CRISPR/Cas9‐mediated mutagenesis to validate the synergy between PARP1 inhibition and chemotherapy in BRCA1‐mutated breast cancer cells

Abstract For patients carrying BRCA1 mutations, at least one‐third develop triple negative breast cancer (TNBC). Not only is TNBC difficult to treat due to the lack of molecular target receptors, but BRCA1 mutations (BRCA1m) also result in chemotherapeutic resistance, making disease recurrence more likely. Although BRCA1m are highly heterogeneous and therefore difficult to target, BRCA1 gene's synthetic lethal pair, PARP1, is conserved in BRCA1m cancer cells. Therefore, we hypothesize that targeting PARP1 might be a fruitful direction to sensitize BRCA1m cancer cells to chemotherapy. We used CRISPR/Cas9 technology to generate PARP1 deficiency in two TNBC cell lines, MDA‐MB‐231 (BRCA1 wild‐type) and MDA‐MB‐436 (BRCA1m). We explored whether this PARP1 disruption (PARP1m) could significantly lower the chemotherapeutic dose necessary to achieve therapeutic efficacy in both a 2D and 3D tumor‐on‐a‐chip model. With both BRCA1m and PARP1m, the TNBC cells were more sensitive to three representative chemotherapeutic breast cancer drugs, doxorubicin, gemcitabine and docetaxel, compared with the PARP1 wild‐type counterpart in the 2D culture environment. However, PARP1m did not result in this synergy in the 3D tumor‐on‐a‐chip model, suggesting that drug dosing in the tumor microenvironment may influence the synergy. Taken together, our results highlight a discrepancy in the efficacy of the combination of PARP1 inhibition and chemotherapy for TNBC treatment, which should be clarified to justify further clinical testing.

endocrine therapy, chemotherapy is usually the only feasible treatment option. 3 The therapeutic outcome is limited, and TNBC tumors often develop resistance. Consequently, TNBC results in the poorest overall survival of any other breast cancer subtype. 4 To achieve more successful prognoses, there is a clinical need to develop more tailored treatments against TNBC.
Overall, 5-10% of breast cancers are attributed to the inheritance of a mutation in the tumor suppressor BRCA1 gene (BRCA1m). 5 Yet, up to 70-90% of BRCA1m carriers develop TNBC. 6 There are variable forms of BRCA1m, which increases the difficulty of potentially targeting those specific mutations for TNBC therapy. The poly (ADPribose) polymerase 1 (PARP1) gene, the synthetic lethal pair of BRCA1, however, is conserved in most of the BRCA1m cancer cells and thus may be a fruitful target for TNBC therapy. 7 Neither PARP1 inhibition alone nor BRCA1 deficiency alone is lethal, but the combination of the two is, suggesting a therapeutic strategy that leverages this synthetic lethality.
PARP enzymes are mainly involved in single-stranded DNA break repair, while BRCA1 plays a role in several pathways of DNA repair, including homologous recombination repair (HR) and nonhomologous end joining repair (NHEJ) of double-stranded DNA breaks. PARP1 inhibition results in the accumulation of single-stranded DNA breaks, which leads to the stalling of replication forks. Since repair mechanisms are not present in BRCA1m cells, these stalled replication forks degrade, forming double-stranded DNA breaks. 8 Typically, the double-stranded DNA breaks would be repaired through either the HR or NHEJ pathway. However, BRCA1m and PARP1 inhibition cause HR initiation failure. The error-prone NHEJ repair pathway predominates, culminating in genomic instability, and ultimately cell death. 7 As a potential approach for treating TNBC with BRCA1m, several PARP1 inhibitors, such as olaparib (AZD-2281), and veliparib (ABT-888), are under investigation in clinical trials. Olaparib has demonstrated clinical efficacy, 9 earning approval for the treatment of germline BRCA1m, metastatic breast cancer. Nevertheless, PARP1 inhibitor monotherapy has shown mixed success in clinical trials. In a 2011 phase II clinical trial (NCT00679783), for example, olaparib monotherapy did not improve the response rate in TNBC patients, including patients with a germline BRCA1 or BRCA2 mutation. 10 Studies of PARP1 inhibition in conjunction with chemotherapy have consequently been tested in clinical trials as a means to improve the therapeutic efficacy, but similarly, limited improvement was found. 11 The combinational therapeutic success may be mediated by several variables including: the type of PARP1 inhibitor, the pharmacokinetic properties of the combinational chemotherapeutic drugs, the suboptimal dosage, and the patients' genetic profiles. While the genetic synthetic lethality paradigm may hold therapeutic promise for TNBC, combining PARP1 inhibitor drugs with chemotherapy to take advantage of this genetic relationship may be more challenging than anticipated.
Given the inconsistent clinical data, CRISPR technology may be an expedient tool to confirm drug specificity in preclinical studies prior to clinical testing. Particularly, compared with other gene manipulation strategies, such as antagonists or RNAi, CRISPR-mediated gene manipulation is more precise 12 and may have comparably fewer offtargeting effects. 13 Recently, despite ongoing clinical trials using Maternal Embryonic Leucine Zipper Kinase (MELK) inhibitors as chemotherapeutics, a study used CRISPR technology to disrupt MELK in vitro, debunking the notion that MELK was necessary for basal breast cancer cell fitness. 14 By undermining the rationale for current clinical trials, this study corroborates the need for using CRISPR technology in preclinical target validation. Inspired by the aforementioned study, we optimize the CRISPR/Cas9 system to target the PARP1 gene for validation of the selective synergism between PARP1 disruption and chemotherapy in TNBC cells. We tested different BRCA1 and PARP1 genetic profiles in an in vitro 2D setting as well as in a 3D tumor-on-a-chip system 15 to better mimic a physiological setting (Scheme 1).
We first tested the response of the BRCA1 wild-type (WT) TNBC cell line, MDA-MB-231, and the BRCA1m line, MDA-MB-436 (containing a c.5396 + 1G > A mutation 16 ) against two PARP1 inhibitors, olaparib and veliparib. In 2D culture, MDA-MB-436 was only slightly more sensitive to both PARP1 inhibitors than MDA-MB-231, while in 3D, the difference in senstivity to veliparib between the cell lines was even smaller ( Figure S1). These results are in accordance with the findings from a previous study that also only showed a minor difference in the drug IC50 dose for these cell lines. 17  To validate this genetic paradigm, we then designed the CRISPR/ Cas9 system to disrupt the PARP1 gene in both cell lines. The guide RNAs (gRNAs) targeting PARP1 (Figure 1a) were selected using the CHOPCHOP algorithm in the default setting. 18 Based on the predicted efficiency and off-targeting effects, the top three resultant gRNA candidates (see Table S1 for the sequences) were synthesized by in vitro transcription with an optimized gRNA backbone 19 and then transfected with Cas9 plasmid in HEK cells for gene disruption evaluation. The result of the T7 endonuclease I (T7EI) assay indicated that gRNA1 was the most efficient among the three candidates; the PARP1 disruption efficiencies with gRNA2 and gRNA3 only reached 82 and 23% of that with gRNA1, respectively ( Figure 1b). The mutation on exon 7 caused by gRNA1 may lead to a frameshift on the domain C of the PARP1 enzyme, disrupting its DNA-binding capability and enzymatic activity. 20 The gRNA1 was subsequently cloned into an all-inone Cas9-T2A-EGFP plasmid 21 using our previously established protocol 22 for PARP1 mutated (PARP1m) TNBC cell generation (sequence verified by Sanger sequencing, Figure S2). Notably, this disruption was similar to that reported in a previous study. 24 Nonetheless, after the introduction of CRISPR editing for these two rounds of enrichment, the edited MDA-MB-436 cells became unstable and formed heterogeneous populations. Since no HR template was introduced during the transfection, as expected, NHEJ was the likely pathway of DNA repair and caused PARP1 mutagenesis.
To assess CRISPR/Cas9 off-target effects, three primers were designed to match the most likely off-target candidates with Cas-OFFinder. 25 A T7EI assay revealed that Cas9 did not induce any gene disruptions at these likely off-target loci ( Figure S4a). In addition to the Cas-OFFinder prediction, we used another machine learningbased algorithm, DeepCRISPR, to find the potential off-targeting sites of our gRNA. 26 According to the DeepCRISPR results, the gRNA that we designed had a relatively low possibility of introducing undesired gene editing ( Figure S4b), yet we still chose the top four potential offtarget sites for further validation. Those sites were verified by amplicon-based next generation sequencing. After removing the lowquality reads, sequence variations at each site were detected with CRISPResso2. 27 Again, the editing at those potential off-targeting sites was minimal (modification rate < 0.5% for both MDA-MB-231-PARP1m and MDA-MB-436-PARP1m; Figure S4c). These results indicated the specificity of this system for PARP1 targeting.  Figure S5). Another study similarly reported that triple negative breast cancer cells were significantly sensitized to cell killing when gemcitabine was introduced in combination with a PARP1 inhibitor drug. 30 At high drug concentrations, it is likely that too many of the cells were dead, making any synergy undiscernible. In contrast, at low drug concentrations, it is possible that the assay was not sufficiently sensitive to ascertain differences in cell viability. As such, the therapeutic window of synergy observed was likely restricted.  20,31 Although there is debate on whether 3D tumor-on-a-chip models can faithfully represent the real tumor microenvironment and ultimately replace animal models, the platform facilitates a more systematic way to study each potential variable component (e.g., extracellular matrix, tumor-stromal interaction, flow and hypoxia) that may affect the drug responses. 32 In addition, the tumor-on-a-chip platform enables screening in a high-throughput manner with reduced sample volume, which may boost the drug screening process and reduce the cost for development. 31 Therefore, a microfluidic model, consisting of the tumor microvasculature with human endothelial cells (Figures 3a,b), 15  presented as mean ± standard deviation (SD). Significance was determined using t-tests and presented as ** p < .01 and *** p < .001 response of cancer cells. 15 These two variables, extracellular matrix and flow, were shown to be important in other studies as well 34 and may explain why we observed a discrepancy in cell viability between the conventional 2D and our 3D tumor-on-a-chip system.
Our 3D tumor-on-a-chip results were consistent with the results from studies testing PARP1 inhibitors in combination with chemotherapy in breast cancer trials, 11 and consistent with results from another in vitro study targeting ovarian cancer. 35 Many TNBC studies have shown that the combinations do not provide benefit beyond the standard of care. Based upon reported in vitro synergism, paclitaxel and olaparib were tested in metastatic TNBC (NCT00707707). The results showed only partial antitumor activity but enhanced overall toxicity, neutropenia, and myelosuppression in patients who received combinational therapy in comparison to those who received either paclitaxel or olaparib alone. 36 In a Phase II trial (NCT01506609), a combination of carboplatin and paclitaxel was compared with a combination of carboplatin, paclitaxel and veliparib. There was no difference in the progression-free survival for the BRCA1m metastatic breast cancer patients. Similarly, in a recent Phase III clinical trial (NCT02032277), veliparib did not improve the efficacy of platinum-based chemotherapy in TNBC patients with BRCA1/2 germline mutations. 37 These results support our findings in the 3D tumor-on-a-chip system, implying that this drug screening platform may be able to provide additional therapeutic validation prior to clinical trials, potentially expediting drug translation.
In summary, CRISPR/Cas9 was designed and optimized to disrupt PARP1, the synthetic lethal pair of BRCA1. While the 2D in vitro results showed that CRISPR/Cas9-mediated PARP1m sensitized the TNBC cells with BRCA1m to chemotherapeutic drugs, there was a dichotomy between the 2D and 3D tumor-on-a-chip results, mirroring inconsistencies found in recent clinical trials. Collectively, our approach combining CRISPR/Cas9-mediated mutagenous and a 3D Green fluorescence represents the apoptotic cancer cells in the unit. Data are presented as mean ± SD. Significance was determined using t-tests and presented as **p < .01 and ***p < .001 tumor-on-a-chip system may represent a better modeling strategy for