161 Acta Chim. Slov. 1998, 45(2), pp. 161-172 (Received 29.5.1998) Stability and folding studies of two homologous proteins, human stefins A and B. Eva Žerovnik1, Roman Jerala2 , Louise Kroon-Žitko1 and Vito Turk1 1 Department of Biochemistry and Molecular Biology, Jo Def Stefan Institute, Jamova 39, 1000 Ljubljana, Slovenia 2 Laboratory for Molecular Modeling and NMR Spectroscopy, National Institute of Chemistry, Hajdrihova 19, 1000 Ljubljana, Slovenia Dedicated to the memory of Professor Savo Lapanje Abstract: In a shorter Introduction we describe the usual experimental means of how to study protein stability and folding. Our work on stability of human stefins A and B, determined by chemical, heat and pH denaturation, is described next. We explain the folding mechanism of human stefin B (as determined thus far) and compare it to homologous human stefin A, in particular, the dependence of folding rates on the concentration of GuHCl. pH denaturation of human stefin B ( E.Zerovnik et al., Eur. J. Biochem. 1997, 245, 364-372 ) and its folding to the native state are described in more detail. Introduction Denaturation [1,2] studies allow the measurement of some basic thermodynamic properties of proteins, which also can be calculated from coordinates of the 3D structures [3]. The mechanism of folding is more difficult to determine as there may be many routes and many intermediate states leading to the native state. The major drive to the native state stems from thermodynamics even though, there are cases where kinetic barriers route proteins into trapped intermediates or meta-stable native states [4]. It is now apparent that 3-state folding with a sequentially placed intermediate on the route from unfolded to the native state is no more an exception [5-7]. The "old" and "new" 162 views of protein folding [8,9] differ mainly in the way they describe this complex reaction. The old view stresses common conformations - states, whereas the new view points to each separate conformation [9]. Experimentalists collect data on thermodynamics and kinetics of protein folding. One usually starts by determining the stability from heat, pH and chemical denaturation. If multiple spectroscopic probes give coincident transitions this means that there are no populated intermediates. 2-state transition is also revealed by identity of Van't Hoff and calorimetrie enthalpies which are obtained by differential scanning calorimetry (DSC). To determine the kinetics of folding and unfolding, lots of data have to be collected under different experimental conditions. The changes should be studied by different spectroscopic probes which monitor different structural aspects of the process. Rate constants should be determined as a function of dénaturant concentration, temperature, pH and selected mutations. Even though the kinetic data can be fit to plausible models these still need to be confirmed. One useful experiment to distinguish between intermediate and native states is measuring unfolding after various times of refolding [10]. New biophysical techniques [11] permit results of high quality and time resolution. For example, advances have been made in multidimensional NMR, where the difficult task of investigating the denatured and partially folded states is achievable [12, 13]. To see which h-bonded structures of protein become protected during folding, quenching of hydrogen to deuterium exchange is followed by NMR or MS measurements. The changes in compactness during folding are getting measured using small angle X-ray scattering [14 ]. Human stefins A and B were chosen for studies of protein stability and folding due to their relative simplicity : small Mr and no disulphide bonds. Both posess a number of prolines which are in the trans conformation. Sequence similarity of stefins A and B amounts to 56 % and they fold in a very similar 3D conformations (less than 1 Â 163 deviation of the backbone C-atoms). Despite these similarities they exhibit marked différencies in stability and mechanism of folding. We have determined stability and kinetics of folding using several spectroscopic probes, and have measured the kinetics by fluorescence as a function of temperature, pH and GuHCl concentration. In order to determine the rate-limiting step, the folding was started from various denatured states. Denaturation and folding studies of human stefins A and B The two proteins were first isolated from human sources [15-17]. Later, many homologues were identified in the animal kingdom [18-20] and in plants [21]. In Table 1, some representative aminoacid sequences are aligned. Human stefins A and B are members of the cystatin superfamily of cysteine proteinase inhibitors [21,22]. They have been cloned and expressed in E.coli [23,24] and the yield of expression subsequently improved [25]. The 3 D structures of each are determined by X-ray crystallography (human stefin B in complex with papain [26 ] and by solution NMR (human stefin A [27, 28]). Initial study was done on folding of human stefin A [29]. GuHCl denaturation [29] , monitored by near UV CD, far UV CD and tyrosine fluorescence, with a midpoint at 2.8 ± 0.1 M GuHCl (25°C, pH 8), showed a reversible, two state unfolding transition. The kinetics of slow folding was measured (10°C, pH 8) by the same three probes. In 2.0 M GuHCl, the kinetics were rather slow ( k = 0.002 s"1 by all three probes). The amplitudes of the observable phases were 90±10 % by tyrosine fluorescence and near UV CD, but only 20% by the far UV CD. This can be interpreted as kinetically 2-state folding from an initial intermediate possessing 80% of the signal in the far UV CD. "Double jump" experiments were performed and have shown that the slow changes in tyrosine fluorescence reach their complete amplitude only after longer times of unfolding. Except for proline isomerization this might be due to slow unfolding of the core. 164 In subsequent years, we have focused on recombinant human stefin B. Its denaturation by heat, acid and GuHCl, monitored by fluorescence, near UV and far UV CD [30], were found noncoincident which was explained by the presence of a molten globule -type [31-34] of intermediate [30, 35]. In contrast to stefin A which is thermally extremely stable, the behavior of stefin B is quite different. Differential scanning calorimetry (DSC) was used to obtain thermodynamic data from the thermal unfolding of both proteins [36] as well as some cystatins [37]. The reasons for the big difference in thermal stability between stefins A and B were sought using structural thermodynamics [38]. Reversible DSC scans of stefin A could be analysed in terms of a 2-state transition whereas stefin B unfolded in an irreversible transition ( at pH 8 and 6.5, 0.13 M salt). The asimetry of the DSC peaks was characteristic of unfolding of a dimer. At pH 5, the transition of stefin B became reversible and permitted thermodynamic analysis [36]. Table 1 Sequence alignment of some representative stefins; the N-terminal extension is omitted from pig leukocyte CPI (5 residues) [20]. The bold letters stress identity to human stefin B, underlined are residues which are the same to human stefin A. Recombinant human stefin B differs in the third residue (Ser3) from the wild type Cys3. resNo 10 20 30 40 50 human r. stef.B MMSGAPSATQ PATAETQHIA DQVRSQLEEK YNKKFPVFKA VSFKSQWAG bovine stefm B44 MMCGGTSATQ PATAETQAIA DKVKSQLEEK ENKKFPVFKA LEFKSQLVAG human stefin A MIPDDLSEAK PATPEIQEIV DKVKPQLEEK TNETYGKLEA VQYKTQWAG bovine stefin A19 MIPGGLTEAK PATIE^QEIA NMVKPQLEEK TNETYEEFTA IEYKSQWAG pig leuko. CPI20 MLAGGLTEPR PATPE^QEIA NKVKPQLEEK TNKTYEKFEA IIYRSQWAG res No 60 7 0 8 0 90 human r.stef.B TNYFIKVHVG DEDFVHLRVF QSLPHENKPL TLSNYQTNKA KHDELTYF bovine stefm B44 KNYFIKVQVD EDDFVHIRVF ESLPHENKPV ALTSYQTNKG RHDELTYF human stefin A TNYYIKVRAG DNKYMHLKVF KSLPGQNEDL VLTGYQVDKN KDDELTGF bovine stefin A19 INYYIKIQTG DNRYIHIKVF KSLPQQSHSL ILTGYQVDKT KDDELAGF pig leuko. CPI20 TNYYIKVHVG GNNYVHIRVF QSLPHQEDPL KLIGYQVDKT KDDELTGF pH denaturation of human stefin B was studied in more detail by spectroscopic and DSC techniques [39]. The pH titration was performed at "low" salt (buffers described in [39], 0.033 M NaCl) or at "high" salt (the same buffers [39], 0.42 M NaCl). Near and far UV CD spectra were recorded as a function of pH. Some representative far UV CD spectra are shown in Figi. The pH denaturation was recorded at 200, 208, 222 and 277 nm, and 165 showed two transitions, at pH 4.1 and 5.7. At pH 5.7 only minor changes at 200, 208 and 222 nm are observable (due to minor changes in secondary structure) which is supported by recent NMR experiments on N15 labeled stefin B. At pH 4.1, rather large and cooperative changes occur at all wavelengths. The final, acid intermediate (denatured) state li (below pH 3.3) has some properties of a "less structured" molten globule [33], in particular, in binding ANS. Adding salt to li produces a transition to a "more structured" molten globule h. All CD and fluorescence data were acquired at 18°C as the population of intermediate states changes with temperature. Instead of 190 200 210 220 230 240 250 nm Figure 1. Far UV CD spectra of human stefin B in various denatured and intermediate states: N (0.01 M phosphate buffer pH 6.5), I2 ( 0.01 glycine 0.25 M sulphate, pH 3.3), i! (0.01 M glycine , pH 3.3); from top to the bottom at 200 nm. studying pH denaturation at several temperatures, we have heated samples at various pH values. DSC data [39] have shown that thermal denaturation deviates from 2-state with decreasing pH which was explained by a rather low enthalpy of unfolding of the acid intermediate state li. As this state is populated to 100 % only below pH 3.3 where 166 enthalpy could not be measured due to the low melting temperature, no firm conclusion on the enthalpy of unfolding of state li could be made. In contrast, the second, more structured acid intermediate h demonstrates a measurable enthalpy of unfolding. lD NMR spectra [39] of the acid induced intermediates of human stefin B have been measured and have confirmed that h posesses some tertiary structure (high field methyl resonances) in contrast to li. It is hoped that by using N and C labelled samples and acquiring 2D and 3D NMR spectra much more detail would be obtained also of the other two acid intermediate states li and h. (R.J., experiments underway). The compactness of the acid intermediate states and their oligomeric nature was probed by SEC [40]. GuHCIdepof Ink 5 &~^ * $ * 3 ^*»^. «^*^ stB 1 -1 -3 c \ stA -O ( D 1 2 3 GuHCI Figure 2. Logarithm of the rate constants against GuHCI molarity. Folding of human stefin B was started from the acid intermediate state at pH 2.7. This was found the same as when started from the GuHCI denatured state at 3.45 M GuHCI. (not shown). Unfolding was started from the native state at pH 6. Folding of human stefin A was started from the GuHCI denatured state (4.4 M GuHCI). The initial values at zero GuHCI have been obtained by folding the acid denatured state to zero GuHCI at otherwise the same conditions (pH 6, 20°C). 167 The folding mechanism of stefin B has been studied [41,42] and compared to that of stefin A. We also have studied the influence of pH and TFE on the folding reaction[41]. Using several spectroscopic probes (CD in the near, CD in the far, tyrosine and ANS fluorescence - Table 2), it was found that the kinetics of folding of human stefin B differ from those of stefin A. Not only is the major phase of folding 40-times faster (at zero GuHCl, - Fig.2), this protein demonstrates an additional slow phase. The slow phase of 25 % amplitude (Table 2) could be ascribed to proline isomerization but pH dependence of the amplitude and double jump experiments do not prove it univocally. An intermediate seems a more likely hypothesis. In the very initial stages of folding, a "burst" phase of the signal at 230 nm was observed [41]. Its amplitude was 60% ± 5% of the total change. In a second paper [42], folding was started from different initial states : GuHCl unfolded, TFE denatured, acid denatured and acid intermediate state. The folding rates were found to be the same, so that the rate limiting step must also be the same. In Fig.3, folding from acid denatured state Ad is shown as monitored by ANS fluorescence. ANS binding on folding of stefin B and many other proteins is very informative. The initial increase can be explained as compactization [32,34] which is supported by X-ray scattering [43]. The decrease of ANS fluorescence occurs along with appearance of the native-like states (Fig.3 B,C). To conclude : Many more experiments still have to be done to be able to understand the folding of these two homologous proteins. We hope to study their mutants and to follow dimerization reactions which seem connected to stefin B folding. In cooperation with other laboratories it will be possible to determine the rate of compactization (time-resolved X-ray scattering [43] and details of structure formation (real time NMR [45]). 168 Table 2 Comparison of the folding kinetics of human stefins A and B as measured by the near and far UV CD, tyrosine and ANS fluorescence. Data are taken from [41]. The folding of stefin B was started by 6-times dilution from 3.45 M GuHCl and of stefin A from 4.4 M GuHCl, both, to pH 6. The measurements were done at 20 °C. Rate constants are in s"1 and amplitudes as % of the observed change. "Burst phase" amplitudes in the far UV CD are 60% and 45 % of the total change, respectively, for stefins B and A. No amplitude is given for ANS fluorescence, only the sign. human stefin B folding probe ki ' ki/Am Tyrfl. 14.2/77% ANS fi. 86(+) 12.8Q CD far 10.5/76% CD near 10.8/75% 0.34(+) k?/Am 0.05/23% 0.022Q 0.045/24% 0.039/25% human stefin A folding probe k/Am Tyrfl. 0.18/100% ANSA. no change CD far 0.18/100% CD near 0.16/100% Acknowledgments We thank the Ministry for Science and Technology of the Republic Slovenia for the financial support throughout this work and for the projects N° (Jl-8742-0106-97) and N° (Jl-8766-97). Grants obtained from the British Council and the Austrian Federal Ministry of Science and Research are acknowledged. We appreciate many fruitful discussions with our colleges Dr. K.Lohner (Institute of Biophysics and X-Ray structure Research, Austrian Academy of Sciences, Graz), Dr. J.P. Waltho (Department of Molecular Biology and Biotechnology, University of Sheffield). Special thanks go to Dr. R.Virden (University of Newcastle upon Tyne) and Dr.M.Carey (Applied Photophysics, U.K.) for the stopped-flow CD measurements. Thanks are due to Professor R.H.Pain for reading the manuscript. 169 Siimi Tin {seconds) 100 îlîij m 4M 500 G.CÎO fi.niîu ifl^wM^ Normalised 41 ar Lance 7 .^E-5 D-îoblv fiprjnpntidJ. Ha^llit? mil Hint Figure 3. Kinetics of folding of human ste fin B recorded by ANS fluorescence. The folding was started from state AD at pH 1.8 ± 0.2 (dialysis against water / HCl pH 1.8) and proceeded into 0.02 M phosphate buffer. After 6-fold dilution the final pH was 6.0 ±0.1. A) the initial ANS fluorescence increase - folding to an intermediate ( k"= 533, k'=92 s"1) B, C) folding to the native state is tri-phasic ( ki = 11 s"1 , k2 = 0.053 s"1 and k3 = 0.004 s"1) 170 Povzetek : Najprej je podan kratek uvod v eksperimentalni del študija stabilnosti in foldinga proteinov. Nato opisujemo naše delo s človeškima stefmoma A in B. Kinetiko foldinga stefina B smo merili z več probami in jo primerjali s kinetiko foldinga stefina A. Merili smo tudi odvisnost hitrosti od koncentracije GuHCl, pH-ja in temperature. pH denaturacija človeškega stefina B (E.Žerovnik et al., Eur.J.Biochem. 1997, 245, 364 - 372) in folding do nativnega stanja iz kislo denaturiranega, sta opisana nekoliko bolj natančno. References [I] C.Tanford, Adv.Prot.Chem. 1968, 23, 121 - . [2] S!apanje,Physicochemical Aspects of Protein Denaturation , Wiley-Interscience, N.York, 1973 [3] K.P. Murphy, E.Freire, Adv.Prot.Chem. 1992, 43, 313 - 361. [4] T.E. Creighton , Nature 1992, 356, 194 - 195. [5] S. Khorasanizadeh, I.D. Peters, H. Roder, Nat.Struct.Biol.1996, 3, 193 - 204. [6] D.K. Heidary, L.A. Gross, M. Roy, P.A. Jennings, Nature Struct.Biol. 1997, 4, 1-10. [7] R.L. Baldwin , Fold.&Design 1996, 1, Rl -R8. [8] K.A. Dill, H.S. Chan, Nat.Struct.Biol. 1997, 4, 10-19. [9] H.S.Chan, K.A.Dill, Proteins 1997, 8, 2-33. [10] F.X. Schmid, In : Protein Folding 1992, T.E. Creighton, ed., Freeman & Co., pp. 197 - 238. [II] K.W. Plaxco, CM. Dobson , Curr.Opin.Struct.Biol. 1996, 6, 630-636. [12] D. Eliezer, J. Yao, J. Dyson, P.E. Wright, Nat.Struct.Biol. 1998, 5, 148 - 155. [13] P. Fan , C. Bracken, J. Baum, Biochemistry 1993, 32, 1573 - 1582. [14] J. Trewhella, Curr.Opin.Struct.Biol. 1997, 7,702 - 708. [15] J.Brzin, M.Kopitar, V.Turk, W. Machleidt, Hoppe Seyler's Z. Physiol.Chem.1983, 364, 1475 - 1480. [ 16] W.Machleidt, U. Borchart, H. Fritz, J.Brzin, A.Ritonj a, V.Turk, Hoppe Seyler's Z.Physiol.Chem. 1983, 364, 1481 - 1486. [17] A.Ritonja, W.Machleidt, A.J.Barrett, Biochim.Biophys.Res.Commun.1985, 131, 1187 - 1192. [18] B.Turk, I.Križaj, B.Kralj, I.Dolenc, T.Popović, J.G. Bieth, V.Turk, J.Biol.Chem. 1993, 268, 7323 - 7329. [19] B.Turk, A.Ritonja, I.Bjork, V.Stoka, I.Dolenc, V.Turk, FEBS Lett. 1995, 101 - 105. [20] B.Lenarčič, A.Ritonja, I.Dolenc, V.Stoka, S.Berbić, J. Pungerčar, B.Štrukelj, V.Turk, FEBS Lett. 1993, 336, 289 - 292. [21] W.M. Brown, K.M. Dziegielewska, Protein Sci 1997, 6, 5 - 12. [22] V.Turk, W.Bode, FEBS Lett. 1991, 285, 213 - 219. [23] R.Jerala, M.Trstenjak, B.Lenarčič, V.Turk, FEBS Lett. 1988, 239, 41 - 44. [24] M. Strauss, J.Stollwerk, B.Lenarčič, V.Turk, K.D.Jany, H.G. Gassen , Hoppe Seyler 1988, 369, 1019 - 1030. [25] RJerala, L.Kroon-Žitko, V.Turk, Protein Expr.Purif. 1994, 5, 65 - 69. [26] M.T.Stubbs, B.Laber , W. Bode, R. Huber, R. Jerala, B. Lenarčič, V.Turk, EMBO J.1990, 9, 1939 - 1947. [27] J.R.Martin, C.J.Craven, R.Jerala, L.Kroon-Žitko, E.Žerovnik, V.Turk, J.P. Waltho, J.Mol.Biol.1995, 246, 331 - 343. [28] J.R.Martin, R.Jerala, L.Kroon-Žitko, E.Žerovnik, V.Turk, J.P. Waltho, Eur.J.Biochem. 1994, 225, 1181-1194. [29] E. Žerovnik, B. Lenarčič, R.Jerala, V.Turk , Biochim.Biophys.Acta 1991, 1078, 313 - 320. [30] E.Žerovnik, R.Jerala, L.Kroon-Žitko, R.H.Pain, V.Turk, J.Biol.Chem. 1992, 267, 9041 - 9046. [31] K.Kuwajima, Proteins : Struct.Fun.& Gen. 1989, 6, 87 - 103. [32] O.B. Ptitsyn, R.H. Pain, G.V. Semisotnov, E. Žerovnik, O.I.Razgulyaev, FEBS Lett. 1990, 262, 20 - 24. [33] O.B. Ptitsyn (1992). In Protein folding, Creighton T.E., Ed., Freeman & Co., New York, pp.243 - 300. 171 [34] H. Christensen, R.H. Pain, In Mechanisms of Protein Folding, R.H Pain., ed., Oxford Univ.Press ,1994, pp. 55-76. [35] E.Žerovnik, R.Jerala, L.Kroon-Žitko, V.Turk, R.H.Pain, Biol.Chem. Hoppe-Seyler 992, 373, 453 - 458. [36] E.Žerovnik, K.Lohner, R.Jerala, P.Laggner, V.Turk, Eur.J.Biochem. 992, 210, 217 - 221. [37] E.Žerovnik, N.Cimerman, J.Kos, V.Turk, K.Lohner, Biol.Chem. 997, 378, 1199 - 1203. [38] RJerala, E.Žerovnik, K.Lohner, V.Turk, Prot.Engineering 994, 7, 977 - 984. [39] E.Žerovnik, R.Jerala, L.Kroon-Žitko, V.Turk, K.Lohner, Eur.J.Biochem997, 245, 364 - 372. [40] E.Žerovnik, R.Jerala, N.Poklar, L.Kroon-Žitko, V.Turk, Biochim.Biophys.Acta 994, 1209, 140 - 143. [41 ] E.Žerovnik, R.Virden, R.Jerala, V.Turk, J.P.Waltho, Proteins: Struct.Fun.&Gen 998, in press [42] E.Žerovnik,R.Jerala, R.Virden, L. Kroon-Žitko, V. Turk, J.P.Waltho, Proteins: Struct.Fun.&Gen. 998, in press [43] L. Chen, G. Wildegger, T. Kiefhaber, K.O. Hodgson, S.Doniach, JJVlol.Biol. 998, 276, 225 - 237. [44] LKrižaj, B.Turk, V.Turk, FEBS Lett.992, 298, 237-239. [45] J.Balbach, V.Forge, W.S. Lau, N.A. van Nuland, K.Brew, C.M.Dobson, Science 996, 274, 1161 -1163..