c-MYC-Induced AP4 Attenuates DREAM-Mediated Repression by p53

Simple Summary Deregulated expression of the c-MYC oncogene activates the tumor suppressor p53, which has been suggested to represent a failsafe mechanism against the uncontrolled expansion of tumor cells. Here, we analyzed the role of the c-MYC-induced TFAP4/AP4 gene in this context using a genetic approach in MCF-7 breast cancer cells. Inactivation of AP4 resulted in elevated levels of both spontaneous and c-MYC-induced DNA damage, senescence, and diminished cell proliferation. Inactivation of p53 in AP4-deficient cells reverted senescence and proliferative defects. Furthermore, loss of AP4 resulted in p53-dependenct, enhanced repression of DREAM and E2F target genes after the induction of c-MYC, which could be abrogated by the concomitant depletion of p21 or the DREAM complex component LIN37. These p53-dependent effects were reflected on the levels of gene expressions and clinical associations in primary breast cancer tumors from patient cohorts. Our results established AP4 as a pivotal factor at the crossroads of c-MYC, E2F, and p53-mediated target gene regulation. Abstract Background: The deregulated expression of the c-MYC oncogene activates p53, which is presumably mediated by ARF/INK4, as well as replication-stress-induced DNA damage. Here, we aimed to determine whether the c-MYC-inducible AP4 transcription factor plays a role in this context using a genetic approach. Methods: We used a CRISPR/Cas9 approach to generate AP4- and/or p53-deficient derivatives of MCF-7 breast cancer cells harboring an ectopic, inducible c-MYC allele. Cell proliferation, senescence, DNA damage, and comprehensive RNA expression profiles were determined after activation of c-MYC. In addition, we analyzed the expression data from primary breast cancer samples. Results: Loss of AP4 resulted in elevated levels of both spontaneous and c-MYC-induced DNA damage, senescence, and diminished cell proliferation. Deletion of p53 in AP4-deficient cells reverted senescence and proliferation defects without affecting DNA damage levels. RNA-Seq analyses showed that loss of AP4 enhanced repression of DREAM and E2F target genes after p53 activation by c-MYC. Depletion of p21 or the DREAM complex component LIN37 abrogated this effect. These p53-dependent effects were conserved on the level of clinical and gene expression associations found in primary breast cancer tumors. Conclusions: Our results establish AP4 as a pivotal factor at the crossroads of c-MYC, E2F, and p53 target gene regulation.


Introduction
The c-MYC transcription factor is encoded by a proto-oncogene, which shows elevated and/or deregulated expression in more than 70% of all cancers [1]. In breast cancer, c-MYC is expressed at elevated levels in 30-50% and amplified in nearly 15% of cases [2].

Assessment of Proliferation by Real-Time Impedance Measurement
Cell proliferation was measured with impedance measurements (X-celligence RTCA DP, Roche). Cells were seeded at a density of 6 × 10 3 cells in a 96-well microtiter plate (E-Plate Cardio 96, Agilent, Santa Clara, CA, USA) in triplicate and subjected to the indicated treatments. Cellular impedance was measured every 60 min for a period of up to 72 h with the X-celligence system (Roche). In parallel, cells were also seeded into 48-well plates in triplicate and counted at the end time point using a Neubauer chamber to validate impedance measurements.

Colony Formation Assay
A total of 1 × 10 3 cells per well were seeded into 6-well plates and cultured for 14 days. Subsequently, cells were stained with crystal violet after fixation.

Transcriptomic Analysis
Total RNA from MCF-7/pRTR-c-MYC cells was isolated using a High Pure RNA Isolation Kit (Roche). Random primed cDNA libraries were constructed and sequenced using the NovaSeq 6000 (Illumina, San Diego, CA, USA) platform by GATC (Konstanz, Germany). Each sample was covered by at least 30 million paired-end read pairs of 150 bp length. RNA-Seq FASTQ files were processed using the RNA-Seq module implemented in the CLC Genomics Workbench v20.0.2 software (Qiagen Bioinformatics, Dusseldorf, Germany) and mapped to the GRCh38/hg38 human reference genome and its associated gene and transcript annotation (ENSEMBL) using the settings mismatch cost = 2, insertion cost = 2, deletion cost = 3, length fraction = 0.8, and similarity fraction = 0.8. RNA-Seq data were filtered to exclude weakly expressed transcripts with less than 20 mapped exon reads in all samples from the analysis and subjected to upper quartile normalization using the R/Bioconductor RUVSeq (remove unwanted variation from RNA-Seq data) package as described in Risso et al. [28]. Differential gene expression analysis was performed with DESeq2 [29] after normalization using the RUVg approach to remove variation between RNA samples resulting from differences in library preparation. Principal component analysis (PCA) was performed using the PCA functionality of the EDASeq R package as implemented in RUVSeq. Gene set enrichment analysis (GSEA) was performed with the fgsea R package [30]. Prior to GSEA, expression changes from low-count genes were adjusted using the ashr (adaptive shrinkage) estimator [31]. The significance of enrichments is presented by normalized enrichment scores (NES) and Benjamini-Hochberg-adjusted p values. Heat-maps were generated with Morpheus (Broad Institute, Cambridge, MA, USA). Gene sets were obtained from the Molecular Signatures database (MSigDB) [32].

Analysis of ChIP-Seq, RNA Expression, and Clinical Data from Public Databases
Direct regulation of DREAM targets was assessed by analysis of publicly available ChIP-Seq data for E2F4 (SRX194566, MCF-7), LIN9 (SRX4213896, MCF-10A), and RBL2 (SRX016031, IMR90) obtained from ChIP-Atlas (https://chip-atlas.org (accessed on 19 December 2022)) [33]. The Integrative Genomics Viewer (IGV) [34] was used for the visualization of ChIP-Seq profiles. For the analysis of human breast cancer samples, we retrieved expression and clinical data from the TCGA-BRCA cohort [35]. The statistics for forest plots were calculated with a log-rank test. For binary classification of cases (high/low expression), the Survminer R package (https://CRAN.R-project.org/package=survminer (accessed on 18 February 2020)) was used to determine optimal cutoff values.

Statistical Analysis
Results are presented as mean +/− standard deviation (SD). Each set of experiments was repeated at least three times. A Student's t-test was used to test the null hypothesis stating no significant differences between each individual parameter were measured. Differences were considered to be significant if p < 0.05. Statistics were performed with Prism 8 (GraphPad Software, San Diego, CA, USA).

Generation and Characterization of AP4-and/or p53-Deficient MCF-7/pRTR-c-MYC Cell Lines
In order to study the role of AP4 downstream of c-MYC in breast cancer cells, AP4 was inactivated in MCF-7/pRTR-c-MYC cells using a CRISPR/Cas9 approach, as described earlier [24]. In addition, we inactivated p53 in these cells by introducing InDel mutations into exon 3 of p53 by CRISPR/Cas9, as described previously [26]. The induction of the ectopic c-MYC allele, activation of AP4, and loss of p53 expression was verified by Western blot analysis ( Figure 1A). The c-MYC-induced elevation of p53 protein levels was similar in AP4 wild-type and AP4-deficient cells, indicating that loss of AP4 had no significant effect on the c-MYC-mediated activation of p53. As expected, basal expression levels, as well as c-MYC-induced up-regulation of the direct p53 targets p21 and MDM2, were highly dependent on the presence of p53 in these cells ( Figure 1A(lower panel),B). Moreover, in AP4-deficient cells harboring wild-type p53, basal expression, as well as c-MYC-induced up-regulation of p21 and MDM2, were elevated when compared to AP4 wild-type cells ( Figure 1A). Similar results were obtained by qPCR analysis of p21 and MDM2 mRNA expression ( Figure 1B). Hence, both p21 and MDM2 expression are antagonistically regulated by AP4 and p53 after induction of c-MYC in MCF-7 cells.

Loss of AP4 Suppresses Induction of Cell Proliferation by Ectopic c-MYC
Activation of ectopic c-MYC by DOX treatment resulted in a significant increase in proliferation in MCF-7 cells with wild-type AP4 and p53 alleles, as shown by cellular impedance measurements. The induction of cell proliferation was severely diminished in AP4-deficient p53 wild-type cells (Figure 2A). In the absence of ectopic c-MYC expression, the inactivation of AP4 resulted in a pronounced decrease in cell proliferation when compared with AP4-proficient cells ( Figure 2B). Interestingly, the deletion of p53 not only reverted the proliferative defect of AP4-deficient cells but led to a strong increase in proliferative capacity independent of the AP4 status ( Figure 2B). Furthermore, an enhancement of cell proliferation by activation of ectopic c-MYC could not be observed after deletion of p53 irrespective of the AP4 status, presumably since p53-negative MCF-7 cells are already proliferating at the maximum rate ( Figure

Loss of AP4 Suppresses Induction of Cell Proliferation by Ectopic c-MYC
Activation of ectopic c-MYC by DOX treatment resulted in a significant increase in proliferation in MCF-7 cells with wild-type AP4 and p53 alleles, as shown by cellular impedance measurements. The induction of cell proliferation was severely diminished in   The p53-dependent proliferative defects of AP4-deficient cells were confirmed by saying colony formation ( Figure 2D). Here, the loss of AP4 in p53 wild-type cells result in decreased colony-forming capacity when compared to AP4 wild-type cells. Converse the loss of p53 dramatically increased colony formation. Furthermore, the deletion of p in AP4-deficient cells reverted their decreased colony-forming capacity and resulted The p53-dependent proliferative defects of AP4-deficient cells were confirmed by assaying colony formation ( Figure 2D). Here, the loss of AP4 in p53 wild-type cells resulted in decreased colony-forming capacity when compared to AP4 wild-type cells. Conversely, the loss of p53 dramatically increased colony formation. Furthermore, the deletion of p53 in AP4-deficient cells reverted their decreased colony-forming capacity and resulted in colony numbers comparable to p53-deficient/AP4 wild-type cells. Interestingly, prolonged activation of c-MYC for two weeks resulted in an overall decrease in the number and size of colonies, irrespective of AP4 and p53 status. However, while this suppression of colony formation was less pronounced in AP4 wild-type/p53-deficient cells compared with AP4/p53 wild-type cells, it was dramatically enhanced in AP4/p53-deficient cells, suggesting that AP4/p53-deficient cells were particularly sensitive to prolonged activation of c-MYC. Taken together, these results showed that the decreased proliferation of AP4-deficient MCF-7 cells was dependent on the presence of wild-type p53, suggesting that p53 acts as a major suppressor of proliferation in these cells. These findings were different from our previously published results obtained in p53-mutant CRC cell lines SW480 and DLD1 [24], where deletion of AP4 resulted in a significantly diminished proliferative capacity.

Loss of AP4 Causes Senescence in Breast Cancer Cells, Which Is Dependent on Wild-Type p53
Next, we analyzed whether the decreased proliferation of AP4-deficient cells was associated with increased senescence, as shown previously in CRC cell lines and MEFs [24,36]. The fraction of senescent, β-galactosidase positive cells was elevated in untreated AP4deficient cells when compared with AP4 wild-type cells ( Figure 3A). Deletion of p53 resulted in a decrease in basal senescence. Furthermore, the deletion of p53 reverted the increased basal senescence levels observed in AP4-deficient cells. Induction of ectopic c-MYC for up to 72 h led to a significant increase in the fraction of senescent, β-galactosidase positive cells independently of the AP4 or p53 status ( Figure 3A,B). However, in AP4-deficient/p53 wild-type cells, the frequency of β-galactosidase positive cells was higher when compared with AP4/p53 wild-type cells after activation of c-MYC for up to 72 h. Moreover, the deletion of p53 reverted the increased senescence levels observed in AP4-deficient cells after activation of c-MYC. Taken together, these results showed that the loss of AP4 results in increased senescence in MCF-7 cells and strongly suggested that functional p53 mediates, at least in part, the induction of a senescent phenotype in AP4-deficient cells. However, the relative increase in senescence after activation of c-MYC appeared to be largely independent of p53.

Deletion of AP4 or p53 Increases Spontaneous and c-MYC-Induced DNA Damage in Breast Cancer Cells
We have previously shown that AP4 suppresses DNA damage, which occurs spontaneously or at an increased rate after c-MYC activation. AP4 suppresses DNA damage by directly and indirectly (via repressing miR-22) inducing MDC1/Mediator of DNA damage checkpoint 1 [24]. Therefore, we determined whether the loss of AP4 in breast cancer cells also increases DNA damage. Indeed, after the deletion of AP4, increased levels of γH2AX-positive nuclear foci were detected in MCF-7 cells (Figures 4A and S2). Induction of ectopic c-MYC for up to 72 h led to a significant and time-dependent increase in the fraction of γH2AX-positive cells irrespective of their AP4 and p53 status ( Figure 4A). Furthermore, both basal and c-MYC-induced levels of DNA damage were elevated in AP4and/or p53-deficient cells when compared with AP4/p53 wild-type cells. These results were corroborated by comet assays (Figures 4B and S3), which also showed that both spontaneous and c-MYC-induced DNA damage was strongly elevated in AP4/p53-deficient compared to AP4/p53 wild-type cells. In addition, a dramatic increase in micronuclei was detected in AP4and AP4/p53 double-deficient cells after induction of c-MYC, which could not be observed in AP4/p53 wild-type or p53-deficient cells ( Figures 4C and S4). Micronuclei result from the missegregation of chromosomes during mitosis. Since AP4and AP4/p53 doubledeficient cells accumulate high levels of DNA damage after c-MYC activation, they may enter mitosis with unrepaired DNA damage, causing the missegregation of chromosomes.

Characterization of AP4-and p53-Dependent Effects in the c-MYC-Regulated Transcriptome
To determine the potential impact of AP4 and/or p53 on c-MYC-induced differential gene expression, we performed a comprehensive Next Generation Sequencing (NGS) analysis after the activation of ectopic c-MYC in the AP4-and/or p53-deficient MCF-7 cells characterized above. To do so, cells were pre-treated with ICI for 72 h in order to downregulate endogenous c-MYC. Ectopic expression of c-MYC was induced by addition of DOX for 48 h in the presence of ICI. For each of the four genotypes, NGS libraries representing RNAs isolated from both DOX-treated (n = 3) and ICI-only (i.e., un-) treated cells Furthermore, the number of bi-nucleated cells was highly elevated in AP4and AP4/p53-deficient cells when compared with AP4/p53 wild-type cells (Figures4D and S5). p53-deficient cells also displayed an increase in the number of bi-nucleated cells, albeit to a lesser extent. Similar to micronuclei, bi-nucleated cells presumably result from unrepaired DNA damage in AP4-deficient cells, which leads to incomplete chromosome segregation and incomplete cytokinesis. These results show that while AP4and AP4/p53-deficient cells accumulate high levels of either spontaneous or c-MYC-induced DNA damage, the de-creased proliferation of AP4-deficient cells is completely reverted by additional inactivation of p53.

Characterization of AP4-and p53-Dependent Effects in the c-MYC-Regulated Transcriptome
To determine the potential impact of AP4 and/or p53 on c-MYC-induced differential gene expression, we performed a comprehensive Next Generation Sequencing (NGS) analysis after the activation of ectopic c-MYC in the AP4and/or p53-deficient MCF-7 cells characterized above. To do so, cells were pre-treated with ICI for 72 h in order to downregulate endogenous c-MYC. Ectopic expression of c-MYC was induced by addition of DOX for 48 h in the presence of ICI. For each of the four genotypes, NGS libraries representing RNAs isolated from both DOX-treated (n = 3) and ICI-only (i.e., un-) treated cells (n = 3) were generated and subjected to RNA-Seq analysis with more than 30 million paired-end reads per library. Principal component analysis (PCA) showed that both untreated and DOX-treated AP4/p53 wild-type, AP4and/or p53-deficient MCF-7/pRTR-c-MYC cells were characterized by distinct transcriptomes ( Figure 5A).
Of note, the majority of variation between DOX-treated and untreated cells was captured by principal component (PC) one in all genotypes, strongly indicating that c-MYCinduced gene expression changes were similar between the different genotypes for a large number of genes.
Differential gene expression analyses using DESeq2 showed that, in MCF-7/pRTR-c-MYC cells (AP4/p53 wild-type), 953 genes were significantly up-regulated, and 1328 genes were down-regulated after treatment with DOX for 48 h ( Figure 5B, Table S5). In AP4deficient/p53 wild-type MCF-7/pRTR-c-MYC cells, 1101 genes were significantly up-regulated, and 1410 genes were down-regulated after activation of c-MYC by DOX treatment ( Figure 5B, Table S6). In AP4-proficient/p53-deficient MCF-7/pRTR-c-MYC cells, 787 genes were significantly up-regulated, and 1051 genes were down-regulated after DOX treatment ( Figure 5B, Table S7). In AP4/p53-deficient MCF-7/pRTR-c-MYC cells, 1225 genes were significantly up-regulated, and 1803 genes were down-regulated after DOX treatment ( Figure 5B, Table S8). Interestingly, the overlap between mRNAs differentially either up-or down-regulated (≥1.5× fold change) genes in AP4-deficient or AP4 wild-type MCF-7/pRTR-c-MYC cells after treatment with DOX was substantial but not complete, irrespective of the p53 status ( Figure 5C,D). Likewise, the overlap between mRNAs differentially either up-or down-regulated (≥1.5× fold change) in p53-deficient or p53 wild-type MCF-7/pRTR-c-MYC cells after treatment with DOX was substantial but not complete, irrespective of the AP4 status ( Figure 5E,F). This suggested that the regulation of molecular and cellular pathways by c-MYC while sharing commonalities, showed differences that were dependent on the AP4 and/or p53 status of the respective cells. However, using this approach, we observed very little or no overlap between genes showing strong opposing regulation (≥1.5× fold change up-or down-regulation) in AP4 wild-type and AP4 KO cells, neither in p53 wild-type or KO background. Notably, one of the three genes displaying opposing regulation in p53 wild-type versus p53 KO cells irrespective of their AP4 status ( Figure 5E,F) was CDKN1A/p21, as shown by qPCR ( Figure 1B), suggesting that its upregulation after activation of c-MYC may be a critical factor for limiting the c-MYC-induced increase in proliferation in p53 wild-type cells.
Next, we employed gene set enrichment analyses (GSEA) in order to identify molecular and cellular pathways which display differences in the regulation of their components in MCF-7/pRTR-c-MYC cells with divergent AP4 and/or p53 status after treatment with DOX ( Figure 6). Direct c-MYC targets were activated after treatment with DOX, irrespective of the AP4 or p53 status. As expected, the activation of p53 targets after treatment with DOX was highly dependent on the presence of functional p53.
(n = 3) were generated and subjected to RNA-Seq analysis with more than 30 million paired-end reads per library. Principal component analysis (PCA) showed that both untreated and DOX-treated AP4/p53 wild-type, AP4-and/or p53-deficient MCF-7/pRTR-c-MYC cells were characterized by distinct transcriptomes ( Figure 5A).  Of note, a direct comparison of basal expression levels (i.e., MCF-7 cells not treated with DOX) indicated that the expression of E2F/DREAM targets, as well as the functional categories represented by these, was increased in p53-deficient cells when compared with AP4/p53 wild-type cells, suggesting an inhibitory effect of wild-type p53 on their expression ( Figure 6). The elevated basal expression of genes involved in cell cycle progression in p53-deficient cells, irrespective of their AP4 status, may explain the increased proliferation of these cells, as well as their lack of responsiveness to ectopic c-MYC. The basal expression of E2F/DREAM targets in AP4-deficient cells was increased compared to AP4/p53 wild-type cells ( Figure 6).
In order to identify genes with similar genotype-dependent differences in regulations after activation of c-MYC, we employed a two-factor (genotype and treatment) interaction analysis design using DESeq2, followed by KMeans clustering (n = 20) ( Figure 7A). In total, we identified 2309 genes that displayed genotype-dependent differences in c-MYCinduced regulation ( Figure S6, Table S9). Next, we determined which functional categories were significantly over-represented in at least one of the identified transcriptional clusters. Thereby, we identified a strong enrichment of E2F/DREAM targets, as well as pathways involved in cell cycle progression in the transcriptional clusters 1 and 2 ( Figure  7B). Interestingly, the DREAM target genes in these clusters were characterized by elevated basal expression in AP4-and/or p53-deficient cells and p53-dependent down-regulation in AP4-deficient cells ( Figure 8A), thus corroborating the findings of GSEA ( Figure  6). A total of 202 E2F/DREAM targets were associated with clusters 1 (129/156) and 2 (73/155) (listed in Table S10). Of note, we also identified a subset of DREAM targets overrepresented in cluster 3 (34/99), which does not display p53-dependent down-regulation in AP4-deficient cells but is induced upon activation of c-MYC. However, induction of the majority of these genes was attenuated in AP4-deficient cells, suggesting potential opposing regulation via the p53-21 axis also for these genes ( Figure 8A). Representative examples of DREAM targets of clusters 1, 2, and 3 are shown in Figure 8B. The repression of p21 by AP4 is predicted to attenuate DREAM complex activity and alleviate the repression of its target genes. This suggested that the activation of AP4 by c-MYC may contribute to the abrogation of p53-mediated gene repression. Interestingly, an up-regulation of E2F-and DREAM-target gene signatures was observed in AP4/p53 wild-type cells after treatment with DOX and was reversed in AP4-deficient/p53 wild-type cells ( Figure 6). Moreover, the differential regulation of E2F/DREAM target genes between AP4 wild-type and KO cells was not observed when p53 was inactivated, suggesting a direct involvement of p53. In addition, the activation of mRNAs belonging to functional categories largely comprising E2F/DREAM targets, such as gene sets representing processes involved in cell cycle progression (e.g., "G 2 /M checkpoint", "Mitotic spindle"), were also abrogated in AP4-deficient cells in a p53-dependent manner. We thus hypothesized that loss of AP4 may lead to enhanced repression of E2F target genes via hyper-activation of the DREAM complex.
Of note, a direct comparison of basal expression levels (i.e., MCF-7 cells not treated with DOX) indicated that the expression of E2F/DREAM targets, as well as the functional categories represented by these, was increased in p53-deficient cells when compared with AP4/p53 wild-type cells, suggesting an inhibitory effect of wild-type p53 on their expression ( Figure 6). The elevated basal expression of genes involved in cell cycle progression in p53-deficient cells, irrespective of their AP4 status, may explain the increased proliferation of these cells, as well as their lack of responsiveness to ectopic c-MYC. The basal expression of E2F/DREAM targets in AP4-deficient cells was increased compared to AP4/p53 wild-type cells ( Figure 6).
In order to identify genes with similar genotype-dependent differences in regulations after activation of c-MYC, we employed a two-factor (genotype and treatment) interaction analysis design using DESeq2, followed by KMeans clustering (n = 20) ( Figure 7A). In total, we identified 2309 genes that displayed genotype-dependent differences in c-MYC-induced regulation ( Figure S6, Table S9). Next, we determined which functional categories were significantly over-represented in at least one of the identified transcriptional clusters. Thereby, we identified a strong enrichment of E2F/DREAM targets, as well as pathways involved in cell cycle progression in the transcriptional clusters 1 and 2 ( Figure 7B). Interestingly, the DREAM target genes in these clusters were characterized by elevated basal expression in AP4and/or p53-deficient cells and p53-dependent down-regulation in AP4-deficient cells ( Figure 8A), thus corroborating the findings of GSEA ( Figure 6). A total of 202 E2F/DREAM targets were associated with clusters 1 (129/156) and 2 (73/155) (listed in Table S10). Of note, we also identified a subset of DREAM targets over-represented in cluster 3 (34/99), which does not display p53-dependent down-regulation in AP4-deficient cells but is induced upon activation of c-MYC. However, induction of the majority of these genes was attenuated in AP4-deficient cells, suggesting potential opposing regulation via the p53-21 axis also for these genes ( Figure 8A). Representative examples of DREAM targets of clusters 1, 2, and 3 are shown in Figure 8B.
We verified the regulation of several exemplary, previously validated DREAM targets (BUB1, CIT1, and BRCA1 [37,38]) by DREAM using publicly available ChIP-Seq data, which confirmed binding of the DREAM complex components E2F4, LIN9, and RBL2 in the promoter regions of the respective genes ( Figure 8C). Notably, we also identified a subset of direct c-MYC targets displaying AP4-dependent differences in c-MYC-induced activation (cluster 10). These genes were characterized by stronger induction by c-MYC in AP4-deficient cells compared with AP4 wild-type cells irrespective of p53 status. This indicated that AP4 is not required for their activation by c-MYC but rather attenuates their induction by c-MYC. Whether their increased expression contributes to the decreased c-MYC-induced proliferation in AP4-deficient cells remains to be determined. Collectively, these results showed that the inactivation of AP4 results in the down-regulation of numerous E2F/DREAM targets after activation of c-MYC in a p53-dependent manner.

Repression of DREAM Targets after c-MYC Activation Is Mediated by p21 and LIN37
Next, we verified the c-MYC-induced downregulation of the DREAM targets BUB1, CIT1, and BRCA1 [37,38] in AP4-deficient/p53 wild-type MCF-7 cells by qPCR ( Figure 9A-C). Indeed, while expression of these genes remained rather unchanged in AP4 wild-type cells, they were significantly repressed in AP4-deficient cells, thus confirming our NGS data. Therefore, we analyzed whether siRNA-mediated down-regulation of either p21 or the DREAM component LIN37 [39] (Figure S7) may revert the effect of loss of AP4 on DREAM target gene repression. Of note, while p21 was induced after activation of c-MYC, LIN37 was repressed after activation of c-MYC in MCF-7 cells ( Figure 9D,E), suggesting that their siRNA-mediated down-regulation may have divergent effects. Interestingly, the three analyzed genes (BUB1, CIT, and BRCA1) where induced by c-MYC activation after RNAi-mediated inactivation of either p21 or LIN37 in AP4 wild-type cells ( Figure 9F-H). However, after activation of c-MYC in AP4-deficient cells repression of these genes was observed which was either converted into an induction (BUB1, CIT) or abrogated (BRCA1) upon RNAi-mediated inactivation of p21 or LIN37 ( Figure 9F-H). While the effect of RNAimediated inactivation of LIN37 was stronger than that of p21 in AP4 wild-type cells, it was weaker in AP4-deficient cells, which may in part be explained by the different regulation of p21 and LIN37 after activation of c-MYC ( Figure 9D,E): the effect of p21 knockdown may be more pronounced in AP4-deficient cells due to elevated p21 levels in these cells compared to AP4 wild-type cells, whereas the levels of LIN37 are presumably already rather low after activation of c-MYC. In summary, these results validated that c-MYCinduced regulation of DREAM target genes is modulated by the opposing effects of AP4 and p53 on the p21-DREAM axis. These context-   We verified the regulation of several exemplary, previously validated DREAM targets (BUB1, CIT1, and BRCA1 [37,38]) by DREAM using publicly available ChIP-Seq data, which confirmed binding of the DREAM complex components E2F4, LIN9, and RBL2 in   Figure S1.

Association of c-MYC and AP4 Expression with p21, DREAM Targets and Patient Survival Is Dependent on p53 Status
Next, we analyzed whether the regulations identified above are conserved in primary breast carcinomas. For this, we analyzed RNA expression data from primary breast carcinomas and their associated clinico-pathological characteristics deposited in the TCGA database [35]. Expression of c-MYC and AP4 showed a positive correlation ( Figure 10A), as shown previously by us for CRC [24,40]. Interestingly, expression of c-MYC displayed a positively correlation with p21/CDKN1A expression. In line with our observations, this was only evident in tumors with wild-type p53 but not those with mutant p53.

Association of c-MYC and AP4 Expression with p21, DREAM Targets and Patient Survival Is Dependent on p53 Status
Next, we analyzed whether the regulations identified above are conserved in primary breast carcinomas. For this, we analyzed RNA expression data from primary breast carcinomas and their associated clinico-pathological characteristics deposited in the TCGA database [35]. Expression of c-MYC and AP4 showed a positive correlation ( Figure  10A), as shown previously by us for CRC [24,40]. Interestingly, expression of c-MYC displayed a positively correlation with p21/CDKN1A expression. In line with our observations, this was only evident in tumors with wild-type p53 but not those with mutant p53. Moreover, the expression of DREAM targets was associated with c-MYC to a lesser extent in p53 wild-type tumors when compared with p53 mutant tumors. Conversely, a negative association of c-MYC with p53 target expression could not be observed in p53 wild-type but only in p53 mutant tumors. Taken together, these strongly imply that in p53 wild-type tumors, high c-MYC levels may induce p53 and p21, which counteracts the c-MYC-induced activation of E2F/DREAM target expression.
Moreover, a high expression of c-MYC or AP4 was associated with shortened relapsefree survival and an increased hazard ratio ( Figure 10B). Interestingly, this association was Moreover, the expression of DREAM targets was associated with c-MYC to a lesser extent in p53 wild-type tumors when compared with p53 mutant tumors. Conversely, a negative association of c-MYC with p53 target expression could not be observed in p53 wild-type but only in p53 mutant tumors. Taken together, these strongly imply that in p53 wild-type tumors, high c-MYC levels may induce p53 and p21, which counteracts the c-MYC-induced activation of E2F/DREAM target expression.
Moreover, a high expression of c-MYC or AP4 was associated with shortened relapsefree survival and an increased hazard ratio ( Figure 10B). Interestingly, this association was only significant in p53 mutant tumors but not in p53 wild-type tumors. Hence, the presence of wild-type p53 appears to suppress the detrimental effects of high c-MYC/AP4 levels in breast carcinomas, possibly via activation of p21 and consequently, DREAM-mediated repression of cell cycle regulatory genes.

Discussion
Deregulated c-MYC expression has been shown to activate the p53 tumor suppressor, either by activation of p14/ARF or the induction of DNA damage due to unscheduled DNA replication. We have previously shown that the c-MYC-induced transcription factor AP4 regulates several pro-tumorigenic processes, including cell proliferation, EMT, and stemness, and suppresses DNA damage and senescence [14,36,40,41]. In addition, AP4 represses several p53 direct target genes, such as MDM2 or p21/CDKN1A [12,15,16]. However, whether the c-MYC-mediated induction of AP4 might play a role in inhibiting the tumor-suppressive effects of p53 activation has not been comprehensively analyzed to date.
Here, we abrogated AP4 expression in the breast cancer cell line MCF-7 harboring an ectopic, inducible c-MYC allele previously generated by us [12] using a CRISPR/Cas9 approach. Ectopic expression of c-MYC activates p53 in these cells, which allowed us to employ this system to analyze the effects of AP4 loss on c-MYC-mediated activation of p53 and on processes downstream of p53.
The E3-ligase MDM2 is a negative regulator of p53 protein levels [42,43]. Since c-MYCinduced levels of p53 protein were highly similar in AP4 wild-type and AP4-deficient cells, we concluded that repression of MDM2 by AP4 is unlikely to account for the activation of p53 after induction of c-MYC. In mouse embryo fibroblasts (MEFs) and murine models of lymphomagenesis, the up-regulation of p53 after activation of c-MYC has been shown to be caused by induction of p19 ARF , as it inhibits Mdm2 and thus leads to stabilization of p53 [6,44]. The MCF-7 cell line used here harbors homozygous deletions of the INK4A/ARF locus [45,46]. Therefore, ARF expression cannot be detected in these cells and does not account for the up-regulation of p53 by c-MYC observed here. Hence, a more likely scenario in this context is that the induction of DNA damage due to DNA replication stress induced by c-MYC ultimately leads to activation of p53 [9,47].
Here, deletion of AP4 resulted in increased spontaneous DNA damage, senescence and reduced proliferation. We have previously shown that AP4 loss has similar effects in colorectal cell lines and mouse embryo fibroblasts [24,36,41]. Interestingly, additional deletion of p53 in AP4-deficient MCF-7 cells fully reverted their proliferative defects, and furthermore rendered these AP4/p53-deficient cells insensitive towards ectopic c-MYC expression, at least with regard to proliferation. Furthermore, even though c-MYC activation in AP4/p53-deficient breast cancer cells resulted in a dramatic increase in DNA damage, it did not result in senescence. These results are different from our previous findings obtained with p53-mutant CRC cell lines, where deletion of AP4 caused a significant decrease in cell proliferation due to the induction of senescence. Potentially, the difference may be due to the complete loss of p53 activity in MCF-7 cells versus the presence of a mutant p53 in the CRC cell lines studied before. Alternatively, cell-type specific differences in the role of AP4 may exist. How AP4/p53-deficient cells regain high proliferative capacity and suppress senescence, even in the presence of high levels of spontaneous and/or c-MYC-induced DNA damage, is currently not understood and remains to be elucidated. It is likely that deletion of p53 allows cell cycle progression and proliferation in the presence of DNA damage, whereas in wild-type p53 cells uncoordinated DNA replication in the absence of AP4 activates p53 and attenuates cell cycle progression. Interestingly, we found recently that AP4 enhances DNA repair by inducing MDC1 expression [24], which may contribute to the positive effect of AP4 on c-MYC-induced proliferation.
The antagonistic regulation of p21 by AP4 and p53 predicted that inactivation of AP4 in p53-proficient cells would result in enhanced repression of DREAM and E2F target genes via the p21-DREAM axis after p53 activation by c-MYC. We generated comprehensive profiles of c-MYC-induced changes in RNA expression to determine which molecular and cellular pathways were affected by loss of AP4 and/or p53. Thereby, we determined that an important role of AP4 after activation of c-MYC in p53-proficient cells lies in the maintenance of E2F/DREAM target gene expression. Gene set enrichment analysis (GSEA) showed that while the DREAM and E2F target gene signatures were up-regulated in AP4 wild-type cells upon activation of c-MYC, they were repressed in AP4-deficient cells. Furthermore, we identified three subsets of DREAM targets that displayed distinct regulatory patterns after activation of c-MYC that were significantly affected by loss of either AP4 and/or p53. For DREAM targets showing a slight induction or repression after activation of c-MYC, deletion of AP4 in p53 wild-type cells caused significantly stronger repression (clusters 1, 2). For DREAM targets showing an induction after activation of c-MYC, deletion of AP4 in p53 wild-type cells caused significantly weaker induction (cluster 3). This effect of loss of AP4 could be abrogated or reverted by additional deletion of p53. Collectively, these differences in regulation all contributed to the regulatory patterns observed in GSEA. For the selected DREAM targets BUB1, CIT, and BRCA1, the effect of AP4 inactivation could be reverted by concomitant siRNA-mediated depletion of p21 or the DREAM complex component LIN37, providing strong evidence that the enhanced activity of the p21-DREAM axis in AP4-deficient cells is causally involved in the repression of these genes by p53 after c-MYC activation. We had previously observed that siRNA-mediated depletion of p21 reduces the number of senescent cells in AP4-deficient MEFs [36], strongly suggesting that the AP4-mediated inhibition of the p53-p21-DREAM axis may also be critical for the suppression of senescence.
The regulation of E2F activity by c-MYC is well established [48]. For example, c-MYC regulates E2F activity via direct transcriptional activation of G1 cyclins, such as Cyclin D1 and cyclin-dependent kinases (e.g., CDK4 [49]), as well as by directly inducing expression of E2F1 [50,51]. This interplay between c-MYC and E2F transcriptional activities is crucial for the control of cell-cycle progression. The results presented here strongly argue for a role of AP4 downstream of c-MYC in the regulation of E2F and DREAM activities via its repression of p21, which presumably contributes to the cell-cycle progression-enhancing effects of AP4 [12,41], as well as to the abrogation of p53 activity by c-MYC [11]. Moreover, the p53-dependent nature of the regulatory relationship between the c-MYC/AP4 axis and p21/DREAM-mediated gene repression was at least in part reflected by RNA expression correlations, as well as clinical associations in primary breast carcinomas. Since AP4 expression has been shown to be elevated in various types of cancer besides breast cancer and is associated with poor prognosis [52], this function of AP4 may also be relevant in other tumor entities.

Conclusions
Here we show that activation of p53 by c-MYC is largely driven by replication stressinduced DNA damage and not mediated by p14/ARF in MCF-7 breast cancer cells. After c-MYC activation, AP4 was necessary to suppress DNA damage and senescence and thereby facilitates cell proliferation. In AP4-deficient cells, p53 mediates senescence and inhibits cell proliferation. Our results show that AP4 represents a pivotal factor required for the balancing of c-MYC, E2F, and p53 activities via repressing p21 and thereby attenuating the activity of the repressive DREAM complex. This function of AP4 appears to be important for a coordinated induction of cell cycle progression by c-MYC and presumably contributes to c-MYC-driven tumorigenesis.