CRISPR/Cas9-mediated knock-in of BRCA1/2 mutations restores response to olaparib in pancreatic cancer cell lines

Pancreatic cancer is one of the most aggressive diseases with a very poor outcome. Olaparib, a PARP inhibitor, as maintenance therapy showed benefits in patients with metastatic pancreatic adenocarcinoma bearing germline BRCA1/2 mutations. However, germline BRCA mutation has been described in only 4–7% of patients with pancreatic adenocarcinoma. A CRISPR/Cas9-mediated system was used to knock-in the c.763G > T p.(Glu255*) and c.2133C > A p.(Cys711*) mutations in cell lines to obtain truncated BRCA1 and BRCA2 proteins, respectively. A CRISPR/Cas9 ribonucleoprotein complex was assembled for each mutation and transfected into two pancreatic cell lines (T3M4 and Capan-2) and into a breast cancer cell lines (MCF7) as control. BRCA protein levels were significantly decreased in all BRCA-depleted cells (P < 0.05), proving the transfection efficiency of our CRISPR/Cas9 systems. As expected, the calculated olaparib IC50 were significantly reduced for all cell lines harbored BRCA1 or BRCA2 mutations compared to wild-type BRCA1/2 cells (P < 0.01). Furthermore, we observed a higher induction of apoptosis after 72 h olaparib treatment in BRCA-depleted cells than in wild-type cells. This strategy might offer new insights into the management of patients with pancreatic cancer and open up new perspectives based on the in vivo use of CRISPR/Cas9 strategy.


Knock-in of BRCA1 E255* and BRCA2 C711* mutations using a CRISPR/Cas9 ribonucleoprotein (RNP) complex
To verify the efficiency of our CRISPR/Cas9 systems and their capacity to induce the desired on-target mutations, we conducted a ddPCR analysis.Fractional abundances (FA) were reported in Table 2.In all tested cell lines, the on-target BRCA1 and BRCA2 mutations were found for transfected cells with the CRISPR/Cas9 BRCA1 (CC_BRCA1) and the CRISPR/Cas9 BRCA2 (CC_BRCA2) respectively, and not for non-transfected (wildtype (WT)) and CRISPR/Cas9 control (CC_ctrl) transfected cells.Regarding MCF7 cells (not tested on PDAC cell lines), we tested the four gRNA alone and in duo.The combination of two gRNAs for the CRISPR/Cas9 Table 1.All guides RNA (gRNAs) designed by the CRISPR LIFEPIPE® tool and the Dharmacon CRISPR Design Tool.For each selected on-target mutation, 5 sequences of gRNAs are proposed with their PAM sequences, cutoff sites, DNA strand, and GC content.The CrisprScan Score are also given for each gRNA.This score is correlated with the gRNA activity 54 and goes from 0 to 100.

Increased sensitivity of BRCA-depleted cell lines to olaparib
The sensitivities of BRCA1 and BRCA2-depleted cells relative to control and WT cells to olaparib were analyzed with a crystal violet assay.Increasing doses of olaparib were tested and viability was evaluated 72 h later.DMSO at the concentrations used for olaparib dilutions had no impact on cell viability.Olaparib IC50 were regrouped in Table 4 and

Increased apoptosis after treatment by olaparib for BRCA-depleted cells
The sensitivity of BRCA-depleted cells to olaparib was investigated by detecting apoptotic cells labeled with Propidium Iodide (PI) and Annexin V. Cells were treated during 72 h with olaparib at 40 µM, a dose above the IC50 of the 3 models.All CC_BRCA1 and CC_BRCA2-transfected cells shown an increased apoptosis after treatment by olaparib at 40 µM concentration compared to BRCA WT cells (Table 5 and Fig. 5).No difference  was observed in the percentage of apoptotic cells after treatment with 0 µM and 40 µM of olaparib concentration for WT and control cells (P > 0.05).

Discussion
The PARPi olaparib is approved for the treatment of patients with breast cancer 25 and its efficacy has been demonstrated on breast cancer cell line such as MCF7 cells 26 .The presence of a pathogenic or deleterious mutations in BRCA1 or BRCA2 is associated with a greater risk of developing breast, ovarian, pancreatic, and prostate cancers but are also predictive to response to olaparib.Nonetheless, a germline BRCA mutation is found in only in 4-7% of patients with pancreatic cancer 22 .In this study, we successfully knocked-in the BRCA1 E255* and the BRCA2 C711* mutations into the Capan-2 and T3M4 PDAC cell lines and in the MCF7 breast cancer cell-line, using CRISPR/Cas9-mediated technology.As previous studies have utilized a CRISPR/Cas9 system to modify MCF7 cells 27 , our initial focus was to optimize and control the effective delivery of our CRISPR/Cas9 systems into MCF7 cells prior to its application in PDAC cells.As expected, BRCA-depleted MCF7 cells exhibited higher sensitivity to olaparib than MCF7 WT cells.This result serves as control and proof of concept for the validation of our CRISPR/Cas9 system.Our CRISPR/Cas9 system is an all-in-one HDR complex, comprising Cas9-RNP (recombinant Cas9 protein complexed with two gRNAs) and donor template DNA (ssODNs) with phosphorothioate modifications at the extremities.This all-in-one strategy increases the HDR efficiency by providing the donor template at the time of DSB occurrence 28 .Furthermore, utilization of ssODN donors along with phosphorothioate modifications at extremities greatly enhances the efficiency of HDR.The design of our ssDNA donors, as proposed by the utilized tools, was symmetric; however, study demonstrated that favoring an asymmetric donor design may facilitate HDR 29 .Therefore, further experiments using an asymmetric donor design are warranted to evaluate their potential to improve HDR efficiency.
For our all gRNAs, the DSBs induced by Cas9 and the donor template insertion-site are separated by less than ten nucleotides and all presented a PAM motif, as recommended 30 .The nature of nucleotides at position -4 from the PAM sequence also influences the editing precision.In our study, the two crRNAs for the BRCA1 E255* mutation harbored a "T" in position -4 and for the BRCA2 C711* mutation a "C" and "A".The presence of a "T" or "A" at position -4 predict efficient insertions at the regions of interest.In contrast, a "G" is synonymous with a more imprecise target 31 , which concerns none of our gRNAs.
The primary limitation of CRISPR/Cas9 technology is the high probability of off-target effects, with a frequency > 50% 32 .However, reducing these off-target effects while maintaining editing efficacy remains a challenge.Several strategies have been suggested for limiting off-targets effects, including improved optimization of Cas9 and guides designs 33 .CRISPR/Cas9 design tools can now be leveraged for obtaining the best designs and identifying potential off-target sites.In our study, no off-target effect was predicted with the CRISPR LIFEPIPE® tool.In contrast, the CrispRGold tool predicted multiple off-target effects for each gRNAs, but mainly are intergenic or intronic.In cultured cells, this category of off-targets are not to be taken into account, at least for functional studies 34 .Utilizing RNP complex for delivering Cas9 protein, gRNAs and ssODN donors, as in our study, appears to be an excellent delivery method for the CRISPR/Cas9 system with decreased off-target effects and a higher rate of on-target mutations 32,33 .
The fractional abundance (FA) of CC_BRCA -transfected MCF7 cells is higher than FA of CC_BRCA -transfected PDAC cells (> 80% and < 80%, respectively) and T3M4 cells exhibit higher FA compared to Capan-2 cells (> 50% and < 20%, respectively); these mutations were not detectable in controls and non-transfected cells as expected.The differences between cell lines stem from the utilization of an identical CRISPR/Cas9 system for all of them.The DharmaFECT™ DUO, a lipid-based reagent used, has already been successfully used for the transfection of our CRISPR/Cas9 systems into MCF7 cells as reported in a previous study 35 .We optimized our transfection protocol, which involved determining the optimal incubation times and reagent concentrations, with MCF7 cells, possibly leading to the higher FA observed compared to PDAC cell lines.Aside to the transfection capacity of each cell line, the epithelial phenotype and the well-differentiated status of Capan-2 cells 36 might explain the difficulty encountered for the transfection of these cells 37 .Furthermore, the difference between T3M4 and Capan-2 cells may be attributed to the HDR activity being restricted to the late S and G2 phases of cell cycle.The Capan-2 cells doubling time is three times longer than that of T3M4 36 , resulting in unsynchronized cell cycles that are staggered during CRISPR/Cas9 transfection.Some chemicals like nocodazole can be used to synchronize cell cycles, leading to an increased HDR.However, these molecules have potential in vivo toxicity 28 .Hence, we made the decision to not incorporate them during the CRISPR/Cas9-system optimization on cell lines with the aim of transposing the protocol in vivo more easily.In an ideal setting to increase the transfection efficiency, the CRISPR/Cas9 transfection protocol should be developed and adapted for each cell line.For simplification purposes, we utilized a universal protocol for all cell lines, leading to the assumption of a heterogeneous population produced through this approach.
Interestingly, the cell lines tested showed a decrease in FA as cells were passaged and underwent freeze-thaw cycles.We were therefore unable to generate a stable BRCA-depleted cell line, as the BRCA-depleted cells did not survive after few passages.One possible explanation is that the introduction of a loss-of-function mutation in BRCA1 or BRCA2 affects the cell viability of modified in vitro cells 38,39 .This phenomenon may results in the survival and expansion of only unedited cells, as estimated by their doubling times (35 40 , 60 and 23 h 36 for MCF7, Capan-2, and T3M4 cells, respectively).After CRISPR/Cas9 transfection, cells were harvested at confluence every 4-5 days, and then either passaged or analyzed.This culture time largely exceeds the doubling time.Thus, the WT cell population expanded while the BRCA-depleted cell population remained fixed or even declined, possibly accounting for the reduction in BRCA-depleted cells after subculturing and thawing.Since we performed transient transfection using a lipid-based reagent, no stable BRCA-depleted colony was generated.To circumvent this problem, co-delivery of a CRISPR/Cas9 RNP complex and a plasmid containing a selectable marker could enrich for edited cells 41 .Moreover, plasmid-based knockout strategies can also generate stable cell lines, particularly with lentiviral transduction.However, transfection of plasmid often leads to random genome integration, www.nature.com/scientificreports/host immunogenic activation in vivo, and generate potentially more off-target effects than RNP complex due to a persistent expression of Cas9 in cells 42 , even if the all-in-one plasmid circumvent some of these side effects 43 .In our study, utilization of ssODN donors and CRISPR/Cas9 RNP complex with a lipid-based transfection is advantageous in vivo but troublesome in vitro, involving CRISPR/Cas9 RNP complex transfection prior to each experiment.For example, carrying out a western blot in our study is challenging due to the transfection performed on a limited number of cells and the decreasing FA, ultimately limiting the generation of a sufficient amount of modified cells.Consequently, we performed ELISA experiments to assess the depletion of BRCA1/2 mediated by our CRISPR/Cas9 systems at the protein level.Moreover, since ddPCR assay is suitable for the detection of mutations with a very low allele frequencies 44 , we decided to use this assay to evaluate the transfection efficiency of our CRISPR/Cas9 systems.It should be noted that the FA of CC_BRCA -transfected MCF7 cells appears to be higher compared to those observed in other study 27 .It is probable that the presence of non-integrated donors DNA could lead to false-positive results.However, this is highly unlikely due to the large difference observed in the FA between cell lines analyzed under identical conditions (transfection time, post-transfection culture time, and day of cell harvest), as well as the larger amplicon size compared to that of the donor DNA.
Olaparib is known to be effective on cancer cells which presented a deficiency in HRR, including those with BRCA mutations.PARPi induce an accumulation of SSBs, which ultimately lead to DSBs and accumulation of DNA damages in HRD tumors, resulting in tumor-cell death 45 .Olaparib is the only FDA-approved PARPi in pancreatic cancer 10 , and thus is the only PARPi tested in our study.Indeed, the therapeutic efficacy of olaparib has now been proven as maintenance treatment in metastatic pancreatic cancer which carrying a germline BRCA mutation 8 .As expected, our study showed an anticipated increase in sensitivity to olaparib in all BRCA -depleted cells.Furthermore, as demonstrated in prior studies 46,47 , the treatment with 40 µM of olaparib significantly increases the apoptosis rate in CC_BRCA1 and CC_BRCA2-transfected cells compared to non-transfected cells (P < 0.01) .Nonetheless, while the difference in sensitivity between transfected and non-transfected cells is statistically significant (P < 0.01), it remains small, particularly for T3M4 cells.It should be noted that we are not working with a pool of 100% mutated cells, meaning that the number of BRCA -mutated cells present during three days of olaparib treatment may be lower than expected.Since these cells are the most sensitive to olaparib, this may potentially alter the response to olaparib and undervalue the final outcome.

Conclusions
In conclusion, we developed CRISPR/Cas9 systems to knock-in BRCA1 and BRCA2 pathogenic mutations into PDAC cell lines, resulting in a premature termination codon in each respective gene.Olaparib sensitivity was found to be increased in BRCA-depleted cells compared to BRCA WT cells.In addition, the Cas9-RNP used displayed short-term activity post-transfection, as its quick degradation by cellular proteases allowed for reduced number of off-target effects and transient genome editing.Therefore, this system might be easily transposed in vivo.As this technology holds the potential of reinstating PARPi sensitivity mutations, it presents an attractive therapeutic option for pancreatic cancer.
The next step in this work is to assess the in vivo translatability, consider a future clinical application.The main objective is to obtain an increase in survival rates after treatment by olaparib in in vivo models of PDAC (e.g.mouse models), which harbored BRCA1/2 mutations induced by our CRISPR/Cas9 systems.The vectorization in a maximum of nucleus and the addressing only in the tumor despite a systemic introduction are the main limitations to the use of CRISPR/Cas9 technology in vivo.The delivery vehicle must be stable and versatile to allow a transient expression of the editing machinery and avoid unintentional off-target mutations within the host genome 48 .Immunogenicity and dose response must be taken into account to achieve a maximal HDR efficiency and a minimal cytotoxicity without immune response 49 .The Cas9-RNP can circumvent such issues, but its direct delivery simultaneously with the donor DNA remains a challenge in vivo.Non-viral vehicles seem to be the most suitable option for the delivering of RNP Cas9:gRNA together with the donor DNA 48 .Nanoparticles, including the CRISPR-Gold system, are good candidates for the delivery of CRISPR/Cas9 machinery.Their efficacy and in vivo safety have already been demonstrated [49][50][51] .Identifying off-target effects at the genomic level remains the ultimate challenge before the clinical application of the CRISPR/Cas9 system.Different methods for identifying off-target effects exist, but GUIDE-seq is the most effective genome-wide detection assay 52 .All these recommendations will allow a future clinical transposition of our CRISPR/Cas9 systems.

Transfection
Capan-2, MCF7 and T3M4 were respectively seeded into 96-wells plates at a density of 8 × 10 3 , 1 × 10 4 and 2 × 10 5 cells per well in triplicate and were maintained in complete medium to reach 70% of confluence on the next day.
Cells were incubated at 37 °C in a 5% CO 2 humidified atmosphere overnight.Transfection of RNP complex into Capan-2, MCF7 and T3M4 cells was next performed following the protocol entitled «Transfection of ssDNA donor oligonucleotides for HDR-mediated gene modifications using the Dharmacon™ Edit-R™ system» and using a lipid-based transfection reagent called DharmaFECT Duo (Dharmacon, Cat #T-2010-01).Briefly, a 2.5 µM Cas9 protein working solution, a 1 µM donor oligo working solution, a 2 µM gRNA transfection complex (with crRNA:tracrRNA) and a 6 µg/mL DharmaFECT Duo working solution were prepared from the stock solutions.RNP complex was then assembled with Cas9 protein at 25 nM, donor oligo at 10 nM and gRNA at 50 nM.The final transfection mixture was consisted of RNP complex, transfection reagent and serum-free medium, and was incubated 18 h under usual culture conditions.Two controls were also prepared: a gene editing control (RNP complex without donor oligo, CRISPR/Cas9 control (CC_ctrl)) and a negative control (non-transfected cells).

Droplet digital PCR (ddPCR)
CRISPR/Cas9-mediated induction of mutations was controlled using droplet digital PCR (ddPCR).Briefly, total genomic DNA was extracted using DNeasy Blood and Tissue Kit (Qiagen, Hilden, Germany) according to manufacturer's protocol for cultured cells.All DNA were quantified using Qubit 3.0 fluorometer (Life Technologies) and Qubit® dsDNA HS Assay Kit (Life Technologies).Bespoken probes were designed to respectively detect BRCA1 c.763G > T p.(E255*) (Bio-Rad, UniqueA-ssayID dHsaMDS901307380) and BRCA2 c.2133C > A p.(C711*) mutations (Bio-Rad, UniqueAssayID dHsaMDS403398316).Reaction mix was prepared using 30 ng of DNA, 11 µL of 2X ddPCR supermix for Probes (no dUTP) (Bio-Rad, Hercules, CA, USA), 1 µL of ddPCR Mutation Assay (that contains primers and probes) and nuclease-free water to a final volume of 22 µL.DNA was replaced by 1 µL nuclease-free water for negative control.The ddPCR mix was loaded on a 96-wells reaction plate (DG32™ Cartridge, Bio-Rad) and placed into the QX200 Automated Droplet Generator (Bio-Rad) for droplets generation.Droplets were then transferred to a 96-wells PCR plate followed by PCR using the C1000 Touch™ Thermal Cycler (Bio-Rad) with the following conditions: 1 cycle for 10 min at 95 °C, 40 two-steps cycles (30 s at 94 °C and 60 s at 55 °C for BRCA1 probe and 52.5 °C for BRCA2 probe; ramp rate 2 °C/second), 1 cycle for 10 min at 98 °C and a 4 °C hold.The plate was finally analyzed using the QX200 Digital PCR Reader (Bio-Rad).Only samples with a number of droplets greater than 10,000 were analyzed according to manufacturer's recommendations.DNA quantities (number of copies per microliter) obtained in the FAM channel (C MUT ) and in the HEX channel (C WT ) were determined using QuantaSoft analysis software version 1.7.4.0917 (BioRad) and the percentage of fractional abundance (FA) was calculated as follows: FA (%) = [C MUT / (C MUT + C WT )] × 100.

Enzyme-linked immunosorbent assay (ELISA)
Sandwich enzyme-linked immunosorbent assays (ELISA) sandwich were performed to quantify BRCA proteins.Protein lysates were obtained using RIPA lysis buffer (Merck, Darmstadt, Germany) containing phenylmethylsulfonyl fluoride (PMSF) (Sigma-Aldrich) and centrifugation at 15,000 g for 20 min at 40 °C.The DCTM Protein Assay kit (Bio-Rad) was used for proteins quantification.
The ELISA kit for BRCA1 MSB2703253 (MyBioSource, San Diego, CA, USA) and the ELISA kit for BRCA2 MSB2703232 (MyBioSource) were used according to the manufacturer instructions.Briefly, 96-wells plates precoated with an antibody specific to BRCA1 or BRCA2 were provided, and 100 µL of diluted protein lysate were added into the appropriate wells with biotin-conjugated antibody specific to BRCA1 or BRCA2.Optical density (O.D.) was read at 450 nm wavelength using a Multiskan Ascent (Labsystems Titertek, Thermo Fisher Scientific).A standard curve for each protein was generated with known concentrations of the standard on the x-axis and corresponding absorbance (O.D. values on the y-axis using GraphPad Prism 9® (GraphPad Software, La Jolla, CA, USA).The line of best fit with an r 2 value (coefficient of determination) ≥ 0.98 was selected.

Crystal violet assay
Capan-2, MCF7 and T3M4 cells were respectively seeded into 96-wells plates at a density of 5 × 10 3 , 1 × 10 4 , and 2 × 10 5 cells per well and were maintained in 200 µL of complete medium for 24 h.Cells were then exposed to 2, 5, 7, 10, 15, 25, 30, 40, and 50 µM concentrations of olaparib (Clinisciences, Nanterre, France) for 72 h.Olaparib solution stock was previously reconstituted in DMSO (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions.Control cells were exposed to medium without olaparib or to 50 µM DMSO.After incubation, cells were fixed with 100 µL of 70% ethanol for 10 min.Next, 100 µL of 0.2% Crystal Violet (Sigma-Aldrich) in 20% ethanol was added into each well for 15 min, followed by four washes of distilled water and elution solution consisting of 100 µL of 0.1% acetic acid in 50% ethanol was added per well.Optical density was determined using a microplate reader (Multiskan Ascent; Thermo Fisher Scientific) at 540 nm.The half-maximal inhibitory concentration (IC50) values of olaparib were determined by a non-linear regression ([inhibitor] vs. normalized response) using GraphPad Prism 9® (GraphPad Software).

Apoptosis detection assay
MCF7 and T3M4 cells were seeded into 6-wells plates at a density of 1 × 10 4 cells/mL per well and Capan-2 cells at a density of 1 × 10 5 cells/mL per well.Cells were maintained in complete medium for 24 h and then were exposed to 40 nM of olaparib or medium.After 72 h, 1 × 10 5 cells were stained with annexin V/propidium iodide (PI) (BD Biosciences, Franklin Lakes, NJ, USA) according to the manufacturer's instructions.Cell apoptosis was quantified within 1 h by flow cytometry using a Accuri C6Plus system (Becton Dickinson, San Jose, CA, USA).

Statistical analysis
All experiments were performed in three independent tests of triplicates.Experimental data were analyzed by using one-way ANOVA or paired t test and differences were considered statistically significant when P value was less than 0.05.All graphs and statistical analysis were performed using GraphPad Prism 9®.

Figure 1 .
Figure 1.Knock-in of BRCA1 E255* and BRCA2 C711* mutations using a CRISPR/Cas9 RNP complex.(a) Guide RNA (gRNA) is composed of two elements: a trans-activating CRISPR RNA (tracrRNA) in purple and a CRISPR RNA (crRNA) in green.(b) A CRISPR/Cas9 RNP complex is first assembled with a gRNA and a Cas9 protein.This complex generates a double-stranded break (DSB) repaired by homology directed repair (HDR) pathway.A donor template of 61 nucleotides containing a sequence homologous to the DSB is required to introduce prematurely the termination codon and obtain a truncated nonfunctional BRCA1 or BRCA2 protein.Created with BioRender.com.

Figure 2 .Figure 3 .
Figure 2. Fractional abundance according to the cell passage number.The fractional abundance (FA) of pools of transfected cells are given depending on the number of cells passaging for MCF7 and T3M4, and for both on-target mutations.No experiments were carried out on the Capan-2 cells.

Figure 4 .
Figure 4. Percentage of cell viability versus logarithm of the concentration of olaparib obtained for nontransfected (wild-type (WT)) cells and cells transfected with a CRISPR/Cas9 control (CC_ctrl) (a CRISPR/Cas9 complex without donor sequences), a CC_BRCA1 and a CC_BRCA2, for MCF7 cell line (a), Capan-2 cell line (b) and T3M4 cell line (c).Increasing concentrations of olaparib (0-50 µM) were applied for 72 h.Data from 3 independent experiments are expressed as mean ± SD.

Table 2 .
Fractional abundances (FA) obtained by ddPCR analysis.Fractional abundances (FA) are given for each pool of cells just after transfection by each CRISPR/Cas9 RNP designed and after one freeze-thaw cycle.Data are represented the mean values and standard derivation (SD) and are representative of three independent experiments.

Table 5 .
Percentage of apoptotic cells of MCF7, Capan-2 and T3M4 cell lines.Population of non-transfected (wild-type (WT)) cells and CRISPR/Cas9 control (CC_ctrl), CC_BRCA1 and CC_BRCA2-transfected cells positive for annexin V-FITC and both for annexin V-FITC and propidium iodide was gated for apoptosis analysis after 72 h treatment with olaparib at concentrations 0 and 40 µM.At least 1 × 10 4 cells/event were evaluated for each analysis.Data are represented as the mean values and standard error of mean (SEM) of 3 independent experiments.**P < 0.01, *P < 0.05 and ns not significant (Student's t-test).