Rucaparib Treatment Alters p53 Oscillations in Single Cells to Enhance DNA-Double-Strand-Break-Induced Cell Cycle Arrest

SUMMARY

DNA double strand breaks induce oscillatory expression of the transcription factor p53 that is dependent on ataxia telangiectasia mutated (ATM) activity and the rate of double strand break resolution. Although p53 dynamics are known to play a role in the regulation of cell fate determination, the consequences of the variability in dynamics associated with differences in repair rates and utilized repair pathways are unknown. Using single-cell time-lapse microscopy, we found that disruption of specific repair pathways has distinct impacts on p53 dynamics. The small-molecule rucaparib, an inhibitor of the alternative end-joining-associated protein poly (ADP-ribose) polymerase (PARP), increased p53 pulse duration, altering the temporal expression of multiple p53 target genes. As a result, combination treatments of the radiomimetic drug neocarzinostatin with rucaparib drove prolonged growth arrest beyond that of DNA damage alone. This study highlights how pharmacological manipulation of DNA repair pathways may be used to alter p53 dynamics to enhance therapeutic regimens.

Graphical Abstract

In Brief

p53 dynamics control the DNA damage response. Hanson and Batchelor show that disruption of distinct DNA repair pathways differentially alter p53 dynamics. The alt-EJ inhibitor rucaparib prolongs p53 expression, deregulating multiple target pathways. Rucaparib treatment prior to DNA damage prolongs growth arrest, suggesting an enhancement for genotoxic therapy regimens.

INTRODUCTION

Mutations in DNA-repair-associated proteins, including ataxia telangiectasia mutated (ATM), breast cancer type 1 susceptibility protein (BRCA1), and breast cancer type 2 susceptibility protein (BRCA2), are associated with increased sensitivity to certain types of DNA damage and increased risk for the development of cancer (Lavin and Shiloh, 1997; Romero-Laorden and Castro, 2017). Paradoxically, targeting defects in DNA repair pathways has proven an effective strategy in some current therapeutic interventions for cancer, such as the observed synthetic lethality that results from poly (ADP-ribose) polymerase (PARP) inhibition in tumors bearing BRCA1 or BRCA2 mutations (Bryant et al., 2005; Farmer et al., 2005). Understanding the function of key DNA repair pathways is crucial not only for improving our understanding of the physiological dysfunction that occurs during cancer development but may also aid in the development of new therapeutic strategies.

Single-cell studies of p53 have shown that p53 expression increases and decreases in distinct temporal patterns in response to different stresses, including oscillations in response to DNA double strand breaks (DSBs) and a single graded pulse in response to UV damage (Batchelor et al., 2011; Geva-Zatorsky et al., 2006; Lahav et al., 2004). These dynamics of p53 expression are shaped by the upstream regulatory kinases ATM, ataxia telangiectasia and Rad3 related (ATR), and DNA-dependent protein kinase (DNA-PK) (Batchelor et al., 2008; Finzel et al., 2016) and the negative regulators mouse double minute 2 (MDM2) and protein phosphatase 1D (WIP1) that feed back to degrade p53 levels (Batchelor et al., 2008). p53 dynamics play a key role in regulating expression patterns of downstream targets involved in cell fate determination (Hafner et al., 2017; Hanson et al., 2019; Porter et al., 2016; Purvis et al., 2012). The dynamics are correlated with the number of DSB foci (Loewer et al., 2013), and recent work has demonstrated that p53 dynamics vary across cell lines depending on intrinsic DNA repair rates and ATM activity (Stewart-Ornstein and Lahav, 2017).

Although a connection between p53 dynamics and DNA repair processes has been identified, several questions remain unanswered. For example, we do not understand how specific repair pathways affect p53 dynamics and subsequent p53 transcriptional activity. DNA DSBs can be repaired through several distinct pathways, including non-homologous end joining (NHEJ), homologous recombination (HR), alternative end joining (alt-EJ), and single strand annealing (SSA) (Chang et al., 2017). Each of these pathways uses unique repair proteins with different dynamic expression patterns (Aleksandrov et al., 2018; Chang et al., 2017; Janssen et al., 2016), potentially regulating p53 dynamics. The impact of DNA-repair-associated alterations on p53 dynamics, subsequent regulation of downstream target genes, and cell fate is also unknown. These questions have significant implications both for understanding the developmental process of cancers harboring the mutations and for the treatment of cancer given that many chemotherapeutic treatments induce DNA damage within both cancer cells and the surrounding normal tissues.

To address these questions, we combined live-cell imaging with targeted inhibition or knockdown of key regulatory proteins involved in specific repair pathways to assess their impact on p53 dynamics. We found that knockdown of BRCA1 and BRCA2 or treatment with the PARP inhibitor rucaparib resulted in specific qualitative alterations to p53 dynamics. Of the tested perturbations, rucaparib, a chemotherapeutic PARP inhibitor clinically approved for the treatment of certain forms of recurrent cancer (Papa et al., 2016), caused the largest impact on p53 dynamics, increasing DNA-PK activity and p53 pulse duration. The alteration to p53 expression resulted in changes to gene expression of multiple downstream stress response targets, enhancing DSB-mediated growth inhibition. Furthermore, we were able to recapitulate these findings using the p53-wild-type cancer cell line, MCF-7. These results highlight a potential mechanism to enhance the p53-mediated cellular response to DNA-damage-inducing therapies.

RESULTS

Inhibition of Specific DSB Repair Pathways Generates Qualitatively Distinct p53 Dynamics

In response to DNA DSBs, p53 levels in individual cells increase and decrease in a series of undamped pulses (Geva-Zatorsky et al., 2006; Lahav et al., 2004). p53 oscillations diversify gene expression patterns based on the stability of the target mRNA, as targets with decay rates greater than the p53 pulse frequency have pulsatile mRNA expression and targets with decay rates lower than the p53 pulse frequency have monotonically rising mRNA expression (Figure 1A). The rate of DNA repair influences p53 dynamics by removing the DNA damage stimulus required for p53 accumulation (Stewart-Ornstein and Lahav, 2017). Depending on the concentration and activity of key DNA-damage-sensing kinases in a cell, one recent model of p53 dynamics predicts that alterations in DNA repair may either increase the periodicity of p53 or push the system away from oscillatory behavior toward sustained p53 expression (Stewart-Ornstein and Lahav, 2017; Figure 1B). Such changes to p53 dynamics would qualitatively change downstream p53-regulated gene expression patterns (Figure 1B) and, consequently, cell fate.

Figure 1. Disruption of Repair-Associated Pathways Alters p53 Dynamics.

(A) Schematic of relationship between DNA damage, p53, and DNA repair. Under conditions of oscillatory p53 expression, gene expression patterns are diversified depending on mRNA stability and p53 pulse frequency (lower panels).

(B) Recent models predict a range of p53 dynamics in response to DNA repair disruption, including increased periodicity or sustained expression. These dynamics are predicted to alter diversification of gene expression patterns.

(C) Selection of specific inhibitions to target select DNA repair pathways: non-homologous end joining (NHEJ; red), alt-end joining (a-EJ; blue), and homologous recombination (HR; green).

(D–H) Upper panels are single cell traces for treatment with NCS (D), SCR7 (E), BRCA1/2 siRNA (F), BRCA2 siRNA (G), and rucaparib (H) with colored lines showing three representative single cells. All cells were treated with 400 ng/mL NCS. Lower panels show median p53 expression (black line) and interquartile range of population (colored area). Total number of cells quantified n is shown for each population plot.

(I) Mean fluorescence intensity ± SEM for γH2AX following NCS treatment and select perturbations of DSB repair.

Data were quantified based on at least 100 cells per condition and time point.

Given that DNA DSBs can be repaired through several alternative pathways, we first determined how p53 dynamics are altered when specific DNA repair pathways are perturbed. We selectively inhibited proteins involved in NHEJ, HR, or alt-EJ. Inhibition was performed using either chemical inhibitors or small interfering RNA (siRNA)-mediated knockdown of key repair-associated proteins (Figure 1C). To observe the associated effects on p53 dynamics, we used the immortalized, non-transformed retinal pigmented epithelial cell line RPE1-hTERT, which expresses wild-type repair networks and in which p53 oscillatory dynamics have been previously characterized (Loewer et al., 2010).

We quantified p53 dynamics via fluorescence microscopy of single living cells expressing a fluorescently tagged p53 (Batchelor et al., 2008; Lahav et al., 2004; Figures 1D-1H and S1A). For each experimental condition, we analyzed a minimum of 50 individual cells. Following induction of DNA DSBs with 400 ng/mL of the radiomimetic drug neocarzinostatin (NCS), we observed oscillations in p53 expression in individual cells (Figure 1D, upper panel), which appeared to be dampened at the population level due to desynchronization of the response between cells (Figure 1D, lower panel). Pre-treatment of cells with SCR7, an inhibitor of the NHEJ-associated protein DNA ligase IV (LigIV), resulted in no obvious changes to p53 oscillations (Figure 1E). siRNA-mediated knockdown of the HR-associated proteins BRCA1 and BRCA2 (Figure 1F) or BRCA2 alone (Figure 1G) increased the variability of p53 expression in individual cells as measured by interquartile range (Figure S1B) and standard deviation (Figure S1C), though a portion of cells still exhibited pulsatile expression. Following live-cell imaging, knockdown of BRCA1 and BRCA2 was confirmed by RT-PCR (Figures S1D and S1E). Pre-treatment of cells with rucaparib, an inhibitor of the alt-EJ-associated protein PARP, resulted in a sustained first pulse in p53 expression (Figure 1H). To confirm that perturbations impacted resolution of DSBs, we performed immunofluorescence staining of phosphorylated H2A histone family member X (γH2AX) at discrete intervals of 0, 1, 4, or 24 h post-NCS treatment (Figures 1I and S2). SCR7 had no impact on breaks, potentially explaining the observed lack of influence on p53 dynamics. In contrast, we found that knockdown of BRCA1 and BRCA2 or treatment with rucaparib generally increased γH2AX staining at 1 and 4 h post-NCS. Prior studies of p53 dynamics reported an increased duration of the first p53 pulse in response to DNA-PK inhibition, which was associated with significant changes to cell fate determination (Finzel et al., 2016) and was similar to what we observed in response to rucaparib treatment (Figure 1H). We therefore chose to further explore how rucaparib functionally altered p53 dynamics and downstream gene expression and cell fate.

Rucaparib Disrupts Oscillatory p53 Dynamics across a Range of DNA Damage Doses

Previous studies of p53 have shown that p53 dynamics vary across different doses of DNA damage (Stewart-Ornstein and Lahav, 2017). To accurately capture the spectrum of disruption of p53 dynamics with rucaparib, we treated RPE1-hTERT cells with a range of NCS doses and analyzed single-cell p53 dynamics (Figures 2A, 2B, and S3A). Consistent with prior studies of p53, we observed variations in dynamics as p53 ranged from weak pulses at low NCS doses to strong oscillations and toward prolonged expression at high NCS doses (Figure 2A). Rucaparib treatment resulted in a first peak of p53 expression approximately 1 to 2 h later than in cells treated with only NCS (Figures 2B and 2C; Videos S1 and S2).

Figure 2. Rucaparib Prolongs the First Pulse of p53 Expression in Response to NCS.

(A and B) Single-cell traces of mean fluorescence intensity for p53-Venus in cells treated with NCS (A) or NCS + rucaparib (B). Each row represents a different dose of NCS ranging from 10 ng/mL to 1,000 ng/mL. Colored lines highlight three individual representative cells. For each experiment, over 50 individual cells were analyzed. Total n for analyzed cells is shown for each plot.

(C) Images highlighting single cells treated with NCS (400 ng/mL) or NCS (400 ng/mL) and rucaparib. Images show a representative cell with a prolonged first pulse following rucaparib co-treatment. Scale bar represents 20 μm.

(D) Graphic highlighting traits quantified, including number of peaks, full-width half-maximum of the first peak, timing of first peak, and amplitude of first peak.

(E) Bar graph showing average number of pulses that occur within 24 h under each condition. Error bars are ± SEM. Total n is shown in (A) and (B).

(F–H) Boxplot showing range of first peak amplitude (F), time to first peak (G), and duration of first peak (H) for p53 under each condition. Median values are highlighted by red line. Boxes show interquartile range. Statistics were performed using one-way ANOVA for multiple comparisons (*p < 0.05; **p < 0.01; ***p < 0.001).

To quantify the impact of rucaparib on p53 dynamics, we measured several dynamic features, including the amplitude of expression at the first peak, timing of the first peak, duration of the first peak (full-width at half-maximum [FWHM]), and number of pulses observed over 24 h (Figure 2D). Prior studies of p53 pulses have shown a linear correlation with the number of DSBs (Loewer et al., 2013). Consistent with these studies of p53, we found cells exposed to higher NCS doses tended to exhibit more pulses until 1,000 ng/mL, a dose at which fewer pulses were observed (Figure 2E). By comparison, rucaparib treatment resulted in fewer pulses over 24 h, likely due to the prolonged duration of the first pulse, which limits the potential number of pulses. Increasing the dose of NCS alone did not dramatically alter first peak amplitude (Figure 2F), consistent with studies showing that p53 pulses tend to show fixed amplitudes and vary in pulse number based on DNA damage (Geva-Zatorsky et al., 2006; Lahav et al., 2004). Treatment with rucaparib modestly increased median expression at the first peak, suggesting a potential enhancement of p53 expression during the first synchronized pulse. In NCS-treated cells, the first p53 pulse occurred between 2 and 3 h post-NCS addition (Figure 2G). Rucaparib treatment delayed the first pulse, and peak expression occurred between 4 and 8 h (Figure 2G) in response to intermediate doses of NCS (100 and 400 ng/mL). Higher doses of NCS (1,000 ng/mL) resulted in individual cells that exhibited peak expression >10 h beyond initial addition of NCS. Consistent with a prolonged time to peak expression, we observed a significant increase in the duration of the first p53 pulse as quantified by FWHM expression at all doses when NCS was combined with rucaparib (Figure 2H). Analysis of second and third subsequent peaks confirms that the major impact of rucaparib occurs during the first peak (Figures S3B-S3D). Interestingly, these effects were specific to rucaparib, as neither olaparib nor niraparib could reproduce this prolonged activation of p53 (Figures S4A-S4C), suggesting that the kinetics and specificity of PARP inhibitors (Carney et al., 2018) may influence the impact on p53. Given that p53 expression is primarily regulated through post-translational modification, we hypothesized the prolonged expression may be due to changes in either the stabilization or degradation of p53.

Rucaparib Treatment Increases Activation of DNA-PKcs

We next sought to determine the potential mechanism driving the rucaparib-dependent prolonged expression of the first p53 pulse. Initially, we hypothesized that the prolonged expression of p53 might result from inhibition of MDM2 or WIP1, negative regulators of p53 stability. However, we observed no distinct changes to expression of either MDM2 or WIP1 with rucaparib treatment (Figures 3A and 3B). Elongation of p53 pulse duration has previously been observed following inhibition of the DNA DSB-sensing kinase DNA-PK, leading to elevated ATM activity (Finzel et al., 2016). However, western blot analysis of ATM and checkpoint kinase 2 (CHK2) showed no increase in ATM activity (Figures 3A and 3B). Finally, we checked whether DNA-PK activity itself was altered with rucaparib treatment, and we found that phosphorylation of the DNA-PK catalytic subunit (DNA-PKcs) Ser-2056, an indicator of active DNA-PK, was increased (Figures 3A and 3B). This result is consistent with previous studies suggesting that PARP inhibitors can drive activation of DNA-PKcs (Patel et al., 2011).

Figure 3. Rucaparib Increases Activation of DNA-PKcs in Response to DNA Damage.

(A) Western blot analysis of positive and negative regulators of p53 in response to NCS (400 ng/mL) alone or in combination with rucaparib (10 μM). Images are representative of three independent biological replicates (n = 3). Actin serves as a loading control for images.

(B) Quantification of western blots for p53 and associated regulators. Expression was normalized based on actin loading and the expression level of NCS-alone-treated samples at time = 0. Error bars ± SEM.

(C) Western blot analysis of DNA-PKcs phosphorylation 2 h post-NCS addition. Additional samples received rucaparib and increasing doses of DNA-PKi inhibitor NU7441 (10 nM–10 μM). Actin serves as a loading control for images.

(D) Population averages ± SD for p53 expression in response to a range of NCS, NCS + rucaparib, or NCS + rucaparib + NU7441. Number of cells n for each condition is shown in upper left of each plot.

(E) Bar graph illustrating mean pulse number observed over 24 h for each condition. Error bars ± SEM (n for each condition shown in D).

(F and G) Boxplot indicating time to first peak (F) and duration of first peak (G) of p53 expression in individual cells for each condition.

Red line indicates median with boxes showing interquartile range (n for each condition shown in D). Statistics for (E)–(G) were performed by one-way ANOVA for multiple comparisons between NCS-, rucaparib-, and rucaparib + NU7441-50 nM-treated groups (*p < 0.05; **p < 0.01; ***p < 0.001; ns, not significant).

Given the increased activation of DNA-PKcs, we sought to examine whether we could reverse the prolonged p53 expression observed with rucaparib treatment by inhibiting DNA-PKcs. Inhibition of DNA-PKcs can also drive prolonged p53 expression, so we anticipated only a relatively narrow dose range could reverse the dynamics phenotype without driving hyperactivation of ATM (Finzel et al., 2016). We identified a concentration of the DNA-PK inhibitor NU7441 that restored DNA-PKcs phosphorylation to levels comparable with NCS alone, rather than complete inhibition. We found that DNA-PKcs phosphorylation was markedly inhibited with a dose of NU7441 between 0.01 μM and 0.1 μM (10–100 nM; Figure 3C).

To test whether DNA-PKcs inhibition restored the oscillatory phenotype observed with p53 alone, we imaged cells treated with 400 ng/mL NCS in the presence of combinations of rucaparib and NU7441 and measured dynamics of fluorescently labeled p53 (Figure 3D). To quantify the effect on p53 dynamics, we focused on the oscillatory characteristics previously observed to change with rucaparib treatment: the number of pulses (Figure 3E); the time to first peak (Figure 3F); and the duration of the first peak (Figure 3G). As in Figure 2, rucaparib and NCS co-treatment resulted in fewer pulses over 24 h (Figure 3E), delayed timing of the first peak (Figure 3F), and increased duration of the first peak (Figure 3G). Additional treatment with 10 nM or 50 nM NU7441 partially rescued the phenotype, generating p53 dynamics close to those observed in response to NCS treatment alone by decreasing the time of the first peak and duration of the first p53 pulse (Figures 3F and 3G). This result supports our hypothesis that the alteration to p53 oscillations due to rucaparib is mediated by activation of DNA-PK. At higher doses of NU7441 treatment (1–3 μM), cells lost oscillatory p53 expression and showed extremely prolonged accumulation of p53 (Figure 3G), suggesting that retention of some basal DNA-PKcs signaling is necessary for the negative regulation of p53 that drives oscillatory dynamics.

Rucaparib Alters the Timing and Magnitude of p53 Target Gene Expression

Previous studies have demonstrated that p53 pulses diversify target gene expression based on the relationship between mRNA stability and p53 pulse frequency (Hafner et al., 2017; Hanson et al., 2019; Porter et al., 2016), and p53 dynamics directly impact cell fate regulation in response to DNA damage (Purvis et al., 2012). Given the observed changes in p53 dynamics upon rucaparib treatment, we sought to determine how the changes alter the expression of downstream transcriptional targets of p53. We focused on a panel of well-characterized p53 target genes important in several cell fate pathways previously shown to have either pulsing mRNA expression dynamics (TRIAP1, PUMA, CDKN1A, TIGAR, MDM2, PLK3, NOXA, WIP1, BTG2, GDF15, GADD45A, and MYC) or rising mRNA expression dynamics (BNIP3L, DDB2, PLK2, PIG3, TP53INP1, RRM2B, PIDD, SFN, and SESN1) in response to 400 ng/mL NCS (Porter et al., 2016). We first confirmed p53 dependence for a subset of these well-characterized p53 target genes using siRNA-mediated knockdown of p53 (Figure S5A). Using qRT-PCR, we quantified the expression of the target genes over 10 h in response to either low (10 ng/mL) or high (400 ng/mL) NCS doses in the presence or absence of rucaparib (Figures 4A and 4B). Treatment with rucaparib alone increased expression (>2-fold) of PLK3 and GDF15 within the first 5 h (Figure S5B), but the majority of p53 targets were not strongly induced.

Figure 4. Rucaparib Treatment Influences Timing and Expression of p53 Target Genes.

(A) Fold change in expression for p53 target genes in response to 10 ng/mL NCS alone or in combination with rucaparib. Data were normalized to expression at the 0-h time point. Error bars ± SEM for three biological replicates (n = 3).

(B) Fold change in expression for p53 target genes in response to 400 ng/mL NCS alone or in combination with rucaparib. Data are normalized to expression at the 0-h time point. Error bars ± SEM for three biological replicates (n = 3).

(C) Min-max normalization of 400 ng/mL NCS (gray) and NCS + rucaparib (blue) for pulsatile p53 targets.

(D) Bar graph showing maximum fold change in expression for p53 targets with rising expression dynamics. Error bars ± SEM (n = 3). Statistical significance was determined by an unpaired t test comparing NCS-treated samples with respective rucaparib-treated samples (*p < 0.05; **p < 0.01; ***p < 0.001).

(E) Bar graph showing maximal fold change in expression for p53 targets with pulsatile expression dynamics.

Error bars ± SEM (n = 3). Statistical significance was determined by an unpaired t test comparing NCS-treated samples with respective rucaparib-treated samples (*p < 0.05; **p < 0.01; ***p < 0.001).

Based on the changes to p53 dynamics we observed in single cells, we predicted that rucaparib treatment would delay the timing of peak gene expression for pulsing target genes. To quantify the timing of peak gene expression, we performed a min-max normalization of the high dose expression data (400 ng/mL NCS; Figure 4C). With rucaparib, we observed a delay in peak expression in most pulsing genes (Figure 4C), consistent with the delay observed in p53 expression in single cells (Figure 2G).

Oscillatory p53 expression discriminates gene expression patterns based on the mRNA degradation rate relative to the p53 oscillation frequency (Porter et al., 2016). We hypothesized that alteration of p53 oscillations due to rucaparib treatment would reduce the ability of p53 to diversify target gene expression patterns. Given the observed delay in peak expression of pulsing genes and the broadened post-peak expression apparent for pulsing targets, such as WIP1 and BTG2 (Figure 4C), we sought to determine whether there is a loss in the distinction between pulsing and rising gene expression normally established during the DSB response. We performed unsupervised hierarchical clustering of min-max normalized expression data. Although cells treated with only NCS showed consistent stratification of target genes based on prior classifications of pulsing and rising dynamics (Porter et al., 2016), co-treatment with rucaparib resulted in disruption of these clusters (Figures S5C and S5D). Taken together, these results indicate that prolonged expression of p53 due to rucaparib delays the timing of pulsing gene targets and reduces the diversity of gene expression patterns, potentially altering the balance of many target genes established in response to DSBs.

Despite single-cell analysis suggesting a modest amplification of p53 expression with rucaparib treatment (Figure 2F), we did not observe increased expression of all p53 target genes examined. Rather, we found that rucaparib treatment tended to preferentially increase expression of slowly decaying rising targets (Figure 4D), whereas rapidly decaying pulsing targets exhibited more variable effects (Figure 4E), with GADD45A, PLK3, and GDF15 having reduced expression. Increasing the NCS dosage (comparing 10 ng/mL to 400 ng/mL) increased gene expression for virtually all p53 targets, suggesting this effect is not due to variation in treatment conditions and may reflect the additional role of PARP in the regulation of transcriptional complexes independent of DNA damage (Gibson et al., 2016; Leutert et al., 2016). Given that rucaparib treatment altered expression of several regulators of cell fate, including PUMA and CDKN1A, we sought to determine whether these changes influenced the DNA DSB-induced cell fate regulation.

Rucaparib Enhances Anti-proliferative Effects of NCS Treatment

Prior studies of p53 dynamics in single cells identified two general dynamic features that determine cell fate. In response to etopo-side treatment, the transition of p53 from oscillatory to monotonic expression dynamics increases the rate of cell death (Chen et al., 2013; Yang et al., 2018). Whereas, in response to DSBs induced by NCS or γ-irradiation, sustained expression of p53 leads to prolonged cell cycle arrest and early activation of senescence (Purvis et al., 2012). Given the extended duration of p53 expression observed with rucaparib treatment and the significant changes in gene expression of key cell fate regulators, we determined whether rucaparib treatment could enhance cell cycle arrest similar to results observed with DNA-PK inhibition (Finzel et al., 2016) and nutlin-3 treatment (Purvis et al., 2012). We fixed cells at 24 and 48 h post-NCS treatment either with or without rucaparib pre-treatment and analyzed cell cycle distribution by propidium iodide staining (Figure 5A). A low dose of NCS (10 ng/mL) did not significantly impact the cell cycle distribution; however, in response to a high dose of NCS (400 ng/mL), we observed rapid depletion of S phase cells and accumulation of cells in G2/M at both 24 and 48 h post-NCS (Figures 5A and 5B). These results are consistent with our microscopy studies, in which extensive cellular division was still observed at a low dose, but not high dose, of NCS. In cells treated with a high NCS dose (400 ng/mL), pre-treatment with rucaparib did not induce additional arrest as assessed by S/G2 ratio (Figure 5B), suggesting a saturating effect on cell cycle arrest activity at high levels of DNA damage. In contrast, rucaparib pre-treatment before low-dose NCS increased the accumulation of cells in G2 phase and depletion of cells in S phase (Figure 5B), suggesting an amplification of growth inhibitory effects through PARP inhibition.

Figure 5. Rucaparib Enhances Growth Inhibitory Effects of NCS.

(A) Mean percentage of cells in each cell cycle phase (G1, S, and G2/M) for each condition tested over 24 and 48 h. Data represent triplicate experiments (n = 3). Error bars ± SEM.

(B) Ratio of S/G2 phases across triplicate experiments (n = 3). Error bars ± SEM. Statistical significance was determined by performing one-way ANOVA. ***p < 0.001.

(C) Cell viability over 96 h in response to the indicated treatment. Error bars ± SEM.

Data represent 4 independent biological replicates for each treatment and time point (n = 4). Statistical significance was determined by unpaired t test comparing NCS-treated time points with respective rucaparib-treated time points. (* comparing NCS [10 ng/mL] samples; # comparing NCS [400 ng/mL] samples). **p < 0.01; ##p < 0.01.

Loss of p53 oscillations and sustained p53 expression has been shown to induce prolonged growth arrest for several days (Purvis et al., 2012). Because rucaparib treatment generates a shift toward sustained p53 expression, we first tested whether rucaparib inhibition during a DSB response elevates levels of cellular senescence, completely removing cells from the cell cycle. Senescence-associated beta-galactosidase (SA-β-Gal) staining showed no difference in the levels of senescent cells with or without rucaparib treatment (Figure S6), indicating that the effects of rucaparib treatment on cell cycle arrest are transient. We next performed a cell viability assay to assess the duration of growth inhibitory effects. Low-dose NCS treatment resulted in minimal effects on cell viability (Figure 5C, gray line); however, pre-treatment with rucaparib significantly attenuated proliferation within 48 h (Figure 5C, light blue line). A high dose of NCS strongly inhibited cell viability (Figure 5C, black line), and addition of rucaparib enhanced the effect by 72 h after DNA damage (Figure 5C, dark blue line). Taken together, these results indicate that rucaparib can effectively enhance cell cycle arrest and reduce cell viability during the cellular DNA DSB response.

Rucaparib Drives Prolonged p53 Expression in the MCF-7 Breast Cancer Cell Line

Our results thus far have demonstrated that rucaparib leads to prolonged accumulation of p53 and enhancement of growth arrest in response to DNA damage in immortalized non-transformed RPE1-hTERT cells. To assess whether these findings were applicable to cancer cells where rucaparib may provide clinical benefit, we examined the single cell dynamics of p53 in response to rucaparib and NCS in the MCF-7 breast cancer cell line using long-term fluorescence microscopy (Figure 6A). These cells express wild-type p53 and retain BRCA function, avoiding potential synergistic killing from PARP inhibition. We found that, in response to 400 ng/mL NCS, additional treatment with rucaparib led to prolonged accumulation of p53 during the first pulse. Similar to RPE1-hTERT cells, this led to reduced pulses over 24 h (Figure 6B), increased amplitude (Figure 6C), increased time of first peak (Figure 6D), and an increased duration of the first pulse (Figure 6E).

Figure 6. Rucaparib Enhances p53 Expression and Growth Arrest in Breast Cancer Cells.

(A) Single-cell traces (top panels) and population averages ± SD (bottom panels) for MCF-7 cells treated with 400 ng/mL NCS (black) or 400 ng/mL NCS + 10 μM rucaparib (blue). Number of cells n for each plot is shown in lower panels.

(B) Average number of pulses over 24 h ± SEM comparing NCS or NCS + rucaparib-treated MCF-7 cells. ***p < 0.001 (one-way ANOVA).

(C) Boxplot of first peak amplitude of p53 expression comparing NCS or NCS + rucaparib-treated MCF-7 cells. *p < 0.05 (one-way ANOVA).

(D) Boxplot comparing timing of the first peak of p53 expression between NCS or NCS + rucaparib-treated MCF-7 cells.

(E) Boxplot comparing duration of first p53 pulse between NCS or NCS + rucaparib-treated MCF-7 cells. ***p < 0.001 (one-way ANOVA).

(F) Cell viability over 96 h in response to the indicated treatment. Error bars ± SEM.

Data represent 6 independent biological replicates for each treatment and time point (n = 6). **p < 0.01 (unpaired t test comparing NCS-treated time points with respective rucaparib-treated time points).

To examine whether treatment with both rucaparib and NCS could enhance growth arrest in MCF-7 cells, we performed growth assays in response to either NCS alone or in combination with rucaparib. Similar to our findings in RPE1-hTERT cells, combination treatment of 400 ng/mL NCS and rucaparib led to further reduction in cell viability (Figure 6F). Rucaparib in conjunction with the lower 10 ng/mL NCS dose was not able to reduce cell viability and may reflect the more modest impact on p53 dynamics observed in MCF-7 cells as compared to RPE1-hTERT cells.

DISCUSSION

The role of DNA repair in the suppression of tumor development is well established, and mutations in numerous DNA-repair-associated proteins have shown a predisposition for cancer development (Lavin and Shiloh, 1997; Romero-Laorden and Castro, 2017). The increasing prevalence of single-cell-based systems has enabled significant insight in the temporal dynamics of the DNA damage response. In particular, the relationship between p53 expression and the resolution of DSBs as assessed by TP53-binding protein 1 (53BP1) has been established (Loewer et al., 2013). More recent work has demonstrated a correlation between variation in p53 dynamics and intrinsic repair rates across multiple cell lines (Stewart-Ornstein and Lahav, 2017); however, a clear mechanistic or functional significance for these variations has not been determined. We examined how inhibition of specific DNA-repair-associated proteins affects the temporal expression of p53 in response to DSBs. We determined that alterations to p53 dynamics are dependent on the targeted repair protein: the PARP inhibitor rucaparib generates a sustained first pulse of p53 in response to DNA damage, whereas BRCA1 and BRCA2 knockdown increases variability in p53 dynamics. Although not explored in this study, the increased variability upon BRCA1 or BRCA2 knockdown may result from the cell cycle dependence of homologous recombination, which occurs only in S and G2 phase after replication of DNA. Single-cell studies of cell cycle regulation in response to DSBs have shown that the timing of DNA damage has pronounced effects on cell fate determination (Chao et al., 2017). As such, cells in different cell cycle phases may generate distinct p53 dynamics in response to the same genotoxic stress. Future studies will be required to address these questions.

Treatment of cells with rucaparib leads to a prolonged first pulse of p53 and, consequently, a disrupted p53 pulse frequency in response to DNA damage. Interestingly, the impact on dynamic expression of p53 appears specific to rucaparib as neither olaparib nor niraparib could reproduce this effect. Although PARP inhibitors primarily target PARP1/2, these differences may result from the impact of compound-specific secondary targets, such as PARP3 (Dockery et al., 2017) or H6PD (Knezevic et al., 2016). Another possibility is that the difference in effect on p53 may be due to differences in the kinetics of different PARP inhibitors (Carney et al., 2018), as timing of treatments has been shown to be important in the modulation of p53 dynamics and cell death (Chen et al., 2016), as well as in other chemotherapeutic approaches (Lee et al., 2012).

We previously showed that the dynamic expression of downstream p53 targets was dependent on the relationship between the p53 pulse frequency, the mRNA stability of the target, and the protein stability of the target. p53 targets with short-lived mRNA and protein have pulsatile expression; p53 targets with long-lived mRNAs or proteins have rising expression (Hanson et al., 2019; Porter et al., 2016). We refer to the p53 targets in these two categories as “pulsing” or “rising,” respectively (Hanson et al., 2019; Porter et al., 2016). Using qRT-PCR, we assessed how the changes in p53 pulse dynamics upon rucaparib treatment affected downstream expression of pulsing and rising p53 targets, identifying two main effects related to the timing and variability of target expression. For pulsing targets, prolonged duration of p53 expression resulted in a delay in the timing of expression. Additionally, rucaparib treatment resulted in consistent upregulation of rising target genes but more variable expression of pulsing targets. For the case of some pulsing targets, such as GADD45A, rucaparib treatment even inhibited the normal induction in response to DNA damage (Figure 4A). Given the lack of correlation between p53 expression and the expression level of the targets, we hypothesize PARP may be playing a p53-independent role on expression of a subset of p53 targets. Several studies have shown that PARP enhances transcriptional pause release (Gibson et al., 2016; Leutert et al., 2016). Several p53 targets, including CDKN1A and GADD45A, have been shown to have a relatively higher frequency of paused polymerases at their promoters (Allen et al., 2014; Gomes et al., 2006). Inhibition of PARP may prevent effective transcription of these targets following DNA damage.

p53 dynamics are important for the regulation of cell fate in response to DNA damage (Purvis et al., 2012). DNA-PKcs inhibition was previously shown to generate a prolonged first p53 pulse, leading to increased rates of cell cycle arrest (Finzel et al., 2016). We showed that treatment with rucaparib is another method to enhance cell cycle arrest features at low doses of DNA damage. At higher doses of damage, rucaparib promoted sustained decreases in cell viability. One possibility is that these effects are the result of a rucaparib-dependent but p53-independent reduction in DNA repair rates, which would be difficult to establish explicitly due to the many interrelated effects of p53 on DNA repair, cell cycle arrest, and cell viability. The similarity of DNA-PKcs and rucaparib inhibition favors an interpretation in which rucaparib functions through a similar mechanism of DNA-PKcs inhibition to hyperactivate ATM. Indeed, prior study of PARP inhibitors have suggested that DNA-PK inhibitors target the same repair pathway (Spagnolo et al., 2012). However, we found instead that rucaparib decreased both ATM and CHK2 phosphorylation, likely through increased activation of DNA-PKcs. Interestingly, although prior studies have highlighted the importance of WIP1 in the maintenance of p53 oscillations in MCF-7 cells, we observed no evidence of oscillatory expression in RPE1-hTERT cells despite observed changes in mRNA expression. This may reflect the poor correlation between mRNA and protein expression (de Sousa Abreu et al., 2009; Schwanhäusser et al., 2011). Although rucaparib treatment was able to prolong p53 expression at the first pulse in MCF-7 cells, these cells were more oscillatory than RPE1-hTERT cells, suggesting that the relative influence of specific p53 regulators varies across cell lines. Consistent with this hypothesis, it has been shown that different cell lines exhibit differences in oscillatory behavior based on intrinsic ATM and DNA repair rates (Stewart-Ornstein and Lahav, 2017). Furthermore, in specific conditions, the expression of MDM2 is the key factor in dictating p53 pulses or accumulation (Yang et al., 2018). Thus, p53 dynamics in RPE1-hTERT cells may be more heavily influenced by DNA-PK and MDM2, whereas in MCF-7 cells, they may be more reliant on ATM and WIP1 for repeated oscillations.

We were able to partially reverse the rucaparib-induced responses in RPE1-hTERT cells through modest inhibition of DNA-PKcs. Given the role of DNA-PKcs in NHEJ, the effects we observe may represent cells increasing repair through NHEJ to compensate for the loss of PARP-dependent repair. Increased DNA-PKcs activity is associated with poor prognosis and radiotherapy resistance in several tumor models (Goodwin and Knudsen, 2014; Hsu et al., 2012). We found that increasing inhibition of DNA-PKcs led to greater increases in p53 accumulation when paired with rucaparib, potentially due to loss of both NHEJ and PARP-dependent repair. These findings suggest that combining PARP and DNA-PKcs inhibition may result in increased DNA damage in tumor cells and thus may improve treatment efficacy, especially for tumors with wild-type p53 expression.

STAR★METHODS

RESOURCE AVAILABILITY

Lead Contact

Further information and requests for resources should be directed to and will be fulfilled by the Lead Contact, Eric Batchelor (ebatchel@umn.edu).

Materials Availability

Plasmid for H2B-Ruby fluorescent nuclear marker is publicly available from the Markus Covert lab on Addgene (Kudo et al., 2018) (pLentiPGK Hygro DEST H2B-mRuby2, Plasmid#90236). Plasmid for the expression of p53-mVenus is available upon request through the Lead Contact, Eric Batchelor.

Data and Code Availability

Single cell p53 expression data for all single cell studies within the manuscript have been deposited in Mendeley Data (“Hanson and Batchelor 2020 Cell Reports,” Mendeley Data,V1, https://doi.org/10.17632/c9fvknxdjf.1) and are available for download as a MATLAB data file. Each file contains a cell array containing single cell p53 expression values over the 24 h time course for each condition. A cell array with the names for each condition and a data matrix containing each time point to facilitate easy reproduction of the plotted data presented within this manuscript. All analysis code was generated using MATLAB R2020a and specific code can be provided upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Plasmids

Plasmid for expression of fluorescently tagged H2B, pLentiPGK Hygro DEST H2B-mRuby2, was obtained from Addgene (Plasmid #90236). Packaging of plasmid into lentivirus was performed using the Takara Lenti-X Packaging Single shot kit (631275) and 293T cells (ATCC, CRL-11268) according to manufacturer specifications. Virus containing media was collected at 48 and 72 h post-infection, filtered, and concentrated 10-fold using Takara Lenti-X concentrator (631231). Virus production was confirmed by Lenti-X GoStix (Takara, 631280). Lentivirus was stored in aliquots at −80°C until use.

Plasmid for expression of p53-mVenus, pDESTN-pMT-p53-Venus, was previously established as described (Batchelor et al., 2008).

Human Cell Lines and Culture

RPE1-hTERT retinal pigmented epithelial cells (female) were obtained from ATCC (CRL-4000) and maintained in DMEM/F12 medium (Fisher Scientific, SH3002301) containing 10% FBS, 10 μg/ml hygromycin, 100 U/mL penicillin, 100 μg/mL streptomycin, and 250 ng/mL amphotericin B (Corning).

To establish cell lines for single cell imaging of p53, we transfected RPE1-hTERT cells with pDESTN-pMT-p53-mVenus in 6-well plates. First 2 μg of plasmid DNA was diluted in 125 μl Opti-MEM I (ThermoFisher, 31985070) with 4 μl P3000. A second tube containing 125 μl Opti-MEM I and 3.75 μl Lipofectamine 3000 (Invitrogen, L3000008). Tubes were mixed 1:1 and incubated at RT for 15 min and then added to cells. Media was replaced after 8 h. Cells underwent selection with 1000 μl/ml G418 for 1-week and then cells were isolated by FACS. Single cell clones were established by limiting dilution and expression confirmed by microscopy. To introduce the nuclear marker H2B-mRuby2, RPE1-hTERT p53-mVenus cells were transduced with lentivirus contained pLentiPGK Hygro DEST H2B-mRuby2 (Addgene, #90236) diluted 1:50 in DMEM/F12 media supplemented with 8 μg/ml protamine sulfate and 10 μM HEPES. Positive cells were isolated by FACS and single clones established by limiting dilution. Expression of both p53-mVenus and H2B-mRuby2 was confirmed by fluorescent microscopy.

MCF-7 breast carcinoma cells (female) were obtained from ATCC (HTB-22) and maintained in a base medium of RPMI containing 10% fetal bovine serum (FBS), 100 U/mL penicillin, 100 μg/mL streptomycin, and 250 ng/mL amphotericin B (Corning). MCF-7 cells expressing fluorescently tagged p53 (p53-Venus) (Batchelor et al., 2008) were grown in base medium supplemented with a G418 at a maintenance concentration of 400 μg/ml All cell lines were maintained at 37°C and 5% CO₂.

METHOD DETAILS

Chemical Inhibition of DNA Repair

The LigIV inhibitor SCR7, PARP inhibitors olaparib, niraparib, and rucaparib, and DNA-PK inhibitor NU7441 used within this study were obtained from SelleckChem (S7742, S1060, S2741, S1098, and S2638, respectively). All pharmacological inhibitors were prepared in DMSO. For inhibition of DNA repair pathways, SCR7, olaparib, niraparib, and rucaparib were all used at 10 uM concentrations. NU7441 was examined across a range of doses (10nM-10μM). Cells were incubated with inhibitors for 2 h prior to start of experiments to ensure entry of inhibitor into cells. Control cells received a vehicle DMSO control of equal volume.

siRNA Knockdown

siRNA knockdown was performed using two siRNA against BRCA1 (Horizon, D-003461-05 and D-003461-06) and BRCA2 (Horizon, D-003462-01 and D-003462-02). Control cells received a non-targeting siRNA control (D-001810-10-05). For each experiment, cells were plated at approximately 50% confluency in 35 mm dishes. 10 μl of 5 μM siRNA was added to 190 μl Opti-MEM I for Tube A. Tube B was prepared containing 4 μl Dharmafect I (Horizon, T-2001-02) and 196 μl Opti-MEM I. Tube A and B were mixed and incubated at room temperature for 20 min and added directly to plated cells. Then 1600 μl of DMEM/F12 was added for final siRNA concentration of 25 nM. 18 h post transfection, the transfection medium was removed, and fresh medium added. After 48 h, cells were imaged via microscopy. Knockdown of BRCA1 or BRCA2 was confirmed by qRT-PCR post imaging. For knockdown of p53, cells received identical treatments except for addition of 10 nM siRNA against p53 (Horizon, L-003329-00-0005).

Double Strand Break Induction

To measure dynamics of p53 and its targets in response to DNA double strand breaks, cells were treated with the radiomimetic neocarzinostatin (NCS; Sigma, N9162). This drug has previously been demonstrated to rapidly induce DNA double strand breaks (Shiloh et al., 1983). Cells were harvested at select time points and lysates were used for western blot analysis.

Live-cell Microscopy of p53-mVenus

For live-cell imaging, cells were plated into glass-bottom dishes (Mattek) and were imaged 24 h after plating. 2 h prior to imaging, medium was replaced with transparent medium lacking riboflavin and phenol red (ThermoFisher) and supplemented with 2% fetal bovine serum (FBS), 100 U/mL penicillin G, 100 mg/mL streptomycin sulfate, and 250 ng/mL amphotericin B (Corning). At this point, cells either received a DSB repair inhibitor (SCR7, olaparib, niraparib rucaparib, or NU7441) or a DMSO control dose. After 2 h, cells were imaged with a Nikon Eclipse Ti-inverted fluorescence microscope equipped with an automated stage (Prior) and a custom chamber to maintain constant 37°C temperature, high humidity, and 5% CO₂. Double strand breaks were induced with neocarzinostatin as described to induce the p53 stress response. Multiple positions were analyzed per experiment with images acquired every 20 min for 24 h using a YFP filter set (Chroma) (488-512 nm excitation filter, 520 nm dichroic beam splitter, and 532-554 nm emission filter) and TRITC filter set (Chroma) (540-580 nm excitation filter, 585 nm dichroic beam splitter, and 593-668 nm emission filter). Images were collected using a 20X CFI Plan Apochromat Lambda (NA =0.75) objective (Nikon). For each condition at least 50 individual cells were tracked. Following imaging, data were exported as individual tiffs for each channel, position, and time point.

γH2AX Immunofluorescence

RPE1-hTERT cells were plated at 25,000 cells/well into Nunc Lab-Tek 4-well chamber slides (ThermoFisher, 177437) and allowed to adhere for 24 h. For siRNA knockdowns, control cells received 25 nM non-targeting siRNA (Dharmacon) at the time of plating. BRCA1 and BRCA2 knockdowns received 25 nM siRNA against each target at the time of plating. For SCR7 and rucaparib treatments, cells received 10 μM of the inhibitor 2-hours prior to induction of DSBs using 400 ng/ml NCS. Media was aspirated, cells were washed with 1X PBS and fixed at 1,4, and 24 h post-NCS treatment, as well as a 0-hour control using 4% PFA for 15 min at room temperature. Following fixation, cells were washed three times in 1X PBS then permeabilized and blocked for 1 h using 0.25% Triton-X and 5% Goat Serum in 1X PBS. After blocking, cells were incubated overnight in a 1:200 dilution of the mouse monoclonal antibody clone JBW301 against γH2AX (Millipore, 05-636) in 0.25% Triton/5% BSA/1X PBS. Cells were then washed 3 times in 1X PBS and incubated with Alexafluor-488 Goat anti-Mouse secondary 1:500 (A28175, ThermoFisher) for 1 h in 0.25% Triton/5% BSA/1X PBS. Nuclei were stained using Hoescht 33342 (ThermoFisher, H3570) 1:10000 for 10 min in 1X PBS. Cells were then washed 3 times in 1X PBS and imaged using a Nikon Eclipse Ti-inverted fluorescence microscope equipped with an automated stage (Prior). Images were collected using a 20X CFI Plan Apochromat Lambda (NA = 0.75) objective (Nikon) from multiple positions.

Western blot Analysis

Cells were cultured in either 35-mm or 60-mm dishes for western blot analysis. Culture medium was removed from cells by aspiration, and cells were washed once with 1X PBS. Cells were then scraped into 1 mL of 1X PBS, and the plate was rinsed with an additional 1 mL of PBS added to the collected cells. Cells were pelleted by centrifugation for 10 min at 13,000 rpm and flash frozen in a dry ice:EtOH bath. Cell pellets were lysed in lysis buffer (50 mM Tris pH 7.5, 100 mM NaCl, 0.5% sodium deoxycholate, 1% Triton-X, 0.1% SDS) supplemented with 50 mM NaF, 1 mM PMSF, and 1:100 phosphatase inhibitor I (Sigma, P2850). Lysates were incubated on ice for 30 min and insoluble debris pelleted by centrifugation. Protein concentration was determined by Bradford assay (Biorad, 5000006). 20 μg of total lysate were run per lane and separated using 4%–20% Tris-Glycine gels (Biorad, 5671095)). Proteins were transferred to Immobilon-FL PVDF membrane (Millipore, IPFL00010). Membrane sections were blocked for 1 h in Li-Cor Blocking Buffer (Li-Cor, 927-4000) and incubated with primary antibodies overnight. Membranes were washed 3X with 0.1% PBST before incubation with IrDye secondary antibodies (1:5000) for 1 h. Membranes were washed three times with PBST and imaged using an Odyssey Imager.

All primary antibodies were used at a concentration of 1 μg/ml unless otherwise indicated. The following antibodies were used: mouse monoclonal anti-p53 DO-1 (Santa Cruz Biotechnology, sc-126) [1:1,000], mouse monoclonal anti-MDM2 SMP14 (Santa Cruz, sc-965) [1:500], rabbit polyclonal anti-WIP1 H-300 (Santa-Cruz,sc-20712), rabbit polyclonal anti- phospho-DNA-PKcs S2056 (Abcam,ab18192) [1:1000], mouse monoclonal anti-DNA-PKcs (Abcam, ab44815)[1:200], rabbit monoclonal anti-phospho-ATM S1981 (Cell Signaling Technology, 5883)[1:1000], rabbit monoclonal anti-phospho-Chk2 T68 (Cell Signaling Technology, 2197)[1:1000], rabbit monoclonal anti-β-Actin (Cell Signaling Technology, 4970)[1:3000], mouse monoclonal against γH2AX (Millipore, 05-636)[1:200]. Secondary antibodies used were anti-Mouse 680 RD and anti-Rabbit 800 CW (Li-Cor) and were used at a dilution of 1:5000.

qRT-PCR

Prior to each experiment, cells were plated onto 35 mm dishes. Once cells reached 70%–80% confluence, they were pre-treated for 2 h with either 10 μM rucaparib or a DMSO vehicle control. After 2 h, DSBs were induced by the addition of either a low dose (10ng/ml) or high dose (400 ng/ml) of NCS. After DSB induction, cells were collected every hour by aspiration of media and scraping of dishes. Cells were spun down and pelleted. After collecting cell pellets, mRNA was isolated using the QIAGEN RNeasy plus kit (QIAGEN, 74134) according to manufacture specifications. RNA concentration was determined by Nanodrop (ThermoFisher) and cDNA were prepared using the High-Capacity cDNA Reverse Transcription kit in 20 μl reactions (ThermoFisher, 4374966) according to manufacturer specifications. After cDNA generation, final volume was brought to 100 μl. RT-PCR was performed in 10 μl reactions containing 5 μl Maxima SYBR green master mix (ThermoFisher, K0222), 2 μl H₂O, 2 μl cDNA, and 1 μl of 10 μM specific primers. For each individual biological replicate, we ran three plates of qRT-PCR reactions each examining seven p53 targets plus measuring GAPDH as a normalization control. Primers used for gene expression of p53 targets were previously described (Porter et al., 2016).

Cell Cycle Analysis

2.5x10⁵ RPE1-hTERT cells were plated in 6 cm plates and allowed to adhere for 24 h. Cells then received a two-hour pre-treatment with either 10 μM rucaparib or DMSO vehicle control. After 2 h, DSBs were induced by the addition of either a low dose (10ng/ml) or high dose (400ng/ml) of NCS. Cells were allowed to grow for either 24 or 48 h. At this time, media was removed and placed into a separate tube. Cells were removed from plate by trypsinization. Once detached, cells were rinsed in media and placed into the same tube as the original growth media. Cells were pelleted by 400x g spin for 4 min and washed twice in PBS. Cells were then resuspended in 300 μl PBS and passed through a 45 μM cell strainer to remove clumps. Cells were then fixed by the addition of 700 μL cold EtOH dropwise while vortexing. Fixed cells were stored at −20°C until experiment. Cells were prepared for analysis by pelleting cells with a 750x g spin for 5 min and two washes in 1 mL PBS. 1x10⁶ cells were then resuspended in a staining solution containing 20 μg/ml propidium iodide and 0.5 mg/ml RNaseA. Cells were incubated at 37°C for 30 min, then brought to 500 μl final volume and passed through a 45 μM mesh. Cells were analyzed using a Sony SA3800 Spectral Analyzer in the Flow Cytometry core facility (NIH Center for Cancer Research, Laboratory of Genome Integrity).

SA-β-Gal Senescence Analysis

1x10⁵ RPE1-hTERT cells were plated on gridded 35 mm Mattek glass-bottom dishes. After cells adhered, they were pre-treated for 2 h with 10 μM rucaparib or a DMSO control. After 2 h, cells received either a low (10 ng/ml) or high (400 ng/ml) dose of NCS to induced DSBs. Control cells received no treatment. After 4 days, cells were fixed and stained for SA-β-Gal activity using the Cellular Senescence Assay kit (EMD Millipore, KAA002) according to manufacturer specifications. After staining, cells were imaged in phase contrast using 20X magnification on a Nikon Eclipse Ti-inverted microscope to quantify blue cells.

Cellular Viability Assay

2x10³ cells were plated into wells of a 96-well plate and pre-treated with either 10 μM rucaparib or a DMSO control. After 2 h, cells received an additional treatment with either low (10 ng/ml) or high (400 ng/ml) NCS dose. At this time point, adherent cells were stained with the Reliablue viability dye (ATCC,30-1014) for 1 hour according to manufacturer specifications. After incubation, wells were measured for fluorescence intensity using a microplate reader (Tecan GENios) with a 550nm excitation and 595 nm emission detection. This time point served as the 0-hour time point. After collection of the 0-hour data, subsequent time points were measured every 24 h for 96 h total. For MCF7 growth curves the alamarBlue high sensitivity assay was used (ThermoFisher, A50100) as Reliablue was discontinued. 1x10⁴ cells were plated into wells of a 96-well plate and received the same treatment regimen as RPE1-hTERT cells. Then cells were stained with alamarBlue for 1 h according to manufacturer specifications. After incubation wells were measured for fluorescence intensity using a microplate reader (Biotek SynergyH1) with 560 nm excitation and 590 nm emission detection.

Hierarchical Clustering

Clustering of gene expression was performed using min-max normalized expression data collected as described in the qRT-PCR methods. Clusters were generated using MATLAB. First, the pairwise Euclidean distance between each gene was measured.using the pdist function in the Statistics and Machine Learning toolbox. Then, genes were clustered based on shortest distance using the linkage function in the Statistics and Machine Learning toolbox. Lastly, dendrograms were generated using the dendrogram function. Once clusters were generated, each gene was color coded based on the dynamic classification previously determined (Porter et al., 2016).

QUANTIFICATION AND STATISTICAL ANALYSIS

Quantification of Nuclear p53-mVenus

Using the exported H2B-mRuby2 images, nuclei were segmented using CellProfiler software (v.3.1.9) to generate a grayscale mask identifying the nuclei of all cells using the IdentifyPrimaryObjects function. Cells were automatically tracked using the Cell Profiler TrackObjects function using the overlap method with a 50 pixel maximum distance to consider a match. Tracked cells were confirmed by manual visualization. Nuclear masks were used to define regions of interest and mean fluorescence intensity for p53-mVenus images was determined in an automated manner using custom MATLAB scripts for these regions. In instances where automatic tracking failed to provide sufficient cell numbers, regions of interest were defined by manually defining nuclei locations using polygons. Mean fluorescence intensity was calculated for these regions using the same MATLAB scripts. For plotting, mean fluorescence intensity was smoothed using a moving mean average with a three-window span (n−1, n, n+1). These analysis tools and pipelines are available upon request. n, represents the number of individual cells analyzed for each plot and are discussed in figure legends. For all single cell analyses a minimum of 50 single cells were analyzed.

Quantification of p53 Dynamic Features

Quantification of pulse number, first peak amplitude, first peak timing, and duration of the first peak was performed in an automated manner using custom MATLAB software. Mean fluorescence intensity for individual cells was first smoothed using a moving mean average with a three-window span (n−1, n, n+1). After data smoothing, peaks were identified using the findpeaks function in the MATLAB Signal Processing toolbox to identify local maxima. To minimize spurious identification of peaks due to noise in expression, we defined that peaks must be a minimum of 9 frames apart. A minimum peak prominence was also defined based on basal and peak expression to ensure minor fluctuations in p53 levels were not identified as pulses. Using these identified maxima, we could further quantify amplitude, timing, and duration based on the timing of these maxima. Based on these criteria, cells exhibited between 1-6 pulses consistent with prior studies of p53 dynamics in both RPE1-hTERT cells and other cell lines (Geva-Zatorsky et al., 2006; Lahav et al., 2004; Loewer et al., 2010; Stewart-Ornstein and Lahav, 2017). For analysis of p53 dynamic features in response to both rucaparib and NU7441, peak identification function was modified to calculate duration based on maximum prominence (peak expression-local minima). This allowed for more accurate quantification of cells that showed prolonged accumulation (ex. rucaparib+3 μM NU7441). Cells that exhibited increased p53 expression throughout the duration of the experiments were assigned a pulse duration of 24 h. Data for pulse number represent the average pulse numbers over 24 h ± SEM. Data for peak amplitude, timing, and duration is presented using boxplots to show the spread of population data. All statistics were performed using one-way ANOVA for multiple comparisons with a p value < 0.05 determined as significant. n, represents the number of individual cells analyzed and is reflected in the figures and figure legends.

γH2AX Immunofluorescence

Nuclei images in the DAPI channel were segmented using the thresholding feature in the NIS Elements software (Nikon) to define regions of interest for each time point and condition. Mean fluorescence intensity for γH2AX was measured for each object using NIS Elements. For each data point a minimum of 100 cells were analyzed. Data presented as the mean fluorescence intensity ± SEM.

qRT-PCR Quantification

For quantification, we normalized expression of each p53 target to its on-plate GAPDH expression control and quantified fold change by ΔΔCt, comparing each time point to the expression at time zero. n, discussed in figure legends indicates the number of independent biological replicates for each condition and time point. Data presented as mean fold change ± SEM. For maximum fold change, the peak fold change was plotted for each individual target. Statistics were performed using unpaired t tests to compare identical NCS doses with or without rucaparib treatment. Significance was defined as p < 0.05. For min-max normalization, we normalized each target according to the formula.

$$MinMa{x}_n=\frac{Expressio{n}_n-\text{min}(Expression)}{\max (Expression)-\text{min}(Expression)}$$

Western Blot Analysis

For quantification of western blots, images were analyzed using ImageJ software (NIH) and all western blots normalized to β-actin levels from the same membrane. Expression for each target was then normalized to expression at time zero to quantify relative expression over each time course. n, discussed in figure legends represents the number of biological replicates for each experiment. Data were plotted as the average relative expression ± SEM.

Cell Cycle Analysis

Single cells were identified from the Sony SA3800 spectral analyzer data based on forward and side scatter measurements. Cell cycle distributions for cells in G1,S, or G2/M phases was performed using FlowJo v.10 and the Watson Pragmatic algorithm. Ratio of S and G2/M ratios were calculated by dividing the percentage of S phase cells by the percentage of G2/M cells for each individual experiment. Statistical analysis was performed by one-way ANOVA for multiple comparisons. Significance was defined as p < 0.05. n, discussed in figure legends represents the number of independent biological replicates performed for each experiment. Data for cell cycle distributions presented as the mean ± SEM.

Cellular Viability Assay

Fluorescence intensity values were collected for each time point and condition. Statistical analysis was performed using unpaired t tests to compare identical NCS doses with or without rucaparib treatment. Significance was defined as p < 0.05. n, discussed in the figure legends represents the number of independent biological replicates measured for each time point and condition. Data plotted as the mean fluorescence intensity ± SEM.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse monoclonal anti-p53 DO-1	Santa Cruz Biotechnology	Cat# sc-126, RRID: AB_628082
Rabbit monoclonal anti-β-actin 13E5	Cell Signaling Technology	Cat# 4970, RRID: AB_2223172
Rabbit polyclonal anti-WIP1 H-300	Santa Cruz Biotechnology	Cat# sc-20712, RRID: AB_2169909
Rabbit polyclonal anti- phospho-DNA-PKcs S2056	Abcam	Cat# ab18192, RRID: AB_869495
Mouse monoclonal anti-DNA-PKcs	Abcam	Cat# ab44815, RRID: AB_731982
Rabbit monoclonal anti-phospho-ATM (Ser1981) D6H9	Cell Signaling Technology	Cat# 5883, RRID: AB_10835213
Rabbit monoclonal anti-phospho-Chk2 (Thr68) C13C1	Cell Signaling Technology	Cat# 2197, RRID: AB_2080501
Mouse monoclonal anti-γH2AX JBW301	Millipore	Cat# 05-636, RRID: AB_309864
Mouse monoclonal anti-MDM2 SMP14	Santa Cruz Biotechnology	Cat# sc-965, RRID: AB_627920
Donkey anti-Mouse 680RD	LI-COR Biosciences	Cat# 926-68072, RRID: AB_10953628
Goat anti-Mouse AlexaFluor-488	ThermoFisher	Cat# A28175, RRID: AB_2536161
Donkey anti-Rabbit 800CW	LI-COR Biosciences	Cat# 926-32213, RRID: AB_621848
Chemicals, Peptides, and Recombinant Proteins
Neocarzinostatin (NCS)	Sigma-Aldrich	Cat# N9162
Rucaparib	Selleckchem	Cat# S1098
Olaparib	Selleckchem	Cat# S1060
Niraparib	Selleckchem	Cat# S2741
SCR7	Selleckchem	Cat# S7742
NU7441	Selleckchem	Cat# S2638
Phosphatase Inhibitor Cocktail I	Sigma-Aldrich	Cat# P2850
DMEM/F12 media	Fisher Scientific	Cat# SH3002301
RPMI media	Fisher Scientific	Cat# SH3002701
Propidium Iodide	Sigma-Aldrich	Cat# P4864
RNase A	ThermoFisher	Cat# EN0531
Opti-MEM I Reduced Serum Media	ThermoFisher	Cat# 31985070
Maxima SYBR green/ROX qPCR Master Mix (2X)	ThermoFisher	Cat# K0222
Lenti-X Concentrator	Takara	Cat# 631231
DharmaFect I	Horizon	Cat# T-2001-02
DMEM/F12, no phenol red	ThermoFisher	Cat# 21041025
RPMI 1640 medium, no phenol red	ThermoFisher	Cat# 11835030
Hoechst 33342	ThermoFisher	Cat# H3570
Protein Assay Dye	Biorad	Cat# 5000006
Li-Cor Blocking Buffer	LI-COR Biosciences	Cat# 927-4000
Critical Commercial Assays
Lenti-X Packaging Single Shot (VSV-G)	Takara	Cat# 631275
Lenti-X GoStix Plus	Takara	Cat# 631280
RNeasy Plus mini kit	QIAGEN	Cat# 74134
High Capacity cDNA Reverse Transcription kit	ThermoFisher	Cat# 4374966
Cellular Senescence Assay	EMD Millipore	Cat# KAA002
Reliablue Viability Dye	ATCC	Cat# 30-1014
AlamarBlue HS Cell Viability Reagent	ThermoFisher	Cat# A50100
Deposited Data
Single Cell p53-mVenus Expression Data	This Paper	Mendeley Data (“Hanson and Batchelor 2020 Cell Reports,” Mendeley Data, V1, https://doi.org/10.17632/c9fvknxdjf.1)
Experimental Models: Cell Lines
RPE1-hTERT	ATCC	Cat# CRL-4000
RPE1-hTERT H2B-mRuby2 p53-mVenus	This Paper	N/A
MCF7	ATCC	Cat# HTB-22
MCF7 p53-mVenus	Batchelor et al. (2008)	N/A
HEK293T	ATCC	Cat# CRL-3216
Oligonucleotides
siGenome BRCA1 siRNA #1	Horizon	Cat# D-003461-05
siGenome BRCA1 siRNA #2	Horizon	Cat# D-003461-06
siGenome BRCA2 siRNA #1	Horizon	Cat# D-003462-01
siGenome BRCA2 siRNA #2	Horizon	Cat# D-003462-02
OnTARGET Plus Human TP53 siRNA SMARTPool	Horizon	Cat# L-003329-00
OnTARGET Plus non-targeting control pool	Horizon	Cat# D-001810-10
Recombinant DNA
pLentiPGK Hygro DEST H2B-mRuby2	Addgene; Kudo et al. (2018)	Cat# 90236
pDESTN-pMT-p53-Venus	Batchelor et al. (2008)	N/A
Software and Algorithms
MATLAB (2020a)	Mathworks	https://www.mathworks.com/products/matlab.html, RRID:SCR_001622
ImageJ	NIH	https://imagej.nih.gov/ij/, RRID:SCR_003070
FlowJo v.10	BD Biosciences	https://www.flowjo.com/solutions/flowjo, RRID:SCR_008520
NIS Elements Advanced Research	Nikon	https://www.microscope.healthcare.mkon.com/products/software/nis-elements, RRID:SCR_014329
Other
4-20% Tris Glycine gels	Biorad	Cat# 5671095
Immobilon FL PVDF membrane	Millipore	Cat# IPFL00010
Mattek 12-well, No 1.5 plates	Mattek	Cat# P12G-1.5-14-F
Nunc Lab-tek 4-well chamber slides	ThermoFisher	Cat# 177437

Highlights.

• Disruption of distinct DNA repair pathways differentially alters p53 dynamics

• Rucaparib treatment elongates the amplitude and duration of the first p53 pulse

• Prolonged p53 expression deregulates gene expression for multiple target pathways

• Rucaparib treatment enhances DNA-damage-induced growth arrest