Supplementary Figure 1: Structural characterization of IL-26.
From: TH17 cells promote microbial killing and innate immune sensing of DNA via interleukin 26
(a) Amino acid sequence of IL-26. Amino acids within the 6 predicted α-helical regions are labeled in red. Positively charged amino acids clustering on one side of the helix bundle are indicated in bold. Hydrophobic residues exposed on the other side of the bundle are underlined. (b) SAXS allows for study of the 3D structure of proteins in solution. Although this method does not reveal atomic details, its resolution is sufficient to determine the multimeric state and shape of a protein. SAXS analysis yields a scattering pattern (black dots). From this pattern, the radius of gyration Rg of the particle in solution can be derived in two model-independent ways: (i) from the Guinier analysis (Rg = 5.68 ± 0.1 nm), and (ii) through the indirect transform algorithm implemented in the program GNOM (5.59 ± 0.1 nm). The algorithm used in GNOM also provides the maximum diameter Dm of the solute particle (here 18.8 ± 0.2 nm). These experimentally determined Rg and Dm values are substantially larger than those calculated from a monomeric IL-26 structural model (Rg = 1.5 nm; Dm = 5.5 nm), or from dimeric IL-26 models (based on IL-22: Rg = 1.8 nm, Dm = 5.9 nm; based on IL-10: Rg = 2.3 nm, Dm = 7.5 nm), indicating that IL-26 forms multimers in our experimental conditions. (c) SAXS data can be used to determine a model-independent ab initio shape of the solute particle. For our data, this ab initio shape resembled four linearly connected spheres (gray beads obtained by the program DAMMIF). Each sphere has the dimensions of one IL-26 monomer. As illustration, one IL-26 molecule (secondary structure representation, color-ramped from blue: N terminus to red: C terminus) was placed by hand in the ab initio SAXS shape. Based on these results, we used the program SASREF to position four IL-26 monomers as a tetramer that best fits the experimental SAXS pattern. The calculated SAXS pattern (red line in b, produced with CRYSOL) of the resulting tetramer fitted the experimental data (black dots, b) with χ = 0.99. (For a good model, χ values should be close to 1. Values much greater than 1 indicate a poor fit of model to data, and χ values much smaller than 1 indicate over-fitting of the data.) SAXS yields the average size of all solute particles. Hence our analysis does not rule out the presence of minor populations of other multimers (3-mers and 5-mers, for example), which are expected to co-exist in a concentration-dependent equilibrium. Repeat runs of SASREF or DAMMIF (both of which use a Monte Carlo algorithm and simulated annealing, and hence can produce different models for each run) produce linear structures with similar, but not identical, curvature, each of which achieves a comparable χ fit. This observation suggests that IL-26 molecules do not form a rigid rod, and retain some flexibility between protomers. (d) Four cysteines are well placed to form disulfide bonds that stabilize the compact helix bundle. The atomic model of IL-26 (colored as in c) was obtained by homology modeling.
