Slow folding of muscle acylphosphatase in the absence of intermediates1

Article No. mb982009

J. Mol. Biol. (1998) 283, 883±891

Slow Folding of Muscle Acylphosphatase in the Absence of Intermediates Nico A. J. van Nuland1, Fabrizio Chiti1, Niccolo' Taddei2 Giovanni Raugei2, Giampietro Ramponi2 and Christopher M. Dobson1* 1

Oxford Centre for Molecular Sciences, New Chemistry Laboratory, University of Oxford, South Parks Road Oxford OX1 3QT, UK

2 Dipartimento di Scienze Biochimiche, Universita' degli Studi di Firenze, Viale Morgagni 50, 50134 Firenze Italy

The folding of a 98 residue protein, muscle acylphosphatase (AcP), has been studied using a variety of techniques including circular dichroism, ¯uorescence and NMR spectroscopy following transfer of chemically denatured protein into refolding conditions. A low-amplitude phase, detected in concurrence with the main kinetic phase, corresponds to the folding of a minor population (13%) of molecules with one or both proline residues in a cis conformation, as shown from the sensitivity of its rate to peptidyl prolyl isomerase. The major phase of folding has the same kinetic characteristics regardless of the technique employed to monitor it. The plots of the natural logarithms of folding and unfolding rate constants versus urea concentration are linear over a broad range of urea concentrations. Moreover, the initial state formed rapidly after the initiation of refolding is highly unstructured, having a similar circular dichroism, intrinsic ¯uorescence and NMR spectrum as the protein denatured at high concentrations of urea. All these results indicate that AcP folds in a two-state manner without the accumulation of intermediates. Despite this, the folding of the protein is extremely slow. The rate constant of the major phase of folding in water, kfH2 O , is 0.23 sÿ1 at 28 C and, at urea concentrations above 1 M, the folding process is slower than the cis-trans proline isomerisation step. The slow refolding of this protein is therefore not the consequence of populated intermediates that can act as kinetic traps, but arises from a large intrinsic barrier in the folding reaction. # 1998 Academic Press

*Corresponding author

Keywords: acylphosphatase; folding; proline isomerisation; real-time NMR; two-state model

Introduction The mechanism by which proteins fold to their native conformation is currently an area of active research in structural biology (Dill et al., 1995; Miranker & Dobson, 1996). One of the major apparent differences emerging from folding studies Present address: N. A. J. van Nuland, Departamento de QuõÂmica-FõÂsica, Facultad de Ciencias, Universidad de Granada, 18071 Granada, Spain. F.C. is on leave from the Dipartimento di Scienze Biochimiche di Firenze, Universita' di Firenze. Abbreviations used: AcP, acylphosphatase; ANS, 8-anilino-1-naphthalenesulphonic acid; far-UV CD, far ultraviolet circular dichroism; HFIP, hexa¯uoroisopropanol; NMR, nuclear magnetic resonance; PPI, peptidyl prolyl isomerase. E-mail address of the corresponding author: [email protected] 0022±2836/98/440883±09 $30.00/0

of a number of proteins is concerned with the presence or absence of intermediates accumulating during the folding process. While many small proteins appear to fold in a two-state fashion, polypeptides longer than 100 residues appear generally to fold by multiphasic processes with one or more relatively stable intermediates forming prior to the ®nal native state. The distinction between twostate or multi-state folding is not necessarily a fundamental one, as this simply depends on whether one or more intermediates are suf®ciently stable to accumulate and to allow experimental detection (Miranker & Dobson, 1996). Nevertheless, the investigation of two-state folding reactions has implications of fundamental importance for the analysis of data for determining thermodynamic and kinetic parameters of protein folding. Another important issue is that fast and highly ef®cient folding has been associated with two-state folding # 1998 Academic Press

884 (Fersht, 1997), while the slow folding of large and more complex proteins has frequently been attributed to the existence of kinetic traps in the form of intermediates. Some proteins, such as the adapter protein drk (Farrow et al., 1995) and, more recently, the SH3 domain of the phosphatidylinositol 30 kinase (Guijarro et al., 1998) have been found to fold slowly in a two-state fashion but these have been regarded as exceptions to the general pattern of behaviour (Plaxco et al., 1998). Here, we study the folding kinetics of human muscle acylphosphatase (AcP), a small enzyme of 98 residues catalysing speci®cally the hydrolysis of acylphosphates (Harary, 1957). AcP has been puri®ed and sequenced from several vertebrates as two isoenzymes sharing about 50% amino acid sequence identity (for a recent review, see Stefani et al., 1997). The two isoenzymes are known as muscle and common-type AcP. The structure of muscle AcP has been determined from 1H NMR data (Saudek et al., 1989; Pastore et al., 1992), and that of the common-type isoenzyme by X-ray crystallography (Thunnissen et al., 1997). In both cases AcP consists of two antiparallel a-helices packed against a ®ve-stranded antiparallel b-sheet. Several properties render AcP particularly suitable for kinetic studies of folding. Among these are its small size and the lack of complicating factors such as intramolecular disulphide bridges, prosthetic groups or cis-proline residues in the native structure. In the present study we have attempted to gain insight into the folding behaviour of muscle AcP upon transfer of protein denatured by urea into solution conditions in which the native state is stable. By means of a variety of techniques, including real-time NMR and stopped-¯ow optical techniques, we show that AcP folds slowly but in a two-state manner.

Folding of Muscle Acylphosphatase

in the present work is referred to as AcP throughout the rest of the text. AcP folding followed by intrinsic fluorescence, far-UV CD and ANS-binding AcP, initially denatured in 7 M urea, was transferred to refolding conditions by 11-fold dilution into a 50 mM acetate buffer (pH 5.5) containing different amounts of urea, and its folding was subsequently followed by ¯uorescence and far-UV CD spectroscopy. Figure 1 shows the recovery of the native ¯uorescence and CD signal monitored at

Results The analysis presented in this work has been carried out with a mutant form of AcP in which the single cysteine residue at position 21 has been replaced by a serine residue (Modesti et al., 1996). Previous studies suggest that a dimer and a complex of AcP with glutathione can form during puri®cation, both originating from intermolecular disulphide bridge formation through a cysteine residue located at position 21 (Berti et al., 1984). Gel-®ltration HPLC con®rms that the wild-type protein consists of two molecular forms and the electrospray mass spectrum of wild-type AcP has also revealed the presence of such a heterogeneous population (data not shown). On the contrary, the mass spectrum of the Cys21Ser mutant revealed a single species with a mass of 11,049.0(0.3) Da, corresponding to the calculated value for the monomeric protein. This prompted us to use the mutant protein to investigate the folding kinetics of AcP. For simplicity, the Cys21Ser mutant used

Figure 1. Refolding kinetics of AcP at different ®nal concentrations of urea. Measurements were at 28 C in 50 mM acetate buffer, pH 5.5. A, Refolding under mildly refolding conditions (1.78 M urea) monitored by the change in ellipticity at 225 nm and the change in ¯uorescence (inset) after initiating refolding by manual mixing. B, Refolding under strongly refolding conditions (0.63 M urea) monitored by the change in ellipticity at 225 nm and the simultaneously recorded change in ¯uorescence (inset). Continuous lines represent best ®ts to a single (A) or double (B) exponential term, and the corresponding best-®t parameters are given in Table 1. Open circles correspond to the ellipticity and ¯uorescence at time zero extrapolated from the equilibrium unfolding post-transition baselines.

885

Folding of Muscle Acylphosphatase

28 C at (A) 1.78 M and (B) 0.63 M urea. These two urea concentrations can be regarded as mildly and strongly refolding conditions, respectively, according to the urea unfolding titration data (see below, Figure 5). Figure 1A shows the analysis of the behaviour in mildly refolding conditions. The folding reaction was here initiated by manual mixing, because of the extremely slow folding of AcP under these conditions. Both the ¯uorescence and CD traces ®t well to single exponential functions with similar time constants (Table 1). In these experiments, fast phases cannot be detected because of the long dead time (ca ten seconds) arising from manual mixing. Nevertheless, the ¯uorescence and CD signals of the unfolded species under these refolding conditions can be easily determined by extrapolating the signal from the equilibrium post-transition unfolding baselines obtained at high urea concentrations. The values at 1.78 M urea obtained in this way correspond to those determined at time zero following extrapolation using the single-exponential ®t; this rules out the existence of an additional fast phase within the ®rst few seconds of the folding process. The data under strongly refolding conditions (Figure 1B) were collected by using a stopped-¯ow device, a situation made necessary because of the higher rate of folding under these conditions. Fluorescence and CD were monitored simultaneously. The ¯uorescence trace is not satisfactorily ®tted to a single exponential function. The analysis of the experimental data requires the use of a double exponential term and the parameters determined from the ®tting reveal two phases with comparable amplitudes and time constants close to each other on the refolding time-scale (Table 1). The ®t to a single exponential term is also not satisfactory for the CD data. These were therefore ®tted to a double exponential function by using the time constants determined by the ¯uorescence analysis; the calculated relative amplitudes determined from this procedure were very similar to those determined by ¯uorescence (Table 1). Again, no burst

phase occurs as is evident from the close agreement between the ¯uorescence and CD signals at time zero and those extrapolated from the equilibrium unfolded AcP baselines. When AcP folding was studied as a function of urea concentration biphasic behaviour was found up to 1.0 M; at higher urea concentrations the ¯uorescence trace could be ®tted satisfactorily to a single exponential term. In an additional set of experiments ANS ¯uorescence was monitored during folding. ANS is known to bind to exposed hydrophobic regions of proteins and has therefore been widely used to detect collapsed intermediate states occurring during folding (Engelhard & Evans, 1995). No ¯uorescence enhancement was observed during AcP folding, a result that complements the above observations providing evidence for the absence of a burst-phase collapsed state. Proline isomerisation The accurate determination of the time constants and relative amplitudes of the two kinetic phases observed under strongly refolding conditions requires that the two phases are well-separated in time-scale. AcP folding was therefore studied in the presence of hexa¯uoroisopropanol (HFIP), known to be a powerful accelerator of the folding of AcP as discussed elsewhere (Chiti et al., 1998a). Figure 2A shows that only the faster phase is accelerated in the presence of HFIP, the slow phase rate constant being independent of the concentration of HFIP. At 2.7% HFIP, when the two phases are well resolved, an average time constant of 13.2 seconds was determined for the slower phase and its relative amplitude accounts only for approximately 13% of the total ¯uorescence change. The low amplitude of the slow phase and the insensitivity to HFIP are suggestive of a cis-trans proline isomerisation rate-limiting event. Native AcP has both proline residues (Pro54 and Pro71) in a trans conformation. Hence, the slow phase might represent the proline isomerisation rate-limited folding of a

Table 1. Relative amplitudes (A) and time constants (t) for AcP folding monitored by a variety of techniques and under different conditions Technique used to monitor folding

Fast phase t1 (s)

Fast phase A (%)

Slow phase t2 (s)

Slow phase A (%)

A. Mildly folding conditions: 1.78 M urea, pH 5.5, 28 C Circular dichroism (225 nm) Intrinsic fluorescence (280 nm excit.)

75.2  0.09 74.7  0.05

100 100

± ±

± ±

B. Strongly folding conditions: 0.63 M urea, pH 5.5, 28 C Circular dichroism (225 nm)a Intrinsic fluorescence (225 nm excit.)a Intrinsic fluorescence (280 nm excit.)b Intrinsic fluorescence with 2.7% HFIP

9.5 9.5  0.2 8.3  0.1 1.03  0.04

46.1  2.4 40.0  1.5 41.8  1.2 87.8  1.2

20.8 20.8  0.4 18.0  0.2 13.2  0.2

53.9  2.4 60.0  1.2 58.2  1.2 12.2  1.2

All experiments were performed by mixing one volume of chemically denatured AcP with ten volumes of refolding buffer. The ®nal conditions following mixing are reported for each block of data. The experimental errors shown are ®tting errors. The time constant (t) is the reciprocal of the rate constant. a Fluorescence and far-UV CD simultaneously detected. The far-UV CD data were ®tted using the time constants obtained from ¯uorescence. b Experiments performed by diluting AcP unfolded in 2 M urea at pH 1.9.

886

Folding of Muscle Acylphosphatase

affect the faster phase, the slower found to be signi®cantly accelerated. to trans proline isomerisation in the PPI the following kinetic expression (Harrison & Stein, 1990):

phase was For the cis presence of is expected

kobs =k0 ˆ …kcat =KM k0 †‰PPIŠ ‡ 1

…1†

where kcat is the rate constant of the enzyme-catalysed isomerisation, KM is the Michaelis constant and k0 is the rate constant of the spontaneous, uncatalyzed isomerisation. The linear relationship between kobs/k0 and [PPI], predicted by this equation is observed for AcP (Figure 2B). By ®tting the data in Figure 2B to equation (1), a kcat/KM value of 2.04.105 sÿ1 Mÿ1 was obtained for the PPI-catalysed cis-trans proline isomerisation of AcP. AcP folding monitored by real-time NMR

Figure 2. The effect of hexa¯uoroisopropanol (HFIP) and peptidyl prolyl isomerase (PPI) on the fast (®lled symbols) and slow phase (open symbols) of AcP folding. A, The effect of HFIP on the rate constants of the folding of AcP at 28 C, 0.63 M urea, 50 mM acetate buffer, pH 5.5. B, Folding of AcP in the presence of catalytic amounts of PPI. Folding was followed by ¯uorescence at 28 C in 0.18 M urea, 1.1% HFIP, 0.1 M Tris buffer, pH 7.5. Data are reported as ratios of the observed rate constant (kobs) and the corresponding value in the absence of PPI (k0). The continuous line corresponds to the best ®t to equation (1).

low population of molecules with one or both proline residues in a cis con®guration. To address this possibility, AcP folding was studied in the presence of peptidyl prolyl isomerase (PPI), an enzyme speci®cally catalysing the cis-trans isomerisation of X-Pro peptide bonds (Fischer et al., 1984). Again HFIP was added to the refolding buffer in order to separate signi®cantly the kinetics of the two events. Figure 2B shows the effect of PPI on the rate constants of both phases. Data are reported as ratios of the observed rate constant, kobs, to the rate constant in the absence of PPI, k0. While catalytic amounts of PPI do not

Figure 3 shows a series of 1D NMR spectra recorded during the refolding of AcP following rapid injection of the protein denatured in 7.0 M urea into a refolding buffer at pH 5.5. The experiments were again carried out in the presence of 1.36% (v/v) HFIP, to resolve kinetically the two observed events, and at a temperature of 15 C in order to slow the overall process for a better acquisition of the NMR spectra. Comparison of the spectra after long refolding times (Figure 3E) with the spectrum of the native protein shows that they are identical, indicating that the native state has been fully regenerated in the refolding experiment. The ®rst spectrum recorded after three seconds (Figure 3A) shows characteristics of both unfolded and native species (cf. spectra B and E). At longer times following the initiation of refolding, peaks corresponding to those of the native state increase in intensity and the opposite behaviour is observed for those of the unfolded state. Figure 4 shows, for example, the changes in intensity of four resolved up®eld-shifted resonances corresponding to protons of residues far apart in the three-dimensional structure of native AcP, including Val20, Leu65, Ile75 and Val96 (Saudek et al., 1989). Analysis of the intensities of these and other well-resolved peaks from different regions of the protein shows that they all develop at rates closely similar to each other and to the rate of the overall appearance of the native spectrum. The kinetics, like those found for the optical experiments under similar conditions, show double-exponential behaviour (Figure 4). When applied to the data in Figure 4, the ®tting procedure discussed in the previous section yields values of 6.1(0.5) and 265(50) seconds for the time constants of the fast and slow phase, respectively, and values of 87  6 and 13  1 for their relative amplitudes. Due to the different experimental conditions, the NMR data cannot be compared directly with the CD and ¯uorescence data reported in the previous section. However, when the ¯uorescence experiments were

Folding of Muscle Acylphosphatase

887 repeated under the conditions used for NMR, very similar values were obtained (5.26(0.02) seconds, 280(200) seconds, 87.4  0.2 and 11.3  0.2 (the experimental errors are ®tting errors)). It is interesting to note that the faster phase observed using 2H2O as a solvent was found to be increased in rate by a factor of 1.8 relative to that observed using H2O (Figure 4, inset). Perturbation of the folding rates by changing H2O to 2 H2O has also been observed for other proteins (Itzhaki & Evans, 1996; Guijarro et al., 1998), although the underlying reason for this has not yet been clearly established (Itzhaki & Evans, 1996; Parker & Clarke, 1997). The kinetic data were then used to reconstruct all NMR spectra at intermediate time-points between the ®rst and the ®nal spectrum by a linear combination of the latter two spectra. The reconstructed spectra are closely similar to the experimental spectra regardless of the time at which the analysis was performed. Using the ®rst and last of the recorded spectra, and the parameters from the kinetic analysis, the spectrum at time zero can be reconstructed (Figure 3C); the chemical shift dispersion in this spectrum is closely similar to that of the fully unfolded protein in 7.0 M urea (Figure 3B). These data indicate that all spectral changes observed during refolding can be attributed solely to a single, highly cooperative two-state transition from the initial unfolded state into the ®nal native state. Moreover, subtraction of the relevant spectra to give the spectrum of the slowly refolding species (Figure 3D) shows that they are largely unstructured and that the biphasic behaviour arises from

Figure 3. The refolding of AcP from 7.0 M urea in H2O into 50 mM d3-acetate buffer/2H2O (pH 5.5), 1.36% (v/v) d2-HFIP at 15 C monitored by NMR. A, Stacked plot of 600 MHz 1D NMR spectra of AcP in 2 H2O at 15 C. Spectra were recorded at incremented time-points between 3.07 seconds and 18 minutes. The region shown contains resonances from methyl and methylene groups. Lines at the earlier time-points are slightly broader, re¯ecting residual inhomogeneity of the solution at early times or slow conformational averaging on the NMR time-scale (Balbach et al., 1995). B, The 600 MHz 1D 1H NMR spectrum of AcP in 7.0 M urea in 50 mM d3-acetate/2H2O, pH 5.5. C, Spectrum at time zero, reconstructed using the ®rst spectrum, recorded 3.07 seconds (average over two scans) after initiation of folding and subtracting the last spectrum (E) weighted by the population of native molecules expected from the kinetics shown in Figure 4. D, Spectrum of the slowly refolding species corresponding to the difference between the kinetic spectrum at 22.1 seconds, in which the fast phase is about 97% complete, and the last, native spectrum (E) weighted by the population fraction of native molecules expected from the kinetics shown in Figure 4. E, The ®nal spectrum following complete refolding (18 minutes after injection). The up®eld region marked by the two dotted lines was used to determine the refolding kinetics shown in Figure 4 and includes resonances of Val20, Leu65, Ile75 and Val96 (Saudek et al., 1989). 2

Figure 4. Refolding kinetics of AcP from 7.0 M urea obtained from the real-time NMR measurement showing the appearance of the native signals between 0.5 ppm and ÿ0.1 ppm indicated by dotted lines in Figure 3E. The continuous line represents the best ®t to a doubleexponential of data points up to 1080 seconds. The inset shows the ¯uorescence traces obtained under the same conditions in H2O and 2H2O.

888 heterogeneity in a largely unfolded state of the protein. Comparison between kinetic and equilibrium unfolding of AcP Figure 5 shows the dependence of the folding and unfolding rates on urea concentration, along with the equilibrium urea denaturation curve obtained under the same conditions (inset). The equilibrium data were analysed using the two-state assumption for the unfolding reaction as described (van Nuland et al., 1998). Using this analysis, a value for
GH2O of 18.8(1.0) kJ molÿ1, an m value of 4.75(0.3) kJ molÿ1 Mÿ1 and a Cm value of 3.97(0.2) M were obtained. The unfolding of AcP was found to be a monophasic process. For a twostate transition, the natural logarithm of the rate constants for folding and for unfolding are linearly dependent on the ®nal urea concentration (Jackson & Fersht, 1991). Values for the folding and unfolding rate constants in the absence of urea (kfH2 O and kuH2 O , respectively) and for the fractional m-values for folding (mf) and unfolding (mu) can be obtained from the two halves of the curve shown in Figure 5. From these data, the values for
GH2O, m and Cm can be calculated (Jackson & Fersht, 1991). This analysis is, however, complicated by the presence of the slower phase ascribed to cis-trans proline iso-

Figure 5. Urea concentration dependence of the natural logarithm of folding (®lled symbols) and unfolding (open symbols) rate constants. Both folding and unfolding appear to be monophasic processes and the rate constant data refer to the single phase observed. Nevertheless, in the range 0 to 1 M urea, two phases can be observed. The rate constant of the faster phase obtained under these conditions has been multiplied by the fraction of molecules with the proline residue(s) in the correct con®guration, as explained in the text (®lled squares). Inset: equilibrium urea unfolding of AcP followed by intrinsic ¯uorescence. The kinetic and equilibrium data were analysed as described (van Nuland et al., 1998). Both the kinetic and equilibrium experiments were carried out at 28 C, 50 mM acetate buffer, pH 5.5.

Folding of Muscle Acylphosphatase

merisation. As shown above, the rate constant of this slower phase is approximately 0.079 sÿ1 (ln k ˆ ÿ2.54). This implies that at urea concentrations higher than 1 M, proline isomerisation is no longer the rate-limiting step in the folding of AcP (Figure 5) and the single observed rate constant kobs corresponds to the rate constant of the molecular population with the proline residues in the correct trans con®guration for folding multiplied by its population fraction (kf ftrans). By contrast, at very low urea concentrations the two phases are well resolved and the observed rate constant of the fast phase corresponds to the actual rate constant of folding kf. Assuming a two-state folding model,
GH2O is then:
GH2 O ˆ ÿ RT ln…Keq † ˆ ÿ RT lnf…‰Utrans Š ‡ ‰Ucis Š†=‰FŠg ˆ ÿ RT ln‰kuH2 O =…kfH2 O ftrans †Š

…2†

where F is the population of folded molecules, and Utrans and Ucis are the populations of unfolded protein with the proline residues in the trans and cis con®gurations, respectively, corresponding to the relative amplitudes of the two phases. For direct comparison of the thermodynamic parameters derived from the two sets of experiments (kinetic and equilibrium), the kf values estimated at very low urea concentrations must be multiplied by ftrans. The values, obtained from the linear ®ts shown in Figure 5, for kfH2 O , kuH2 O , mf and mu are 0.23 sÿ1, 1.1  10ÿ4 sÿ1, 1.61 Mÿ1 and 0.42 Mÿ1, respectively. From these, the values for
GH2O, m and Cm were calculated to be 19.0(1.0) kJ molÿ1, 5.07(0.30) kJ molÿ1 Mÿ1 and 3.75(0.3) M, respectively, in close agreement with the values obtained from the equilibrium measurements. Substantially identical values are obtained by excluding the folding rate constant data at very low urea concentration. The values of mf and mu can be used to obtain a value for a{ ˆ mu/(mu ‡ mf), the solvent-exposed surface area of the transition state relative to the unfolded and folded states (Jackson & Fersht, 1991). For AcP the value of this is 0.20  0.02, suggesting that the transition state is highly compact and native-like.

Discussion Our results all show that the folding of AcP can be well described as a two-state reaction. Folding appears as a monophasic process (except for the minor phase due to proline isomerization) regardless of the probe employed to follow the reaction. Further, evidence comes from the linearity of a plot of the natural logarithm of folding and unfolding rate constants against denaturant concentration (Figure 5) and from the very good agreement between the thermodynamic parameters calculated from kinetic and equilibrium experiments. Although recent studies have shown that devi-

889

Folding of Muscle Acylphosphatase

ations from two-state behaviour are more likely to be found under strongly refolding conditions (Matouschek et al., 1990; Schreiber & Fersht, 1993; Khorasanizadeh et al., 1996), our kinetic studies allow us to exclude this possibility for AcP and to be con®dent about the validity of a two-state model for describing the folding and unfolding of AcP. Urea-denatured AcP does not show characteristics of any signi®cant non-random residual structure (Taddei et al., 1994). This situation persists after abrupt transfer of AcP into refolding conditions; the 1D 1H NMR spectrum, the far-UV CD ellipticity and the intrinsic ¯uorescence extrapolated to time zero in all cases correspond to those of the protein denatured in 8 M urea. Moreover, the lack of ¯uorescence enhancement when folding is monitored by ANS-binding argues against the formation of an intermediate with the characteristics of a compact partially folded state. These observations indicate that after initiation of folding AcP is highly unstructured without speci®c and persistent non-native or native-like interactions, and that the structural changes accompanying AcP folding occur in a highly cooperative manner on a time-scale of seconds. The proline isomerisation limited phase of AcP is catalysed by PPI. The calculated value of kcat/ KM, an indicator of the ef®ciency of PPI in promoting the cis-trans isomerisation reaction, is 2.04  105 sÿ1 Mÿ1. The corresponding value estimated for model pentapeptides and containing a single proline residue ranges from 6  105 to 3  106 sÿ1 Mÿ1, whereas the kcat/KM value is often 1000-fold lower for PPI-catalysed proline isomerisation in proteins (Schmid et al., 1991). Many slow phases known to involve proline isomerisation are indeed not affected by PPI. The low ef®ciency of PPI in accelerating proline isomerisation in many proteins has been attributed to a lack of accessibility to PPI within a collapsed state of the proteinsubstrate (Schmid et al., 1991). The high ef®ciency of PPI in catalysing the cis-trans isomerisation of AcP is in agreement with the lack of residual structure in the species formed directly after initiating of folding, as concluded from the other data presented above. AcP is a protein structurally related to the activation domain of procarboxipeptidase A2 (ADA2h) and the histidine-containing phosphocarrier protein (HPr). They all show a classical openfaced b-sandwich fold comprised of two or three antiparallel a-helices packed against a four or ®vestranded antiparallel b-sheet, but do not share sequence homology. The folding rate constant in the absence of denaturant (kfH2 O ) varies enormously within the set of three proteins; 897 sÿ1 for ADA2h (Villegas et al., 1995), 14.9 sÿ1 for HPr (van Nuland et al., 1998), and 0.23 sÿ1 for AcP, although their stabilities are fairly similar (
GH2O values are 17.0, 18.5 and 19.0 kJ molÿ1, respectively). A wide variety of rate constants was found for ®ve all-b-sheet SH3 domains, all sharing the same fold but in gen-

eral displaying a low level of sequence identity (Plaxco et al., 1998). It was found that the most stable protein within the class of SH3 domains folds most rapidly. In the set of three proteins discussed here, however, the one with the lowest stability (ADA2h) refolds most rapidly. Moreover, the wide range in folding rates displayed within this set of proteins (e.g. ADA2h folds ca 3600 times faster than AcP) suggests that indeed, in addition to thermodynamic stability, other factors such as long-range interactions or intrinsic stability of secondary structure elements play a fundamental role in de®ning the folding and unfolding rates of proteins displaying a similar fold. Moreover, the data for AcP show that intrinsic barriers to folding can be high even in the case of proteins that fold in a two-state manner.

Materials and Methods Materials Wild-type and mutant AcP were expressed and puri®ed as described (Modesti et al., 1995). Protein purity was found to be higher than 98% as judged from SDSPAGE and electrospray mass spectrometry. PPI was a generous gift from Sophie Jackson. Far-UV CD Folding of AcP was studied by monitoring the far-UV ellipticity at 225 nm on either a Jasco J-720 spectropolarimeter (manual mixing) or on a Biologic stopped-¯ow machine equipped with double detection for both CD and ¯uorescence. AcP (2.5 mg/ml in 7 M urea) was rapidly mixed to give an 11-fold dilution in 50 mM acetate buffer (pH 5.5) containing various amounts of urea to initiate refolding at 15 or 28 C. AcP was ®rst incubated for one hour in 7 M urea, 50 mM acetate buffer (pH 5.5), allowing complete denaturation to be attained. For the stopped-¯ow experiments, the ellipticity at 225 nm as well as the total ¯uorescence above 314 nm (excitation wavelength of 225 nm) were simultaneously detected. The ¯uorescence emission arising from excitation at 225 nm is the result of an absorption maximum of tryptophan residues at this wavelength (Creighton, 1993). Refolding kinetics were ®tted to a single or double exponential function using the Kaleidagraph software (Synerge Software, PCS Inc.). Intrinsic fluorescence Total ¯uorescence above 314 nm was followed during the folding reactions as described above when monitored simultaneously with CD, or were performed on either a Perkin Elmer LS 50 B spectro¯uorimeter (manual mixing) or on an Applied Photophysics SX.17MV stopped¯ow spectro¯uorimeter (APP) using an excitation wavelength of 280 nm. The ®nal protein concentration in the latter experiments was typically 0.04 mg/ml. The refolding experiments at ®nal urea concentrations up to 0.18 M were carried out by 11-fold dilution of initially unfolded AcP in 2 M urea (pH 1.9). Under identical ®nal conditions, the folding kinetics were shown to be independent of the initial conditions of denaturation of AcP. The unfolding experiments were performed on either the APP or the Perkin Elmer spectro¯uorimeter. The kinetic

890 traces were analysed using the Kaleidagraph program. The equilibrium experiments of urea-induced unfolding were performed as reported by Chiti et al. (1998b). 8-Anilino-1-naphthalenesulphonic acid (ANS) fluorescence For ANS ¯uorescence experiments an excitation wavelength of 350 nm was used, and total ¯uorescence above 390 nm was monitored using the APP instrument. Typically, 1 mg/ml AcP solution unfolded in 7 M urea was diluted 11-fold with 100 mM ANS in 50 mM acetate (pH 5.5) to initiate refolding. Real-time NMR spectroscopy NMR spectra were recorded using a home-built NMR spectrometer operating at 600.2 MHz, and processed using FELIX (Hare Research). The spectral width was 8000 Hz. The residual water was saturated by weak onresonance irradiation during the 0.7 second relaxation delay. Chemical shifts are expressed relative to sodium2.2-dimethyl-2-silapentane-5-sulphonate (DSS). For all refolding experiments 50 ml of 160 mg/ml AcP in 7.0 M d4-urea, 50 mM d3-acetate buffer/2H2O (pH 5.5) was diluted into a 500 ml volume of refolding buffer containing hexa¯uoroisopropanol (HFIP). Final conditions were 0.63 M d4-urea, 1.36% d2-HFIP, 50 mM d3-acetate/2H2O buffer (pH 5.5), 15 C. The ®nal protein concentration in all NMR samples was 1.3 mM. In all, 512 1D NMR spectra with two scans each were recorded for refolding at 15 C with an acquisition time of 0.256 second. To initiate the refolding of AcP, denatured protein was injected at a ®xed time-point during the recording of the 512 1D FIDs (Balbach et al., 1995). The ®rst spectrum was acquired 3.07 seconds after injection and subsequent spectra were then recorded at 2.06 second intervals. Refolding kinetics were ®tted to a double exponential function using the Kaleidagraph software. Folding in the presence of peptidyl prolyl isomerase (PPI) For the proline isomerisation study, AcP initially denatured in 2 M urea (pH 1.9) was diluted with ten volumes of 0.1 M Tris buffer (pH 8.0) with PPI at concentrations ranging from 0 to 3.5 mM. The ®nal urea concentration and pH were 0.18 M and 7.5, respectively, conditions optimal for the activity of PPI (Harrison & Stein, 1990). For the PPI-mediated refolding experiments, total ¯uorescence above 314 nm was monitored using the APP stopped-¯ow instrument.

Acknowledgements We are grateful to Claudia Icardi for her technical support in the expression and puri®cation of acylphosphatase, to Anabel Azuaga for the gel-®ltration experiments and to Evonne Chung and Carol Robinson for performing electrospray mass spectrometry experiments. We thank Inaki Guijarro and Massimo Stefani for useful discussions and Sophie Jackson for providing PPI. N.A.J.v.N. and F.C. were supported by the European Community, TMR and Biotechnology programmes, respectively. This is a contribution from the Oxford Centre for Molecular Sciences, which is supported by

Folding of Muscle Acylphosphatase EPSRC, BBSRC and MRC. The work has been supported also by funds from Italian CNR (Target Project Structural Biology), from MURST (fondi 40%) and from the European Community (Biotechnology Unit). The research of C.M.D. is supported in part by an International Research Scholars award from the Howard Hughes Medical Institute, and by the Wellcome Trust.

References Balbach, J., Forge, V., van Nuland, N. A. J., Winder, S. L., Hore, P. J. & Dobson, C. M. (1995). Following protein folding in real time using NMR spectroscopy. Nature Struct. Biol. 2, 865±870. Berti, A., Stefani, M., Camici, G., Manao, G., Cappugi, G., Degl'Innocenti, D. & Ramponi, G. (1986). Puri®cation and characterisation of rabbit muscle acylphosphatase in the thiol (-SH) form. Int. J. Pept. Protein Res. 28, 15 ± 21. Chiti, F., Taddei, N., van Nuland, N. A. J., Magherini, F., Stefani, M., Ramponi, G. & Dobson, C. M. (1998a). Structural characterisation of the transition state for folding of muscle acylphosphatase. J. Mol. Biol. 283, 893± 902. Chiti, F., van Nuland, N. A. J., Taddei, N., Magherini, F., Stefani, M., Ramponi, G. & Dobson, C. M. (1998b). Conformational stability of muscle acylphosphatase. The role of temperature, denaturant concentration and pH. Biochemistry, 37, 1447± 1455. Creighton, T. E. (1993). Proteins, 2nd edit., p. 15, Freeman and Co., New York. Dill, K. A., Bromberg, S., Yue, K. Z., Fiebig, K. M., Yee, D. P., Thomas, P. D. & Chan, H. S. (1995). Principles of protein folding: a perspective from simple exact models. Protein Sci. 4, 561± 602. Engelhard, M. & Evans, P. A. (1995). Kinetics of interaction of partially folded proteins with a hydrophobic dye: evidence that molten globule character is maximal in early folding intermediates. Protein Sci. 4, 1553± 1562. Farrow, N. A., Zhang, O., Forman-Kay, J. D. & Kay, L. E. (1995). Comparison of the backbone dynamics of a folded and unfolded SH3 domain existing in equilibrium and aqueous buffer. Biochemistry, 34, 868± 878. Fersht, A. R. (1997). Nucleation mechanisms in protein folding. Curr. Opin. Struct. Biol. 7, 3 ± 9. Fischer, G., Bang, H. & Mech, C. (1984). Nachweis einer enzymkatalyse fur die cis-trans-isomerisierung der peptidbindung in prolinhaltigen peptiden. Biomed. Biochim. Acta, 43, 1101. Guijarro, J. I., Morton, C. J., Plaxco, K. W., Campbell, I. D. & Dobson, C. M. (1998). Folding kinetics of the SH3 domain of PI3 kinase by real time NMR combined with optical spectroscopy. J. Mol. Biol. 276, 657± 667. Harary, I. (1957). The hydrolysis of 1,3-diphosphoglyceric acid by acylphosphatase. Biochim. Biophys. Acta, 26, 434± 442. Harrison, R. K. & Stein, R. L. (1990). Substrate speci®cities of peptidyl prolyl cis-trans isomerase activities of cyclophilin and FK506 binding protein: evidence for the existence of a family of distinct enzymes. Biochemistry, 29, 3813± 3816. Itzhaki, L. S. & Evans, P. A. (1996). Solvent isotope effects on the refolding kinetics of hen egg-white lysozyme. Protein Sci. 5, 140± 146.

Folding of Muscle Acylphosphatase Jackson, S. E. & Fersht, A. R. (1991). Folding of chymotrypsin inhibitor 2. 1. Evidence for a two-state transition. Biochemistry, 30, 10428± 10435. Khorasanizadeh, S., Peters, I. D. & Roder, H. (1996). Evidence for a three-state model of protein folding from kinetic analysis of ubiquitin variants with altered core residues. Nature Struct. Biol. 3, 193±205. Matouschek, A., Kellis, J. T., Jr, Serrano, L., Bycroft, M. & Fersht, A. R. (1990). Transient folding intermediates characterised by protein engineering. Nature, 346, 440± 445. Miranker, A. D. & Dobson, C. M. (1996). Collapse and cooperativity in protein folding. Curr. Opin. Struct. Biol. 6, 31 ±42. Modesti, A., Taddei, N., Bucciantini, M., Stefani, M., Colombini, B., Raugei, G. & Ramponi, G. (1995). Expression, puri®cation and characterisation of acylphosphatase muscular isoenzyme as fusion protein with glutathione S-transferase. Protein Express. Purif. 6, 799± 805. Modesti, A., Taddei, N., Chiti, F., Bucciantini, M., Magherini, F., Rigacci, S., Stefani, M., Raugei, G. & Ramponi, G. (1996). Properties of Cys21-mutated muscle acylphosphatases. J. Protein Chem. 15, 27 ± 34. Parker, M. J. & Clarke, A. R. (1997). Amide backbone and water-related H/D isotope effects of the dynamics of a protein folding reaction. Biochemistry, 36, 5786± 5794. Pastore, A., Saudek, V., Ramponi, G. & Williams, R. J. P. (1992). Three-dimensional structure of acylphosphatase. Re®nement and structure analysis. J. Mol. Biol. 224, 427± 440. Plaxco, K. W., Guijarro, J. I., Morton, C. J., Pitkeathly, M., Campbell, I. D. & Dobson, C. M. (1998). The folding kinetics and thermodynamics of the FynSH3 domain. Biochemistry, 37, 2529± 2537.

891 Saudek, V., Boyd, J., Williams, R. J. P., Stefani, M. & Ramponi, G. (1989). The sequence-speci®c assignment of the 1H NMR spectrum of an enzyme, horse muscle acylphosphatase. Eur. J. Biochem, 182, 85 ± 93. Schreiber, G. & Fersht, A. R. (1993). The refolding of cisand trans-peptidyl prolyl isomers of barstar. Biochemistry, 32, 11195± 11203. Schmid, F. X., Lang, K., Kiefhaber, T., Mayer, S. & Schonbrunner, E. R. (1991). Prolyl isomerase: its role in protein folding and speculation on its function in the cell. In Conformations and Forces in Protein Folding (Nall, B. T. & Dill, K. A., eds), p. 198, American Association for the Advancement of Science, Washington, DC. Stefani, M., Taddei, N. & Ramponi, G. (1997). Insights into acylphosphatase structure and catalytic mechanism. Cell. Mol. Life Sci. 53, 141±151. Taddei, N., Buck, M., Broadhurst, R. W., Stefani, M., Ramponi, G. & Dobson, C. M. (1994). Equilibrium unfolding studies of horse muscle acylphosphatase. Eur. J. Biochem. 225, 811± 817. Thunnissen, M. M. G. M., Taddei, N., Liguri, G., Ramponi, G. & Nordlund, P. (1997). Crystal structure of common type acylphosphatase from bovine testis. Structure, 5, 69 ± 79. van Nuland, N. A. J., Meijberg, W., Warner, J., Forge, V., Scheek, R. M., Robillard, G. T. & Dobson, C. M. (1998). Slow cooperative folding of a small globular protein HPr. Biochemistry, 37, 622± 637. Villegas, V., Azuaga, A., Catasus, Ll. , Reverter, D., Mateo, P. L., Aviles, F. X. & Serrano, L. (1995). Evidence for a two-state transition in the folding process of the activation domain of human procarboxypeptidase A2. Biochemistry, 34, 15105± 15110.

Edited by P. E. Wright (Received 12 February 1998; received in revised form 10 June 1998; accepted 12 June 1998)