Correlation analysis of Smad3 mutant protein binding assays and rescue of the Smad3-dependent reporter gene assay.

Protein binding data from Tables 2 and 3 and reporter gene activation data (RESCUE) from Table 2 were tested pairwise for their ability to correlate based on each sample's results with all of the Smad3 mutants using Kendall's rank correlation analysis. The data were analyzed using Mstat 5.4. Tau is the correlation coefficient. The p-value is shown with the false discovery rate (FDR) correction, which accounts for the multiple comparisons. The first three rows of data confirm expected results that, for the panel of hot-spot mutations examined, binding to Smad3 and Smad4 correlate, binding to full length SkiW274E and the Ski peptide correlate, and binding to full length SARA and the SARA peptide correlate. There is a positive correlation, not predicted by any prior studies, that a similar pattern of Smad3 hot-spots are involved in binding to both the SKI peptide and the SIP1 peptide. In contrast, there are significant negative correlations between Smad3 mutants binding to SARA versus Smurf2 peptide as well as between Smad3 mutants binding to Smad4 versus SARA peptide. The analysis also was used to correlate the protein binding properties of the hot spot mutations (LUMIER assay in 293 cells, Tables 2 and 3) to function (reporter gene assay in JEG3 cells, Figure 2). As expected, rescue of the Smad3-dependent reporter gene activation correlated with binding to Smad4. Rescue also significantly correlated with how the hot-spot mutations affected binding to the SIP1 peptide or SKI274E. In contrast, rescue of the reporter gene did not correlate with how the Smad3 mutant hot-spots affected binding to Smurf2 peptide or to SARA.