Fig 2: E2f1 recruits the Pontin/Reptin complex in TKO HCC.(a) Immunoprecipitation (IP) for IgG, E2f1 and E2f3 in TKO HCC cells. Endogenous E2f1 and E2f3 were pulled-down and the presence of Reptin was detected in the pulled-down fractions by immunoblotting. (b) IP for HA and Reptin in TKO HCC cells. Endogenous Reptin was pulled-down and the presence of E2f1, E2f2, E2f3, Pontin and H2a.z was detected in the pulled-down fractions by immunoblotting. H2a.z could only be detected upon IP/western blot. (c) IP for IgG, E2f1 and E2f3 in TKO HCC cells. Endogenous E2f1 and E2f3 were pulled-down and the presence of Pontin was detected in the pulled-down fractions by immunoblotting. (d) IP for IgG and Pontin in TKO HCC cells. Endogenous Pontin was pulled-down and the presence of E2f1 and Reptin was detected in the pulled-down fractions by immunoblotting. (e–g) GST constructs encompassing domains of human E2f1 were generated (e) and their interaction with endogenous Reptin from TKO HCC cells was determined by immunoblotting (f). (g) The endogenous IP described in a was repeated and increasing amounts of GST-E2f1-1-121 were added to test its capacity to displace the E2f1/Reptin interaction. The presence of endogenous Reptin in the E2f1 pulled-down fraction was detected by immunoblotting. The E2f1 antibody used in this experiment (C-20) recognizes the C-terminal domain of E2f1. (h) Nuclear extracts from TKO HCC cells were deposited at the top of a gradient and allowed to migrate through the gradient by ultracentrifugation. Twenty fractions were collected from the gradient and aliquots were run on a polyacrylamide gel to detect the migration pattern of Pontin, Reptin, E2f1 and H2a.z by immunoblotting. (i,j) Detection by IP of the interaction between E2f1 and Pontin/Reptin in human cancer cell lines: HepG2, Hep3B and SNU4499 (hHCC), NBLS and Be2c (neuroblastoma), DAOY (medulloblastoma), U2OS (osteosarcoma), T47D (breast cancer), K562 (chronic myeloid leukemia-CML) and EW8 (Ewing sarcoma). Asterisk marks cell lines where Pontin and/or Reptin interacts with E2f1. (k) E2f1, Pontin and Reptin expression levels in the different cell lines used in i,j as detected by immunoblotting. Normalization was assessed by Ponceau staining to ensure equivalent loading. *P<0.05; **P<0.01; ***P<0.001.

Fig 3: Progressive amplification of E2f response during cancer progression.Model for the evolution of E2f transcriptional output in TKO HCC. Upon Rb family loss, TKO liver cells enter cell cycle activity. Early lesions composed of proliferative Sca1+ progenitor-like cells can be detected after 2–5 weeks and gradually turn into late lesions macroscopically visible by ~4 months of age (TKO HCC). In quiescent cells, Rb family represses the transcriptional activity of E2f and the dimerization partner Dp1. Upon acute Rb family loss, E2f2 and E2f3 activate the transcription of cell cycle genes and E2f1. During TKO HCC progression, E2f1 gradually binds to the promoter region of E2f target genes and recruits Pontin and Reptin to incorporate H2a.z and amplify E2f transcriptional response. This mechanism leads to increased transactivation of cell cycle target genes (early genes displaying high-affinity E2f-binding sites in their promoter regions) and activation of non-cell-cycle target genes (late genes displaying low-affinity E2f-binding sites in their promoter regions), including genes that regulate the Warburg effect, Notch signalling and most likely other oncogenic features.

Fig 4: Progressive recruitment of Pontin/Reptin by E2f1 in TKO HCC.(a) E2f1, E2f2, E2f3, E2f4 and Gapdh (normalization) expression levels in control liver and TKO HCC, as detected by immunoblotting. (b) IP for HA, E2f1 and E2f3 in primary liver extracts from cTKO (d0), Rosa26-CreERT2 TKO 4 days after Tamoxifen treatment (d4) and TKO HCC. Reptin (upper panels) and Pontin (middle panels) were detected in the pull-down fractions at each time point by immunoblotting. Expression levels of E2f1, E2f3, Reptin and Pontin were detected at the different time points in the corresponding input fractions by immunoblotting (lower panels). (c) H2a.z, Pontin, Reptin and Actin protein expression in control liver and TKO HCC, as detected by immunoblotting. Ponceau staining is used for normalization purpose. (d) mRNA expression for Reptin and Pontin in control liver (n=5) and TKO HCC (n=9), as detected by qPCR. (e) mRNA expression for H2a.z in control liver (n=5) and TKO HCC (n=9), as detected by qPCR. (f) Binding of E2f1 to the promoter region of H2a.z (following the identification of an E2f-binding site in the promoter of H2a.z), as shown by ChIP assay in TKO HCC cells (n=3). (g) Repression of H2a.z expression in two clones of TKO HCC cells (2.1 and 1.1) upon infection with retroviruses either control (MigR1) or expressing the Rb-7LP protein (n=3). (h) Left: Frequency of E2F1 expression in tissue microarray (TMA) encompassing human HCC cases from stages I–IV. Right: Representative E2F1 staining of two independent HCC cases classified as stage IV HCC. Error bars represent standard deviation *P <0.05; **P <0.01; ***P<0.001. WB, western blot.

Fig 5: Accumulation of the H2a.z histone variant at E2f target genes in TKO HCC.(a) ChIP assay to detect the enrichment of common epigenetic marks at the promoter region of the panel of E2f target genes (see Fig. 2b for gene list) in TKO HCC cells, as determined by quantitative PCR (qPCR) analysis (n=3). Green: marks positively enriched; red: marks negatively enriched. (b) ChIP assay to detect the enrichment for the H3k9ac and the H3k4me3 marks (left) as well as the H2a.z histone variant (right) at the promoter region of the panel of E2f target genes in TKO HCC (black bar, long term Rb family loss) and in Rosa26 Cre-ERT2 TKO liver 4 days after Tamoxifen treatment (white bar, acute Rb family loss), as determined by qPCR analysis (n=3). The relative fold enrichment is compared for both conditions with the fold enrichment observed in control liver and arbitrarily set at 1 for the Acute Rb family loss condition. (c) ChIP-Seq was performed to identify the genomic sequences bound by E2f1 in TKO HCC. The graph determines the distance of E2f1 bound regions from the closest gene. (d) Venn diagram showing the overlap between E2f1 and H2a.z-bound genes in TKO HCC. P-value obtained by hypergeometric test. (e,f) E2f1 and E2f3 were pulled-down from TKO HCC lysates and the pull-down fractions were allowed to migrate on an acrylamide gel subsequently stained with Coomassie Blue for visualization (represented in f) and processed by mass spectrometry (n=3). The relative abundance of E2f1 (arbitrarily set at 100%), Dp1 (positive control) and Pontin/Reptin in the E2f1 pull-down fraction is indicated on the right of the display. All these proteins were isolated from the same band migrating at ~50 kDa. (e) Summary of the principal findings of the MS performed with E2f1 and E2f3 in TKO HCC. The complete results are displayed in Supplementary Data 5. Error bars represent standard deviation *P <0.05; **P<0.01; ***P<0.001.