Generation and characterization of isogenic BRCA mutant cell lines

We hypothesized that HRR deficiency would be a key regulator of cell death in response to talazoparib (Tala) + decitabine (Deci) treatment. To test this we selectively targeted the BRCA1 and BRCA2 loci in the HRR proficient triple negative breast cancer cell line MDA-MB-231 (M231wt), using standard CRISPR-Cas9 techniques [28] to generate an isogenic cell line panel. Single guide RNAs targeting the 5′end of BRCA1 between the nuclear export sequences (amino acids (aa) 81–99) and the nuclear import sequences (aa 200–300) generated modest biallelic mutations in clone B1.85 (M231BRCA1; Supplementary Fig. S1A; [29]). A 42 base pair deletion in allele 1 produced a leucine > histidine substitution at aa 165, and deleted 14 subsequent amino acids (to aa 180). This potentially compromises the MB2 myc binding site (aa 175–300; [30]). A simple 3 base pair deletion at the CRISPR guide target site in allele 2 caused deletion of glutamine at aa 169. We found BRCA1 glutamine 169 variants on ClinVar [31] but no deletion mutations at aa 169, nor leucine to histidine conversion at 165, nor larger deletions containing the missing14 amino acids in our clone.

Our BRCA2 guides targeted sequence within the transactivation domain (aa 15-105; [32]), just 3′ of the PALB2 binding site. Clone B2A (M231BRCA2) also had biallelic BRCA2 mutations: the first, a 22-base pair deletion that induced a frameshift stop after aa 90 (Supplementary Fig. S1B). The second allele had a 33 base pair deletion that removed aa 86–96 from the transactivation domain. Thus, it is possible that at least transactivation domain function in this clone may be compromised. These deletion mutations are not found on ClinVar.

We tested for compromised HRR proficiency by assessing RAD51 repair foci increases in response to DNA damage by carboplatin treatment. Without treatment, foci numbers in the M231BRCA1 and M231wt cells were similar (average 12.8, 10.7; p = 0.13), while the M231BRCA2 showed significantly more foci (versus M231wtp = 1.3E−8; versus M231BRCA1; p = 2.2E−4; Fig. 1A). However while carboplatin treatment induced abundant new foci in M231wt cells (p < 0.0001), foci numbers in M231BRCA1 and M231BRCA2 did not significantly increase. We conclude that we have generated HRR hypomorphic alleles in the M231 HRR-proficient background.

Fig. 1

Talazoparib and decitabine synergize in treatment of M231wt and isogenic BRCA1 and BRCA2-targeted mutants. A M231 cells with targeted mutations in the BRCA1 or BRCA2 loci do not assemble significant numbers of RAD51 foci in response to 16-h treatment with 8 uM carboplatin. M231wt, M231 cells with intact HRR genes; M231BRCA1, cells with targeted mutations in BRCA1; M231BRCA2, cells with targeted mutations in BRCA2 mean ± SEM calculated and significance determined by two-way ANOVA with Tukey’s adjustment for multiple testing. B–E Drug effects on final population sizes measured by SRB assay at day 6 of culture, means ± SD calculated and significance determined via 1-way ANOVA with Tukey’s adjustment for multiple testing. F–I Cell cycle profiles at 48 h of drug treatment as indicated. B–I Drugs: Deci,14 nM decitabine; Tala, 5 nM talazoparib; or 14 nM Deci + 5 nM Tala; or DMSO. A–I All experiments used at least triplicate samples, and were performed at least twice. ****, p < 0.0001; ***, p < 0.001; **, p < 0.01; ns, not significant

We tested the growth inhibitory effects of Tala with and without Deci in this cell line panel, using the triple negative breast cancer cell line SUM149PT as a control with a naturally occurring pathogenic BRCA1 2288delT mutation [33] in a different genetic background. We found that these drugs synergized to produce smaller final populations in all 4 cell lines in a 4 day endpoint SRB assay (M231wt, SynergyFinder ZIP score 23.39; M231BRCA1, ZIP score 16.37; M231BRCA2, ZIP score 12.82; SUM149PT, ZIP score 18.379; Supplementary Fig. S1C-F; [26]). These scores are well above the “likely synergistic” score boundary of > 10. As expected, at the SUM149PT IC50 doses, the BRCA1 and BRCA2-targeted M231 cells were more sensitive to talazoparib than M231wt cells (Fig. 1B–D) final population sizes: M231wt Tala treated were 74% of controls versus M231BRCA1 ~ 48%, and M231BRCA2 33%. They were also more sensitive to Tala + Deci treatment (M231wt 29% of controls versus M231BRCA1 ~ 23%, and M231BRCA2, 16%). In SUM149PT controls this drug combination almost completely prevented growth (Fig. 1E; reduction to 4% of control treatment). Accordingly, prior studies testing whether this drug combination would be useful for HRR proficient tumors found synergistic M231wt responses to Tala ± Deci, and Tala ± azacytidine in SRB endpoint assays [22].

Next, we found that BRCA mutations per se did not affect the ability of these mutants to transit the cell cycle normally under standard culture conditions (Fig. 1F-H; compare first columns, < 5% differences), and 48 h of either Tala or Deci produced only minor profile variations (Fig. 1F–H, second and third columns versus first). However in the drug combination, the M231wt cells generated a larger S-Phase (from ~ 27 to ~ 40%,) and G2/M fraction (23% to 32%; Fig. 1F), suggesting that cells might be slowing DNA synthesis to repair drug-induced DNA damage when homologous recombination can be used to seamlessly mend DNA lesions. In contrast, M231BRCA1 and M231BRAC2 cells had increased the G2/M fraction (M231BRCA1 from 22 to ~ 40%; M231BRAC2 from 19 to 27%; Fig. 1G, H), similar to the expansion seen in SUM149PT (Fig. 1I; 21–47%). G2/M stalling suggests that combination therapeutic treatment potentiates DNA damage, and that cells may be attempting DNA repair by inaccurate mechanisms available to cells with compromised HRR (review: [34]).

Talazoparib plus decitabine treatment potentiates DNA damage in HRR deficient cells

We tested for increased DNA damage signaling by counting phosphorylated histone H2A.X (p-H2A.X) foci in our isogenic M231 cell lines and in SUM149PT control cells over a 4 day time course. As expected from their ability to accurately repair DNA, the HRR proficient M231wt cells maintained a relatively low, tonic level of p-H2A.X foci at all timepoints and tratments, only briefly rising over 10 per cell at 48 h with Tala + Deci (Fig. 2A, blue, ~ 12 per nucleus average). In contrast, p-H2A.X foci numbers in the M231BRCA1, M231BRCA2 and the naturally occurring mutant SUM149PT treated with Tala + Deci increased significantly over Tala alone from 48 h onwards (Fig. 2B–D, blue versus red icons; asterisks: significant difference between Tala + Deci versus Tala).

Fig. 2

Talazoparib plus decitabine treatment potentiates cumulative DNA damage in HRR deficient cells. A–D p-H2A.X foci per nucleus in the M231 variants and in SUM149PT quantitated over 4 days. Each icon represents 50 nuclei analyzed in 3-D z-stack assemblies, dotted line, average foci numbers in Tala + Deci treated M231wt cells from 48 h onwards. Means ± SEM shown, significant difference from Tala alone treatment determined by two-way ANOVA with Tukey’s adjustment for multiple testing. E Foci numbers from Tala + Deci treatment plotted to directly compare M231 BRCA variants. Curves fit using one site binding, nonlinear least squares fits, significant differences from M231wt determined by two-way ANOVA, with Tukey’s adjustment for multiple testing, and indicated by asterisks over icons for M231BRCA1 and M231BRCA2. A–E, G; ****, p < 0.0001; ***, p < 0.001; **, p < 0.01; *, p < 0.1; ns, not significant. F Representative fluorescence images of p-H2A.X staining with talazoparib + decitabine treatment. G Percent of cells with micronuclei in control or Tala + Deci treatment visualized by DAPI staining of coverslips at treatment day 4. At least 500 cells per group examined. M231wt versus M231BRCA1 foci increases significant at p = ****. A–G Drugs: Tala, 5 nM talazoparib; Deci, 14 nM decitabine; or Tala + Deci, 14 nM decitabine + 5 nM talazoparib; or Control, DMSO. All experiments except for G used at least triplicate samples, performed at least twice

Comparison of Tala + Deci responses between the M231 isotypes revealed significantly more p-H2A.X foci in M231BRCA1 cells from 48 h onwards, and in the M231BRCA2 cells at 96 h (Fig. 2E, F example images, complete image panel Supplementary Fig. S2). This signaling was accompanied by an increase in cells with abnormal chromatin morphology and content, including a 3.5 fold increase in cells with micronuclei in M231BRCA1 versus a 1.5 fold increase in the M231wt cells at treatment day 4 (Fig. 2G).

Talazoparib plus decitabine treatment potentiates death in BRCA mutant cells

Endpoint assays such as Sulforhodamine B or 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS) cannot determine whether cell population size differences at the end of an assay are due to cell cycle arrest/slowing or cell death. To make this distinction, we directly tested the effects of BRCA mutation on drug-induced cell death in our M231 variants using the Calcein AM Live and Dead Cell Assay in which live cells generate a green-fluorescent dye, while dead cells stain with the nucleic acid intercalating dye propidium iodide. We found that Tala + Deci treatment produced significantly more death than treatment with either drug alone regardless of BRCA allele status (Fig. 3A; each isotype p < 0.0001). But in M231wt cells, the combination effects were minimal (16% maximum death), while death in the BRCA1 mutants was 30.7% with Tala and increased to 73% with Tala + Deci. In BRCA2-targeted cells, Tala-induced death was 9.6% and increased to 39% in combination with Deci. Similarly, in SUM149PT Tala- induced death was 41%; Deci 32.6%; and the drug combination killed 89% of cells (Fig. 3B. p < 0.0001 each comparison; Supplementary Fig. S3 representative FACS plots). Given that the M231 variants were isogenic, these data directly implicate BRCA1 and BRCA2 mutation in the death response.

Fig. 3

Talazoparib plus decitabine treatment potentiates death in BRCA mutant cells. A–F Death analysis using Calcein AM versus Propidium Iodide (PI) staining at day 6 of drug treatment. Values indicate mean ± SD of PI positive cells. A, B Cell death in the M231 isogenic variants, and in SUM149PT. A Significance determined by two-way ANOVA with Tukey’s adjustment for multiple testing. C, D Little cell death in the HRR proficient non-transformed TNBC line MCF10A or the HRR proficient breast cancer cell line BT20. E, F Death in BRCA1 mutant ovarian cancer cell lines COV362; JHOS2. B–F significance determined by one-way ANOVA with Tukey’s adjustment for multiple testing. A–F ****, p < 0.0001; ***, p < 0.001; **, p < 0.01; ns, not significant. All experiments use at least triplicate samples, and performed at least twice. Drugs: Tala, 5 nM talazoparib; Deci, 14 nM decitabine

This was further supported by the observation of minimal death in an HRR proficient non-transformed breast cell line (Fig. 3C; MCF10A; > 95% live with any treatment), or in another HRR proficient breast cancer cell line (Fig. 3D; BT20, 17% maximum death). In contrast, Tala + Deci also potentiated death in two independent BRCA1 mutant ovarian cancer cell lines (Fig. 3E; COV362, up to ~ 58% death; and Fig. 3F; JHOS2 up to 35% death). Thus, the enhanced death phenotype we saw in our isogenic M231-derived variants was not limited to the M231 genetic background, or to breast cancer. We note that SUM149PT was always more sensitive to all drug treatments, suggesting that it may also express other modifiers that further potentiate these responses.

In subsequent studies we observed minimal and insignificant increases in Annexin V + /PI- staining at various timepoints in response to Tala + Deci in all M231 variants (data not shown).

Decitabine co-treatment allows the use of lower talazoparib doses to affect similar therapeutic outcomes

Clinical studies suggest that PARPi have toxicities often requiring dose reduction to ineffective concentrations. We tested whether the addition of a DNMTi could allow use of lower dose PARPi, using colony formation assays. We found that treatment with our standard drug dose combination (5 nM Tala, 14 nM Deci) reduced colony formation for all M231 variants and the independent BRCA1 mutant cell line SUM149PT, but with more limited effects on MDA-231wt cells (Fig. 4A–D representative images; E–H, # icon is over standard Tala (5 nM) doses with and without 14 nM Deci). This is concordant with the cell number reductions seen in SRB assays, and with the death response data in Fig. 3.

Fig. 4

Colony formation attenuated by treatment with talazoparib + decitabine. A–D Representative crystal violet stained culture plate images at day 14, illustrating the effects of Tala, Deci, and combination treatment on colony growth for each M231 BRCA isotype and SUM149PT. Drugs: C, DMSO; T, 5 nM talazoparib; D, 14 nM decitabine; T + D, both drugs. E–H Deci treatment reduces the colonies formed at each Tala dose in all M231 isotypes and SUM149PT. (# indicates the 5 nM Tala ± Deci values for each cell line for comparison). Colony quantitation used BioRad ChemiDoc imaging System (BioRad®), analyzed by ImageJ, values indicate mean ± SEM. I Values from Tala + Deci treatment plotted to directly compare M231 BRCA variants. Non-linear curve fit, asterisks below red icons represent significant differences for M231BRCA1 versus M231wt cells, and asterisks beside green icons represent significant differences for M231BRCA2 versus M231wt cells. Significance determined by two-way ANOVA with Tukey’s adjustment for multiple testing. ****, p < 0.0001; ***, p < 0.001; **, p < 0.01; *, p < 0.1; ns, not significant. All experiments used at least triplicate samples, and performed at least twice. Prior studies testing colony formation effects of 150 mM azacytidine + 10 nM Tala in M231wt cells concur with our results. Combination treatment reduced colonies to ~ 60% of controls, while in their SUM149PT controls, colonies were reduced to about 30% of controls

In Tala titration studies, Deci addition reduced colony formation at each Tala dose in all M231 variants (Fig. 4E–G), and potently in SUM149PT -where Tala doses could be reduced to 1 nM (IC50 5.2 nM as single agent), and co-treatment with Deci still strongly inhibited colonies (Fig. 4H). Direct comparisons between M231 isotypes revealed the MDA-231BRCA1 cells were significantly more sensitivity to Deci + low dose Tala than M231wt cells (Fig. 4I, from 1 to 7 nM; red icons and curve fit, asterisks indicate p-values from 0.001 to 0.0001). All effects were more pronounced in M231BRCA1 cells than in M231BRCA2 mutants. We conclude tht Tala + Deci treatment impairs the ability of cells to produce viable colonies, and that this drug sensitivity is enhanced by BRCA mutation. We note that others reported cooperative colony reduction in MDA-231wt cells treated with azacytadine doses ranging from 100 to 400 nM combined with Tala doses ranging from 5 to 20 nM [22].

Talazoparib + decitabine treatment significantly inhibits xenograft and PDX growth

We tested the effects of clinically relevant concentrations of talazoparib and decitabine [35] in xenografts and PDX models. At these doses, the M231wt xenografts showed no effects of Deci treatment (Fig. 5A green versus black lines) and Deci did not add benefit to Tala treatment (red versus blue lines). In contrast, in SUM149PT xenografts, each drug reduced tumor volumes ~ 50% relative to controls (Fig. 5B; red, green, versus black lines) and combined, they produced tumors that were only ~ 9% of controls (blue; p < 0.0001, all comparisons). Xenografts of these two cell lines had been previously shown to be moderately sensitive to Azacytidine + Tala [22], but with much smaller growth inhibition, which we speculate may have to do with the use of azacytidine instead of decitabine.

Fig. 5

Talazoparib plus decitabine treatment significantly impair growth of BRCA mutant xenografts and PDX. A M231wt xenografts. B SUM149PT xenografts. C M231BRCA2 xenografts. D PDX TM00089, n = 6 animals per group. E PDX TM00091. A, B, C, E n = 5 animals per group. A–F Xenografts recieved 0.1 mg/kg talazoparib (Cat. #S7048, SelleckChem®) given by oral gavage or carrier, and/or 0.2 mg/kg decitabine (Cat. #S1200, SelleckChem®) or carrier given by intraperitoneal (IP) injection. PDX received the identical talazoparib dose, and ½ dose of decitabine. Treatment schedule for all mice was 5-days on/2-days off. A–E Values indicate mean ± SEM. Significance determined by two-way ANOVA with Tukey’s adjustment for multiple testing. ****, p < 0.0001; ***, p < 0.001; **, p < 0.01; *, p < 0.1; ns, not significant. E final tumor images for TM00091; #, animal died 2 days before tumor harvests, tumor was unrecoverable

In M231BRCA2 variant xenografts, Deci and Tala significantly reduced growth versus control treatment (Fig. 5C; Tala, red, p = 0.03; Deci, green, p = 0.002), and their combined efficacy was significantly greater than that of either individual drug (blue, p < 0.0001 versus all others). Efficacy was also tested in two independent BRCA1 mutated breast cancer PDX models. PDX TM00089 is derived from a patient with a germline pathogenic v757fs*BRCA1 mutation (truncation) who had received prior cisplatin/taxol therapy for a fallopian tube carcinoma ([31]; Mouse Models of Human Cancer Database, Jackson Research Labs). TM00089 growth was sharply reduced by either drug treatment (Fig. 5D; red, green, p < 0.0001 for each) and was nearly halted by Tala + Deci (blue, p < 0.0001).

The second BRCA1 PDX, TM00091, expresses the common pathogenic C61G missense mutation, with 100% variant allele frequency. This decreases BRCA1-BARD1 heterodimer formation and E3 ubiquitin ligase activity [36], with minimal response to cisplatin [37]. We found that TM00091 PDX were also extremely sensitive to Tala + Deci, exhibiting little tumor growth over the treatment course (Fig. 5E, blue p < 0.0001 versus controls; F, dissected tumors) which is significantly improved control over treatment with either drug alone (versus talazoparib, red, p = 0.0003; versus decitabine, green, p < 0.0001). In all xenograft and PDX experiments, the drug combination was well tolerated, as demonstrated by animal behavioral observations and weights (Supplementary Fig. S4A–E).