2.2 Binding of 15N L-hIGFBP2 to IGF-1 and IGF-2
Dissociation constants (K d) of 4.1 ± 2.2 µM for IGF-1 and 3.7 ± 1.5 µM for IGF-2 were estimated from surface plasmon resonance (SPR) based on the association and dissociation rates (Figs. 3a, b ). The binding of IGF-1 and IGF-2 to L-h IGFBP2 was also monitored using NMR. To identify L-h IGFBP2 residues involved in binding IGFs, a sample of15N-labeled L-h IGFBP2 was titrated with unlabeled IGF-1. Fig. 3c shows an overlay of the 2D [15N-1H] HSQC spectrum of the complex with that of free (unbound) L-h IGFBP2, with residues undergoing significant shifts highlighted in the inset. For 1 mM IGFBP2 protein, the peaks stop moving after the addition of 1 mM IGF-1, showing a clear 1:1 ratio for binding. Therefore, the binding is both very dynamic (fast on and off rates) and of moderate affinity (K d of 4.1 ± 2.2 µM as measured from SPR,Fig. 3a ). However, a substantial change in the overall dynamics of the protein was observed, as discussed below. Assignment of the shifted resonances was based on inspection of their position for the unbound L-h IGFBP2 spectrum. The magnitude of15N/1H chemical shift perturbations for each residue of L-h IGFBP2 (Fig. 3d ) shows that residues K150-E161 and Q165 undergo the largest chemical shift changes upon binding IGF-1, with an average shift of 0.032±0.017 ppm. Similar shifts were observed when IGF-1 bound to full-length IGFBP2 (Fig. S6 ). A region containing these residues (K150-E161) is predicted to be a molecular recognition feature (MoRF) (a short binding region located within a longer intrinsically disordered region) and is located adjacent to the protease cleavage sites of h IGFBP2 (Figs. 3e ). To confirm the involvement of these residues in binding IGF-1, we prepared a mutant form of L-IGFBP2 with deletion of K150-E161 (L-hIGFBP2[desK150-E161]), which showed weaker binding to IGF-1 (Fig. S7 ).
We also identified residues in IGF-1 involved in binding L-h IGFBP2. For this purpose, 15N-labeled IGF-1 in 50 mM Na-phosphate buffer (pH 6.0) was titrated at 20 ºC with a solution of unlabeled L-h IGFBP2. An overlay of the 2D [15N-1H] HSQC spectrum of the L-h IGFBP2: IGF-1 complex with that of free IGF-1 is shown inFig. S8 . Chemical shifts for most residues of IGF-1 remained unchanged, but cross-peaks for residues E3, T4, A13, V17, C18, G19, D20, R21, G22, M59, Y60, C61, and A62, which were very weak or absent in free IGF-1, exhibited a significant increase in intensity in the bound form (Fig. S8 ). This implies that the dynamics of IGF-1 are also significantly affected upon binding L-h IGFBP2, even though its average conformation is unchanged (Fig. S8 ).