(A, B) CEST profiles (circles) of G51, N53, T54, and G61 reporting on the transient native dimerization process (I) measured at 2 B1 fields (31 and 59 Hz). Solid lines are fits of CEST data for each residue independently to a two-state model of exchange, including profiles from both B1 fields (A) or for each B1 field separately (B). A range of excited state populations and lifetimes from 1.5–4.5% and 2.2–4.3 ms is obtained. (C, D) CEST profiles at 2 B1 fields (31 and 59 Hz) and CPMG profiles at two static magnetic fields (corresponding to 1H resonance frequencies of 600 and 800 MHz) for T135 and T137 reporting on the local folding process (II). Solid lines are fits of both CEST and CPMG data to a two-state model (C) or to a three-state model (D) of conformational exchange. (E) Lowest energy ( ∼2) physically meaningful three-state models describing process II, where N is the native state of apoSOD12SH, H is the transiently populated helical state and I is an intermediate not directly visible in CEST measurements, but required for data fitting. Each of the shapes is meant to represent the absence (circle) or presence of the helix (triangle) that is formed via process II, as established on the basis of 15N chemical shifts; note that very large 15N chemical shift changes (∼10 ppm) are observed for helix formation for both T135 and T137 so that these shifts can be used as direct reporters of structure. Exchange parameters for each model are indicated on the plot, where listed exchange rates refer to the sums of forward and backwards processes (for example, 1400 s−1 = kIN + kNI in the model on the left). Lifetimes (τH) of the excited state H, corresponding to the conformer with the native-like electrostatic loop helix, were calculated according to the relations τH = 1/koff,H, where koff,H = kHI = kex,HI × (pI/(pH + pI)) for the on-pathway model (left) and koff,H = kHN = kex,HN × (pN/(pH + pN)) for the off-pathway model (right). Values of τH ranged between 12–14 ms, depending on the model used and are reported in the text (Figure 9) as an average over the two models. Note that the on-pathway and off-pathway model above cannot be distinguished, as both models fit the data equally well.