Extant Trx Stabilities and Folding/Unfolding Kinetics.

Previous studies of stability of Trxs indicated that these proteins are highly stable (41, 48). Indeed, even at pH 2, the transition temperatures range between 40 and 95 °C (Fig. 2A). This variation in the transition temperatures seems to correlate with the optimal growth temperatures of the corresponding organisms. For example, Trxs from thermophilic organisms, such as TrxTT, TrxPY, TrxMJ, and TrxST, have much higher Tm values than their mesophilic homologs TrxEC and TrxHS. The mesophilic proteins, in turn, have higher melting temperatures that the proteins from psychrophilic organisms TrxSB and TrxCP. This correlation of the growth temperature and protein stability is also evident from the denaturant-induced unfolding studies. Fig. 2B shows urea-induced unfolding profiles for the same set of proteins. The midpoint for the urea-induced unfolding transitions ranges from 1.5 to 7 M. For one protein, thermophilic TrxST, the midpoint is probably even higher, because no indication of unfolding was observed at 10 M urea.

Open in a separate window

Fig. 2.

Trx protein family has diverse thermodynamic and kinetic properties. (A) Comparison of the thermal stabilities of Trx proteins at pH 2.0 as monitored by DSC. Cp, partial molar heat capacity. (B) Comparison of the stabilities of Trx proteins against urea-induced unfolding at pH 2.0. Experimental points are shown as symbols, whereas the fit to a linear extrapolation model is shown by solid lines. (C) Chevron plots of lnk_obs vs. urea concentration of extant Trx variants. (D) Chevron plots of lnk_obs vs. urea concentration of ancestral Trx variants. Solid lines in C and D are the results of the global fit to Eq. 2. Symbols and colors are as follows: brown triangle, TrxSB; green triangle (only in A), TrxCP; turquoise-blue triangle, TrxEC; purple triangle, TrxHS; dark-blue triangle, TrxTT; teal-green triangle, TrxPY; greenish-gray triangle, TrxMJ; gray triangle, TrxST; light-green circle, TrxLAFCA; gray circle, TrxLECA; red circle, TrxLBCA; blue circle, TrxLGPCA; pink circle, TrxAECA; yellow circle, TrxLPBCA; black circle, TrxLACA.

How are the differences in thermodynamic stabilities manifested in the kinetic stabilities? Changes in thermodynamic stability can be caused by the differences in the folding rates, unfolding rates, or both. We thus performed characterization of the folding and unfolding of extant Trx variants using standard stopped flow methods. Fig. 2C shows the experimentally measured folding and unfolding rate, k_obs, in the form of chevrons for the extant Trx variants. Each data point on the chevron is an average of at least five independently measured traces, and their SDs are reported as the error bars. It is evident from these chevron plots that there is minimal variation in the rates of Trx folding. They all fold in the absence of denaturant in 80 ± 60 ms, and the difference between the fastest (20 ms) and the slowest (140 ms) is only sevenfold.

More notably, inspection of the chevron plots shown in Fig. 2D reveals that the variation in the unfolding rates is dramatic. Extrapolation to a zero denaturant concentration gives unfolding rate constants in the range between 7 and 22,000 s [6 d; i.e., ∼3,000-fold difference between the fastest (TrxCP) and the slowest (TrxPY) unfolding Trx proteins]. Importantly, the unfolding rates correlate with stability, whereas folding rates show little correlation with stability within the extant Trx protein family (Fig. 3). This observation suggests that the differences in stabilities of extant Trx are largely caused by the differences in the unfolding rates of these proteins.

Open in a separate window

Fig. 3.

Thermodynamic stability of Trx variants correlates with the unfolding rates and does not correlate with the folding rates. Equilibrium stabilities, ΔG_eq, are plotted vs. folding (ΔGfol#; black regression line) and unfolding (ΔGunf#; red regression line) barriers. The folding/unfolding rates are extrapolated to zero denaturant concentration using data for extant and ancestral Trx variants shown in Fig. 2 C and D. Symbol/color coding for different proteins is the same as in Fig. 2.

Ancestral Trx Stabilities and Folding/Unfolding Kinetics.

Similarity of the folding rates among extant Trx proteins is not an unexpected finding, because we are assessing the kinetics of a specific fold, and there are several reports correlating various topology metrics to folding kinetic rates (49–52). Moreover, the observed robust folding rates suggest that the Trx energy landscapes are unfrustrated (3, 5, 29, 36, 53, 54). This observation, in turn, implies that the folding of extant proteins of the Trx family obeys the principle of minimal frustration and that variations in stability are caused by the unfolding rates. Was this part of the evolutionary considerations?

To answer this question, we turned our attention to the analysis of the thermodynamics and kinetics of the Trx proteins for which sequences were obtained from ASR (41, 55). ASR is a bioinformatics technique that allows sequence extrapolation from extant protein sequences back to their common ancestors. It has been used to resurrect ancestral nodes for a number of proteins and shown to produce catalytically and/or functionally active proteins (40–42, 56–58). The sequences for the ancestral Trx proteins were obtained using maximum likelihood sequence reconstruction, targeting several Precambrian nodes during Trx evolution (41). The seven sequences that were studied here are last animal and fungi common ancestor (TrxLAFCA), last eukaryotic common ancestor (TrxLECA), last bacterial common ancestor (TrxLBCA), last γ-proteobacteria common ancestor (TrxLGPCA), archaea/eukaryota common ancestor (TrxAECA), last cyanobacterial, deinococcus and thermus common ancestor (TrxLPBCA), and last archaeal common ancestor (TrxLACA). These sequences were dated back to over 4 billion years, thus enabling us to probe the thermodynamics and kinetics of folding/unfolding for proteins with sequences that arguably represent a wide evolutionary timespan (41). Furthermore, the crystal structures of these proteins have been determined and support the high conservation of the Trx protein fold (55).

Thermodynamic stabilities of the ancestral Trx proteins at pH 2 are compared in Fig. 2A. As with the extant Trx proteins, ancestral Trx variants showed significant variation in their transition temperatures. Proteins that are dated earlier on the evolutionary tree (e.g., LACA and AECA) have higher T_m values than the more recent proteins, such as LECA and LAFCA, which in turn, are more stable than TrxEC. Equilibrium urea-induced unfolding profiles also support this correlation (Fig. 2B). The differences in transition temperatures at pH 2 agree well with the T_m values obtained at pH 7 (Fig. S1), indicating a generality of the stability trend (41).

Open in a separate window

Fig. S1.

Correlation of transition temperatures as measured by DSC for extant and ancestral Trx proteins at pH 2 and pH 7. Symbols are as follows: brown triangle, TrxSB; green triangle, TrxCP; turquoise-blue triangle, TrxEC; purple triangle, TrxHS; dark-blue triangle, TrxTT; teal-green triangle, TrxPY; greenish-gray triangle, TrxMJ; gray triangle, TrxST; light-green circle, TrxLAFCA; gray circle, TrxLECA; red circle, TrxLBCA; blue circle, TrxLGPCA; pink circle, TrxAECA; yellow circle, TrxLPBCA; and black circle, TrxLACA. The solid line is the linear fit with the correlation coefficient of 0.92.

The results of folding/unfolding experiments for ancestral Trx variants measured using standard stopped flow methods are presented in the form of chevron plots in Fig. 2D. Again, as in the case of the extant Trx proteins, the folding rates of ancestral Trx are very similar, whereas the unfolding rates vary dramatically. This observation might suggest that the principle of minimal frustration played a role in the sequence selection early on in evolution.

Both ancestral and extant Trx proteins maintain strong correlation of increasing apparent unfolding barrier (i.e., decrease in the unfolding rate) with increase in thermodynamic stability, whereas the apparent folding barrier (as defined by the folding rates) remains independent of the thermodynamic stability. The thermal and kinetic stabilization conferred to the oldest ancestral variants is comparable with the most stable extant Trx proteins (e.g., TrxPY and TrxMJ).

The observation that both extant and ancestral Trx proteins fold with similar rates directly supports the assumption that protein sequences have evolved with minimal energetic frustration so as to rapidly fold into their stable 3D states (36). It is also clear that the differences in thermodynamic stabilities are primarily caused by differences in unfolding rates for this protein family.

Differential Scanning Calorimetry Measurements.

Differential scanning calorimetry (DSC) experiments were performed on a VP-DSC instrument (Microcal/GE-Healthcare) at a scan rate of 1.5°/min using protein concentrations of about 1.0 mg/mL. Buffers used were 30 mM glycine-HCl for pH values 2–3.5 experiments and 30 mM sodium-cacodylate for pH 7.0 experiments (67). Proteins were extensively dialyzed against the corresponding buffer. Raw DSC data were analyzed using the Origin-DSC (OriginLab) software package. Global fitting of the heat capacity profiles was done using in-house scripts using the nonlinear regression (NLREG) routine (68, 69).

Equilibrium Stability Measurements.

Initially, all Trx protein stocks were extensively dialyzed in acidified water and diluted 20-fold into 30 mM glycine-HCl or buffered urea solutions adjusted to pH 2. Dilutions were done to final protein concentrations of 4–10 μM depending on their fluorophore quantum yields. Stock protein samples were diluted into buffered urea with concentrations incrementing up to 9.5 M and incubated at room temperature for 24 h. Tryptophan emission spectra were collected as a function of urea concentration using a fluorescence plate reader (TECAN Infinite M1000 Pro) in a 384-well black plate. Plate and samples were thermostated in the instrument at room temperature (23 °C ± 1 °C) for 5 min before data collection. Excitation was done at 295 nm with a gain of 150. Fluorescence emission spectra were recorded from 305 to 450 nm in 1-nm increments. TrxMJ does not have a tryptophan; therefore, tyrosine emission fluorescence was used instead. Trx proteins that did not produce a significant intensity change were additionally characterized via far-UV CD.

Urea unfolding profiles (intensity at 355 nm vs. urea concentration) were fitted globally for all Trx variants according to the linear extrapolation model (70, 71) using in-house written NLREG scripts as previously described (72).

Kinetic Stopped Flow Experiments.

All Trx protein samples were prepared as described above. Data for chevron plots were collected by standard stopped flow methods on a JASCO J-815 spectropolarimeter operating in fluorescence mode that was equipped with an SFM 300 mixing module (BioLogic Science Instruments) containing a high density (HDS) mixer and a 30-µL FC-15 observation cuvette (73). Buffer containing 30 mM glycine-HCl adjusted to pH 2 was used for all kinetic experiments. Folding and unfolding reactions were initiated by a 20-fold dilution of protein-buffered stock solutions (80–200 µM) with the desired buffered urea concentration. Because of slow kinetics of unfolding, equilibration in buffered urea was required before refolding experiments could be successfully initiated. The reagent syringes, mixing chamber, and observation cuvette were thermostated at 20 °C using a circulating water bath. Fluorescence emission intensity from an N-WG 320-nm (N-WG 305 nm for TrxMJ) cutoff filter (BioLogic Science Instruments) was collected after excitation at 295 nm (280 nm for TrxMJ) through a 3-nm slit from a mercury lamp source. Voltages applied to the photomultiplier tube were set constant based on the fluorescence signal intensity at maximum amplitude (∼920 V).

The Trx model system required multiexponential fits with a slow phase related to proline isomerization (74–76). Thus, the pertinent fastest folding and unfolding phases are reported. Rate constants, k_obs, were obtained using stretched exponential as previously described (73):

where I(t) is fluorescence intensity as a function of time, y_o is the initial fluorescence intensity, A is the amplitude of the change between initial and final fluorescence intensities, k_obs is the observed kinetic rate constant associated with the fluorescence intensity relaxation, and pb is the sloping baseline correction for the photobleaching effect used in the Bio-Kine32 software. All traces were corrected for instrumental dead time (6 ms) before fitting, because some Trx protein variants showed significantly fast k_obs.

The natural logarithms of the k_obs are plotted as a function of urea concentration in the form of chevron plots. Each data point on the chevron plot is an average of five individual traces, and errors are taken as the SDs. Extrapolated k_f(H₂O) and k_u(H₂O) values were obtained by globally fitting the chevrons of all Trx variants to Eq. 2 below (77) using in-house scripts for the NLREG data-fitting package:

where k_f(H₂O) and k_u(H₂O) are the folding and unfolding rates in the absence of denaturant, respectively, and m_f and m_u are the kinetic folding and unfolding m values (measured in kilojoules per mole per molar), respectively (77). The apparent folding/unfolding barriers have been calculated as ΔGfol/unf#=RTln(kf/u), where R is the universal gas constant (8.184 J/mol K), and T is the temperature (Kelvin).

We thank Jose M. Sanchez-Ruiz for providing the plasmids for ancestral Trxs and Catherine Royer for the usage of the high-throughput fluorescence plate reader. This work was supported by US National Science Foundation Grant 1506468/1330249.