Although combinations of DNA damage response inhibitors (DDRi) and DNA-damaging chemotherapy enhance cytotoxicity in cell-based systems, clinical success has been limited by overlapping bone marrow toxicities. Here, we show that the tumour-targeted nanoparticle camptothecin CRLX101, when administered concurrently with DDRi, enhances anti-tumour efficacy but also increases bone marrow toxicity in preclinical models. In NCI-H417 small cell lung cancer cells, PARPi, ATRi, and ATMi all potentiated CRLX101 activity, with the PARPi olaparib reducing the CRLX101 growth inhibition (GI50) by 3.2-fold, ATR inhibition by 3.6-fold, and ATM inhibition by 7.8-fold. Following assessment of the most promising combinations in a rat bone marrow safety model, we identified olaparib as the best tolerated combination with CRLX101. However, concurrent dosing of CRLX101 with olaparib still increased hematological toxicity relative to either single agent alone. Using rat bone marrow progenitor cells as biomarkers and leveraging the differential repair kinetics of CRLX101 induced DNA damage in tumour and bone marrow, we identified a delayedor gap schedule of olapariband CRLX101 in which separation of dosing by 48 hours reduced the hematological impact of the combination while maintaining enhanced anti-tumour efficacy over single agents in xenograft models. Extended olaparib dosing after CRLX101 further improved tumour control, consistent with prolonged tumour retention of CRLX101. A clinical trial has been designed using the gap schedule identified here, and this approach may provide a template for combining DDR inhibitors with tumour-targeted DNA-damaging chemotherapy in a clinically translatable manner.

DNA damaging chemotherapy has been at the centre of cancer therapy for over seven decades (Cheung-Ong et al., 2013; DeVita and Chu, 2008). Underlining the preferential sensitivity of tumours compared to normal tissues are deficiencies in the DNA damage response (DDR), a collective term for the intra- and inter-cellular signaling events involved in the detection and repair of DNA damage (Ciccia and Elledge, 2010; Jackson and Bartek, 2009). In spite of the cancer-specific DDR deficiencies (Jackson and Bartek, 2009), the levels of exogenously generated DNA damage causedby systemic chemotherapies still results in significant and unwanted side effects. For this reason, there has been an attempt over the last twenty years or so, to identify specific inhibitors of the DDR in order to exploit the same cancer-specific DDR deficiencies (Curtin, 2012; O'Connor, 2015). In the case of inhibitors of poly (ADP-ribose) polymerase (PARP), this has led to the first approved DDR-based medicines, where PARP inhibitors (PARPi) have demonstrated monotherapy activity (Coleman et al., 2017; de Bono et al., 2020; Golan et al., 2019; Ledermann et al., 2014; Litton et al., 2018; Mirza et al., 2016; Robson et al., 2017) through a mechanism described as synthetic lethality(Fong et al., 2009; Lord and Ashworth, 2017). However, the potential for PARP inhibitors goes well beyondmonotherapy in HRR-deficient cancers and includes the potentiation of DNA damaging chemotherapy in a broader range of tumours (Pilie et al., 2019). Moreover, in addition to inhibitors of PARP, there are  now  several  other  DDR  targeted  agents  being developed in the clinic (O'Connor, 2015), including those agents that inhibit WEE1 (Hirai et al., 2009), ATR (Min et al., 2017; Reaper et al., 2011; Wengner et al., 2020), CHK1 (Angius et al., 2020), ATM (Durant et al., 2018; Riches et al., 2020), DNA-PK (Fok et al., 2019; Wise et al., 2019) and Aurora Kinase B (Ashton et al., 2016).

While there are no shortagesof preclinical exampleswhere DDR inhibitors (DDRi) enhance the anti-tumour activityof chemotherapies, in most cases, poor tolerability for these combinations is underestimated becauseof the use of sub-optimal preclinical models. For example, mouse tumour efficacy models will not accurately reflect the potential impact on bone marrow since mice preferentially use different (more efficient but error-prone) DDR pathways in bone marrow stem cells than either rats or humans (Lane and Scadden, 2010; O'Connor et al., 2016).For this reason, rats represent a better preclinical safety model than mice to assess bone marrow toxicity. When DDRi are combined with chemotherapy, the experience from the clinic is that the overlapping toxicities with DDRi present a significant challenge and currently there have been few successful trials demonstrating efficacy coupled with acceptable tolerability. This point is highlighted by the fact that currently there isn’t a single approval for a targeted DDR agent in combination with chemotherapy.

One approach to improve the therapeutic window of chemotherapy is to target the DNA damaging agent to the tumour and/or avoid the bone marrow compartment. Approaches to achieve this currently in development include re-formulation (e.g. through liposomal or nanoparticle approaches (Wang et al., 2012)) or conjugation of the chemotherapy war-headto an antibody that preferentially targets a tumour surface marker (antibody-drug-conjugates or ADCs; (Beck et al., 2017; Thomas et al., 2016)). However, even tumour-targeted delivery may not fully prevent overlapping toxicity when combined concurrently with DDR inhibitors.

Here, we investigated whether a rationally designed dosing schedule could widen the therapeutic window of a tumour-targeted topoisomerase I inhibitor when combined with DDR inhibition. We evaluated CRLX101, a nanoparticle camptothecin formulation CRLX101(Weiss et al., 2013), which we chose for combination studies based on its monotherapy activity, preferential accumulation in tumors, and favorable activity compared to conventional topoisomerase I inhibitors (TOP1i) (Clark et al., 2016; Pham et al., 2015; Weiss et al., 2013). For the DDRi combinations, we assessed the PARPi olaparib (Menear et al., 2008), WEE1i adavosertib/AZD1775 (Hirai et al., 2009), ATRi ceralasertib/AZD6738 (Foote et al., 2018), ATMi AZD0156 (Riches et al., 2020), DNA-PKi AZ’6119, a precursor of AZD7648 (Fok et al., 2019), and Aurora kinase Binhibitor AZD1152 (Wilkinson et al., 2007). After triaging combinations in vitro to identifythose with enhancedanti-tumour activity, we evaluated the most promising regimens in a rat bone marrowsafety model and in mouse xenograft efficacy studies. Our objective was to determine whether differential tumour and bone marrow damage-repair kineticscould be exploited to develop a gap schedule that reduces hematological toxicity without compromising efficacy, therebyproviding a clinically translatable starting point for combination testing.

Cell culture and chemicals

All cell lines were grown at 37°C in humidified incubator with 5% CO2 and maintained in phenol-free Dulbecco Modified Eagle Medium (Gibco) supplemented with 10 percentage fetal bovine serum and 2mM GlutaMax (Gibco). AZD2281 (olaparib; PARPi), AZD1775 (adavosertib; WEE1i), AZD0156 (ATMi), AZD6738 (ceralasertib; ATRi), AZ’6119

(DNA-PKi), and AZ1152 (Aurora B kinase inhibitor) were synthesized internally at AstraZeneca. All compounds were dissolved in DMSO at stock concentration of 10 mM.

Cell growth inhibition assay

SCLC NCI-H417 cells were seeded in 96-well plates and allowed to adhere overnight before treatment with DMSO, CRLX101, or CRLX101 in combination with the indicated DDR inhibitors. For CRLX101 single-agent assessment, cells were exposed for 48h before drug washout. DDR inhibitors were added concurrently with CRLX101 and replenished after 48h. Growth inhibition (GI50) values were determined on day 7 using the MTT assay,in which tetrazolium MTT substrate (Sigma) was added to a final concentration of 0.25 mg/mL. Following 6–8h incubation, formazan crystalproducts were solubilized in 5% SDS and 0.005 M HCl (final concentrations). Optical density was read at 570nm, and absorbance values were normalized to the DMSO control. Relative absorbance values were analysed in GraphPad Prism to derive GI50 values.

Cell line immunoblotting and antibodies

Whole cell lysates were prepared by lysing cell pellets directly in 2x Laemmlisample buffer (4% SDS, 20% glycerol, 125 mM Tris-HCl pH 6.8) and vortexing samples at highest speed for 20 seconds. Following protein concentration measurement using DC Protein Assay (Bio-Rad), protein samples were supplemented with sample reducing agents at 1x (Invitrogen) and 0.01% bromophenol blue final concentration. Samples were boiled at 95°C for 5 minutes before proteins were separated on NuPAGE 4-12% Bis-Tris protein gels (Invitrogen) by SDS-PAGE.Following protein transferonto nitrocellulose membranes, immunoblotting analyses were performed using antibodies listed below.

Cell cycle analysisby flow cytometry

Cells treated with indicated compounds were harvested and fixed in ice-cold70% ethanol overnightat -20 °C. Fixed cells were then washed with PBS before incubated in PBS containing 100 µg/ml RNase A (Thermofisher) and 50 µg/ml propidium iodide (Sigma) for 30 minutes at 37 °C. FACSCalibur (BD Biosciences) was employed to analyze samples, and data were plotted using FlowJo software.

In vivo efficacy and pharmacodynamic studies

All in vivo studies were performed within AstraZeneca in the United Kingdom, and study protocols were reviewed and approved by the Home Office. All studies were conducted in accordance with the Animal Scientific Procedures Act 1986 (ASPA) and AstraZeneca Global Bioethics policy. Data were reported in line with the Animal Research: Reporting In Vivo (ARRIVE) guidelines (Kilkenny et al., 2010). For efficacy studies, human small cell lung cancer cells (NCI-H417) were implanted subcutaneously in athymic Fox1-nu mice. Animals were dosed with individual agents or combinations of CRLX101 (once weekly) with or without DDR inhibitors. Animals were randomized into vehicle and treatment groups based on a mean tumour volume of approximately 0.1–0.2cm3. Treatment groups received CRLX101 alone at either 4mg/kg or 5mg/kgonce weekly, while DDR inhibitor daily oral dosing was 100mg/kg for olaparib, 120mg/kg for AZD1775, and  5mg/kg  for  AZD0156.  Animals  were monitored throughout the study according to institutional welfare guidance, and tumour and tissue sampling time points were selected to capture both early and sustained pharmacodynamic effects following CRLX101 exposure. Statistical significance was evaluated using a one-tailed t test.

In vivo rat studies

RccHan:WIST rats (n = 7 per treatment group) were obtained from Harlan UK. Animals were approximately 11 weeks old at the start of dosing, were acclimatized for at least 1 week, and were group-housed at up to 5 per cage. Water and RM1 (E) SQC pelleted diet (Special Diet Services Ltd, England) were provided ad libitum. Environmental enrichment included nesting material and polycarbonate tunnels. CRLX101 was formulated as a 2mg/mL PBS nanosuspension and dilutedto the required concentration in PBS on the day of use. Animals receiving CRLX101 were given a single intravenous administration on day 1. Olaparib was formulated in dimethyl sulphoxide diluted 1:10 with 10% hydroxypropyl-β-cyclodextrin in phosphate-buffered saline (pH 7.4) and administered once daily by oral gavage. When AZD2281 and CRLX101 were dosed on the same day, the oral olaparib dose was given approximately 1h after the intravenous CRLX101 dose. Topotecan was supplied by Accord Healthcare Limited, formulated in 0.9% saline, and administered daily by oral gavage. All animals were euthanized by administration of halothane.

Flow cytometry analysisof rat bone marrow cells

Rat femurs were removed at necropsy and both ends were trimmed. Bone marrow cells were immediately flushedout with 3 ml PBS containing 50 percentage fetal calf serum (FCS). Cell suspension was syringed and filtered through 100 μm strainer, collected by centrifugation (300 g/ 7 min/ 4ºC) and washed once in HBSS containing 2 percentage FCS and 10 mM HEPES (staining buffer). Total cell count of isolated cells was determined by automated cell counter (Countess, Invitrogen). Cell concentration was adjusted to 1x107 cells/ ml in staining buffer and processed for antibody staining. CD71 and CD45 cocktail: 100 μl cell suspension wasresuspended in 100 μl stainingbuffer containing anti-rat CD71-FITC (dilution 1:10) and CD45-PE (1:10) from Serotec. Cells were incubated with antibodies for 30 minutes RT and washed twice in stainingbuffer. Cell pellet stained with CD71-CD45 antibodies was resuspended in 200 μl staining buffer. This was followed by additionof 10 μl of LDS-751 cell-permeant nuclear stain (Life Technologies). CD90.1 and Lineages cocktail: 1 ml of cell suspension was resuspended in 100 μl staining buffer containing anti-ratCD90.1-APC (dilution 1:100),CD6-FITC (1:100),CD3-FITC  (1:100),  CD11b-FITC  (1:200), Granulocytes-FITC (1:200) from BD Pharmingen and CD45RC-FITC (1:100) purchased from Serotec. After final centrifugation, 1 ml of staining buffer was added to CD90.1-Lineages-stained cells. Cells were incubated in dark for 30 minutes prior to flow cytometry analysis. Data (at least 10,000 events) were acquired on FACS AriaII (BD). Analysis was performed in FlowJo software.

Immunohistochemistry (IHC) analysis of gH2AX in rat bone marrow and xenofraft tumor

Rat femurs or human tumour samples implanted in mice were removed at necropsy and fixed in formalin in situ for 48 hours prior to coring. Preserved tissue was processed into wax blocks, sectioned and stained by immunohistochemistry for the presence of gH2AX (Ventana™, Omin-UltraMap HRP, Doscovery XT Staining module; gH2AX CST 25777 @ 1:100). Sections were counter stained with haematoxylin. The slides providedfor analysis were scanned using the Aperio Scanscope, converted to .TIF images (x 10 magnifications), and analysed on the KS400 image analyser.

Haematology analysis of rat plasmasamples

Blood samples to be taken from the tail vein (0.4 mL into EDTA). Haematology analysis was performed on the same day using the Siemens Advia 2120i automated haematology analyser.

Statistical analysis

GI50 values from cell growth inhibition assays were calculated using GraphPad Prism by fittingnormalized dose-response curvesto the experimental data. For in vivo efficacy studies, tumour volume data were summarized over time, and statistical comparisons between treatment groups were performed using a one-tailed t test, as described in the study methods. A p value <0>

Data Availability Statement

The data supporting the findings of this study are contained within the manuscript and its supplementary materials. Additional underlying data may be made available by the corresponding author upon reasonable request, subject to institutional and confidentiality requirements.

DDR inhibitors differ in both the extent and mechanism of CRLX101 potentiation

CRLX101is a targeted topoisomerase I inhibitor (TOP1i)in clinical development based on a nanoparticle camptothecin formulation (Clark et al., 2016). Small cell lung cancer (SCLC) is a tumour type of interest for CRLX101 development because these tumours are responsive to TOP1 inhibitors, including the standard-of-care (SoC) therapeutic topotecan and remains an area of high unmet clinical need. To evaluate the benefit of combining CRLX101 with different DDR inhibitors, we chose to use the SCLC cell line model NCI-H417, because the responseto single agent CRLX101 had previously been characterized and importantly, the NCI-H417 cell line could be used to generate an in vivo xenograft model.

We treated the NCI-H417 cells with CRLX101 in combination with a PARP inhibitor (olaparib), a WEE1 inhibitor (AZD1775), an ATM inhibitor(AZD0156), an ATR inhibitor (AZD6738), a DNA-PK inhibitor (AZ’6119)and an Aurora kinase B inhibitor (AZD1152). Cells were exposed to CRLX101 for 48h, followed by drug wash-out. A fixed dose of DDR inhibitor (shownto provide greaterthan 95% target inhibition) was added concurrently to varying concentrations of CRLX101 on day 1 and replenished after 48h. Cell viability was determined at day 7 using an MTT assay (Figure 1A). The PARPi, ATRi and ATMi all significantly potentiated the activity of CRLX101, resulting in a decrease in the GI50 of CRLX101 by 3.2-fold, 3.6-fold and 7.8-fold respectively (Figure 1A).

Figure 1(A): Potentiation of CRLX101 by different DDR inhibitors. SCLC cell line NCI-H417a treatedwith CRLX101 for 48h before washing out the drug. DDR inhibitors were added concurrently with CRX101 and replenished after 48h. Growth inhibition (GI50 values) were determined at day 7 using the MTT assay.

A significant proportion of cells (~20-30%) were not killed by CRLX101 treatment alone, even at the higher concentrations. Combinations with the PARPi, ATMi, or ATRi shifted the GI50 curve but did not result in any increase in this maximumcell kill value. Combination with the WEE1 inhibitor (AZD1775), although not significantly shifting the GI50 curve, did result in a more potent cell kill (>90

Many SoC chemotherapies act by generating DNA damage that has the potential to be enhanced by inhibitors of the DDR. However, overlapping toxicities, in particular bone marrow toxicity, has hampered the ability to combine these agents. Attempting to identify an effective and sufficiently well tolerated dose and schedule in the clinic cantake many years and still fail to identify a combination that is clearly better than the full dose SoC chemotherapy alone. A good illustration of this is the combination of the PARPi olaparib with carboplatin and paclitaxel, where a Phase I trial (NCT00516724) began in 2007, took many years to complete and involved 189 patients. Only recently has this trial been reported (van der Noll et al., 2020) and the eventual conclusion was that olaparib in combination with carboplatin and/or paclitaxel resulted in increased hematologic toxicities, making it challenging to establish a dosing regimen that could be tolerated for multiple cycles without dose modifications. Anotherexample, where there is a clear mechanistic rational for combination but where enhanced bone marrow toxicity has meant no optimal recommended Phase II dose has been identified that could be taken forward, is the combination of PARPi with TOP1i such as irinotecan and topotecan (Murai et al., 2014). Several clinical trials (Chen et al., 2016; Dhawan et al., 2017; Hendrickson et al., 2018; Kummar et al., 2011; LoRusso et al., 2016; Rajan et al., 2012; Samol et al., 2012) have tried and failedto combine PARPiand TOP1i at doses close to their individual MTDs with dose limiting myelosuppression precluding administration of >20%of a PARPi MTD dose.

One limitation of the work presented here is that the experiments were performed in selected tumour models, and the magnitude of benefit may vary according to tumour biology, including DNA repair status and intrinsic sensitivity to TOP1 inhibition. Moreover, the feasibility of this approach is likely to depend on the pharmacokinetic and pharmacodynamic properties of the DNA-damaging partner, particularly the extent to which sustained tumour retention can be separated from normal tissue exposure.

Nonetheless, the data presented in this preclinical study deploying a gap schedule that could deliver 80% of the MTD equivalent of CRLX101 with an MTD equivalent of olaparib, was used to design a clinical trial (Clinical trials.gov reference NCT02769962, where CRLX101 was renamed EP0057). This clinical study has recently been published and provides encouraging validation of our gap scheduling approach, with escalation of both the PARP inhibitor and TOP1 inhibitor to approximately 80% of their respective single-agent MTDs, consistent with our preclinical data, and higher than previous clinical trials were able to achieve and with encouraging clinical activity in heavily pretreated patients (Thomas et al, 2025).

In addition to the specificexample of CRLX101that was the focus of this study, the gap scheduling approach exemplified here also has the potential to work for PARPi with other targeted DNA damaging agents and for other DDRi. One exciting opportunity will be the combination of DDRi with antibody drug conjugates (ADCs) and targeted radioligands that could benefit from gap combination schedules with DDR inhibitors (Yap et al., 2026).

In summary, DDR inhibitor combinations, either with DNA damaging agents or other DDR inhibitors, have significant therapeutic potential but are nevertheless challenging. The work presented here has highlighted how an assessment of preclinical models of both tumour and normal tissue responses, such as those of rat bone marrow, can provide dose and scheduling insights that can significantly enable combination testing in the clinic.

Acknowledgements

We would like to thank those currentand past AstraZeneca scientists, who are not co-authors but who nevertheless contributed towards the work, namely Jamie Reens, Catherine Wilkinson, Anna Cronin, Richard Knights and Pete Newham.

Author Contributions

Original conceptualization and methodology for the preclinical studies was provided by MOC and LOC. Experimental investigation was provided by LOC, AW, CS, JB, RO, AS, GH, JR, AL and EC, with overall supervision from MOC.

Declaration of interests

MOC, LOC, AW, AS, GH, AL and EC are employees and shareholders of AstraZeneca, while CS, JB, and RO are former employees of AstraZeneca.