RP-3500: A Novel, Potent, and Selective ATR Inhibitor that is Effective in Preclinical Models as a Monotherapy and in Combination with PARP Inhibitors

This article is featured in Highlights of This Issue, p. 243

Many cancers harbor defects in DNA damage response (DDR) pathways leading to incomplete DNA repair, circumvention of DNA damage checkpoints, and/or the dependence of tumor cells on alternative DNA repair mechanisms. Tumors bearing DNA repair defects therefore possess important vulnerabilities that can be exploited for therapeutic intervention (1).

Ataxia telangiectasia and Rad3-related (ATR) kinase is an important therapeutic target due to its pivotal role in DNA replication and repair. It is a member of the phosphatidylinositol 3′ kinase-related kinase (PIKK) family which also includes the ataxia telangiectasia mutated (ATM) and DNA-dependent protein kinase catalytic subunit (DNA-PKcs) repair kinases. While ATM and DNA-dependent protein kinase (DNA-PK) repair double-strand breaks (DSB) generated from radiation/chemotherapy or collapsed DNA replication forks, the primary role of ATR is to regulate DNA replication and cell cycle checkpoint activation to ensure complete and intact cell division (2). ATR, with its regulatory partner ATR interacting protein (ATRIP), are recruited to blocked or damaged DNA replication forks. ATR protects stalled replication forks through helicases that remodel their structure to prevent cleavage and then promotes fork repair via homologous recombination (HR) factors such as BRCA2 and partner and localizer of BRCA2 (PALB2) (3). ATR also acts by phosphorylating checkpoint kinase 1 (CHK1) on Ser-345 (pCHK1), causing its activation. Once activated, pCHK1 mediates cell cycle arrest in S or G₂–M phases (2) allowing DNA repair prior to cell cycle progression. Following ATR inhibition, replication stress is increased by compromised fork repair and accumulation of dormant origins of replication. This leads to massive formation of DSBs, activation of ATM and DNA-PKcs followed by phosphorylation of their substrates KRAB-associated protein 1 (KAP1) and H2AX, and irreversible collapse of DNA replication known as replication catastrophe (3–5). ATR inhibition also impairs cell cycle checkpoint control, which prematurely forces cells into mitosis in the presence of DNA damage and an incompletely replicated genome, resulting in chromosome fragmentation and mitotic cell death (6).

Several orally available ATR inhibitors (ATRi) including AZD6738 (7, 8), M4344 (9), BAY1895344 (10, 11), M1774 (12), ATR0380 (NCT04657068), and ATRN-119 (NCT04905914) are currently undergoing clinical evaluation and have tolerability profiles as single agents with toxicities including myelosuppression, fatigue, and gastrointestinal effects (13). Clinical single-agent ATRi efficacy has been reported for BAY1895344 and AZD6738 in a small number of patients with selected DDR deficiencies including BRCA1/2 mutations, ATM loss of function, or tumors with unreported genomic alterations, suggesting an early proof of concept for this drug class (8, 14, 15).

Given the role of ATR during DNA replication, it is not surprising that upon ATR inhibition cells rely on genes such as ATM and BRCA1/2 to mediate DSB repair caused by collapsed replication forks and loss of HR (16–21). In addition, several recent studies have demonstrated synergy between ATRi and PARP inhibitors (PARPi) in ATM and BRCA1/2-deficient preclinical models (22–26).

Synergy between ATRi and PARPi has been widely reported and is a promising clinical strategy; however, continuous combination treatment has been poorly tolerated preclinically (22–24) and clinically due to myelosuppression (13, 27). Inhibition of PARP activity not only impairs single-strand DNA break repair but also stabilizes PARP-DNA complexes (PARP trapping), leading to replication fork stalling and collapse (28). Both effects cause the accumulation of DSBs during DNA replication and a dependency on HR for repair. Furthermore, BRCA1 and BRCA2 mutant cells that have become resistant to PARP inhibition by bypassing the replication fork protection roles of these HR proteins, are also dependent on ATR for essential DNA repair functions (22, 25).

ATR inhibition in the presence of PARPi results in premature entry into mitosis, leading to mitotic catastrophe and cell death (26, 29, 30). In preclinical cancer models, ATRi and PARPi combinations demonstrated synergistic enhancement of cell death and tumor regression (10, 22, 25, 26); however, tolerability was limited by both the dose level and duration of administration. Clinically, combining an ATRi and PARPi leads to overlapping toxicities, including myelosuppression, limiting tolerability (30). Additional work is needed to establish dose scheduling to optimize these combinations for clinical use.

This is the first disclosure of the structure of RP-3500, a novel, oral ATRi currently under clinical evaluation (NCT04497116). RP-3500 is highly selective and inhibits ATR at picomolar to nanomolar concentrations in biochemical and cell-based assays. RP-3500 is highly efficacious as a monotherapy in several preclinical models that are BRCA1-deficient or that have deficits of ATM pathway function. Here, we demonstrate that the therapeutic index of RP-3500 is widened when administered on an intermittent dosing schedule to mitigate on-target hematologic toxicity associated with continuous dosing. Furthermore, the benefits of intermittent dosing extend to PARPi combinations where attenuated doses of RP-3500 with either niraparib or olaparib administered concomitantly were substantially more efficacious and better tolerated than sequential administration of the combination.

RP-3500 (1R,3R,5S)-3–6-[(3R)-3-methylmorpholin-4-yl]-1-(1H-pyrazol-3-yl)-1H-pyrazolo[3,4-b]pyridin-4-yl-8-oxabicyclo[3.2.1]octan- 3-ol was prepared as described for Example 121 of WO2020/087170 (Supplementary Methods; ref. 31). Gemcitabine, bleomycin (Sigma, #G6423, #B5507), niraparib, and olaparib (MedChem Express, #HY-10619, #HY-10162) were purchased.

Capan-1, CW-2, NCI-H23, LoVo, MDA-MB-436, SNU-601, and HT-29 cells were obtained from ATCC, Granta-519 cells from DSMZ, and SUM149PT from Asterand in 2018, and were not further authenticated. RPE1-hTERT TP53KOATM—wild-type (WT) and ATM-knockout (KO) cells were made as described in the Supplementary Methods. Cells were cultured as recommended and were verified Mycoplasma and murine pathogen free, quarterly by PCR (internally and at Idexx Bioanalytics). Cells were cultured up to 10 weeks and were used at passages of less than 20 from the initial source vial.

Cells were plated 24 hours before treatment at optimal densities for each line (200–2,500 cells per well) in 96-well plates (Corning, #3903), then treated with compound for 120 hours before determining cell viability using CellTiter-Glo (Promega, # G7571). Luminescence was read on either Envision 2105 (Perkin-Elmer), SpectraMax M3, or FlexStation 3 (Molecular Devices) plate readers. Cell viability was calculated by normalizing each blank-corrected luminescence value to untreated cells. Synergy between RP-3500 and PARPi was determined with SynergyFinder (https://synergyfinder.fimm.fi) using the Zero interaction potency (ZIP) model (32).

Cells and tumor fragments were lysed in a buffer (Mesoscale discovery, #R60TX-2) with protease and phosphatase inhibitors (Thermo Fisher Scientific, #78437; Thermo Fisher Scientific, #78420). Lysates were separated by SDS-PAGE (Bio-Rad, #5678085); then transferred to polyvinylidene fluoride membranes. Membranes were incubated with primary antibodies: anti-pCHK1 (Ser345; Cell Signaling Technology, #2348), anti-pKAP1(S824; Abcam; catalog no. ab133440), anti-KAP1 (Abcam; catalog no. 10484), anti-ATM (Novus; NB100–104SS) and anti-pDNA-PKcs (Abcam; catalog no. 124918), anti-CHK1 (Cell Signaling Technology, #2348), or anti-γH2AX (EMD Millipore; catalog no. 05–636). Signal detection was carried out with secondary anti-rabbit (Jackson Immuno Research, #711–035–152) or anti-mouse (Jackson Immuno Research, #115–035–174) horseradish peroxidase (HRP) antibody and SuperSignal West Dura Extended Duration Substrate (Thermo Fisher Scientific, #34076) using a Bio-Rad ChemiDoc Touch Imaging System (Bio-Rad, #1708370).

ATR biochemical assays used human recombinant tagged proteins ATR/ATRIP (Eurofins, Item#14–953) purified from mammalian cells. Assays were conducted using glutathione S-transferase–tagged p53 substrate (Enzo Lifesciences; catalog no. BML-FW9370–0050) and 3 μmol/L ATP in 50 mmol/L HEPES, 0.01% Brij, 0.1 mmol/L EGTA, and 1% glycerol. Phosphorylated substrate was incubated with anti-phospho-p53 (Ser15) antibody (NEB; catalog no. 9284), donor (Perkin Elmer, 6765301), and acceptor (Perkin Elmer, 670137M) beads for 4 to 5 hours in 60 mmol/L EDTA, 50 mmol/L Tris pH 7.4 with 1.0 mg/mL BSA. Phosphorylation was quantified by fluorescence measured in a BMG Polarstar plate reader (excitation 680 nm, emission 520–620 nm). Standard PIKK and related kinase assays (Kinase Profiler) were performed at Eurofins Discovery, using standard assay protocols. See Supplementary Methods.

ATR, mTOR, and PI3Kα assays were conducted in LoVo cells as described in the Supplementary Methods, using AlphaScreen and AlphaLISA SureFire assay kits carried out according to the manufacturers' instructions.

HT-29 cells were plated in 24-well plates (VWR, #10062–896) at 350,000 cells/well. The following day, RP-3500 was added for 60 minutes followed by 3 mU/mL bleomycin (Sigma, #B5507) for 2 hours. Cells were washed with PBS and lysed in MSD Lysis Buffer (Mesoscale discovery, #R60TX-2) supplemented with phosphatase and protease inhibitors (Thermo Fisher Scientific, #78437 and #78420). Phosphorylated ATM substrate pKAP1 (Ser824) was detected by immunoblotting, and DNA-PK activation measured with pDNAPKcs-S2056 (Abcam, #124918) and an HRP-coupled secondary antibody (Jackson Immuno Research, #711–035–152). Membranes, first probed to detect pKAP1 or pDNA-PK, were then stripped with Restore Plus Western Blot Stripping Buffer (Thermo Fisher Scientific, #46430) and reprobed as described above to detect total KAP1 (Abcam, #ab10484) or total DNA-PKcs (Sigma, #MABC1236).

Treated RPE1-hTERT Cas9 TP53-KO ATM WT or ATM-KO cells at the experimental endpoint, were rinsed in PBS, fixed with 4% paraformaldehyde/PBS, and permeabilized in 0.3% Triton X-100/PBS. Cells were subsequently blocked in polybenzyl glutamate (0.2% cold water fish gelatin, 0.5% BSA in PBS) and incubated with a primary mouse anti–γH2AX antibody clone JBW301 (Millipore Sigma, 05–636). Cells were rinsed and incubated with an Alexa Fluor 488-conjugated goat anti-mouse IgG secondary antibody (Invitrogen) and 0.5 mg/mL 4′,6-diamidino-2-phenylindole, and then rinsed and imaged on an Operetta automated high-content confocal microscope (Perkin Elmer). Image analysis was performed using Harmony software (Perkin Elmer). Built-in algorithms were used to quantify the mean nuclear γH2AX intensity.

All cells were implanted subcutaneously (at 5 × 10⁶ cells/mouse) into the flanks of female mice (6–8 weeks old; Charles River) in a suspension of 1:1 media:Matrigel (Matrigel Corning; catalog no. CB35248). Lovo cells were implanted into CB17 severe combined immunodeficiency (SCID) mice at Repare Therapeutics. CW-2 cells were implanted at NuChem Sciences Inc. and SUM149PT cells at Oncodesign into NOD/SCID. The above studies were conducted in Canadian Council on Animal Care–accredited vivariums with Institutional Animal Care Committee–approved protocols.

Granta-519 cells (1 × 10⁷ cells/mouse) were implanted into female NOD/SCID mice in 1:1 matrigel:PBS at Crown Biosciences Inc. Patient-derived xenograft (PDX) tumor fragments were cut into small pieces (approximately 2–3 mm in diameter) and implanted subcutaneously into female BALB/c nude mice and subsequently passaged by implantation into the efficacy cohort at Crown Biosciences Inc. The procedures involving the care and use of animals for Granta-519 and PDX studies were reviewed and approved by the Institutional Animal Care and Use Committee of Crown Biosciences Inc. prior to execution. During the study, the care and use of animals was conducted in accordance with the regulations of the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC).

Animals were monitored for tumor volume, clinical signs, and body weight three times per week. Tumor volume was measured using a digital caliper and calculated using the formula 0.52 × L × W². Mice were randomized to treatment groups when tumors reached 100 to 200 mm³. RP-3500, olaparib, and niraparib were administered in 0.5% methylcellulose/0.02% SDS vehicle. See Supplementary Methods for data analysis and statistics.

Mouse whole blood was collected at 0.5, 1, 3, 8, and 24 hours postdose by tail snip and diluted 1:3 with 0.1 mol/L citrate buffer. Plasma was collected from whole blood drawn by cardiac puncture at 1 and 3 hours postdose into tripotassium (K₃) EDTA tubes (Sarstedt; catalog no. 41.1504.105), then centrifuged at 12,000 relative centrifugal force for 10 minutes at 4°C. The concentration of RP-3500 in whole blood and plasma was determined by high-performance LC/MS (see Supplementary Methods). Excised tumors were flash-frozen for protein extract or preserved in 10% formalin. Frozen fragments were homogenized in lysis buffer (Mesoscale discovery, #R60TX-2) with protease and phosphatase inhibitors (Thermo Fisher Scientific, #78437; Thermo Fisher Scientific, #78420) using 2.8-mm ceramic bead-containing tubes (OMNI, #19–628) and a Bead Ruptor 24 (OMNI International). The levels of pCHK1(Ser345), total CHK1, pKAP1(Ser824), total KAP1, and pDNA-PKcs (Ser2056) were determined by immunoblotting relative to untreated controls as described above. The LoVo tumor IC₈₀ for pCHK1(Ser345) was determined by a nonlinear least-squares regression to a normalized dose-response four-parameter fit model (GraphPad Prism v9).

At the designated endpoint, whole blood was collected from anesthetized mice into K₃-EDTA BD Microtainer MAP Microtubes (Thermo Fisher Scientific, 22–253–270) by cardiac puncture. Hematology analysis was performed with a Sysmex XN-1000^TM hematology Analyzer (Sysmex America Inc.).

To determine the potency of RP-3500 (Fig. 1A), a biochemical assay for ATR/ATRIP inhibition was established using tagged human ATR and ATRIP purified from mammalian cells, with p53 (Ser15) as the phosphorylation substrate. In this assay, the Km for ATP was 30 nmol/L and biochemical assays were conducted at 3 μmol/L ATP where the IC₅₀ of RP-3500 was 1.00 ± 0.40 nmol/L (Table 1) and the Ki was 0.022 ± 0.002 nmol/L. In a LoVo cell-based assay, the IC₅₀ of RP-3500 to inhibit gemcitabine-stimulated ATR phosphorylation of its substrate pCHK1(Ser345) was 0.33 nmol/L (Table 1). The potency in biochemical assays is consistent with low nanomolar growth inhibitory activity (IC₅₀ = 8–64 nmol/L) in cell viability assays in a panel of cell lines with defective ATM or BRCA1/2 function (Fig. 1B; Supplementary Table S1). These results demonstrate that RP-3500 is a potent ATRi in both biochemical and cell-based assays.

To determine selectivity within the kinome, RP-3500 was incubated with a Jurkat whole-cell lysate, and the ability of an ATP-binding pocket probe to displace RP-3500 was determined by mass spectrometry (Supplementary Fig. S1A and S1B; ref. 33). RP-3500 demonstrated high selectivity for ATR. At 3 nmol/L, RP-3500 remained bound to 82.8% of ATR in the lysate but not to any of the other evaluable kinases. At 300 nmol/L, RP-3500 showed no binding activity to more than 300 kinases outside of the PIKK family. Weak binding was detected to five PIKK or PI3K family members. To confirm RP-3500 selectivity, specific cell-based kinase assays were developed to determine the intracellular inhibitory activity within the PIKK family. In each assay, RP-3500 showed weak inhibition of PIKK family cellular kinase activity resulting in 30-fold selectivity over mTOR and more than 2,000-fold selectivity over ATM, DNA-PK, and PI3Kα kinases (Table 1).

To characterize the DDR to RP-3500–mediated ATR inhibition, LoVo and CW-2 human colon cancer cell lines were treated with 1 μmol/L of RP-3500 for various durations up to 24 hours, the dose being chosen to inhibit more than 90% of endogenous CHK1 phosphorylation. RP-3500 inhibits CHK1(Ser345) phosphorylation from 1 to 3 hours after treatment initiation. However, starting at 4 hours, CHK1(Ser345) became rephosphorylated as DNA-PKcs became activated in treated cells, along with its substrates KAP1 and H2AX (Fig. 1C), demonstrating induction of DSB DNA damage (34).

RP-3500 showed single agent activity in several xenograft models of gastric and colon cancer with dysfunctional ATM pathways. LoVo cells express WT ATM but have an MRE11A splicing deficiency (functionally ATM-compromised; ref. 35), while CW-2 cells have an ATM missense mutation (A2843V; ref. 36) resulting in low ATM expression (Fig. 1C). Administration of RP-3500 once daily to mice bearing LoVo xenografts produced dose-dependent tumor growth inhibition with an MED of 7 mg/kg (Fig. 2A). Similarly, administration of RP-3500 at a dose of 5 and 10 mg/kg once daily produced statistically significant tumor growth inhibition in the CW-2 colon xenograft model (Supplementary Fig. S2A). Despite tumor-induced cachexia in both models, RP-3500 doses of 3 to 10 mg/kg daily were well tolerated with minimal additional body weight loss relative to vehicle control (Supplementary Fig. S2B and S2C). At 15 mg/kg, 1 mouse was sacrificed due to body weight loss (after day 12) and the remaining mice experienced more than 15% body weight loss. Therefore, the MTD was 10 mg/kg once daily on a continuous dosing schedule. In a PDX model of gastric cancer with ATM protein loss (Supplementary Fig. S2D), RP-3500 was highly efficacious, eliciting complete tumor regressions at 5 and 10 mg/kg daily (Fig. 2B) without body weight loss at either dose level (Supplementary Fig. S2E). These results demonstrate a low single-agent MED of RP-3500 (5–7 mg/kg daily) in multiple tumor models (Supplementary Table S2), with potential for complete regression.

To verify target engagement in vivo and establish a pharmacokinetic (PK)/pharmacodynamic (PD) marker relationship, plasma and tumors were sampled from the LoVo tumor model at 1 and 3 hours posttreatment and analyzed for circulating levels of RP-3500 and tumor pCHK1 by immunoblotting. A clear proportional relationship between pCHK1 inhibition and circulating free plasma levels of RP-3500 was demonstrated with an estimated tumor target engagement IC₈₀ of 18.6 nmol/L [85% confidence interval (CI), 7.23–145 nmol/L] (Fig. 2C).

To determine the exposure levels needed to achieve minimum efficacy, the PK profile of RP-3500 was measured at steady state at several doses (Fig. 2D). The circulating free plasma levels of RP-3500 at the MED range of 5 to 7 mg/kg support the idea that an RP-3500 free concentration above the in vivo tumor IC₈₀ for 10 to 12 hours is sufficient for efficacy. This is considered a conservative estimate based on the modest RP-3500 sensitivity of the LoVo xenograft relative to other cell line tumor models (Fig. 1B).

DNA DSBs induced by ATR inhibition cause DNA damage marked by phosphorylated H2AX (γH2AX) and KAP1 (pKAP1; ref. 2), and are proposed as useful clinical pharmacodynamic biomarkers for ATRi (12, 14, 37). Mice bearing LoVo tumors were treated for 7 days, and tumor tissue harvested 24 hours after the last RP-3500 treatment, consistent with the timing of the DDR seen in Fig. 1C. Dose-dependent KAP1 and DNA-PKcs phosphorylation was detected by immunoblotting resulting in 8.1- and 2.7-fold inductions, respectively, over background levels at 7 mg/kg RP-3500 (Fig. 2E; Supplementary Fig. S2F). Similarly, gastric PDX tumor tissues with biallelic ATM loss sampled 24 hours after three daily treatments showed dose-proportional IHC staining of γH2AX and pKAP1 (Fig. 2F). Quantitation of the signal in tumor tissue indicated a 2.8- and 4.4-fold induction for γH2AX and a 6.7- and 18.4-fold induction for pKAP1 over background with 5 and 10 mg/kg RP-3500, respectively, confirming the induction of DNA damage and reinforcing the proposed use of these molecular markers of tumor response in early clinical trials.

To optimize an in vivo RP-3500 administration schedule, washout experiments were conducted in vitro to determine the exposure required to generate irreversible DNA damage. These experiments were performed in an engineered isogenic cell line pair (RPE1-hTERT Cas9 TP53-KO ATM-WT and ATM-KO) where the sensitivity of the ATM-KO cell line was six-fold greater than the ATM-WT line (Supplementary Fig. S3A and S3B). Exposure to 30 nmol/L RP-3500 for 48 and 72 hours increased γH2AX staining to similar levels in both the ATM-WT and ATM-KO cells. However, when compound was removed after a 72-hour exposure, the ATM-WT cells mostly recovered from γH2AX–marked DNA damage after 24 hours, whereas the cells lacking functional ATM had sustained DNA damage for at least 48 hours (Fig. 3A and B). Consistent with these results, treatment of Granta-519 cells (ATM LOF mutation) with RP-3500 for longer than 2 to 3 days did not result in significant lowering of the IC₅₀. Together, these results suggest that short-term exposure to RP-3500 is sufficient to cause irreparable DNA damage and induce growth inhibition (Supplementary Fig. S3C and S3D).

To examine the effects of intermittent RP-3500 administration on tolerability and efficacy, the MTD was determined for weekly treatment schedules of 2 days on/5 days off, 3 days on/4 days off, and 5 days on/2 days off. The daily MTDs for the three schedules were 30, 30, and 25 mg/kg, respectively, demonstrating that increased dose levels of RP-3500 are tolerated with intermittent schedules compared with continuous daily administration (MTD of 10 mg/kg; Supplementary Fig. S3E). Firstly, three daily doses weekly is sufficient to maximize antitumor efficacy of the 10 mg/kg dose, consistent with in vitro results (Supplementary Fig. S3F and S3G). Secondly, a more profound antitumor effect occurred at higher doses on the 3 days on/4 days off and 5 days on/2 days off schedules compared with consecutive daily administrations at a lower dose (Fig. 3C; Supplementary Table S2). This suggests that a dose of RP-3500 (30 mg/kg) providing IC₈₀ target inhibition for 24 hours/day for 3 days is more efficacious than a dose (10 mg/kg) providing target inhibition approximately 16 hours/day for 7 days (Fig. 3C; Supplementary Fig. S3H).

Anemia is a known adverse effect of ATRi, observed in both the preclinical and clinical settings (8, 14, 38). Therefore, hematologic toxicity was evaluated in mice after 2 weeks of RP-3500 treatment as a more sensitive measure of tolerability than body weight loss. Although body weight loss was somewhat increased for the longer-duration weekly dosing schedules (Supplementary Fig. S3E), these schedules (5 and 7 days on per week) also resulted in increasingly dramatic reductions in red blood cell (RBC) count (Fig. 3D). Intermittent dosing schedules with a dose holiday of at least 4 consecutive days was sufficient to allow reticulocyte regeneration above the vehicle control values (Fig. 3E), demonstrating that erythroid toxicity resulting from sustained target inhibition over 3 days is reversible given a sufficient recovery time. These results imply that intermittent dose holidays increase the therapeutic index and support the clinical evaluation of intermittent administration schedules in monotherapy settings.

To investigate whether intermittent schedules of RP-3500 and PARPi combination therapy could provide the same tolerability and efficacy benefits of single-agent intermittent dosing, synergy was first examined in vitro. The SUM149PT BRCA1-mutated breast carcinoma cell line (39) was exposed to a 3-day concomitant or 3-day sequential combination of RP-3500 and olaparib with endpoint viability measured at day 8 (Fig. 4A). The overall combined effects of RP-3500 and olaparib were more synergistic (maximum ZIP synergy score >10) when administered concomitantly compared with sequential olaparib treatment first (Fig. 4B). Importantly, synergy between RP-3500 and olaparib occurred at lower concentrations of each agent (Fig. 4C) and a greater tumor cell-killing effect was observed (Fig. 4D) with concomitant treatment compared with sequential treatment.

Given the synergy observed between RP-3500 and a PARPi in cell-based assays, concomitant and sequential administration schedules were also compared in xenograft models with olaparib and niraparib to evaluate efficacy and tolerability. Mice were treated on weekly concomitant (both agents days 1–3 on/4 days off) or sequential (PARPi for 3 days followed by RP-3500 for 3 days, then 1 day off) schedules (Fig. 5A). Mice bearing SUM149PT tumors were treated for 5 weeks with either a concomitant or sequential intermittent RP-3500 + olaparib combination at dose levels below the MTD of each agent alone. All combinations were well tolerated with no significant body weight loss (Supplementary Fig. S4A) or PK drug–drug interactions (Supplementary Fig. S5A and S5B) for the treatment duration. As observed in vitro, the efficacy of the RP-3500/olaparib combination was significantly better when administered concomitantly compared with sequentially and was better than RP-3500 alone at its MTD (Fig. 5B; Supplementary Table S3). Importantly, intermittent concomitant administration was tolerated over 5 weeks of treatment and resulted in sustained antitumor activity for 3 weeks after dosing cessation in contrast to sequential administration where tumor growth continued over the course of the experiment.

Similarly, in mice bearing Granta-519 tumors, concomitant administration of RP-3500 and niraparib was more effective than sequential administration over a 3-week period, with no terminal body weight loss or difference in RBC count (Fig. 5C and D; Supplementary Fig. S4B). In this model, the combination was more efficacious and better tolerated (body weight and anemia) than high-dose niraparib as a single agent (70 mg/kg 5 days weekly) and did not cause meaningful drug–drug interactions (Supplementary Fig. S5B). A surge in the appearance of reticulocytes was observed with the 3 days on/4 days off concomitant schedule after 4 days off drug, suggesting that the intermittent RP-3500 and PARPi schedule permits regeneration of the RBC compartment, in contrast to sequential administration that causes more profound anemia and lower reticulocyte count relative to vehicle (Fig. 5E). For comparison, the same low dose levels of RP-3500 and niraparib were administered continuously for 7 days weekly in combination and were not tolerated, with 2 of 6 mice sacrificed on day 19 due to body weight loss and remaining mice showing severe anemia and ablation of reticulocytes on day 21 (Supplementary Fig. S4C–S4E). Together, these results demonstrate that an intermittent concomitant rather than sequential administration of RP-3500 + PARPi generates stronger antitumor responses without increased anemia.

In this first disclosure, we show that RP-3500 is a highly potent and selective ATRi in biochemical and cell-based assays with a preclinical pharmacology profile that justifies clinical evaluation. In cell assays, RP-3500 inhibits ATR with an IC₅₀ of 0.33 nmol/L, and is 30-fold selective over mTOR and more than 2,000-fold selective over other PIKK family members including ATM, DNA-PK, and the related kinase PI3Kα, showing that RP-3500 is among the most potent (9) and selective published ATRi (5, 10, 40–42). RP-3500 is orally bioavailable in vivo with compelling efficacy when administered daily as a single agent in the LoVo model with an MED of 7 mg/kg once daily compared with the published MED for AZD6738 of 25 mg/kg once daily in the same model (5). Similarly, in the Granta-519 model, RP-3500 at 30 mg/kg once daily provided equivalent efficacy to that reported for BAY1895344 at 50 mg/kg twice daily when both were administered on a 3 days on/4 days off schedule (10).

Insights derived from preclinical studies facilitate optimization of the RP-3500 therapeutic index as a single agent and in combination with PARPi; firstly, by enrolling patients whose tumors have genetic vulnerabilities to ATR inhibition. For example, the ATM-deficient RPE-1 cells were six-fold more sensitive to RP-3500 than WT RPE-1 cells. Further, ATM-deficient cells sustained irreversible RP-3500–mediated DNA damage over a 3-day exposure, unlike their WT counterparts. This is consistent with the results of Lloyd and colleagues (23), where AZD6738-induced DNA damage in the form of micronuclei was observed more quickly and at lower doses in FaDu ATM KO cells compared with FaDu ATM WT cells. Additional genetic vulnerabilities to ATRi have been identified through functional genomic screens (17–20) some of which now serve as clinical patient selection biomarkers. Furthermore, genetic profiles of innate replication stress have been identified in patients responsive to ATRi (43) or ATRi + topotecan (21) indicating a wide potential patient population for this drug class.

Secondly, the therapeutic index can be optimized with dose scheduling. The RP-3500 PK/efficacy relationship defined here illustrates that efficacy is influenced by the duration of daily target engagement and the duration of each treatment cycle. Our preclinical in vitro and in vivo studies suggest that, when using a continuous dosing schedule, 10 to 12 hours of 80% target engagement daily is required for minimum efficacy, whereas a 3- to 5-day treatment weekly is sufficient to maximize tumor cell killing if 80% target engagement is maintained during the treatment period. The increase in γH2AX that is proportional to efficacy suggests this to be a feasible molecular PD marker. The ability to maximize efficacy with a short duration of treatment allows flexibility for dosing holidays and increases tolerability by allowing recovery of bone marrow cells, given the reversible erythroid toxicity. Continuous daily administration of RP-3500 is limited by a reduction in RBC count and reticulocyte regeneration, consistent with the clinical findings of anemia with other ATRi (8, 14, 38, 44). Intermittent treatment is better tolerated with less anemia on a 3 days on/4 days off compared with 5 days on/2 days off or continuous daily administration schedules.

Optimizing the dosing schedule is particularly important in the context of ATRi/PARPi combinations, where overlapping clinical toxicities of single agent PARPi and ATRi include anemia, neutropenia, and thrombocytopenia (8, 14, 30). We demonstrate that continuous daily coadministration of RP-3500 and niraparib for several weeks in animal models is not tolerated, even at doses well below single-agent MTDs, consistent with previous preclinical (23, 24, 45) and clinical combination trials (44) with other ATRi. The concept of intermittent dosing of PARPi and ATRi was reported by Fang and colleagues and demonstrated good tolerability and efficacy with sequential PARPi followed by ATRi administration in vivo, compared with continuous administration of both agents together (24). However, this work did not provide a comparison of the proposed sequential schedule with concomitant intermittent scheduling. In contrast, our in vitro and in vivo results consistently demonstrate superior efficacy for concomitant versus sequential administration of PARPi followed by RP-3500. Supporting our observations in ATM-deficient tumor models, Lloyd and colleagues demonstrated sustained efficacy and less impact on body weight with full-dose olaparib in combination with half-dose AZD6738 with 3 days on/4 days off versus 5 days on/9 days off concomitant intermittent schedules (23). Our results demonstrate that, for RP-3500/PARPi combinations, introducing a weekly 4-day dose holiday allows for a resurgence of reticulocytes to regenerate the erythroid compartment; and that doses below the MTDs of each as single agents are sufficient for maximal efficacy. The increased tolerability and efficacy with this schedule are likely the result of two factors: (i) selective, irreparable DNA damage to DDR-defective tumor cells, and (ii) intermittent dose holidays allow for hematopoietic precursor cell maturation, minimizing anemia. Erythroid cell maturation, from early progenitor cells to mature enucleated reticulocytes, requires 2 to 7 days in mice and 7 to 9 days in humans (46, 47). A better understanding of the impact of RP-3500 on hematopoietic precursor cell injury and regeneration time following a dose holiday will dictate the intermittent scheduling of RP-3500 monotherapy or in combination with PARPi in clinic. The shorter half-lives of olaparib (12–18 hours; ref. 48) and niraparib (48–50 hours; ref. 49) among the PARPi are particularly amenable to intermittent clinical administration while talazoparib, with a half-life of 50 to 90 hours (50), is better suited to continuous administration. These results extend to the scheduling of ATRi combinations with other replication stress-inducing chemotherapeutic agents such as gemcitabine, platins, and topoisomerase inhibitors (14, 21, 51). Similar to PARPi, these agents are myelosuppressive and investigations incorporating intermittent ATRi administration warrant further investigation.

In summary, the results presented here demonstrate that RP-3500 is a potent ATRi with excellent selectivity over mTOR and other kinases. RP-3500 is highly efficacious in several ATM and BRCA1 LOF mouse models, with potential for tumor regression as a monotherapy. In xenograft models, intermittent administration 3 days on/4 days off weekly maximized efficacy while minimizing anemia. RP-3500 is highly effective in combination with PARPi when administered concomitantly on a short-term intermittent schedule rather than on continuous or sequential administration schedules. Preclinically, low-dose PARPi can be combined with RP-3500 to achieve dramatic and sustained antitumor responses on a schedule tolerated over several weeks. These results support the investigation of RP-3500 in an advanced phase I clinical study (NCT04497116) examining the short-term intermittent administration of RP-3500 as a monotherapy and in combination with low-dose PARPi as a potential means of achieving efficacy while alleviating anemia in clinical settings.

A. Roulston reports other support from Repare Therapeutics Inc. outside the submitted work; in addition, A. Roulston has a patent for WO2021/119523 pending; and is an employee of Repare Therapeutics Inc. M. Zimmermann reports personal fees from Repare Therapeutics Inc. during the conduct of the study; personal fees from Repare Therapeutics Inc. outside the submitted work; in addition, M. Zimmermann has a patent for WO2021/119523 pending. R. Papp reports other support from Repare Therapeutics Inc. outside the submitted work; in addition, R. Papp is an employee of Repare Therapeutics Inc.; and received stock options. A. Skeldon reports other support from Ventus Therapeutics Inc. during the conduct of the study. S. Dorich reports a patent for PCT17/245,574 pending. S. Fournier reports other support from Repare Therapeutics Inc. outside the submitted work; in addition, S. Fournier is an employee of Repare Therapeutics Inc. M.-E. Leclaire reports other support from Repare Therapeutics Inc. outside the submitted work; in addition, M.-E. Leclaire is an employee of Repare Therapeutics Inc. S.Y. Yin reports other support from Repare Therapeutics Inc. outside the submitted work; in addition, S.Y. Yin is a full time employee of Repare Therapeutics Inc. C. Fugère-Desjardins is a former employee of Repare Therapeutics. S. Vignini-Hammond reports other support from NuChem Sciences Inc. outside the submitted work. K. Skorey reports other support from NuChem Sciences Inc. during the conduct of the study. A. Mulani reports other support from NuChem Sciences during the conduct of the study. A. Veloso reports other support from Repare Therapeutics Inc. outside the submitted work; In addition, A. Veloso is an employee of Repare Therapeutics Inc. R. Stocco reports other support from Repare Therapeutics Inc. outside the submitted work; in addition, R. Stocco is an employee of Repare Therapeutics Inc. Y. Mamane reports personal fees and other support from Repare Therapeutics Inc.outside the submitted work; in addition, Y. Mamane is an employee of Repare Therapeutics Inc. Z. Li reports other support from NuChem Sciences Inc. outside the submitted work. J.T.F. Young reports other support from Repare Therapeutics Inc. outside the submitted work; in addition, J.T.F. Young is an employee of Repare Therapeutics Inc. M. Zinda reports other support from Repare Therapeutics Inc. during the conduct of the study; other support from Repare Therapeutics Inc. outside the submitted work; in addition, M. Zinda is an employee of Repare Therapeutics Inc. and owns equity in the company. W.C. Black reports personal fees from CQDM; and other support from Repare Therapeutics Inc. outside the submitted work; in addition, W.C. Black has a patent for WO2020/087170 pending and a patent for WO2021/119523 pending; and is an employee of Repare Therapeutics Inc. No disclosures were reported by the other authors.

A. Roulston: Conceptualization, formal analysis, supervision, writing–original draft, writing–review and editing. M. Zimmermann: Conceptualization, data curation, formal analysis, supervision, writing–review and editing. R. Papp: Conceptualization, data curation, formal analysis, supervision, writing–review and editing. A. Skeldon: Conceptualization, data curation, formal analysis. C. Pellerin: Conceptualization, data curation, formal analysis. É. Dumas-Bérube: Data curation, formal analysis. V. Dumais: Data curation, formal analysis. S. Dorich: Conceptualization, data curation, formal analysis. L.D. Fader: Conceptualization, supervision. S. Fournier: Data curation, formal analysis. L. Li: Data curation, formal analysis. M.-E. Leclaire: Data curation, formal analysis. S.Y. Yin: Data curation, formal analysis. A. Chefson: Data curation, formal analysis. H. Alam: Data curation, formal analysis. W. Yang: Data curation, formal analysis. C. Fugère-Desjardins: Data curation, formal analysis. S. Vignini-Hammond: Data curation, formal analysis. K. Skorey: Conceptualization, data curation, formal analysis. A. Mulani: Data curation, formal analysis. V. Rimkunas: Conceptualization, data curation, formal analysis. A. Veloso: Data curation, formal analysis. M. Hamel: Conceptualization. R. Stocco: Conceptualization. Y. Mamane: Conceptualization, supervision. Z. Li: Formal analysis, supervision. J. Young: Supervision. M. Zinda: Conceptualization, supervision, writing–review and editing. W.C. Black: Conceptualization, supervision, writing–review and editing.

The authors would like to thank Maria Koehler, Adrian Fretland, Stephen Morris, and Jeffrey Brown for helpful discussions and critical reading of the manuscript. Medical writing support was provided by Prime Global, supported by Repare Therapeutics Inc. and in accordance with Good Publication Practice guidelines (Link).

This research was funded by Repare Therapeutics Inc. All authors accept final responsibility for the content of the manuscript and for the decision to submit the manuscript for publication.