PROTEINS: Structure, Function, and Genetics 30:435–441 (1998) Proteolysis as a Probe of Thermal Unfolding of Cytochrome C Leyu Wang, Robert X. Chen, and Neville R. Kallenbach* Department of Chemistry, New York University, New York, New York ABSTRACT Recent hydrogen exchange experiments on native cytochrome c implicate a sequential unfolding pathway in contrast to a simple two-state process. We have studied the heat-induced unfolding of this protein by using spectroscopic measurements to detect changes in conformation and proteolytic enzyme digestion to identify regions of the protein that are labile. Several spectroscopic profiles were monitored: CD at 222 nm, a measurement of secondary structure change in the protein, the absorbance at 280 nm, involving the local environment of Trp 59, and absorbance at 420 nm, the Soret band of the heme. The apparent Tm values for these probes differ, consistent with an unfolding pathway containing intermediates. The limited digestion by proteinase K is consistent with population of an intermediate state in unfolding. We find a single strong region of cleavage at low temperature with retention of structure in each fragment. Proteins 30:435– 441, 1998. r 1998 Wiley-Liss, Inc. Key words: cytochrome c; thermal unfolding; proteolysis; proteinase K; thermolysin INTRODUCTION Many single-domain proteins can spontaneously refold to their native active states after unfolding in dilute solution. The mechanism of this reaction remains a matter of considerable debate, although there is an emerging consensus that globular proteins above a critical size fold via formation of intermediate structured subdomains or specific nucleation regions.1–3 Recent theoretical studies by Srinivasan and Rose4 argue that important aspects of folding can be modeled as a hierarchical process, in which stable intermediates accumulate. Others have proposed formation of a framework of sequential intermediates,2 based on various hypothetical subunits and mechanisms for interaction among subunits.5 What remains unclear at present is the detailed mechanism of folding and the nature of the barriers separating intermediates.6 The hypothesis that folding proceeds via a series of distinct intermediates is strongly supported by the recent analysis of peptide hydrogen exchange in horse heart cytochrome c (cyt c), which reveals a r 1998 WILEY-LISS, INC. specific set of subdomains that unfold (or refold) cooperatively within the native structural manifold.7 The intermediates in cyt c are identified as clusters or groups of peptide NH that share a common opening channel; four such sets have been identified in the protein to date, each associated with a particular free energy and sensitivity to denaturing solvent. Within each subdomain, clusters of NH groups exchange initially via a local process, which exhibits minimal sensitivity to GuHCl. Others in the cluster exchange only when released by a larger scale cooperative unfolding reaction, involving perhaps a dozen NH groups or more. Because the pathway(s) by which NH groups exchange from proteins is not yet elucidated in detail, the issue we address here is whether subdomains can be detected directly, by using reactive probes of the backbone as an alternative to NH exchange. In a sense, NH exchange represents a reactive probe of the peptide backbone that perturbs the bond minimally, substituting D for H or vice versa. Scission by a protease is a more severely perturbing probe but one that allows facile detection of reactive sites. The native states of proteins tend to resist cleavage by proteases. Limited proteolysis has been used to probe the surface accessibility of particular side chains and the flexural dynamics of proteins.8–10 Proteinase K, an extremely stable protease with a relatively broad cutting specificity,11 provides a probe for conformational flexibility in proteins that is able to monitor structural changes during protein unfolding. An enzyme, such as trypsin, on the other hand, cleaves charged groups most likely to occur on the protein surface and would thus be less sensitive as a probe of local unfolding reactions. Cyt c is a small stable globular protein that has served as a model for numerous folding and unfolding studies.12 Horse heart cyt c contains 104 amino acids, covalently linked to a heme group (Fig. 1). It acts as a soluble electron carrier in oxidative phosphorylation, transferring electrons from cytochrome reductase to the terminal oxidase complex. Recent Contract grant sponsor: National Institutes of Health; Contract grant number: GM40746. *Correspondence to: Neville R. Kallenbach, Department of Chemistry, New York University, New York, NY 10003. E-mail: firstname.lastname@example.org Received 29 August 1997; Accepted 4 November 1997 436 L. WANG ET AL. experiments on the solvent-induced unfolding of native cyt c have been interpreted in terms of a sequential, progressive unfolding model, in contrast to a simple all-or-none or fully cooperative unfolding process. Figure 1 illustrates the structure of cyt c in ribbon form, with regions of the protein colored red, yellow, green, and blue to indicate the sequentially unfolding subdomains in the molecule identified by hydrogen exchange rate measurements.7 Determination of the rates of exchange of NH peptide hydrogens of the backbone of cyt c as a function of added denaturant suggests that the protein unfolds in discrete cooperative subdomains, Fig. 1. Structure model of cytochrome c, colored to indicate the subdomains in the molecule that exchange cooperatively with different free energies, according to Bai et al.7 The lowest energy opening set of groups is red, with yellow, green, and blue representing higher energy opening sets of groups, respectively. The blue set of NH groups merge at higher guanidine concentration into an opening corresponding to unfolding of the entire protein. Fig. 2. Schematic illustration of the sequential, cooperative unfolding and refolding reaction identified in cytochrome c, according to Bai et al.7 On the left, the native protein includes NH groups from four distinct regions (represented as blocks) of the molecule shown in Figure 1. The lowest energy opening reaction includes NH groups from the loop colored in red. Once the red tier has exchanged, members of the yellow group unfold, followed by the identified by the different colors shown in Figure 1. As urea or guanidinium chloride is added, the first opening reaction detected occurs in the region colored red, the second is in the domain indicated by yellow, followed by the green and blue regions. Exchange of NH in the latter subdomain corresponds to complete unfolding or denaturation of the protein (see diagram in Fig. 2). A corresponding hierarchy of states can be detected in thermal unfolding of the protein, implying that these subdomains represent intrinsic units in the folding or unfolding of the protein. HD exchange measurements monitored by NMR allow determining the extent of unfolding within each subdomain indirectly, from the protection factors by which exchange of the slowest NH within that subdomain are retarded relative to unstructured NH of the same sequence. The opening reactions responsible for release of peptide NH for solvent exchange have not been characterized; hence, the picture of unfolding that emerges from NH exchange rates does not tell us what fluctuations are involved. The question we address here is, can a stepwise unfolding reaction be monitored more directly by using protease digestion as a reactive probe, for example? Several studies indicate that the accessibility of certain proteases to target sites in a protein is determined by fluctuational opening reactions or local dynamics.8 In this work we describe traditional spectroscopic measurements of thermal unfolding of cyt c as well as monitor the ability of proteinase K and thermolysin to digest cyt c at various temperatures. These enzymes retain activity up to 100°C. The pattern of cutting is consistent with the presence of an intermediate in the opening process as cyt c unfolds, the initial site of attack lying within the yellow subdomain of the molecule (Fig. 2). MATERIALS AND METHODS Proteins and Enzymes Horse heart cyt c was purchased from Sigma (St. Louis, MO) and used without further purification. Proteinase K was purchased from USB. Thermolysin two green domains and finally by the blue group, including the ends of the chain. Once the blue tier has exchanged, the protein is fully unfolded as shown on the right. Refolding is thought to begin with cooperative formation of the blue set, progressing stepwise to green, yellow, and red, generating the native state. The three intermediates in the center of the figure probably correspond to molten globule states of the protein. PROBE OF THERMAL UNFOLDING OF CYT C 437 TABLE I. Molecular Weight of Fragments of Horse Heart Cytochrome c Produced by Limited Proteolysis With Proteinase k as a Function of Temperature in Figure 6 Temperature (°C) On ice Room temperature 30 40 50 60 70 80 90 100 Molecular weight of fragments (kD)† 5.8, 5.4 5.8, 5.4 5.8, 5.4 5.8, 5.4, 4.4, 3.2 5.8, 5.0, 4.4, 3.2 5.8, 5.0, 4.4, 3.2 5.8, 5.0, 4.4, 3.2 9.7, 7.0, 5.8, 5.0, 4.4, 3.9, 3.2, 2.9 Same as 80°C Same as 80°C †The molecular weight of the fragments was got from its mobility on SDS-PAGE compared with the protein markers: insulin (a and b chain, 2,900), bovine trypsin inhibitor (5800), lysozyme (14,600), b-lactoglobin (18,500), carbonic anhydrase (28,860), ovalbumin (44,000). The molecular weight of cytochrome c (12.4 kD) estimated from the gel was 13.4 kD. Fig. 3. SDS-PAGE of the proteolysis of cytochrome c by proteinase K. The protein was treated on ice and at room temperature with proteinase K at 1:50, 1:25, 1:10, 1:5 ratios (w/w) of protease to cytochrome c in 20 mM Tris-HCl buffer, pH 5 7.5. The reaction time is 24 hours, and the reaction was stopped by PMSF. TABLE II. Molecular Weight of Fragments of Horse Heart Cytochrome c Produced by Limited Proteolysis With Thermolysin as a Function of Temperature in Figure 7 Temperature (°C) On ice Room temperature 30 40 50 60 70 80 90 Molecular weight of fragments (kD)† No cut 5.5, 4.3 5.5, 4.3 5.5, 4.3 5.5, 4.3 No significant fragment No significant fragment 9.7, 6.7, 5.5, 4.3, 3.1 Same as 80°C †The molecular weight of cytochrome c (12.4 kD) estimated from the gel was 13.0 kD. was purchased from Sigma. All other reagents were highest purity grade. Limited Proteolysis and Electrophoresis Digestion of cyt c was carried in 20 mM Tris-HCl, pH 5 7.5, with an enzyme/protein ratio at 1:190, 1:50, 1:25, 1:10, and 1:5 for proteinase K, with an enzyme/protein ratio at 1:200 for thermolysin. Before mixing protease with cyt c, the cyt c solution was equilibrated at each temperature in a water bath for 3–5 minutes. The proteolysis reaction was stopped by addition of PMSF for proteinase K with EDTA for thermolysin. SDS-PAGE was used to monitor the digestion. We used the high-resolution tricine buffer electrophoresis system.13 Gels were stained with Coommassie Brilliant Blue R-250. Mixtures of proteins of known molecular mass were used as a calibration standard in SDS-PAGE. Fig. 4. SDS-PAGE of the proteolysis of cyt c by proteinase k at 1:50 enzyme/protein ratio at 37°C. Lane 1: protein markers; lane 2: proteinase k control; lane 3: cyt c control; lane 4: 4 min; lane 5: 8 min; lane 6: 16 min; lane 7: 32 min; lane 8: 1 hour; lane 9: 2 hours; lane 10: 3 hours; lane 11: 4 hours. Transblotting and N-Terminal Sequencing The protein fragments were transferred to PVDF membrane by using the MilliBlot-Graphite Electroblotter System. The membrane then was stained with Coommassie Brilliant Blue R-250. The fragment bands were cut and submitted to a protein-sequencing analyzer for N-terminal sequencing. Thermal Unfolding Measurements Both 20 µM cyt c in 20 mM Tris-HCl buffer (pH 5 7.5) containing 20 mM NaCl were used in melting curve measurement. CD versus temperature was performed at 222 nm on AVIV 60 DS instrument. A cell of 1-mm length was used for the far UV studies, with an equilibration time of 1 minute (1 degree one point). The corresponding absorption versus temperature profiles were moni- 438 L. WANG ET AL. Fig. 5. SDS-PAGE of the proteolysis of cyt c by proteinase k at 1:190 enzyme/protein ratio at 60°C. Lane 1: 32 min; lane 2: 16 min; lane 3: 15 min; lane 4: 10 min; lane 5: 8 min; lane 6: 5 min; lane 7: 4 min; lane 8: 2 min; lane 9: 1 min; lane 10: cyt c control; lane 11: protein markers. Fig. 7. SDS-PAGE of the proteolysis of cyt c by thermolysin at 1:200 enzyme/protein ratio as a function of temperatures. Lane 1: cyt c control; lane 2: cyt c control; lane 3: 40 min on ice; lane 4: 40 min at room temperature; lane 5: 20 min at 30°C; lane 6: 10 min at 40°C; lane 7: 5 min at 50°C; lane 8: 5 min at 60°C; lane 9: 5 min 70°C; lane 10: 5 min at 80°C; lane 11: 5 min at 90°C. ments with those of a set of markers. The apparent molecular weight of protein markers was used to generate a curvilinear line relating log molecular weight of the protein to the migration distance on the separating gel using a least-squares analysis. The molecular weights of the fragments in Figures 6 and 7 are listed in Tables I and II, respectively. According to this method, the apparent molecular weight of cyt c is 13.4 kD in Figure 6 and 13.0 kD in Figure 7, 5–7% higher than its formula molecular weight of 12.4 kD. Limited Proteolysis Fig. 6. SDS-PAGE of the proteolysis of cyt c by proteinase k at 1:50 enzyme/protein ratio as a function of temperatures. Lane 1: 4 min at 90°C; lane 2: 3 min at 80°C; lane 3: 3 min at 70°C; lane 4: 4 min at 100°C; lane 5: 4 min at 60°C; lane 6: 8 min at 50°C; lane 7: 16 min at 40°C; lane 8: 16 min at 30°C; lane 9: 32 min at room temperature; lane 10: 32 min on ice; lane 11: protein markers; lane 13: proteinase k control; lane 14: cyt c control. tored at 280 nm and 420 nm by using a Perkin Elmer 552 UV-VIS spectrophotometer. A 10-mm path cell was used, and the equilibrium time was 2 minutes (2 degree per time point). RESULTS Calibration of Fragment Sizes Estimates of fragment molecular weight were made by comparing the relative mobility of frag- Figure 6 shows the results of cutting by proteinase K as a function of temperature; the sizes of the fragments are listed in Table I. The catalytic activity of proteinase K drops as a function of temperature, so that at 60°C the enzyme is about 80% as active as at 37°C.14 Cyt c becomes increasingly labile to protease as the temperature increases, and above 80°C, the protein can be cut in each of the colored regions shown in Figure 1. We also conclude that the increased scission does not result from a change in activity of proteinase K, but from conformational changes in the protein, by comparing Figures 3, 4, and 5. At temperatures below room temperature, two fragments result from the digestion, regardless of length of exposure to the enzyme. The N-terminal sequence of band II is TDANKNKGITWK, which results from breaking the peptide bond between amino acid Tyr 48 and Thr 49. The N-terminus of band III is ANKGITWKE, which comes from cutting between amino acid Asp 50 and Ala 51. Because the N-terminal of cyt c itself is acetylated, we cannot see A B C D Fig. 8. Unfolding profiles of cytochrome c as a function of increasing temperature. The protein concentration is 20 µM in 20 mM Tris-HCl buffer (pH 5 7.5) containing 20 mM NaCl. A: CD measurement at 222 nm. B: Van’t Hoff plot of data in A: ln Ku as function of 1/T. C. Absorption at 280 nm (triangle) and 420 nm. D. Van’t Hoff plot of data in C: ln Ku as function of 1/T. The data in A and C were initialized by using the corresponding value at 10°C. 440 L. WANG ET AL. the corresponding N-terminal fragments by this sequencing. CD and Absorption Unfolding Profiles The CD profile shown in Figure 8 indicates that cyt c loses its helical structure at pH 7.5 when the temperature exceeds 80°C. The Tm of the protein is close to 83°C measured by [u] 222. By contrast, the values of A280 and A420 show that a conformational change sets in at temperatures as low as 40°C; the Tm detected by A280 is 67.1°C, whereas that for A420 is 68.3°C. The absorption at 280 nm decreases by about 15% from 20°C to 80°C, which indicates there is a significant change in the neighborhood of the chromophore, the side chain of Trp59. Thus, the loop region of the cyt c molecule opens before the helical structure in the molecule unfolds significantly. It is known that a change in heme ligation occurs on thermal unfolding of cyt c, detected by a change in heme absorbance at 695 nm. From the above analysis, we conclude that the thermal unfolding of cyt c at pH 7.5 is not a twostate, or concerted, reaction as it is at pH 5. The loop region of cyt c appears to unfold before the bulk of the helical structure in the molecule. Proteinase K shows a wide temperature range in its activity, remaining active at 100°C. This makes it possible to discriminate structured subdomains as folding and unfolding progresses, even under conditions in which U is stable. Below the major unfolding reaction, specific regions of proteolysis are seen, the fragments showing resistance to further proteolysis. Digestion at elevated temperature reveals several fragments with proteinase K and thermolysins (Figs. 6 and 7). DISCUSSION The thermal unfolding profiles (Fig. 8) show that cyt c does not unfold in a two-state process. Instead, intermediates of some kind must be involved, because the profiles at 222 and 420 nm are not superimposable. Earlier studies had revealed this effect, without fully explaining it. As the protein unfolds, side chains other than the normal ones can interact with the heme,15 and one picture of the unfolding of the protein consistent with the spectroscopic data is that the heme cavity undergoes rearrangements at temperatures below the unfolding of the secondary structure.16,17 The fragmentation data show that proteinase K initially attacks a region in the molecule located in the sequence within the subdomain indicated by yellow in the sketch in Figure 1. The resistance to proteolysis of each resulting fragment suggests that the fragmented molecule retains substantial structure. Subsequent cleavage proceeds from the fragments resulting from this initial event, consistent with a sequential opening mechanism. The scission data are consistent with the results of reconstitution experiments, which indicate that fragments with chain length greater than that of the native protein reassemble to form a native-like core if the cut site lies within the region we have identified.18–20 In addition, Fontana and his co-workers21 have reported that thermolysin cleaves cyt c selectively at the Gly56-Ile57 bond in the presence of TFE. This bond lies within the yellow subdomain, indicating that even in the distorted state induced by TFE, containing additional helical structure, a sequential unfolding mechanism dominates. Do these results support the detailed unfolding pathway delineated in cyt c by HX analysis?7 Initial scission is detected here within the yellow subdomain, rather than the predicted red sequence (70–85 in the sequence, Fig. 2). However, this might reflect a lack of target sequences for the enzyme within the red subdomain relative to yellow. Despite its broad specificity, proteinase K has sequence preferences. Cuts in the red domain certainly are detected in the unfolded protein, however, so the issue is not settled. The red subdomain is considerably smaller in size than yellow. The breakdown of the fragments released following initial cutting is consistent with subsequent opening at sites within the blue and green domains shown in Figure 2, although we have not proven this as yet. Overall, our observations are consistent with a sequential unfolding mechanism, centered on the loop within the subdomain colored yellow, but do not yet establish the process defined by HX experiments in detail. The nature of the breathing reactions that release NH hydrogens for exchange might differ fundamentally from those that expose segments of the backbone to proteolysis in principle. The agreement seen thus suggests a common pathway for at least the cooperatively exposed NH groups in the yellow subdomain. Additional reactive probes might reveal further details of the opening pathway in cyt c. ACKNOWLEDGMENTS L.W. is a recipient of a Henry Mitchell MacCracken Fellowship at NYU. We thank Dr. Martine Cadene for help in N-terminal sequencing at NYU medical center and Angelo Fontana (CRIBI, University of Padua, Padua, Italy) for helpful discussion. REFERENCES 1. Rose, G.D. Hierarchic organization of domains in globular proteins. J. Mol. Biol. 134:447–470, 1979. 2. Kim, P.S., Baldwin, R.L. Intermediates in the folding reactions of small proteins. Annu. Rev. Biochem. 59:631– 660, 1990. 3. Jeng, M.F., Englander, S.W. Stable submolecular folding units in a noncompact form of cytochrome c. J. Mol. Biol. 221:1045–1061, 1991. 4. Srinivasan, R., Rose, G.D. LINUS: A hierarchic procedure to predict the fold of a protein. Proteins 22:81–99, 1995. 5. Karplus, M., Weaver, D.L. Protein-folding dynamics. Nature 260:404–406, 1976. 6. Sosnick, T.R., Mayne, L., Englander, S.W. Molecular collapse: The rate-limiting step in two-state cytochrome c folding. Proteins 24:413–426, 1996. 7. Bai, Y., Sosnick, T.R., Mayne, L., Englander, S.W. Protein PROBE OF THERMAL UNFOLDING OF CYT C 8. 9. 10. 11. 12. 13. 14. folding intermediates: Native-state hydrogen exchange. Science 269:192–197, 1995. Fontana, A., Fassina, G., Vita, C., Dalzoppo, D., Zamai, M., Zambonin, M. Correlation between sites of limited proteolysis and segmental mobility in thermolysis. Biochemistry 25:1847–1851, 1986. Hubbard, S.J., Eisenmenger, F., Thornton, J.M. Modelling studies of the change in conformation required for cleavage of limited proteolysis sites. Protein Sci. 3:757–768, 1994. Cohen, S.L., Ferre-D’Anare, A.R., Burley, S.K., Chait, B.T. Probing the solution structure of the DNA-binding protein Max by a combination of proteolysis and mass spectrometry. Protein Sci. 4:1088–1099, 1995. Lebherz, H.G., Burke, T., Shackelford, J.E., Wilson, K.J. Specific proteolytic modification of creatine kinase isoenzymes. Implication of C-terminal involvement in enzymatic activity but not in subunit-subunit recognition. Biochem. J. 233:51–56, 1986. Kallenbach, N.R. Breathing life into the folding pathway of cytochrome c. Nat. Struct. Biol. 2:813–816, 1995. Schaegger, H., von Jagow, G. Tricine-sodium dodecyl sulfatepolyacrylamide gel electrophoresis for the separation of proteins in the range from 1–100 Kda. Anal. Biochem. 166:368–379, 1987. Bajorath, J., Saenger, W., Papa Pal, G. Autolysis and 15. 16. 17. 18. 19. 20. 21. 441 inhibition of proteinase K, a subtilisin-related serine proteinase isolated from the fungus Tritirachium album Limber. Biochim. Biophys. Acta 954:176–182, 1988. Drew, H.R., Dickerson, R.E. The unfolding of the cytochrome c in methanol and acid. J. Biol. Chem. 253:8420– 8427, 1978. Myer, Y.P., MacDonald, L.H., Verma, B.C., Paude, A. Urea denaturation of horse heart ferricytochrome c. Equilibrium studies and characterization of intermediate forms. Biochemistry 19:199–207, 1980. Brems, D.N., Stellwagen, E. Manipulation of the observed kinetic phases in the refolding of denatured ferricytochromes. J. Biol. Chem. 258:3655–3660, 1983. Hantgan, R.R., Taniuchi, H. Formation of a biologically active, ordered complex from two overlapping fragments of cytochrome c. J. Biol. Chem. 252:1367–1374, 1977. Hantgan, R.R., Taniuchi, H. Conformational dynamics in cytochrome c. A fragment exchange study. J. Biol. Chem. 253:5375–5380, 1978. Juillerat, M., Parr, G.R., Taniuchi, H. A biologically active, three-fragment complex of horse heart cytochrome c. J. Biol. Chem. 255:845–853, 1980. Fontana, A., Zambonin, M., De Filippis, V., Bosco, M., Polverino de Laureto, P. Limited proteolysis of cytochrome c in trifluoroethanol. FEBS Lett. 362:266–270, 1995.