Peptidyl-prolyl isomerization reactions can make for rate-limiting steps in protein folding due to their high activation energy. Onconase, an unusually stable ribonuclease A homologue from the Northern leopard frog, contains four trans proline residues in its native state. During the refolding from its guanidine hydrochloride unfolded state, which includes the formation of a folding intermediate, the slowest of the three phases has earlier been attributed to a cis-to-trans peptidyl-prolyl isomerization reaction. We thus substituted all four proline residues individually by alanine and investigated the effect of the amino acid substitutions on the folding and stability of the onconase variants. All onconase variants proved to adopt a tertiary structure comparable with that of the wild-type protein. Although the slow phase was not eliminated for any of the variants, the P43A substitution resulted in an increase in the rate constant of the fast folding phase, i.e. a faster formation of the folding intermediate. This variant also exhibits a significant increase in thermodynamic stability. As residue 43 belongs to those residues that are protected from hydrogen exchange with the solvent in the folding intermediate, the increase in the rate constant and stability of the P43A variant emphasizes the importance of the intermediate for the folding of onconase.
Onconase (EC 184.108.40.206)
Among the 20 canonical amino acids, proline plays a special role. The peptide bond with the preceding amino acid residue lacks the amide hydrogen so that this peptide bond cannot act as hydrogen bond donor. Consequently, proline residues are rarely found in α-helices at positions > 3. For steric reasons the cis configuration is energetically more accessible for Xaa-Pro peptide bonds than for Xaa-non-Pro peptide bonds  resulting in a slower cis-to-trans isomerization reaction  and a more common occurrence in natively folded proteins (5.5%–6.5% versus 0.03%–0.05% found in crystal structures) [3, 4]. Thus, proline is a prevalent amino acid in β-turn structures, an especially favorable way for the polypeptide chain to reverse its direction . Replacement of such ‘cis prolines’ with other amino acids often results in a trans configuration of the respective peptide bond and a distortion of the (β-turn) structure or in a cis Xaa-Yaa peptide bond, both of which decrease the protein's stability . In turn, fixation of the β-turn by structural mimetics such as 5,5-dimethyl-l-proline can result in an increase in stability [7-9]. Moreover, the isomerization of Xaa-Pro peptide bonds, which results in considerable structural rearrangements, was found to act as a molecular switch by changing the biological activity of the protein [10-12].
In the unfolded polypeptide chain, the Xaa-Pro peptide bond isomerizes (cis-to-trans ratio ~ 20 : 80) as its native configuration is no longer supported by the native three-dimensional protein structure . Despite the possibility for the peptide bonds to rotate in the unfolded state (U) of the protein, the relatively high activation energy barrier for the isomerization of Xaa-Pro peptide bonds (~ 80–100 kJ·mol−1) can result in unfolded subspecies that even fold on different pathways to the native protein . Moreover, this high energy barrier can impede protein folding resulting in the formation of folding intermediates . Interestingly, the comparison of thermophilic proteins with their mesophilic counterparts revealed an increase in proline content in thermophilic proteins (except for helical structures) [16, 17]. The restricted conformational freedom of Xaa-Pro peptide bonds in U (due to the high energy barrier) entropically destabilizes U by raising the energy of U [16, 17], and moreover Xaa-Pro peptide bonds may stabilize the native state [18, 19] thereby increasing the thermodynamic stability (Gibb's free energy ΔG) of the protein.
Ribonuclease A (RNase A), which has served as a model protein in enzymology, biophysics, biochemistry and protein chemistry for decades [20-23], contains four proline residues. Two of them – Pro42 and Pro117 – form trans peptide bonds and replacement with alanine has no (P42A) or only limited (P117A) effect on the stability and folding of the protein [13, 21, 24]. In contrast, Pro93 and Pro114 are located in β-turns and form cis peptide bonds in the native protein. The slow trans-to-cis isomerization of these Xaa-Pro peptide bonds during the folding process results in the accumulation of folding intermediates [21, 25]. Replacement of these proline residues with alanine or glycine results in a decrease in stability and a changed folding behavior [1, 13, 24, 26, 27]. Whereas the crystal structures of the Pro→Gly variants indicated trans 92–93 and 113–114 peptide bonds [1, 6], data on the Pro→Ala variants indicated cis peptide bonds [24, 28]. Replacement of Pro114 with 5,5-dimethyl-l-proline [9, 29] or (2S,4R)-4-fluoroproline (U. A., unpublished results) results in an increase in stability by 3.5 °C (6.0 kJ·mol−1) or 1.3 °C (2.5 kJ·mol−1), respectively, by favoring the native cis configuration of the Asn113-Pro114 peptide bond.
Onconase (ONC), which surpasses the homologous RNase A by ~ 25 °C or ~ 20 kJ·mol−1 in stability [30, 31], likewise contains four proline residues but all four form trans peptide bonds in the native state . Pro41 and Pro43 are found at positions 1 and 3 of helix III, respectively. Pro74 is located in a loop and Pro95 is part of the C-terminal β-sheet system (Fig. 1). In the unfolded state, these trans Xaa-Pro peptide bonds (partially) isomerize to the non-native cis configuration. Thus, the slowest of the three refolding phases of ONC was described as an Xaa-Pro cis-to-trans isomerization reaction . However, which of the four proline residues may cause the slow phase or whether one or more proline residues are involved remained unclear. Pro41, Pro43 and Pro95 are located in regions that are involved in the (fast) formation of the folding intermediate of ONC but the β-sheet system flanking the loop with Pro74 gains its native conformation not before the conversion of the folding intermediate to the native state [33, 34].
Here, we replaced all proline residues of ONC individually by alanine and studied the effect of the substitution on the stability and folding of the proteins. None of the substitutions resulted in an obvious alteration of the protein structure but, whereas P95A-ONC was less stable than ONC by 4.6 °C, the unusual stability of ONC was further increased by 6.5 °C in P43A-ONC.
Design, expression, renaturation and purification of ONC variants
According to the analysis of the crystal structure of ONC (1ONC), all four proline residues in native ONC adopt a trans configuration (Fig. 1). Pro41 and Pro43 occupy positions 1 and 3 of helix III, respectively. Pro74 is located in a loop, and Pro95 is part of the C-terminal β-sheet. As none of the four proline residues is in cis configuration in native ONC, the individual replacement by alanine was expected to result in no severe disturbance of the tertiary structure.
ONC and its variants were expressed as inclusion bodies, renatured, and purified as described in Materials and methods. The purified proteins proved to be homogeneous by SDS/PAGE and re-chromatography on a SOURCE S column (not shown). Mass spectrometry confirmed the correct amino acid substitutions as well as the cyclization of the N-terminal Gln residue (Table S1).
Biophysical characterization of the native ONC variants
Besides the determination of the catalytic activity, which is a very sensitive measure for the maintenance of the native structure of an enzyme, circular dichroism (CD) and NMR spectroscopy were used to investigate the structural influence of the Pro→Ala substitutions in comparison with wild-type ONC. The kcat/KM values for ONC and its variants, determined with 6-carboxyfluorescein-dArU(dA)2-6-carboxytetramethylrhodamine (FAM-AUAA-TAMRA) as substrate, revealed that all ONC variants were as active as the wild-type enzyme (Table S1) indicating a proper fold.
As deduced from the concordant CD spectra in the far- and near-UV region, all the investigated ONC variants acquire a secondary and tertiary structure comparable with that of ONC (Fig. 2). Likewise, comparison of the NMR spectra of the four Pro→Ala ONC variants with that of the wild-type enzyme shows no significant shift in the resonances (Fig. 2). All 1D 1H NMR spectra showed a good dispersion in the amid proton region and well resolved signals for the upfield shifted methyl groups (Fig. 2) indicating a folded protein. Despite some changes in signal intensity, which may indicate marginal local changes in structure or dynamics, the overall structure is preserved.
Thermally induced conformational transition
The thermal stability of ONC and its variants was investigated by temperature-induced unfolding using the change in CD signal at 289 nm (Figs 3 and S1). Thermal unfolding proved to be reversible and follows a two-state transition model as judged from the fit of the data. Whereas the Tm values of P41A-ONC and P74A-ONC are comparable with that of ONC (+1.7 °C and +1.1 °C, respectively), P95A-ONC is less stable than ONC by 4.6 °C. In contrast, P43A-ONC shows an increase in Tm by 6.5 °C in comparison with ONC (Fig. 3).
Refolding experiments followed by stopped-flow fluorescence spectroscopy
The refolding reaction of ONC exhibits three phases [33, 35]. Whereas two phases show a dependence on the denaturant concentration, the slow phase was assigned to a cis-to-trans peptidyl-prolyl isomerization reaction . Replacement of proline residues with alanine was anticipated to eliminate the slow phase in one of the variants (in the simplest case). Thus, refolding of the Pro→Ala-ONC variants was followed in 1.25 m guanidine hydrochloride (GdnHCl). Under these conditions the rate constants for the three refolding phases of ONC differ by about one order of magnitude each and the amplitude of the slowest phase is maximal ([33, 35] and C. S., unpublished data).
In contrast to the anticipation for the simplest case all four Pro→Ala-ONC variants showed three phases during refolding from the GdnHCl-denatured state (Table 1, Figs 4 and S2). Whereas the folding rates of both P41A-ONC and P74A-ONC are hardly changed in comparison to ONC (kfast of P41A-ONC is slightly larger than that of ONC), both kfast and kmedium of P95A-ONC are slightly smaller than those of ONC. Surprisingly, during the second phase of the refolding reaction of P95A-ONC the signal increased (Figs 4 and S2). The only considerable deviation from the rate constants of ONC can be found for kfast of P43A-ONC, which is about 10-fold higher than kfast of ONC.
|kfast (s−1)||Relative Afast (%)||kmedium (s−1)||Relative Amedium (%)||kslow (s−1)||Relative Aslow (%)|
The role of proline residues in proteins is multifaceted. The high activation energy barrier for the cis–trans isomerization of Xaa–Pro peptide bonds is exploited by nature for molecular switches [10, 12, 36] but can result in parallel folding pathways and/or the accumulation of folding intermediates in vitro as well . Moreover, cis-prolines favor the formation of β-turn structures whereas trans-prolines can increase the stability of proteins by a stabilization (rigidification) of the native conformation and/or by an entropic destabilization of the unfolded state . The unusually stable protein ONC contains four trans-prolines, which were replaced individually by alanine in this study. Refolding of GdnHCl-unfolded ONC proceeds via an on-pathway intermediate . A slow phase with a rate constant independent of the GdnHCl concentration (at low concentrations of GdnHCl) and an Arrhenius activation energy of ~ 75 kJ·mol−1 had been assigned to an Xaa–Pro cis-to-trans isomerization reaction in the unfolded state . However, this slow phase was not accelerated in the presence of the peptidyl-prolyl isomerases cyclophilin 18 or Thermus SlyD . From the structural point of view, Pro74 and Pro95 in particular appeared to be candidates for the origin of the slow refolding phase  as a slow isomerization might limit the formation of the respective β-sheet systems. According to the CD and 1D NMR spectra (Fig. 2) none of the replacements resulted in a noticeable alteration of the native protein structure and also the catalytic activity was found to be unaffected (Table S1). The analysis of the refolding of GdnHCl-unfolded ONC and its variants, however, surprisingly revealed that for none of the variants was the slow phase, which had been described as an Xaa-Pro cis-to-trans isomerization reaction , eliminated (Figs 4 and S2). As the number of observed folding phases remained constant, obviously no fundamental change of the folding behavior was induced by the amino acid substitutions. Notably, however, the P95A substitution resulted in a deceleration of the refolding (kfast and kmedium, Table 1) and likewise in a destabilization by 4.6 °C. Moreover, in contrast to the folding process of ONC and the other Pro→Ala ONC variants an increase in the fluorescence signal during the medium phase was detected for P95A-ONC. Residue 95 is in close spatial vicinity to Trp3, which predominantly is responsible for the fluorescence properties of ONC. Thus, the P95A substitution might alter the structural arrangement of the C-terminal region of P95A-ONC in comparison with ONC during the folding process. In contrast, kfast of P43A-ONC was found to be about 10-fold higher than kfast of ONC (Table 1), which coincides with the effect of the substitution on Tm (ΔTm = +6.5 °C).
The continued existence of the slow phase for all Pro→Ala ONC variants indicates that the slow phase in the folding process of wild-type ONC is obviously not caused by an (or at least not by a single) Xaa-Pro cis-to-trans isomerization reaction. Instead, the isomerization of an Xaa-non-Pro peptide bond – either at the introduced alanines or at residues beyond the mutated prolines – might be responsible for the (persistent) occurrence of the slow folding phase as has been described for the folding of a proline-free variant of tendamistat . Moreover, slow folding kinetics might be caused by a rigid framework of disulfide bonds as has been found for cystine knots [38, 39]. Whereas the slow folding phase of RNase A, which leads to conformational rearrangements late in the folding process, originates from a trans-to-cis isomerization reaction [24, 25], the slow folding phase of ONC, which is caused by processes in the unfolded state , is apparently not caused by an Xaa-Pro cis-to-trans isomerization reaction.
The effect of the substitutions on the thermal stability of the respective ONC variants is diverse (Fig. 3). The substitutions P41A and P74A resulted in almost no change in Tm. P95A-ONC is less stable than ONC by 4.6 °C (thereby indirectly supporting the concept of protein stabilization by the introduction of proline residues). Surprisingly, the substitution P43A further increases the Tm of ONC by 6.5 °C. An explanation for this stabilizing effect could be a faster formation as well as an extension of helix III. In P43A-ONC this helix could be elongated N-terminally by two more residues so that Pro41 would then be at position 3. However, an increase in α-helical content could not be detected by CD spectroscopy (Fig. 2). As mentioned above, Pro43 and Pro95 in particular are located in structural regions that are involved in the (fast) formation of the folding intermediate , which is essential for the efficient formation of native ONC .
Although the folding process did not become simpler for any of the ONC variants investigated and the slow phase, which occurs during the refolding of GdnHCl-unfolded ONC, apparently is not caused by an Xaa-Pro cis-to-trans isomerization reaction, it was possible to further increase the stability of this unusually stable protein by a Pro→Ala substitution. The increase in both kfast and Tm of P43A-ONC underscores the importance of the efficient formation of the folding intermediate for the folding and stability of ONC. Further experiments are required to pinpoint the molecular reason for the slow refolding phase of ONC.
Materials and methods
Oligonucleotides (Table S2) and FAM-AUAA-TAMRA were from Metabion International AG, Martinsried, Germany. Restriction enzyme DpnI was from New England Biolabs, Frankfurt/Main, Germany, growth media were from Difco Laboratories, Detroit, MI, USA, and Escherichia coli strains XL-1 Blue and BL21(DE3) were from Stratagene, Heidelberg, Germany. All other chemicals were of the purest grade commercially available.
The onc gene in pET-26b(+)  was modified by use of the QuikChange™ site-directed mutagenesis kit (Stratagene) to obtain the mutations P41A, P43A, P74A and P95A, respectively. The introduction of the desired mutations was verified by DNA sequencing (MWG Biotech, Ebersberg, Germany).
Expression, renaturation and purification of the enzyme variants
The experimental procedures were performed as described previously [31, 35]. Briefly, cultures of E. coli strain BL21(DE3) (Stratagene) that had been transformed with a plasmid (pET-26b(+)) directing the expression of ONC or the individual ONC variants were grown in terrific broth containing 50 μg·mL−1 kanamycin at 37 °C. Cells were harvested 4 h after induction with 1 mm isopropyl β-d-thiogalactopyranoside. Cell lysis was performed by treatment with lysozyme and sonication. The inclusion bodies were isolated and resolubilized. After renaturation, ONC and its variants were purified on a SOURCE S column (Amersham Biosciences, 50 mm Tris/HCl, pH 7.5, with a linear gradient of 0–500 mm NaCl). Cyclization of the N-terminal glutamine residue was achieved by dialysis of the purified proteins against 200 mm sodium phosphate buffer, pH 7.0, for 3 days at room temperature. The protein concentration was determined using the absorption at 280 nm (A280 = 0.87 mL·mg−1·cm−1) . Finally, buffer was exchanged to 100 mm sodium acetate buffer, pH 5.5. Determination of the molecular mass to verify cyclization of the N-terminal pyroglutamate was performed by electrospray ionization (Esquire) or matrix-assisted laser desorption ionization mass spectrometry (Reflex; Bruker-Franzen, Bremen, Germany) after desalting of the protein samples using ZipTip pipette tips (Millipore, Schwalbach, Germany).
Values of kcat/KM for the enzymatic cleavage of the fluorogenic substrate FAM-AUAA-TAMRA were determined as described previously . Activity was measured at 20 °C in 100 mm 2- (N-morpholino)ethanesulfonic acid/NaOH buffer, pH 6.0, containing NaCl (100 mm).
CD spectra of ONC and its variants were recorded in 100 mm sodium acetate buffer, pH 5.5, containing 1 mg·mL−1 of protein, on a CD spectrometer J-810 (Jasco, Groß-Umstadt, Germany) at 20 °C. Cuvettes of 1 cm and 0.01 cm path length were used for CD spectroscopy in the near-UV region (250–350 nm) and in the far-UV region (190–250 nm), respectively.
1D NMR spectra
NMR spectra of ONC and its variants were recorded on a Bruker AvanceIII 600 MHz spectrometer (Bruker BioSpin GmbH, Rheinstetten, Germany). Experiments were performed at 20 °C with 50 μm of unlabeled protein in 100 mm sodium acetate buffer, pH 5.5, containing 10% D2O (v/v). The water resonance was suppressed by presaturation and Watergate techniques . Spectra were processed by bruker topspin 2.1 software.
Thermally induced transition
Values of Tm were determined by CD spectroscopy (CD spectrometer J-810, Jasco) at 289 nm using a heating rate of 1 K·min−1. Measurements were carried out in 100 mm sodium acetate buffer, pH 5.5, containing 1 mg·mL−1 of protein. The signal y was fitted as described by Pace et al.  to obtain Tm .
Refolding experiments followed by stopped-flow fluorescence spectroscopy
Refolding of GdnHCl-unfolded proteins was followed by stopped-flow fluorescence spectroscopy using an SX20 stopped-flow spectrometer (Applied Photophysics Ltd, Leatherhead, UK). Emission was recorded as integral fluorescence using a 320 nm cut-off filter with excitation at 280 nm at 20 °C. Refolding was initiated by 11-fold dilution of denatured protein samples (in 6.0 m GdnHCl) into 0.775 m GdnHCl. The final protein concentration was typically 5 μm, and final buffer condition was 100 mm sodium acetate, pH 5.5, containing 1.25 m GdnHCl. Kinetic traces were fitted to single-, double- or triple-exponential equations. Amplitudes and rate constants were usually the average from five to ten measurements. The experimental error was < 5% for the triple-exponential fits.
Dr A. Schierhorn (Martin-Luther University Halle-Wittenberg, Germany) is acknowledged for performing mass spectrometry measurements and the European Regional Development Fund (ERDF) by the European Union for significant investments into the NMR facility of the University Halle-Wittenberg. The Graduiertenkolleg 1026 ‘Conformational transitions in macromolecular interactions’ of the DFG and the initiative ProNet-T3 of the German Federal Ministry of Education and Research (BMBF) are acknowledged for financial support.