Supplementary Figure 7: Experimental and simulated EPR (DEER) data.
EPR measurements were used to obtain very long-range distance restraints. These are imperative for the definition of the discoidal shape and the confirmation of a fixed arrangement of the MSPΔH5 molecules within the rdHDL assembly. (a)-(f) Left panels: Intensity-normalized, phase-corrected primary DEER data (i.e. data before background correction). Experimental time traces (shown as a black line in the left panel) were acquired using the Q-band DEER experiment. Middle panels: corresponding DEER distance distributions for paramagnetic MTSL attached to the indicated residue. Experimental distance distributions (black line in the corresponding middle panel) were computed using Tikhonov regularization, as implemented in DeerAnalysis, assuming a homogeneous spatial distribution of the observer spins (3D background model). The light blue lines in the middle panels represent the ensemble averaged simulated distance distribution using the 10 structures with the lowest CYANA target function. The dotted red line in the middle panels represent the ensemble averaged distances found within the 10 conformers of our structure between the Cα dummy atoms representing the N-O nitroxide moiety determined by the rotamer approach. Right panels: error estimation for the experimental distance distributions performed with the Validation tool of DeerAnalysis (here possible distance distributions are confined between the dark blue and the red lines. Black lines represent corresponding generalized normalized distributions. Grey areas indicate an extent of possible distance distribution models seen during the error computation. The light blue lines in the right panels represent simulated distance distributions for each of the 10 structures with the lowest CYANA target function. Simulations were done with the rotamer library approach as implemented in the MMM package. The rotamer library R1A_298K_UFF_216_CASD.mat was used. (g)-(i) The Q-band DEER traces were acquired on the mutant with MTSL at position 213 with full (black) and strongly reduced (blue) inversion efficiency of the pump pulse to determine whether more than two MSP monomers assemble to rdHDL particles (see Supplementary Information for more details). (g) Normalized and phase corrected primary DEER traces. (h) DEER form factors. (i) Overlay of two form factors after normalization to the same modulation depth. They correspond perfectly. Experimental and simulated distance distribution show a good match. The broadest distance distribution is obtained between the more flexible sites 67. For sites 166 an apparent long-distance shift is fully within the experimental uncertainty. A long-distance shift between the experimental and simulated distributions for sites 213 can partially be explained by overweighted simulated rotamers, which extend fully into the solvent phase. Simulation of the distribution of rotamers at the solvent-lipid interface is complicated and has, to our knowledge, not been done yet. In the rotamer library, no distinction is made between spin labels that experience a homogeneous environment (solvent or lipid) or inhomogeneous environment (e.g. located at the lipid-solvent interface), resulting in a larger uncertainty of the predicted rotamer distribution for such a case compared to the case of a homogeneous environment.
