Supplementary Figure 2: Genotype-phenotype relationship in STAT3 alterations.

The predicted effects of the STAT3 alterations were modeled in PDB structure 1BG1 (mouse STAT3/DNA complex) using SWISS-MODEL and visualized in the Swiss-PdbViewer. (a) Overview of the STAT3 dimer bound to DNA; STAT3 chains are shown in ribbon form, with residues N646 (red) and N647 (green) shown as space-filling residues on the left chain only; DNA strands are shown as blue and turquoise ribbons. (b) As in a, but expanded to show the proximity of residues N646 and N647 to both the DNA-binding and dimerization surfaces. (c) Predicted molecular surfaces of wild-type STAT3 (wt) and mutants N646K, N647D and N647I; surfaces are colored for positive charge (blue; top row), negative charge (red; middle row) and hydrophobicity (brown (most polar) to blue (most hydrophobic); bottom row); structures have been rotated compared to in a and b to show relevant groups more clearly. The N646K alteration reported here results in increased positive charge (circled, N646K column, upper row) at the DNA-binding surface; this is likely to result in higher DNA binding affinity due to electrostatic interaction with the DNA backbone and, hence, increased STAT3 activity. Conversely, the N647D substitution, previously reported as a loss-of-function alteration in HIES, leads to increased negative surface charge in this region (circled, N647D column, middle row) and is likely to inhibit DNA binding and/or dimerization. By comparison, a different substitution at this position, N647I, has been previously reported as an activating alteration in LGLL; it has been postulated that STAT3 mutations in LGLL promote STAT3 dimerization and, hence, biological activity, as a result of increased hydrophobicity at the dimerization surface. This is consistent with protein modeling in silico, which predicts increased hydrophobicity in this region (circled, N646I column, bottom row) compared to wild-type STAT3 or other variants.