Supplementary Figure 6: Goodness of global fitting for four-step and six-step reaction models.

(a) For the four-step scheme, the lower limit for k_–1 is undefined, the estimated k₁ is much lower than estimated from the concentration dependence of k₁ in Supplementary Fig. 2b, and a monomolecular step of about 100 s⁻¹ corresponding to GTP hydrolysis 19,33 is not defined. Visual inspection indicates deviations of the fit to the four-step model from the experimental data (not shown), suggesting that global fitting to a four-step scheme does not provide a consistent fit.

(b) For the six-step scheme, the upper limits for k₁ and k_–1 are poorly defined; the rate of additional step 1a is about 300 s⁻¹, which is not observed directly, but cannot be excluded. Regardless of the model, global analysis robustly finds three steps, with the global minima at about 33-43 s⁻¹, 14-15 s⁻¹, and 4 s⁻¹, corresponding to steps 3, 4, and 5 of the five-step model; step 2 with the rate of 80-100 s⁻¹ is well-defined in the five- and six-step models and consistent with the previous data. Red dashed lines show the lower and upper boundaries as calculated for a χ² threshold of 1.025 for the four-step model (the calculated χ² threshold of the fit is 1.0025) and 1.032 for the six-step model (the calculated χ² threshold of the fit is 1.0032) (refs. 31,45,46).