column (0.9 X 60 cm) equilibrated with 100 mM Tris, 1 mM EDTA, pH 7.5; and aldolasecontaining fractions (free of carboxypeptidase) were collected. The degraded aldolase had a specific activity of 0.7 unit/mg for Fru-1,6-P2 and 0.6 unit/mg for Fru-1-P. At the above conditions carboxypeptidase A releases 3.4 Tyr and 1.8 Ala residues from the aldolase tetramer and no histidine.When the degraded enzyme was intended for cross-linking, the aldolase sample was eluted from the Sephadex G-75 column with 50 mM Tris, pH 7.5.
Kinetic Measurements-A Specord "40 spectrophotometer was used for the investigation of the kinetics of Nbs release from the ald~lase-Cys~~'-Nbs derivative. The instrument was equipped with a kinetic data acquisition and analysis accessory. The reaction was started by addition of 50 pl of 20 mM glutathione to a cuvette containing 2 ml of the ald~lase-Cys~~~-Nbs derivative (2.4 PM), and the progress of the reaction was monitored at 412 nm. For each curve, -300 data points were usually collected, and pseudo first-order rate constants (k) were obtained by nonlinear regression analysis of the collected data. The reaction obeys pseudo first-order kinetics at least up to 95% of completion.
The kinetics of CysZ3' modification by Nbs2 was followed on a Durrum D-110 stopped-flow spectrophotometer interfaced with an IBM PC XT compatible computer for data acquisition and analysis. For each curve, 500 data points were collected, and pseudo first-order rate constants for CysZ3' modification were obtained as described previously (Heyduk and Kochman, 1985). Nbs2 was present at 2 mM, and experiments were performed in 100 mM Tris, 1 mM EDTA, pH 7.5, at 20 "C.
Control experiments on the modification of thiol groups of bovine serum albumin in the presence or absence of the ligands studied were performed as described for aldolase.
Fluorescence Quenching-All fluorescence experiments were performed in 100 mM Tris, 1 mM EDTA, pH 7.5, at 20 "C using a Perkin-Elmer Cetus MPF-44 spectrofluorometer. To a cuvette containing 2.5 ml of the ald~lase-Cys~~'-EDANS derivative (1.08 pM) with or without ligand were added 25-pl portions of 8 M acrylamide (or 6 M KI with 0.1 mM Na,SOS). The contents of the cuvette were mixed with a glass rod, and fluorescence intensity at 490 nm (Aex = 340 nm) was measured.
Binding of Active-site Ligands to Aldolase-Binding of ligands to aldolase was studied by measuring fluorescence quenching of ANS bound to aldolase (Kasprzak and Kochman, 1980). All measurements were carried out in 100 mM Tris, 1 mM EDTA, pH 7.5. Fluorescence titrations were performed according to the following procedure. To 2.5 ml of aldolase solution were added 100 pl of ANS solution. The solution was thermostatted at 25 "C; and after 15 min, 10-2O-pl aliquots of the appropriate ligand solution were added. The content of the cuvette was gently stirred and after 1 min, the fluorescence intensity at 485 nm was measured (Aex = 350 nm). For the determination of dissociation constants, the enzyme concentrations were kept much lower than Kd, whereas the opposite conditions were used when the number of binding sites was estimated. ANS concentrations were in the range of 5-40 p~ depending on aldolase concentration. For each experiment, -15 data points were obtained. Fluorescence changes were corrected for sample dilution and for fluorescence of unbound ANS.
Parameters of ligand binding to aldolase were obtained by nonlinear least-squares fitting as described elsewhere (Kasprzak and Kochman, 1980). The effect of ANS concentration on the observed dissociation constants was also taken into account using the equation: are dissociation constants for ligand measured in the presence and absence of ANS, respectively; [ANSI is the label concentration in a particular experiment; and &(ANS) is the dissociation constant for the aldolase-ANS complex (Kasprzak, 1980). Analytical Procedures-The following absorption coefficients were used for the spectrophotometric determination of concentration: Nbs, 13,600 M" cm" (Ellman, 1959); I-EDANS, 6100 M" cm" (Hudson and Weber, 1973); and aldolase A; 0.91 mg" cm" ml (Baranowski and Niederland, 1949). The concentration of Nbs2 was determined from the absorbance at 412 nm after reduction of Nbs2 with an excess of 2-mercaptoethanol. Concentrations of DHAP, Fru-1-P, and Fru-1,6-P2 were determined enzymatically (Kasprzak, 1980).  Table I. Although the presented results for demodification are for experiments performed with glutathione, the same results (R values) were obtained with different thiols (2-mercaptoethanol, cysteine, dithiothreitol) and also when CysZ3' was modified with 2,2'- Control experiments showed that the ligands used in this study did not change the reactivity of thiol groups of bovine serum albumin (data not shown).
The results obtained for modification and demodification are almost identical. The exception is Pi binding, where the R value for demodification is clearly lower. The pattern of the changes in Cys2"' reactivity seems to be a function of the distribution of phosphate groups in the ligand molecule. Active-site ligands possessing a phosphate moiety bound only to the C-1 phosphate-binding site on the enzyme increase the R value, whereas those having only a C-6 phosphate or both C-1 and C-6 phosphates decreased the R value or caused no change at all. There seems to be an apparent discrepancy between the effects of Fru-1,6-P2 and hexitol-1,6-Pz since the former results in an R value of essentially 1, whereas the latter decreases the R value. This can be explained if one remembers that mixing aldolase with Fru-l,6-Pz results in an equilibrium mixture of Fru-l,6-P2 andphosphotrioses (DHAP and glyceraldehyde 3-phosphate). The R value in the presence of Fru-1,6-P2 is therefore averaged over all the different complexes present in solution. To illustrate this, we have taken advantage of the high dependence of the percentage of Fru-1,6-Pn in the equilibrium mixture on the starting concentration of this substrate (Fig. 1). When the Fru-1,6-P2 concentration is low, the products of the reaction (DHAP and glyceraldehyde-3-P) prevail, and the aldolase-DHAP complex should be a major species among different enzyme-substrate complexes. Therefore, the R value is >1 (Fig. 1). When the starting concentration of Fru-1,6-Pz is high (and therefore percent of this substrate in the equilibrium mixture), the R value decreases to a value of 4 , as found for hexitol-1,6-P2. So, we conclude that Fru-l,6-P2 has an effect on the reactivity of CysZ3' similar to hexitol-1,6-P,.
The changes of the reactivity of CysZ3' observed upon ligand binding (measured by both the rate of modification and demodification of this residue) display saturation kinetics when studied as a function of ligand concentration. This is illustrated in Fig. 2 for DHAP. The dissociation constant for DHAP calculated from these data by nonlinear regression for the equation describing binding of ligand to four independent sites is 4.9 p~. This is in agreement with the values obtained by different methods (Grazi and Trombetta, 1974). Therefore, it can be concluded that the measured changes in thiol reactivities are the result of formation of the specific enzymeligand complex.
The reactivity of the second exposed thiol group in aldolase, Cys2*', was also measured. In contrast to the results for CysZ3', Z of fru-l,6-P, in equilibrium mixture active-site ligands did not change the reactivity of CysZn9 (data not shown). These results indicate that the region of the aldolase molecule comprising the environment of CysZn9 is not affected by ligand binding. Fluorescence Quenching of Ald~lase-Cys~~~-EDANS Deriuatiue-The fluorescence of the ald~lase-Cys~~~-EDANS derivative (both intensity and Amax) is not changed significantly upon binding of active-site ligands (data not shown). However, changes in the microenvironment of CysZ3' can be seen if fluorescence quenching of aldolase-CysZ3'-EDANS is studied.
The quenching curves for both acrylamide and KI are nonlinear (Fig. 3, A and B ) . A curvature of this type in Stern-Volmer plots can in general be caused either by a heterogenous population of fluorophores or by poor quenching efficiency of a quencher (Eftink and Ghiron, 1981). This latter possibility should be discarded since the plots for the cysteine-EDANS Acrylamide (M) derivative are linear both for acrylamide and KI (Fig. 3, A  and B ) . The heterogeneity of the population of fluorophores can originate either from the heterogeneity in the chemical modification of the protein with a fluorescence probe or from multiple conformations of protein in solution. Again, this latter possibility should be discarded since the plots remain curved in 6 M guanidine HCI, which unfolds the protein. The simplest scheme that gave satisfactory fit for all curves (Fig. 3, A and B ) was the one with n = 2 and with no static quenching (V, = 0). For all curves, the distribution of fluorophores was 80-90% of one kind and 10-20% of the other kind. It is reasonable to ascribe the fluorophore that is prevailing to aldolase-Cys'"-EDANS since modification of the protein with 1,5-I-EDANS correlates well with the disappearance of fast-reacting cysteine residues (see "Experimental Procedures"). The location of the remaining 10-20% of the fluorophore is unknown. Fitting the quenching data gives fi, fi, and K1 and K2 values. However, their physical meaning remains questionable since they appear to be highly correlated as judged by correlation matrix analysis (Johnson and Frazier, 1985). Therefore, to compare the accessibility of the fluoro-phores to quenching, the apparent quenching constants were used that were calculated as weight averages of fitted parameters, i.e. Kapp = fl*Kl + f2*K2 (Table 11). Since aldolase-CysZ3'-EDANS is a dominant fraction of the fluorophores, K,,, mostly reflects the accessibility of this fluorophore.
The results obtained show that in the absence of substrates, CysZ3'-EDANS is somewhat shielded from the solvent since the Kapp value is less than the one for cysteine-EDANS. The addition of hexitol-1,6-P2 or DHAP increases the accessibility of this fluorophore (K,,, increases). In the case of KI quenching, CysZ3'-EDANS in the absence of ligands is more quenched than cysteine-EDANS. This shows that there is some favorable distribution of positive charges around the fluorophore in Cys2"-EDANS that makes negatively charged I-so effective. In the case of KI, contrary to acrylamide quenching, addition of substrates dramatically decreases Kapp values.
In the presence of 6 M guandidine HCl, all these effects of ligands disappear, i.e. quenching curves with or without ligands are the same (Fig. 3, A and B ) .
Effect of Cross-linking of Aldolase on Reactivity of Thiol Groups-Cross-linking of aldolase by the intrasubunit -S-Sbond between Cys7' and Cys338 results in the increase of reactivity of the fast-reacting CysZ3' residue and of the slowly reacting CysY8' residue (Table 111). No effect of hexitol-1,6-P2 on the reactivity of CysZ3' or CysZR9 was observed. In this aldolase derivative, fluorescence titration showed that crosslinked aldolase is unable to bind DHAP, hexitol-1,6-P2, ATP, or Pi (Table IV). Cross-linked aldolase after reduction with 0.5 mM dithiothreitol, 100 mM Tris, 1 mM EDTA, pH 7.5, for 160 min recovers 97% of the native enzyme activity and the ability of binding active-site ligands, indicating that crosslinking does not cause irreversible change in aldolase structure.    The Reactivity of CYS~~' and CysZ8' in Des-Ty?'j3-aldolase-The de~-Tyr~~'"aldolase derivative has about four binding sites for hexitol-1,6-P2 and DHAP and exhibits decreased affinities for these ligands as compared to the native enzyme ( Table IV). As shown in Table 111, removal of Tyr3'j3 results in a large increase in the reactivity of CysZ3' and a substantial increase in the reactivity of CysZ8'. Because the change in the reactivity of CysZ3' in de~-Tyr~~'-aldolase was almost identical to that observed in the cross-linked enzyme, the reactivity of cysteine residues was investigated in cross-linked des-Tyr3'j3aldolase. As shown in Table 111, cross-linking of d e~-T y r~~'aldolase via the Cys7' and Cys338 disulfide bridge results in a further increase in the reactivity of both CysZ3' and CysZS9 residues. This indicates that conformation changes induced by removal of the C-terminal Tyr3'j3 residue are distinct from those induced by cross-linking of aldolase.

Syncatalytic Nature of Substrate-induced Conformational
Changes in Aldolase A-The reactivity of thiol groups was successfully used in the detection of temperature-induced conformational transition in the aldolase molecule (Heyduk and Kochman, 1985). In this study, this technique was applied to the investigation of conformational changes upon binding of active-site ligands. Conformational changes induced by ligand binding have already been proposed for aldolase. These suggestions were based on experiments with proteolysis in the presence of substrates (Adelman et al., 1968) and on differences in UV spectrum after binding of arabinitol1,5-bisphosphate, a strong inhibitor of aldolase (Crowder et al., 1973). However, there are interesting features of these changes that become apparent from this work.
All reactive cysteine residues of aldolase are localized outside the active site of the enzyme (Sygusch et al., 1987) (also see Fig. 4). Conformational changes induced by active-site ligand binding can be detected at CysZ3' (by cysteine reactivity and fluorescence measurements) and at Cys72-Cys338 (by cysteine reactivity (Steinman and Richards, 1970)) and are undetectable at Cys2". This shows that perturbation of the enzyme conformation by substrate binding is transmitted from the active site to remote parts of a molecule (see Fig. 4). It also seems that it is not a global rearrangement of the protein structure since some regions (e.g. Cys2") are unaffected by substrate binding. CysZ3' is positioned close to the intersubunit contact region (Sygusch et al., 1987) (Fig. 4), indicating that the signal of substrate binding can be transmitted to another subunit (it does not necessarily mean that it is transmitted to the second active site). Aldolase A is not an allosteric enzyme and displays pure Michaelis-Menten kinetics for substrate binding (Penhoet et al., 1969;Horecker FIG. 4. Schematic localization of aldolase reactive cysteine residues with respect to active-site residues (Lysl", L y~l~~, LysZz9) within enzyme subunit. The scheme was drawn based on Fig. 1 from Sygusch et al. (1987). The solid bar depicts the region involved in intersubunit contacts. et al., 1972). Other active-site ligands also seem to bind to the enzyme in a noncooperative mode (Kasprzak and Kochman, 1980;Palczewski et al., 1983Palczewski et al., , 1985Palczewski and Kochman, 1987). On the other hand, binding of inositol polyphosphates to aldolase A is sigmoidal (Koppitz et al., 1986). Hybridization studies between aldolase A4 and C4 show that antibody binding to C-type subunits can inhibit enzyme activity by as much as 60% in A3C tetramer (a hybrid composed of three subunits of aldolase A and one of aldolase C) (Penhoet and Rutter, 1971). The removal of C-terminal tyrosine residues from lobster or snail muscle aldolase with carboxypeptidase is much slower than the loss of activity of these enzymes (Guha et al., 1971;Buczylko et al., 1980). All of the above can be interpreted as the existence of cooperative interactions between aldolase subunits.
The most interesting feature of these changes is that the enzyme response seems to be differential with respect to the substrate used. Ligands with a phosphate group at C-1 only induce the conformation of the enzyme with increased accessibility of the fast-reacting CysZ3' residue, whereas ligands with C-6 phosphate or both C-1 and C-6 phosphates induce the conformation of aldolase with decreased reactivity of this group. The mechanism of the aldolase reaction could not be unequivocally determined (Lai et al., 1974;Periana et al., 1980). One of the interesting features of the aldolase reaction is the pronounced difference in the kinetic properties observed for Fru-1,6-P2 and Fru-1-P (Penhoet et al., 1967). On the basis of published mechanisms of the reaction, it is difficult to explain this difference (especially the -10 times higher value of V, , , for Fru-1,6-Pz in comparison to Fru-1-P) (Buczylko et al., 1980).
Our results offer a simple explanation for this phenomenon.
We propose that the chemical mechanism of fructose-1,6-P2 and fructose-1-P cleavage is the same and that the enzyme discriminates between these two substrates by adopting a different conformation upon the formation of an enzymesubstrate complex, resulting in different kcat values for these substrates. In this sense, substrate-induced conformational changes in aldolase are syncatalytic in nature (Birchmeier and Christen, 1977). The old concept of "induced-fit" (Koshland, 1958) would be, in this case, used by the enzyme to control substrate specificity.
Long-range Effects Induced by Intrasubunit Cross-linking and C-terminal Removal-Cross-linking of the aldolase subunit with the Cys72-Cys338 bridge results in the significant perturbation of enzyme conformation that is propagated to CysZ3' and Cys289. Therefore, this is a good example of longrange effects in aldolase. This modification also results in the complete loss of enzyme activity and a loss of ability to bind active-site ligands. This is quite remarkable since Cys7' and Cys338 are not in the active site of aldolase Takahashi et al., 1989;Sygusch et al., 1987) (Fig. 4). It can then be concluded that cross-linking of the enzyme results in a very dramatic rearrangement of the activesite region and also that binding of active-site ligands results in perturbation of the Cys7* and Cys338 region since they become unaccessible to chemical modification. Therefore, these two regions of the molecule are tightly coupled. Interestingly, it has been shown, in the case of serine proteases, that there is a well-defined region of a protein outside the active site that is coupled to the active site and is important in determining the specificity of the enzyme (Liebman, 1986). Very similar changes in the enzyme are observed upon Tyr3'j3 removal, although the enzyme retains some activity and ability to bind active-site ligands. This removal of a single amino acid is also "felt" by the enzyme at CysZ3' and C~S~*~.
The aldolase tetramers are very stable because, in opposition to glyceraldehyde-3-phosphate dehydrogenase (Kochman et al., 1974) or lactate dehydrogenase (Miller et al., 1971), aldolase tetramers stay intact within a rather wide range of ionic strength. The exchange of aldolase subunits can be achieved at extreme pH values (below pH 3 or above pH 10) or in >1.2 M solutions of MgClz (Stellwagen and Schachman, 1962;Deal et al., 1963;Westhead et al., 1963;Rudolph et al., 1976;Kent and Lebherz, 1984). On the other hand, within tetramers, the aldolase molecule exhibits substantial flexibility that is reflected in the reversible temperature-induced conformational transition (at 26 "C) observed at Cy$', Cys"', CysZa9, Cys338, and Trp147 and at the active-site region (Heyduk and Kochman, 1985). The portion of the polypeptide chain including 18 C-terminal residues is invisible in x-ray studies of crystalline aldolase (Sygusch et al., 1987), indicating its static and/or dynamic disorder. Aldolase can crystallize in many different crystal forms what can reflect this freedom in macroscopic scale (Eagles et al., 1969;Heidner et al., 1971;Miller et al., 1981;Sawyer, 1972;Goryunov et al., 1969;Brenner-Holzach and Smit, 1982).
Tyr363, which was located by chemical modification studies close to Lys107, which binds the C-6 phosphate, is in this flexible peptide. Also, the C y~~~--C y s~~" pair is close to a hinge region connecting the H2 helix with this flexible peptide. Therefore, the formation of a Cys72-Cys338 bridge is likely to affect the flexibility of this peptide.
It seems that freedom of movement of the C-terminal portion of the polypeptide chain in the aldolase molecule is a vital element of enzyme catalysis and is required for the correct transition state structure of the enzyme-substrate complex.