|Home | About | Journals | Submit | Contact Us | Français|
Ribonuclease A (RNase A) undergoes more rapid conformational folding with its disulfide bonds intact than during oxidative folding from its reduced form. In this study, the effect of the mutants Y92G, Y92A, and Y92L on both the conformational and oxidative folding pathways was examined in order to determine the role of native interactions in different types of conformational searches to obtain the biologically active structure of a protein. These mutations did not affect the overall conformational folding pathway of RNase A. However, in the mutants Y92G and Y92A, a key structured disulfide-bonded species, des-[65-72], involved in the oxidative folding pathway of RNase A, was destabilized. These results demonstrate the importance of native interactions in the folding process, namely, protection of a native (40-95) disulfide bond by a nearby tyrosyl-prolyl stacking interaction, when disulfide bonds are allowed to undergo SH/S-S reshuffling.
The process of how the four-disulfide-bond-containing protein RNase A† obtains its biologically active or native structure from an unfolded state has been studied extensively (1-9). The goal of these studies was to understand how interactions among amino acids drive the conformational search to fold a protein into its native structure. Acquiring native structure in disulfide-intact proteins is referred to as conformational folding. Acquiring native structure, in addition to the correct disulfide-bond-linkages in the presence of an oxidizing agent from a state in which the disulfide bonds are reduced, is referred to as oxidative folding. The goal of this study is to understand how these two different folding mechanisms utilize some of the same native interactions. In the conformational space that is searched in the disulfide-intact folding mechanism, disulfide bonds restrict the freedom of the backbone, and, as a result, folding takes place in minutes. In oxidative folding, the backbone is less restricted and the search of conformational space is influenced by the formation of disulfide bonds. In contrast to conformational folding, oxidative folding takes place in hours. Therefore, the degree to which the same side-chain/side-chain interactions drive the folding process is different in these two types of folding mechanisms.
The conformational folding pathway of RNase A has been studied extensively in many laboratories (1-4) and in our own (5-7). An equilibrium population of different unfolded states appears in GdnHCl-denatured RNase A, U-RNase A, because of proline isomerization (10, 11); each state folds concomitantly to N at different rates when the GdnHCl is rapidly diluted to impose folding conditions.
The oxidative folding mechanism by which RNase A concomitantly attains its correct disulfide linkages as well as its biologically active structure has been identified (8, 9), Figure 1. In the presence of the redox couple, DTTox/DTTred, a pre-equilibrium is first established among unstructured ensembles that contain zero, one, two, three or four native and non-native disulfide bonds. Through a redox independent SH/S-S reshuffling step, two structured species, des-[65-72] and des-[40-95] (8, 9), are preferentially populated from the unstructured 3S ensemble in the rate-determining step along two different pathways (8, 9), and then oxidized rapidly to N-RNase A.
In order to observe how these two mechanisms may differ in their utilization of side-chain/side-chain interactions, tyrosine 92 was mutated to remove a ring-stacking interaction with proline 93, and the extent of the effects of this mutation on the oxidative folding and conformational folding pathways relative to those of wild-type RNase A is measured and compared to that of the wild-type protein.
Previous work (12, 13) with mutations of tyrosine 92 showed that they affect the redox chemistry around the [40-95] disulfide bond. In the crystal structure (14) in Figure 2, Y92 and P93 form a ring-stacking interaction in which the tyrosine residue covers the (40-95) disulfide bond. By removing residue Y92 in the mutants, the (40-95) disulfide bond becomes more exposed. In a previous study (13), the mutation at Y92 to G, A or L affected the reductive unfolding pathway of RNase A whereby the [40-95] disulfide bond was preferentially reduced over the (65-72) disulfide bond, i.e., the rate of reduction of the (40-95) disulfide bond is increased, and des-[40-95] is the only three-disulfide-bond species observed on the reductive unfolding pathway. For wild-type RNase A, both the (40-95) and (65-72) disulfide bonds are reduced along two different reductive unfolding pathways (15). By making the [40-95] disulfide bond more easily reducible in the mutants, the oxidative folding of mutant RNase A is also affected. The mutant Y92G of RNase A recovers its biologically active form more slowly than WT RNase A (12). A similar increase in the reducibility of the (40-95) disulfide bond is also observed in the mutant P93A (16). In this mutant, the ring-stacking interaction was not present to help residue Y92 protect the (40-95) disulfide bond, thus making the latter more reactive.
In order to make a comparison of a conformational search between the oxidative folding mechanism and the disulfide-intact folding mechanism, the effects of these mutants on the redox-independent reshuffling of the 3S ensemble of native and non-native disulfide bonds to des-[65-72] and des-[40-95] was studied. In this step of the oxidative folding pathway of WT RNase A (Figure 1), the unstructured ensemble, 3SU, undergoes SH/S-S reshuffling along two different pathways to populate des-[65-72], and des-[40-95], both of which have native-like structure (17, 18), and oxidizes rapidly to form N-RNase A. In the rate-determining step that forms native-like structure in the oxidative folding pathway (9), the 3SU → 3SF reaction involves reshuffling but no oxidation or reduction. Our focus was placed here on the conformational folding in the transition from 3SU to 3SF. Therefore, the effects of these mutants on this conformational folding in oxidative folding pathways can be compared with conformational folding of disulfide-intact folding pathways.
Desired mutations in RNase A were achieved by PCR using the QuikChange Site-Directed Mutagenesis Kit from Stratagene. The required designed primers for PCR were obtained from Sigma-Genosys.
GdnHCl, required for unfolding of the proteins in kinetic studies, was obtained from Mallinckrodt Chemicals. Tris(hydroxymethyl)aminomethane hydrochloride (TRIZMA HCl, 99.9%), used in pH 8 buffers for kinetic studies, was obtained from Sigma-Aldrich. EDTA, used in buffers for kinetic studies, was obtained from Fischer Scientific. All other materials were obtained as the highest grade available.
Mutant primers were designed for site-directed Y92 substitutions in order to generate DNA plasmids for the RNase A mutants by PCR. The plasmid for wild-type RNase A in a pET22b(+) vector, obtained from previous studies (19), was used as a template for PCR along with the mutant primers. The protocol for PCR was taken from the QuikChange Site-Directed Mutagenesis Kit from Stratagene. Post-PCR, the sequencing of all the resultant mutant DNA plasmids was carried out at the Cornell Biotechnology Resource Center. The plasmid DNAs for the correct desired tyrosine-to-leucine, tyrosine-to-alanine, and tyrosine-to-glycine substitutions were used for protein over-expression and purification. Wild-type RNase A (Type 1-A, Sigma-Aldrich) was purified by ion exchange chromatography (13). The protocols for the over-expression and purification of the mutant proteins from inclusion bodies have been described earlier (19).
3SU ensembles of RNase A and Y92 mutants were produced by first populating a three-disulfide-bond-containing intermediate species from reductive unfolding of each protein. For wild-type RNase A, the two species from N-WT-RNase A each with three native disulfide bonds, des-[65-72] and des-[40-95], was produced by incubating N-WT-RNase A in a 0.1 M Tris buffer at a concentration of 1 mg/mL at 15 °C for 12 hours at pH 8.0 in the presence of 100 mM DTTred (15). For the mutant Y92A, the des-[40-95] species from N-Y92A-RNase A was produced by incubating N-Y92A-RNase A under the same conditions as used for wild-type RNase A except at a concentration of 0.5 mg/mL for 160 minutes (13). For the mutant Y92G, N-Y92G-RNase A was incubated under the same conditions used for N-Y92A-RNase A except that the reductive unfolding was allowed to proceed for 50 min (13). After the reductive unfolding reaction proceeded for the prescribed amount of time for each protein, the reaction was quenched by addition of glacial acetic acid.
The species of each protein with three native disulfide bonds was purified from the reductive unfolding mixtures by RP-HPLC using a water/acetonitrile buffer system with 0.09% TFA using a 25 cm × 4.6 mm SULPELCO Discovery® BIO Wide Pore C18, 5 μm particle size column with the eluted species monitored at 210 nm. In the procedure used, buffer salts, DTTred and DTTox eluted first with 85 % water/ 15 % acetonitrile, followed by the protein using a gradient of 25 - 34 % acetonitrile over 55 minutes. This gradient was sufficient to separate des-[65-72] and des-[40-95] from each other and from the native and reduced forms of both Y92A-RNaseA and Y92G-RNase A. All of these purified species were collected separately, frozen, and lyophilized to remove TFA, water and acetonitrile. After lyophilization, each these three-disulfide species were reconstituted in 50 mM acetic acid and stored in -70 °C.
To generate an unstructured ensemble, 3SU, of Y92A-RNase A and of Y92G-RNase A, the solutions of purified des-[40-95] from Y92A-RNase A and Y92G-RNase A were scrambled separately to the 3SU ensemble by incubating the solutions of des-[40-95] from each protein in a pH 8.2 0.1 M Tris buffer containing 5.6 M GdnHCl for 2 hours over humidified argon at a protein concentration of 0.5 mg/mL to allow SH/S-S reshuffling to form non-native disulfide bonds. The 3SU ensemble of WT-RNase A was generated similarly from des-[40-95] and des-[65-72] except that solutions containing purified des-[40-95] and des-[65-72] were combined before incubation in the same buffer. These buffer conditions, pH 8 and 5.6 M GdnHCl, are sufficient to unfold the native forms of WT-RNase A, Y92A-RNase A and Y92G-RNase A and, presumably, their less stable des species. The buffer containing 5.6 M GdnHCl was previously purged of oxygen with humidified argon. After two hours, the SS/S-H reshuffling reaction was stopped by adding glacial acetic acid to lower the pH to 3. GdnHCl and buffer salts were removed from the scrambling mixtures on a G25 size-exclusion column equilibrated with 50 mM acetic acid, and the eluting species were monitored at 280 nm. Acetic acid was removed from the 3SU ensemble by lyophilization. After lyophilization was complete, the 3SU ensemble for each protein was reconstituted in 50 mM acetic acid at a concentration of 5 mg/mL and stored at -70 °C until further use.
The 3SU to 3SF transition was initiated by diluting a 5 mg/mL stock solution of 3SU with a 0.1 M pH 8.0 Tris buffer, previously purged of oxygen with humidified argon, to a final protein concentration of 0.5 mg/mL, and the transition was allowed to proceed for 40 minutes at pH 8.0 under humidified argon, with the reaction progress determined every 5 minutes.
To determine the amount of structured species that was present in the reaction mixture at different times, quantitatively, an aliquot was removed and was subjected to a reduction pulse (8) with 5 mM DTTred. The 5 mM DTTred was allowed to reduce the unstructured three-disulfide-bond-containing species for two minutes to the reduced form of each of the proteins, and the reduction was then quenched with glacial acetic acid and stored frozen at - 70 °C until analysis by RP-HPLC. Under these conditions, structured three-disulfide-bond-containing species are not reduced by DTTred. Each of the aliquots (at different reshuffling times) was analyzed by RP-HPLC as described above. The progress of the transition was followed quantitatively by integrating the area under the chromatographic curves of the structured species and the reduced protein, with the reduced protein corresponding to the unstructured species that were reduced by DTTred during the reduction pulse. The species involved in the transition, the 3SF species and R, were detected at 210 nm and each was assumed to have the same absorptivity because, under the conditions of separation by RP-HPLC, all the species are unfolded. Using the precautions against oxidation by oxygen for the duration of the experiment, a maximum air oxidation of species of no more than 3% was observed.
According to the mechanism in Fig. 1, the system of differential equations that describes the formation of two structured 3SF species from the unstructured 3SU ensemble are:
where k1 and k2, and F1 and F2 correspond to the two structured des species. The analytical solution for this system of differential equations is:
where [3SU]0 is the initial concentration of 3SU. For WT-RNase A, as shown in Fig. 1, the two structured 3SF species are des-[65-72] and des-[40-95]. Since the progress of this reaction was followed quantitatively by using RP-HPLC to calculate the relative fraction of each species, with the quantity [3SU]0 taken as 1.0.
While it was shown previously (15) that these two 3SF species have a tendency to reshuffle back to the unstructured 3SU ensemble, the rate of this back reshuffling from 3SF to 3SU is much slower than the reshuffling of the 3SU ensemble to the 3SF species (15). This was verified by obtaining the solutions of the system of differential equations, 1-3, with and without additional back reshuffling reactions for the 3SF species. There was no discernable difference between the time-dependent concentrations of the two 3SF species with and without back reshuffling, within the time used to study this transition. Therefore, the back reshuffling of the 3SF species to the 3SU ensemble was not taken into account in this analysis.
To identify the 3SF species formed in the SH/S-S reshuffling of 3SU to 3SF, the Wu-Watson cyanylation method was used to identify the free cysteines (20). The disulfide-bond connectivity of the wild- type and mutant forms was determined to be native because the 3SF species produced in this transition for each of these proteins have the same chromatographic behavior (8, 15) as the corresponding species that are also populated in the reductive unfolding pathways of the native wild-type and mutant proteins.
All refolding studies were carried out at 15 °C in order to observe the entire kinetic processes by minimizing burst phase kinetics, if any. All single-jump measurements on the native forms of the wild-type and the Y92 mutant proteins, N-WT-RNase A, N-Y92G-RNase A, N-Y92A-RNase A, and N-Y92L-RNase A, were carried out on a stopped-flow mixing module from HI-TECH Scientific using absorbance at 287 nm as the optical probe. For refolding studies, the proteins were first unfolded in unfolding buffer (5.9 M GdnHCl, 0.1 M Tris, 1 mM EDTA, pH 8.0) at room temperature for 4 hours prior to refolding to ensure complete unfolding, and then equilibrated at 15 °C prior to the start of kinetic experiments on the stopped-flow mixing module. Each unfolded protein was diluted 11-fold in refolding buffer (0.1 M Tris, 1 mM EDTA, pH 8.0, 15 °C) on the stopped flow mixing module so that the final refolding condition was pH 8.0 and 0.53 M GdnHCl, 15 °C. The refolding kinetics were acquired on a log-time base in which the acquisition is faster in the millisecond timescale and slower in the second timescale, thereby allowing accurate measurements to be made in the fast kinetics regime. The procedure described earlier (21) was used to minimize viscosity effects and thereby ensure proper mixing at the cost of an increase in dead time. An increase in the dead time did not give rise to any burst-phase kinetics since the mixings were carried out at 15 °C.
The observable refolding kinetics of the wild-type and the Y92 mutant proteins under the conditions mentioned above are described by a four-exponential process when monitored by absorbance at 287 nm by the following equation:
where a(t) and a(∞) are the observed amplitudes at times t and at infinite time, respectively; k1, k2, k3, and k4, are the rate constants of the fast, medium, slow and very slow phases, respectively; and a1, a2, a3, and a4 are the respective amplitudes. The very fast phase cannot be distinguished from the fast phase in a single-jump experiment but only in a double-jump experiment (5).
Analysis of the effects of these mutants should recognize that the 3SU to 3SF transition was examined here, independent of the rest of the oxidative folding pathway shown in Fig. 1. Irrespective of whether the 3SU ensemble is populated in the oxidative folding pathway or is produced by the methods introduced here, it is still an unstructured ensemble; however, the distribution of species in the 3SU ensemble, and the transition to 3SF, could differ between these two circumstances. When the 3SU ensemble is populated in the oxidative folding pathway and participates in a pre-equilibrium, some of the species are reduced to populate the 2S ensemble or oxidized to populate the 4S ensemble. These redox reactions could affect the distributions of the 3SU ensemble. There were no oxidizing or reducing agents present in the investigation carried out here. According to Table 1, there is a difference in the rate constant associated with the formation of the des [40-95] species in this study, 5.9 ± 0.3 × 10-3 min-1, and that calculated in ref. 9, 1.4 ± 0.1 × 10-2 min-1. However, there is essentially no difference in the rate constant associated with the formation of des-[65-72]; the value determined in this study was 1.8 ± 0.1 × 10-3 min-1 while that reported in ref. 9 is 2.1 ± 0.2 × 10-3 min-1. This discrepancy in k1 could arise from the presence of DTTox and DTTred during oxidative folding, but their absence in the present study of the isolated 3SU to 3SF reshuffling. Since des-[40-95] and des-[65-72] are formed from 3SU in different pathways (9), with different inter-residue interactions in each pathway, it is possible that the absence of this discrepancy in the k2 pathway is due to such different inter-residue interactions. Hence, given that the conformational search in the 3SU ensemble in the present study or in the oxidative folding pathway finds two structured species, des-[40-95] and des-[65-72], with rate constants k1 and k2 within the same magnitude, the results of the present study can be applicable to effects in the oxidative folding pathway. Yet, the effects of the mutants Y92A-RNase A and Y92G-RNase A on the conformational search within 3SU should be compared to the results under the same conditions used for the wild-type protein, i.e., the conformational search within the 3SU ensemble should be independent of the oxidative folding pathway.
The Y92A-RNase A and Y92G-RNase A mutants affect the 3SU to 3SF transition of RNase A such that only one 3SF species is observed within the time-scale of the reaction. However, within the same time-scale, both des-[65-72] and des-[40-95] are observed for wild-type RNase A. Figure 3 shows three chromatograms for 24-minute transitions of Y92A, Y92G and wild-type RNase A. After 24 minutes of SS/S-H reshuffling, des-[65-72] and des-[40-95] are detectable for the wild-type protein, but only des-[40-95] is observed for the mutants Y92A and Y92G. As seen in Table 1, the rate constants calculated for the formation of des-[40-95], 7.0 ± 0.4 ×10-3 min-1 for Y92A-RNase A and 9.1 ± 0.6 × 10-3 min-1 for Y92G-RNase A, are comparable to the rate constant calculated for WT-RNase A, 5.9 ± 0.3 × 10-3 min-1. A rate constant for the formation of the des-[65-72] species was not calculated for this species for each of the mutants, Y92A-RNase A and Y92G-RNase A, because des-[65-72] was not observed.
While the value of the rate constant for the formation of des-[65-72] in wild-type RNase A is much smaller than that for des-[40-95], the des-[65-72] species is still important for the formation of N-RNase A because 20 % of the biologically-active form of RNase A is obtained from direct oxidation of des-[65-72] (9). This des species and des-[40-95] each have substantial native-like structure (17, 18). Therefore, the observation that this species is destabilized in these mutants is not trivial.
These Y92 mutants affect local structure during the conformational folding pathway of disulfide-intact RNase A instead of disrupting more global structure as seen in the 3SU to 3SF transition. According to Figure 4, each protein, including the mutants and wild-type, recovers the biologically active form in the same amount of time. However, these mutations do affect the slow phase, US, involved in the conformational folding pathway because of the isomerization of Pro93 (5).
According to column 4 of Table 2, as the size of the side chain of the mutated residue became smaller in size, L → A→ G, the time constant associated with the folding of the slow phase decreases, which corresponds to an increase in the rate of folding for this phase. By decreasing the size of the side chain that interacts with Pro93, Pro93 isomerizes at an increased rate because the steric hindrance between this residue and the side chain at residue 92 is lower. Since isomerization of Pro93 from the non-native trans conformer to the native cis conformer is a rate-limiting event in the folding of the US phase, an increase in the rate of isomerizaiton leads to an increase in the rate of folding for the US phase. Slower isomerization of Pro93 due to steric hindrance and accompanying slower rates of folding can be seen in the slow-phase folding in WT-RNase A and the large-leucine mutant, Y92L-RNase A, with time constants of similar magnitude (column 4).
Despite these effects on the slow phase, the time constants and amplitudes of the other phases are unaffected by mutation of Y92, as seen in the rest of the columns in Table 2. The diversity of unfolded states that gives rise to different phases in the folding pathway is due to the isomerization states of different prolines in RNase A. In phases other than the slow phase, Pro93 is in its native cis state. Removing the tyrosyl-prolyl interaction with the cis Pro93 does not affect the rate of folding for the other phases. This means that this interaction is not critical to the conformational folding of disulfide-intact RNase A and removing this interaction affects species that are only locally involved in the folding pathway.
While a structured intermediate in the mutants Y92A and Y92G is destabilized compared to that of the wild-type protein in the 3SU to 3SF transition, the mutants Y92A, Y92G and Y92L affect a species only in the unfolded state locally in the conformational folding pathway. Therefore, the 3SU to 3SF transition is more sensitive to changes in native interactions than the conformational folding pathway. Native interactions become more important when disulfide bonds are allowed to interchange rather than when they are fixed during the folding process.
As mentioned in the “Introduction”, Y92 and P93 form a ring-stacking interaction which shields and protects the (40-95) disulfide bond from reduction (13). In the 3SU to 3SF transition, thiolates attack disulfide bonds during the SH/S-S reshuffling. Therefore, this ring-stacking interaction can prevent reshuffling reactions from thiolates. In order to stabilize des-[65-72], each of its disulfide bonds, including the (40-95) disulfide bond, must be stabilized. Since this disulfide bond is stabilized by this ring-stacking interaction in the wild-type protein, removal of this ring-stacking interaction removes the extra protection of the (40-95) disulfide bond from thiolate attack. Once this ring-stacking interaction was removed, the (40-95) disulfide bond could not be stabilized and, thus, the structured species, des-[65-72] could not be populated within the time observed for this experiment during which the (40-95) disulfide bond is preferentially reduced. For des-[40-95], this species lacks the (40-95) disulfide bond, and protection of this disulfide bond is not required for its stabilization. As a result, des-[40-95] is observed in the 3SU to 3SF transition in the absence of this ring-stacking interaction.
In the 3SU to 3SF transition in the structural homolog ONC, only one 3SF species is populated, des-[30-75] (22), which is structurally homologous to des-[40-95] in RNase A. In the native structure of ONC, Y92 is replaced by R73 which does not participate in any interaction that would shield this disulfide bond (23). In fact, the (30-75) disulfide bond is approximately 4 times more exposed in ONC than the (40-95) disulfide bond in RNase A (23). Therefore, as ONC undergoes the 3SU to 3SF transition, the (30-75) disulfide bond is the least protected and thus cannot stabilize a 3SF species other than des-[30-75].
In the conformational folding pathway of disulfide-bond intact RNase A, the loss of this ring-stacking interaction does not affect other interactions that are required for U- RNase A to fold; the time constants and amplitudes of the other phases for the Y92 mutants and wild type RNase A listed in Table 2 are virtually unchanged by the removal of this ring-stacking interaction. Therefore, there are enough interactions other than the ring-stacking interaction of P93 and Y92 for RNase A to fold into its biologically active form.
By observing how the removal of a tyrosyl-prolyl stacking interaction affects the oxidative folding and disulfide-intact conformational folding of RNase A, this study demonstrates how native interactions become more important along a folding pathway when SH/S-S reshuffling is allowed to occur. This reshuffling allows more conformational freedom in the backbone during a folding pathway; as a result, more favorable interactions between the side-chains are required as the native structure is formed.
This research was supported by NIH grant GM - 24893
†Abbreviations: RNase A, wild-type bovine pancreatic ribonuclease A; N-x-RNase A, N indicates that native disulfide bonds are intact, x indicates a Y92 mutant or wild-type protein; R-RNase A, RNase A without disulfide bonds; U-RNase A, disulfide-intact RNase A denatured by GdnHCl; 3SU, unstructured ensemble of species containing three native and/or non-native disulfide bonds; 3SF, three-disulfide structured species with only native disulfide bonds. Des-[x-y], a disulfide-bond species that contains native disulfide bonds except for the (x-y) disulfide bond; GdnHCl, guanidine hydrochloride; TFA, trifluoroacetic acid; Tris, tris(hydroxymethyl)aminomethane; DTTox and DTTred, the oxidized and reduced forms, respectively, of dithiothreitol; EDTA, Ethylenediamine tetraacetic acid; PCR, Polymerase Chain Reaction.