CX-5461 is a DNA G-quadruplex stabilizer with selective lethality in BRCA1/2 deficient tumours

Abstract

G-quadruplex DNAs form four-stranded helical structures and are proposed to play key roles in different cellular processes. Targeting G-quadruplex DNAs for cancer treatment is a very promising prospect. Here, we show that CX-5461 is a G-quadruplex stabilizer, with specific toxicity against BRCA deficiencies in cancer cells and polyclonal patient-derived xenograft models, including tumours resistant to PARP inhibition. Exposure to CX-5461, and its related drug CX-3543, blocks replication forks and induces ssDNA gaps or breaks. The BRCA and NHEJ pathways are required for the repair of CX-5461 and CX-3543-induced DNA damage and failure to do so leads to lethality. These data strengthen the concept of G4 targeting as a therapeutic approach, specifically for targeting HR and NHEJ deficient cancers and other tumours deficient for DNA damage repair. CX-5461 is now in advanced phase I clinical trial for patients with BRCA1/2 deficient tumours (Canadian trial, NCT02719977, opened May 2016).

Figure 1. BRCA2 deficient cells are highly sensitive to CX-5461 in different human and murine cell types.

(a) The colony formation capacity of BRCA2^−/− HCT116 cells was greatly reduced by treatment with CX-5461. Experiments were repeated twice with similar results. Scale bar, 1 cm. (b) The hypersensitivity of BRCA2^−/− cells to CX-5461 in HCT116 validated by WST-1 assay. Representative experiment #3 (see Supplementary Fig. 1a for full experimental panels, n=9) is displayed as individual data points and as fitted sigmoid dose response curves (green for BRCA2 deficient and red for BRCA2 wild type). Dashed vertical lines are the IC50. (c) CX-5461 induced more apoptosis in BRCA2^−/− cells as indicated by FACS analysis. A representative result is shown on the left panel and the right panel shows mean apoptotic fraction with 95% CI of cells in early apoptosis under different drug concentrations. ***P<0.0001, t-test; n=3, 2 or more replicates per condition. See Supplementary Table 5 for more statistical analyses. (d) BRCA2^−/− in DLD1 isogenic cell line pairs displayed hypersensitivity to CX-5461 by WST-1 assay. Representative experiment #3 is shown (see Supplementary Fig. 2a for full experimental panels; n=4). Green fitted sigmoid curves are for BRCA2 homozygous (HOM) and red for BRCA2 wild type and heterozygous (HET). (e) BRCA2 deficient ovarian cancer PEO1 cells exhibited increased sensitivity to CX-5461 relative to BRCA2 proficient C4-2 cells by WST-1 assay. A representative experiment #1 is displayed (see Supplementary Fig. 2b for full experimental panels, n=3). (f) RNAi knockdown of BRCA2 increased sensitivity to CX-5461 in p53^−/− HCT116 cells by WST-1 assay (4 days in drug). The results of all three experiments are summarized by the green (BRCA2 knockdown) and red (non-targeting control) super-smoother fit lines. (g) 45S pre-rRNA level measured by RT-PCR after CX-5461, CX-3543 and BMH-21 treatment in WT and BRCA2^−/− HCT116 cells. Drug incubation time was 24 h. Fold change estimates and unadjusted 95% CIs of 45s pre-rRNA levels under drug treatment condition versus vehicle control are shown. P values (by F-test) are shown in Supplementary Table 7. (h) BRCA2 knockout cells are not more sensitive to BMH-21 in HCT116 through WST-1 assay. One representative experimental result is shown (more replicates are shown in Supplementary Fig. 3a).

Figure 2. DNA damage is induced in cells with CX-5461 and CX-3543 treatment.

(a) The formation of γ-H2AX, 53BP1, RPA and RAD51 foci was monitored upon CX-5461, CX-3543 and BMH-21 treatment at 10⁻⁷ M in U2OS cells. Drug treatment time is 24 h for all drugs. Scale bar, 10 μM. (b) Beanplots of U2OS cells showing five or more γ-H2AX foci for the indicated drug treatment condition after 24 h. **P<0.01 (one tailed randomization tests adjusted for multiple comparisons relative to vehicle control); n=2 experiments, (35–150 cells were counted each time per each condition). DMSO was the vehicle control solvent for CX-3543 and NaH₂PO₄ was the vehicle control solvent for BMH-21 and CX-5461. (c) Beanplots of HCT116 cells showing three or more 53BP1 foci for the indicated drug treatment conditions after 24 h. **P<0.01, ***P<0.0001 (one-tailed randomization tests adjusted for multiple comparisons relative to vehicle control); n=3 experiments; >100 cells per replicate condition. (d) Quantification of alkaline comet and neutral comet assay result upon CX-5461 treatment (30 min) in HCT116. Tail moments were determined as described in Methods; (n=3 experiments, 100 cells were counted in each replica), mean and 95% CI are shown. Right panels show pair-wise comparison of condition contrasts and 95% CIs of tail moment difference. (e) Representative images of alkaline comet assay of cells treated with CX-5461 (10⁻⁶ M, 30 min) and no drug control. Scale bar, 20 μM.

Figure 3. CX-5461 and CX-3543 induced DNA damage is replication-dependent.

(a) Active replication decreased upon CX-5461 treatment in WT and BRCA2^−/− HCT116. Cells were treated with CX-5461 for the time indicated before incubating with EdU (10 μM) for 1 h. Cells were analysed by FACS with the intensity of EdU and PI recorded. Left panel shows one representative FACS profile when cells were treated with CX-5461 at 10⁻⁶ M; right panel shows the mean percentage of cells in S phase (with 95% CIs) under different CX-5461 concentrations at different time points; n=3 experiments. Cell cycle distributions at more time points and drug concentrations are shown in Supplementary Fig. 5a and Supplementary Table 6. (b) CX-5461 induced 53BP1 foci enriched in S phase (positive for EdU labelling), and APH greatly suppressed CX-5461 induced DNA damage in HCT116. WT Cells were treated with EdU (20 μM) for 30 min, then EdU was washed out and the cells were treated with CX-5461 (10⁻⁷ M) for 1 h. For APH treatment, after EdU labelling, APH (5 μM) was added for 2 h before incubating with CX-5461 (10⁻⁷ M) for 1 h. Scale bar, 10 μM. (c) The percentage of 53BP1 foci positive cells within EdU positive and EdU negative population with or without APH was quantified in HCT116 cells. Experimental conditions were the same as stated in b. Bars show the mean of three time course experiments (>100 cells each replica) and 95% CIs. (d) Replication rate is reduced by CX-5461 in BRCA2 deficient cells at higher level than in BRCA2 proficient cells. CIdU (30 min) treated HCT116 cells were chased with or without CX-5461 for 30 min in the presence of IdU, then the cells were processed for DNA fibre analysis; n=2. Median fork rate and the number of tracks analysed are shown. The box extends from the 25th to 75th percentiles. P value was calculated by Mann–Whitney U test.

Figure 4. The repair of CX-5461 and CX-3543 induced DNA damage relies on BRCA pathway.

(a) CX-5461 induces higher levels of DNA damage in BRCA2^−/−cells as manifested by the increase of γ-H2AX and RPA phosphorylation in BRCA2^−/− cells. HCT116 BRCA2^+/+ and BRCA2^−/− cells were incubated with vehicle (Ve), 10 μM CX-5461 (CX) or 10uM PDS for 4 h after 1 h release from double thymidine block. Whole-cell lysates or chromatin bound fractions were analysed by Western blotting. BRCA2^+/+ cells treated with 2 mM HU for 4 h were immunoblotted as a control. Increased γ-H2AX and RPA phosphorylation happened before apoptosis as shown by the absence of Parp1 degradation. Uncropped western blotting pictures are shown in Supplementary Fig. 11. (b) BRCA2^−/− HCT116 cells accumulate more chromosome abnormalities in the presence of CX-5461 (10⁻⁸ M 48 h) demonstrated by mitotic chromosome spread. Scale bar, 10 μM. Arrows point to chromosome structure abnormalities. (c) Percentage of cells with chromosome abnormalities with experimental conditions stated in b. N≥3, >50 cells each replica. 95% CIs are shown for each data point. (d) 53BP foci after pulse CX-5461 treatment were resolved in WT HCT116 cells after 72 h but not in BRCA2^−/− HCT116 cells. Cells were pulse treated with CX-5461 at 10⁻⁸ M for 2 h, and then the drug was washed out. Damage foci were monitored after 24, 48 and 72 h. Scale bar, 10 μM. (e) Plot displays the percentage of HCT116 cells with 53BP1 foci with experimental conditions stated in d. At least three independent experiments were done (>100 cells were counted each time). P values were calculated using two-tailed randomization tests.

Figure 5. CX-5461 and CX-3543 stabilize G4 sequences.

(a) In vitro FRET melting assay with three different G4 forming DNA fragments and a non-G4 forming dsDNA control. Vertical axis, changes in melting temperature; horizontal axis, drug concentration (μM). Error bars denote the s.d.; n=3. The solid lines represent the interpolation of the values with a single binding curve model. (b) Progression of DNA polymerase was stalled by CX-5461 and CX-3543 when incubating with G4 forming sequence cKit1. Full gel image is displayed in Supplementary Fig. 6c. (c) CX-5461 and CX-3543 bind to and stabilize G4 structure as demonstrated by the increased number of immunofluorescence foci with G4 binding antibody, BG4. Scale bar, 10 μM. Right panel shows the quantification. Median BG4 foci per nucleus is shown. The box extends from the 25th to 75th percentiles. (d) Co-localization between 53BP1 foci and BG4 foci. Drug treatment time is 24 h, N=2.∼500 cells per condition were counted. Scale bar, 10 μM. Right panel shows the quantification. Error bars denote the s.d.

Figure 6. CX-5461 induces chromosome instability at G4 sequences in human and yeast systems.

(a) The CX-5461 induces increased GCR rates in yeast. Left panel shows the GCR assay setup. Right panel shows increased GCR rates induced by CX-5461 compared to untreated control and to a non-G4 forming G-rich control sequence (represented as per 10⁻⁸ mutations/generation). N=3. (b) Effect of CX-5461 on telomere fragility in BRCA2^+/+ and BRCA2^−/−HCT116 cells. Arrows point to telomere defects with either fragile telomeres or missing telomeres. N=3, >100 cells each replica. The data were modelled using a logistic regression model. Scale bar, 5 μM. (c) RAD51-ChIP after CX-5461 treatment identified more peaks than with vehicle control and IgG-CHIP in U2OS cells. Three biological replicates per condition and two for IgG backgrounds. Peaks were classified as unique if occurred in isolation, or reoccurring if overlapped with at least one other peaks±500 bp. Error bars depict the range of peak numbers for three biological replicates. The exact peak numbers are shown in Supplementary Tables 8 and 9. (d) Peaks identified from RAD51-ChIP with CX-5461 treatment enrich for G4 sites. The number of G4 sites in unique and reoccurring peaks are shown for three ChIP conditions. G4 sites normalized by peak length are shown in Supplementary Fig. 6d. Peak length distribution is shown in Supplementary Fig. 6e. A screen shot of the peak is shown in Supplementary Fig. 6f.

Figure 7. CX-5461 effectively kills tumour cells deficient for a number of DNA damage repair pathways.

(a) PRKDC^−/− HCT116 cells are more sensitive to CX-5461 compared with PRKDC wild type HCT116 cells. The dose sensitivity of CX-5461 (6 days in drug) was measured by WST-1 assay with representative experiment #2 shown (see Supplementary Fig. 7a for all three experiments). Vertical dashed lines indicate the estimated IC50 values. (b) LIG4^−/− HCT116 cells are more sensitive to CX-5461 and PDS compared with LIG4 wild type HCT116 cells. The drug sensitivity (4 days in drug) was measured by a WST-1 assay. n≥4. IC50 difference for CX-5461 between LIG4^+/+ and LIG4^−/− cells is 5.73 (P<10⁻¹⁰, t-test). Sigmoid fits were unavailable for PDS treated cells so no IC50 estimates or fold change was available. Assessment via a two-way ANOVA test showed that Lig4 deficient cells were more sensitive to PDS than Lig4 proficient cells (F-test P<10⁻¹⁰). Results of drug sensitivity to BMH-21 and cisplatin for LIG4^+/+ and LIG4^−/− cells are shown in Supplementary Fig. 7b. (c) Genotype specific sensitivity to CX-5461 in C. elegans. The percentage of viable embryos observed in the progeny of animals treated for∼20 h with 100 μM CX-5461 or carrier (50 μM NaH₂PO₄). Error bars depict the s.e.m. of at least three experiments. Statistically significant difference by t-test was discovered for all mutants shown in this figure (P<0.05) comparing carrier-treated to CX-5461-treated animals. (d) CX-5461/CX-3543 exhibits antiproliferative potency against a panel of breast cancer cell lines by SRB cytotoxicity assay. Heatmap represents IC50 values for 50 breast cancer cell lines treated with Olaparib, BMH-21, CX-3543 and CX-5461 for 5 days. The ward-linkage method of ‘hclust' (R function) was used to compute the dendrogram. The panel on the left is ordered by intrinsic subtype of cell lines, the panel on the right is ordered by ranked value of CX-5461 sensitivity. The legends indicate the categorical values for intrinsic subtype, BRCA mutation status, DNA repair mutation status, and cisplatin sensitivity (data from Sanger cell line project).

Figure 8. CX-5461 selectively suppresses growth of BRCA deficient tumours in murine xenografts and chemo-resistant PDX model.

(a,b) The effect of CX-5461 on tumor growth with xenografted tumours from isogenic WT and BRCA2 knockout HCT116 (a) and DLD1 (b) cells. For HCT116 xenograft model, three drug doses were administered at 12.5 mg kg⁻¹, 25 mg kg⁻¹ and 50 mg kg⁻¹ together with vehicle control. For DLD1 xenograft model, CX-5461 was administered at 50 mg kg⁻¹. Vertical dashed lines show the end of drug treatment. Each coloured line represents individual mouse in a. Solid lines in b represent the mean tumour volume with 95% CI (shown by shadow around solid lines) from a linear model fitted to the tumour volumes. (c,d) The administration of CX-5461 greatly extended the survival of mice with tumours from BRCA2 deficient but not BRCA2 proficient in xenograft models with tumours formed from HCT116 cells (c) and DLD1 cells (d). The survival of mice in experimental panel is shown as moribund-free survival time post randomization. The significance of survival differences is indicated with the log rank test (See Statistical methods). Dose-dependent trend significance is indicated with the log rank test for trend. (e) CX-5461 is effective for taxane resistant BRCA1/2 deficient TNBC in PDX model. Growth curve of tumours grafted from individual patients with BRCA1 and BRCA2 WT (CTG-1019), BRCA1 germline and BRCA2 somatic mutation (CTG-0888, gBRCA1m/sBRCA2m), and BRCA2 germline and BRCA1 somatic mutation (CTG-0012, sBRCA1m/gBRCA2m). Tumour volume average curves (lines) with pointwise 95% CIs (shaded regions) are shown. Mice were treated with vehicle, Olaparib, CX-5461 or the combination of Olaparib and CX-5461. See Supplementary Fig. 8c for body weight and drug dosing schedule. (f) CX-5461 is effective in cisplatin pretreated and BRCA1/2 deficient TNBC in PDX model. Tumour volume average curves (lines) with pointwise 95% CIs (shaded regions) are shown. Tumours are grafted from patients with BRCA1 (CFIB-NB02), and BRCA2 (CFIB-70620) germline mutation. See Supplementary Fig. 8d for body weight and drug dosing schedule.