Supplementary Figure 4: Structural and dynamics characterization of rdHDL based on NMR data.

(a) Sequence-specific ¹³C^α secondary chemical shifts (δΔ¹³C^α) along the amino acid sequence of MSPΔH5 (black bars) overlaid with the TALOS-N secondary structure prediction probability (pink points) based on ¹H^N, ¹⁵N, ¹³CO and ¹³C^α chemical shifts. Both entities suggest a helical secondary structure throughout the protein. δΔ¹³C^α is the difference between measured chemical shift and the corresponding random coil chemical shift of a perdeuterated protein at a temperature of 42°C and a pH of 7.4 determined using the database from the University of Copenhagen (see Materials and Methods). Positive δΔ¹³C^α values indicate an α-helical conformation. (b) [¹H]¹⁵N-hetero-nuclear NOE data (hetNOE) along the amino acid sequence (black bars). Negative or near zero values indicate highly flexible ¹⁵N-¹H moieties with motions faster than ca. 1 ns. The error of the measurement is shown as red markers, whereas the back-calculated values from Fast ModelFree are shown as blue markers. Values close to 1 indicate the absence of such motions. (c) Longitudinal ¹⁵N R1 relaxation rates (black bars) versus the amino acid sequence. ¹⁵N-¹H moieties with larger R1 rates indicate the presence of higher flexibility in the ps-ns time range. In particular, the termini show some mobility. The error of the measurement is shown as red markers, whereas the back-calculated values from Fast ModelFree are shown as blue markers. (d) Transverse ¹⁵N R2 (R1ρ) relaxation rates versus the amino acid sequence (black bars) recorded with a spin-lock field of 2 kHz, which effectively removed exchange contributions slower than 80 μs. The error of the measurement is shown as red markers, whereas the back-calculated values from Fast ModelFree are shown as blue markers. (e) Representation of the calculated order parameter S² (left axis) and internal correlation time τ_e (right axis). Both values were calculated using relax (green markers for the order parameter and green bars for the internal correlation time) and Fast ModelFree (blue markers for the order parameter and blue bars for the internal correlation time). The back-calculated hetNOE, R1 and R1ρ values from Fast Modelfree are used for panels (b)-(e). (f) Transverse relaxation ¹⁵N R2_β relaxation rates versus the amino acid sequence measured on the slowly relaxing β-component at 900 (black circles) and 700 (red circles) MHz ¹H frequency. A few ¹⁵N-¹H moieties (such as at the N-terminus) have faster relaxation rates at 900 MHz than at 700 MHz, indicating the presence of local chemical exchange. (g) Transverse cross-correlated relaxation, η_xy, was determined as described previously and converted into the rotational correlation time in a similar way as with the TRACT experiment. An average correlation time τ_c of 34 ns was determined, identical to previous TRACT measurements. The homogeneity of τ_c along the sequence indicates uniform tumbling as one entity. (h) Presaturation experiments were performed on a 900 MHz spectrometer (black bars) and a 600 MHz spectrometer (red dots) using a presaturation time of 1 sec. Two different protein batches were used to test the reproducibility. Error bars in panel b, c, d and f correspond to one standard deviation from 30 single-exponential fits of the experimental relaxation rates.