PTBP1 enforces ATR-CHK1 signaling determining the potency of CDC7 inhibitors

Summary CDC7 kinase is crucial for DNA replication initiation and fork processing. CDC7 inhibition mildly activates the ATR pathway, which further limits origin firing; however, to date the relationship between CDC7 and ATR remains controversial. We show that CDC7 and ATR inhibitors are either synergistic or antagonistic depending on the degree of inhibition of each individual kinase. We find that Polypyrimidine Tract Binding Protein 1 (PTBP1) is important for ATR activity in response to CDC7 inhibition and genotoxic agents. Compromised PTBP1 expression makes cells defective in RPA recruitment, genomically unstable, and resistant to CDC7 inhibitors. PTBP1 deficiency affects the expression and splicing of many genes indicating a multifactorial impact on drug response. We find that an exon skipping event in RAD51AP1 contributes to checkpoint deficiency in PTBP1-deficient cells. These results identify PTBP1 as a key factor in replication stress response and define how ATR activity modulates the activity of CDC7 inhibitors.


INTRODUCTION
The replication of genomic DNA is highly regulated both at the level of initiation, which occurs in a coordinated manner from multiple origins of replication, as well as at each individual replication fork, which can encounter obstacles leading to replication fork stalling and collapse. [1][2][3][4] Defining the mechanisms that coordinate replication initiation and elongation is an important task to understand how genome stability is maintained during DNA replication and it can also have therapeutic implications for cancer therapy.
Two kinases, CDC7 and ATR, play essential roles in this process with both being involved in controlling initiation and elongation. [5][6][7][8][9][10][11] In initiation, CDC7 phosphorylates several subunits of the MCM helicase allowing for the formation of the CDC45-MCM-GINS (CMG) complex and origin activation. 6,11,12 CDC7-dependent phosphorylation of the MCM complex is counteracted by RIF1-PP1 phosphatase; thus, the efficiency of initiation is dictated by a balance between phosphatase and kinase activity. 13,14 ATR was also reported to phosphorylate the MCM complex and in yeast it can act as a priming event for subsequent CDC7 phosphorylation suggesting a role in promoting origin firing. 9, 15 However, multiple studies have since shown that ATR mainly acts as a negative regulator of origin firing and in human cells it does so by downregulating CDK1 activity, thus reinforcing the RIF1/PP1 axes, both during unperturbed S-phase and in response to genotoxic agents. 5, 16,17 During elongation, CDC7 is involved in DNA damage tolerance pathways, like translesion synthesis (TLS), in yeast as well as human cells. 18 More recently, we showed that human CDC7 kinase physically associates with ongoing and stalled replication forks, where it promotes fork processing and restart in a MRE11dependent manner. MRE11, together with EXO1 and other nucleases, generate extended ssDNA in the proximity of the fork resulting in RPA accumulation and activation of ATR kinase. 19,20 ATR is then responsible for the phosphorylation of many factors, which help in preserving fork integrity and arresting the cell cycle, thus limiting DNA damage and genome instability. 16,21 with TAK-931 have suggested that tumors experiencing high levels of replication stress are more sensitive to CDC7 inhibition. 27 This raises the expectation that the combination of CDC7is and checkpoint inhibitors, including ATRis, CHK1 inhibitors (CHK1is), and WEE1 inhibitors (WEE1is), would be an attainable and effective strategy by increasing underlying replication stress of cancer cells.
Nevertheless, the relationship between CDC7 and the ATR pathway is not well understood, with reports indicating that CDC7is and checkpoint inhibitors have additive or synergistic effects in cell killing, whereas others indicate antagonism. 10, 27 Furthermore, mechanistically, it is highly debated whether CDC7 should be considered as an effector or a direct target of the ATR pathway. [28][29][30][31] Such discrepancies particularly affect the rational clinical development of CDC7is and to a lesser extent of ATRis.
With the aim of identifying genes, which contribute to the antiproliferative activity of CDC7is, we conducted a genome-wide CRISPR/Cas9 KO screen with cells that had been treated with high doses of the CDC7i XL413. In that study, we identified ETAA1 as the relevant subunit activating ATR, restraining DNA synthesis and proliferation in response to CDC7 inhibition. However, we also reported that co-inhibition of CDC7 and ATR leads to cell death through mitotic catastrophe, 32 an observation that has since been recapitulated by others. 33 Such apparent contradiction likely indicates an ambiguous relationship between the two kinase activities and that the phenotypes observed may strongly depend on the levels of inhibition of the two kinases, achieved by either genetic or pharmacological means.
Here we reveal the role of Polypyrimidine tract binding protein 1 (PTBP1), which was identified in the aforementioned CRISPR screen, as a novel modulator of the ATR pathway. PTBP1's main function, as a member of the heterogeneous nuclear ribonucleoprotein (hnRNP) subfamily, is in RNA metabolism: controlling mRNA alternative splicing, polyadenylation, translocation, and stability. [34][35][36][37][38] It has also been shown to influence the initiation of translation through internal ribosome entry site-mediated translation. 39 It has four tandem RNA recognition motifs (RRM) and an N-terminal nuclear localization signal (NLS). 40 PTBP1 interacts with mRNA at polypyrimidine-rich regions, namely UCUU-rich sequences in exonic and flanking intronic areas 39 and it usually promotes the inclusion of the exon. 41,42 Because of its biochemical activity, PTBP1 has a pleiotropic role in cells and has been implicated in many biological processes, from the regulation of the pro-inflammatory senescence-associated secretory phenotype (SASP) and B-cell maturation to cancer cell proliferation and invasion. [43][44][45][46] Intriguingly, chemo-genetic screens have suggested that PTBP1 may affect cell viability to some genotoxic agents in RPE1 cells. 47 However, the exact involvement of PTBP1 in the response to these agents has not been studied yet.
We now show that PTBP1 loss of function causes partial resistance to CDC7 inhibition by reducing ATR activation in response to CDC7is as well as DNA damaging agents. We also show that low levels of ATR inhibition, similar to PTBP1 loss of function, allows more efficient DNA synthesis in the presence of CDC7is. Impaired ATR activation in PTBP1-deficient cells correlates with deficient RPA chromatin binding and focal recruitment suggesting altered frequency of fork stalling or processing. Finally, we identify RAD51AP1, which is defectively spliced and expressed in PTBP1-deficient cells, as an important component of the replication stress response.

RESULTS
We previously reported that in MCF10A cells CDC7 inhibition can cause cell growth arrest, replication stress and ATR activation. 32 To better understand how the antiproliferative activity of CDC7is may be affected by ATR, we treated MCF10A cells with a well characterized CDC7i (XL413) and an ATRi (AZD6738) at increasing concentrations either alone or in combination. Cell viability was assessed after three days using a resazurin assay. We discovered that at most doses ATRi and CDC7i have an additive (d score: À10 to 10) or synergistic (d score: >10) effect in preventing proliferation, possibly leading to cell death with a maximum d score of 26.73 ( Figure 1A). However, low doses of XL413 suppressed AZD6738-induced loss of viability ( Figures 1A and 1B), while low doses of AZD6738 allowed cells to grow better in presence of high doses of XL413 ( Figures 1A and 1C). More specifically, we noticed that 0.156 mM AZD6738, allows MCF10A cells to better proliferate in the presence of 20 mM XL413, whereas doses of R0.625 mM AZD6738 drastically decreased the number of living cells ( Figure 1C). Using a flow cytometry-based assay, we observed that while, on its own, 20 mM XL413 drastically reduced DNA synthesis, causing cells to accumulate in mid-to-late S-phase of the cell cycle, the addition of 0.156 mM AZD6738 allows DNA synthesis to proceed more efficiently and cells distribute more normally throughout the phases iScience Article of the cell cycle. In contrast, cells treated with high doses of both compounds dispersed symmetrically around G1 DNA content with many cells not incorporating further EdU into their DNA ( Figures 1D-1F), a phenotype that we previously correlated with mitotic catastrophe. 32 DNA fiber assay was then performed to provide a more in-depth analysis of the replication dynamics of cells treated with CDC7is and ATRis. MCF10A EditR cells were treated with 20 mM XL413 or DMSO and labeled with CldU for 30 min followed by the addition of either 0.156 mM or 5 mM AZD6738 and labeling with IdU in the continued presence or absence of XL413 ( Figure 1G). We found that the addition of AZD6738, at both low and high doses, significantly increased origin firing both in the presence or absence of CDC7i whereas we did not observe a statistically significant change in the number of terminating forks ( Figures 1G-1I and S1).
These results indicate that the efficiency of CDC7is in limiting DNA synthesis and restraining proliferation requires robust ATR activity, but a severely compromised ATR response can eliminate cells exposed to CDC7is.

PTBP1 deficiency suppresses the antiproliferative activity of CDC7 inhibitors
In a previously published CRISPR/Cas9 KO screen, we identified several genes that when edited may confer resistance to XL413, among which were RIF1, ETAA1, and PTBP1. We showed that ETAA1 depletion hinders the ATR response triggered by CDC7is allowing more origin firing and more efficient DNA synthesis to occur, similarly to RIF1 depletion and low levels of ATR inhibition by AZD6738 (ref. 32 and Figure 1).
To validate that loss of PTBP1 provided cells with a proliferative advantage when CDC7 is inhibited, a 16-day competitive proliferation assay was performed. MCF10A EditR cells, stably expressing Cas9, were transduced with a lentiviral vector that co-expressed GFP and one of two short guide RNAs (sgRNAs) targeting PTBP1, sgRNA4 or sgRNA6, or alternatively with an empty vector (EV) control ( Figure S2A). Starting with a mixed population of 5-10% GFP-positive cells, we assessed the percentage of GFP-positive cells following a treatment with 20 mM XL413 for 16 days. Relative to the EV control, sgRNA4 and sgRNA6 transfected cells showed an increase in the percentage of GFP-positive cells in the population ( Figure S2A). This indicates that editing the PTBP1 gene confers a proliferative advantage on CDC7 inhibition.
From the pools of transduced cells, we isolated two independent clones, one for each sgRNA, with mutations in the PTBP1 gene ( Figure S2B). Surprisingly, we found that neither cell line harbored a null mutation in the PTBP1 gene: clone 4.1 had a 3-nucleotide homozygous deletion in exon 5 resulting in the loss of Ile99 within the RRM1 leading to protein instability, as PTBP1 protein was barely detectable by western blotting Figure 1. Partial inhibition of ATR confers resistance to CDC7is while full inhibition results in sensitisation (A) MCF10A cells were treated with indicated doses of XL413 and AZD6738 for 72 h followed by resazurin assay to analyze cell viability. 2D synergy map was generated using normalized viability scores of three independent resazurin assays. Antagonistic (green, d score <À10), additive (d score À10 to 10) or synergistic (red, d score >10) dose regions are highlighted. (B) Data described in A were further analyzed. Viability was determined relative to samples treated with 5 mM AZD6728 (dotted line). Data represent three independent experiments performed in technical triplicates, mean G SD. Statistical analysis was performed using one-way ANOVA and Dunnett's multiple comparison test (****p < 0.0001).
(C) Additional analysis of data presented in A. Viability was determined relative to samples treated with 20 mM XL413 (dotted line). Data represent three independent experiments performed in technical triplicates, mean G SD. Statistical analysis was performed using one-way ANOVA and Dunnett's multiple comparison test (**p < 0.01; ****p < 0.0001  Figure S1. (H) Percentages of IdU only tracks (new origin firing) are plotted for the DNA fiber assay described in G. At least 150 replication tracks were analyzed for each condition of three independent experiments. Data is presented as mean G SD. Statistical analysis was performed using one-way ANOVA, with Tukey's multiple comparison test (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001).
(I) Percentages of CldU only tracks (terminated/collapsed forks) are plotted for the DNA fiber assay described in G. At least 150 replication tracks were analyzed for each condition of three independent experiments. Data is presented as mean G SD. iScience Article and immunofluorescence microscopy ( Figures S2B-S2E). In contrast, clone 6.6 carried a wild type allele and an allele with a 9-nucleotide loss leading to an in-frame deletion of three amino acids within the RRM2 in heterozygosis ( Figure S2B). This small in-frame deletion likely also leads to protein instability as the level of PTBP1 protein was decreased compared to the parental MCF10A EditR cells ( Figures S2C and S2D).
To further confirm, that PTBP1 is required to restrain proliferation in cells treated with CDC7is, MCF10A EditR cells and PTBP1 mutant clones 4.1 and 6.6 were treated with DMSO, 10 mM or 20 mM XL413 for 72 h. Cells were then stained with crystal violet to assess cell viability. The intensity of crystal violet stain was quantified for each treatment and normalized to its DMSO control. Treatment with XL413 restricted cell proliferation in the EditR cells and to a lesser extent in clones 4.1 and 6.6; notably the extent to which the antiproliferative effect of XL413 was suppressed correlated with the residual amount of PTBP1 protein in these cells ( Figures 2B and 2D).
To confirm that the proliferative advantage of the PTBP1 mutant clones persists even after prolonged treatment with CDC7is, we extended the treatment to eight consecutive days in a competitive proliferation assay with MCF10A EditR cells. Equal numbers of MCF10A EditR and GFP-expressing PTBP1 mutant clone 4.1 ( Figure S2E) were mixed and cultured in DMSO or increasing doses of XL413 and TAK-931 over eight days before assessing the percentage of GFP-positive cells. In absence of CDC7is we observed that MCF10A EditR cells were accumulating in the population, but importantly the treatment with either one of the CDC7is resulted in a significant increase in the percentage of GFP-expressing cells, indicating that PTBP1 mutant clone 4.1 is outcompeting MCF10A EditR cells only in the presence of CDC7is ( Figures 2E and 2F).
To understand if the resistance to CDC7is in PTBP1 mutant cells was related to DNA replication, MCF10A EditR and clones 4.1 and 6.6 cells were treated with either XL413 or TAK-931 for 24 h and DNA synthesis as well as cell cycle distribution was analyzed by flow cytometry. In parental cells CDC7i treatment led to the characteristic accumulation of cells in late S-phase with a lower rate of DNA synthesis, however PTBP1 mutant clone 4.1, and to a lesser extent clone 6.6, displayed near normal levels of DNA synthesis despite the presence of CDC7is ( Figures 3A and 3B). Similarly, PTBP1 depletion by siRNA increased the rate of DNA synthesis and reduced the accumulation of cells in late S-phase upon CDC7 inhibition ( Figures S3A-S3D).
As shown for the combination of ATRis and CDC7is, unscheduled origin firing can partially overcome the effects of CDC7is on the rate of DNA synthesis ( Figures 1D-1I All together, these data provide compelling evidence that PTBP1 contributes to restricting DNA synthesis and origin firing, thus delaying S-phase progression when CDC7 is inhibited. As the magnitude of this effect is dependent on the residual amount of functional PTBP1, all further experiments were performed with clone 4.1.

ATR signaling is partially impaired in PTBP1-deficient cells
Our analysis indicates that PTBP1 deficiency phenocopies the partial inhibition of ATR by AZD6738 suggesting that in PTBP1 mutant clone 4.1 the ATR response may be impaired. To test this hypothesis, we Thus, PTBP1 determines the amount of RPA-coated ssDNA that is generated at replication forks, both during normal and challenged DNA replication, which in turn limits the extent of ATR activation.
Consistent with poor ATR signaling and problems in DNA replication, many micronuclei were observed in PTBP1 mutant clone 4.1 cells, and their number decreased to almost normal levels on ectopic expression of PTBP1-SM ( Figures 5A-5C). More interestingly, PTBP1 mutant clone 4.1 cells displayed a significant increase in large 53BP1 foci in G1 cells compared to either parental or PTBP1-SM cells ( Figures 5D and  5E). Such large foci are also known as 53BP1 bodies and have been associated with repair of unreplicated DNA from the previous cell cycle. 48,49 Since micronuclei can also be generated by defective mitosis independently from DNA replication, we scored approximately 600 mitotic cells in MCF10A EditR, PTBP1 mutant clone 4.1 and PTBP1-SM cells. Chromatin bridges and lagging chromosomes, were detected in PTBP1-deficient clone 4.1 cells and their number were increased compared to parental and PTBP1-SM cells ( Figure S5), although the effect did not reach statistical significance. Altogether, these data indicate that PTBP1 function is required to maintain the normal control of DNA replication and its impairment can lead to genome instability.

PTBP1 deficiency severely alters the transcriptome
As the canonical function of PTBP1 is in mRNA processing, we hypothesized that the disrupted DNA replication stress response in PTBP1 mutant cells might be due to the differential gene expression and/or iScience Article alternative splicing in one or more of the genes regulating these processes. Thus, we performed an RNA-seq experiment to compare the transcriptomes of PTBP1 mutant cells and EditR cells. The analysis of differentially expressed genes identified 876 genes significantly upregulated and 414 significantly downregulated in PTBP1-deficient cells compared to the parental cells ( Figure 6A and Table S1). Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) analysis indicated that PTBP1 disfunction can affect multiple pathways and processes among which were complement and coagulation cascades, cell adhesion as well as Ras signaling and cytoskeletal regulation, but, to our surprise, the terms and pathways related to DNA replication, DNA repair or cell cycle regulation were not highlighted ( Figure S6). Such widespread alteration in gene expression was also matched by many statistically significant alterations in exon usage across numerous genes with skipped exons being the most frequent event occurring in the PTBP1deficient cells ( Figure 6B).
To determine how PTBP1 mutation could possibly confer resistance to CDC7is, we used the GO database 50,51 and generated a list of 1243 genes associated with the terms of DNA replication, DNA damage repair, S-phase checkpoint signaling, S-phase progression, and G1-S-phase transition (Table S2). These genes were then compared to the ones either differentially expressed or alternatively spliced in PTBP1 mutant clone 4.1 compared to EditR cells. We found that some of these genes were partially deregulated in PTBP1-deficient cells, but in most cases the differences in the level of expression were limited (Table S1).
Intriguingly, we observed an imbalance in the expression of subunits of several protein complexes, such as the MRN complex where MRE11 was downregulated by approximately 30%, whereas RAD50 expression was increased by 40%. In the fork protection complex CLASPIN was upregulated 2-fold, whereas TIMELESS was downregulated approximately 20%. Similarly, although ATR was downregulated by 25%, ETAA1 was instead upregulated (Table S3). Thus, even though small changes in the expression of single specific genes are unlikely to be solely responsible in determining a strong phenotype, the imbalance in the expression of proteins forming functional complexes may lead to a more accentuated disruption of their function.
When we examined the genes that were alternatively spliced, we observed that in most cases, although statistically significant, such alternate splicing events either occurred at a low frequency or in exons which are seldom utilized, and were thus unlikely to be responsible for the phenotypes described above. However, we found one exception, a prevalent exon skipping event affecting exon 8 (Chr12: 4556353-4556502), coding for one of the DNA binding domains in the RAD51AP1 protein ( Figure 6C). As RAD51AP1 is involved in both homologous recombination and the response to transcription-replication conflicts [52][53][54][55] and since the loss of RAD51AP1 has been associated with reduced S-phase checkpoint activation, 56 we reasoned that RAD51AP1 could be a key factor linking PTBP1 deficiency to altered drug response. To assess how impaired splicing may affect RAD51AP1 levels and function, MCF10A EditR and PTBP1 deficient clone 4.1 cells were either mock-treated or treated with CDC7i or MMC for 24 h before harvesting. Lysates were then fractionated into soluble and chromatin-enriched samples and analyzed by western blot. We found that most of RAD51AP1 was detected in the chromatin-enriched fraction as an immunoreactive band running at $33 kDa, however three additional bands with slower migration, possibly because of post-translational modifications, [57][58][59] were also seen in the chromatin and soluble fractions (Figure 6D). The overall levels of RAD51AP1 species were strongly reduced in PTBP1-deficient cells and their distribution was not affected by drug treatment, consistent with the hypothesis that exon skipping leads to the synthesis of a truncated and likely unstable protein ( Figure 6D).  Figure 7A). Using flow cytometry-based analysis, we determined the importance of RAD51AP1 in restraining DNA synthesis following CDC7i treatment. Similar to PTBP1-SM cells, RAD51AP1 cells displayed reduced EdU incorporation in late S-phase cells compared to MCF10A PTBP1 mutant clone 4.1, suggesting at least a partial re-sensitization of both cell lines to CDC7 inhibition ( Figures 7B-7D). Similarly, RAD51AP1 expression in PTBP1-deficient cells partially rescues ATR activation, as demonstrated by the increased phosphorylation of CHK1 following MMC treatment ( Figure 7E). iScience Article Thus, we can conclude that PTBP1 modulates the response to CDC7 inhibition and replication stress in part through splicing of RAD51AP1, enforcing DNA synthesis inhibition and stimulating ATR signaling.

DISCUSSION
In this work, we define the relevance of the ATR driven checkpoint response to determine the antiproliferative activity of CDC7is. Careful titration of CDC7 and ATR inhibitors reveals that when substantial levels of inhibition of both kinases is reached, cells are unlikely to complete DNA synthesis within S-phase and, in the absence of checkpoint control, undergo premature and lethal mitosis, which was previously reported by us and others. 32,33 However, we find two conditions in which ATRis and CDC7is show antagonism: the first condition is when low levels of CDC7i suppress the antiproliferative effects of high doses of ATRi. Fork instability and collapse is a feature in cells with impaired ATR signaling 60,61 and is substantially driven by nucleases acting on unprotected ssDNA. 62, 63 We have shown that CDC7 promotes MRE11 activity at forks and that CDC7 inhibition suppresses fork collapse induced by long treatment with hydroxyurea. Intriguingly, in the same study we have seen that CDC7 inhibition reduces H2AX phosphorylation in ATRi-treated cells in absence of other replication interfering agents, 32 suggesting CDC7 inhibition may also partially suppress fork collapse occurring in ATRi-treated cells, thus reducing DNA damage and loss of viability upon ATR inhibition (ref. 32 and Figure 1).
The second condition is when low doses of ATRi, which per se have no visible effects on the cell cycle, counteract the antiproliferative effects of high levels of CDC7i. This effect is mediated by the unleashing of origin firing even in the presence of CDC7is, thus accelerating the progression of cells through S-phase. PTBP1 loss of function, similarly to ETAA1 depletion, mimics this second condition allowing additional origin firing and cells to replicate more efficiently in presence of CDC7is.
Phenotypically, PTBP1 loss of function impairs RPA recruitment on chromatin and RPA focal formation, which explains why ATR signaling, which is dependent on RPA-coated ssDNA, 20 is deficient. Impaired RPA recruitment could be caused by several reasons, either by fewer forks being generated, which is unlikely as ATR impairment promotes origin firing and fork generation, by reduced fork stalling or by defective processing of forks. We suggest that a combination of the last two may occur as DNA fiber analyses show normal levels of origin firing and fork progression in parental and PTBP1-deficient cells. Consistent with problems in DNA replication, PTBP1-deficient cells show increased 53BP1 G1 nuclear bodies and micronuclei, which are a hallmark of genome instability, 48,64 although we cannot exclude that aberrant mitotic events also contribute to the generation of micronuclei in these cells.
Our analysis identifies RAD51AP1 as a target of PTBP1 regulation of pre-mRNA splicing and indeed ectopic expression of RAD51AP1 partially rescues checkpoint deficiency of PTBP1 defective cells. In vitro RAD51AP1 facilitates RAD51 D-loop formation, thus it is possible that RAD51AP1 deficiency alters the dynamics/residence time of RPA and RAD51 on ssDNA at stalled forks. 52,53,65 Recent work has shown that at telomeres RAD51AP1 also stabilizes R-loops, thus promoting the formation of G4 DNA structures, which can then promote break-induced replication and alternative lengthening of telomeres in the absence of telomerase. 55,66 By analogy, formation and stabilization of R-loops by RAD51AP1 may also occur in other regions of the genome, thus providing obstacles to the replication machinery and increasing the iScience Article dependency on ATR signaling. Such a hypothesis is consistent with previous work in DT-40 chicken cells showing that in unchallenged S-phase RAD51AP1 knock-out alters progression of replication forks and increases origin usage, which also in this system could be because of defective ATR signaling. 56 Furthermore, proteomics studies using proximity-based biotinylation have identified RIF1 and TOPBP1, which are important for determining the efficiency of origin firing and ATR activation, as part of the RAD51AP1 proximal interactome. 66 This may suggest an alternative mechanism to couple the events occurring at replication forks with late/dormant origin activation. As the reintroduction of RAD51AP1 only partially suppresses the replication and checkpoint phenotypes of PTBP1 deficiency, other mechanisms are involved and we speculate that, on fork stalling, further defects in signal generation and amplification could be because of misfunction of nucleases, such as MRE11, and the imbalance in the ratio of the ATR kinase activating subunits ( Figure S7).
Although this work positions PTBP1 as a novel and important factor in the replication stress response, so that the efficacy of CDC7is in suppressing DNA synthesis and proliferation is profoundly diminished in PTBP1-deficient cells within the first cell cycle, it did not escape our attention that PTBP1 also has an important role in driving cellular senescence, which can result from either excessive or prolonged DNA damage or replication stress. CDC7 inhibition can drive senescence in several cell types 67,68 and CDC7 inhibition in combination with agents that eliminate senescent cells was shown to have a very profound anti-tumor activity specifically in p53 mutated liver cancer models, suggesting a possible strategy to target these very aggressive tumors for which there is still a major unmet medical need. 67 As a note of caution, if such a therapeutic strategy is pursued, based on this study, it is predicted that resistance and relapse may arise by preexisting and acquired mutations in PTBP1 or in factors controlling PTBP1 expression.

Limitations of this study
Though our mechanistic study positions PTBP1 as an important factor in the replication stress response, this study has some limitations. Our data were solely generated in MCF10A cells where PTBP1 deficiency had a broad effect on mRNA splicing; in other cell lines PTBP1 deficiency may differently affect gene expression, thus the phenotypes described here for PTBP1 deficiency could be either more or less penetrant. In addition, this study does not include preclinical experimentation to conclude that the anti-tumor activity of CDC7 inhibitors is decreased in an in vivo setting. Here, we limited our investigation to RAD51AP1 as downstream effector of PTBP1's replication and checkpoint phenotype. However, additional factors contributing to these phenotypes are likely to exist, and their identification would complete the data presented in this study.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:   Cell culture was performed in a Class II Bio-safety cabinet and all cell lines were maintained at 37 C in a humidified atmosphere containing 5% CO 2 . Cell counts and viability were determined using a countess (Invitrogen) or LUNA II cell counter (Logos biosystems) and trypan blue exclusion. If not specified otherwise, all culture media and reagents were obtained from Merck. MCF10A cell line (human, female origin) was purchased from ATCC and authenticated via genome sequencing. MCF10A EditR cells (stably expressing Cas9) were generated as previously described in Rainey et al., 2017. 26 MCF10A PTBP1 mutant clone 4.1 and 6.6 were derived via CRISPR/Cas9 genome editing from MCF10A EditR cells and monoclonal expansion following a competitive proliferation assay described in method details. Mutation of PTBP1 was verified via targeted PCR and Sanger sequencing. MCF10A PTBP1-SM and MCF10A RAD51AP1 cells were generated for this publication as detailed in method details from MCF10A PTBP1 mutant clone 4.1. MCF10A cells and derivatives were cultured using DMEM supplemented with 5% (v/v) horse serum, 25 ng/ml cholera toxin, 10 mg/ml insulin, 20 ng/ml epidermal growth factor (Peprotech), 500 ng/ml hydrocortisone, 50 U/ml penicillin and 50 mg/ml streptomycin. MCF10A PTBP1-SM and MCF10A RAD51AP1 cells were additionally kept in 0.5 mg/ml G418 for selection. Lenti-Xä 293T cell line (human, female origin) were purchased from TaKaRa and used without further authentication. Routine culture was performed using DMEM + GlutMAX TM -I (ThermoFisher) supplemented with 10% (v/v) fetal bovine serum, 50 U/ml penicillin and 50 mg/ml streptomycin.

Drugs treatments
If not otherwise indicated, XL413 (synthesized in house) was used at a concentration of 20 mM and TAK-931 (Chemietek) was used at a concentration of 0.625 mM. Etoposide was used at 10 mM, camptothecin at 50 nM, cis-platin at 5 mM, and MMC at 200 nM (all acquired from Merck).

Competitive proliferation assay
To assess the effects of gene editing PTBP1 with PTBP1-sgRNA4 and PTBP1-sgRNA6, we adapted a competitive proliferation assay previously described in Rainey et al., 2020. 19 Oligonucleotides coding for sgRNAs were cloned into BsmBI restriction site of the LRG vector (Lenti_sgRNA_EFS_GFP; Addgene) 69 to generate PTBP1-g4-LRG and PTBP1-g6-LRG plasmids. MCF10A EditR cells were plated at 120,000 cells per well in 6-well plates. After 24 hours cells were transduced with viral particles containing either empty LRG, PTBP1-g4-LRG or PTBP1-g6-LRG at a Multiplicity of Infection (MoI) of 0.1. 24 hours after transduction, the culture media was changed and after an additional 24 hours cells were harvested and divided into two samples. The first sample was fixed in 4% (w/v) PFA in PBS for 20 min at room temperature before being washed once and resuspended in PBS. Cells were analysed with an AccuriC6 flow cytometer to determine cell count and levels of GFP expression as a readout for transduction efficiency. Cells were replated at 80,000 cells/well in the presence of 20 mM XL413. Samples were incubated over 16 days in the continued presence of 20 mM XL413 with a change of media every 4 days. Alternatively, to assess the ability of PTBP1 mutant clone 4.1 to resist long-term CDC7i treatment, MCF10A EditR and GFP expressing PTBP1 mutant clone 4.1 or RAD51AP1 expressing cells were harvested, counted and a sample was fixed as described above. The GFP expression was analysed using a BD FACS Canto II flow cytometer. Based on cell counts, 40,000 cells of MCF10A EditR and PTBP1 mutant clone 4.1 cells were combined and plated per well in the presence of XL413 or TAK-931 to achieve approximately 50% GFP-positive cells per well. Plates were incubated in a humid chamber for 8 days in the continued presence of XL413 or TAK-931. Analysis of GFP expression and replating was conducted as described once cells reached $80% confluency.

Establishing monoclonal cell lines with PTBP1 mutations
MCF10A EditR cells expressing PTBP1-sgRNA4 and PTBP1-sgRNA6 from the competitive proliferation assay were subject to limited dilution and incubated in 200 ml of complete media in 96-well plates for 10-14 days. Cells from wells that contained only single colonies were expanded and genotyped to assess PTBP1 mutational status. For screening, adherent cells in 96-well plates were washed with PBS and lysed in 50 ml of lysis buffer (10 mM Tris-HCl pH 7.5, 10 mM EDTA, 10 mM NaCl, 0.5% (w/v) N-Lauryl sarcosine, 10 mg/ml proteinase K and 20 mg/ml glycogen) for 2 hours at 60 C. Genomic DNA was precipitated by adding 3x volumes of 150 mM NaCl in 96% (v/v) EtOH before mixing and incubation at room temperature for iScience Article 30 min. DNA was pelleted by centrifugation (15 min, 800 x g), washed with 70% (v/v) EtOH and re-pelleted prior to air drying and resuspension in 100 ml TE buffer (10 mM Tris-HCl pH 7.5, 1 mM EDTA).
5 ml genomic DNA were used in PCR reactions using Taq DNA Polymerase and screening primers (Table S4: PTBP1-sg4 fwd/rev, PTBP1-sg6 fwd/rev). PCR products were purified with MACHEREY-NAGEL NucleoSpinâ Gel and PCR clean-up kit and sequenced using PTBP1-sg4 fwd or PTBP1-sg6 fwd primer (Table S4) by Eurofins genomics. The web-based tool TIDE (https://tide.nki.nl) 71 was used to assess genome editing of the PTBP1 locus. The chromatogram files from a sanger sequencing analysis of a control sample and a test sample were uploaded to TIDE alongside the 20-nucleotide sequence of the small guide RNA used to direct Cas9 to the target genomic locus. Using the quantitative sanger sequence trace data, TIDE analysis discriminated between wild-type, homozygous, heterozygous, and complex genotypes and provided predictions for potential indels at the CRISPR cut site. TIDE analysis indel predictions were confirmed by manual analysis of the Sanger sequencing reads.

Generation of lentiviral transfer plasmids
Total RNA was isolated from 1x10 6 MCF10A cells using the RNeasy mini kit (Qiagen) according to the manufacturer's instructions. Subsequently, 1 mg of purified total RNA was subject to first-stand cDNA synthesis using Oligo(dT) primers and the SuperScriptä II Reverse Transcriptase kit (Invitrogen). The resulting cDNA library was used as a template for PCR amplification using gene specific primers with adaptors. The primers were designed to amplify PTBP1 (PTBP1-fwd/rev) and RAD51AP1 (RAD51AP1-fwd/rev) from start to stop codon and add a NotI restriction site for cloning (Table S4). PCR products were cloned into the Not1 restriction sites of the lentiviral transfer plasmid pCDH-EF1-MCS-IRES-Neomycin (System Biosciences) and inserts were verified by Sanger sequencing (Eurofins genomics). To generate a sgRNA4-resistant variant of the cloned PTBP1 coding sequence, primers for site-directed mutagenesis (PTBP1-SDM fwd/rev) were designed (Table S4)

Small scale lentiviral packaging
Lenti-Xä 293T cells were seeded at 140,000 cells per well in a 6-well plate. After 24 hours, transfections were performed using a 1:2 ratio (w/v) of DNA to polyethyleneimine ''MAX'' MW 40,000 (1 mg/ml, Polysciences). 2 mg of Lentiviral Packaging Plasmids (Cellecta) and 0.4 mg lentiviral transfer plasmids were added to 100 ml of 150 mM NaCl prior to mixing with polyethyleneimine (4.8 ml) that was diluted separately in 100 ml of 150 mM NaCl. Reagents were vortexed and incubated at room temperature for 20 min before dropwise addition to cells. Media was changed after 24 hours and transfected Lenti-Xä 293T cells were allowed to produce virus for a further 24 hours. 2 ml of media was harvested, centrifuged (1,500 rpm for 5 min) and passed through a 0.45 mm sterile filter disc before being used for transduction of target cell lines.

Transduction of transgene expression vectors
MCF10A PTBP1-deficient clone 4.1 cells were seeded at 120,000 cells per well in a 6-well plate and after 24 hours were transduced with viral supernatant in the presence of polybrene (5 mg/ml). After 24 hours, cells that had undergone transduction were selected by adding fresh media containing G418 (0.5 mg/ml). Expression of the transgenes in these polyclonal cell populations was confirmed by western blotting.

siRNA transfections
MCF10A cells were seeded at 80,000 cells per well in a 6-well plate and transfected 24 hours after plating. siRNA transfections were performed using JetPRIME transfection reagent (Polyplus Transfection). For each individual well, siRNAs (50 nM final concentration) were prepared by mixing with JetPrime buffer (200 ml) and JetPrime reagent (4 ml) and the mixture was incubated for 15 min at room temperature prior to dropwise addition to the cells. Following a 5 hours incubation at 37 C, 5% CO 2 the culture media was exchanged for fresh, incubator-equilibrated media and cells were returned to the incubator. In this study, treatment of siRNA transfected cells was performed 48 hours after transfection. Proliferation assay with crystal violet MCF10A cells were seeded at 50,000 cells per well in a 6-well plate and treated 24 hours after plating. Cells were allowed to proliferate for 72 hours in media containing DMSO, 10 mM XL413 or 20 mM XL413. Media was discarded and plates were washed with PBS and allowed to dry completely. Cells were fixed by the addition of 2 ml 100% methanol and incubation for 30 min at room temperature. The methanol was discarded and the plates were air dried for 10 min. Cells were stained with 3 ml crystal violet solution (0.5% (w/v) crystal violet in 25% (v/v) methanol) for 30 min at room temperature and washed with H 2 O until any excess staining solution was removed. Plates were left to dry overnight. The plates were analysed using the Odyssey Infrared Imaging System and the intensity of cell stain was calculated using Image Studio software (LI-COR Biosciences). All values were normalised to corresponding DMSO controls.

96-well plates resazurin reduction assays
MCF10A cells were seeded at 2,500 cells per well in 96-well plates. Each well already contained 50 ml of inhibitors either alone or in combination. Each experiment was performed in technical triplicates. The effect of the CDC7 inhibitor TAK-931 was tested at six different concentrations using a 2-fold dilution that ranged from 5 to 0.0195 mM. The effect of the CDC7 inhibitor XL413 was tested at different concentrations using a 2-fold dilution that ranged from 40 to 2.5 mM or alternatively from 80 to 0.3125 mM. The effect of the ATR inhibitor, AZD6738, was tested at different concentrations using a 2-fold dilution series that ranged from 10 to 0.0391 mM. After 72 hours of drug treatment, the resazurin reduction assays were performed by addition of 0.56 mM resazurin sodium salt in PBS to each well to a final concentration of 93 mM. Plates were incubated for 6 hours at 37 C with 5% CO 2 and fluorescence intensity was determined using a Victor 3V 1420 multilabel counter plate reader with 530 nm excitation and 595 nm emission. Background from wells that contained only media and resazurin solution was subtracted and fluorescence intensity values were normalised to values from untreated controls to determine the relative surviving fraction. The half-maximal inhibitor concentration values (IC 50 ) were determined using GraphPad 9.5.0 by plotting the relative survival fraction of three biological repeats against the Log10 inhibitor concentration and using a nonlinear regression (curve fit) and log(inhibitor) versus normalized response-variable slope equation.
Additionally, the average surviving fraction for each individual and combination dose across three biological repeats was analysed using SynergyFinder 3.0 for interactive analysis and visualization of multi-drug and multi-dose combination response data (https://synergyfinder.fimm.fi). 72 Data was submitted to SynergyFinder as % viability and the recommended default options were used for the four-parameter logistic regression (LL4) curve fitting algorithm to fit single-drug dose-response curves. To investigate the degree of combination synergy or antagonism, the Zero interaction potency (ZIP) model was used to compare the change in the potency of the dose-response curves between individual drugs and their combinations. 2D and 3D synergy maps were downloaded from the SynergyFinder website and highlight synergistic and antagonistic dose regions in red and green colours, respectively. Delta scores obtained from the interactive 3D synergy plot were used as a guide to describe the interaction between the drug combinations: Antagonistic (d score <À10), additive (d score from À10 to 10) or synergistic (d score >10).

Flow cytometry
To analyse cell cycle distribution and rate of DNA synthesis, MCF10A cells were seeded at 160,000 per well in 6-well plates. Following treatment with indicated inhibitors or DMSO, nascent DNA was labelled by incubating cells with 10 mM EdU for 30 min at 37 C prior to harvesting. Cells were washed with PBS, which if not specified otherwise involves centrifugation at 400 x g for 5 min at 4 C and the removal of the supernatant from the cell pellet. Samples were fixed by resuspension in 0.3 ml PBS and dropwise addition of 0.7 ml 100% ethanol while vortexing followed by incubation for at least 1 hr at -20 C. After a PBS wash, cells were washed with 1% (w/v) BSA/PBS and incorporated EdU was labelled by incubating cells in click reaction buffer (PBS containing 10 mM Sodium-L-ascorbate, 2 mM Copper-II-Sulfate and 10 mM AF647-TEG-azide (Jena, CLK 1299-1) or alternatively 10 mM 6-Carboxyfluorescine-TEG-azide) for 30 min at room temperature protected from light. Cells were then incubated in PBS containing 1% (w/v) BSA and 0.5% (v/v) Tween-20 for 5 min at room temperature followed by an additional wash with PBS. DNA was stained with 1 mg/ml DAPI in 1% (w/v) BSA/PBS and to reduce RNA interference 0.5 mg/ml RNase A was added to each sample until measurement. Fluorescence intensity data for DAPI (405_450_50 nm) and AF647-TEG-azide (633_660_20 nm) or ll OPEN ACCESS

Fluorescence microscopy
MCF10A cells were seeded at a density of 150,000 cells per well in a 6-well plate on poly-L-lysine coated coverslip. Following drug treatment, coverslips were washed once with PBS and cells were fixed in PBS containing 4% (w/v) PFA for 10 min at room temperature. For the detection of nuclear RPA2, a cytoplasmic extraction with CSK buffer (see section: protein samples preparation) was performed for 15 min on ice prior to PFA fixation. Following the fixation step, samples were washed three times with PBS to remove residual PFA. Cells were permeabilized with PBS-TX (0.1% (v/v) Triton X-100 in PBS) for 20 min at room temperature iScience Article or alternatively for 53BP1 detection, cells were permeabilized with PBS-TX (0.125% (v/v) Triton X-100 in PBS) for 3 min at room temperature, followed by incubation for 30 min in blocking buffer (PTBP1, RPA2, beta-Tubulin staining: 10% (v/v) FBS, 0.5% (w/v) BSA in PBS-TX; 53BP1 staining: 1%.(w/v) BSA in PBS) at room temperature. Cells were incubated for 1 hr at 37 C with either mouse anti-PTBP1 (sc56701; Santa Cruz; 1:500), mouse anti-beta-Tubulin (#05-661; Millipore; 1:2000), rabbit anti-53BP1 (NB100-304; Novus Biologicals; 1:200) or mouse anti-RPA2 (NA19L, Calbiochem; 1:500) primary antibody diluted in blocking buffer. Following three washes with PBS-TX or PBS for 53BP1 detection, coverslips were incubated for 1 hr at 37 C with goat anti-mouse AlexaFluor 546 (A11003; ThermoFisher; 1:500) or goat anti-rabbit AlexaFluor 546 (A11010; ThermoFisher; 1:500) secondary antibody diluted in blocking buffer, which additionally contained DAPI (1:600 dilution in PBS of 0.5 mg/ml stock) to stain nuclei. Cover slips were washed three times in PBS-TX or PBS for 53BP1 detection, once in PBS and dipped in ddH 2 O before being mounted onto slides using Slow Fade Gold Antifade Reagent (ThermoFisher).
Microscopy for PTBP1, 53BP1 and beta-Tubulin detection was performed using IX71 Olympus microscope using a 40X objective lens or 60X oil-immersion objective as indicated. Mean fluorescence of nuclear PTBP1 signal was analysed for a minimum of 300 cells per sample in two independent experiments using ImageJ Fiji software. 70 Background signal was calculated from three independent measurements of empty areas for each picture and subtracted from the PTBP1 fluorescence intensity.
The number of micronuclei per cell were manually counted using DAPI staining. A minimum of 300 cells per sample of three independent experiments were analysed. Due to the uniform staining protocol for PTBP1 and beta-Tubulin, micronuclei counts of both experiments were combined. 53BP1 foci analysis was performed using ImageJ Fiji software. Intensity of the DAPI staining was used to identify cells in G1-phase by their low DNA content compared to S-or G2-phase cells. Low DAPI intensity was defined as less than the mean DAPI intensity of at least 610 cells for each sample. 53BP1 nuclear bodies were counted with ImageJ Fiji using the find maxima function and selection for high intensity foci. 53BP1 foci for 300 G1-phase cells were analysed for each condition and three independent experiments.
Mitotic aberrations were analysed manually by identifying mitotic cells and counting the following aberrant structures in samples with DAPI (DNA) and beta-Tubulin staining (microtubules; mitotic spindle): Multipolar spindles/spindle defects, chromatin bridges (broken and intact), unaligned or lagging chromosomes, chromosome fragments. At least 150 mitotic cells were analysed per sample in four independent experiments.
RPA2 analysis and microscopy were performed using the 60X objective lens on the Operetta CLS High Content Analysis System (PerkinElmer, London, UK). Programmed analysis using Harmony analysis software 3.1.1 was used to quantify the number of nuclear RPA2 foci per cell for a minimum of 200 cells per sample in three independent experiments. The analysis sequence used was as follows: identify nuclei based on DAPI intensity, exclude border cells, exclude doublet cells, detect foci based on signal to background ratio, output foci/nucleus counts.
After quantification and analysis, the brightness of representative images was adjusted to the same extent for all samples in an experiment to aid visualisation in figures.

DNA fibre spreading
Cells were seeded at a density of 300,000 cells per well in a 6-well plate and allowed to recover overnight.
To assess the effect of ATRi on MCF10A EditR cells in the presence of CDC7i, cells were treated with DMSO or 20 mM XL413 and 20 mM CldU for 30 min before the media was exchanged and cells were washed three times with pre-warmed culture media followed by treatment with DMSO, 0.156 mM or 5 mM AZD6738 and labelled with 200 mM IdU for 30 min in the continued presence or absence of XL413. Additionally, one sample was left untreated and unlabelled for later use. Cells were harvested, resuspended in ice-cold PBS, counted, and diluted to 2.5x10 5 cells/ml. Labelled cells were diluted 1:4 with unlabelled cells and gently mixed. 2.5 ml of the cell solution was removed and placed onto a coverslip before adding 7.5 ml of spreading buffer (0.5% (w/v) SDS, 200 mM Tris-HCl pH 7.4, 50 mM EDTA) and incubating the coverslip at room temperature for 8 min. The coverslips were tilted by 15 to allow the DNA to slowly run down the slide and air dried before being fixed in methanol/acetic acid (3:1) at 4 C overnight. Coverslips were removed from the fixative, incubated in 2.5 M HCl for 1 hr at room temperature to denature the DNA followed by three washes ll OPEN ACCESS iScience Article