Figure 5. Analysis of epistasis between Brn3a and Islet1 mutations.

Epistatic interactions between the Brn3a KO and Islet1 DKO genotypes were examined for 479 transcripts which are at least 3-fold changed relative to WT in at least one of the three mutant genotypes. (A) Two-dimensional epistasis plot in which the x-axis represents the change in expression due to the loss of Brn3a function in the presence of Islet1, based on a the difference between the WT and Brn3aKO expression levels, normalized to the maximum expression in any genotype: (E_WT - E_B3aKO)/max. The y-axis represents the change in expression due to the loss of Brn3a in the absence of Islet1, based on the difference between the Islet1 CKO and DKO expression levels: (E_Isl1CKO - E_DKO)/max. Transcripts for which the effects of the Brn3a KO and Islet1 CKO genotypes are additive appear on the diagonal. Transcripts showing negative epistasis are shifted to the left and right of the diagonal, and those showing positive epistasis are shifted above and below the diagonal. (B) Individual regulated transcripts exhibit negative epistasis, additive effects, and positive epistasis between the Brn3a KO and Islet1 CKO genotypes. (C) Distribution plots of difference values for comparisons between WT and DKO ganglia for the most-changed 479 transcripts and a control set of randomly selected transcripts. (D) Histogram plots of distributions of (E_WT - E_B3aKO)/max and (E_Isl1CKO - E_DKO)/max. (E) Distributions of (E_WT - E_Isl1KO)/max and (E_Brn3aCKO - E_DKO)/max. Comparisons of WT to single knockouts exhibit broader distributions than comparisons between single knockouts and DKO ganglia, confirming the predominant effect of global negative epistasis. P-values are derived from K-S tests and represent the probability of observing the null hypothesis that the distributions do not differ significantly.