Portal de Periódicos da CAPES

Folding kinetic pathway of phosphofructokinase-2 from Escherichia coli : A homodimeric enzyme with a complex domain organization

2011; Wiley; Volume: 585; Issue: 14 Linguagem: Inglês

10.1016/j.febslet.2011.05.041

1873-3468

Mauricio Báez, Christian A. M. Wilson, Jorge Babul,

Protein Structure and Dynamics

Pfk-2 binds to Pfk-2 by circular dichroism (View interaction) Phosphofructokinase-2 (Pfk-2) belongs to the ribokinase-like family of sugar kinases [1] whose dimeric members show an intricate architecture. Pfk-2 is a medium size homodimer (about 66 kD) and each subunit can be divided in two parts: a large α/β/α domain and an additional β-sheet structural element that protrudes from it (Fig. 1 ). Each β-sheet works as a lid for the active site and also creates the interface between the subunits, denominated β-clasp. In contrast with the typical subunits interfaces, the β-clasp seems like a bimolecular domain created by an orthogonal packing of each β-sheet forming an intertwined structure similar to a flatten β-barrel fold [2, 3]. However, the β-clasp and the α/β/α domain of each monomer do not seem to behave independently with respect to their conformational stability and structural integrity. Under equilibrium conditions Pfk-2 follows a three state mechanism of unfolding [N2 ↔ 2I ↔ 2U] with the accumulation of an expanded monomelic intermediate that retains a low content of secondary structure [4]. By this mechanism, most of the dimer stability and their native contacts are lost during the dissociation step [N2 ↔ 2I] supporting the fact that the β-clasp and an important portion of the α/β/α domain unfold cooperatively. However, equilibrium data alone are insufficient to describe the interdependence between the β-clasp and the α/β/α domain since the equilibrium cooperativity depends on the kinetic behavior [5]. Indeed, folding kinetic mechanisms are often more complex than their equilibrium counterpart, showing multiple pathways and transient intermediates whose formation not only depends on their overall stability but also on the kinetic barriers between their conformational steps [6-8]. Although folding pathways of oligomers can be as complex as those of monomeric proteins, a common theme is the coordination between the subunit interface formation and the folding process of each subunit [for a recent review see [9]]. In this respect, several structural properties at the oligomer interfaces have been proposed as indicators to predict the degree of folding during the interface formation, like large ratios of interfacial to intrasubunit contacts [10], high mean hidrophobicity [11, 12] together with the degree of structural similarity with respect to the hydrophobic core of globular proteins [13, 14]. Taking into account that the β-clasp interface of Pfk-2 resembles a β-barrel fold with a hydrophobic core, its formation is expected to follow a two-step folding mechanism. Nevertheless, there are additional constraints for its formation: the main chain that conforms the β-barrel domain is inserted discontinuously into the main chain of the large α/β/α domain of each subunit (Fig. 1). Thus, the β-clasp interface together with the α/β/α domain seem to be subject to mutual constraints to reach the native state of Pfk-2. The folding pathway of Pfk-2 and those of closer homologues have not been characterized and hence the transient intermediates and rate limiting steps involved in the formation of this kind of multidomain-dimeric architectures are unknown. In this work, the folding mechanism of Pfk-2 was characterized to determine the kinetic barriers and transient intermediates behind the thermodynamic coupling between the β-clasp interface and the α/β/α domain. Escherichia coli Pfk-2 was purified and stored as described by Babul in [15]. Previous to the unfolding or refolding experiments, the storage buffer was changed to the standard buffer (50 mM Tris pH 8.2 5 mM MgCl2 and 2 mM DTT) using a HiTrap desalting column (Amersham Biosciences, Uppsala, Sweden) or Micro Bio-Spin chromatography columns P-6 (Bio-Rad Laboratories, Inc., California, USA). The enzyme was concentrated with a centricon-60 concentrator (Amicon, Beverly, USA) and further concentration was archived using a microcon ultracel YM-10 (Millipore Corporation, Billerica, USA). Protein concentration was determined by the Bradford assay [16] using BSA as a standard and is expressed in terms of the monomer concentration. Circular dichroism measurements were performed with a 410 AVIV Biomedical, Inc., Lakewood, NJ, USA or a Jasco 810 Jasco Corp. spectropolarimeter, both equipped with a thermostated cell holder. Fluorescence measurements were performed with a Perkin Elmer RF-5301PC or a FluorMax-3 (Horiba Jobin Yvon) spectrofluorimeter. The equilibrium experiments were performed as in [4]. For the renaturation experiments, the enzyme was exposed to 3 or 1.5 M guanidine hydrochloride (GdnHCl) (Pierce, molecular biology grade, Rockford, IL, USA) for at least 5 or 24 h respectively at 20 °C. Under these conditions Pfk-2 was completely unfolded, as indicated by the absence of catalytic activity and CD and intrinsic fluorescence spectra. Kinetics of folding and unfolding reactions of Pfk-2 were obtained by manual mixing of unfolded or native protein with the standard buffer containing the desired chaotropic agent concentrations at 25 °C. The dead time of mixing was about 20 s. The temporal changes of the intrinsic fluorescence of Pfk-2 were followed at 346 nm with excitation at 295 nm with an integration time of 0.1 s and the temporal changes of circular dichroism were followed at 222 nm using a cell with an optical path of 0.1 cm. The integrated law of velocity for an irreversible bimolecular process of subunits association predicts a hyperbolic function with an apparent constant which depends on the protein concentration (Eq. (1)). This property, coupled with the temporal changes in spectroscopic properties, has been used to determine the occurrence of bimolecular rate-limiting steps during the folding pathway of several dimeric proteins [9, 17]. Fig. 2 shows a representative folding reaction of Pfk-2 monitored by intrinsic fluorescence (Fig. 2A) and CD measurements (Fig. 2B). The kinetic traces obtained either by CD or intrinsic fluorescence were better described by a hyperbolic function than by a single exponential function (see residuals distribution in upper panel of Fig. 2). Fitting to a single exponential function becomes worse as the final GdnHCl concentration is lowered. Fig. 3 shows the k obs values determined from either intrinsic fluorescence or CD measurements using several protein concentrations at 0.15 M GdnHCl. The k obs values determined by both methods were similar and increased with the protein concentration in a similar fashion. The kinetics traces fit well using a two exponential function, but the observed rate constants do not follow a defined pattern against the protein concentration, while adjusting to a bimolecular function gives a straight line when the observed constants were plotted against the protein concentration. These results indicate that both intrinsic fluorescence and CD measurements detect conformational changes that sense the same rate-limiting step of folding, which could involve the association of Pfk-2 subunits. However, folding reactions followed by CD show an additional unresolved phase during the dead time of the manual mixing experiments. The unresolved phase can be deduced from the differences between the signal calculated at zero time and the expected initial signal measured at 1.5 M GdnHCl (see arrow in Fig. 2B). Fig. 4 A compares the amplitude of the slow CD phase with its equilibrium counterpart. As shown, the resolved phase of folding accounts for about 65 % of overall change of ellipticity at 222 nm measured between the native dimer “N2” and the unfolded monomer “U”. Therefore, it is likely that the burst phase reflects the endpoint of a fast transition from for the unfolded state to the transient intermediate step. Stopped-flow experiments, designed to detect the formation of the intermediate species, were unable to resolve the burst phase when samples were diluted from 3 M GdnHCl to concentrations where the intermediate is observed (between 0.5 and 2 M, data not shown). On the other hand, the intrinsic fluorescence increment observed in the folding kinetics equals the difference in the fluorescence obtained between the native and the unfolded states of Pfk-2 measured under equilibrium conditions (Fig. 4B). Thus, the transient intermediate detected by CD cannot be distinguished from the unfolded state through fluorescence measurements since the fluorescence amplitudes of the folding reactions are the same under conditions where the intermediate is populated and where is not. This characteristic has been described for the monomeric intermediate detected under unfolding equilibrium conditions of Pfk-2 [4]. The unfolding reactions measured by CD and intrinsic fluorescence at 2 M GdnHCl are shown in Fig. 5 A and B, respectively. As is observed, the decrease of the CD signal at 222 nm is poorly described by a single exponential function but can be fitted to the sum of two exponential functions (although residuals still show a small degree of inequality, a third phase provokes function over-parameterization). The kinetic constants of both phases were 1.1 × 10−3 and 2 × 10−2 s−1, where the amplitude of the faster one (k u-fast) is about the 40% of the overall change of the CD signal between 1.5 and at 2.5 M of GdnHCl (data not shown). Over 2.5 M GdnHCl the kinetic traces can be well fitted to a single exponential function and hence the amplitude of the fast reaction could not be calculated. This suggests that over 2.5 M GdnHCl, the fast phase occurs within the dead time of mixing (15–20 s) and explains the decrease of the overall kinetic amplitude observed over 2.5 M GdnHCl with respect to the expected base line of the equilibrium transition (Fig. 4A). Unlike the two phases observed by CD, unfolding kinetics followed by intrinsic fluorescence (Fig. 5B) can be adjusted using a single exponential decay function with a kinetic constant close to the slowest unfolding constant measured by CD (k u-slow = 1.03 × 10−3 s−1). No burst phase was detected by intrinsic fluorescence; the kinetic amplitudes cover quite well the equilibrium base lines (Fig. 4B). As is observed in Fig. 6 , the kinetic constants calculated from the unfolding and folding curves follow a semilogarithmic behavior with respect to the GdnHCl concentration. Consequently, Eq. (2) was used to calculate the kinetic parameters shown in Table 1 . The folding pathway of Pfk-2 was characterized by two kinetic phases but only one of them could be resolved. The rate constants calculated from the folding reactions observed either by fluorescence or CD measurements were quite similar between 0.1 and 0.3 M GdnHCl (Fig. 6) and hence both probes yield similar kinetic parameters (Table 1). In average, the variation of the folding constants with the denaturant concentration yields bimolecular constants of 6 × 10−3 s−1 M−1 at zero GdnHCl concentration together with denaturant dependences (m f values) of 11 kcal mol−1 M−1. On the other hand, unfolding of Pfk-2 involves two phases characterized by a slow (k u-slow) and fast (k u-fast) kinetic constants. CD and intrinsic fluorescence-monitored values of k u-slow appear to be similar between 1.5 M and 4.5 M GdnHCl and yield similar kinetic parameters (Table 1). Thus, the slow phase of unfolding is characterized by a kinetic constant at zero molar GdnHCl () of about 5 × 10−5 s−1 and a denaturant dependence of 0.89 kcal mol−1 M−1. The fast unfolding step, which is only detected by CD measurements, was characterized by values of and denaturant dependence (m u-fast) slightly higher than the kinetic parameters calculated from the slow step of unfolding (Fig. 6). The estimated values of k u-fast and m u-fast were 1.1 ± 0.3 × 10−4 s−1 and 1.4 ± 0.05 kcal mol−1 M−1, respectively. Folding of Pfk-2 can be described by two processes: a rapid accumulation of an intermediate detected as a burst increment on the ellipticity at 222 nm (during the dead time of mixing) without changes in the intrinsic fluorescence of the unfolded state, and a slow phase which amplitude describes the N2 ↔ 2I equilibrium transition. Thus, most part of the ellipticity at 222 (about 65%) and the full intrinsic fluorescence of Pfk-2 were regained simultaneously during the slow phase of folding (compare the values of the kinetic amplitudes in Fig. 4). Since the N2 ↔ 2I transition accounts for about 90% of the equilibrium ΔASA of Pfk-2 (17–19 kcal M−1 mol−1) and most part of the dimer stability (about 75%) [4], the slow phase of folding is expected to involve the formation of most of the native contacts of Pfk-2. Moreover, and taking into account the positive dependence of the slow phase with the protein concentration, the acquisition of most part of the native contacts of Pfk-2 seem to be rate-limited by the association step. In this sense, the association step of Pfk-2 resembles the two-sate behavior of association and folding of small dimers stabilized by a bimolecular hydrophobic core such as the ARC repressor [18, 19] and others dimers [9]. The localization of the single tryptophan (trp-88) also supports this hypothesis. In the Pfk-2 structure obtained with MgATP, each subunit presents a tryptophan localized at the hinge that connects the β-clasp with each α/β/α domain and is close to the subunit interface hiding 128 Å2 with the partner subunit (Fig. 1, [3]). In this way, to reach the native environment of trp-88, non local folding that involves the β-clasp interface and its correct interaction with the α/β/α domain must occur. In agreement with these structural features, most part of the secondary structure is regained along with the intrinsic fluorescence and both processes depend on the protein concentration. Several features of subunit interfaces, such as hydrophobic character, hidden surface and geometric considerations, have been used to predict the folding mechanism of oligomeric proteins [11, 12]. However, in the framework of the minimal frustration model [20] the mechanism and the transition state of protein folding can be predicted by the topology of the native state. Accordingly, coarse grained simulations performed with potentials reproduce the observations on whether folding and binding are coupled in one step or whether intermediates occur during the folding pathway of several small homodimers [10]. The overall architecture of Pfk-2 and closer homologues can be described as a two-domain dimer with an important topological feature. A large part of the α/β/α domain together with the β-clasp interface of Pfk-2 can be considered as discontinuous domains since the backbone of both β-strands that form the β-clasp interface interrupt the polypeptide chain of the α/β/α domain in two places (Fig. 1). This reentrant topology would originate a mutual constrain between the interface (the β-clasp structure) and folding of the large domain, which in turn could prevent partial folding of Pfk-2 and hence the presence of folded monomer intermediates. A reentrant domain connection is also observed between the C and N terminal domains that conform the T4-lysozyme. The release of this topological constrain upon circular permutation of the T4-lysozyme decouples the folding pathway allowing the C-domain to fold independently of the rest of the protein [21]. Events that occur initially during unfolding can be expected to be similar to the final events during folding, and hence it is interesting that the unfolding kinetics of Pfk-2 show deviations from a two-state mechanism. Unfolding kinetics of Pfk-2 were described by two events characterized by constants of 5 × 10−5 and 1 × 10−4 s−1 (values extrapolated at zero molar GdnHCl, Table 1). Both phases were apparent from kinetic traces measured by CD at 222 nm, but through the fluorescence kinetic traces it was only possible to detect the slower one. In the framework of a parallel mechanism of unfolding, the fast and slow reactions would arise from parallel unfolding reaction originated from two forms of Pfk-2 present under native conditions: one form devoided of intrinsic fluorescence and another form with the native environment of the trp-88, i.e. the native dimer. In this scenario, shifting the equilibrium between both forms would modify the relative amplitude of the slow and fast phases of unfolding (see examples in [22]). Stabilization of Pfk-2 with ligands or salts did not modify the relative amplitude of unfolding kinetics detected by CD (Supplementary data, S1). There were also no changes in the relative amplitudes when the enzyme was destabilized with GdnHCl or supplemented with dithiotreitol (DTT) [data not shown]. These results argue against the hypothesis that parallel pathways of unfolding would originate two kinetic phases during unfolding of Pfk-2. Other scenery could be a sequential mechanism of unfolding. In this case, the rapid phase can be interpreted as a loss of secondary structure that occurs without changes in the intrinsic fluorescence of Pfk-2 and hence would indicate the accumulation of a partially unfolded dimeric intermediate that retains the intrinsic fluorescence of the native dimer. Furthermore, this intermediate could be unfolded and dissociated along the slow phase to originate the unfolded state or the monomeric intermediate species observed under equilibrium conditions. This situation can occur if part of the large domain is disrupted prior the dissociation and unfolding step. Interestingly, Cabrera et al. have postulated that the large α/β/α domain of each subunit can be divided into two modules: the Rossmann module, interrupted by the insertion that originates the β-clasp interface, and the β-meander module localized at the C-terminal of the α/β/α domain ([3], Fig. 1). Thus, since the β-meander module is not interrupted by the β-sheet that originates the β-clasp it could unfold independently from the dissociation of Pfk-2. The amplitude of the N2 ↔ 2I equilibrium transition is well described by kinetic amplitudes measured by intrinsic fluorescence and hence is plausible a correspondence between equilibrium and kinetic parameters by means of the microscopic reversibility principle. Using the kinetic parameter values from the folding and the slow unfolding steps observed by intrinsic fluorescence or CD, a free energy difference of about 11 kcal mol−1 is obtained. This value is close to the value obtained for the N2 ↔ 2I equilibrium transition (12 ± 1 kcal mol−1, Table 1) suggesting that the folding and unfolding reactions measured by intrinsic fluorescence behave as elementary kinetic steps of the N2 ↔ 2I reaction. In this framework, the m f and m u-slow values could be used as a tool to evaluate the compactness of the transition state for the association step of Pfk-2 by means of the β T value. The β T value measures the average fractional change in the degree of exposure of residues between the initial and the transition states of a folding reaction [23, 24]. A β T value of about 0.8 was determined considering the kinetic m-values (m f/(m f + m u-slow)) obtained from the slow unfolding transition and a lower value of β T ∼ 0.6 when the m eq of the N2 ↔ 2I transition was used (m f/m eq). This difference occurs because the average m eq value for the N2 ↔ 2I transition (17–19 kcal mol−1 M−1) is close but larger than the expected m value calculated from kinetic measurements (about 13 kcal mol−1 M−1). In spite of these differences, the β T values for the N2 ↔ 2I step suggest that the transition state is closer to the native dimer in terms of the fractional degree of accessible surface area (ASA) or degree of compactness. Since the m eq of the N2 ↔ 2I transition accounts for about 90% of the ΔASA value of unfolding [4], the transition state of the association step would involve most part of the ΔASA of the native state of Pfk-2. By inference, a large portion of the structure of Pfk-2 seems to be formed at the transition state in agreement with the mutual constraint impose by its interrupted domain architecture. Nevertheless, this analysis should be taken with caution. In absence of a comprehensive mechanism of folding, the above mentioned kinetics constants are apparent values and hence is unknown how the microscopic rates and m-values are connected along with the overall folding or unfolding pathway of Pfk-2. In order to obtain a comprehensive folding mechanism it is necessary to fit the entire set of kinetic data with specifics models of folding. A global fit procedure was performed with several models involving one or two on-pathway or off- pathway monomeric or dimeric intermediates using the software COPASI [26]. We were unable to find a full correspondence between kinetic traces and kinetic models, probably because the lack of data to describe the fast folding transition. However, models containing on-pathway intermediates show better fit than models with off-pathway intermediates, particularly models with a dimeric intermediate (Supplementary data, S2). Further experiments are necessary to describe the rapid formation of the intermediate and thus obtain a comprehensive mechanism of folding for Pfk-2. Folding experiments show an unstructured monomeric intermediate with similar spectroscopic properties to the intermediate observed under equilibrium conditions. However, most part of the dimer structure is reached as a concerted folding/association event represented by a slow step with a transition state quite folded in terms of solvent exposure. On the other hand, unfolding kinetics show the accumulation of a transient intermediate which spectroscopic properties that cannot be inferred from the equilibrium experiments. Assuming that the unfolding intermediate is formed on-pathway and taking into account the key localization of trp-88, a partially unfolded dimer could be transiently accumulated during Pfk-2 unfolding. These characteristics could be a consequence of the reentrant topology that connects the β-clasp interface with the Rossmann module that conform each subunit in this kind of dimers. This work was supported by a grant from the Comisión Nacional de Investigación Cientifica y Tecnológica, FONDECYT 1090336, Chile. CAMW is recipient of a Comisión Nacional de Investigación Científica y Tecnológica (CONICYT) graduate fellowship and a Beca de Apoyo a Tesis Doctoral. We thank Susan Marqusee, the Marqusee laboratory, University of California Berkeley and Richard Garratt, Laboratorio de Cristalografía, Instituto de Fisica de São Carlos, Universidad de São Paulo, Brazil, for the use of their instrumentation facilities, and Sergio Kaufman, Universidad de Buenos Aires, Facultad de Farmacia y Bioquímica, for his help with the COPASI software and curve fitting. Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.febslet.2011.05.041. Supplementary data. Supplementary Table. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.