Combinatorial MAPK states drive opposing gene programmes: a reanalysis of MAP3K-driven transcriptional responses in MCF10A cells
TL;DR
It is demonstrated that MAP3Ks co-activating ERK and JNK drive approximately tenfold more cell cycle entry than MAP3Ks activating ERK alone, and that both kinase activities are independently required for this proliferative response.
Abstract
Background The MAPK signalling network coordinates opposing cell fates — proliferation versus apoptosis — by combining the activities of ERK, JNK, and p38 kinases in patterns that depend on upstream MAP3K identity. Using multiplexed kinase translocation reporter (KTR) biosensors and functional assays in MCF10A mammary epithelial cells, Peterson et al. (Cell Systems, 2023) demonstrated that MAP3Ks co-activating ERK and JNK drive approximately tenfold more cell cycle entry than MAP3Ks activating ERK alone, and that both kinase activities are independently required for this proliferative response. The transcriptional mechanism underlying this quantitative difference was not identified in the original study. Results We performed a systematic computational re-analysis of the RNA-seq data deposited by Peterson et al. (GEO: GSE213882), encompassing thirteen MAP3K perturbations in doxycycline-induced MCF10A mammary epithelial cells. Based on signalling responses reported by Peterson et al., MAP3Ks were grouped into ERK-only (RAF1, COT) or ERK+JNK co-activating kinases (MLK1, MLK3, ZAK, TAK1, MEKK2, MEKK3). This stratification identified JNK co-activation as the feature most strongly associated with E2F/MYC transcriptional polarity. ERK-only MAP3Ks showed negative enrichment of E2F TARGETS, G2M CHECKPOINT, and MYC TARGETS Hallmark gene sets, whereas ERK+JNK MAP3Ks showed positive enrichment (each MAP3K individually FDR < 0.05 by GSEA). Consistently, inferred activities of E2F1, E2F2, E2F4, and MYC shifted from negative mean ULM t-scores in the ERK group to strongly positive values in the ERK+JNK group (FDR < 0.05 using the CollecTRI regulon). The observed E2F/MYC polarity was robust to alternative transcription factor inference strategies, being reproduced using the DoRothEA A+B regulon and the multivariate model (MLM) framework (Pearson r = 0.77). The signal remained stable under leave-one-out sensitivity analysis and was independently reproduced in MAP3K overexpression signatures from the LINCS L1000 atlas (GEO: GSE92742). Conclusions ERK activation alone was associated with repression of E2F/MYC transcriptional programmes, whereas JNK co-activation was associated with their activation. These opposing transcriptional states provide a potential explanation for the ∼10-fold differences in cell cycle entry reported by Peterson et al. and supports a model in which pRb/E2F regulation integrates combinatorial MAPK co-activation states beyond the canonical ERK–cyclin D–CDK4/6 axis.