- Email: [email protected]
Cooperativity in the unfolding transitions of cysteine proteinases. Calorimetric study of the heat denaturation of chymopapain and papain
121
Biochimica et Biophysica Acta, 1203 (1993) 121-125 © 1993 Elsevier Science Publishers B.V. All rights reserved 0167-4838/93/$06.00
BBAPRO 34607
Cooperativity in the unfolding transitions of cysteine proteinases. Calorimetric study of the heat denaturation of chymopapain and papain Silvia Solls-Mendiola, Arturo Rojo-Domlnguez and Andr6s Hernfindez-Arana
*
Departamento de Qu[mica, UniuersidadAutdnoma Metropolitana-Iztapalapa, Apartado Postal 55-534, Iztapalapa, D.F. 09340, M~xico City (M~xico) (Received 9 March 1993) (Revised manuscript received 9 July 1993)
Key words: Cysteine proteinase; Thermal unfolding; Differential scanning calorimetry; Domain interaction
Differential scanning calorimetry (DSC) was employed to study the thermal unfolding of chymopapain (EC 3.4.22.6) and papain (EC 3.4.22.2), two highly homologous cysteine proteinases from Carica papaya. Under all pH conditions used, both enzymes showed irreversible thermal denaturation. However, results from experiments performed at two different scanning rates suggest that interpretation of data in terms of equilibrium thermodynamics is not unreasonable. For papain, the ratio of calorimetric (AHca~) to van't Hoff (AHvH) enthalpies approximated to 2.0. This value indicates that papain domains unfold almost independently, as it has been reported previously. In contrast, chymopapain displayed a more cooperative behavior with a AHcaj to AHvH ratio of 1.3-1.4. DSC curves were analyzed in terms of a mechanism that includes domain-domain interactions. The results showed a negligible interdomain free energy in the case of papain, but a significant value of approx. 1.0 kcal/mol (1 cal = 4.184 J) for chymopapain. These two proteins also differed in the unfolding heat-capacity change, ACp, which suggests that their native structures bury different amounts of nonpolar surface area.
Introduction
It is well known that many small globular proteins behave as single cooperative units on unfolding, showing no stable intermediates during this process [1,2]. In those cases, a decisive test for the absence of stable intermediates has relied on the comparison of the enthalpy change as measured calorimetrically, AHca~, with the van't Hoff enthalpy, AHvn, deduced from the temperature dependence of the equilibrium constant [1,3-5]. If the unfolding transition follows a two-state mechanism, the ratio AHcal/AHvH should be close to 1.0 [1,4]. On the other hand, thermal denaturation of multidomain proteins is usually a more complex process where several cooperative transitions may be present [6,7]. Although in some cases the transitions are strongly overlapped, deviation from the two-state mechanism is clearly indicated by a value of Anca I higher than AHvH. It should be noted, however, that certain proteins whose three-dimensional structures consist of two distinct domains unfold as single cooper-
* Corresponding author. Fax: +52 5 7244666.
ative units, without stable intermediates [1,7,8]. Apparently, the precise behavior of a protein on unfolding depends on the delicate balance between domain-domain interactions and intrinsic stabilities of the domains [8,9]. The thermal unfolding of papain, a cysteine proteinase whose structure shows two domains separated by a deep cleft [10], has been studied by calorimetric and spectroscopic methods [11,12]. These studies indicate that papain domains unfold almost independently. Crystalline structures of two other plant cysteine proteinases (actinidin from Actinidia chinensis, and proteinase O from Carica papaya) have been determined in detail [13,14]; in both cases the two-domain architecture of papain is clearly observed. It is probable that the same type of folding pattern will be found in other members of this protein family, since these enzymes display a high amino-acid sequence identity [15,16]. However, it is not known yet whether other cysteine proteinases show the unfolding behavior characteristic of papain. In this work, we studied the thermal unfolding of chymopapain and papain by differential scanning calorimetry (DSC). These proteinases are obtained from the same source and their sequences are 58%
122 identical [16]. In spite of this similarity, circular dichroism studies suggest that parts of the structure of native chymopapain might differ from the papain structure [17]. Furthermore, it is reported here that these proteinases show different cooperativities and heat capacity changes during thermal unfolding.
K-'g-'~ _ _
I lOcol
Materials and Methods
Materials. Partially purified chymopapain and 2times crystallized papain were obtained from Sigma (St. Louis, MO, USA). The main chymopapain form was isolated by cation-exchange liquid chromatography as previously described [17]. To avoid autolysis during unfolding experiments, the active site of both enzymes was blocked by treatment with iodoacetamide [18]. Protein concentration was determined by means of the absorption coefficient ~A~% Xl-lcmJ~ at 278 nm. For papain the value 25.0 was employed [19]. The corresponding value for the main chymopapain form was 18.2, determined according to the method of Robinson [20]. In the calculation of molar quantities, molecular weights of 23 500 (papain) and 24 000 (chymopapain) were used [10,16,21]. Differential scanning calorimetry. Calorimetric scans were performed with a MicroCal MC-2 (Microcal, Northampton, MA, USA) differential scanning calorimeter, in the pH range 1.20-4.10. Protein solutions of 2.0-5.0 m g / m l were extensively dialyzed against 0.05 M glycine buffer of the desired pH value. After dialysis, protein concentration and pH of the samples were checked. All solutions were degassed under vacuum before being loaded into the calorimeter cells. Most calorimetric experiments were conducted at a scan rate of 60 C°/h; in some cases the scan rate used was 30 C°/h. Buffer-buffer base lines were obtained under the same conditions and subtracted from sample curves. The DA-2 software package (Microcal) was used for automatic data collection and data analysis that included baseline subtraction and calculation of AH¢a~ and AHvH. Curve fitting involving nonlinear least-squares was done employing a modified version of the GRIDLS program [22]. Results
Differential scanning calorimetry The DSC curves of papain and chymopapain, corrected for buffer-buffer tracings, obtained at pH 3.18 and 60 C ° / h are shown in Fig. 1. In the pH range studied, the thermal unfolding of both proteins was apparently irreversible, as indicated by the fact that no endotherm was observed by rescanning of the samples after being cooled from the first scan. We tested the effect of scan rate on the temperature at the maximum of the heat capacity curve, Tm, by performing some
I
I
50
60
I
I
70 80 Temperature (*C)
I
I
go
I00
Fig. 1. DSC recordings of chymopapain (upper curve) and papain (lower curve) after subtraction of buffer-buffer baseline. Experiments were done at pH 3.18 with a scan rate of 1.0 C°/min. Protein concentration was 3.28 and 2.90 mg/ml for chymopapain and papain, respectively.
experiments at heating rates of 30 C°/h. For the two enzymes it was observed that the value of Tm varied by as much as 1.0°C when the scan rate was doubled. These results suggest that the source of irreversibility is a process that takes place at temperatures higher than those at which the unfolding transition occurs [23]. Therefore, application of equilibrium thermodynamics to the analysis of our data will lead to unfolding parameters close to the true equilibrium values [23]. To determine the enthalpy of unfolding, calorimetric curves, such as those shown in Fig. 1, were corrected for the difference in permanent heat capacity, ACp, by subtraction of a linear base line connecting the initial and final temperatures of the transition; i.e., an excess heat capacity function, Cex, was calculated and from it AHcaI and AHvH were determined. Essentially the same results were obtained when sigmoidal base lines were used in this procedure. A summary of the thermodynamic parameters that characterize the thermal unfolding of papain and chymopapain is presented in Table I. It can be seen that for papain the ratio AHcal/AHvH approaches a value of 2,0, in agreement with previous results of Tiktopulo and Privalov [11]. This quotient of enthalpies is an indication that papain domains undergo quasi-independent unfolding transitions. In contrast, chymopapain displays a markedly different cooperative behavior, since in this case AH~a~/AHvH is 1.3-1.4 in the entire pH range studied.
Analysis of domain-domain interactions In order to describe quantitatively the differences in cooperativity just mentioned, we analyzed our calorimetric data in terms of the model proposed by Brandts
123 TABLE I Unfolding thermodynamic characteristics of papain and chymopapain
Calorimetric and van't Hoff enthalpies were calculated from excess heat-capacity functions and, therefore, are referred to T = Tm. pH
Papain
Chymopapain
AHvH
AHca 1
(°C) (kcal/mol)
Tm
AHca I
(kcal/mol)
/AHvH
4.10 3.80 3.18 2.90 2.55
81.8 78.5 74.6 60.3 51.0
210.8 197.5 186.0 133.0 107.9
115.8 108.5 96.0 78.0 65.4
1.82 1.82 1.94 1.71 1.65
3.45 3.18 2.90 2.56 2.35 2.20 1.91 1.20
83.7 81.1 78.2 75.2 67.9 62.0 56.9 43.0
198.7 193.4 186.2 174.0 161.2 143.8 138.2 104.6
144.8 142.7 138.3 135.8 123.1 108.3 99.2 73.5
1.37 1.36 1.35 1.28 1.31 1.33 1.39 1.42
that each d o m a i n behaves as a single cooperative unit), and the 4) term includes all interdomain interactions. T h e K i terms can be expressed as functions of temperature by the known relationships K i = exp( - zaGi/RT) AG
i =
AH
i -- TAS
i =
(6)
AH i -
TAH
i /
(7)
Tm,i
in which the transition enthalpy for the unfolding of the isolated domain, A H i, is introduced t o g e t h e r with its transition t e m p e r a t u r e , Tm,i. T h e 4) term is related to the free energy of d o m a i n - d o m a i n interaction, AG~, t h r o u g h an equation analogous to Eqn. 6. At any given t e m p e r a t u r e , the excess enthalpy of unfolding, ( A H ) , is o b t a i n e d according to Eqn. 8 which includes the enthalpy associated with the disruption of interdomain contacts (AH~). Finally, Cex is evaluated as the temp e r a t u r e derivative of ( A H ) . ( a l l ) = ( a l l A + AH+)PA + ( A H B + AHco)P a
et al. [8] which is a particular case of the general model of R a m s a y and Freire [24]. F o r the unfolding of a two-domain protein, the four states considered in the model are: the native (N) and completely u n f o l d e d (U) states; and two partially u n f o l d e d states (A and B) associated with forms of the molecule in which one d o m a i n is folded and the o t h e r unfolded. T h e fractional populations, P, of the states can be calculated from the following equations: PN=I/Q
(1)
PA = K A & / Q
(2)
PB = KBga/Q
(3)
e u = KAKBd~ / Q
(4)
w h e r e Q, the equilibrium partition function, is given by (5)
Q = 1 + KA~b + KB~b + KAKB~b
In the above equations, K A and K B represent the intrinsic unfolding constants of the domains (assuming
+ (alia + AHB + AH~)PD
(8)
In the analysis of an experimental Cex curve, the p a r a m e t e r s to be adjusted are A H A, A H B, TIn,A, Tin,a, ~IG6 and AH6. T o keep the m o d e l as simple as possible we considered AG6 as a constant equal to AH~, thus assigning all entropy changes solely to d o m a i n denaturations. Similarly, transition enthalpies for the individual domains were taken as constant quantities. Calorimetric data for papain ( p H 3.80) and c h y m o p a pain ( p H 2.90) were analyzed according to the procedure outlined above by m e a n s of non-linear leastsquares fitting. Calculated values of Cex were, in general, in good a g r e e m e n t with experimental data (Fig. 2) as indicated by the s t a n d a r d deviation of fitting that was less than 5% of the m a x i m u m Cex value in both cases. P a r a m e t e r s obtained from curve fitting are summarized in Table II. For papain, the negligibly small AG6 indicates that the domains of this protein unfold almost completely i n d e p e n d e n t from each other. Besides, the values of A H A and A H B do not differ too
TABLE II Thermodynamic parameters obtained from the deconvolution of calorimetric curves
DSC data for papain (pH 3.80) and chymopapain (pH 2.90) were fitted to an unfolding mechanism that considers two interacting cooperative units. SDF, standard deviation of fitting. Transition 1
Papain Chymopapain Chymopapain a a
Tin,A (°C) 75.5 ± 0.3 71.1 + 0.3 73.1 + 0.2
Transition 2
Interaction term
AHA
Tm.a
AH a
AG~,
(kcal/mol) 97 ± 7 67 ± 6 88 ± 8
(°C) 78.9 ± 0.2 77.8 ± 0.1 77.8 _+0.2
(kcal/mol) 126 ± 9 134 ± 8 112 ± 7
(kcal/mol) 0.005 ± 0.04 0.97 + 0.04 1.15 ± 0.05
In this case, a restriction was imposed to force the ratio of enthalpies to a value similar to that of papain.
SDF (cal/mol per K) 923 1053 1228
124 much (AHA//tH B = 1.29), as would be expected for a protein with domains of similar size [6]. In the case of chymopapain, the difference between AHA and AHB was considerably larger (ZaHA/AH8 = 1.98) and the interaction term acquired a significant value of 0.97 kcal/mol. However, given the high sequence identity and nearly equal molecular weight of the two enzymes, it is reasonable to assume that the molecular structure of chymopapain is formed by domains of a size akin to those of papain. Therefore, we further analyzed the chymopapain curve, introducing a restriction that limited the largest transition enthalpy to a value not larger than 1.29-times the value of the smallest enthalpy. Results obtained with this restriction are shown in the bottom row of Table II. The standard deviation of fitting was again less than 5% of the maximum value of the curve, indicating a satisfactory agreement between calculated and experimental data. With this approach, ziG, is slightly larger than in the previous analysis.
Temperature dependence of the unfolding enthalpy Fig. 3 illustrates the influence of temperature on the specific unfolding enthalpy, AHca1, of the two proteins. For comparison, data reported by Tiktopulo and Privalov [11] are also shown. As can be seen, there is a good linear relationship between AHcaI and the temperature, in concordance with results for several other small proteins [1,3]. Indeed, it is generally accepted that in a broad temperature range the enthalpy of unfolding is a linear function of temperature with a slope equal to the heat capacity difference between unfolded and native states [1,25]. From Fig. 3, heat capacity differences of 0.096 c a l / K per g (2.3 kcal/K
2520o
15I0_
¢J
5-
%
O-
I
65
I
?(3
I
75 Temperature
[
80 (*C)
I
8,5
Fig. 2. Excess heat capacity functions for chymopapain at pH 2.90 ( ) and papain at pH 3.80 ( . . . . . . ). Fitting the data to an unfolding mechanism that considers two interacting domains gave values of Cex indicated as triangles (chymopapain) and circles (papain).
IO 0
o
<:1 6
4 30
40
50
60
70
80
9(3
I00
Temperoture (*C)
Fig. 3. Temperature dependence of the enthalpy change (per gram of protein) for chymopapain ( A ) and papain (11). For papain, data reported by Tiktopulo and Privalov [11] are also shown ( [] ).
per mol) and 0.142 c a l / K per g (3.3 kcal/K per mol) were calculated for chymopapain and papain, respectively. These values agree well with those determined directly from the original calorimetric curves (cf., Fig. 1) which were 2.1 + 0.5 kcal/K per mol (chymopapain) and 3.0 + 0.4 kcal/K per real (papain). These latter determinations were done by linearly extrapolating the pre- and postdenaturational parts of each calorimetric recording and evaluating the difference between them at Tm [1,3]. Discussion
The results of this study showed that the unfolding transition of chymopapain is not a two-state process, as indicated by the ratio of calorimetric to van't Hoff enthalpies. In this respect, the unfolding behavior of chymopapain resembles that of papain, although a difference in the cooperativity of these processes is clearly noticed (Table I). Apparently, the domains of the papain molecule unfold in an independent way which leads to a value of AHcal/ZlHvH close to 2.0, as previously reported by Tiktopulo and Privalov [11]. The analysis of calorimetric data, based on a mechanism that includes domain-domain interactions [8,24], yielded results in agreement with this proposal since the interaction free energy, AG,, is negligible. For chymopapain, in contrast, the value of ziG, obtained from the same type of analysis was significantly larger (approx. 1.0 kcal/mol). It must be noted that AG, had little variation when the ratio of the two domain-unfolding enthalpies was restricted to a value similar to that observed for papain denaturation (Table II). A more complex method of analysis (i.e., one including different ch terms for the unfolding of different domains) would probably provide more information on the cooperative behavior of these macromolecules during denaturation. However, before attempting an anal-
125 ysis of this type it would be convenient to have a better understanding of the irreversible nature of these phenomena. In any case, calorimetric data of chymopapain seem to be consistent with a bilobal molecular structure like that of papain yet with stronger domain-domain interactions. Furthermore, it is possible that these different interdomain contacts may be the cause of small variations in the folding of the polypeptide chain, thus explaining the dissimilar circular dichroism spectra these proteins show [17]. Further calorimetric studies with other cysteine proteinases of known three-dimensional structure will probably help to clarify this point. In that way, a detailed correlation between molecular structure and unfolding behavior could be established for the members of this family, as it has been recently done with some other proteins [9,26]. Another point that deserves comment is the different heat capacity change associated with the unfolding of papain and chymopapain. This suggests that these macromolecules possess unlike hydrophobic cores, since it is considered that the value of ACp is, to a first approximation, proportional to the nonpolar area buried in the native structure of a protein [1,25]. In addition to its important connection with the hydrophobic effect, ACp is a fundamental thermodynamic parameter that determines the temperature dependence of entropy and enthalpy changes. Within a relatively wide temperature range ACp can be regarded as a constant [25] and, as a consequence, the unfolding enthalpy follows a linear dependence on temperature which can be written as AH = AH*
+ ACp(T - T~)
(9)
where T~ is a reference temperature at which AH = A H * . The constant T~ has a special meaning because the enthalpy changes (expressed on a per gram or per residue basis) for a number of proteins extrapolate to an approximately common value at this temperature, as first noted by Privalov and Khechinashvili [3]. It has been proposed that at the reference temperature the contribution of the hydrophobic effect to AH is zero, i.e., at this temperature the unfolding enthalpy is only due to polar interactions (hydrogen bonds and polar van der Waals contacts) [1,9]. Therefore, proteins having equal numbers of polar interactions per gram should show the same enthalpy value (AH *) when T = Try, as Eqn. 9 indicates. However, the precise value of Tr~ has been a matter of controversy. From the analysis of thermodynamic properties of proteins, different authors have assigned values of 112 [27], 100 [9] or 84°C [28] to TI~. Nevertheless, in this work we found that the specific unfolding enthalpies of chymopapain and papain coincide approximately at 64°C (Fig. 3), a temperature lower than any of the values of T~ indicated above. This finding suggests that members of a protein
family may have thermodynamic properties that distinguish them from other proteins. In other words, it is possible that the values of TI~ and AH* may vary from family to family in response to the particular network of hydrogen bonds and hydrophobic interactions that characterize each type of structural folding pattern.
Acknowledgements This work was supported in part by the Consejo Nacional de Ciencia y Tecnologia, Mdxico (Convenio No. 0609-N9110).
References 1 Privalov, P.L. (1979) Adv. Protein Chem. 33, 167-241. 2 Kim, P.S. and Baldwin, R.L. (1982) Annu. Rev. Biochem. 51, 459-489. 3 Privalov, P.L. and Kechinashvili, N.N. (1974) J. Mol. Biol. 86, 665-684. 4 Sturtevant, J.M. (1987) Annu. Rev. Phys. Chem. 38, 463-488. 5 Cantor, C.R. and Schimmel, P.R. (1980) Biophysical Chemistry, part III, pp. 1075-1088, W.H. Freeman, New York. 6 Privalov, P.L. (1982) Adv. Protein Chem. 35, 1-104. 7 Hu, C.Q. and Sturtevant, J.M. (1987) Biochemistry 26, 178-182. 8 Brandts, J.F., Hu, C.Q., Lin, L.N. and Mas, M.T. (1989) Biochemistry 28, 8588-8596. 9 Murphy, K.P. and Freire, E. (1992) Adv. Protein Chem. 43, 313-361. 10 Kamphuis, I.G., Kalk, K.H., Swarte, M.B.A. and Drenth, J. (1984) J. Mol. Biol. 179, 233-256. 11 Tiktopulo, E.I. and Privalov, P.L. (1978) FEBS Lett. 91, 57-58. 12 Hermlndez-Arana, A. and Soriano-Garcla, M. (1988) Biochim. Biophys. Acta 954, 170-175. 13 Baker, E.N. and Dodson, E.J. (1980) Acta Cryst. A36, 559-572. 14 Pickersgil, R.W., Rizkallah, P., Harris, G.W. and Goodenough, P.W. (1991) Acta Cryst. B47, 766-771. 15 Baker, E.N. and Drenth, J. (1987) in Biological Macromolecules and Assemblies, Vol. 3 (Jurnak, F.A. and McPherson, A., eds.), pp. 313-368, Wiley, New York. 16 Jacquet, A., Kleinschmidt, T., Schnek, A.G., Looze, Y. and Braunitzer, G. (1989) Biol. Chem. Hoppe-Seyler 370, 425-434. 17 Solls-Mendiola, S., Arroyo-Reyna, A. and Hern~ndez-Arana, A. (1992) Biochim. Biophys. Acta 1118, 288-292. 18 Solis-Mendiola, S., Zubillaga-Luna, R., Rojo-Domlnguez, A. and Hermlndez-Arana, A. (1989) Biochem. J. 257, 183-186. 19 Glazer, A.N. and Smith, E.L. (1971) in the Enzymes, 3rd Edn. (Boyer, P.D., ed.), pp. 502-545, Academic Press, New York. 20 Robinson, G.W. (1975) Biochemistry 14, 3695-3700. 21 Goodenough, P.W. and Owen, J. (1987) Phytochemistry 26, 75-79. 22 Bevington, P.R. (1969) Data Reduction and Error Analysis for the Physical Sciences, pp. 204-246, McGraw-Hill, New York. 23 Freire, E., van Osdol, W.W., Mayorga, O.L. and S~nchez-Rulz, J.M. (1990) Annu. Rev. Biophys. Chem. 19, 159-188. 24 Ramsay, G. and Freire, E. (1990) Biochemistry 29, 8677-8683. 25 Privalov, P.L. and Gill, S.J. (1988) Adv. Protein Chem. 39, 191234. 26 Murphy, K.P., Bhakuni, V., Xie, D. and Freire, E. (1992) J. Mol. Biol. 227, 293-306. 27 Murphy, K.P., Privalov, P.L. and Gill, S.J. (1990) Science 247, 559-561 28 Spolar, R.S., Livingstone, J.R. and Record, M.T., Jr. (1992) Biochemistry 31, 3947-3955.
Biochimica et Biophysica Acta, 1203 (1993) 121-125 © 1993 Elsevier Science Publishers B.V. All rights reserved 0167-4838/93/$06.00
BBAPRO 34607
Cooperativity in the unfolding transitions of cysteine proteinases. Calorimetric study of the heat denaturation of chymopapain and papain Silvia Solls-Mendiola, Arturo Rojo-Domlnguez and Andr6s Hernfindez-Arana
*
Departamento de Qu[mica, UniuersidadAutdnoma Metropolitana-Iztapalapa, Apartado Postal 55-534, Iztapalapa, D.F. 09340, M~xico City (M~xico) (Received 9 March 1993) (Revised manuscript received 9 July 1993)
Key words: Cysteine proteinase; Thermal unfolding; Differential scanning calorimetry; Domain interaction
Differential scanning calorimetry (DSC) was employed to study the thermal unfolding of chymopapain (EC 3.4.22.6) and papain (EC 3.4.22.2), two highly homologous cysteine proteinases from Carica papaya. Under all pH conditions used, both enzymes showed irreversible thermal denaturation. However, results from experiments performed at two different scanning rates suggest that interpretation of data in terms of equilibrium thermodynamics is not unreasonable. For papain, the ratio of calorimetric (AHca~) to van't Hoff (AHvH) enthalpies approximated to 2.0. This value indicates that papain domains unfold almost independently, as it has been reported previously. In contrast, chymopapain displayed a more cooperative behavior with a AHcaj to AHvH ratio of 1.3-1.4. DSC curves were analyzed in terms of a mechanism that includes domain-domain interactions. The results showed a negligible interdomain free energy in the case of papain, but a significant value of approx. 1.0 kcal/mol (1 cal = 4.184 J) for chymopapain. These two proteins also differed in the unfolding heat-capacity change, ACp, which suggests that their native structures bury different amounts of nonpolar surface area.
Introduction
It is well known that many small globular proteins behave as single cooperative units on unfolding, showing no stable intermediates during this process [1,2]. In those cases, a decisive test for the absence of stable intermediates has relied on the comparison of the enthalpy change as measured calorimetrically, AHca~, with the van't Hoff enthalpy, AHvn, deduced from the temperature dependence of the equilibrium constant [1,3-5]. If the unfolding transition follows a two-state mechanism, the ratio AHcal/AHvH should be close to 1.0 [1,4]. On the other hand, thermal denaturation of multidomain proteins is usually a more complex process where several cooperative transitions may be present [6,7]. Although in some cases the transitions are strongly overlapped, deviation from the two-state mechanism is clearly indicated by a value of Anca I higher than AHvH. It should be noted, however, that certain proteins whose three-dimensional structures consist of two distinct domains unfold as single cooper-
* Corresponding author. Fax: +52 5 7244666.
ative units, without stable intermediates [1,7,8]. Apparently, the precise behavior of a protein on unfolding depends on the delicate balance between domain-domain interactions and intrinsic stabilities of the domains [8,9]. The thermal unfolding of papain, a cysteine proteinase whose structure shows two domains separated by a deep cleft [10], has been studied by calorimetric and spectroscopic methods [11,12]. These studies indicate that papain domains unfold almost independently. Crystalline structures of two other plant cysteine proteinases (actinidin from Actinidia chinensis, and proteinase O from Carica papaya) have been determined in detail [13,14]; in both cases the two-domain architecture of papain is clearly observed. It is probable that the same type of folding pattern will be found in other members of this protein family, since these enzymes display a high amino-acid sequence identity [15,16]. However, it is not known yet whether other cysteine proteinases show the unfolding behavior characteristic of papain. In this work, we studied the thermal unfolding of chymopapain and papain by differential scanning calorimetry (DSC). These proteinases are obtained from the same source and their sequences are 58%
122 identical [16]. In spite of this similarity, circular dichroism studies suggest that parts of the structure of native chymopapain might differ from the papain structure [17]. Furthermore, it is reported here that these proteinases show different cooperativities and heat capacity changes during thermal unfolding.
K-'g-'~ _ _
I lOcol
Materials and Methods
Materials. Partially purified chymopapain and 2times crystallized papain were obtained from Sigma (St. Louis, MO, USA). The main chymopapain form was isolated by cation-exchange liquid chromatography as previously described [17]. To avoid autolysis during unfolding experiments, the active site of both enzymes was blocked by treatment with iodoacetamide [18]. Protein concentration was determined by means of the absorption coefficient ~A~% Xl-lcmJ~ at 278 nm. For papain the value 25.0 was employed [19]. The corresponding value for the main chymopapain form was 18.2, determined according to the method of Robinson [20]. In the calculation of molar quantities, molecular weights of 23 500 (papain) and 24 000 (chymopapain) were used [10,16,21]. Differential scanning calorimetry. Calorimetric scans were performed with a MicroCal MC-2 (Microcal, Northampton, MA, USA) differential scanning calorimeter, in the pH range 1.20-4.10. Protein solutions of 2.0-5.0 m g / m l were extensively dialyzed against 0.05 M glycine buffer of the desired pH value. After dialysis, protein concentration and pH of the samples were checked. All solutions were degassed under vacuum before being loaded into the calorimeter cells. Most calorimetric experiments were conducted at a scan rate of 60 C°/h; in some cases the scan rate used was 30 C°/h. Buffer-buffer base lines were obtained under the same conditions and subtracted from sample curves. The DA-2 software package (Microcal) was used for automatic data collection and data analysis that included baseline subtraction and calculation of AH¢a~ and AHvH. Curve fitting involving nonlinear least-squares was done employing a modified version of the GRIDLS program [22]. Results
Differential scanning calorimetry The DSC curves of papain and chymopapain, corrected for buffer-buffer tracings, obtained at pH 3.18 and 60 C ° / h are shown in Fig. 1. In the pH range studied, the thermal unfolding of both proteins was apparently irreversible, as indicated by the fact that no endotherm was observed by rescanning of the samples after being cooled from the first scan. We tested the effect of scan rate on the temperature at the maximum of the heat capacity curve, Tm, by performing some
I
I
50
60
I
I
70 80 Temperature (*C)
I
I
go
I00
Fig. 1. DSC recordings of chymopapain (upper curve) and papain (lower curve) after subtraction of buffer-buffer baseline. Experiments were done at pH 3.18 with a scan rate of 1.0 C°/min. Protein concentration was 3.28 and 2.90 mg/ml for chymopapain and papain, respectively.
experiments at heating rates of 30 C°/h. For the two enzymes it was observed that the value of Tm varied by as much as 1.0°C when the scan rate was doubled. These results suggest that the source of irreversibility is a process that takes place at temperatures higher than those at which the unfolding transition occurs [23]. Therefore, application of equilibrium thermodynamics to the analysis of our data will lead to unfolding parameters close to the true equilibrium values [23]. To determine the enthalpy of unfolding, calorimetric curves, such as those shown in Fig. 1, were corrected for the difference in permanent heat capacity, ACp, by subtraction of a linear base line connecting the initial and final temperatures of the transition; i.e., an excess heat capacity function, Cex, was calculated and from it AHcaI and AHvH were determined. Essentially the same results were obtained when sigmoidal base lines were used in this procedure. A summary of the thermodynamic parameters that characterize the thermal unfolding of papain and chymopapain is presented in Table I. It can be seen that for papain the ratio AHcal/AHvH approaches a value of 2,0, in agreement with previous results of Tiktopulo and Privalov [11]. This quotient of enthalpies is an indication that papain domains undergo quasi-independent unfolding transitions. In contrast, chymopapain displays a markedly different cooperative behavior, since in this case AH~a~/AHvH is 1.3-1.4 in the entire pH range studied.
Analysis of domain-domain interactions In order to describe quantitatively the differences in cooperativity just mentioned, we analyzed our calorimetric data in terms of the model proposed by Brandts
123 TABLE I Unfolding thermodynamic characteristics of papain and chymopapain
Calorimetric and van't Hoff enthalpies were calculated from excess heat-capacity functions and, therefore, are referred to T = Tm. pH
Papain
Chymopapain
AHvH
AHca 1
(°C) (kcal/mol)
Tm
AHca I
(kcal/mol)
/AHvH
4.10 3.80 3.18 2.90 2.55
81.8 78.5 74.6 60.3 51.0
210.8 197.5 186.0 133.0 107.9
115.8 108.5 96.0 78.0 65.4
1.82 1.82 1.94 1.71 1.65
3.45 3.18 2.90 2.56 2.35 2.20 1.91 1.20
83.7 81.1 78.2 75.2 67.9 62.0 56.9 43.0
198.7 193.4 186.2 174.0 161.2 143.8 138.2 104.6
144.8 142.7 138.3 135.8 123.1 108.3 99.2 73.5
1.37 1.36 1.35 1.28 1.31 1.33 1.39 1.42
that each d o m a i n behaves as a single cooperative unit), and the 4) term includes all interdomain interactions. T h e K i terms can be expressed as functions of temperature by the known relationships K i = exp( - zaGi/RT) AG
i =
AH
i -- TAS
i =
(6)
AH i -
TAH
i /
(7)
Tm,i
in which the transition enthalpy for the unfolding of the isolated domain, A H i, is introduced t o g e t h e r with its transition t e m p e r a t u r e , Tm,i. T h e 4) term is related to the free energy of d o m a i n - d o m a i n interaction, AG~, t h r o u g h an equation analogous to Eqn. 6. At any given t e m p e r a t u r e , the excess enthalpy of unfolding, ( A H ) , is o b t a i n e d according to Eqn. 8 which includes the enthalpy associated with the disruption of interdomain contacts (AH~). Finally, Cex is evaluated as the temp e r a t u r e derivative of ( A H ) . ( a l l ) = ( a l l A + AH+)PA + ( A H B + AHco)P a
et al. [8] which is a particular case of the general model of R a m s a y and Freire [24]. F o r the unfolding of a two-domain protein, the four states considered in the model are: the native (N) and completely u n f o l d e d (U) states; and two partially u n f o l d e d states (A and B) associated with forms of the molecule in which one d o m a i n is folded and the o t h e r unfolded. T h e fractional populations, P, of the states can be calculated from the following equations: PN=I/Q
(1)
PA = K A & / Q
(2)
PB = KBga/Q
(3)
e u = KAKBd~ / Q
(4)
w h e r e Q, the equilibrium partition function, is given by (5)
Q = 1 + KA~b + KB~b + KAKB~b
In the above equations, K A and K B represent the intrinsic unfolding constants of the domains (assuming
+ (alia + AHB + AH~)PD
(8)
In the analysis of an experimental Cex curve, the p a r a m e t e r s to be adjusted are A H A, A H B, TIn,A, Tin,a, ~IG6 and AH6. T o keep the m o d e l as simple as possible we considered AG6 as a constant equal to AH~, thus assigning all entropy changes solely to d o m a i n denaturations. Similarly, transition enthalpies for the individual domains were taken as constant quantities. Calorimetric data for papain ( p H 3.80) and c h y m o p a pain ( p H 2.90) were analyzed according to the procedure outlined above by m e a n s of non-linear leastsquares fitting. Calculated values of Cex were, in general, in good a g r e e m e n t with experimental data (Fig. 2) as indicated by the s t a n d a r d deviation of fitting that was less than 5% of the m a x i m u m Cex value in both cases. P a r a m e t e r s obtained from curve fitting are summarized in Table II. For papain, the negligibly small AG6 indicates that the domains of this protein unfold almost completely i n d e p e n d e n t from each other. Besides, the values of A H A and A H B do not differ too
TABLE II Thermodynamic parameters obtained from the deconvolution of calorimetric curves
DSC data for papain (pH 3.80) and chymopapain (pH 2.90) were fitted to an unfolding mechanism that considers two interacting cooperative units. SDF, standard deviation of fitting. Transition 1
Papain Chymopapain Chymopapain a a
Tin,A (°C) 75.5 ± 0.3 71.1 + 0.3 73.1 + 0.2
Transition 2
Interaction term
AHA
Tm.a
AH a
AG~,
(kcal/mol) 97 ± 7 67 ± 6 88 ± 8
(°C) 78.9 ± 0.2 77.8 ± 0.1 77.8 _+0.2
(kcal/mol) 126 ± 9 134 ± 8 112 ± 7
(kcal/mol) 0.005 ± 0.04 0.97 + 0.04 1.15 ± 0.05
In this case, a restriction was imposed to force the ratio of enthalpies to a value similar to that of papain.
SDF (cal/mol per K) 923 1053 1228
124 much (AHA//tH B = 1.29), as would be expected for a protein with domains of similar size [6]. In the case of chymopapain, the difference between AHA and AHB was considerably larger (ZaHA/AH8 = 1.98) and the interaction term acquired a significant value of 0.97 kcal/mol. However, given the high sequence identity and nearly equal molecular weight of the two enzymes, it is reasonable to assume that the molecular structure of chymopapain is formed by domains of a size akin to those of papain. Therefore, we further analyzed the chymopapain curve, introducing a restriction that limited the largest transition enthalpy to a value not larger than 1.29-times the value of the smallest enthalpy. Results obtained with this restriction are shown in the bottom row of Table II. The standard deviation of fitting was again less than 5% of the maximum value of the curve, indicating a satisfactory agreement between calculated and experimental data. With this approach, ziG, is slightly larger than in the previous analysis.
Temperature dependence of the unfolding enthalpy Fig. 3 illustrates the influence of temperature on the specific unfolding enthalpy, AHca1, of the two proteins. For comparison, data reported by Tiktopulo and Privalov [11] are also shown. As can be seen, there is a good linear relationship between AHcaI and the temperature, in concordance with results for several other small proteins [1,3]. Indeed, it is generally accepted that in a broad temperature range the enthalpy of unfolding is a linear function of temperature with a slope equal to the heat capacity difference between unfolded and native states [1,25]. From Fig. 3, heat capacity differences of 0.096 c a l / K per g (2.3 kcal/K
2520o
15I0_
¢J
5-
%
O-
I
65
I
?(3
I
75 Temperature
[
80 (*C)
I
8,5
Fig. 2. Excess heat capacity functions for chymopapain at pH 2.90 ( ) and papain at pH 3.80 ( . . . . . . ). Fitting the data to an unfolding mechanism that considers two interacting domains gave values of Cex indicated as triangles (chymopapain) and circles (papain).
IO 0
o
<:1 6
4 30
40
50
60
70
80
9(3
I00
Temperoture (*C)
Fig. 3. Temperature dependence of the enthalpy change (per gram of protein) for chymopapain ( A ) and papain (11). For papain, data reported by Tiktopulo and Privalov [11] are also shown ( [] ).
per mol) and 0.142 c a l / K per g (3.3 kcal/K per mol) were calculated for chymopapain and papain, respectively. These values agree well with those determined directly from the original calorimetric curves (cf., Fig. 1) which were 2.1 + 0.5 kcal/K per mol (chymopapain) and 3.0 + 0.4 kcal/K per real (papain). These latter determinations were done by linearly extrapolating the pre- and postdenaturational parts of each calorimetric recording and evaluating the difference between them at Tm [1,3]. Discussion
The results of this study showed that the unfolding transition of chymopapain is not a two-state process, as indicated by the ratio of calorimetric to van't Hoff enthalpies. In this respect, the unfolding behavior of chymopapain resembles that of papain, although a difference in the cooperativity of these processes is clearly noticed (Table I). Apparently, the domains of the papain molecule unfold in an independent way which leads to a value of AHcal/ZlHvH close to 2.0, as previously reported by Tiktopulo and Privalov [11]. The analysis of calorimetric data, based on a mechanism that includes domain-domain interactions [8,24], yielded results in agreement with this proposal since the interaction free energy, AG,, is negligible. For chymopapain, in contrast, the value of ziG, obtained from the same type of analysis was significantly larger (approx. 1.0 kcal/mol). It must be noted that AG, had little variation when the ratio of the two domain-unfolding enthalpies was restricted to a value similar to that observed for papain denaturation (Table II). A more complex method of analysis (i.e., one including different ch terms for the unfolding of different domains) would probably provide more information on the cooperative behavior of these macromolecules during denaturation. However, before attempting an anal-
125 ysis of this type it would be convenient to have a better understanding of the irreversible nature of these phenomena. In any case, calorimetric data of chymopapain seem to be consistent with a bilobal molecular structure like that of papain yet with stronger domain-domain interactions. Furthermore, it is possible that these different interdomain contacts may be the cause of small variations in the folding of the polypeptide chain, thus explaining the dissimilar circular dichroism spectra these proteins show [17]. Further calorimetric studies with other cysteine proteinases of known three-dimensional structure will probably help to clarify this point. In that way, a detailed correlation between molecular structure and unfolding behavior could be established for the members of this family, as it has been recently done with some other proteins [9,26]. Another point that deserves comment is the different heat capacity change associated with the unfolding of papain and chymopapain. This suggests that these macromolecules possess unlike hydrophobic cores, since it is considered that the value of ACp is, to a first approximation, proportional to the nonpolar area buried in the native structure of a protein [1,25]. In addition to its important connection with the hydrophobic effect, ACp is a fundamental thermodynamic parameter that determines the temperature dependence of entropy and enthalpy changes. Within a relatively wide temperature range ACp can be regarded as a constant [25] and, as a consequence, the unfolding enthalpy follows a linear dependence on temperature which can be written as AH = AH*
+ ACp(T - T~)
(9)
where T~ is a reference temperature at which AH = A H * . The constant T~ has a special meaning because the enthalpy changes (expressed on a per gram or per residue basis) for a number of proteins extrapolate to an approximately common value at this temperature, as first noted by Privalov and Khechinashvili [3]. It has been proposed that at the reference temperature the contribution of the hydrophobic effect to AH is zero, i.e., at this temperature the unfolding enthalpy is only due to polar interactions (hydrogen bonds and polar van der Waals contacts) [1,9]. Therefore, proteins having equal numbers of polar interactions per gram should show the same enthalpy value (AH *) when T = Try, as Eqn. 9 indicates. However, the precise value of Tr~ has been a matter of controversy. From the analysis of thermodynamic properties of proteins, different authors have assigned values of 112 [27], 100 [9] or 84°C [28] to TI~. Nevertheless, in this work we found that the specific unfolding enthalpies of chymopapain and papain coincide approximately at 64°C (Fig. 3), a temperature lower than any of the values of T~ indicated above. This finding suggests that members of a protein
family may have thermodynamic properties that distinguish them from other proteins. In other words, it is possible that the values of TI~ and AH* may vary from family to family in response to the particular network of hydrogen bonds and hydrophobic interactions that characterize each type of structural folding pattern.
Acknowledgements This work was supported in part by the Consejo Nacional de Ciencia y Tecnologia, Mdxico (Convenio No. 0609-N9110).
References 1 Privalov, P.L. (1979) Adv. Protein Chem. 33, 167-241. 2 Kim, P.S. and Baldwin, R.L. (1982) Annu. Rev. Biochem. 51, 459-489. 3 Privalov, P.L. and Kechinashvili, N.N. (1974) J. Mol. Biol. 86, 665-684. 4 Sturtevant, J.M. (1987) Annu. Rev. Phys. Chem. 38, 463-488. 5 Cantor, C.R. and Schimmel, P.R. (1980) Biophysical Chemistry, part III, pp. 1075-1088, W.H. Freeman, New York. 6 Privalov, P.L. (1982) Adv. Protein Chem. 35, 1-104. 7 Hu, C.Q. and Sturtevant, J.M. (1987) Biochemistry 26, 178-182. 8 Brandts, J.F., Hu, C.Q., Lin, L.N. and Mas, M.T. (1989) Biochemistry 28, 8588-8596. 9 Murphy, K.P. and Freire, E. (1992) Adv. Protein Chem. 43, 313-361. 10 Kamphuis, I.G., Kalk, K.H., Swarte, M.B.A. and Drenth, J. (1984) J. Mol. Biol. 179, 233-256. 11 Tiktopulo, E.I. and Privalov, P.L. (1978) FEBS Lett. 91, 57-58. 12 Hermlndez-Arana, A. and Soriano-Garcla, M. (1988) Biochim. Biophys. Acta 954, 170-175. 13 Baker, E.N. and Dodson, E.J. (1980) Acta Cryst. A36, 559-572. 14 Pickersgil, R.W., Rizkallah, P., Harris, G.W. and Goodenough, P.W. (1991) Acta Cryst. B47, 766-771. 15 Baker, E.N. and Drenth, J. (1987) in Biological Macromolecules and Assemblies, Vol. 3 (Jurnak, F.A. and McPherson, A., eds.), pp. 313-368, Wiley, New York. 16 Jacquet, A., Kleinschmidt, T., Schnek, A.G., Looze, Y. and Braunitzer, G. (1989) Biol. Chem. Hoppe-Seyler 370, 425-434. 17 Solls-Mendiola, S., Arroyo-Reyna, A. and Hern~ndez-Arana, A. (1992) Biochim. Biophys. Acta 1118, 288-292. 18 Solis-Mendiola, S., Zubillaga-Luna, R., Rojo-Domlnguez, A. and Hermlndez-Arana, A. (1989) Biochem. J. 257, 183-186. 19 Glazer, A.N. and Smith, E.L. (1971) in the Enzymes, 3rd Edn. (Boyer, P.D., ed.), pp. 502-545, Academic Press, New York. 20 Robinson, G.W. (1975) Biochemistry 14, 3695-3700. 21 Goodenough, P.W. and Owen, J. (1987) Phytochemistry 26, 75-79. 22 Bevington, P.R. (1969) Data Reduction and Error Analysis for the Physical Sciences, pp. 204-246, McGraw-Hill, New York. 23 Freire, E., van Osdol, W.W., Mayorga, O.L. and S~nchez-Rulz, J.M. (1990) Annu. Rev. Biophys. Chem. 19, 159-188. 24 Ramsay, G. and Freire, E. (1990) Biochemistry 29, 8677-8683. 25 Privalov, P.L. and Gill, S.J. (1988) Adv. Protein Chem. 39, 191234. 26 Murphy, K.P., Bhakuni, V., Xie, D. and Freire, E. (1992) J. Mol. Biol. 227, 293-306. 27 Murphy, K.P., Privalov, P.L. and Gill, S.J. (1990) Science 247, 559-561 28 Spolar, R.S., Livingstone, J.R. and Record, M.T., Jr. (1992) Biochemistry 31, 3947-3955.