Differential Scanning Calorimetry of APO-, Apophosphoryl, and Metalloalkaline Phosphatases*

Differential scanning calorimetry has been applied to assess the nature of the thermally induced denaturation of the metal-free apo and apophosphoryl derivatives of the dimeric ZnY+ metalloenzyme alkaline phosphatase (Esche- richia COW. The transitions of both species deviate markedly from the two-state model, giving rise to detectable intermediates on renaturation as detected on repetitive calorimetric scans. Consistent with the temperature, enthalpy, and cooperativity of the observed transitions the reversible denaturation of both apoproteins can be described as a four-state process (N $ N’ + U) and the thermodynamic parameters characterizing the native and intermediate states (N,N’,N”) relative to the unfolded state (U) have been determined. Thus for apoalkaline phosphatase while for apophosphoryl alkaline and AG,. 2.4 molP’. The destabilization of the native state of the apoprotein arising as a consequence of specific


Differential
scanning calorimetry has been applied to assess the nature of the thermally induced denaturation of the metal-free apo and apophosphoryl derivatives of the dimeric ZnY+ metalloenzyme alkaline phosphatase (Escherichia COW. The transitions of both species deviate markedly from the two-state model, giving rise to detectable intermediates on renaturation as detected on repetitive calorimetric scans. Consistent with the temperature, enthalpy, and cooperativity of the observed transitions the reversible denaturation of both apoproteins can be described as a four-state process (N $ N' = N" + U) and the thermodynamic parameters characterizing the native and intermediate states (N,N',N") relative to the unfolded state (U) have been determined.
Thus for apoalkaline phosphatase at 30", AC, = 19.9 kcal mol-' and AG,. = 2.3 kcal mol-' while for apophosphoryl alkaline phosphatase AG, = 5.6 kcal mol -I, AGN, = 22.2 kcal mol-', and AG,. = 2.4 kcal molP'. The destabilization of the native state of the apoprotein arising as a consequence of specific phosphorylation at a single seryl residue is compatible with the postulated subunit interactions (negative cooperativity) documented for this enzyme.
The binding of the native metal ion Zn'* at the two active center binding sites increases the stability of the enzyme by -70 kcal mo1F'. The binding of the first two Zn'+ ions appears to be cooperative.
Successive addition of a second pair of Zn'+ ions and 1 eq of Mg'+ further stabilizes the metalloenzyme by 30 and 10 kcal molP'. Metal ion binding does not, however, markedly increase the cooperativity of the observed transitions.
Determination of the entropic contribution to the relative free energy of the metalloproteins indicates that metal ion association results in the generation of structures of high internal order relative to the apoenzymes. The application of differential scanning microcalorimetry (l-3) to the investigation of thermally induced conformational transitions of proteins has allowed a rigorous assessment of * This work was SUDDOrted bv Grants AM 09070-13. GM 05168-03. and GM 04725-25 fro-4 the National Institutes of Health and by Grants GB 43481 and GB 36346X from the National Science Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "raduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. the thermodynamic parameters describing the unfolding process (4-6). These results have led to an understanding of the nature and magnitude of the forces contributing to the stabilization of the folded polypeptide at ambient temperatures (7,8). To date these studies have been performed on a variety of monomeric proteins for which detailed structural information is available. The ability of the dimeric Zns+ metalloenzyme, alkaline phosphatase, and the metal-free apoenzyme to undergo reversible dissociation and unfolding render it a suitable protein for the utilization of scanning calorimetry as a probe of enzyme secondary, tertiary, and quaternary structure. Alkaline phosphatase (EC 3.1.3.1) from Escherichia coli is a dimer (M, = 86,000) composed of identical subunits symmetrically disposed about a 2-fold rotation axis (g-11). The enzyme is a nonspecific phosphate monoester hydrolase with two separate active centers/dimer. In the native state, the protein contains four Zn" and one to two Mg'+ ions (12). The metal ions can be reversibly removed from the protein with concomitant loss of catalytic function which can be completely restored on addition of 2 eq of Zn')+ ion/dimer (13)(14)(15)(16). The native enzyme possesses a remarkable thermostability and is relatively insensitive to hydrogen ion concentration, displaying only minor structural perturbations (as monitored by optical spectroscopy) between pH 4 and pH 8 (9, 17). Removal of the intrinsic metal ions at neutral pH does not result in dissociation of the protein dimer (18,19) and only minor perturbations in secondary structure are detected (19,20). The importance of the metal ions in maintaining the conformational integrity of the enzyme and its resistance to thermally induced structural transitions has been suggested (21, 221, but little experimental evidence has been obtained documenting these proposals. The mechanism of action of alkaline phosphatase involves the intermediate phosphorylation of the enzyme at a unique serine residue (23, 24). a process subject to allosteric regulation as evidenced by the existence of negative homotropic interactions between the subunits (25)(26)(27). Under certain conditions apophosphoryl enzyme containing 1 phosphoseryl residue/enzyme dimer can be isolated (28). The symmetrical relation of the subunits is altered in the phosphorylated apoenzyme (27). Differential scanning calorimetry applied to alkaline phosphatase provides a sensitive method for: (a) comparing the nature of thermally induced structural changes of a dimeric enzyme to those of the monomeric proteins previously studied; (6) assessing the consequences of disruption of the symmetric disposition of the subunits on covalent phosphorylation; (e) determining the role of the metal ions in stabilizing protein structure; and Cd) probing the proposed structural perturbations induced by variation of the hydrogen ion concentration.

Apoenzyme
-Calorimetric scans of solutions of apoalkaline phosphatase at pH 7.5 are shown in Fig. IA. Over a ZO-fold range of protein concentration (0.31 to 6.57 mg ml ') a single transition is observed with a transition temperature (T,,,) of 57.5". The differential heat change, AQ, is a linear function of concentration (Fig. 2) corresponding to a specific enthalpy, Ah, of 5.57 cal g-' (Table Il. Following cooling to 20", the samples were immediately rescanned under the same conditions to yield the results shown in Fig. l?3. In all but the most dilute samples two transitions are clearly resolved, occurring at transition temperatures of 57.5" (identical to that observed on initial heating1 and 43.5". Repetitive scans of the reheated sample duplicate the heat capacity profile observed on second heating with respect to the position and relative amplitude of the transitions. The initial heating of the protein thus appears to convert the sample into two species undergoing thermally induced transitions at characteristic temperatures. The species undergoing the 43.5" transition is metastable. If the interval between successive thermal scans is increased from 0.5 to 9 h, the amplitude of the 43.5" transition is reduced, while the 57.5" transition has increased and approaches that observed on initial heating ( Fig. 3A; Table I    and 30 + 3% has returned to the native state. This proportion is constant over almost a lo-fold concentration range. The absence of a concentration dependence suggests that the conversion of the intermediate to the native form can be described as a first order process occurring with a half-time of -7 h (k Cu"Lenl"n = 4 x lo-" ~~'1. At the very lowest concentrations the apparent deviation (Fig. B) from this behavior may result from a reduction in the rate of conversion of protein into the native state (i.e. the 57.5" transition) on cooling. The reason for this slower conversion of the refolded apoprotein from the metastable state to the native state is unclear.
Apophosphoryl Enzyme -Calorimetric scans of the apophosphoryl enzyme in which the serine of one of the subunits has been specifically phosphorylated are given in Fig. 4. A single transition occurring at a temperature higher than that of the apoenzyme is observed on initial heating. As in the case of the apoenzyme the transition temperature CT,,,= 65") and specific enthalpy (M = 4.20 cal g-') show no dependence on protein concentration (Figs. 2 and 4A; Table I). On immediate reheating two transitions are observed (Fig. 4B). In contrast to the behavior of the apoenzyme, however, neither transition temperature (44 and 60") corresponds to that observed initially.
The low temperature transition can be identified with a metastable intermediate which slowly converts to the form undergoing the 60" transition. As with the apoenzyme, following a prolonged interval between successive scans, the amplitude of the low temperature transition is decreased while that of the high temperature transition is increased ( Fig. 3B; Table I). The initial 65" transition, and by inference the native state of the apophosphoryl enzyme, does not reappear even on prolonged equilibration.
The failure of the apophosphoryl enzyme to return to the native state is consistent with the reduction in the catalytic activity which can be restored to the enzyme on addition of Zn')+. In contrast to the apoenzyme, only -50% of the specific activity is restored to the apophosphoryl enzyme on the addition of ZnY+ following heating.
Since neither of the transitions detected on reheating correspond to the transition observed in the initial thermal scan, analysis of the system is not straightforward.
If it is assumed that no irreversible denaturation has occurred, the data can be fit by assigning values of specific enthalpy, Ah, of 1.90 cal g-' to the 44.0" transition and Ah = 4.00 cal g-' to the 60.0 where n = I3 tAH,.,lAH,) + 11/13@H,.,I~,.) ~ 11. ' Thermodynamic parameters are derived from the equations given below (7) calculated as the difference in enthalpy (Ah), entropy (Ass), and free energy CAg) between the indicated species and unfolded (U) state at 30" 1303.  sample is cooled. Paralleling the calorimetric result, the nonlinear transition is abolished on addition of 2 eq of Zn')+ to the sample. These results are qualitatively similar to analogous experiments reported for the temperature-dependent circular dichroism spectra of pancreatic ribonuclease (31).

Cooperativity of Transitions
Mechanism of Reversible Unfolding -While the mechanism of protein unfolding and folding remains a highly controversial subject, the denaturation of small globular proteins, as analyzed on the basis of thermodynamic criteria, can be approximated by a two-state model (7,32) For the two-state model the ratio of the van't Hoff enthalpy change (MZ,.,,) to the calorimetric enthalpy change W&J is unity. Deviations from this value can be interpreted as arising from the successive formation of intermediate states. The total number, n, of states present in the conversion of a species to the denatured state is related to the ratio of the van't Hoff and calorimetric enthalpies by the expression SZ,.,,/AHC = (n + 1)/3(n -1) (4). By this criterion, the thermally induced unfolding of a number of small 04, < 26,000) monomeric proteins can be described by the two-state model CAH,,,/AHH, = 1.05) (7).
In contrast, the reversible transitions of apo-and apophosphoryl alkaline phosphatase deviate markedly from this model. The value of MZ,.,,/MZ, for the transition of native apoalkaline phosphatase (I, Fig. 3A) observed on initial heating is 0.53 corresponding to II = 4.42 (Table III). The twostate model precludes the formation of intermediates of significant thermodynamic stability in either the unfolding or folding process; thus the appearance of a metastable intermediate (II, Fig. 3A) on reheating of the sample is consistent with n > 2 for the unfolding of I. While SZ,,,/SZC for the high temperature transition is unaltered on reheating. .W,.,,/ AHC associated with the low temperature transition is 1.02 (n = 1.97) in excellent agreement with the two-state model.
For apophosphoryl alkaline phosphatase, the induced transition of the native form (III, Fig. 3B) of the protein yields a value of AH,,,,/AHC of 0.54 (n = 4.22), similar to that observed for the apoenzyme and again deviating from the predicted behavior of the two-state model. As for the apoenzyme, an  (45), it is striking that numerous small regions of p structure scattered throughout the sequence are predicted by such an analysis. The predicted absence of large structural domains may be related to the high number of proline residues in the molecule (46) which are known to destabilize both u helical and p structures (43). While the energy of isomerization of proline residues is not sufficient to account for the stabilization of refolded intermediates (351, it is possible that, in a structure comprised of a large number of ordered structural domains. proline isomerization could be coupled to conformational changes involving a number of amino acid residues. Amplification of the energetic requirements for such a transition to permit kinetic and thermodynamic stabilization of intermediates is plausible in a highly segmented but ordered structure. The temperature-dependent changes in the circular dichroism spectrum of the apoenzyme reflect a large but not complete loss of ordered structure (Fig. 9). As in the corresponding calorimetry experiment (Fig. 5). the spectroscopic transition is abolished by the addition of 2 eq of Zn'+/enzyme dimer to the system. Retention of structural elements on thermal denaturation, in contrast to the total unfolding of the polypeptide induced by guanidinium chloride, has been documented for other proteins (47). Acid denaturation of alkaline phosphatase, occurring on exposure of the protein to an environment of pH < 4.0 differs from thermal unfolding, yielding completely dissociated subunits which are completely unfolded (17). In agreement with these observations, no thermal transitions are detected in the calorimetric scans of solutions of protein at pH < 4.0. The hysteresis accompanying recovery of structure on reversal of the exposure of the protein to low pH has also been confirmed in these studies. This hysteresis has been attributed to alterations in the reversibility of the titration of carboxyl groups which become exposed in the unwound, dissociated structure resulting in an apparent shift in the pI$, of these groups (17). Deprotonation, apparently required for refolding to the native state, becomes possible only at high pH and the consequent alteration of the surface charge is suggested as responsible for structural alterations preventing full recovery of quaternary structure, i.e. reassociation of monomers is complete only at pH > 6.0.
The observation of a hysteresis in the reversal of the thermal denaturation as monitored by circular dichroism might be construed as evidence that the mechanism of refolding of thermally and acid-denatured protein are related. The marked differences in the nature of the denatured states (i.e. degree of apparent association of subunits and the extent to which structure is retained) suggests that such a comparison, in the absence of more detailed information, is unwarranted. Effects of Metal Zon Binding on Heat-induced Transitions -Addition of Zn"' to solutions of the apoenzyme results in a reduction of the amount of apoprotein undergoing the thermally induced transitions (Figs. 5 and 8; Table II). The decrease in the observed transition enthalpy is proportional to the amount of metal ion added and is complete at a Zn'+ concentration corresponding to the presence of 2 eq of ZrP/ enzyme dimer. A variety of experimental methods have demonstrated that population of specific metal ion binding sites of high affinity and unusual structure is complete at a metal ion equivalence of Biapoprotein dimer (15,16). Similarly 2 eq of metal iomapoprotein dimer are necessary and sufficient to restore full functional capacity, as measured by the stoichiometry of ligand (c.g. phosphate, aryl phosphonate) binding to the enzyme (27). These results suggest that metal ion binding is a cooperative process, i.c. that formation of the two metal ion protein selectively occurs as opposed to random population of high affinity binding sites. The calorimetric titration is consistent only with the preferential formation of the two Zn" enzymes. This is true even in the absence of ligand and the presence of excess apoprotein (Fig. 5). Thus, association of metal ion at the active center binding sites must be a positively cooperative process. The same conclusion appears to apply to formation of the Cd" enzyme (Fig. 5). In contrast. the stabilization of the enzyme afforded by Mg" requires the initial presence of the Zn" ions ( Figs. 5 and 8).
The specific enthalpies of all of the metallophosphatases (VI to VIII, Fig. 8; Table III) are substantially greater than that of the apoenzyme. while the values of .K$ are markedly reduced (0.02 t,o 0.04 cal deg ' g '1. The calculated thermodynamic parameters (Table Ill) indicate that metal ion binding stabilizes the folded protein relative to the denatured state by 90 to 130 kcal mol ' in contrast to the modest (20 kcal molt '1 stabilization of the protein in the absence of metal ion. The value of the entropy change, &. between the states is also increased for the metalloenzymes, indicating a substantial increase in the degree of ordered structure relative to the apoprotein. The decrease in the value of U?;i is consistent with a reduction of nonpolar interactions on metal ion binding. The dramatic changes in the thermostability and thermodynamic properties are primarily a consequence of the binding of the first 2 eq of Zn" to the apoprotein (Fig. 81. Subsequent metal ion additions produce additional stabilization (Table III). The binding of metal ions to apoalkaline phosphatase has been shown to result in alterations in the conformational stability of the enzyme reflected in changes in solvent (H,O) access to structural elements of the protein. The binding of 2 or 4 eq of Zr?+ increases from 120 to 200 the number of protons which are not in rapid exchange with solvent as measured by hydrogen-tritium exchange (22). A number of these protons are susceptible to exchange occurring at a slow rate, leaving an inert "core" of protons which are virtually completely shielded from solvent. The number of shielded protons present in the apoenzyme is not increased on binding 2 eq of Zn'+. However, the presence of 4 eq of Zn" increases the number of exchange-inert protons by a factor of 3 (22). Biosynthetic incorporation of m-fluorotyrosine and ly-"'Clhistidine into alkaline phosphatase has permitted assay by '!'F and "'C NMR methods (48,49) of the effect of metal ion binding on the environment of individual amino acid residues. Well resolved resonances are observed in spectra of the native enzyme dimer corresponding to the number of tyrosine or histidine residues per monomer. Thus, corresponding residues on separate subunits appear to be symmetry-related.
The magnitude of the chemical shifts reflect the local environment of the amino acids. Removal of the metal ions results in changes in the spectra consistent with a reduction in ordered structure and an increased susceptibility of the protein structure to solvent penetration.
Restoration of the well resolved pattern of resonances is largely complete on addition of 2 eq of Zn'+ to the apoprotein, but the presence of 4 eq of Znr+ appears necessary to duplicate precisely the spectrum of the native enzyme.
The interpretation of the calorimetric data in thermodynamic terms is consistent with these observations. There appears to be a collapse of ordered elements of structure on removal of the metal ion. To compensate for the loss of defined structure, the apoproteins adopt conformations which increase the association of hydrophobic groups as reflected in the increase in heat capacity (AC$ (Table III) (7). There is a concomitant exposure of polar groups to enhanced solvent interaction as indicated by the NMR data (48,49). Nevertheless, the removal of the metal ions results in an enormous decrease in thermodynamic stability. The multistate mechanism for the unfolding of alkaline phosphatase is maintained on metal ion binding. In fact, the addition of the first 2 eq of metal ion decreases the cooperativity of the observed transition (AH,.,/AZ& (VI) = 0.45). Subsequent metal ion additions increase the cooperativity moderately (AH.,,/m, (VII) = 0.53, AH,.,/AH, (VIII) = 0.62), but in no case does the system correspond to the two-state model. This suggests that as structural constraints are imposed on the enzyme, as a consequence of sequential population of metal ion binding sites, the energetic separation of thermodynamically significant states is decreased. These data indicate that occupation of distinct metal-ion binding sites results in formation of chemically distinct species (VI to VIII). Binding of Zn" at the tight binding sites associated with the two active centers of the dimer is sufficient to restore functional capacity and structural integrity to the protein and is necessary before the consequences of metal ion association at other binding sites can become effective (Fig.  8). Both structure and function depend primarily on occupancy of the active center binding sites (501, but both structural stability and catalytic efficiency appear to be positively affected by occupancy of the additional binding sites for Zn'+ and Mg" (12,51,52).
Experiments conducted on samples of native enzyme gave results qualitatively similar to those observed for the reconstituted metalloalkaline phosphatases. However, the value for the transition enthalpy was significantly higher (A/z = 16.0 cal go '1, and multiple transitions, occurring at high temperatures CT,,, > 90?, were resolved as an apparent function of decreasing enzyme concentration.
As isolated, alkaline phosphatase is known to be heterogenous with respect to metal ion and phosphate content. Given the effect of metal ion stoichiometry on the transition temperature and transition enthalpy it is reasonable to attribute the detected differences in the native and reconstituted protein to the heterogeneity of the enzyme as isolated.
The relationship of these studies to the nature of the folding and subunit association of alkaline phosphatase occurring in uiuo is difficult to assess. The biosynthetic generation is a complex process. The site of enzyme synthesis appears to be at polyribosomes associated with the plasma membrane (53). Preliminary reports suggest that alkaline phosphatase may be synthesized as a preenzyme which is processed to yield the form of the protein isolated from the periplasmic space (53). The temporal and spatial relationships of protein synthesis, molecular processing, transport to and across the membrane, subunit association, and metal ion binding are incompletely documented.
The presence of Zn"+ does not appear to be required for assembly of the dimer in Go. Cultures of Escherichia coli grown in a medium depleted of metal ions synthesize alkaline phosphatase, which is delivered to the periplasmic space and can be isolated as the apoprotein (19). Apoalkaline phosphatase generated in vivo is identical to apoenzyme produced on removal of metal ions from the native protein.
In contrast to the above, treatment of cultures of cells with 1 2 3 lysozyme and EDTA produces spheroplasts capable of synthesizing and extruding enzyme monomers which show no evidence of dimerization by antigenic criteria (54). The failure to form dimeric structures indicates that the protein generated from the spheroplast is not identical to the enzyme as it appears in the periplasmic space. While the mode of formation and lifetime of apoprotein is unclear, the sequestration of metal ions generating the metalloenzyme in viva clearly results in a remarkable stabilization of the enzyme. The detailed pattern of structural changes preceding and accompanying generation of the biosynthetic product remain to be determined.