An obligatory intermediate in the folding pathway of cytochrome c552 from Hydrogenobacter thermophilus.

The folding mechanism of many proteins involves the population of partially organized structures en route to the native state. Identification and characterization of these intermediates is particularly difficult, as they are often only transiently populated and may play different mechanistic roles, being either on-pathway productive species or off-pathway kinetic traps. Following different spectroscopic probes, and employing state-of-the-art kinetic analysis, we present evidence that the folding mechanism of the thermostable cytochrome c552 from Hydrogenobacter thermophilus does involve the presence of an elusive, yet compact, on-pathway intermediate. Characterization of the folding mechanism of this cytochrome c is particularly interesting for the purpose of comparative folding studies, because H. thermophilus cytochrome c552 shares high sequence identity and structural homology with its homologue from the mesophilic bacterium Pseudomonas aeruginosa cytochrome c551, which refolds through a broad energy barrier without the accumulation of intermediates. Analysis of the folding kinetics and correlation with the three-dimensional structure add new evidence for the validity of a consensus folding mechanism in the cytochrome c family.

As for any other chemical process, a meaningful description of the protein folding mechanism demands careful analysis of its kinetics. A wealth of experimental studies has shown that proteins may fold through processes of different complexity, with or without partially folded intermediate states (1). Evidence for folding intermediates is usually provided by the ob-servation of complex multiphasic time courses. Intermediate states can either accumulate in the time window accessible to classical stopped-flow instruments (ms) or be detected in a sub-ms time range by ultrarapid mixing (2)(3)(4) or temperature jump techniques (5,6). However, simple kinetics is not always a sufficient indication of a two-state mechanism, because hidden intermediates are sometimes revealed by more sophisticated kinetic experiments, which may allow assignment of their mechanistic significance.
To highlight general rules in protein folding, an increasingly employed strategy is to compare the folding mechanism of different proteins belonging to the same fold family. From the few examples available to date, it emerges that native state topology plays a dominant role in determining the folding process, because the structure of the transition state is generally conserved among different members of a protein family (7)(8)(9)(10). However, it has been shown that differences in local propensity to form elements of secondary structure may modulate the stabilities of transient species, such as intermediates and/or transition state(s), and therefore introduce variations in the order whereby structural motifs are formed during folding (11,12). Such effects may result in different topologies of folding transition states in homologous proteins, as demonstrated by Baker and co-workers (13) for the IgG binding domains of proteins G and L.
We focus here on c-type cytochromes, a protein family in which the folding mechanism seems to involve an elusive intermediate species in the ms time window (14,15). Starting from the identification of an on-pathway intermediate observed in the folding of Thermus thermophilus cytochrome c 552 (16), a critical re-evaluation of published kinetic data on different prokaryotic and eukaryotic c-type cytochromes led to the proposal (17) that proteins belonging to the cytochrome c family share a consensus folding mechanism involving, in all cases, the population of an on-pathway intermediate with conserved structural features.
To obtain additional experimental evidence to strengthen the hypothesis of a consensus folding mechanism in this protein family, we report herein an analysis of the folding kinetics of cytochrome c 552 from the thermophilic bacterium Hydrogenobacter thermophilus. The three-dimensional structure of ferrous H. thermophilus cytochrome c 552 has been previously solved by solution NMR (18). In this study, we present the crystal structure of the ferric derivative, which is generally used in folding studies on c-type cytochromes. Characterization of the folding mechanism of this cytochrome c is particularly interesting for the purpose of comparative folding, because contrary to T. thermophilus cytochrome c 552 , H. thermophilus cytochrome c 552 has a typical class I cytochrome c fold (Fig. 1) and shares high sequence identity (57%) and structural homology with its mesophilic counterpart from the bacterium Pseudomonas aeruginosa cytochrome c 551 . The latter was found to refold through a broad energy barrier with two transition states separated by a high energy intermediate (15). On the basis of the substantial difference in thermodynamic stability between these two structurally homologous cytochromes (18), we speculated that an intermediate species with properties similar to those proposed for the high energy intermediate of P. aeruginosa cytochrome c 551 (and other evolutionary distant c-type cytochromes) would be populated in the folding of H. thermophilus cytochrome c 552 .
We show here that, despite a canonical V-shaped chevron plot derived following secondary structure formation, H. thermophilus cytochrome c 552 transiently populates a compact obligatory intermediate during refolding. Analysis of the folding kinetics, followed by Trp fluorescence, and correlation with the three-dimensional structure add new evidence for the validity of a consensus folding mechanism in the cytochrome c family.

MATERIALS AND METHODS
Protein Expression and Purification-H. thermophilus cytochrome c 552 was expressed and purified as described previously (19).
Crystallization and Structure Determination-The protein was crystallized using the hanging-drop method at 294 K under the following conditions: drops (1.6 l) were prepared by mixing equal volumes of protein solution and crystallization solution consisting of 50 -60% 2-methyl-2,4-pentanediol, 0.8 -0.9 M ammonium sulfate, HEPES at 0.1 M, pH 7.0, and allowing the mixture to be equilibrated against 0.5 ml of the same reservoir solution. Crystals grew in ϳ9 months and were flash-frozen in liquid nitrogen without further addition of cryoprotectants.
A complete data set at 2.0 Å of resolution was collected at the ELETTRA Synchrotron (Trieste, Italy). Reflection intensities were integrated and scaled using DENZO/SCALEPACK (20). The crystal belongs to the P4322 space group, with unit cell dimensions of a ϭ b ϭ 56.7 Å, c ϭ 220.18 Å ( Table I).
The structure was determined by molecular replacement with the program AMORE (21), using as a template the structure of P. aeruginosa cytochrome c 551 (Protein Data Bank code 451C). The structure was completed in alternating cycles of model building using QUANTA (Molecular Structure, The Woodlands, TX) and refinement using Refmac5 (22). Water residues were added into the F o Ϫ F c density map contoured at 4 with the X-SOLVATE tool of QUANTA. Several cycles of refinement, manual rebuilding, and addition of solvent molecules led to an R-factor of 0.17 and an R free value of 0.22. The refined model was monitored for geometrical quality by using PROCHECK (23) (Table I).
Coordinates and structure factors have been deposited in the Protein Data Bank (PDB code 1YNR). Structural alignments were performed with software CE (24).
Equilibrium Experiments-Chemical denaturation was monitored by fluorescence emission at 354 nm (excitation at 280 nm; 10-mm light path) using a JobinYvon fluorometer and by circular dichroism at 222 nm using a Jasco spectropolarimeter (2-mm light path). Samples (1 M for fluorescence and 11 M for CD 1 measurements) were thermostatted at 10 Ϯ 0.1°C.
Kinetic Experiments-Kinetic experiments were carried out either with a *-180 (single mixing experiments) or with an SX-18 stoppedflow instrument (double mixing experiments), both from Applied Photophysics, Leatherhead, UK.
Fluorescence emission was measured using a Ͼ320 nm cutoff filter (excitation was at 280 nm). Time-resolved circular dichroism was followed at 225 nm (light path 10 mm). Temperature was set at 10 Ϯ 0.1°C. Folding and unfolding were initiated by an 11-fold dilution of the denatured or native protein in the appropriate GdnHCl solution.
Data Analysis-Equilibrium fluorescence and far-UV CD measurements, as a function of denaturant concentration, were fitted according to a two-state model in which only the native and the fully unfolded states are populated, using the equation, where GdnHCl 1 ⁄2 is the GdnHCl concentration in which 50% of the protein is unfolded and m D-N is the m-value. The free energy of denaturation in the absence of denaturant (⌬G 0 D-N ) was calculated assuming a linear dependence of the free energy of unfolding on denaturant concentration.
Kinetic Measurements-In a simple three-state reaction scheme, the observed kinetics is governed by the two apparent rate constants 1 and 2 , and the denaturant dependence of the observed rate constant 2 shows a typical distortion (roll-over effect seen at low denaturant) from the classical V-shaped chevron plot 1 ϭ (p ϩ q)/2 and 2 ϭ (p Ϫ q)/2, . However, discrimination between the on-pathway (U 7 I 7 N) Model 1 and off-pathway (I 7 U 7 N) Model 2 can be achieved only when the two apparent phases are kinetically coupled and all the relaxation rates are measured over a wide range of denaturant concentration, where k IJ is the microscopic rate constant for the formation of state J from state I, and m IJ defines the denaturant dependence of this step.

RESULTS AND DISCUSSION
Structure of Ferric H. thermophilus Cytochrome c 552 -We solved the structure of oxidized H. thermophilus cytochrome c 552 (Fig. 1) by x-ray crystallography at 2.0 Å of resolution (crystallographic parameters are listed in Table I). Comparison of the three-dimensional structures of the ferrous (18) and ferric derivatives shows that the two forms are very similar, as expected. The root mean square deviation of the C-␣ atoms between the crystal structure and the average of the 20 NMR solution structures of H. thermophilus cytochrome c 552 is 0.98 Å. Some interesting features, which were not detected by solution NMR, were identified in the crystal structure of the ferric Interestingly these residues were shown by site-directed mutagenesis to be important for the thermostability of H. thermophilus cytochrome c 552 (25).
Furthermore, comparison between the crystal structures of P. aeruginosa cytochrome c 551 (26) and H. thermophilus cytochrome c 552 , over and above the similarity of the main chain folding (root mean square deviation of the C-␣ atoms is approx. 0.7 Å), highlighted major packing differences for the two proteins, with the mesophilic cytochrome displaying a greater volume of internal cavities (18). Indeed it has been shown by site-directed mutagenesis that such differences can account for the enhanced thermodynamic stability of H. thermophilus cytochrome c 552 (25).
Equilibrium Unfolding-Characterization of the (un)folding transition of H. thermophilus cytochrome c 552 has been carried out with the ferric derivative at different pH values from pH 3.0 to 9.0 (data not shown). GdnHCl- Folding and Unfolding Kinetics-Kinetic characterization was carried out at pH 4.7 in 50 mM sodium acetate buffer to facilitate comparison of the results with those previously obtained for P. aeruginosa cytochrome c 551 (27). The kinetics of folding and unfolding followed by far-UV CD were single exponential (Fig. 3, lower trace) at all denaturant concentra-tions, with no evidence for protein concentration dependence (from 0.5 to 5.0 M) of the first order rate constants (data not shown), suggesting the absence of transient aggregation events (28). The semilogarithmic plot of the observed folding and unfolding rate constants, as a function of denaturant concentration (chevron plot) reported in Fig. 4, top panel), corresponds to a simple V-shape, apparently consistent with a two-state folder. The thermodynamic stability derived from quantitative analysis of the kinetic experiments (⌬G DN ϭ Ϫ8.6 Ϯ 0.1 kcal mol Ϫ1 ) is in very good agreement with the value obtained from equilibrium experiments.
When the refolding time course of H. thermophilus cytochrome c 552 was monitored by Trp fluorescence, extensive emission quenching yields a kinetic phase with k ϭ 75 s Ϫ1 and an additional slower phase (ϳ3 s Ϫ1 ) with very small amplitude (ϳ5% of the signal) (Fig. 3, upper trace). This slower process may have escaped detection by far-UV CD because of the lower signal to noise ratio with respect to fluorescence quenching. The shallow denaturant dependence and the small amplitude of this slower process initially suggested that it may be as-  signed to proline cis-trans isomerization (29); however, additional experiments allowed us to conclude that this slow phase reflects, in fact, the formation of native molecules from an intermediate state, as detailed below.
A powerful experimental approach to the testing of a folding mechanism is afforded by double-mixing interrupted kinetic experiments. As shown by Kiefhaber (30) for lysozyme, this technique allows the fraction of native protein formed during the refolding process to be monitored. In the case of H. thermophilus cytochrome c 552 , unfolded molecules were allowed to begin refolding by denaturant dilution in the first mix and then were unfolded again in a second mix after a variable delay time. The results showed that the major fraction of native protein is formed in a first-order kinetic reaction (Fig. 4, inset) with k ϳ 5 s Ϫ1 , similar to the slower process seen in fluorescence-monitored single-mixing experiments (see previous paragraph). Therefore, at the same [GdnHCl], the faster process (k ϳ 80 s Ϫ1 at 1 M GdnHCl) associated with the major amplitude should reflect formation of a partially folded intermediate, in which the fluorescence and CD properties are very similar to those of the native state (implying only a small additional fluorescence quenching associated to the transition I 3 N).
The Folding Intermediate Is On-pathway-Detection of a folding intermediate is not sufficient, per se, to distinguish between two fundamentally different mechanisms that may involve (i) a productive on-pathway obligatory intermediate or (ii) a misfolded off-pathway state that must unfold to achieve the native protein. It has been shown previously that such an issue may be addressed only if all the microscopic rate constants are measured over a wide range of denaturant concentration (3,16,31). In the case of H. thermophilus cytochrome c 552 , interrupted refolding experiments allowed us to measure the unfolding limb of the intermediate, thereby defining its kinetic role. By this approach, the intermediate was populated to a considerable extent in the first mixing step and thereafter challenged with high and variable denaturant concentrations in the second mix to measure its unfolding rates. ics of H. thermophilus cytochrome c 552 monitored by fluorescence and depicts the dependence of the rate constants for the two identified phases over a broad range of denaturant concentrations. Global analysis of the kinetic data (see "Materials and Methods") is consistent with the observed intermediate being an on-pathway species in the folding to the native state, following the approach used for other proteins (3,16,31). The calculated Tanford ␤-value for this intermediate, reflecting the buried surface area relative to the unfolded (␤ T ϭ 0) and native (␤ T ϭ 1) states, is ␤ T ϭ 0.71 Ϯ 0.04. This value is close to that (␤ T ϭ 0.66 Ϯ 0.05) recently outlined to be characteristic of the cytochrome c family on the basis of extensive analysis of available kinetic data (17), adding general validity to the consensus folding mechanism proposed for the c-type cytochromes.
Folding Mechanism of Thermophilic and Mesophilic Cytochromes-As recalled above, the folding mechanism of P. aeruginosa cytochrome c 551 involves an energy barrier with at least two transition states (TS1 and TS2) and a high energy intermediate in between (15). On the other hand, our data show that a low energy intermediate is accumulated in the folding pathway of H. thermophilus cytochrome c 552 . Can we identify some structural determinants of the thermophilic protein responsible for the accumulation of a kinetically well resolved intermediate contrary to the homologous P. aeruginosa cytochrome c 551 ? The two proteins have essentially the same molecular weight and contain the same fraction of hydrophobic residues; moreover, they are structurally very similar (root mean square deviation of the backbone atoms ϳ 0.7 Å). The difference in their thermodynamic stabilities (⌬⌬G ϳ 2.9 kcal mol Ϫ1 at pH 4.7) suggests that, for H. thermophilus cytochrome c 552 , the stabilization of the native state and of the partially folded intermediate may be related to the same characteristic structural features. In this respect, it is interesting that, contrary to what was observed for mitochondrial cytochrome c (32), also the apoform of this thermophilic cytochrome c is surprisingly stable and partially folded; indeed, apocytochrome c 552 is able to bind the heme group in vitro, suggesting that it may contain a nascent heme binding site (33,34).
It should be noticed that, over and above the similarity of the main chain fold, several local conformational differences emerge from the comparison between these two bacterial cytochromes (18). Notably, cavity-filling effects due to a few selected substitutions were claimed to be responsible for the enhanced thermostability of the thermophilic protein (25). It is therefore possible that packing defects might be involved in controlling the stability of the folding intermediate(s), in addition to dictating overall protein stability.
Concluding Remarks-Two decades of extensive studies on protein folding have shown that proteins seem to fold alternatively via two distinct mechanisms: (i) the framework mechanism (35), whereby secondary structural elements form first, giving rise to high/low energy intermediates, followed by tertiary structure formation and (ii) the nucleation-condensation mechanism (36), whereby folding is driven by formation of a weak nucleus stabilized by long range interactions (nucleation) and simultaneously consolidated by extended structure formation (condensation), following a two-state process. Interestingly, a recent study on the homeodomain superfamily demonstrated that the framework and nucleation-condensation models are different manifestations of a common mechanism (37). Inherent stabilities of local structural elements modulate the folding process; as the propensity for forming secondary structure increases, the mechanism slides from the nucleationcondensation to the diffusion-collision model (38). This observation is mirrored by detection of two-state folding for proteins with low secondary structure propensity versus multistate fold-ing for proteins with high secondary structure propensity (11,37).
In the case of the c-type cytochromes, a comparison of inherent propensities of helical elements in P. aeruginosa cytochrome c 551 and H. thermophilus cytochrome c 552 is meaningful, given the notable structural homology between these two proteins. Travaglini-Allocatelli et al. (17) reported that, although a distinctly higher overall helical propensity for the thermophilic cytochrome can be excluded, the distribution of helical propensity is indeed different. In the case of H. thermophilus cytochrome c 552 , the C-terminal helix, which is part of a crucial folding nucleus in all c-type cytochromes (15, 39 -41), displays remarkably higher helical propensity compared with the mesophilic P. aeruginosa cytochrome c 551 . This observation, together with the reduced volume of packing defects, might explain the increased stability of the folding intermediate, which involves crucial contacts at the interface between the N-and C-terminal helices. Appropriate mutagenesis of the helical segments in the two proteins may allow verification of the hypothesis that accumulation of the intermediate depends on the balance of helical propensities for the three major ␣-helices in the cytochrome c fold.