Introduction

Intratumor heterogeneity (ITH) is driven by (epi)genomic remodeling and microenvironmental changes, and present therapeutic resistance hurdles that must be overcome to optimize cancer treatments.

Over the past decade, technological advances have enabled the exploration of ITH at single-cell resolution, revealing a multitude of functional genetic and non-genetic cell states within the same tumor1,2,3. Certain cell states, such as those harboring stem cell traits (hereafter named cancer stem cells, CSCs), have been repeatedly associated with tumor progression and therapeutic failure4,5,6. Malignant cells can adapt to various stresses, including cancer treatment, by transitioning between states, a phenomenon known as cell plasticity. This transition between cell states is thought to be the major source of drug-resistance adaptation7. In this context, epigenetic remodeling, such as histone modifications, plays a significant role in tumor evolution processes8,9,10,11. For example, breast cancer cells can reach a drug-tolerant state by reducing H3K27me3 histone marks9. Conversely, inhibition of H3K27me3 demethylation in combination with chemotherapy prevents the transition to this drug-tolerant state. These studies emphasize the fact that the epigenetic heterogeneity and plasticity act as reservoir of cell states and therefore as a determinant of cell fate upon treatment exposure12. In this context epigenetic modifiers represent potential therapeutic targets to overcome therapy resistance.

From a fundamental perspective, each cell state may reflect a distinct configuration of gene regulatory networks (GRNs), which emerge from the complex interplay among chromatin structure, transcription factors, and gene expression13. The differential response of cancer cell states to treatment may be explained by variations in their chromatin architecture itself and the resulting activation of specific GRNs. While most chemotherapies used in cancer treatment act as DNA-damaging agents, several studies have demonstrated the role of chromatin features—such as chromatin folding, nucleosome remodeling, and histone modifications—in influencing DNA repair responses. Conversely, chromatin aberrations that confer either epigenetic restriction or increased plasticity can drive adaptive cell fate transitions14. More recently, studies have highlighted the crucial role of the DNA repair machinery in modulating chromatin marks, organization, and mobility15,16. This suggests that DNA double-strand breaks (DSBs) induced by treatment, along with their repair, can reshape chromatin organization, ultimately altering intracellular signaling pathways. These changes influence the cell’s capacity for adaptation and contribute to the evolutionary dynamics of cancer.

In this perspective, we discuss recent evidence for the interplay between DNA repair, the epigenetic landscape, and cancer cell plasticity. We discuss how the epigenome can impact the DNA damage repair response and, conversely, how DNA repair induces chromatin mis-restoration with direct effects on ITH. Finally, we highlight the promising therapeutic implications that could result from elucidating this coordinated process.

Interplay between DNA damage repair and epigenetic landscapes

Accumulating evidences suggest that DNA damage and the subsequent activation of DNA repair machinery depends on the chromatin structure that specify cell identity. Here, we have summarized studies that relate the impact of cell identity on DNA damage mapping and the response to these damages.

Influence of cell identity on the spatial mapping of DNA damage

Cell identity is determined by specific (epi)genetic landscape, which governs the activity of a given gene regulatory network (GRN). As a result, the identity of a cell has a unique genomic DNA packaging that restricts the localization of DNA damage. Indeed, genome-wide mapping of DSB, using BLESS or DSB Capture methodologies, demonstrates a relationship between genomic instability and nucleosome density. DSB are enriched in regions bearing epigenetic marks of transcriptionally active genes (H3K4me2/3), enhancer loci (H3K27ac, H3K9ac, and H3K4me1), regions rich in structural proteins (such as CCCTC-binding factor (CTCF)), or with motifs from several transcription factors (e.g. FOS, JUN, P53), and RNA polymerase II17,18,19,20. Thus, DSB do not appear randomly, but their localization is impacted by numerous cellular processes, DNA structures and sequences, histone modifications, and ultimately cell identity (Fig. 1A). The mapping of genomic breaks or “breakome” would therefore be influenced by cell identity17,21,22,23,24. Indeed, such heterogeneity in DNA damage mapping has been observed in tumors with glioblastoma cancer stem cells (CSCs) that presents a high expression activity of genes located at common fragile sites (CFS) compared to the glioblastoma cells composing the tumor bulk. This transcriptional activity promotes a transcription-replication conflict due to the specific DNA structure at CFS and leads to DSB formation25. Another source of genomic breaks that may be directly linked to cell identity is the accumulation of single-strand breaks (SSBs) hotspots that tend to occur in the immediate vicinity around the TSSs of genes that are actively transcribed26. Thus, SSBs hotspots location will be dependent on the activity of a given GRN and inextricably linked to cell identity. If the mechanistic reason behind this phenomenon is still unknown, we can suppose that the repair efficiency of breaks around TSSs is less efficient due to the special chromatin environment around TSSs compared to elsewhere in the genome. Another explanation may reside in Transcription-Replication conflicts that occur when the two critical cellular machineries responsible for gene expression and genome duplication collide with each other on the same genomic location. A recent study reports that DNA loci of hyper-transcribed genes accumulate DNA damage due to the TOP1 cleavage complex trapped in the R-loop that interferes with the resolution of these supercoiling events, ultimately leading to DSBs27.

Fig. 1: From breaks to restoration: the of chromatin architecture in the DNA Damage Response.
figure 1

A, B During the DNA repair process the chromatin structure is playing an essential by influencing the location of DSB accumulation (Breakome) and the recruitment of the DDR machinery to the DSB locus (Repair). A DSBs preferentially accumulate in specific chromatin contexts, including common fragile sites (CFS), transcriptionally active regions, DNA loops enriched in structural proteins such as CCCTC-binding factor (CTCF), and active enhancers (B). The epigenetic landscape critically influences the recruitment of DNA repair complexes. Changes in nucleosome composition enhance DSB mobility toward the nuclear pores, which are areas of increased repair activity. Nucleosome remodelers, such as the SWI/SNF complex, facilitate the recruitment of NHEJ factors by remodeling chromatin flanking DSBs. Similarly, the choice between homologous recombination (HR) and NHEJ is influenced by the specific pattern of histone post-translational modifications (PTMs) around the DSB sites. C Restoration and maintenance of chromatin structure following DNA repair are essential for preserving cellular function. Among the emerging mechanisms, the deposition of new histones H3 and H4 by DNAJC9 and MCM2, as well as H3.3 incorporation by HIRA in collaboration with CAF-1, appear to play a central role in chromatin reassembly post-repair. Created in BioRender. Mitoyan, L. (2025) https://BioRender.com/5je9jd0.

Full size image

Until now, the concept of breakome is mainly explained by transcription-induced DNA breaks but we cannot exclude that some DNA damage hotspots are not related to transcription initiation but still represent breaks consistently occurring in different cell states. It may be caused by exogenous factors that has a strong preference for specific chromatin structures. In line with this hypothesis, the use of clickable cisplatin derivatives revealed a unique genomic distribution of induced DNA-Pt lesions according to chromatin structure28. Indeed, the modification of chromatin folding by histone deacetylase inhibitor increases the number of induced DNA-Pt lesions. Thus, chromatin relaxation due to histone hyperacetylation reveals new genomic targets for cisplatin. This observation sustains an influence of chromatin structure on DSB mapping following exposition to exogenous DNA damaging agents.

Cell identity guide the DNA repair pathway choice

In addition to its influence on DNA damage mapping, cell identity also has a direct impact on the DNA repair response. The type of DNA repair pathway activated in response to DNA lesions has always been considered to be the result of the type of damage and the phase of the cell cycle in which the injured cell is located. However, over the past few years, several studies have highlighted the influence of cell state on the DNA damage response and repair capacity.

First, several regulators that maintain cell identity present a dual role with the capacity to activate specific DDR pathways. The anticlastogenic function of these cell-fate regulators directly linked cell identity to a singular DNA damage response. Second, the chromatin structure itself appears to have an impact on DNA repair pathway activation. These both elements will strongly influence the ability of each cancer cell state to maintain genome integrity, respond to genotoxic agents, and will impact tumor adaptability to treatment.

Dual role of cell identity safe guarders

Among the various regulators of cell identity and the DNA damage response, the Polycomb group complexes (PRC1 and PRC2) play essential roles. Their synergistic activity leads to the formation of transcriptionally repressive Polycomb domains, characterized by compacted chromatin enriched in the histone modifications H2AK119ub1 (catalyzed by PRC1) and H3K27me3 (catalyzed by PRC2). Initial models proposed a hierarchical recruitment mechanism, where PRC2 is first recruited to target loci to deposit H3K27me3. This mark is then recognized by canonical PRC1 complexes via chromodomain-containing subunits, facilitating H2AK119ub1 deposition and further chromatin compaction29,30,31. Importantly, components of both PRC1 and PRC2 have been implicated in stem cell identity regulation as well as in the detection and response to DNA lesions, suggesting their dual role in maintaining genome integrity and cellular plasticity. Among these components, BMI1 (PRC1) is the most prominent. BMI1 is associated with self-renewal capacity of various adult stem cells32,33. The preponderant role of BMI1 in maintaining stemness in malignant cells has been demonstrated in different cancers such as breast, colon, head and neck, or lung34,35,36,37. The inhibition of BMI1 is sufficient to decrease the proportion of CSC and to limit tumor progression in colon or prostate cancers35,38. Besides its role as an epigenetic regulator of cell identity, BMI1 appears to contribute to the DNA damage response by deposing H2AK119ub mark at the DNA lesions. It allows the recruitment of CtIP and DNA end resection to promote DNA repair via HR30 (Fig. 2, line1). This central regulatory node connecting cell identity and DDR activation may explain the higher capacity of CSC compared to the tumor bulk to resist to genotoxic agents as demonstrated in glioblastoma or breast tumors39,40. Similar to PRC1, components of PRC2 are also involved into the maintenance of cell identity and the activation of DNA damage response41,42,43 (Fig. 2, line2). Enhancer of zeste homolog 2 (EZH2), the PRC2 catalytic component, is able to increase H3K27me3 mark during chemotherapeutic treatment to regulate the expression of the DNA/RNA helicase SLFN11, inhibit transcriptional activity at DNA damage site and promote DNA repair44,45. Because EZH2 is frequently overexpressed in CSCs46, we can assume that EZH2 capacity to control gene expression during treatment will contribute to CSC resistance. Actually, it was demonstrated that the MELK-FOXM1-EZH2 signaling axis is essential for GSC radioresistance47.

Fig. 2: Dual role of cell identity safe-guarders.
figure 2

(line 1) To control cell identity, BMI1 (a polycomb repressor complex 1 (PRC1) protein) increase monoubiquitinylation of H2AK119 to inhibit the transcription of differentiation-associated genes. To activate DNA repair, BMI1 ubiquitinylate histones next to DSB. It leads to local transcription inhibition and recruitment of (C-terminal binding protein) interacting protein (CtIP) to promote DNA end resection and homologous recombination (HR). (line2) To control cell identity, EZH2 (a PRC2 protein) increase H3K27me3 marks to inhibit the transcription of differentiation-associated genes. To activate DNA repair EZH2 promote HR though the downregulation of Schlafen11 gene expression (SLFN11, an DNA/RNA helicase), and increase H3K27me3 marks on histones neighboring the DSB to induce local transcription silencing. Moreover, EZH2 is also known to inhibit REV7, hence favoring the HR repair pathway choice. (line 3) To control cell identity, BRD4 (Bromodomain-containing protein 4) regulates the transcription of genes-related to stemness by promoting enhancer-promoter interaction. To activate DNA repair, BRD4 interact with BRG1 to increase histone eviction and bind to histones next to DSB favoring CtIP recruitment and HR repair activity. In addition, BARD4 binds super-enhancers with MED1 and TEAD, to promote the recruitment of Rad51 and DNA repair on high transcript loci (line 4) To control cell identity, ZEB1 (Zinc finger E-box binding homeobox 1) play as a major regulator of the epithelial-to-mesenchymal transition (EMT) program. To activate DNA repair, ZEB1 promote HR and inhibit Alt-EJ by regulating the gene expression of ATM and polθ. Moreover, ATM phosphorylation enhance ZEB1 interaction with USP7 and CHK1 to promote HR pathway activity. Created in BioRender. Mitoyan, L. (2025) https://BioRender.com/5je9jd0.

Full size image

More recently, it was demonstrated that other epigenetic regulators such as the BET protein BRD4 known to bind to active enhancers and control cell identity gene induction48 may also play a role in regulating HR during DSB repair (Fig. 2, line3). Actually, the interaction between BRD4, BRG1, and CtIP appears to be required to achieve homology-directed repair of DSBs49. Moreover, the regulatory machinery at super-enhancers involving BRD4, MED1, and TEAD appears to recruit RAD51 to repair DSBs generated by the high transcriptional activity of these loci20.

This dual role of PRC1/2 complex or BRD4 in conferring efficient DNA damage response to specific cancer cell states is not restricted to epigenetic effectors. The Epithelial-to-Mesenchymal transition (EMT) transcription factor ZEB1 is a critical regulator of cancer cell plasticity50 but also strongly contributes to DNA damage response and repair (Fig. 2, line4). ZEB1 is required for DNA repair and the clearance of DNA breaks by controlling the expression of ATM, and its phosphorylation enhance ZEB1 interaction with USP7 and CHK1 to promote HR pathway activity51. ZEB1 can also inhibit polθ expression leading to a lower error-prone Alt-EJ pathway activity and consequently increasing genome stability of EMT-like cells52. As a consequence, ZEB1 inhibition is sufficient to impair DNA damage repair in CSCs and sensitize tumors to radiotherapy51.

Cell state-dependent chromatin structure and DNA repair

Beyond the dual role of certain cell identity safe-guarders to lineage-restrict DNA damage response and repair, they also guarantee a unique chromatin folding. This cell state-dependent chromatin structure will also constrain DNA repair. Indeed, it has been demonstrated that repair proteins spread according to chromatin topological features (Fig. 1B), using detailed analysis of DSB repair factor localization in single cells19. Several studies reported a preponderant role of the chromatin structure in governing DNA repair pathway choice, principally between homologous recombination (HR) and non-homologous end joining (NHEJ) pathways. We can identify HR restrictive domain as lamina-associated domain (LAD) that preferentially mobilized the error-prone repair pathway (NHEJ, Alt-EJ) to repair DSB53,54. Thus, the re-localization of DSB in HR permissive domain (euchromatin) is essential for genome stability and cell state maintenance55. Diverse chromatin remodeling factor are involved in increasing chromatin accessibility as BRG1 or INO8056. These chromatin remodelers promote the break relocation at the nuclear periphery by the incorporation of H2A.Z that increases the interaction of DSB with nuclear pore57. The SWI/SNF complex (with BAF sub-unit), is also required for efficient DNA repair pathways activation including NHEJ, by re-organizing the chromatin flanking the DNA lesion to promote the DNA repair58,59,60,61. If the chromatin mobility during DNA repair is essential, histone modifications are also preponderant to reshape chromatin landscape at DSBs. Using combined repair proteins (RAD51, XRCC4, 53BP1) and histone marks ChIP-seq with well-annotated DSB map, the group of Gaelle Legube offered a comprehensive picture of the DNA repair pathway choice according to the chromatin landscape62. HR and NHEJ appears to conceivably require very different chromatin settings. Concordant to previous studies, HR-competent chromatin contained elevated levels of H3K36me3. This histone mark, associated with transcription elongation machinery, is deposed by the trimethyl transferase SETD2 (SET domain containing 2)63 and the NSD family members64. The H3K36me3 is bound by LEDGF (lens epithelium-derived growth factor) that promotes recruitment of RAD51 and Ct-IP to facilitate DNA damage repair by HR63,65. Mechanistically, K354 deacetylation by SIRT1 (a HDAC protein) promotes SAMHD1 (Sterile alpha motif and HD domain-containing protein 1) recruitment to DSB and binding to ssDNA at DSB, which in turn facilitates Ct-IP ssDNA binding allowing genome integrity through the promotion of HR66.HR-competent chromatin is also associated with H3K79me2, H4K20me2/3, H2BK120ub, H3K4me2 near the DSB and low level of H2AZ. Conversely, DSB repair by NHEJ exhibits high levels of H4K20me1 and H2AXK15ub62. These experimental approaches represent the first step in understanding how chromatin structure guides the choice of DNA repair pathway. Because the 3D chromatin architecture undergoes considerable remodeling during cell state transition67, we can suspect that DNA damage response is constantly adjusted during cell plasticity. For example, the loss of H2A.Z near the TSS or the increase of H3K36me3 (implicated in HR pathway) modulate EMT by promoting expression of mesenchymal genes involved in first step of development68. The regulation of chromatin plasticity is essential to maintain cell identity in physiological and pathological conditions14. As a direct consequence, in tumors comprising various cell states (i.e. various epigenome), we can suspect a heterogeneity in terms of DNA damage response that may explain dissociated tolerance to DNA-damaging agents. Indeed, cancer cells undergoing EMT acquire stem cell traits69, accompanied by a massive chromatin reprogramming. Although DNA methylation remains unchanged during EMT, cells undergo global chromatin remodeling including an increase in the transcriptional mark H3K36me3 known to be enriched in HR-competent chromatin70. Of note, our current understanding of H3K36me1/2/3 writers and erasers remains limited compared to other well-characterized marks (e.g., H3K9me), and further studies will be required to clarify their contribution to cell plasticity. During treatment, chemoresistant breast cancer cell activates persister transcriptional program, due to the loss of bivalent chromatin (H3K27me3) in favor of active chromatin mark (H3K4me3)9. As the histone mark H3K4me3 is known to promote recruitment of the Xeroderma Pigmentosum Complex (XPC) and nucleotide excision repair (NER) machinery to repair DNA damage, it could explain the drug-tolerant capabilities of these persisters cells71. In addition, several studies report a strong HR activity in CSC compared to their differentiated counterpart40,72. As a result, glioblastomas stem cells represent the radio-resistant sub-population in GBM tumor bulk and breast CSCs tackle more efficiently DNA lesions and replicative stress generated by genotoxic treatment than non-bCSCs.

Overall, these studies highlight how elucidating the molecular bases of DNA repair in the context of chromatin and cell identity can help unravel the non-genetic mechanisms of therapeutic resistance in cancer.

Chromatin maintenance after DNA repair

While the chromatin landscape defines cell states with different susceptibility to DNA damage accumulation and with specific DNA repair pathways activation, the DNA repair process itself is not neutral on chromatin structure. Instead, it induces chromatin remodeling. Similarly, to replication, DNA repair processes provoke substantial epigenetic modification, due to chromatin disassembly needed to increase access to repair protein complex to the DNA lesion73,74. In fact, nucleosome is partially or totally disassembled around DSB in nucleolin-dependent manner, to allow the recruitment of repair protein such as RPA75. After DNA repair, the epigenetic landscape must be restored, following the access-repair-restore model73, to maintain the transcriptional activity, and the subsequent cellular identity (Fig. 1C). Despite several decades of research, how transcription restarts after DNA damage repair in a chromatin context is not fully elucidated. Most studies described transcription coupled repair (TCR) pathway in the context of NER of UV damage76. The histone chaperone chromatin assembly factor-1 (CAF-1) is recruited at the UV damage locus to facilitate new histone deposition. Then, the histone chaperone HIRA (histone regulator A) recruit new histone 3.3 at UV and DSB damage chromatin to act as a chromatin bookmarking process to facilitate recovery of transcription activity77,78,79,80. The histone variant H3.3 and its dedicated chaperone (CAF-1, HIRA, DAXX) play an essential role in the regulation of promoter and enhancer activity, whereas the variants H3.1 and H3.2 are usually present in transcriptionally repressed region during S phase81. Thus, the restoration of H3.3 is crucial for the maintenance of transcription activity immediately after DNA damage repair. Recently, new players in this process, the DNAJC9 histone chaperone and MCM2, has been demonstrated to provide new H3-H4 histones to CAF-1 and HIRA for its deposition into chromatin, and also stimulates old H3-H4 histone recovery82. Thus, in addition of parental histone recovery74 the integration of new histone on DNA damage locus is essential to preserve the epigenetic memory and cell identity after DNA repair.

Nevertheless, in a malignant context with substantial stalled replication fork associated with replicative stress, mis-restoration of epigenetic marks after DNA replication and repair might not be a rare event, but a common failure. The loss of initial chromatin architecture could lead to cell plasticity and participate to shape the non-genetic tumor heterogeneity.

Epigenetic damage scar challenge cell identity

While DNA repair pathways typically restore the DNA sequence to its original state before damage occurred, the accuracy of chromatin restoration remains unclear. In several context, including cancer or aging, restoration of the epigenetic landscape after DNA damage is not always allowed leading to the generation of “epigenetic damage scar”15,83. It was first demonstrated that during DSB repair, silencing proteins (e.g. SIRT1, EZH2, DNMT1, and DNMT3B) are recruited to the damage site with enrichment of their corresponding histone marks (hypoacetyl H4K16, H3K9me2, H3K9me3, and H3K27me3) and are maintain after repair84,85,86,87,88. Although promoter’s activity in the immediate vicinity of DSB is mainly preserved, in some case promoter regions harbored increased DNA methylation in the CpG island or on promoter close to the recombination site, leading to heritable silencing86,89. In addition to histone mark and DNA methylation, the chromatin condensation must be restored after DNA repair in part by the reestablishment of nucleosome following the reincorporation of new histone variant, such as H3.376,77. Importantly, this H3.3 distribution and relative abundance profoundly impact cellular identity and plasticity by epigenetically regulating gene expression90. These new incorporated histones present diverse posttranslational modifications (PTMs) that differ from the parental ones91. As an example, new H3.3 present accumulation of K9/K14ac2 and K9me2 that could participate to the establishment of an epigenetic damage scar91.

A direct consequence of these epigenetic scars linked to DNA lesions is an impact on the transcriptional activity of neighboring genes. It was first demonstrated in cancer cells using the reporter construct DRGFP (Direct-Repeat GFP) to monitor epigenetic modifications following DSB repair by HR in the GFP locus. Even though the damaged GFP locus was properly repaired, epigenetic scars, including DNA methylation and H3K9me2/3 modifications, appeared and generate cells with different but heritable GFP expression levels88. More recently, using a mouse model (named ICE, ERT2-HA-I-PpoI-IRES-GFP) that induce non-mutagenic DSB repair following the induction of the endonuclease I-PpoI, it was observed that ICE cells had relatively less chromatin-bound H3K27ac and H3K56ac (2% and 5%, respectively)92. This epigenetic erosion was sufficient to weaken insulation and disordered promoter-enhancer (P-E) communication accelerating the epigenetic clock and age-related changes to chromatin, gene expression, and cellular identity. These epigenetic damage scars can be assimilated to post-repair chromatin fatigue which can affect numerous genes expression within the topologically defines chromatin neighborhood that recovered from a single DNA breakage93. This notion of aging‒driven epigenetic changes could causally contribute to tumor initiation. Indeed, transient depletion of PRC1 cellular subunits is associated with irreversible activation of genes that promote cell growth, proliferation, migration, cell polarity, promoting neoplastic transformation94. This work introduces the concept of epigenetically initiated cancers (EICs) with epigenetic dysregulation that can lead to inheritance of altered cell fates sufficient to initiate tumor. In this context several reports suggest that loss of epigenetic regulation mitigates tissue homeostasis causing a susceptible state, in which cells are more prone to be transformed95,96.

Mechanistically this post-repair chromatin fatigue may be explain by the three-dimensional chromatin structure that is established by hierarchical folding at multiple scales starting from small functional loops, followed by larger topologically associated domains (TADs)97. The chromatin loop that contains DSBs presents a local DNA replication attenuation98. The persistence of DSB in this context could change the replication timing, and at the end, the transcriptome. DSBs directed to specific locations within an entire TAD have a lasting impact on transcription even if the lesion has been generated (and subsequently properly repaired) at megabase distances from the gene itself93. In addition, a recent study demonstrated that, the formation of DSB in heterochromatin rapidly moved outside the polycomb body (compact nuclear condensation) associated with a reduction of H3K27me399. This could persist and influence the transcriptional program. Importantly, these 3D epigenetic damage scars are inherited to the next generations of daughter cells and inevitably lead to a derail cell identity.

The inheritance of these epigenetic damage scars is surprising knowing the complex mechanisms that drive post-mitotic chromatin reconfiguration to maintain chromatin integrity and eliminate chromatin alterations to prevent the spread to the progeny100. One mechanistic hypothesis could reside in the process named mitotic bookmarking which assure the transmission of cell identity. It consists in the persistence of transcription factor DNA binding during mitosis allowing the rapid transcriptional activation upon mitotic exit101,102,103. Mitotic bookmarking is dependent on SWI/SNF complex that is required for appropriate reactivation of bound genes after mitosis104. Interestingly, it was demonstrated that SWI/SNF complex is enriched in DSB-flanking chromatin60. Thus, we can hypothesize that accumulated SWI/SNF complex on the epigenetic damage scar loci will allow the inheritance of transcriptional damage memory to the progeny and could challenge cell fate. How epigenetic scar could promote tumor recurrences by favoring the emergence of treatment-resistant cell states is a fascinating area that deserved to be explored.

Therapeutic opportunities

The intricate relationship between DNA damage repair and cell identity offers novel therapeutic opportunities to tackle adaptive mechanisms fueling the dynamism integral to tumor heterogeneity. Several studies indicate that CSCs have developed a robust replicative stress response (RSR) to reduce and tolerate replicative stress observed in neoplasia105. CSCs benefit from a super-active HR system including RAD51 upregulation. This cellular state of super-active homologous recombination appears to resolve replication stress by promoting efficient stressed replication fork stabilization, reversal and restart. This stress tolerance, which benefits from the DDR, is in fact a sign of targeted vulnerability in the CSCs. The use of different RSR inhibitors appears to be highly effective to eradicate the CSC-state and limit tumor progression. The combined inhibition of ATR and PARP provided GSC-specific cytotoxicity and complete abrogation of GSC radiation resistance25. In colorectal cancers, the combined treatment of MRE11 and RAD51 inhibitors (respectively mirin and B02) eradicate CSC by inducing mitotic catastrophe106. Similarly, in breast cancers, RAD51 inhibition increase replicative stress in CSC and sensitize these tumor-initiating cells to cisplatin, thus reducing tumors’ ability to relapse40. Recently, it was demonstrated that nifuroxazide treatment, a prodrug that is specifically bioactivated in breast CSC, induces a chemical HRDness in breast CSCs that (re)sensitizes breast cancers with innate or acquired resistance to PARP inhibitor (PARPi)107. Preclinical and clinical development of antitumor agents targeting the RSR machinery is extending with the new generation of ATMi, ATRi, WEE1i, CHK1i, DNA-PKi, RAD51i, or POLθi108 (Table 1).

Table 1 Registered clinical trials of drugs targeting replicative stress response machinery in cancers
Full size table

In addition to super-active HR system, other targetable DNA repair mechanisms may be selectively active in cancer stem cells (CSCs), such as templated-sequence insertions (TSIs)—a form of DNA double-strand break (DSB) repair that depends on hTERT activity109,110. Notably, hTERT expression is tightly restricted to stem and progenitor cells. Inhibition of hTERT using imetelstat has been shown to sensitize leukemic stem cells to genotoxic agents by suppressing this telomerase-mediated DNA repair pathway109,111,112. Under these premises, it becomes a priority to enhance our understanding of the mechanisms driving cell state specific DNA repair. This knowledge is crucial for identifying targets for intervention to limit tumor adaptation and evolution. Another way of exploiting the association between cell identity and the choice of DNA repair mechanisms is to corrupt the GRN of the cancer cell-state in order to reprogram cells into states that respond to treatment113. This last decade, the genetic concept of synthetic lethality has been exploited with success to target tumors HR-deficient with PARP inhibitors114,115. In tumors without any germline (or even somatic) mutation in HR-related genes, we can propose reprogramming cells in a state of low HR-activity by promoting a loss of epigenetic heterogeneity. A few years ago, following an epidrug screen, it was demonstrated that bromodomain (BRD) and extraterminal domain inhibitors (BETis) were able to sensitize HR-proficient breast cancer cells to PARPi. BETi-treated cells presented repression of HR-related genes expression leading to the induction of chemical HRDness116. One possible explanation of this synthetic lethality could lie in a phenotypic change induced by BETi, leading to a homogenized tumor in a cell state with limited HR activity. Indeed, BETi treatment is able to reduce the CSC pool by inducing a cell state transition to a non-CSC state, known to be less HR-active117. Similarly, the use of HDAC inhibitor has been repeatedly observed to be synthetic lethal with PARPi in tumors initially HR-competent118,119,120. In this case again, we observe a reduction of HR-related genes such as RAD51 or FANCD2 in HDACi-treated cells and HDACi have been identified as one of the first differentiation therapies121 capable to induce CSC differentiation122. EZH2 inhibitors (EZH2i) have been also consistently reported as a potent therapeutic strategy for targeting CSCs in both hematologic and solid tumors123. Although the precise mechanism of action remains unclear, EZH2’s role in regulating HR activity may partly explain this effect, as several studies have shown that EZH2i sensitize cancer cells to PARP inhibitors (PARPi)124,125. A similar outcome may also be achieved by inhibiting other histone methyltransferases, such as G9a/GLP, which appear to regulate cancer stemness and potentially synergize with PARPi treatment126,127 or by inhibiting DNA methyltransferase using cytosine analogs 5-azacitidine (5-AZA) that impairs leukemia or breast cancer stem cells function128,129 and induces an HRDness phenotype that sensitizes cancer cells to PARPi130.

In this context, we can propose that differentiation therapy is a new opportunity to induce a chemical HRDness, due to the shift from a HR-competent state to a state presenting poor HR functionality.

Beyond the various hurdles that must be overcome to introduce DNA damage response inhibitors (DDRi) into clinical practice—such as the identification of predictive biomarkers of response and the frequent emergence of acquired resistance due to DDR restoration—one promising strategy to enhance the efficacy of antitumor DDR therapies is to leverage the interplay between epigenetics and DDR, tailoring treatment to the specific cell state.