Statistical analysis

All RT-PCR, western, and immunoprecipitation experiments were done at least 3 times. All IHC and IF pictures are representative of at least 5 mice in each genotype. Pictures from the first human patient sample tumor array are representative of 60 case pairs of normal and tumor samples. All data are expressed as the mean ± SD. All statistical tests were 2-sided. P-values of less than 0.05 were considered statistically significant. For each measurement variable, the distribution of normality was conducted before statistical analysis of testing group differences. We chose either parametric or non-parametric version of statistical procedure as appropriate based on the data distribution.

Differences between two groups were analyzed using two-sample Student's t-test or Wilcoxon rank sum test; differences among more than two groups were analyzed using one-way ANOVA or Kruskal-Wallis test as appropriate. Multiple comparisons of mean post hoc test for one-way ANOVA were adjusted using Tukey's method [52], while multiple comparisons of mean ranks post hoc test with a Kruskal-Wallis analysis were adjusted using Nemenyi (Tukey-type) test (P. Nemenyi, Distribution-Free Multiple Comparisons, State University of New York, Downstate Medical Center, 1963 (cited in Wilcoxon and Wilcox, 1964)). For equal group sizes or Dunn's test [53]. For unequal group sizes or some ranks tied [54]. Statistical analyses were performed using SAS version 9.4 (SAS Institute, Inc., Cary, NC, USA).