Targeting G1–S-checkpoint-compromised cancers with cyclin A/B RxL inhibitors

Cell lines and cell culture

NCI-H446, NCI-H69 cells, NCI-H1048, NCI-H526, NCI-H82, WI-38, RPE1 and HEK293T cells were originally obtained from American Type Culture Collection (ATCC) and authenticated by ATCC. A549, HCC4006 and NCI-H1299 cells were a gift from P. Janne’s laboratory at DFCI and originally obtained and validated by ATCC. NCI-H1048, Jurkat, NCI-H82 and MDA-MB-231 cells with DOX-on inducible E2F1 expression were a gift from W. G. Kaelin’s laboratory at DFCI and were originally obtained and authenticated by ATCC prior to genetically manipulation. G-CSF mobilized human peripheral blood CD34⁺ cells were obtained from StemCell Technologies (700060.1, 230472503C) and validated by flow cytometry. No commonly misidentified cell lines were used in this study.

NCI-H69, NCI-H446, HCC4006, NCI-H1299, NCI-H526, A549, NCI-H82 and Jurkat cells were maintained in RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS) (Gemini), 100 U ml⁻¹ penicillin and 100 μg ml⁻¹ streptomycin. NCI-H1048 cells were cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS), penicillin–streptomycin and insulin–transferrin–selenium (Gemini). WI-38 cells were maintained in DMEM supplemented with 10% FBS. HEK293T cells were maintained in DMEM medium with 10% FBS and penicillin–streptomycin. MDA-MB-231 and RPE1 cells were maintained in DMEM/F12 with 10% FBS and penicillin–streptomycin. G-CSF mobilized human peripheral blood CD34⁺ cells were thawed into StepSpan Serum Free Expansion Medium (SFEM II) Hematopoietic Stem Cell Medium (StemCell Technologies, 09655) supplemented with StemSpan CD34⁺ Expansion Supplement (StemCell Technologies, 02691). Early-passage cells of all of the cell lines listed above were tested for mycoplasma (Lonza, LT07-218) and then were frozen using Bambanker’s freezing medium (Bulldog Bio) and maintained in culture no more than 4 months in cases in which early-passage vials were thawed. iHCT116 cells were previously described³⁷ and cultured in 10% fetal bovine serum in DMEM medium (Sigma-Aldrich) supplemented with 2 mM l-glutamine (Sigma-Aldrich).

Pharmacological inhibitors

Macrocycle peptides offer a feasible strategy to block broad and shallow protein-protein interaction surfaces that until recently have been considered undruggable. Over the past decade, a growing body of research has demonstrated that peptidic macrocycles can be engineered to be both passively cell permeable^67,68,69,70 and orally bioavailable^26,28,71,72, therefore offering promise for this modality to access previously undruggable intracellular targets²⁷. Our approach to identify passively permeable cyclin RxL inhibitors included the incorporation of peptidomimetic features based on our previous research^67,68,73,74. Detailed descriptions of our medical chemistry efforts to achieve passive permeability, cyclin binding affinity and selectivity of an earlier series of distinct tool cyclin RxL macrocycles not included in this manuscript are reported in ref. ³¹. A detailed description of the synthesis of the cyclin RxL inhibitors in this manuscript is provided in the Supplementary Methods.

Where indicated, the following chemicals (stored at −20 °C or −80 °C) were also added to the medium: CIRc-004 (cyclin A/B RxL inhibitor, stock 10 mM in DMSO, obtained from Circle Pharma), CIRc-005 (inactive enantiomer of CIRc-004, stock 10 mM in DMSO, obtained from Circle Pharma), CIRc-001 (cyclin A/B RxL inhibitor, stock 10 mM in DMSO, obtained from Circle Pharma), CDK2i (PF-07104091, stock 10 mM in DMSO, chemically synthesized and validated to be the same PF-07104091 in the published patent), CIRc-018 (cyclin-A-selective RxL inhibitor, stock 10 mM in DMSO, obtained from Circle Pharma), CIRc-019 (cyclin-B-selective RxL inhibitor, stock 10 mM in DMSO, obtained from Circle Pharma), MPS1 inhibitor (BAY-1217389, Selleck chemicals, S8215, stock 10 mM), doxycycline hydrochloride (Sigma-Aldrich, 3447, stock 1 mg ml⁻¹), RP-6306 (MedChemExpress, HY-145817A, stock 20 mM), roscovitine (Selleck Chemical, S115350MG, stock 10 mM), staurosporine (MedChemExpress, HY-15141, stock 20 mM), nocodazole (Sigma-Aldrich, M1404, stock 2 mg ml⁻¹), RO-3306 (ENZO, ALX-270-463-M001, stock 10 mM) and MK-1775 (Selleck Chemical, NC1097015, stock 10 mM).

Biochemical activity assays

The binding potency of macrocycles for cyclin–CDK complexes was determined by a fluorescence polarization competitive assay based on previously established protocols^25,75 with the following modifications: (1) a higher-affinity fluorescent-labelled probe was identified referred to as CIR7-2706 (Supplementary Methods); (2) cyclin–CDK protein complexes were sourced as follows: cyclin A2–CDK2 (CRELUX Protein Services; www.crelux.com), cyclin B1–CDK1 (Eurofins, Discovery, 14-450) and cyclin E1–CDK2 (Eurofins, Discovery, 14-475). Fluorescence polarization binding assays were performed in 25 mM HEPES pH 7.5, 100 mM NaCl, 1 mM DTT, 0.01% NP-40 and 1 mg ml⁻¹ BSA for all 3 protein complexes in black 96-well plates (Costar, 3356). After experimental plates were set, they were equilibrated by gentle mixing by placing them on an orbital shaker at 100 rpm for 2 h at room temperature and then read on the SpectraMax i3X Multi-Mode Microplate Detection platform. The 50% inhibitory concentration (IC₅₀) of binding of a fluorescence labelled probe CIR7-2706 at 2 nM to the cyclin–CDK complex was determined from an eight-point threefold serial dilution curve of CID-078. The protein concentrations used were 8 nM for cyclin A2–CDK2, and 10 nM for cyclin B1–CDK1 and cyclin E1–CDK2. Under these conditions, the assay dynamic range was about 120 mP units between 100% binding of CIR7-2706 to the proteins and its complete displacement by an unlabelled competitor compound. All experiments showed a Z′ factor of >0.80. The reported IC₅₀ values are the average of all experiments described in Fig. 1b.

RxL inhibitor affinity determinations were performed in a WaveDelta instrument (Malvern Panalytical). Each of the cyclin–CDK complexes were immobilized to the chip surface by amine coupling on one of four channels of a 4PCP chip (9060001, Malvern Panalytical) in 10 mM MES pH 6.5 buffer for cyclin A2–CDK2 and cyclin E1–CDK2 and 10 mM MOPS pH 7.0 for cyclin B1–CDK1. For coupling, the proteins were diluted at 25 mg ml⁻¹ and applied to the corresponding chip channel for 600 s at a flow rate of 10 ml min⁻¹, followed by quenching by injecting 1 M ethanolamine-HCl pH 8.5 for 420 s at a flow rate of 10 ml min⁻¹. Compounds were dissolved at 300 nM in 20 mM HEPES pH 7.5, 300 mM NaCl, 0.5 mM TCEP, 0.05% Tween-20 and 3% DMSO. An eight-point 1:3 serial dilution of the compounds was sequentially injected into the chip at flow rate of 100 ml min⁻¹ for 120 s to measure the association rates, followed by injection for 600 s of buffer alone to measure the dissociation rates. Background readings were subtracted using an empty channel on the chip on which no protein was immobilized. The data were analysed using a 1:1 binding kinetic model using the instrument software (Wave Control, v.4.5.17). The reported K_d, k_a and k_d are the average of two or more experiments as indicated. ClogP values were calculated using the calculator in ChemDraw. MDCK monolayer cell permeability, kinetic solubility (K_sol), and logD experimental values were generated by Quintara Discovery using standard experimental conditions.

sgRNA cloning to make lentiviruses

sgRNA sequences were designed using the Broad Institutes sgRNA designer tool (http://portals.broadinstitute.org/gpp/public/analysis-tools/sgrna-design) and chosen from the Brunello CP0043 sgRNA library and synthesized by IDT technologies. The sense and antisense oligonucleotides were mixed at equimolar ratios (0.25 nmol of each sense and antisense oligonucleotide) and annealed by heating to 100 °C in annealing buffer (1× annealing buffer: 100 mM NaCl, 10 mM Tris-HCl, pH 7.4) followed by slow cooling to 30 °C for 3 h. The annealed oligonucleotides were then diluted at 1:400 in 0.5× annealing buffer. For CRISPR–Cas9 knockout experiments in cells, the annealed oligos were ligated into pLentiCRISPR Puro V2 (Addgene, 52961). Ligations were done with T4 DNA ligase for 2 h at 25 °C. The ligation mixture was transformed into HB101 competent cells. Ampicillin-resistant colonies were screened by restriction digestion of miniprep DNAs and subsequently validated by DNA sequencing.

The following sgRNA oligos were used for LentiCRISPRV2 Puro vector for CRISPR knockout experiments: non-targeting 1 sense (5′-CCGTCTCCGCATCGTCTTTT-3′); non-targeting 1 antisense (5′-AAAAGACGATGCGGAGACGG-3′); non-targeting 2 sense (5′-TATTTTGACTTGACGCAGGC-3′); non-targeting 2 antisense (5′-GCCTGCGTCAAGTCAAAATA-3′); human CCNB1 1 sense (5′-CATCAGAGAAAGCCTGACAC-3′); human CCNB1 1 antisense (5′-GTGTCAGGCTTTCTCTGATG-3′); human CCNB1 2 sense (5′-GAGGCCAAGAACAGCTCTTG-3′); human CCNB1 2 antisense (5′-CAAGAGCTGTTCTTGGCCTC-3′); human CDK2 1 sense (5′-CAAATATTATTCCACAGCTG-3′); human CDK2 1 antisense (5′-CAGCTGTGGAATAATATTTG-3′); human CDK2 2 sense (5′-CATGGGTGTAAGTACGAACA-3′); human CDK2 2 antisense (5′-TGTTCGTACTTACACCCATG-3′); human KNTC1 1 sense (5′-AAACATTCGGAACACTATGG-3′); human KNTC1 1 antisense (5′-CCATAGTGTTCCGAATGTTT-3′); human KNTC1 2 sense (5′-AAGCTAACGATGAAAATCGG-3′); human KNTC1 2 antisense (5′-CCGATTTTCATCGTTAGCTT-3′); human LIN54 1 sense (5′-GTTCTTTCACAGTCTACTCC-3′); human LIN54 1 antisense (5′-GGAGTAGACTGTGAAAGAAC-3′); human LIN54 2 sense (5′-AAGTAGTTACCATTGGAGGG-3′); human LIN54 2 antisense (5′-CCCTCCAATGGTAACTACTT-3′); human MAD1L1 1 sense (5′-GAAGAAGCGCGAGACCCACG-3′); human MAD1L1 1 antisense (5′-CGTGGGTCTCGCGCTTCTTC-3′); human MAD1L1 2 sense (5′-GAGAGACTGGACCAGACCAT-3′); human MAD1L1 2 antisense (5′-ATGGTCTGGTCCAGTCTCTC-3′); human ZWINT 1 sense (5′-CACCTACAGGGAGCACGTAG-3′); human ZWINT 1 antisense (5′-CTACGTGCTCCCTGTAGGTG-3′); human ZWINT 2 sense (5′-CAGTGGCAGCTACAACAGGT-3′); human ZWINT 2 antisense (5′-ACCTGTTGTAGCTGCCACTG-3′); human RB1 2 sense (5′-CGGTGGCGGCCGTTTTTCGG-3′); human RB1 2 antisense (5′-CCGAAAAACGGCCGCCACCG-3′); human TP53 1 sense (5′-CCATTGTTCAATATCGTCCG-3′); and human TP53 1 antisense (5′-CGGACGATATTGAACAATGG-3′).

cDNA expression plasmids

CDC20 (NM_001255.3) or CDC20 with an N-terminal 3×Flag tag was cloned into the pLVX IRES Hygro vector (Clontech). The CDC20 R445Q mutation was generated using site-directed mutagenesis (Pfu polymerase). pTwist Lenti SFFV Puro WPRE lentiviral vectors encoding: (1) CCNB1 wild-type-HA, (2) CCNB1 triple mutant-HA (E169K/Y170H/Y177C) or a (3) negative control that encodes a functionally inactivated EED mutant-HA with mutations to inactivate EED (F97A/Y148A/Y365A) were codon optimized and synthesized by Twist Biosciences. All plasmids were sequence verified by sanger sequencing. To make the DOX-on pTripz neo E2F1, E2F2 or E2F3 construct, pDONR223 E2F1 (W. G. Kaelin’s laboratory), pCMVHA E2F2 (Addgene, 24226) and pCMV Tag2B E2F3 (Addgene, 202522) were used as templates for overhang PCR to introduce the attB1 and attB2 sites onto the 5′ and 3′ ends of E2F1 (sense primer, 5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTTCACCATGGCCTTGGCCGGGGCCCCTGCG; and antisense primer, 5′-GGGGACCACTTTGTACAAGAAAGCTGGGTTGAAATCCAGGGGGGTGAGGTCCCC-3′), E2F2 (sense primer, 5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTTCACCATGCTGCAAGGGCCCCGGGCCTTG-3′; and antisense primer, 5′-GGGGACCACTTTGTACAAGAAAGCTGGGTTATTAATCAACAGGTCCCCAAGGTC-3′) and E2F3 (sense primer, 5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTTCACCATGAGAAAGGGAATCCAGCCCGCT; and antisense primer, 5′-GGGGACCACTTTGTACAAGAAAGCTGGGTTACTACACATGAAGTCTTCCACCAG-3′). The PCR product was then gel extracted according to the manufacturer’s instructions (Qiagen). Purified E2F1, E2F2 and E2F3 double-stranded DNA fragments containing attB1 and attB2 sites were introduced into the pDONR223 vector by homologous recombination using BP clonase (Life Technologies, 11789020) for 1 h at 25 °C according to the manufacturer’s instructions. The reaction mixture was then transformed at a ratio of 1:10 (reaction volume/volume competent cells) into HB101 competent cells (Promega). Spectinomycin-resistant colonies were screened by restriction digestion of miniprep DNA and subsequently validated by DNA sequencing. A homologous recombination reaction was then performed using LR Clonase II (Life Technologies, 11791100) with the pDONR223-E2F1, pDONR223-E2F2 or pDONR223-E2F3 plasmids and DOX-on pTripz-NEO DEST vector that was modified to be used as destination vectors for Gateway recombination cloning. The LR reaction was transformed into HB101 competent cells (Promega). Kanamycin-resistant (pTripz-NEO) colonies were screened by restriction digestion of miniprep DNA and subsequently validated by DNA sequencing. Lentivirus was then generated and then transduced into cells to make stable DOX-on cell lines for inducible E2F1, E2F2 or E2F3 expression.

Lentivirus production

Lentiviruses were made by Lipofectamine-2000-based co-transfection of HEK293FT cells with the respective lentiviral expression vectors and the packaging plasmids psPAX2 (Addgene, 12260) and pMD2.G (Addgene, 12259) at a ratio of 4:3:1. Virus-containing supernatant was collected at 48 and 72 h after transfection, pooled together (7 ml total per 6 cm tissue culture dish), passed through a 0.45-µm filter, aliquoted and frozen at −80 °C until use.

Lentiviral infection

The cells were counted using a Vi-Cell XR Cell Counter (Beckman Coulter) and 2 × 10⁶ cells were resuspended in 1 ml lentivirus with 8 μg ml⁻¹ polybrene in individual wells of a 12-well plate. The plates were then centrifuged at 931g (Allegra X-15R Centrifuge (Beckman Coulter), rotor SX4750A, 2,000 rpm) for 2 h at 30 °C. Then, 16 h later, the virus was removed and cells were grown for 72 h before being placed under drug selection. Cells were selected in puromycin (0.5 μg ml⁻¹) or G418 (500 μg ml⁻¹).

RB1 TP53 CRISPR–Cas9 double-knockout experiments in RPE1 cells

RPE1 cells were first infected with lentivirus encoding an sgRNA targeting RB1 (sgRB1) or a non-targeting sgRNA (sgC0111, sgControl) in LentiCRISPR-V2 Puro vector. RPE1 sgRB1 or sgControl cells were superinfected with lentiviruses encoding an sgRNA targeting TP53 (sgTP53) or a non-targeting sgRNA (sgC0111, sgControl) and a G418 resistance gene in the pLentiCRISPR-V2 vector. Cells were selected in G418 (500 µg ml⁻¹) until all mock-infected cells were killed and immunoblot analysis was performed for RB1 and p53 in the presence or absence of CIRc-004 at the concentrations indicated.

Cyclin A/B RxL inhibitor, CDK2i and MPS1 inhibitor dose–response assays

For dose–response assays with cyclin A/B RxL inhibitor (CIRc-004), inactive enantiomer of CIRc-004 (CIRc-005) and CDK2i (PF-07104091), the cells were counted on day 0 using a Vi-Cell XR Cell Counter and plated in a tissue-culture-treated six-well plate at 50,000 cells per ml in 2 ml of complete medium. Cells were treated with the compounds (CIRc-004, CIRc-005, PF-07104091) at the indicated concentrations for 6 days. Normalized cell counts were calculated as (day 6 count for treated samples)/(day 6 count for DMSO treated sample). The EC₅₀ was calculated using GraphPad Prism v.10.0.0.

For DOX-inducible (DOX-on) E2F1-, E2F2- or E2F3-overexpressing cell lines, cells were counted on day 0 and plated in tissue-culture-treated six-well plates at 250,000 cells per ml for 3-day assays or 50,000 cells per ml in 2 ml of complete medium for 6-day assays. For 3-day cyclin A/B RxL inhibitor assays, E2F1 expression was induced with DOX (1 µg ml⁻¹) for 24 h followed by drug treatment for 3 days. For 6-day cyclin A/B RxL inhibitor assays, E2F1 expression was induced with DOX for 24 h followed by drug treatment for 6 days, with replacement with fresh DOX every 3 days.

MTT proliferation assays

NCI-H1048, NCI-H446 and WI-38 cells were counted using a Countess II FL (Invitrogen) and plated at 5,000 cells per well in 96-well plates. NCI-H69 cells were grown to confluency in a T150 flask. Cells were collected by centrifugation at 283g (Thermo Fisher Scientific X4R Pro-MD centrifuge, Sorvall TX1000 rotor, 1,100 rpm) and resuspended in 30 ml medium of which 10 ml was used to seed six 96-well plates. NCI-H526 cells were grown to confluency in a T75 flask, collected by centrifugation at 283g and resuspended in 10 ml medium of which 2.5 ml was used to seed six 96-well plates. The next day, cells were dosed with cyclin A/B/E RxL inhibitor (CIRc-001), cyclin A/B RxL inhibitors (CIRc-004; CIRc-028), cyclin-A-selective RxL inhibitor (CIRc-018), cyclin-B-selective RxL inhibitor (CIRc-019) or inactive enantiomer of CIRc-004 (CIRc-005). Roscovitine and staurosporine were used as plate controls to define the top and bottom of the growth inhibition curves, respectively. Inhibitors were dosed in duplicate in either an 8-point (WI-38) or 10-point (SCLC cell lines) 1:3 serial dilution with 10 µM maximum concentration. Roscovitine was dosed in singlet in an 8- or 10-point 1:2 serial dilution with 100 µM maximum concentration. Staurosporine was dosed in singlet in an 8- or 10-point 1:2 serial dilution with 1 µM maximum concentration. After dosing, plates were maintained in tissue culture incubators (37 °C, 5% CO₂) for 3 days (WI-38) or 5 days (SCLC cell lines) to allow for at least two cell doublings before processing in an MTT proliferation assay (R&D systems, 4890-050-K) performed according to the manufacturer’s instructions. The average absorbance value obtained with the highest two concentrations dosed for roscovitine and staurosporine was used for background subtraction. The top of the assay (100% growth) was determined by normalization with the top of the roscovitine curve. The GI₅₀ was determined by nonlinear regression analysis using log(inhibitor) versus response-variable slope (four parameters) using GraphPad Prism (v.10.1.0).

For MTT proliferation assays with the MYT1 inhibitor RP-6306, NCI-H1048, NCI-H446 and NCI-H69 cells were seeded in 96-well plates as described above. The next day, cells were dosed in duplicate in a 10-point 1:3 serial dilution with CIRc-018, CIRc-019, or CIRc-004 alone or in combination with RP-6306 at 100 nM for NCI-H1048 cells or 500 or 1,000 nM for NCI-H446 and 1,000 nM for NCI-H69 cells. After 5 days, cells were processed in an MTT proliferation assay and GI₅₀ values were determined using GraphPad Prism (v.10.1.0) as described above.

Dose–response studies in iHCT-116 cells

Viability of cells were quantified using Cell Titer Glo according to the manufacturer’s instructions (Promega) and the IC₅₀ values were determined using GraphPad Prism (log inhibitor versus response four parameters).

FACS-based cleaved PARP apoptosis assays

Cells were plated at 250,000 cells per ml density and grown in cyclin A/B RxL inhibitor (CIRc-004), MPS1 inhibitor (BAY-1217389) or DMSO at the indicated concentrations for 72 h. Cells were then collected and fixed in 4% paraformaldehyde followed by a PBS wash, and then permeabilized in ice-cold 90% methanol (added dropwise) for 30 min on ice. The cells were then centrifuged at 206g (Eppendorf centrifuge 5804R, A-4-44 rotor, 2,000 rpm) washed once in PBS, centrifuged again at 206g and then incubated with Pacific blue conjugated cleaved PARP antibody (Cell Signaling, Asp214, D64E10, 60068) at a dilution of 1:100 for 1 h at room temperature protected from light. Cells were then washed again in PBS containing 0.5% BSA, resuspended in PBS containing 0.5% BSA and analysed on the LSR Fortessa flow cytometer (Becton Dickinson). Analysis for cleaved-PARP-positive cells was carried out using FlowJo v.10.8.1. The FACS gating strategy is provided in Supplementary Fig. 3.

FACS-based p-histone H3 (Ser10) analysis

Cells were plated at 500,000 cells per ml and grown in cyclin A/B RxL inhibitor (CIRc-004) or DMSO at the indicated concentrations for 24 h. The cells were then washed once in room temperature PBS and then fixed in 4% paraformaldehyde for 15 min at room temperature. The cells were then centrifuged at 206g (Eppendorf centrifuge, 5804R, A-4-44 rotor, 2,000 rpm) for 5 min at room temperature, washed once in PBS, centrifuged again at 206g and then permeabilized with ice-cold methanol at 4 °C for 30 min. The cells were then washed again in PBS and then incubated with Alexa-647-conjugated phospho-histone H3 (Ser10) antibody (Cell Signaling, 3458) at a dilution of 1:100 for 1 h at room temperature, then washed once in PBS containing 0.5% BSA, centrifuged at 400g and then counterstained with DAPI) (Cell Signaling, 4083) for 15 min at room temperature. FACS analysis was then performed to determine the percentage of positive p-histone H3 (Ser10) cells using FlowJo v.10.8.1. The FACS gating strategy is provided in Supplementary Fig. 3.

FACS-based PI cell cycle analysis

Cells were plated at 500,000 cells per ml and grown in cyclin A/B RxL inhibitor (CIRc-004) or DMSO at the indicated concentrations for 24 h. After incubation, cells were washed once in ice-cold PBS and then fixed in ice-cold 80% ethanol (added dropwise) for at least 2 h at −20 °C. The cells were then centrifuged at 206g (Eppendorf centrifuge 5804R, A-4-44 rotor, 2,000 rpm) for 5 min, washed once in PBS, centrifuged again at 206g and then washed again in PBS containing 0.5% BSA. Finally, cells were stained with propidium iodide (BD, 550825) for 15 min at room temperature. FACS analysis for PI was then performed on the LSR Fortessa flow cytometer (Becton Dickinson) and analysis was performed using FlowJo v.10.8.1. The FACS gating strategy is provided in Supplementary Fig. 3.

FACS-based EdU incorporation and DNA content analysis

NCI-H1048 cells were plated at 500,000 cells per ml in a 12-well (Fig. 4a) or 6-well (Fig. 4l and Extended Data Fig. 7b) plate and allowed to adhere overnight. Cells were dosed with cyclin A/B RxL inhibitor (CIRc-004) or DMSO for 24 h. Cells were pulsed for 1 h with 10 µM EdU (Click-It Plus EdU Flow Cytometry Assay kit, Invitrogen, C10632) before collection. After incubation, cells were washed once with ice-cold PBS and fixed by incubation in ice-cold 80% ethanol for at least 2 h at −20 °C. EdU was fluorescently tagged with Alexa Fluor 488 by click reaction (Click-it Plus EdU flow cytometry assay kit, Invitrogen, C10632). DNA content was monitored by FxCycle Violet stain (Invitrogen, F10347) or 1 μg ml⁻¹ DAPI (4,6-diamidino-2-phenylindole, Cell Signaling) for 5 min. For Fig. 4a, flow cytometry analysis for EdU and DNA content was performed by full spectral flow cytometry on the Northern Lights cytometer (Cytek Biosciences) and analysis was carried out using FlowJo v.10.9.0. For Fig. 4l and Extended Data Fig. 7b, flow cytometry analysis for EdU and DNA content was performed using the LSR Fortessa flow cytometer (Becton Dickinson) and analysis was carried out using FlowJo v.10.8.1. FACS gating strategy is given in Supplementary Fig. 3.

FACS-based EdU cell cycle and apoptosis analysis in haematopoietic stem and progenitor cells

G-CSF mobilized human peripheral blood CD34⁺ cells were plated in a 24-well plate at 25,000 cells per ml after counting with a VI-Cell BLU (Beckman Coulter). Cells were then allowed to recover for 48 h before dosing cells with CIRc-004, staurosporine (100 nM) or DMSO for 24 h. Cells were pulsed for 1 h with 10 µM EdU (Click-It Plus EdU Flow Cytometry Assay kit, Invitrogen, C10634) before collection. After incubation, cells were collected, washed once with PBS and resuspended in 1% BSA-PBS before staining with CD34 (8G12) Per-CP (BD, 340666) for 30 min at a 1:10 dilution protected from light. The cells were centrifuged for 200g (Eppendorf centrifuge, 5424R) for 5 min at all wash steps. Cells were then washed in 1% BSA-PBS. Cells were then fixed by resuspending cells in 50 µl of Click-it Fixitive (4% PFA) and incubated for 15 min. Cells were then washed with 1% BSA-PBS and resuspended in permeabilization and wash buffer prepared as directed from Click-It Plus kit (Click-It Plus EdU Flow Cytometry Assay kit, Invitrogen, C10634) for 15 min. EdU was fluorescently tagged with Alexa Fluor 647 by click reaction (Click-It Plus EdU Flow Cytometry Assay kit, Invitrogen, C10634). Cells were then resuspended in permeabilization buffer and stained with FITC mouse anti-cleaved PARP (Asp214) (BD, 558576) for 1 h at a 1:5 dilution protected from light. DNA content was monitored by FxCycle Violet stain (Invitrogen, F10347). Flow cytometry analysis for EdU and DNA content was performed by full spectral flow cytometry on a Northern Lights cytometer (Cytek Biosciences) and analysis was carried out using FlowJo v.10.9.0. The FACS gating strategy is given in Supplementary Fig. 3.

Immunoblotting

Cell pellets were lysed in a modified EBC lysis buffer (50 mM Tris-Cl pH 8.0, 250 mM NaCl, 0.5% NP-40, 5 mM EDTA) supplemented with a protease inhibitor cocktail (Complete, Roche Applied Science, 11836153001) and phosphatase inhibitors (PhosSTOP, Sigma-Aldrich, 04906837001). Soluble cell extracts were quantified using the Bradford protein assay. Then, 20 µg of protein per sample was boiled after adding 3× sample buffer (6.7% SDS, 33% glycerol, 300 mM DTT and Bromophenol Blue) to a final concentration of 1×, resolved by SDS–PAGE using either 10% or 6% SDS–PAGE, semi-dry transferred onto nitrocellulose membranes, blocked in 5% milk in Tris-buffered saline with 0.1% Tween-20 (TBS-T) for 1 h, and probed with the indicated primary antibodies overnight at 4 °C. The membranes were then washed three times in TBS-T, probed with the indicated horseradish peroxidase (HRP)-conjugated secondary antibodies for 1 h at room temperature, and washed three times in TBS-T. Bound antibodies were detected with enhanced chemiluminescence (ECL) western blotting detection reagents (Immobilon, Thermo Fisher Scientific, WBKLS0500) or Supersignal West Pico (Thermo Fisher Scientific, PI34078). The primary antibodies and dilutions used were as follows: mouse anti-β-actin AC-15 (Sigma-Aldrich, A3854, 1:25,000), mouse anti-vinculin hVIN-1 (Sigma-Aldrich, V9131, 1:10,000), rabbit anti-p-CASC5/KNL1 Thr943/Thr1155 D8D4N (Cell Signaling, 40758, 1:1,000), rabbit anti-CASC5/KNL1 E4A5L (Cell Signaling, 26687, 1:1,000), rabbit anti-cyclin B1 D5C10 (Cell Signaling, 12231, 1:1,000), rabbit anti-E2F1 (Cell Signaling, 3742, 1:1,000), CDC20 (Cell Signaling, 4823S, 1:1,000), rabbit anti-p-KAP1 Ser824 (Cell Signaling, 4127, 1:1,000) and Flag (Sigma-Aldrich, A8592). The secondary antibodies and dilutions were as follows: goat anti-mouse (Jackson ImmunoResearch, 115-035-003, 1:5,0000) and goat anti-rabbit (Jackson ImmunoResearch, 111-035-003, 1:5,000).

To detect p-stathmin S38, total stathmin, p-cyclin B1 (Ser126), p-nucleolin (Thr84), p-RPA2 (Ser33) and p-FOXM1 (Thr600), after incubation with the indicated cyclin RxL inhibitors or DMSO control, NCI-H1048 cells were washed once in ice-cold PBS and lysed in 1% Triton X-100 buffer (40 mM HEPES pH 7.5, 1 mM EDTA pH 8.0, 10 mM sodium pyrophosphate, 10 mM β-glycerophosphate, 50 mM sodium fluoride, 120 mM sodium chloride and 1% Triton X-100) supplemented with a protease inhibitor cocktail (cOmplete, Sigma-Aldrich, 5892970001) and phosphatase inhibitors (PhosSTOP, Sigma-Aldrich, 4906837001). Cell lysates were centrifuged at 9,391g (Eppendorf centrifuge 5424R, FA-45-24-11 rotor, 10,000 rpm) for 10 min. The supernatant (cell extract) was removed and the protein content was estimated by performing a Bradford assay. Cell extracts (30 μg of protein) were boiled after adding 4× Laemmli buffer and separated by SDS–PAGE and transferred to PVDF membranes. Membranes were blocked in TBS plus 0.2% Tween-20 (TBS-T) containing 5% skimmed milk and incubated with relevant primary antibodies overnight at 4 °C in blocking buffer containing 5% BSA. After incubation with primary antibodies, the membranes were washed in TBS-T and then incubated with HRP-labelled secondary antibodies for 1–-2 h at room temperature. Membranes were washed again in TBS-T, incubated with ECL reagent (LI-COR, WesternSure, 92695000) and scanned on the Odyssey Fc Analyzer (LI-COR Biosciences). The primary antibodies and dilutions used were as follows: rabbit anti-p-stathmin (Ser38) (Cell Signaling, 3426S, 1:1,000), rabbit anti-total stathmin (Abcam, ab52630), rabbit anti-p-cyclin B (Ser126) (Abcam, ab3488, 1:1,000), rabbit anti-p-nucleolin (Thr84) (Abcam, ab155977, 1:1,000), rabbit anti-p-FOXM1 (Thr600) (Cell Signaling, 14655S, 1:500), rabbit anti-p-RPA2 (Ser33) (Fortis Life Sciences, A300-246A, 1:2,000), rabbit anti-tubulin-HRP (Cell Signaling, 9099S, 1:5,000). The secondary antibodies used and dilutions were as follows: goat anti-rabbit-IgG-HRP (Cell Signaling, 7074S, 1:5,000).

Immunoblots were quantified using Fiji (Image J2, v.2.9.0) or LICOR Odyssey built-in Image Studio Software (v.5.2). Quantified values from immunoblots are provided in Supplementary Tables 8 and 9. Raw immunoblots are included as Supplementary Fig. 1.

Immunoprecipitation analysis

For immunoprecipitation of endogenous cyclin A and cyclin B in Figs. 2o, 3h and 4c and Extended Data Figs. 8u and 9d–f,h, NCI-H1048 cells were seeded at 2.5 × 10⁷ cells per 15 cm plate and were allowed to adhere overnight. The next day, cells were incubated with the indicated cyclin RxL inhibitors at 300 nM or DMSO for 2 h before rinsing once in ice-cold PBS and lysing in 1% NP-40 buffer (40 mM HEPES pH 7.5, 1 mM EDTA pH 8.0, 10 mM sodium pyrophosphate, 10 mM β-glycerophosphate, 50 mM sodium fluoride, 120 mM sodium chloride and 1% NP-40) supplemented with a protease inhibitor cocktail (cOmplete, Sigma-Aldrich, 5892970001) and phosphatase inhibitors (PhosSTOP, Sigma-Aldrich, 4906837001). The cell lysates were centrifuged at 9,391g for 10 min. The supernatant (cell extract) was removed and the protein content was estimated by performing a Bradford assay. Then, 2 mg cell extract was rotated overnight at 4 °C with 2 µg anti-cyclin A (Santa Cruz Biotechnology (SCBT) sc-271682) or 2 µg anti-cyclin B (SCBT sc-245) antibodies. The next day, 20 µl packed protein-A- and protein-G-coated magnetic beads (Dynabeads Protein A and Dynabeads Protein G, Invitrogen, 10002D and 10003D, respectively) were added and lysates were rotated for an additional 1 h at 4 °C. The protein A/G beads were collected using a magnet, washed three times with 1% NP-40 cell lysis buffer, boiled and denatured in 1× Laemmli sample buffer, separated by SDS–PAGE and immunoblotting was performed as described above. The primary antibodies and dilutions used were as follows: anti-MYT1 (Fortis Life Sciences, A302-424A, 1:1,000), rabbit anti-WEE1 (Cell Signaling, 4936, 1:1,000), mouse anti-cyclin B (Cell Signaling, 4135 1:1,000), mouse anti-E2F1 (SCBT, sc-251, 1:1,000), mouse anti-cyclin A (Cell Signaling, 4656, 1:1,000), mouse anti-CDK2 (Origene, TA502935), rabbit anti-CDC2 (Cell Signaling 28439). The secondary antibodies used and dilutions were: mouse anti-rabbit (confirmation specific)-IgG HRP (Cell Signaling, 5127, 1:2,000), goat anti-rabbit IgG-HRP (Cell Signaling, 7074S, 1:5,000), horse anti-mouse IgG-HRP (Cell Signaling, 7076S, 1:5,000), goat anti-mouse-IgE-HRP (Southern Biotech, 1110-05, 1:5,000).

For immunoprecipitations of cyclin B–HA and CDK2 in Fig. 3i and Extended Data Fig. 9k, HEK293T cells were stably infected with pTwist Lenti SFFV Puro WPRE lentiviral vectors (SYNTHESIZED by Twist Biosciences) encoding (1) CCNB1 wild-type-HA, (2) CCNB1 triple mutant-HA (E169K/Y170H/Y177C) or a (3) negative control encoding a functionally inactivated EED mutant-HA with mutations to inactivate EED (F97A/Y148A/Y365A). In total, 2.0 × 10⁷ cells were seeded at 1 × 10⁶ cells per ml density and, the next day, were treated with the cyclin A/B RxL inhibitor (CIRc-004) at 300 nM or DMSO for 2 h before cell collection. Cells were lysed in 700 µl lysis buffer (Cell Signaling, 9803) supplemented with protease inhibitor cocktail (Complete, Roche Applied Science, 11836153001) and phosphatase inhibitors (PhosSTOP, Sigma-Aldrich, 04906837001). Samples were sonicated (Branson Digital Sonifier) three times for 5 s each to ensure homogenization and centrifuged at 14,000g for 10 min at 4 °C (Eppendorf centrifuge 5424R, FA-45-24-11 rotor). The supernatant was precleared by rotating 20 µl protein A agarose beads (Cell Signaling, 9863) for 1 h at 4 °C. The samples were then briefly centrifuged to remove the beads and protein concentration was determined. Then, 60 µl of lysate was removed as input (~10% total for ~2.5% input per immunoblot) and boiled after adding 3× sample buffer (6.7% SDS, 33% glycerol, 300 mM DTT and Bromophenol Blue) to a final concentration of 1×. Then 300 µl of lysate was immunoprecipitated by rotating overnight at 4 °C with rabbit anti-HA epitope (Cell Signaling, 3724, 1:100) for exogenous cyclin B, rabbit anti-CDK2 E8J9T (Cell Signaling, 18048, 1:100) or equivalent amount of rabbit (DA1E) monoclonal antibody IgG XP isotype control (Cell Signaling, 3900). The next day, 20 µl protein A agarose beads was added to each sample tube and incubated for 1 h at 4 °C with rotation. The beads were collected by centrifugation and washed five times with cold lysis buffer followed by boiling in 3× sample buffer. The samples were run on SDS–PAGE and immunoblotting was performed as described above. Primary antibodies used for immunoblotting were as follows: rabbit anti-cyclin B1 D5C10 (Cell Signaling, 12231, 1:1,000), rabbit anti-CDK2 E8J9T (Cell Signaling, 18048, 1:1,000) and mouse anti-HA.11 epitope tag (BioLegend, 901501, 1:1,000).

For immunoprecipitation of endogenous cyclin A with E2F3 in Extended Data Fig. 10f, NCI-H1048 cells were seeded at 2.5 × 10⁷ cells per 15 cm plate and allowed to adhere overnight. The next day, cells were incubated with the indicated cyclin RxL inhibitors at 300 nM or DMSO for 2 h before rinsing once in ice-cold PBS and lysing in 1% Triton buffer (40 mM HEPES pH 7.5, 1 mM EDTA pH 8.0, 10 mM sodium pyrophosphate, 10 mM β-glycerophosphate, 50 mM sodium fluoride, 120 mM sodium chloride and 1% Triton X-100) supplemented with a protease inhibitor cocktail (cOmplete, Sigma-Aldrich, 5892970001) and phosphatase inhibitors (PhosSTOP, Sigma-Aldrich, 4906837001). Cell lysates were centrifuged at 9,391g for 10 min. The supernatant (cell extract) was removed and the protein content estimated by performing a Bradford assay. Anti-cyclin A antibody (SCBT, sc-271682) was covalently coupled to M-270 Expoxy dynabeads (Dynabeads Antibody Coupling Kit, Invitrogen 14311D) at a ratio of 10 μg antibody per mg of beads and rotated overnight at 37 °C. Next, 2 mg cell extract was then rotated overnight at 4 °C with 1 mg coupled dynabeads. The dynabeads beads were collected using a magnet, washed three times with 1% Triton cell lysis buffer, boiled and denatured in 1× Laemmli sample buffer, separated by SDS–PAGE, and immunoblotting was performed as described above. The primary antibodies were mouse anti-E2F3 (SCBT, sc-56665, 1:1,000), and mouse anti-cyclin A (Cell Signaling, 4656, 1:1,000). The secondary antibodies were horse anti-mouse IgG-HRP (Cell Signaling, 7076S, 1:5,000) and goat anti-mouse-IgE-HRP (Southern Biotech, 1110-05, 1:5,000).

Nocodazole cell synchronization experiments

To synchronize cells in mitosis, as depicted in Fig. 3k, a preliminary evaluation was conducted to determine the minimal nocodazole concentration required to induce mitotic arrest and the optimal time post-washout for facilitating the re-entry of the majority of cells into the cell cycle. In brief, 500,000 cells per ml of NCI-H69 cells were seeded in six-well plates and treated with increasing concentrations of nocodazole (60, 120, 240 ng ml⁻¹) or DMSO as a control. After overnight incubation, cells were collected, fixed and subjected to propidium iodide-based FACS to assess mitotic arrest. A nocodazole concentration of 60 ng ml⁻¹ was sufficient to induce mitotic arrest in around 90% of cells. To determine the optimal wash-out time for mitotic exit, cells were plated at 500,000 cells per ml density and treated with 60 ng ml⁻¹ nocodazole. After overnight incubation, nocodazole-containing media was removed, and cells were washed three times with DPBS (Life Technologies, 14190250). Fresh warm medium was then added, and cells were collected at 0, 3, 6 and 12 h after washout. A 6 h incubation period was optimal to allow the majority of cells to complete mitosis and re-enter the subsequent cell cycle. For the immunoblot analysis shown in Fig. 3k, cells were seeded in 6 cm tissue culture dish at a density of 500,000 cells per ml and incubated overnight with nocodazole (60 ng ml⁻¹). The next day, cells were washed three times with PBS and incubated with CDK2i (PF-07104091, 500 nM), cyclin A/B RxL inhibitor (CIRc-004, 20 nM), nocodazole (60 ng ml⁻¹) or the combinations as indicated using DMSO as control. At 6 h after incubation, cells were collected for immunoblot analysis.

Histone extractions and histone immunoblot analysis

For histone extractions in Fig. 4, cells were counted on day 0 using the Vi-Cell XR Cell Counter and plated in 6 cm plates at 250,000 cells per ml in 4 ml of medium per plate for all cell lines along with indicated concentration of cyclin A/B RxL inhibitor. Then, 3 days later, cells were pelleted by centrifugation at 400g for 3 min at 4 °C in a 15 ml conical tube. The cell pellets were then washed once in 1 ml of ice-cold PBS, transferred to a 1.5 ml Eppendorf tube and pelleted by centrifugation at 400g for 3 min at 4 °C. The PBS was removed by gentle aspiration and the soluble proteins from the cell pellets were extracted by incubation in nucleus lysis buffer (250 mM sucrose, 60 mM KCL, 15 nM NaCl, 15 mM Tris pH 7.5, 5 mM MgCl₂, 1 mM CaCl₂, 1 mM DTT, 0.3% NP-40 supplemented with a cocktail of protease inhibitor cocktail and phosphatase inhibitors) for 10 min at 4 °C. The extracts were then centrifuged at 10,000g for 1 min at 4 °C and the supernatant was removed by aspiration. The cell pellets were again resuspended in nucleus lysis buffer and incubated for 10 min at 4 °C and then centrifuged at 10,000g for 1 min at 4 °C. The supernatant was removed by aspiration and the histones in the insoluble pellet were extracted by overnight incubation in 60 µl of 0.2 N HCl followed by centrifugation at 16,800g for 15 min at 4 °C. The supernatant was transferred to a fresh 1.5 ml Eppendorf tube and quantified using the Bradford protein assay. The pH was neutralized by adding 1/5 volume of 1 N NaOH, as confirmed by the colour of the samples after the addition of sample buffer containing Bromophenol Blue. Then, 1 μg of protein in sample buffer was boiled and resolved by SDS–PAGE on a 15% SDS–PAGE gel. Immunoblot analysis was performed as described above. The primary antibodies were as follows: rabbit anti-histone-3 DH12 (Cell Signaling, 4499, 1:5,000) and rabbit anti-p-histone H2A.X Ser 139 (Cell Signaling, 2577, 1:1,000). For all experiments, total histone H3 protein was run as a sample processing control on a separate gel owing to the low amount of protein lysate loaded onto the gel (1 μg) for histone blots.

Cancer cell line sensitivity screening with cyclin A/B RxL inhibitors

A panel of 46 SCLC cell lines was tested for their sensitivity to cyclin A/B RxL inhibition in a Cell Titer Glo (CTG) proliferation assay performed at Shanghai ChemPartner (Fig. 1d). Cells were plated in 384-well format and dosed in duplicate with CIRc-004 in a 10-point, 1:3 serial dilution dose–response assay with a 10 μM starting concentration. Staurosporine was used as an assay control per cell line and was dosed in duplicate from 2 μM maximum concentration in a 10-point 1:3 serial dilution. Cells were exposed to inhibitor for 4–8 days depending on the length of time required for at least two cell doublings to occur. A time 0 h read and an end-point DMSO control read were obtained per cell line for calculation of GI₅₀ values. SCLC panel GI₅₀ values are included in Supplementary Table 1.

CIRc-001 was profiled in Horizon Discovery’s High Throughput OncoSignature screening platform containing 302 cell lines across 18 indications (Extended Data Fig. 2a). Cells were seeded in 384-well tissue culture plates at 250–1,500 cells per well. Assay plates were dosed in triplicate with inhibitor in a 9-point, 1:3 serial dilution dose–response assay with 10 μM maximum concentration. Cells were incubated for 3 days and then analysed using CellTiter-Glo 2.0 (Promega). At the time of treatment, a set of assay plates that did not receive treatment were collected and ATP levels measured by adding CellTiter-Glo 2.0 (Promega). All datapoints were collected by automated processes and were subject to quality control. Potency and efficacy metrics were derived from logistics curves fitted to growth inhibition using Horizon’s Chalice software. Horizon Discovery OncoSignature panel cell line GI₅₀ values are provided in Supplementary Table 2.

GSEA

To identify differentially expressed gene sets associated with sensitivity and resistance to CIRc-001 across the Horizon Discovery OncoSignature cell line panel, and to CIRc-004 across the SCLC panel (Fig. 1e), GSEA was first performed between sensitive and resistant cell lines using the MSigDB hallmark pathways using mRNA expression data obtained from the Broad Institute’s Cancer Cell Line Encyclopaedia (CCLE) database (2019 release: DepMap 2019q3). Sensitive and resistant are defined as the top and bottom quartile of the area over the curve to identify hallmark pathways that were differentially enriched in sensitive versus resistant lines based on FDR < 0.05 For the Horizon Discovery OncoSignature cell line panel, 42 out of 46 SCLC lines and 288 out of 302 cell lines across Horizon Discovery OncoSignature cell line panel had mRNA expression that was available for the analysis.

The GSVA method³³, using the MSigDb Hallmark collection of RNA-seq data, was then used to calculate GSVA scores for E2F targets and G2/M checkpoint pathway using the GSVA Bioconductor package (v.1.50.5). To visualize the top selected pathways from GSEA analysis, gene expression data were first aggregated as GSVA scores based on gene set annotations obtained from MSigDB. For each pathway, pathway scores were transformed to z scores across samples to standardize the data. The heat map was constructed using the ComplexHeatmaps package (v.2.18.0) in R (v.4.3.2), where rows represented Hallmark pathways, columns represented samples and the colour intensity reflected the mean z score of each pathway for each sample and was hierarchically clustered by pathway to reveal patterns of enrichment. For Fig. 1e (right), the median E2F target or G2M pathway score across the SCLC panel was used as the threshold to define low versus high groups and P values were calculated using Mann–Whitney t-tests.

CRISPR–Cas9 resistance screen with cyclin A/B RxL inhibitor, cyclin A/B/E RxL inhibitor and CDK2i

On day 0, NCI-H1048 cells were counted using the Vi-Cell Counter. In total, 1.50 × 10⁸ cells (which would yield representation of at least 500 cells per sgRNA if multiplicity of infection (MOI) = 0.3) were pelleted and resuspended at 2 × 10⁶ cells per ml in complete medium containing 50 µl ml⁻¹ of human genome-wide Brunello sgRNA library (CP0043) (purchased from the Broad Institute) lentivirus, which contains both the sgRNA and Cas9, and 8 µg ml⁻¹ polybrene. The lentiviral titre was determined empirically in pilot experiments with a goal MOI of 0.3–0.5. CP0043 contains 77,441 sgRNAs targeting 4 sgRNAs per gene with 1,000 non-targeting sgRNAs controls. The cells mixed with polybrene and lentivirus were then plated in 1 ml aliquots onto 12-well plates and centrifuged at 931g (Allegra X-15R Centrifuge (Beckman Coulter), rotor SX4750A, 2,000 rpm), for 2 h at 30 °C. In parallel, 2 × 10⁶ cells were also centrifuged under the same conditions but without lentivirus as a control for puromycin selection (Mock). Around 16 h later (day 1), the cells were collected, pooled and centrifuged again at 448g to remove the lentivirus and polybrene, and the cell pellet was resuspended in complete medium and plated into twenty-three 15 cm tissue-culture-treated plates at 0.3 × 10⁶ cells per ml or in 6-well plates at the same density for the control (mock) cells. The cells were then cultured for 72 h at which time (day 4) the cells were counted and plated at 0.4 × 10⁶ cells per ml in 15 cm tissue-culture-treated plates with fresh medium containing puromycin (0.25 µg ml⁻¹) to select for puromycin-transduced cells. A parallel experiment was performed on day 4 to determine the MOI of the screen. To do this, cells infected with the sgRNA library or mock-infected cells were plated at 0.4 × 10⁶ cells per ml in 6-well plates in the presence or absence of puromycin (0.25 µg ml⁻¹). After 72 h (day 7), cells were counted using the Vi-Cell XR Cell Counter and the MOI was calculated using the following equation: (number of puromycin-resistant cells infected with the sgRNA library/total number of cells surviving without puromycin after infection with the sgRNA library) − (number of puromycin-resistant mock-infected cells/total number of mock-infected cells). The calculated MOI was 0.33. On the same day (day 7), the puromycin-containing medium was exchanged with fresh complete medium not containing puromycin.

On day 10, cells were pooled, counted and split into two replicates. Each replicate was split into four arms: (1) CIRc-004 (cyclin A/B RxL inhibitor, 200 nM); (2) CIRc-001 (cyclin A/B/E RxL inhibitor, 200 nM); (3) CDK2i (PF-07104091, 500 nM); or (4) CIRc-005 (inactive enantiomer of CIRc-004, 200 nM). CIRc-005 had no effect on cellular proliferation (Extended Data Fig. 2d) and therefore was used as a negative control comparison for all drug treated arms. A total of 4 × 10⁷ cells were plated at 0.2 × 10⁶ cells per ml in 15 cm tissue culture plates in complete medium containing the respective drugs maintaining 500 cells per sgRNA. At the same time, the remaining cells were collected and divided in aliquots of 4 × 10⁷ (again to maintain representation of 500 cells per sgRNA), washed in PBS and cell pellets were frozen for genomic DNA isolation for the ETP before drug selection. Mock-infected cells were maintained at the same density for all drug conditions and carried out in parallel for the entire screen described below.

The medium on the plates under drug selection was exchanged every 3 days with complete medium containing fresh drug. Treatment arms were counted and passaged when cells were about 80% confluent and split back to a representation of 4 × 10⁷ cells at a density of 0.2 × 10⁶ cells per ml onto 15 cm tissue culture plates. This was done for the CIRc-005 inactive enantiomer every 3 days as CIRc-005 had no effect on cellular proliferation, and it happened only once or twice during the screen for all of the other drug treatment conditions as the drugs exhibited strong cytotoxicity. When cells were passaged, all cells were pooled, counted and replated maintaining 500 cells per sgRNA. The remaining cells were divided into aliquots of 4 × 10⁷ cells, washed in PBS and cells pellets were frozen for genomic DNA isolation. The screen was terminated on day 26 (LTP) when there was an obvious growth advantage in the CIRc-004, CIRc-001, CDK2i arms infected with the sgRNA library relative to the mock infected cells treated with the same compounds.

After completion of the screen, genomic DNA was isolated using the Qiagen Genomic DNA maxi prep kit (51194) according to the manufacturer’s protocol. Raw Illumina reads were normalized between samples using: log₂[(sgRNA reads/total reads for sample) × 1 × 10⁶) + 1]. The initial common ETP (day 10) was then subtracted from the LTP after drug treatment (day 26) to determine the relative enrichment of sgRNAs in the CIRc-004, CIRc-001 and CDK2i arms relative to the CIRc-005 inactive enantiomer control arm for each replicate. Apron analysis, hypergeometric analysis and STARS analysis (GPP web portal, https://portals.broadinstitute.org/gpp/screener/) were performed using the normalized counts of the drug-treated arms LTP–ETP (Apron) or the log-normalized counts of the drug-treated arms LTP–ETP (hypergeometric, STARS) comparing the drug-treated condition to the CIRc-005 enantiomer control at the matched LTP or compared to the ETP. The averaged data from the two replicates were used for all analyses shown in the manuscript. All raw data and outputs of each analysis listed above are provided in Supplementary Table 3.

Dose–response assays for screen hit validation with CRISPR–Cas9 sgRNA knockout

For validation studies, we focused on the cyclin A/B RxL inhibitor (CIRc-004) and not CIRc-001, as CIRc-004 spared cyclin E and, perhaps as a result, demonstrated modestly increased cellular potency. NCI-H1048 and NCI-H446 cells were first infected with pLentiCRISPR V2-Puro lentiviruses encoding sgRNAs targeting the selected screen hits and selected with puromycin. Stably infected cells were counted on day 0 using a Vi-Cell XR Cell Counter and plated onto tissue-culture-treated 6-well plates at 50,000 cells per ml in 2 ml of complete medium. For dose–response assays with CIRc-004 and the CDK2i (PF-07104091), the cells were treated with the compounds at the indicated concentrations for 6 days and counted using the Vi-Cell system. Normalized cell counts were calculated as day 6 counts for treated samples/day 6 counts for DMSO sample. Analysis was performed using GraphPad Prism v.10.0.0.

Forward genetic screen with cyclin A/B RxL inhibitor in iHCT-116 cells

The Nijhawan laboratory developed a forward genetic screening system using an engineered colorectal cancer cell line, termed iHCT116, that identifies mutations that confer compound resistance using HCT116 cells engineered for inducible protein degradation (with the addition of IAA) of mismatch repair protein, MLH1³⁷. As a result, cells cultured without IAA have a lower mutation rate (mut-low) than cells cultured with IAA (mut-high). These cells are diploid and, therefore, resistance mutations are more likely to be the result of a gain of function. For the forward genetic screen, 5 million iHCT116 cells that had previously been treated with vehicle or indole acetic acid were cultured in 10 μM of CIRc-004 for 17 days, with the medium replenished every 3 days. Six and twelve clones from the mut-low and mut-high conditions, respectively, were isolated. The surviving clones were analysed for resistance to CIRc-004 in a 3-day proliferation assay in which CIRc-004 was added in dose–response with a final concentration of DMSO of 0.5%. The EC₅₀ for all clones were at least 100-fold higher than parental cells. All clones tested were resistant to an unrelated toxin, MLN4924, suggesting that non-specific mechanisms of CIRc-004 resistance were unlikely (Extended Data Fig. 5d). After evaluating the barcode sequence in all 18 clones, 5 and 12 founder clones were identified in the mut-low and mut-high condition, respectively (Extended Data Fig. 5e). Eight distinct clones were selected for exome sequencing. Barcode sequences were PCR amplified and sequenced as described previously³⁷. Genomic DNA was collected from select clones (Qiagen) and processed for whole-exome sequencing. Mutations unique to individual clones were identified using the previously described analysis³⁷. Only four genes were mutated in more than 1 of 8 clones and heterozygous CDC20 mutations were found in 5 of 8 clones.

Rescue experiments with MPS1 inhibitor

For dose–response assays with MPS1 inhibitor (BAY-1217389) to empirically identify concentrations of the MPS1 inhibitor that effectively blocked MPS1 activity (measured by immunoblotting for p-KNL1) without gross effects on cellular proliferation, the cells were counted on day 0 using the Vi-Cell XR Cell Counter and plated in a tissue-culture-treated 6-well plate at 250,000 cells per ml in 2 ml of complete medium. Cells were treated with the drug (BAY-1217389) at the indicated concentrations for 3 days. The normalized cell count was calculated as (day 3 count for treated samples)/(day 3 count for DMSO treated sample). Analysis was performed using GraphPad Prism v.10.0.0. These experiments identified 3 nM of BAY-1217389 blocks p-KNL1 without grossly affecting proliferation.

For rescue experiments with the MPS1 inhibitor, cells were counted using the Vi-Cell XR Cell and plated at 250,000 cells per ml for NCI-H1048 cells or at 50,000 cells per ml for NCI-H69 and NCI-H446 cells and then treated with cyclin A/B RxL inhibitor (CIRc-004) at the indicated concentrations with or without MPS1 inhibitor (BAY-1217389 at 3 nM) and collected after 3 days for dose–response assays using the Vi-Cell system. For FACS-based cleaved PARP apoptosis assay, cells were counted and plated at 250,000 cells per ml and then treated with CIRc-004 at the indicated concentrations with or without MPS1 inhibitor (BAY-1217389 at 3 nM) and, 3 days later, were collected, stained and analysed as described above.

For immunoblot analysis of p-KNL1 levels after BAY-1217389 treatment, about 2 million cells were counted using the Vi-Cell XR Cell and plated at 500,000 cells per ml and then treated with CIRc-004 at the indicated concentrations with or without BAY-1217389 at 3 nM, and collected 24 h later for immunoblot analysis.

Immunofluorescence

Cells were seeded onto glass coverslips at 250,000 cells per ml and treated with CIRc-004 as indicated. Coverslips were fixed with ice-cold 100% methanol for 15 min on ice. Cells were washed with PBS and incubated in blocking buffer (5% BSA in PBS) for 1 h. Coverslips were then incubated with the following primary antibodies diluted in fresh blocking buffer: rabbit anti-p-histone H2A.X Ser139 (Cell Signaling, 2577, 1:400), rabbit anti-p-CASC5/KNL1 Thr943/Thr1155 D8D4N (Cell Signaling 40758, 1:400), ACA (Antibodies, 15-235, 1:500) and incubated overnight at 4 °C. Cells were rinsed three times with PBS + 0.05% Triton X-100 and then incubated with secondary antibodies: Alexa Fluor 568 goat anti-rabbit (Thermo Fisher Scientific, A11011, 1:300) and Alexa Fluor 647 goat anti-human (Thermo Fisher Scientific, A21445, 1:300) for 1 h at room temperature. The coverslips were then washed three times in PBS containing 0.05% Triton X-100. Cells were then incubated with 1 μg ml⁻¹ DAPI (Cell Signaling) for 5 min. After rinsing with PBS coverslips, the coverslips were mounted onto glass slides with ProLong Glass mounting reagent (Invitrogen, P36930). Images were acquired using the Zeiss LSM980 laser-scanning microscope (Oberkochen) with ZEN 2.3 SP1 software. Image analysis was performed using Image J2 v.2.9.0.

Fluorescence quantification of p-KNL1 confocal images

Image quantification was performed using Fiji (Image J2 v.2.9.0). In brief, regions of interest were defined by identifying mitotic cells and drawing an area centred on the DAPI stain. Background values measured in an adjacent region were subtracted and the resulting corrected total cell fluorescence (CTCF) was calculated using the formula CTCF = (integrated density) − (area of selected cell × mean fluorescence of background). Each individual mitotic cell CTCF was plotted in a dot plot and CIRc-004-treated cells were compared with vehicle-treated cells.

Live-cell imaging

Live-cell imaging was performed on a Ti2 inverted microscope fitted with a CSU-W1 spinning-disk system (Nikon). To perform live-cell imaging, NCI-H1048 cells were transduced to stably express both mCherry-PCNA and PGK-H2B-eGFP (Addgene, 21210). Two rounds of FACS sorting were performed to obtain a pure population of eGFP and mCherry double-positive cells. To capture mitotic events for these experiments, we focused our live-cell imaging experiments on H2B–eGFP. Cells were seeded onto 35 mm µ-Dish (iBiDi, 80137) at 500,000 cells per ml and allowed to attach and grow for 48 h. CIRc-004 (20 nM) was added 2 h before live-cell imaging. z stacks (+6 (above) and −4 (below) planes at 0.5 μm spacing) were captured every 5 min for 10 h, using a Zyla 4.2 sCMOS camera (Andor), and a ×20/0.95 NA Plan Apochromat Lambda objective with the correction collar set to 0.17. An environmental enclosure was used to maintain cell culture conditions (37 °C and humidified 5% CO₂) for all live-cell confocal imaging. Images were analysed on NIS Elements Viewer v.4.2 (Nikon). The timing of mitosis was measured from DNA condensation (beginning timepoint) to cytokinesis completion (end timepoint).

For FUCCI reporter LIC, RPE1 cells were stably transduced with Incucyte cell cycle green/red construct (47741, Sartorius). After selection with puromycin (2 µg ml⁻¹), cells were seeded in a 96-well plate at 1,000 cells per well. CIRc-004 (2,000 nM) was added before initiating live-cell imaging. Plates were imaged every 20 min for 24 h. Imaging and analysis was performed using Incucyte Zoom/S3 live-cell imagers (Essen Biosciences) and quantified using the Incucyte analysis software.

CRISPR–Cas9 base editor screens with the cyclin A/B RxL inhibitor

To establish the adenosine or cytosine base editor Cas9-expressing cell line for the screen with the tiling library, NCI-H1048 cells were transduced with pRDA_867 (A>G base editor Cas9) and pRDB_092 (C>T base editor Cas9), respectively. Successfully transduced cells were selected with blasticidin (5 µg ml⁻¹) to make stable adenosine or cytosine base editor cell lines.

The base editor library (CP1985) was custom designed using the Broad Institute’s CRISPR base editor library designer Beagle (https://portals.broadinstitute.org/gppx/beagle/screener) to include a total of 2,802 sgRNAs tiling CCNB1, NM_031966.4 (572 sgRNAs), CCNA2 NM_001237.5 (577 sgRNAs), CDK2 NM_001798.5 (476 sgRNAs) and CDC20 NM_001255.3 (897 sgRNAs). As controls the library contained 56 guides against essential genes (EEF2, DBR1, PLK1, KIF11, GAPDH, PCNA, RPA3, RPL3, POLR2B and PSMB1), 112 guides against intergenic sites and 112 guides as non-targeting controls.

First a pilot experiment was performed to empirically determine the lentiviral titre needed for a goal MOI of 0.3–0.6. Cells were mixed with polybrene and increasing amounts of the lentiviral CP1985 sgRNA library and plated in 1 ml aliquots onto 12-well plates and centrifuged at 931g (Allegra X-15R Centrifuge (Beckman Coulter), rotor SX4750A, 2,000 rpm), for 2 h at 30 °C. In parallel, 2 × 10⁶ cells were also centrifuged under same conditions but without lentivirus as a control for puromycin selection (Mock). Then, about 16 h later (day 1), the cells were collected, pooled and centrifuged again at 448g to remove the lentivirus and polybrene, and the cell pellet was resuspended in complete medium and plated into 15 cm tissue-culture-treated plates at 0.4 × 10⁶ cells per ml or in 6-well plates at the same density for the control (mock) cells. The cells were then cultured for 72 h at which time (day 4) the cells were counted and plated at 0.4 × 10⁶ cells per ml in 15 cm tissue-culture-treated plates with fresh medium containing puromycin (0.5 µg ml⁻¹) to select for puromycin-transduced cells. After 72 h (day 7), cells were counted using the Vi-Cell XR Cell Counter and the MOI was calculated using the following equation: (number of puromycin-resistant cells infected with the sgRNA library/total number of cells surviving without puromycin after infection with the sgRNA library) − (number of puromycin-resistant mock-infected cells/total number of mock-infected cells). These experiments demonstrated that 25 µl ml⁻¹ of CP1985 virus was needed to achieve a MOI of about 0.3–0.6.

For the screen, adenosine or cytosine base editor Cas9-expressing NCI-H1048 cell lines were counted using the Vi-Cell counter on day 0. Two replicates of 2.4 × 10⁷ cells (which would yield representation of at least 2,000 cells per sgRNA if MOI = 0.3) were pelleted and resuspended at 2 × 10⁶ cells per ml in 1 ml of complete medium containing 25 µl ml⁻¹ of the pooled sgRNA library (CP1985) lentivirus and 8 µg ml⁻¹ polybrene. A parallel experiment was performed on day 4 to determine the actual MOI of the screen as described above, which showed that the MOI of the screen was about 0.6. On the same day (day 7), puromycin-containing medium was exchanged with fresh complete medium not containing puromycin for all cells in the screen.

On day 10, cells from each replicate were pooled, counted and then split into two additional replicates (four replicates in total). In total, 6 × 10⁶ cells from each replicate were collected for an ETP for representation of 2,000 cells per sgRNA. This was done by washing cells in PBS and freezing cell pellets at −80 °C for genomic DNA isolation. The remaining cells from each replicate were then split into two treatment arms: (1) CIRc-004 (cyclin A/B RxL inhibitor, 200 nM) and (2) CIRc-005 as a negative control and comparator (inactive enantiomer of CIRc-004, 200 nM) with 4.5 × 10⁶ cells in each treatment arm plated in 15 cm tissue culture plates at 0.2 × 10⁶ cells per ml in complete medium containing the respective drugs listed above to maintain representation of 1,500 cells per sgRNA. Mock-infected cells were maintained at the same density for all drug conditions and carried in parallel for the entire screen described below.

The medium on the plates under drug selection was exchanged every 3 days with complete medium containing fresh drug at the concentrations described above. Treatment arms were counted and passaged when cells were about 80% confluent and split back to a representation of 4.5 × 10⁶ cells at a density of 0.2 × 10⁶ cells per ml onto 15 cm tissue culture plates. This occurred for the CIRc-005 inactive enantiomer every 3 days as CIRc-005 had no effect on cellular proliferation. As the CIRc-004 exhibited strong cytotoxicity, this treatment arm required medium exchange only every 3 days. When cells were passaged, the cells were pooled, counted and replated maintaining 1,500 cells per sgRNA. The remaining cells were divided into aliquots of 6 × 10⁶ cells, washed in PBS and the cell pellets were frozen for genomic DNA isolation. The screen was terminated on day 26 when there was an obvious growth advantage in the CIRc-004 arm infected with the sgRNA library relative to the mock infected cells treated with the same compounds.

After completion of the screen, genomic DNA was isolated using the Qiagen Genomic DNA midi prep kit (51183) and mini prep kit (51106) according to the manufacturer’s protocol. Raw Illumina reads were normalized between samples using log₂[(sgRNA reads/total reads for sample) × 1 × 10⁶) + 1]. LFC values in Supplementary Fig. 2a,b were calculated by subtracting the plasmid DNA (pDNA) log-normalized value from the CIRc-005 LTP (day 26) log-normalized value and this was used for replicate reproducibility and dropout of essential sgRNA versus negative-control sgRNAs. To calculate the z-scored LFC values in Fig. 3b and Supplementary Fig. 2c–e, we first subtracted the log-normalized value of the negative enantiomer control (CIRc-005) at the LTP (day 26) from the log-normalized value of the cyclin A/B RxL inhibitor (CIRc-004) at the LTP (day 26). This LFC value (CIRc-004 minus CIRc-005) for each sgRNA was z-scored, a transformation that factors in the performance of the control guides, using the following equation: z score = (x − μ)/σ, where x, μ and σ correspond to the LFC of an individual guide, the mean LFC of all negative control guides and the s.d. of all control guides, respectively. These z-scored LFC values for each sgRNA were then matched to the annotated CP1985 library where Beagle (Broad Institute) was used to predict the expected nucleotide edit and plotted for each tiled gene as indicated in each figure panel listed above. The averaged data from the four replicates were used for all analyses shown in the manuscript. All raw data and outputs of each analysis listed above are provided in Supplementary Table 5.

CRISPR–Cas9 base editor screen with the cyclin A/B RxL inhibitor with focused CCNB1 target validation

To validate the exact mutations enriched after CIRc-004 treatment in CCNB1 corresponding to the amino acids 117–285, which contained several enriched CCNB1 variants from our original base editor screen in Fig. 3b, we repeated the screen as in Fig. 3a at scale with a representation of 1,500 cells per sgRNA, and at the LTP (day 26), mRNA was isolated from the cyclin A/B RxL inhibitor (CIRc-004)-treated and enantiomer control (CIRc-005)-treated arms using the Quick-RNA Miniprep kit (Zymo Research) according to the manufacturer’s instructions. The RNA concentration was determined using the Nanodrop 8000 (Thermo Fisher Scientific) system. A cDNA library was synthesized using iScript Reverse Transcription Supermix for RT-qPCR (BioRad, 1708841) according to the manufacturer’s instructions with 3,000 ng of RNA. The target site on CCNB1 was amplified by two-step PCR using KOD XTREME HOT START DNA POLYMERASE (Sigma-Aldrich, 71975-3) using 25% of the total cDNA reaction above. For the first PCR step, CCNB1 1F, 5′-GGAACGGCTGTTGGTTTCTG-3′; and CCNB1 1R, 5′-AAGTAAAAGGGGCCACAAGC-3′ primers were used to amplify a 1,514 bp region. Then, 0.2 μl of the first PCR reaction was used for a second round of PCR amplification using the following primers: 2F, 5′-CGCCTGAGCCTATTTTGGTTG-3′; and CCNB1 2R, 5′-CCATCTGTCTGATTTGGTGCTTAG-3′ to yield a 510 bp product. This final PCR product was column purified using the QIAquick Gel Extraction kit (Qiagen, 28704) and sent for amplicon sequencing using next-generation sequencing by the MGH DNA Core Facility and analysed using MGH DNA Core’s complete amplicon sequencing pipeline for data analysis of >600 bp amplicons using the following CCNB1 NM_031966.4 refseq amplicon: CGCCTGAGCCTATTTTGGTTGATACTGCCTCTCCAAGCCCAATGGAAACATCTGGATGTGCCCCTGCAGAAGAAGACCTGTGTCAGGCTTTCTCTGATGTAATTCTTGCAGTAAATGATGTGGATGCAGAAGATGGAGCTGATCCAAACCTTTGTAGTGAATATGTGAAAGATATTTATGCTTATCTGAGACAACTTGAGGAAGAGCAAGCAGTCAGACCAAAATACCTACTGGGTCGGGAAGTCACTGGAAACATGAGAGCCATCCTAATTGACTGGCTAGTACAGGTTCAAATGAAATTCAGGTTGTTGCAGGAGACCATGTACATGACTGTCTCCATTATTGATCGGTTCATGCAGAATAATTGTGTGCCCAAGAAGATGCTGCAGCTGGTTGGTGTCACTGCCATGTTTATTGCAAGCAAATATGAAGAAATGTACCCTCCAGAAATTGGTGACTTTGCTTTTGTGACTGACAACACTTATACTAAGCACCAAATCAGACAGATG.

To analyse variants from the CCNB1 refseq above, the percentage of mismatched reads at individual base positions were normalized to the percentage of sequencing reads to get the percentage of variants at each position and then averaged across two independent experiments. Then the enrichment of percentage of variants in the CIRc-004-treated arm was compared to the variants in the CIRc-005-treated arm to determine the fold change in CIRc-004 versus CIRc-005 at the LTP. All data outputs are in Supplementary Table 5.

AlphaFold protein co-folding prediction

Co-folding of proteins were done by AlphaFold 2 implemented in ColabFold⁷⁶. The default parameters were used to predict their relative positions (parameters: msa_mode (MMseqs2_UniRef_Environmental), pair_mode (unpaired_paired), model_type (auto), number of cycles (3), recycle_early_stop_tolerance(auto), relax_max_iterations (200), and pairing_strategy (greedy)). The resulting co-folding structure along with the PAE files was analysed using ChimeraX software⁷⁷. The command ‘alphafold contacts /A to /B distance 3’ was used to identify residues potentially interacting within a putative predicted distance of less than 3.0 Å.

Modelling the impact of cyclin B mutations on CIRc-004 binding

Schrodinger software suite (2024-3 version) was used to conduct modelling studies on cyclin B. Initially, the binding model of CIRc-004 to wild-type cyclin B1 was generated using the crystal structure of the cyclin B–p-CDK2 complex (PDB: 2JGZ). The crystal structure was superimposed onto the cyclin A–CIRc-004 model by performing backbone alignment using the Protein Structure Alignment tool in Maestro, and was prepared using the Protein Preparation Workflow with the default settings. Starting from the binding pose in the cyclin A model, CIRc-004 was minimized in the cyclin B’s RxL site using Prime MM-GBSA with the VSGB solvent model, OPLS4 force field and a rigid protein, followed by docking using Glide SP with the ‘refine-only’ option. The resulting binding model was then used as the input structure for free-energy perturbation (FEP) calculations^78,79 to assess the binding of CIRc-004 to various single and double mutants of cyclin B. Using the protein FEP for ligand selectivity module, the relative binding free energies of CIRc-004 to various cyclin B mutants were computed with the wild-type cyclin B treated as the reference. The default settings were used for the FEP calculations with the SPC water model and OPLS4 force field.

Generation of endogenous CCNB1 mutations using Cas9 base editing

sgRNA sequences targeting specific base edits in CCNB1 were selected from the CP1985 sgRNA library based on z-scored LFC enrichment after CIRc-004 treatment in our primary screens, synthesized by IDT technologies and cloned using the same approach as for sgRNA oligo cloning described above. These sgRNAs were integrated into the GFP-expressing pMV_AA013 lentiviral vector (Broad Institute) using sgRNA Cloning method described above. NCI-H1048 cells, expressing either pRDA_867 (A>G base editor Cas9) or pRDB_092 (C>T base editor Cas9), were transduced with the pMV_AA013 plasmid carrying the sgRNA sequences and GFP and subsequently subjected to puromycin selection to establish stable edited cell lines.

The following sgRNA oligos were used for A>G edits: non-targeting sense (5′-CAACGTCGCGAACGTCGTAT-3′); non-targeting antisense (5′-ATACGACGTTCGCGACGTTG-3′); human CCNB1 Tyr170His sense (5′-ACATATTCACTACAAAGGTT-3′); human CCNB1 Tyr170His antisense (5′-AACCTTTGTAGTGAATATGT-3′); human CCNB1 Tyr177Cys sense (5′-TATGCTTATCTGAGACAACT-3′); human CCNB1 Tyr177Cys antisense (5′-AGTTGTCTCAGATAAGCATA-3′); human CCNB1 Gln213Arg/Met241Val sense (5′-GGTTCAAATGAAATTCAGGT-3′); human CCNB1 Gln213Arg/Met241Val antisense (5′-ACCTGAATTTCATTTGAACC-3′); human CCNB1 Val227Ala/Ser228Pro sense (5′-ATGGAGACAGTCATGTACAT-3′); human CCNB1 Val227Ala/Ser228Pro antisense (5′-ATGTACATGACTGTCTCCAT-3′).

The following sgRNA oligos were used for C>T edits: non-targeting sense (5′-CAACGTCGCGAACGTCGTAT-3′); non-targeting antisense (5′-ATACGACGTTCGCGACGTTG-3′); human CCNB1 Glu169Lys sense (5′-CATATTCACTACAAAGGTTT-3′); human CCNB1 Glu169Lys antisense (5′-AAACCTTTGTAGTGAATATG-3′); human CCNB1 Glu314Lys sense (5′-TGTACCTCTCCAATCTTAGA-3′); human CCNB1 Glu314Lys antisense (5′-TCTAAGATTGGAGAGGTACA-3′); human CCNB1 Ala346Thr/Gly347Arg sense (5′-CTCCTGCTGCAATTTGAGAA-3′); human CCNB1 Ala346Thr/Gly347Arg antisense (5′-TTCTCAAATTGCAGCAGGAG-3′); human CCNB1 Glu374Lys sense (5′-GAGATTCTTCAGTATATGAC-3′); human CCNB1 Glu374Lys antisense (5′-GTCATATACTGAAGAATCTC-3′).

FACS-based competition assay

The GFP-expressing stable base-edited cells were combined with their parental base editor counterpart cells (without the sgRNA and GFP) at a ratio of 1:4 with the exact ratio confirmed by FACS analysis for GFP and seeded at a density of 0.1 × 10⁶ cells per ml in 10 cm dishes. The cells were cultured in the presence of CIRc-004 (20 nM or 200 nM) or DMSO. Cells were collected for FACS analysis on days 6 and 13. After each timepoint, cells were replated and cultured in the presence of drug with fresh drug added every 3 days. On day 13, the remaining cells after FACS were collected, and genomic DNA was extracted. Data from the FACS-based competition assay at day 13 is shown in Fig. 3e.

To verify the exact mutations in CCNB1 made by the base-editing sgRNAs and to determine whether CIRc-004 caused enrichment of these mutations, genomic DNA from the competition assay above at day 13 was collected and PCR amplification of the CCNB1 target region (forward 3F, 5′-GCCCAATGGAAACATCTGGATG-3′; reverse 3R, 5′- GCGATCTCTTAAGAAATGCTGCCC-3′) was performed followed by next-generation sequencing using CRISPR amplicon sequencing conducted at the MGH DNA Core Facility as previously described⁸⁰. Variant reads are listed in Supplementary Table 5.

MS cyclin B1 CIRc-004-dependent interactome sample preparation

To identify proteins that are modulated by CIRc-004, we performed cyclin B immunoprecipitation followed by mass spectrometry in NCI-H1048 cells treated with CIRc-004 (50 nM),CIRc-005 (50 nM) or DMSO for 2 h. The cyclin B1 immunoprecipitation was performed as described above with the following modifications: 10 mg cell extract was rotated overnight at 4 °C with 10 μg anti-cyclin B antibodies. After cyclin B1 immunoprecipitation, beads were washed with PBS before freezing at −80 °C. Dry beads were resuspended with 200 μl of Smart digest buffer (Thermo Fisher Scientific, 60113-101) before adding 3 μl of Smart trypsin. Beads were digested in a Thermo block (Thermo Fisher Scientific) for 1 h 30 min at 70 °C with 1,400 rpm (~250g) agitation. Trypsin digestion was then stopped by adding trifluoroacetic acid at 1% final concentration. The samples were then desalted using SOLA HRP SPE cartridges (Thermo Fisher Scientific, 60109-001) according to the manufacturer’s instructions. Purified peptide eluates were dried by vacuum centrifugation and re-suspended in 0.1% formic acid and stored at −20 °C until analysis.

LC–MS/MS analysis

Liquid chromatography–tandem mass spectrometry (LC–MS/MS) analysis was performed using an Ultimate 3000 HPLC connected to an Orbitrap Ascend Tribrid instrument (Thermo Fisher Scientific) and interfaced with an EASY-spray source. Then, 0.5% of the tryptic peptides were loaded onto an AcclaimPepMap100 trap column (100 µm × 2 cm, PN164750; Thermo Fisher Scientific) and separated on a 50 cm EasySpray column (ES903, Thermo Fisher Scientific) using a 60 min linear gradient from 2 to 35% B (acetonitrile, 0.1% formic acid) and at a 250 nl min⁻¹ flow rate. Both trap and column were kept at 50 °C. The Orbitrap Ascend was operated in data-independent mode (DIA) to automatically switch between MS1 and MS2, with minor changes from previously described method^81,82,83. In brief, MS1 scans were collected in the Orbitrap at a resolving power of 45,000 at m/z 200 over an m/z range of 350–1,650 m/z. The MS1 normalized AGC was set at 125% (5 × 10⁵ ions) with a maximum injection time of 91 ms and a RF lens at 30%. DIA MS2 scans were then acquired using the tMSn scan function at 30,000 orbitrap resolution over 40 scan windows with variable width, with a normalized AGC target of 1,000%, maximum injection time set to auto and a 30% collision energy.

Mass spectrometry data analysis

DIA-NN software (V8.1) was used to analyse the raw data in library-free mode and using the recommended default settings⁸³. In brief, a library was created from human UniProt SwissProt database (downloaded 22 July 2022 containing 20,386 sequences) using deep learning. Trypsin was selected as the enzyme (1 missed cleavage), N-term M excision and methionine were enabled. The identification and quantification of raw data were performed against the in silico library applying 1% FDR at precursor level and match between runs (MBR). Perseus (v.1.6.2.2) was used to further analyse the DIA-NN protein group output. Protein intensities were log₂-transformed and normalized by median subtraction (by column) after applying a pre-processing filter based on valid numbers (3 out of 3 in at least one group). Subsequently, missing values were imputed following the normal distribution down shifted. A two-sample Student’s t-test combined with multiple-testing correction using permutation-based FDR (5%) was applied. For ranking the proteins based on intensity, the average intensity was calculated for all values in the three treatment groups, log₁₀-transformed and ranked. CIRc-004-dependent interactome is included in Supplementary Table 6.

Animal studies

All mouse experiments complied with National Institutes of Health guidelines. The PDX studies were approved and performed at the Dana-Farber Cancer Institute Animal Care and Use Committee. The NCI-H69 and NCI-H1048 xenograft studies were approved and performed at Labcorp Early Development Laboratories and Pharmaron (Ningbo) Technology Development, respectively. Housing conditions for mice include a 12 h–12 h day–night cycle where the temperature is maintained at 72 °F.

SCLC cell line xenograft studies

For NCI-H69 studies, 5 × 10⁶ NCI-H69 cells were suspended in serum-free RPMI1640 medium, combined with an equal volume of ECM gel (Sigma-Aldrich, E1270), and implanted subcutaneously into the flanks of 7–8-week-old female Nude mice (Envigo, Hsd:Athymic Nude-Foxn1^nu). After 13 days, when tumours were between 88–200 mm³ in size, mice were randomized and put into treatment groups of CIRc-028 or vehicle (5% DMSO, 10% Solutol HS15, in 85% D5W). Mice were treated daily with i.v. injections of 100 mg per kg CIRc-028 or vehicle for 14 days. Tumours were measured with callipers 3 times per week for the duration of the study. Tumours were collected 18 h after the last dose, formalin fixed for 24 h and paraffin embedded for IHC analysis.

For NCI-H1048 studies, 3 × 10⁶ NCI-H1048 cells were suspended in a 1:1 mixture of HITES medium and Matrigel (Corning, 354234) and implanted subcutaneously into the flanks of 6–8-week-old female BALB/c nude mice (GemPharmatech Biotech, D000521). After 16 days, when tumours were between 180–315 mm³ in size, the mice were randomized and put into treatment groups. Treatment with CIRc-028 and vehicle was carried out as described above for NCI-H69 xenografts. Tumours were measured with callipers twice per week for the duration of the study. For NCI-H69 and NCI-H1048 xenograft studies, the investigator was not blinded but was agnostic to the outcome of the study. Animal use protocols specified that animals with tumour volumes larger than 2,000 and 2,500 mm³, respectively, or experiencing greater than 20% body weight loss or other severe clinical signs of deteriorating condition would be euthanized and these end-point criteria were not exceeded for any mice on study without prompt euthanasia.

Bulk RNA-seq, PCA and differential gene expression analysis

RNA-seq was performed on untreated SCLC DFCI-393 and DFCI-402 PDX tumours that were used for the PDX studies. For RNA-seq analysis of experimental PDX, DFCI-393 tumours were treated with the orally bioavailable cyclin A/B RxL inhibitor CIRc-014 at 100 mg per kg orally three times per day or vehicle (30% PEG400, 20% Solutol HS15, 50% Phosal 53 MCT) three times a day for 4 days and tumours were collected 8 h after the morning dose on the last day of treatment and snap-frozen. Six independent tumours from six independent mice were used for CIRc-014 and five independent tumours from five independent mice were used for vehicle. For RNA-seq of NCI-H1048 cell line, cells were plated at 500,000 cells per ml in the presence of DMSO (vehicle), cyclin A/B RxL inhibitor (CIRc-004 or CIRc-014, 200 nM), inactive enantiomer (CIRc-005, 200 nM) or a CDK2i (PF-07104091, 500 nM). Then, 24 h after drug treatment, cells were collected for RNA. The NCI-H1048 cell line experiments were done in two biologically independent experiments. RNA was extracted using the RNeasy mini kit (Qiagen, 74106) including a DNase digestion step according to the manufacturer’s instructions. Total RNA was submitted to Novogene. The libraries were prepared using the NEBNext Ultra II non-stranded kit. Paired-end 150 bp sequencing was performed on NovaSeq 6000 sequencer using S4 flow cell. Sequencing reads were mapped to the hg38 genome by STAR. Statistics for DEGs were calculated by DESeq2 (v.1.36.0) (Supplementary Tables 4 and 7).

Principal component analysis (PCA) in Extended Data Fig. 11g was performed using the top 500 genes with the largest s.d. of log-transformed FPKM (fragments per kilobase of transcript per million mapped fragments) values and these genes were subjected to PCA using the removeBatchEffect function in the limma package (v.3.58.1) and prcomp function of R software (v.4.3.3). The log-transformed sum of ASCL1, NEUROD1, CHGA, CHGA, INSM1 and SYP FPKM values was calculated as the neuroendocrine score for each sample. FPKM values are included in Supplementary Table 7. Heat maps were generated by calculating the z score using log-transformed FPKM values. To calculate the z score, a log-transformed FPKM value was subtracted from the mean and then divided by the s.d. The genes were then sorted for the log₂-transformed fold change of treatment versus control samples. z-score values are provided in Supplementary Tables 4 and 7. For HALLMARK pathway enrichment analysis, differentially expressed genes were tested if over-represented against the HALLMARK pathways from the MSigDB using R packages msigdbr (v.7.5.1) and clusterProfiler (v.4.8.3). List of differentially expressed genes are included in Supplementary Tables 4 and 7.

PDX studies

PDXs were generated from either tumour biopsies or pleural effusions of patients with SCLC. All patient samples were collected after obtaining written informed consent. PDX tumour DFCI-393 originated from a surgical biopsy at resistance from a patient who received previous chemotherapy and radiation. The surgical biopsy of DFCI-393 was directly implanted directly into the subrenal capsule of NSG mice for expansion. DFCI oncopanel genomic DNA sequencing of DFCI-393 showed an RB1 E280* mutation with a single-copy deletion of RB1 at 13q14.2, and a TP53 K132N mutation with single-copy deletion of TP53 at 17p13.1 together showing biallelic inactivation of RB1 and TP53. DFCI-402 was derived from pleural effusion of a patient who was treated with first-line carboplatin/etoposide and then second-line nivolumab and the pleural effusion was removed as part of routine clinical care. The pleural fluid from the effusion was immune depleted and implanted subcutaneously to make DFCI-402. DFCI oncopanel genomic DNA sequencing of DFCI-402 showed an RB1 R358* mutation with copy loss of 13 probably contributing to biallelic loss of RB1. It also showed a TP53 G154Afs*16 mutation. RNA-sequencing of DFCI-393 and DFCI-402 with confirmation by immunoblot analysis showed that both PDX models are neuroendocrine-high SCLCs and DFCI-393 dominantly expresses ASCL1 while DFCI-402 expresses both ASCL1 and NEUROD1. RNA-seq data are included in Supplementary Table 7.

PDX model generation

Studies and procedures were performed at Dana Farber Institute in accordance with Institutional Animal Care and Use Committee (IACUC) guidelines in an AAALAC accredited vivarium. PDX models were generated using 7- to 8-week-old female NSG (NOD.Cg-Prkdcscid Il2rgtm1WjI/SzJ) mice purchased from Jackson Laboratories. After initial implantation of patient samples, all PDX tumours were expanded and passaged continuously in NSG mice as subcutaneous tumours.

In vivo efficacy and PD studies in SCLC PDX models with the orally dosed cyclin A/B RxL inhibitor CIRc-014

Female NSG mice (aged 7–8 weeks) were used for all in vivo studies. PDX tumour fragments were dipped in Matrigel and implanted subcutaneously in NSG mice. Tumour growth was measured twice weekly when palpable. Once tumour volumes reached approximately 200 ± 50 mm³, mice were randomized using Studylog software into the treatment groups as: vehicle (30% PEG400, 20% Solutol HS15, 50% Phosal 53 MCT) or orally bioavailable cyclin A/B RxL inhibitor CIRc-014 (100 mg per kg, Circle Pharma). For the DFCI-393 model, CIRc-014 was administered orally three times a day for 28 days continuously and, for the DFCI-402 model, it was administered orally twice a day for 28 days continuously. For the DFCI-393 model, tumour samples were collected on day 28 after 8 h and for DFCI-402, tumours were monitored for regrowth after treatment withdrawal. Animals were euthanized if the tumour was necrotic/ulcerated or if the tumour volume exceeded 2,000 mm³. For the PD study in DFCI-393, mice bearing PDX tumours were treated with either vehicle or CIRc-014 (100 mg per kg orally three times per day) for 4 days. All tumour samples were collected 8 h after the morning dose on the last day of treatment. Tumour samples were fixed and snap-frozen for further downstream IHC analysis (see below). For DFCI-393 and DFCI-402 PDX studies, the investigator was not blinded but was agnostic to the outcome of the study. Animal-use protocols specified that animals with tumour volumes larger than 2,000 or experiencing >15% body weight loss or other severe clinical signs of deterioration would be euthanized, and these endpoint criteria were not exceeded for any mice on study without prompt euthanasia.

Pharmacokinetic plasma analysis with the cyclin A/B RxL inhibitor CIRc-014 in the DFCI-393 SCLC PDX model

For the pharmacokinetic studies with the orally bioavailable cyclin A/B RxL inhibitor CIRc-014, DFCI-393 PDX-tumour-bearing NSG mice were used from the PD study above. After 4 days of CIRc-014 being dosed at 100 mg per kg orally three times per day, plasma timepoint sampling was conducted after the last dose in the CIRc-014 treatment group (n = 6). Animals 1–3 were bled at 0.5 and 2 h and animals 4–6 at 1 and 4 h followed by terminal bleed through cardiac puncture at 8 h. Blood was collected in K2EDTA tubes, centrifuged and plasma was collected, frozen and sent for analysis. The concentration of CIRc-014 was quantified in the mouse plasma samples using LC–MS/MS at Charles River Laboratories. CIRc-014 was used as a reference standard and glyburide (Sigma-Aldrich) was used as the internal standard. Blank matrix CD-1 mouse plasma, pooled/mixed gender with K₂EDTA anticoagulant (BioIVT) was used to prepare calibration standards and quality-control samples.

Sample extraction was performed on ice. A 10 µl aliquot of sample (calibration standards, quality control samples, blanks and study samples) was added to a 96-well plate. Then, 60 µl of internal standard solution (100 ng ml⁻¹ glyburide in acetonitrile) was added to each well, except for wells designated for matrix blanks to which 60 µl of acetonitrile was added. The samples were covered, vortexed and centrifuged for 5 min. Then, 50 µl of supernatant was transferred to a new 96-well plate and the samples were diluted with 50 µl of MilliQ water before analysis. A Waters Acquity UPLC BEH C18, 50 × 2.1 mm (1.7 µm) column was used and maintained at 50 °C during analysis. 5% acetonitrile/water with 0.1% formic acid was used as mobile phase A and 5% acetonitrile/methanol with 0.1% formic acid was used as mobile phase B. The chromatography condition with flow rate at 0.8 ml min⁻¹ was started with 60% B, ramped to 90% B in 1 min and further increased to 95% B in 0.05 min and held for 0.25 min and the column was re-equilibrated at the initial conditions for 0.4 min. Analyte and internal standard were detected using an Applied Biosystems Sciex API 6500 triple quadrupole mass spectrometer. The instrument was equipped with an electrospray ionization source (550 °C) operated in the positive-ion mode. The analyte and internal standard were monitored in the multiple-reaction-monitoring (MRM) scan mode. The selected [M+H]⁺ precursor ions were m/z 962.4 for CIRc-014, and 494.1 for glyburide, and the product ions monitored were at m/z 423.4 and 169.0 for CIRc-014 and glyburide, respectively. Linear fitting with 1/x² weighting for a calibration range from 1.00 to 10,000 ng ml⁻¹ was used for data analysis. Analytical results for all samples met acceptance criteria.

Pharmacokinetic plasma analysis with the i.v. dosed cyclin A/B RxL inhibitor CIRc-028 in NCI-H69 and NCI-H1048 SCLC xenografts

For the pharmacokinetic studies with the i.v. dosed cyclin A/Bi CIRc-028, NCI-H69- and NCI-H1048-tumour bearing Nude mice were used from the cell line xenograft efficacy studies described above. After 14 days of dosing with CIRc-028 at 100 mg per kg i.v. once per day, plasma timepoint sampling was conducted after the last dose in the CIRc-028 groups. For the NCI-H69 study, animals 1–5 were bled at 5 min followed by terminal bleed by cardiac puncture at 4 h, and animals 6–10 were bled at 0.5 and 2 h followed by terminal bleed by cardiac puncture at 18 h. For the NCI-H1048 study, animals 1–8 were bled at 5 min and at 0.25, 0.5, 1, 2, 4 and 8 h followed by terminal bleed by cardiac puncture at 24 h. Blood was collected in K2EDTA tubes, centrifuged and plasma was collected, frozen and sent for analysis.

For the NCI-H69 study, the concentration of CIRc-028 was quantified in the mouse plasma samples by LC–MS/MS at Integrated Analytical Solutions. CIRc-028 was used as a reference standard and dextromethorphan (Sigma-Aldrich) was used as the internal standard. Blank matrix mouse plasma, with K2EDTA anticoagulant (Bioreclamation) was used to prepare the calibration standards and quality-control samples. Sample extraction was performed on ice. A 20 µl aliquot of sample (calibration standards, quality control samples, blanks and study samples) was added to 2 volumes of cold internal standard solution except samples designated for matrix blanks to which 40 µl of acetonitrile was added. The samples were covered, vortexed and centrifuged for 30 min at 6,100g. An aliquot of each supernatant was transferred to an autosampler plate and diluted with 2 volumes of 0.2% formic acid in water before analysis. A Waters Acquity UPLC BEH C18, 50 × 2.1 mm (1.7 µm) column or equivalent was used and maintained at room temperature during analysis. Water with 0.2% formic acid was used as mobile phase A and acetonitrile with 0.2% formic acid was used as mobile phase B. The chromatography condition with a flow rate at 1.2 ml min⁻¹ was started with 5% B, ramped to 95% B in 1.5 min and held for 0.2 min at 95% B with a flow rate of 0.8 ml min⁻¹ and the column was re-equilibrated at initial conditions for 0.5 min. The analyte and internal standard were detected using an Applied Biosystems Sciex API 4000 triple quadrupole mass spectrometer. The instrument was equipped with an electrospray ionization source operated in the positive-ion mode. The analyte and internal standard were monitored in the MRM scan mode. The selected [M+H]⁺ precursor ions were m/z 976.4 for CIRc-028, and 272.2 for dextromethorphan, and the product ions monitored were at m/z 596.0 and 215.2 for CIRc-028 and dextromethorphan, respectively. Power fit regression with no weighting for a calibration range from 25 to 50,000 ng ml⁻¹ was used for data analysis. Analytical results for all samples met acceptance criteria.

For the NCI-H1048 study, the concentration of CIRc-028 was quantified in the mouse plasma samples by LC–MS/MS at Pharmaron Laboratories. CIRc-028 was used as a reference standard. Blank BALB/c nude mice matrix mouse plasma with K2EDTA anticoagulant was used to prepare calibration standards and quality-control samples. Sample extraction was performed on ice. A 10 µl aliquot of sample (calibration standards, quality control samples, blanks and study samples) was added to 200 volumes of cold internal standard solution in acetonitrile except samples designated for matrix blanks to which 200 µl of acetonitrile was added. The samples were covered, vortexed and centrifuged for 15 min at 4,000 rpm. An aliquot of each supernatant was transferred to an autosampler plate and diluted with 2 volumes of with water before analysis. A phenomenex Halo ES-CN 2.7 µ (30 × 3 mm) column or equivalent was used for analysis. 95:5 water:acetonitrile with 0.1% formic acid was used as mobile phase A and 95:5 acetonitrile:water with 0.1% formic acid was used as mobile phase B. The chromatography condition with flow rate at 0.6 ml min⁻¹ was started with 10% B, ramped to 100% B in 1.7 min and held for 0.3 min at 95% B and the column was re-equilibrated at initial conditions for 0.4 min. The analyte and internal standard were detected using an Shimadzu LCMS-8050 quadrupole mass spectrometer. The instrument was equipped with an electrospray ionization source operated in the positive-ion mode. The analyte and internal standard were monitored in the MRM scan mode. The selected [M+H]⁺ precursor ion was m/z 976.4 and the product ion monitored was m/z 596.35 for CIRc-028. An appropriate internal standard was used. An appropriate linear fit was used for calibration range from 0.5 to 2,000 ng ml⁻¹ was used for data analysis. Analytical results for all samples met acceptance criteria.

IHC staining

IHC staining was performed with a Bond RX Autostainer (Leica Biosystems) using a Polymer Refine Detection kit (Leica Biosystems). Sections (thickness, 4 μm) were serially cut from formalin-fixed paraffin-embedded mouse tumour samples to perform single IHC studies using antibodies recognizing p-KNL1 (1:100, rabbit monoclonal antibody, Cell Signaling, 40758), and cleaved caspase-3 (1:400, rabbit monoclonal antibody, Cell Signaling, 9664). The immunostained slides were scanned at (×40) using the PhenoImager HT instrument (Akoya Biosciences). For each slide, tumour areas were identified by a pathologist (Y.N.L.) and manually annotated using the HALO system. The number of positive and negative tumour cells was then determined using the HALO platform multiplex-IHC v.3.2.5 algorithm (Indica laboratory) and the percentage of positive tumour cells was subsequently calculated.

Statistical analysis

For the CRISPR–Cas9 screen in Fig. 2, the data were analysed using the Apron, STARS and Hypergeometric analysis algorithms on the GPP portal (https://portals.broadinstitute.org/gpp/screener/). Genes were considered hits if q < 0.25, using the STARS analysis.

For all experiments with statistical data, the number of independent biological replicates is indicated in each figure legend. For immunoblots shown in Figs. 2i,k,n,o, 3i–k, 4b,e,g,j,k and 5d and Extended Data Figs. 2k, 5h,j, 6f–h,l, 7a, 8g,i, 9f,g and 10b–e,g,i,k, at least three biological replicates were performed and representative immunoblots are shown. For immunoblots shown in Figs. 3h and 4c and Extended Data Figs. 8h,u, 9d,e,h–k and 10f, at least two biological replicates were performed and representative immunoblots are shown. Raw immunoblots are included in Supplementary Fig. 1, and quantification of immunoblots is included in Supplementary Tables 8 and 9. For dose–response assays in Figs. 1f, 2d–g,l and 4f, h, i and Extended Data Figs. 2d,f, 4a–h, 5k and 8a,c,e, the mean of three independent biological replicates is shown unless otherwise specified. For dose–response assays shown in Fig. 2p and Extended Data Fig. 8j–t, the representative mean of two technical replicates is shown and each experiment was performed in three biological replicates. For dose–response assay shown in Extended Data Fig. 8v–x, the representative mean of two technical replicates is shown and each experiment was performed in two biological replicates. For the live-cell imaging experiment in Extended Data Fig. 6a, representative micrographs are shown from two biological replicates and, for Extended Data Fig. 2l, average data are shown of six technical replicate wells per treatment given, performed in two biological replicates. For the immunofluorescence experiment in Extended Data Figs. 6c and 10a, representative micrographs are shown from three biological replicates. For the IHC experiments in Fig. 5c,d,k,l, representative micrographs are shown from at least five independent tumours per treatment group and the data are quantified in Fig. 5e,f,m,n.

For in vivo xenograft experiments in Fig. 5a,b,i,j, two-way ANOVA was used to compare drug treatment versus vehicle. For all other experiments, statistical significance was calculated using unpaired two-tailed Students t-tests. P values were considered to be statistically significant if P < 0.05. Error bars represent s.d. unless otherwise indicated. No statistical methods were used to pre-determine sample sizes, but sample sizes are similar to those used previously for in vivo experiments.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.