Reply to Candel et al.: Evidence for evolutionary conservation of folding kinetics in the thioredoxin protein family.

The letter by Candel et al. (1) does not address a potential problem with their own experimental set-up. In particular, the use of guanidinium hydrochloride (GdnHCl) is not the best choice of denaturant for studying a folding reaction. There is an ample amount of evidence to suggest that the variations in the ionic strength (with GdnHCl being an ionic compound) can lead to unexpected effects (2), including an increase in protein stability at low denaturant concentration (3–6) and changes in apparent kinetics of folding (7, 8). Therefore, we performed our studies using urea (9). Moreover, Candel et al. (1) do not mention their previously published results with the Escheria coli thioredoxin, using urea as a denaturant (see figure 7 of ref. 10), which does not show rollover behavior. This provides additional compelling evidence that the use of GdnHCl can lead to erroneous results. Whereas indeed our folding–unfolding experiments were performed at pH 2 (to observe the transition in urea-induced unfolding for thioredoxins with higher stability), we have provided sufficient evidence that the general features observed at pH 2 are consistent with those at neutral pH. The first piece of evidence is that the stabilities obtained from kinetic and equilibrium urea-induced folding–unfolding experiments are very similar [see figure 3 in Tzul et al. (9)]. The second piece of evidence is that the transition temperatures at pH 2 and pH 7 are highly correlated, suggesting that pH has similar effects on the stabilities of these proteins [see figure S1 in Tzul et al. (9)]. The third piece of evidence is that the proteolytic resistance, which can be used as a proxy for the unfolding rate, of thioredoxins at pH 2 and pH 7, correlates well with the unfolding rates [see figure 4 in Tzul et al. (9)]. Taken together, these data suggest that the behavior observed at pH 2 can be extrapolated to pH 7. The thorough experimental data of our report (9) is fully consistent with the hypothesis of the principle of minimal frustration. We have provided a substantial dataset of thioredoxin variants other than the two variants Candel et al. (1) have called into question. These additional 13 thioredoxin variants strongly support the trend seen from both extant and ancestral thioredoxin groupings. The main focus of our publication (9) is the evolutionary kinetic behavior of the thioredoxin protein family. Although we acknowledge there may be alternative explanations to our data, the principle of minimal frustration is a most consistent explanation and we fully stand by this conclusion.