Reply to Marchenko et al.: Flux analysis of GroEL-assisted protein folding/unfolding.

Using NMR-based relaxation experiments, we showed that exchange between the folded state (F) of a metastable SH3 domain and a folding intermediate (I) is an order of magnitude faster when the SH3 domain is bound to apo GroEL than in free solution (1). We did not consider fluxes through the apo GroEL-assisted and unassisted pathways. Marchenko et al. (2) note that the approximate rate constants for the GroELassisted interconversion between the F and I states (kF↔F-G↔I-G↔I and kI↔I-G↔F-G↔F) are slower than the corresponding rate constants (kFI and kIF) for direct interconversion, as the binding of I to GroEL is slower than the interconversion between the GroEL-bound F and I states under the conditions of the NMR experiments [i.e., ðkapp on + kIG off Þ< ðkFI + kIFÞ, where kapp on is a pseudo-first-order association rate constant given by kon[G]; see scheme in Fig. 1]. On this basis, Marchenko et al. (2) conclude that our data provide “strict experimental evidence that apo GroEL does not accelerate protein folding, although it does accelerate one of its steps,” and therefore corroborates their earlier hypothesis that the interaction of GroEL with folding intermediates hinders the formation of native structure (3). However, Marchenko et al. (2) fail to take into account that binding of F and I to GroEL are second-order processes dependent upon the concentration of apo GroEL. The relative contributions of GroEL-assisted and unassisted pathways can be assessed by steadystate flux analysis (4). The flux through parallel and serial reaction paths is given by Fparallel = ΣFi and Fserial = [Σ(1/Fi)], respectively, where Fi is the flux of the ith reaction step. For the kinetic scheme in Fig. 1, the fluxes between states F and I through the GroEL-assisted and unassisted pathways are given by