The temperature-dependences of the second-order rate constants (k) of the reactions of the catalytic site thiol groups of two cysteine peptidases papain (EC 22.214.171.124) and actinidin (EC 126.96.36.199) with a series of seven 2-pyridyl disulphide reactivity probes (R-S-S-2-Py, in which R provides variation in recognition features) were determined at pH 6.7 at temperatures in the range 4–30 °C by stopped-flow methodology and were used to calculate values of ΔS‡, ΔH‡ and ΔG‡. The marked changes in ΔS‡ from negative to positive in the papain reactions consequent on provision of increase in the opportunities for key non-covalent recognition interactions may implicate microsite desolvation in binding site–catalytic site signalling to provide a catalytically relevant transition state. The substantially different behaviour of actinidin including apparent masking of changes in ΔH‡ by an endothermic conformational change suggests a difference in mechanism involving kinetically significant conformational change.
- activation parameter
- conformational change
- microsite desolvation
The least well understood aspects of enzyme catalysis and active-centre chemistry include dynamic aspects of non-covalent interactions that are involved in substrate and inhibitor recognition and their effects on catalytic site reactivity characteristics. Members of the papain family of cysteine peptidases have proved to be useful in contributing to the detection and understanding of these phenomena (see below).
Cysteine peptidases are endopeptidases whose catalytic activity relies on the thiol group of a cysteine residue (for a comprehensive review, see , and for more recent reviews on their catalytic mechanism and active-centre chemistry, see [2,3]). These enzymes are widely distributed in animals, plants, bacteria and viruses, and constitute a superfamily composed of six enzyme families . Most of the well-characterized and intensively studied cysteine peptidases are members of the papain family. Those that have been subjected to detailed structural and mechanistic study contain a Cys-S−/His-Im+H (imidazolium) ion pair as the major feature of the catalytic site. The thiolate anion component of this interactive system becomes transiently acylated during catalysis, assisted by general acid-catalysed expulsion of the leaving group from a tetrahedral species, the acid catalysis being provided by the Im+H component.
Papain (EC 188.8.131.52) was long considered to be the cysteine peptidase archetype, and mechanistic studies were concentrated largely on this member of the family for many years. Studies that included the use of natural variants of papain, however, led to the discovery of considerable variation in catalytic site reactivity characteristics within the family. These differences demonstrate that the traditional view of the archetypal nature of papain is correct only for low-resolution aspects of structure–function relationships such as the central mechanistic roles of the common Cys25-S−/His159-Im+H ion pair (papain numbering) and that much is to be gained by study of natural variants. Particularly noteworthy variation exists in (i) the discrepancy between the pKa values of -S−/-Im+H ion-pair formation and the essentially common pKa value for the development of catalytic activity (kcat/Km) leading to the important concept of an essential electrostatic modulator, (ii) the coupling of key non-covalent molecular recognition interactions with catalytic site chemistry and their interplay with electrostatic effects, and (iii) both postacylation– and free enzyme–protein dynamics. As a result of these differences, gradations in functional characteristics and mechanistic phenomena that are more easily revealed in some members of the family than in others were identified. Papain exhibits characteristics at one end of a spectrum of chemical behaviour with actinidin (EC 184.108.40.206) at the other. Caricain [papaya (Carica papaya) proteinase Ω, EC 220.127.116.11] and ficin (EC 18.104.22.168) each exhibit behaviour intermediate between these extremes.
Many of the advances indicated above were obtained by using substrate-derived 2-pyridyl disulphide time-dependent inhibitors as two-protonic state reactivity probes  and pH-dependent kinetics . In the present study, the temperature-dependences of the kinetics of selected cysteine peptidase–probe reactions were determined to ascertain whether changes in transition geometry occasioned by some of the specific non-covalent binding interactions and identified by the use of our probes [2,3] might be accompanied by selective active-centre desolvation. The crystal structures of papain  and actinidin  each reveal an ordered network of water molecules in their active centres and some binding interactions might be predicted to interrupt these networks. The present paper reports the activation parameters (ΔS‡ and ΔH‡) for the reactions of papain and actinidin with a series of 2-pyridyl disulphide reactivity probes in which individual specific non-covalent binding sites are introduced in various combinations into the probe molecule. The striking changes in ΔS‡ from large and negative to large and positive in the reactions of papain, which are mirrored only partially in the reactions of actinidin, and the enthalpy–entropy relationships for the reactions of both enzymes are discussed and interpreted in terms of selective active-centre desolvation and, in the case of actinidin, masking of decrease in ΔH‡ by an endothermic conformational change, possibly related to those deduced by using a specific thionoester substrate.
MATERIALS AND METHODS
Enzymes and reactivity probes
Methods of purification of actinidin  and papain , which include the production of fully active enzymes by covalent chromatography (reviewed in ), have been described previously, as have their active-centre content and evaluation by spectroscopic titration at 343 nm (ϵ343=8080 M−1·cm−1) using 2,2′-dipyridyl disulphide as titrant .
The syntheses and characterization of the seven reactivity probes used in the present work (shown in Figure 1) have been described previously in connection with pH-dependence studies: ethyl 2-pyridyl disulphide (1) ; 2-(acetoxy)ethylene 2′-pyridyl disulphide (2) ; 2-(acetamido)ethylene 2′-pyridyl disulphide (3) ; 2-(N′-acetyl-L-phenylalanyl)hydroxyethylene 2′-pyridyl disulphide (4)  and its D-enantiomer (5) ; 2-(N′-acetyl-L-phenylalanylamino)ethylene 2′-pyridyl disulphide (6)  and its D-enantiomer (7) .
SF (stopped-flow) kinetics
Kinetic studies on the reactions of the catalytic site thiol groups of papain and actinidin with the reactivity probes were performed by using a Hi-Tech Scientific SF spectrophotometer, kinetics workstation, and data acquisition and analysis software. Monochromator entrance and exit slit widths were set at 0.5 mm. The sample handling unit was fitted with a UG5 bandpass filter to eliminate stray light and was configured for thermostatically controlled temperature cycling. Temperature control was achieved using a Grant LTD6 water bath.
Reactions were carried out in the low-temperature coefficient (dpKa/dT=0.0028 ) KH2PO4/NaOH buffer, pH 6.7, I 0.1 M, at ten different temperatures in the range 4–30 °C under (pseudo) first-order conditions with [probe]≥20 times [enzyme thiol]. The stock solution of the phosphate buffer was prepared at I 0.3 M and was diluted to provide a solution containing the reactivity probe at I 0.1 M in one of the syringes of the SF spectrophotometer. The other syringe contained the enzyme in 0.1 M KCl. Mixing of equal volumes of these two solutions maintained I 0.1 M in the reaction chamber. The pH was checked to be 6.7 in the effluent from the reaction chamber and in dummy reaction mixtures. The release of pyridine-2-thione was monitored at 343 nm. Values of the observed first-order rate constants (kobs) were obtained by fitting the equation for a single exponential process, where P1=A∞−A0, P2=kobs and P3=A∞, to the absorbance (A)−time (t) data collected as multiple superimposable traces by the SF system.
Determination of activation parameters
Values of entropy of activation (ΔS‡) and enthalpy of activation (ΔH‡) were determined by linear regression of the left-hand side of eqn (1) on 1000/T, where T is the absolute temperature and 1000 is a convenient scaling factor: (1) Eqn (1) is a convenient transform of the Eyring equation (see the Results and Discussion section) in which R is the gas constant (8.13145 J·mol−1·K−1), kB is Boltzmann's constant, h is Planck's constant and the value of ln(kB/h) is 23.76. The regression provides −ΔH‡ as the slope of the linear plot and ΔS‡ as the ordinate intercept.
RESULTS AND DISCUSSION
Temperature-dependent kinetics: aspects of analysis
The well-known Arrhenius and Eyring equations used in chemical kinetics can be properly applied to reactions of enzymes and sensibly interpreted providing that: (i) a restricted temperature range that avoids denaturation (often taken to be 0–50 °C) is used, (ii) account is taken of the temperature-dependence of acid dissociation constants, and (iii) the mechanistic significance of the rate parameter is known and lacking in complexity. In the present study, it was possible to: (i) restrict the temperature range to 4–30 °C, where the enzymes are conformationally stable without detriment to the analysis, (iii) study the reactions at pH 6.7, a value around which the rate constants change only slightly with change in pH in the range ∼6–8, and (iii) determine the second-order rate constant (k) for the enzyme–probe reactions whose mechanistic significance (reaction of the catalytic site cysteinyl thiol group with the neutral probe molecule) is well defined and is the same for each probe.
For many reactions, the increase in k with increase in temperature is described by the Arrhenius equation (eqn 2) where Ea, the activation energy, in terms of transition state theory, is equal to ΔH‡+RT. (2) Transition state theory leads to the Eyring equation (eqn 3, in which ΔG‡ is the free energy of activation) or eqn (4) and division throughout of eqn (4) by T leads to the logarithmic form, eqn (5). (3) The form of the Eyring equation (eqn 1, see the Materials and methods section) convenient for analysis is obtained from eqn (5) by multiplying through by and collecting terms in R. Regression analysis provides values of ΔS‡ and ΔH‡ directly.
Despite the well-known long extrapolation to 0 K required to obtain the value of ΔS‡ and the relatively large standard errors on this parameter, conclusions drawn from temperature-dependence studies that, as in the present study, rely on the sign of ΔS‡ (positive or negative) and not only on its magnitude are usually not affected by the error inherent in the extrapolation process (see, for example, ).
Dependence on non-covalent interactions of ΔS‡ and ΔH‡ for the reactions of papain and actinidin with the 2-pyridyl disulphide reactivity probes 1–7 at pH 6.7
The values of ΔS‡ and ΔH‡ for the two sets of enzyme–probe reactions, together with those of ΔG‡ at 298 K calculated as ΔH‡−TΔS‡, are collected in Table 1, and a typical linear plot derived from eqn (1) (for the reaction of papain with the substrate-analogue probe 6) is shown in Figure 2. The variations, particularly in ΔS‡, for the reactions of papain are considered first and then compared with those for the reactions of actinidin.
Reactions of papain
For the reactions of papain with the ‘featureless’ probe, CH3-CH2-S-S-2-Py (where 2-Py is 2-pyridyl) (1), ΔS‡ has a large negative value (−79.9±4.0 J·mol−1·K−1), as expected, because of the restriction of the motion of the probe in the transition state, and the probable minimal disruption of the water network in the active centre of the enzyme owing to the lack of recognition features in the small alkyl group of 1. When provision is made for the possibility first of one hydrogen bond [between the backbone N-H of Gly66 and the carbonyl O of CH3CO-O-[CH2]2-S-S-2-Py (2)] and then also of a second hydrogen bond [between the backbone carbonyl O of Asp158 and the N-H of the amide group of CH3CO-NH-[CH2]2-S-S-2-Py (3)], the values of ΔS‡ become progressively less negative (from (−79.9±4.0 to −60.5±4.0 to −37.1±2.0 J·mol−1·K−1). In these three cases, the negative values of ΔS‡ are characteristic of transition states that are more ordered than the corresponding ground states. The progressive decrease in their negativity could be accounted for by the increased probability of an appropriate transition-state geometry. The changes in transition-state geometry consequent on provision of various recognition interactions in these reactions may be inferred from marked changes in shapes of pH–k profiles [16,17].
The most striking changes in ΔS‡ arise when the probe is equipped with phenylalanine at P2 as a potential occupant for the major recognition site for papain, the hydrophobic S2 subsite. For the reaction of CH3CO-[L-Phe]-NH-[CH2]2-S-S-2-Py (6), the probe containing both a P1-P2 amide bond and L-phenylalanine at P2, which has a pH–k profile with a shape strikingly similar to that of a pH–kcat/Km profile, i.e. maximal activity around pH 6, ΔS‡ has become large and positive (+57.7±7.0 J·mol−1·K−1). The effectiveness of this reaction at pH values around 6 suggest a high probability of a well-organized transition state in which the thiolate component of the catalytic site ion pair and the Im+H component are correctly positioned to act as nucleophile and hydrogen bond donor respectively. The fact that this reaction is characterized also by a large positive entropy of activation, however, suggests a third component which provides for a less-ordered structure in the transition state than in the ground state. The most obvious candidate for this additional component is the water network of the active centre. This leads to the view that desolvation resulting from binding of the substrate-like probe 6 in the transition state conformation with consequent release of bound water molecules allows both components of the ion pair to be involved synchronously in the reaction, and is most likely to be the cause of the large positive value of ΔS‡.
The effect on ΔS‡ of removing the possibility of hydrogen bonding of the probe to the backbone carbonyl O of Asp158 while retaining L-phenylalanine at P2 was investigated by comparison of the value for the reaction of CH3CO-[L-Phe]-O-[CH2]2-S-S-2-Py (4) with that for the reaction of probe 6. Removal of the possibility of the (P1)-NH···O=C<(Asp158) hydrogen bond results in a substantial decrease in the value of ΔS‡ from +57.7±7.0 to +18.7±4.1 J·mol−1·K−1. The fact that ΔS‡ for the reaction of probe 5, the D-enantiomer of 4 (−19.0±4.4 J·mol−1·K−1) is substantially less negative than that for the reaction of probe 2 which lacks the D-phenylalanine moiety (−60.5±4.0 J·mol−1·K−1) provides evidence for the binding of (P2)-D-phenylalanine to papain (see  for a discussion of this phenomenon deduced by model building). It is noteworthy that ΔS‡ for the reaction of probe 5 is still negative and becomes positive for the D-enantiomer only when the possibility of hydrogen bonding to Asp158 is restored in probe 7. This supports the conclusion suggested by the relative values of difference kinetic specificity energies , that transition state geometry in reactions of papain is determined by interdependent binding interactions in the S2-subsite, S1-S2 intersubsite region and catalytic site. The present study also suggests that selective desolvation of the regions of the active centre participates in the binding site–catalytic site coupling process.
Reactions of actinidin
ΔS‡ for the reactions of the featureless probe (1) is even more negative (−112.1±6.9 J·mol−1·K−1) than for the analogous reaction of papain (−79.9±4.0 J·mol−1·K−1), possibly indicating even more restriction of the motion of the probe in the transition state. As in the case of the papain reactions, incorporation of the P1-P2 hydrogen bonding systems in the P1-P2 ester probe (2) and the P1-P2 amide probe (3) results in the progressive decrease in the negativity of ΔS‡ from −112.1±6.9 to −42.2±3.8 to −34.5±6.3 J·mol−1·K−1, presumably reflecting the increased probability of binding with appropriate orientation in the transition state. Although the probability of an appropriate transition-state geometry is increased by the opportunity to form the additional trans-cleft hydrogen bond, the mechanism of the reaction of actinidin is different from that of papain. Thus, whereas for papain there is substantial reaction involving activation of the leaving group by hydrogen bond donation by the Im+H component of the catalytic site ion pair, the reaction of actinidin is mainly with the formally protonated probe (pKa ∼2.5). This conclusion arises from the observation of maximal activity at pH 6 for the papain reaction, but at pH 3–4 for the actinidin reaction (see [16,17]).
A particularly marked difference between the reactions of actinidin and papain occurs when the hydrogen bond acceptor (the ester carbonyl O in probe 2) is augmented by (P2)-L-phenylalanine in probe 4. In the papain reaction, this addition results in ΔS‡ changing from markedly negative (−60.5±4.0 J·mol−1·K−1) to substantially positive (+18.7±4.1 J·mol−1·K−1). In the actinidin reaction, however, it remains substantially negative (−21.4±8.0 J·mol−1·K−1), although less negative than the value for the amide probe 3 devoid of the L-phenylalanine moiety. This suggests that, although binding of the L-phenylalanine moiety contributes to a correct transition-state geometry for nucleophilic attack of the thiolate anion of actinidin, desolvation is not as effective as in the papain reaction. Another striking difference between the reactions of the two enzymes is that changing the chirality of the phenylalanine moiety from L-phenylalanine in 4 to D-phenylalanine in 5 results in a marked change in ΔS‡ from +18.7±4.1 J·mol−1·K−1 to −19.0±4.4 J·mol−1·K−1 for the papain reaction, but essentially no change for the actinidin reaction (both approx. −21 J·mol−1·K−1). This suggests that binding either of the enantiomeric phenylalanine moieties to actinidin does not achieve a good fit of probes 4 and 5 in the hydrophobic pocket of the S2 subsite. This is in accord with the prediction from the crystal structure of actinidin  which reveals a shorter pocket than that in papain . A significantly better fit is achieved for the L-phenylalanine enantiomer; however, when provision is made also for the (P1)>N-H···O=C<(Asp161) hydrogen bond by changing the P1-P2 ester bond in 4 to the P1-P2 amide bond in 6. Reaction of actinidin with 6 is characterized by a positive ΔS‡ (+12.1±1.2 J·mol−1·K−1), indicating more effective desolvation, although less effective than in the papain reaction (ΔS‡=+57.7±7.0 J·mol−1·K−1). The change in structure from the L-phenylalanine enantiomer 6 to the D-enantiomer 7 which results in the decrease in ΔS‡ from +57.7±7.0 to +11.0±5.5 J·mol−1·K−1 for the papain reaction, results in a similar change for the actinidin reaction, in this case from +12.1±1.2 to −18.2±3.3 J·mol−1·K−1. In both cases, therefore, the binding mode of the D-phenylalanine moiety differs from that of the L-phenylalanine moiety with respect to the (P2)-binding pocket. This contrasts with the lack of dependence on P2 chirality for actinidin when the possibility of the (P1)>N-H···O=C<(Asp158) hydrogen bond is missing (in probes 4 and 5). This emphasizes the interdependence of P2/S2 subsite and P1-P2/S1-S2 intersubsite binding.
Differences in the relative changes in ΔH‡ and ΔS‡ that characterize reactions of papain and actinidin with the substrate-derived reactivity probes 2–7
Linear relationships between enthalpies and entropies occur widely in molecular associations particularly those involving hydrogen bonding (see, for example, [20–22]). Whether such relationships arise at least in part from chemical causation (extrathermodynamic relationships, isokinetic effect, enthalpy–entropy compensation effect) as against a dominant statistical compensation pattern has been the subject of much debate (see, for example,  and references therein). The analysis of enthalpy–entropy compensation reported by Cornish-Bowden  necessitates extreme caution in discussion of this phenomenon. In the present paper, we report the differences observed in the temperature-dependent kinetic characteristics of some analogous reactions of papain and actinidin that are influenced by non-covalent interactions and point out possible correlations with our results from other types of kinetic study reported previously.
Linear dependence of ΔH‡ on ΔS‡ was observed for the reactions of papain with the substrate-derived probes 2–7 (plot not shown; data in Table 1). Evidence for chemical causation suggested by Krug et al. [24,25] includes a substantial difference between the compensation temperature, Tc (the approximate value of which is provided by the slope of the ΔH‡ against ΔS‡ plot) and the average experimental temperature, and a linear relationship between ΔH‡ and ΔG‡. These two criteria appear to be met in the present work (Tc=200 K, average temperature 295 K; linear plot not shown, results in Tables 1 and 2).
Such chemical causation as may exist in enthalpy–entropy compensation is generally considered to be at least partly due to the perturbation, release or change in the state of water consequent upon ligand binding by the protein. This may reflect the responsiveness of proteins to ligand binding , and the compensation effect for the reactions of papain correlates well with the responsiveness of papain to changes in the structure of the substrate-derived probes 2–6 detected by the changes in the pH–k profiles [16,17]. The much lower sensitivity of the compensation effect for the corresponding reactions of actinidin (Table 1) and the clustering of the ΔG‡/ΔH‡ data (Table 2) contrast markedly with the results for papain and correlate with the relative insensitivity of the pH–k profiles for the actinidin reactions to respond to structural change in the probes. The absence of substantial differences in ΔH‡ for the reactions of the substrate-derived probes with actinidin could result from the masking of an enzyme-mediated decrease in enthalpy of activation by an endothermic conformational change as suggested by Jencks  for the deacylation of a series of acyl α-chymotrypsins. Such an interpretation of the temperature-dependent kinetic data for the reactions of actinidin might be related to the kinetically significant change(s) in conformation detected uniquely for actinidin by using a specific thionoester substrate [28,29].
We thank the Engineering and Physical Sciences Research Council and the Biotechnology and Biological Sciences Research Council for support of this work, including postdoctoral research assistantships for S.G. and G.W.M. respectively.
Abbreviations: Im+H, imidazolium; 2-Py, 2-pyridyl; SF, stopped-flow
- The Biochemical Society, London