Prolonged loss of the core PRC1 subunit PH (ph-KD) triggers genome instability
Recent studies showed that knocking down the PRC1 subunit PH for a short time (24 h, Fig. 1, transient ph-KD) during L1 larval stage is sufficient to induce EIC formation in third instar larvae (L3), and these EICs do not exhibit DNA repair defects or genomic instability (Parreno 2024). These studies used an efficient thermosensitive ph-RNAi fly system to acutely deplete PH with a 24 h incubation at 29°C, and normal levels were restored within 48 h after switching to 18°C(Parreno 2024). The same system was used to address the effect of prolonged PRC1 inactivation (constant ph-KD), thus enabling direct comparisons with transient ph-KD. Additionally, white-KD or larvae maintained at 18°C (no ph-KD) were used as controls (Fig. 1). Constant ph-KD was obtained by incubating the larvae at 29°C during the whole larval development for 5 days. Similar to transient ph-KD(Parreno 2024), constant ph-KD also results in tumor formation in 100% of eye-antennal imaginal discs (EDs) of L3 larvae (Parreno 2024).
Prolonged ph-KD results in H2AK118ub and H3K27me3 loss at Polycomb target sites
Given that both transient and constant ph-KD results in tumors characterized by loss of polarity and differentiation, we asked whether these tumors differ at the epigenetic level. We plotted the genome-wide enrichments of PH, H2AK118ub, and H3K27me3 from control eye-antennal discs (EDs, no ph-KD), EICs after transient (24 h) ph-KD and tumors obtained after constant (5 d) ph-KD, using published ChIP-seq and CUT&RUN data sets (GSE222193 from (Parreno 2024), Supplementary table 1). As shown in Fig. 2a, Hilbert curve plotting shows that PH expression and recruitment to chromatin are restored after transient ph-KD but not after constant ph-KD. Consistently, the analysis of H2AK118ub and H3K27me3 enrichments around PRC1 target genes (PRC1-bound) relative to PRC1 non-target genes (PRC1-unbound) shows that these modifications are largely restored after transient ph-KD, but not after constant ph-KD (Fig. 2b and Supplementary Fig. 1a). The most significant difference between EICs derived from transient ph-KD and constant ph-KD tumors is associated with H2AK118ub, consistent with this histone modification being the primary modification established by PRC1. We conclude that tumors resulting from prolonged ph-KD are characterized by extensive loss of H2AK118ub and H3K27me3 at PcG target genes, while this is not the case for EICs resulting from transient ph-KD.
Prolonged ph-KD results in up regulation of DNA replication and DNA repair genes
Given the major epigenetic differences between EICs generated by transient and constant ph-KD, we examined the differential gene expression between these tumors compared to control tissues (no ph-KD) and temperature-matched white-KD, using the published datasets derived from RNA-seq analyses (Parreno 2024). As shown in Fig. 3a and Supplementary Fig. 1b, we found significant differences in gene expression profiles between transient and constant ph-KD tumors.
Remarkably, gene clusters corresponding to Gene Ontology (GO) terms related to DNA replication, DNA damage and DNA repair were mostly up-regulated in constant ph-KD conditions relative to transient ph-KD tumors (Fig. 3a). Consistently, a fold-change analysis of all the genes classified as “DNA replication” (n=111) or “DNA damage response” (n=242) shows a significantly higher level of transcription for both categories in constant ph-KD tumors relative to control, and also compared to all genes (Supplementary Fig. 1c). This indicates that DNA replication and DNA damage response genes are overall more transcriptionally active in tumors derived from sustained ph-KD.
Within this general trend, 21 genes required for “DNA replication" and 28 genes required for the “DNA damage response” were the most affected, displaying at least a 2-fold change in expression specifically in constant ph-KD tumors relative to controls, most of which (18 and 26 genes, respectively) were up-regulated (Fig. 3b, d).
Most of the DNA replication and DNA damage response genes up-regulated in tumors derived from constant ph-KD are not associated with PRC1 enrichments in normal tissues (no ph-KD), suggesting that they are not direct targets of PH and their up regulation is an indirect effect of PRC1 loss (e.g., Fig. 3e, CG10336, or TIPIN in mammals). The most notable exception is the replication, repair, and transcription factor Fkh (FOXA2 and FOXA1 in mammals) (Knott, Peace et al. 2012, Li, Coic et al. 2012, Dummer, Su et al. 2016, Jin, Liang and Lou 2020, Hoggard, Hollatz et al. 2021), which is enriched for PRC1 in normal tissues, suggesting that PRC1 down-regulation directly affects the transcription of this gene (Fig. 3e).
The replication genes affected in constant ph-KD tumors include key replication components: the MCM complex, origin firing factors, and several DNA polymerases (Supplementary Table 2). This might result from an overall induction of replication in the tissue. Thus, we investigated the proliferation state of the cells by EdU incorporation and labeling in these tumors. As shown in Fig. 3c, control EDs show a few replicating cells, mostly localized at the morphogenetic furrow (Avellino, Peng and Lin 2023, Parreno 2024). Conversely, tumors derived from constant ph-KD are characterized by massive EdU incorporation, indicating a complete switch to an uncontrolled over-proliferating state (Fig. 3c). Of note, DNA replication-associated genes are found over-expressed also in transient ph-KD tumors (Supplementary Fig. 1c), albeit to a lesser extent compared to constant ph-KD tumors. Similarly, transient ph-KD tumors are also highly enriched for replicating cells (Parreno 2024).
Together, these results underscore that constant ph-KD leads to tumors characterized by the upregulation of several DNA replication genes, which is likely a consequence of cell hyper-proliferation. This up-regulation is more pronounced than that observed in EICs, and might reflect an even higher proliferation rate. Upregulation of components required for replication initiation and progression can also contribute to the acquisition of the hyper-proliferative state (Yu, Wang et al. 2020). In addition, we observed dysregulation of several DNA damage response genes upon constant depletion of PH, most of which are likely the indirect consequence of PH loss. These genes are mostly expressed at normal levels in transient ph-KD tumors, representing a major difference between the effects of short-term and long-term PH depletions.
Prolonged ph-KD leads to defective DSB repair and increased genomic instability compared to EICs
DNA repair genes over-expressed in constant ph-KD tumors include several components previously linked to damage accumulation, cancer formation, and/or poor cancer prognosis (Table 1), like Mms4 (Dewalt, Kesler et al. 2014), RecQ4 (Maire, Yoshimoto et al. 2009, Su, Meador et al. 2010, Xu, Chang et al. 2021), PolH ( Tomicic, Aasland et al. 2014, Sonobe, Yang et al. 2024), Tipin/Timeless (Zhou, Zhang et al. 2020, Chen, Zhang et al. 2022), Claspin (Choi, Yang et al. 2014), MRNIP (Staples, Barone et al. 2016, Bennett, Wilkie et al. 2020, Wang, Zhao et al. 2022), FANCI (Smogorzewska, Matsuoka et al. 2007, Li, Yu et al. 2023), MMR proteins (Msh2, Mlh1, Msh6) (Shcherbakova and Kunkel 1999, Velasco, Albert et al. 2002, Li, Liu et al. 2008, Wagner, Webber et al. 2016, Wilczak, Rashed et al. 2017, Chakraborty, Dinh and Alani 2018, Donis, Gonzalez et al. 2021, Zhou, Xiao and Chen 2024), and Rif1 (Liu, Mei et al. 2018, Mei, Liu et al. 2018, Sad, Mohamed et al. 2021). Similarly, genes down regulated in constant ph-KD tumors include known components required for DNA repair and replication fork protection in the presence of replication damage, like the PCNA variant PCNA2 (Feng, Xia et al. 2023) (Table 1). Collectively, misregulation of these genes is expected to lower fork protection, increase DSB formation in response to stalled fork, and impair DSB repair.
We directly tested this by investigating DNA break formation through immunofluorescence (IF) analysis of γH2Av foci in tumors dissected from L3 larvae after constant ph-KD or in EDs from the temperature-matched wRNAi control. Constant ph-KD results in a higher number of γH2Av foci in the tissue, indicating a higher level of endogenous DNA damage (Fig. 4a, b). This likely derives from the higher number of replicating cells, which typically experience a higher baseline level of damage than non-replicating cells, along with defective fork protection and repair.
In addition, we investigated the DSB repair response by treating constant ph-KD tumors and their controls with 5Gy ionizing radiation (IR), and by quantifying the kinetics of γH2Av focus formation and resolution. Both tumor and ED control tissues showed a significant increase in the number of γH2Av foci 30 min after IR, indicating DSB induction and checkpoint activation. The higher level of repair foci in ph-KD tumors relative to the control reflects the higher baseline level of damage. Importantly, constant ph-KD tumors display a significantly higher number of γH2Av foci compared to control EDs 4 hours after irradiation, and this difference is much more pronounced than what observed in untreated (UNT) tissues or after 30 min from IR. This indicates that, unlike transient ph-KD tumors (Parreno 2024), constant ph-KD tumors are defective in DSB repair.
Given the higher amount of DNA damage and defective repair, we hypothesized that constant ph-KD tumors might accumulate unrepaired DSBs over time, resulting in chromosome rearrangements and genome instability. We tested this by karyotype analysis of tumors from constant ph-KD and EDs from wRNAi control in L3 larvae, using pericentromeric fluorescence in situ hybridization (FISH) probes specific for individual chromosomes (Fig. 4d). Remarkably, we observe a 4-fold increase in the frequencies of chromosome rearrangements in constant ph-KD tumors relative to the temperature-matched ED controls (Fig. 4d-f). Rearrangements include a large number of chromosome fusions, aneuploidies and abnormal number of satellites (Fig. 4d-f). Moreover, we observe a massive increase in a very rare form of rearrangements characterized by fusions across several chromosomes (“broad rearrangements”), which are so severe that they prevent clear discrimination between the chromosomes (Fig. 4d-f).
In conclusion, tumors induced by PH depletion over 5 days during larval stages are characterized by misregulation of genes required for fork protection and repair, DSB repair defects, and widespread genome instability, which was not observed in EICs derived from transient ph-KD.