SAXS data acquisition, analysis and modelling

SAXS data were measured at the Australian Synchrotron on the SAXS/WAXS beamline⁵⁷ with the data collection parameters presented in Table S2. Data were reduced to I(q) versus q (I(q) = 4πsinθ/λ and 2θ is the angle between the incident and scattered X-rays, λ their wavelength) using the software scatterBrain [ http://www. synchrotron. org.au/aussyncbeamlines/saxswaxs/software-saxswaxs]. Intensities were placed on an absolute scale using the known scattering from H₂O. All samples were prepared as described above under Recombinant protein expression and purification. SAXS measurements were performed on two consecutive elution fractions from size exclusion chromatography (representing the front half and back half of the peak profile, termed fractions A and B) after dilution to a starting concentration of 1.5 mg/mL according to BCA assays. SAXS data were collected on a concentration series representing 100%, 80%, 60%, 40%, 20% and 10% of the starting concentration of each of the four samples (wild type rIgSF-CTD and the r664i6H fractions A and B). All dilutions were made in gel filtration buffer.

A solvent scattering blank was measured on the gel filtration buffer alone. The scattering profiles for wild type rIgSF-CTD and the r664i6H variant were obtained by subtraction of the corresponding solvent blank scattering from the samples (protein + solvent) scattering. There was no discernable concentration dependence to the scattering profiles, nor any difference in the highest concentration measurement from fractions A and B from either wild type rIgSF-CTD or the r664i6H variant. These highest concentration measurements were therefore averaged to improve signal to noise and used for all further analysis. Molecular weight (MW) estimates for the proteins were made using the method of Orthaber et. al.⁵⁸. Values for contrast and partial specific volumes were determined using the program MULCh⁵⁹. SAXS data analysis and modeling used the ATSAS program package⁶⁰, with the specific programs used are specified in the Table S2 along with the data ranges used to determine the given structural parameters.

Both constructs were shown to be monodisperse in solution by virtue of the linear log I(q) versus q² plot (Guinier plot, Fig. S5), good agreement between R_g values determined by Guinier or P(r) analysis (Table S2) and molecular weight estimates that are within 10% of the expected values from the forward scattering intensity (I(0)) and from shape reconstructions (modelled using DAMMIF) (Table S2).

DAMMIF modelling calculations were submitted to ATSAS on-line, with P1 symmetry for 20 independent runs that yielded a set of similar shapes for each construct with NSD values all significantly less than 1 as follows (standard deviations in parentheses): rIgSF-CTD (wild type) 0.689(0.032), r664i6H 0.551(0.045). BUNCH modelling used default parameters with the structures for the IgSF (residues 579–662) and CTD (residues 672–736) domains from the crystal structure (PDB code 5AG8), with no structure assigned to the respective linkers for the wild type (residues 663–671 inclusive) and for the r664i6H variant (residues 663–671 inclusive plus the hexahistidine insert) so that the domains were free to adopt positions that optimised the fit to the scattering data. Ensemble Optimisation Modelling (EOM) used the same domain structures and flexible linkers and was run using the ATSAS on-line server with default parameters.