key: cord-0309452-focgutoq
authors: Chien, Jasper Che-Yung; Tabet, Elie; Pinkham, Kelsey; da Hora, Cintia Carla; Chang, Jason Cheng-Yu; Lin, Steven; Badr, Christian Elias; Lai, Charles Pin-Kuang
title: A multiplexed bioluminescent reporter for sensitive and non-invasive tracking of DNA double strand break repair dynamics in vitro and in vivo
date: 2020-07-09
journal: bioRxiv
DOI: 10.1101/2020.03.30.015271
sha: 152d81350a12f4f05e081dd12a79b94f84979a89
doc_id: 309452
cord_uid: focgutoq

Tracking DNA double strand break (DSB) repair is paramount for the understanding and therapeutic development of various diseases including cancers. Herein, we describe a multiplexed bioluminescent repair reporter (BLRR) for non-invasive monitoring of DSB repair pathways in living cells and animals. The BLRR approach employs secreted Gaussia and Vargula luciferases to simultaneously detect homology-directed repair (HDR) and non-homologous end joining (NHEJ), respectively. BLRR data are consistent with next-generation sequencing results for reporting HDR (R2 = 0.9722) and NHEJ (R2 = 0.919) events. Moreover, BLRR analysis allows longitudinal tracking of HDR and NHEJ activities in cells, and enables detection of DSB repairs in xenografted tumours in vivo. Using the BLRR system, we observed a significant difference in the efficiency of CRISPR/Cas9-mediated editing with guide RNAs only 1-10 bp apart. Moreover, BLRR analysis detected altered dynamics for DSB repair induced by small-molecule modulators. Finally, we discovered HDR-suppressing functions of anticancer cardiac glycosides in human glioblastomas and glioma cancer stem-like cells via inhibition of DNA repair protein RAD51 homolog 1 (RAD51). The BLRR method provides a highly sensitive platform to simultaneously and longitudinally track HDR and NHEJ dynamics that is sufficiently versatile for elucidating the physiology and therapeutic development of DSB repair.

Tracking DNA double strand break (DSB) repair is paramount for the understanding and therapeutic development of various diseases including cancers. Herein, we describe a multiplexed bioluminescent repair reporter (BLRR) for non-invasive monitoring of DSB repair pathways in living cells and animals. The BLRR approach employs secreted Gaussia and Vargula luciferases to simultaneously detect homology-directed repair (HDR) and non-homologous end joining (NHEJ), respectively. BLRR data are consistent with next-generation sequencing results for reporting HDR (R 2 = 0.9722) and NHEJ (R 2 = 0.919) events. Moreover, BLRR analysis allows longitudinal tracking of HDR and NHEJ activities in cells, and enables detection of DSB repairs in xenografted tumours in vivo. Using the BLRR system, we observed a significant difference in the efficiency of CRISPR/Cas9mediated editing with guide RNAs only 1-10 bp apart. Moreover, BLRR analysis detected altered dynamics for DSB repair induced by small-molecule modulators.

Finally, we discovered HDR-suppressing functions of anticancer cardiac glycosides in human glioblastomas and glioma cancer stem-like cells via inhibition of DNA repair protein RAD51 homolog 1 (RAD51). The BLRR method provides a highly sensitive platform to simultaneously and longitudinally track HDR and NHEJ dynamics that is sufficiently versatile for elucidating the physiology and therapeutic development of DSB repair.

Repairing DNA damage plays a key role in maintaining genome integrity and cell viability. One DNA repair mechanism, DNA double strand break (DSB) repair, comprises two major pathways; error-prone non-homologous end joining (NHEJ) and template-dependent homology-directed repair (HDR) (1, 2) . The NHEJ pathway repairs DSBs by rejoining the two broken ends, which introduces random insertions or deletions at the DSB site, resulting in disruption of the gene sequence. By contrast, the HDR pathway repairs DSBs via homologous recombination when a donor template with a homologous sequence is available, thereby enabling insertion of desired nucleotides into the target DNA region. Importantly, the cellular preference for particular repair pathways can affect the choice of sensitizer employed in cancer treatment, as well as the efficiency of introducing therapeutic genes (3, 4) .

Cancer treatment often includes radiation and chemotherapy (chemoradiotherapy), which targets tumour cells by causing DNA damage, including introducing DSBs in some cases. However, this damage is recognised and often repaired by the intrinsic DNA damage response (DDR), which reduces DNA damageinduced cell death (5) . Consequently, active DNA repair mechanisms can promote therapy resistance and recurrence in various tumour types. For instance, DNA repair protein RAD51 homolog 1 (RAD51) overexpression in breast and brain cancer cells can lead to increased HDR activity, resulting in resistance to chemoradiotherapy (6) (7) (8) .

Fortunately, small-molecule modulators of DNA repair mechanisms have since been reported to increase the efficacy of DNA-targeting therapeutics against cancers (4) , and genome editing tools are being actively investigated for therapeutic and precision diagnostic applications. Meganucleases, zinc-finger nucleases (ZFNs), transcription activator-like effector nuclease (TALEN) and clustered regularly interspaced short palindromic repeat (CRISPR)-associated protein 9 (Cas9) (9) create DSBs at target DNA sites to introduce therapeutic genes by HDR, or to knockout disease-associated genes by NHEJ (10) . Much effort in gene therapy development has focused on enhancing HDR over NHEJ during DSB repair to introduce functional genes, either by controlling genome editing tools, the cell cycle (11, 12) , optimising donor templates (13) , or using small molecules to inhibit NHEJ-related proteins (14) (15) (16) .

However, investigating DSB repair outcomes can be time-consuming, and typically requires disruption of cells for subsequent DNA sequence analyses. This challenge has impeded high-throughput HDR optimisation for the development of cancer and gene therapies (3) .

Conventional sequencing methods involve genomic DNA extraction, PCR amplification of DSB sequences, and subsequent sequence analysis methods such as Sanger sequencing and next-generation sequencing (NGS) (17) . Meanwhile, mismatch cleavage nucleases such as T7 Endonuclease I (T7E1) and Surveyor nuclease have been applied to quantify insertion and deletion (indel) frequencies (18, 19) . However, nuclease-based methods often underestimate indel frequencies, and are unreliable when the indel frequency is over 30% or below 3% (19) (20) (21) (22) . In parallel, PCR products amplified from DSB sites can be cloned into bacterial vectors by ligation, and numerous (>48) clones must be picked for Sanger sequencing to obtain precise DSB repair results, including mutation type and indel frequency (23) . In recent years, alternative strategies including tracking of indels by decomposition (TIDE) and

tracking of insertions, deletions and recombination events (TIDER) have been developed (24, 25) . Such strategies provide a simpler analysis method for detecting indels by directly decomposing Sanger sequencing results for 500−1,500 bp PCR products of CRISPR-Cas9-edited cells. By contrast, NGS analyses of amplified PCR products provide information on the type of DSB repair, including the type and frequency of mutation sequences, as well as long mutations (9, 17) . NGS data are often studied using NGS analysis tools such as CRISPResso (26) to assess CRISPR-based editing results. Although NGS can detect mutation frequencies as low as 0.01%, it is costly and time-consuming, requiring days to generate results (27) .

Reporter genes such as fluorescent proteins and bioluminescent luciferases are commonly used for cost-effective analysis of DSB repair results (28, 29) . DSB repair events can be quantified by knocking down fluorescent/bioluminescent reporter genes expressed in cells, and HDR efficiency can be measured by introducing reporter genes into target sequences. Fluorescent reporter-based methods do not require cell lysis and genomic DNA extraction, and instead use flow cytometry and/or a microplate reader for detection. However, most of these reporters are designed to reveal either HDR or NHEJ events in cells (28, 30) . By contrast, traffic light reporters (TLRs) developed by Certo et al. (2011) use an inactivated enhanced green fluorescent protein (EGFP)

bearing an I-SceI site followed by a T2A peptide sequence and an out-of-frame mCherry to report HDR and NHEJ activities simultaneously (31) . However, TLRs require flow cytometry analysis in order to quantitate DSB repair events, which limits their use for non-disruptive, longitudinal monitoring of DSB repair events.

Herein, we describe a non-invasive and highly sensitive bioluminescence repair reporter (BLRR) for longitudinal tracking of HDR/NHEJ both in vitro and in vivo.

The BLRR method employs the naturally secreted Gaussia luciferase (Gluc) and

Vargula luciferase (Vluc) (32) Blood collection and luciferase measurement were carried out as previously described (36) . Briefly, ~30 µL of blood was collected following a small incision in the tail and immediately mixed with ethylenediaminetetraacetic acid (EDTA; 10 mM) to prevent coagulation. A 5 µL sample of blood was used for Gluc and Vluc activity measurement by adding 100 μ L coelenterazine (50 μ g/mL; Gluc substrate) or 100 μ l of vargulin (2.5 μ g/mL; Vluc substrate), respectively. Photon counts were acquired for 10 s using a GloMax Discover System GM300. 

The BLRR consists of secreted Gluc and Vluc for simultaneous monitoring of HDR and NHEJ, respectively. HDR and NHEJ activities can thus be detected by assaying each reporter activity in a small volume (i.e. a few µl) of conditioned medium or blood, keeping cells and animals unperturbed for subsequent molecular analyses such as sequencing and proteomics (Figure 1A, B) . To create the BLRR system, we replaced the Q105−E110 (QGGIGE) sequence in Gluc with a 39 bp fragment containing an I-SceI endonuclease targeting site, two spacers, and a stop codon, thereby generating early translational termination and an inactive Gluc protein (Supplementary Figure   1) . We next inserted a 2 bp frame-shifted T2A peptide sequence (37) reporting HDR activity. Meanwhile, in the absence of the trGluc donor template, one of three frameshifts from NHEJ indels will correct the frameshifted T2A-Vluc sequence, causing it to become in-frame, thereby enabling subsequent Vluc expression to report NHEJ activity ( Figure 1A) . To verify BLRR function, we used two positive control constructs, BLRR-(+)NHEJ and BLRR-(+)HDR, to simulate NHEJ and HDR repair, respectively, and confirmed the specificity of BLRR signals ( Figure 1C, D) .

To examine whether the BLRR reflects endogenous DSB repair, 293T cells stably expressing BLRR (BLRR cells) were transfected with or without trGluc for 48

h to express I-SceI. Aliquots of conditioned medium were then assayed for Gluc and Vluc activities to detect HDR and NHEJ events, respectively. Importantly, the Vluc signal increased in the presence of I-SceI expression, and the Gluc signal was elevated only under co-expression of I-SceI and the trGluc donor template ( Figure 1E ).

As an alternative to I-SceI-mediated activation of BLRR, we investigated whether the BLRR can also report CRISPR/Cas9-induced DSB repair. Based on scores predicted by Benchling (http://www.benchling.com) and CHOPCHOP (38) (Supplementary Table 2 ), we selected six gRNA target sites within the I-SceI cut site to examine BLRR sensitivity for reporting gRNA editing efficiency (Figure 2A) .

We first performed in vitro cleavage assays with gRNAs to estimate the editing efficiency and correlate with Benchling and CHOPCHOP on-target scores, and BLRR assay data are consistent with NGS results. To examine BLRR assay sensitivity, increasing amounts of pX330-gRNA and trGluc were introduced into BLRR cells to examine whether BLRR activity rises as DSB repair is increased. Both

Gluc and Vluc signals rose when the total number of transfected plasmids increased ( Figure 3A) , demonstrating that the BLRR can quantitatively measure HDR and NHEJ. Next, we performed NGS analysis on the same cells used to generate the results shown in Figure 3A , and observed a similar increase in HDR and NHEJ measured by the BLRR assay ( Figure 3B) . By comparing the two assays, we verified the detection limit of Vluc to be around 14.7 ± 1.41% of NHEJ, suggesting this may be the NHEJ detection limit of BLRR ( Figure 3B , 90+90 ng). By contrast, the Gluc signal has a detection limit of 1.23 ± 0.32% of HDR ( Figure 3B , 60+60 ng),

indicating that the BLRR system is more sensitive for detecting HDR than NHEJ.

Notably, we observed a robust correlation between BLRR signals and NGS results;

the coefficient of determination (R 2 ) between the Gluc signal and HDR% was 0.9722 ( Figure 3C ) and the R 2 value between the Vluc signal and NHEJ% was 0.919 ( Figure 3D) . To further validate BLRR sensitivity for reporting the type and frequency of DSB repair, an increasing amount of trGluc combined with a fixed quantity of pX330-gRNA were transfected into BLRR cells. BLRR analysis showed that the Gluc signal rose as trGluc was increased, indicating elevated HDR events ( Figure 3E) . Concurrently, NGS analysis of the same cells used to generate the results shown in Figure 3E demonstrated an increase in HDR events ( Figure 3F) . Although several DDR reporters have been established, their applications have been largely restricted to cell culture models. Hence, we tested whether the BLRR could detect HDR/NHEJ in small animal models through ex vivo monitoring of Gluc and Vluc activities in blood samples ( Figure 5A) . We stably transfected 293T cells with BLRR+trGluc+I-SceI (active BLRR reporter) or BLRR+trGluc (negative control), and subcutaneously implanted the resulting cells in the flanks of nude mice.

As the tumour size increased (Supplementary Figure 5) , an increase in Gluc (HDR)

and Vluc (NHEJ) activities was observed starting on Day 21 post-implantation in mice bearing 293T-BLRR+trGluc+I-SceI tumours, and signals increased significantly over time (Figure 5B, C) . By contrast, low BLRR signals were detected in the 293T- (45) . Following NU7441 treatment, the Gluc signal increased as the Vluc signal decreased in a dose-dependent manner ( Figure 6A) . The BLRR ratio (Gluc activity divided by Vluc activity) exhibited a dose-dependent increase, suggesting that it can be applied to assess the dynamics between HDR and NHEJ events (Figure 6B) . The same cells were further analysed by TIDER assay (Supplementary Figure 6A) , and the value of HDR%/NHEJ% was strongly correlated with the BLRR ratio (R 2 = 0.9594; Figure 6C and Supplementary Figure   6B ). To support the BLRR results, we also examined the expression levels of key components in HDR and NHEJ pathways, namely RAD51 and phosphorylated DNA-Pkcs, and observed a dose-dependent decrease in the percentage of phosphorylated DNA-PKcs (Supplementary Figure 6C) . By contrast, treatment with B02 resulted in a dose-dependent decline in Gluc activity in BLRR cells ( Figure 6D) . Although

Vluc activity also decreased with an increasing dose of B02, the BLRR ratio showed a dose-dependent decrease, suggesting that HDR was suppressed by B02 ( Figure 6E ).

TIDER analysis corroborated the BLRR assay findings, and revealed a correlation between the BLRR ratio and HDR%/NHEJ% (R 2 =0.7411; Figure 6F and Supplementary Figure 7A, B) . In addition, we observed reduced DNA-PKcs expression following B02 treatment, which likely resulted in the decreased Vluc signals, especially at higher dosages (Supplementary Figure 7C) . These results indicate that BLRR signals and the BLRR ratio can be applied to investigate the effect of small molecules or other modalities in modulating DSB repair, which is of relevance to high-throughput screening and preclinical studies.

Genomic instability and enhanced DNA repair are defining features of tumour cells (46) . In fact, upregulation of DDR contributes to increased therapeutic resistance in stem-like tumour populations (7, 47, 48) . Therefore, we tested whether BLRR can detect modulated DSB repair events in patient-derived GBM cancer stem cells (GSCs) ( Figure 7A) . As a positive control for BLRR detection of HDR and NHEJ activities,

GSCs were transfected to co-express BLRR, trGluc and I-SceI, and a marked increase in Gluc activity (400-fold) was observed (Supplementary Figure 8) . By contrast, only Vluc activity could be readily detected following co-expression of BLRR and I-SceI. Background Gluc and Vluc signals were detected in BLRR+trGluc and mock controls. These results indicate that the BLRR reports NHEJ and HDR events in GSCs with high specificity.

We recently reported that pharmacological inhibition of stearoyl-CoA desaturase 1 (SCD1) with CAY10566 (CAY) downregulates the HDR protein RAD51 in GSCs as an anticancer strategy (49) . Therefore, we first examined whether treating GSCs with CAY impairs HDR function. Notably, applying CAY to GSCs expressing

BLRR+trGluc+I-SceI at sub-toxic nanomolar concentrations revealed a significant reduction in Gluc activity and BLRR ratio as the amount of applied CAY increased, thereby indicating an HDR-suppressing effect for CAY in GSCs (Figure 7B, C) .

sulphoxide (DMSO) controls. These results suggest that the BLRR accurately reports the effects of compounds on DNA DSB repair in GSCs.

We previously identified cardiac glycosides as potential glioma therapeutics, but their involvement in DSB repair remains poorly understood (50, 51) . To investigate the possible DSB repair-modulating effects of cardiac glycosides, human U87 GBMs as well as GSCs stably expressing BLRR were treated with low nanomolar doses of ouabain, lanatoside C, or digoxin, and BLRR assays were performed. Remarkably, cardiac glycosides significantly reduced Gluc activity and the BLRR ratio, while Vluc activity remained similar in both U87 and GSC cells, demonstrating suppression of HDR in both cell types (Figure 7D−G) . To elucidate the mechanism of cardiac glycoside-mediated HDR inhibition, we examined RAD51 expression in treated cells, and discovered that all three cardiac glycosides triggered a dose-dependent downregulation of RAD51 protein expression, thus corroborating the decrease in HDR observed by BLRR assay (Figure 7H, I) . These findings reinforce the antineoplastic properties of cardiac glycosides, and unveil a novel HDR-suppressing function of these natural compounds as modulators of DDR in tumour and tumour stem-like cells. Figure 1E and Figure Herein, BLRR analysis revealed that gRNA-3 exhibited a significantly higher HDR% and NHEJ% than gRNA-1 with the two gRNAs only 30 bp apart, demonstrating that it may be used for screening optimal gRNAs for Cas9-based editing. All tested gRNAs except gRNA3 displayed similarly low Gluc activity, and TIDER analysis revealed that gRNA1 and gRNA2 yielded the highest HDR%, while gRNA5 gave the lowest HDR%. The differences between the two analyses may be attributed to fewer HDR events, below the optimal detection limit of the assays. Meanwhile, the BLRR results also demonstrated that the closer the distance between the DSB site and the HDR arm, the higher the HDR efficiency, thereby corroborating previous findings (57) . Moreover, we found that the Vluc signal declined in cells transfected with pX330-gRNA compared with the other group ( Figure 4C) . Given that NHEJ events can be elevated in the presence of donor templates(40), we speculated that the amount of transfected trGluc would decrease over the course of the experiment as cells proliferate. Consequently, cells carrying less pX330-gRNA+trGluc may proliferate faster than their counterparts, thereby resulting in an increased ratio of low plasmidcontaining to high plasmid-containing cells (i.e. an increased low NHEJ:high NHEJ cell population ratio), and consequently a decrease in Vluc signal at the latter time points. By contrast, the NHEJ activity of the pX330-gRNA group was not potentiated by the presence of trGluc donor template from the start of the experiment, hence a slower increase in Vluc signal was observed without a decline before the end of the experiment as NHEJ accumulates. Consistently, NGS analysis showed that HDR and NHEJ events decreased at 48 h (Figure 4D, E) , in line with the increased low plasmid-containing to high plasmid-containing cell ratio. Of note, the assay exhibited a ~6 h delay in reporting significantly increased NHEJ and HDR events compared with NGS analysis, though the general trends were similar between the two assays.

The time delay of BLRR is likely a result of the time required for the translation and release of Gluc and Vluc luciferases following DSB repair. Hence, whereas the BLRR cannot facilitate real-time detection, it enables time-lapsed monitoring of the trends of HDR and NHEJ events while keeping cells intact. By taking advantage of the high signal-to-noise ratio of Gluc and Vluc activity and the secreted luciferases, we showed that the BLRR platform can be used for longitudinal and non-invasive monitoring of HDR and NHEJ in vivo. We speculate that the significant increase in the BLRR signal from day 21 to day 28 likely reflects Gluc/Vluc reaching a detectable level in the blood during this period. As tumours grew, BLRR luciferases were constantly secreted, and the signals could only be detected in the blood once the signal-to-noise ratio is >1. We predict that an engineered mouse model with tissuespecific activation of BLRR could be established to study precise genome editing, including targeted delivery of transgenes, editing activity, and DDR dynamics. Efforts are currently underway to evaluate the ability of the BLRR multiplex assay to predict the efficacy of HDR inhibitors in mouse orthotopic GSC brain tumour models. (14, 62, 63) . Radiation therapy and chemotherapeutics such as the alkylating agent temozolomide (TMZ) induce lethal DSB. However, increased HDR repair is identified as a common feature of several malignancies such as GBM, as well as recurrent tumours (64, 65) . By repairing DSB, an increase in HDR contributes significantly to acquired radioresistance (7) and TMZ resistance (65) . Furthermore, GSCs are more resistant to DNA damage than their non-GSC counterparts (66, 67) . For instance, RAD51 contributes to the resistance of GSCs to TMZ (8) , and confers resistance to radiation therapy in GBMs and GSCs. To first confirm whether BLRR can detect altered DSB repair induced by small-molecule modulators, we applied NU7441 and B02 and observed dose-dependent HDR enhancing and suppressive effects, respectively. Notably, we found that when HDR was enhanced at higher NU7441 concentrations, NHEJ was reduced, suggesting an inverse correlation between HDR and NHEJ when the repair dynamic is significantly shifted. On the other hand, we observed that both HDR and NHEJ were reduced when HDR was suppressed by B02 at higher concentrations. Consistently, we observed a decrease in DNA-PKcs expression at higher B02 concentrations, which coincides with the reduced NHEJ events (Supplementary Figure 7) . Although the presented Gluc and Vluc values were normalised against cell viability, we also speculate that the decrease in both HDR and NHEJ may be partly attributed to cell stress and/or cell death induced by high concentrations of B02 (68, 69) . Furthermore, we found that the BLRR ratio (i.e. Gluc:Vluc) may prove to be a more accurate assessment of the ability of compounds to influence DNA repair mechanisms. Taken together, the results imply that the BLRR enables analysis of the altered dynamics of DSB repair induced by small-molecule modulators.

We recently showed that inhibition of fatty acid desaturation mediated by SCD1 depletes RAD51, thereby increasing DNA damage and sensitivity to TMZ in patientderived GSCs (28) . However, whether HDR efficiency is affected by inhibition of fatty acid desaturation remains unknown. In the current study, the BLRR assay revealed dose-dependent HDR reduction induced by CAY treatment, thereby validating these findings, and confirming that pharmacological inhibition of SCD1 Significance was calculated using one-way ANOVA as indicated, followed by Tukey's post-hoc test (*p <0.05, **p <0.01, ***p <0.001, ****p <0.0001). Vluc (C) activities compared with the BLRR+trGluc control group. BLRR signals are shown as the fold change relative to day 7 (presented as mean ± SEM of three mice).

Significance was calculated by two-way ANOVA as indicated, followed by Ś ídák's multiple comparisons test (**p <0.01, ****p <0.0001). 

The mechanism of double-strand DNA break repair by the nonhomologous DNA end-joining pathway

Nonhomologous DNA end joining and alternative pathways to double-strand break repair

Therapeutic genome editing: prospects and challenges

DNA double-strand break repair inhibitors as cancer therapeutics

Targeting DNA Double-Strand Break Repair Pathways to Improve Radiotherapy Response

RAD51 up-regulation bypasses BRCA1 function and is a common feature of BRCA1-deficient breast tumors

RAD51 Is a Selective DNA Repair Target to Radiosensitize Glioma Stem Cells

Rad51 inhibition is an effective means of targeting DNA repair in glioma models and CD133+ tumor-derived cells

RNA-guided human genome engineering via Cas9

A guide to genome engineering with programmable nucleases

Post-translational Regulation of Cas9 during G1 Enhances Homology-Directed Repair

Enhanced homologydirected human genome engineering by controlled timing of CRISPR/Cas9 delivery

Increasing Cas9-mediated homology-directed repair efficiency through covalent tethering of DNA repair template

Inhibition of homologous recombination in human cells by targeting RAD51 recombinase

Pharmacological inhibition of DNA-PK stimulates Cas9-mediated genome editing

Small molecules enhance CRISPR genome editing in pluripotent stem cells

Quantifying genome-editing outcomes at endogenous loci with SMRT sequencing

Detection of mutations by cleavage of DNA heteroduplexes with bacteriophage resolvases

Mutation detection using Surveyor nuclease

A Survey of Validation Strategies for CRISPR-Cas9 Editing

Death receptor-based enrichment of Cas9-expressing cells

Comparison of T7E1 and surveyor mismatch cleavage assays to detect mutations triggered by engineered nucleases

Genome engineering using the CRISPR-Cas9 system

Easy quantitative assessment of genome editing by sequence trace decomposition

Easy quantification of template-directed CRISPR/Cas9 editing

Analyzing CRISPR genome-editing experiments with CRISPResso

Quantifying on-and off-target genome editing

Chimeric nucleases stimulate gene targeting in human cells

Structure-based redesign of the dimerization interface reduces the toxicity of zinc-finger nucleases

Improving CRISPR-Cas nuclease specificity using truncated guide RNAs

Tracking genome engineering outcome at individual DNA breakpoints

Bioluminescence in the sea

A versatile viral system for expression and depletion of proteins in mammalian cells

Production of Purified CasRNPs for Efficacious Genome Editing

CRISPR RNAs trigger innate immune responses in human cells

A secreted luciferase for ex vivo monitoring of in vivo processes

Development of 2A peptide-based strategies in the design of multicistronic vectors

CHOPCHOP v3: expanding the CRISPR web toolbox beyond genome editing

Multiplex genome engineering using CRISPR/Cas systems

Enhancing homology-directed genome editing by catalytically active and inactive CRISPR-Cas9 using asymmetric donor DNA

Nonhomologous DNA increases gene disruption efficiency by altering DNA repair outcomes

cDNA cloning and characterization of a secreted luciferase from the luminous Japanese ostracod, Cypridina noctiluca

Advances in genome editing through control of DNA repair pathways

Identification of a highly potent and selective DNA-dependent protein kinase (DNA-PK) inhibitor (NU7441) by screening of chromenone libraries

Identification of specific inhibitors of human RAD51 recombinase using high-throughput screening

The DNA damage response and cancer therapy

Glioma stem cells promote radioresistance by preferential activation of the DNA damage response

Stem cell marker CD133 affects clinical outcome in glioma patients

Stearoyl CoA Desaturase Is Essential for Regulation of Endoplasmic Reticulum Homeostasis and Tumor Growth in Glioblastoma Cancer Stem Cells

Functional drug screening assay reveals potential glioma therapeutics

Lanatoside C sensitizes glioblastoma cells to tumor necrosis factor-related apoptosis-inducing ligand and induces an alternative cell death pathway

Quantitative live cell imaging reveals a gradual shift between DNA repair mechanisms and a maximal use of HR in mid S phase

Regulation of homologous recombination in eukaryotes

A Comparison of Techniques to Evaluate the Effectiveness of Genome Editing

Dramatic enhancement of genome editing by CRISPR/Cas9 through improved guide RNA design

Rational design of highly active sgRNAs for CRISPR-Cas9-mediated gene inactivation

Double Nicking by RNA-Guided CRISPR Cas9 for Enhanced Genome Editing Specificity

Surrogate reporter-based enrichment of cells containing RNAguided Cas9 nuclease-induced mutations

DNA double-strand breaks: signaling, repair and the cancer connection

DNA repair pathways as targets for cancer therapy

DNA repair dysregulation from cancer driver to therapeutic target

DNA-PK inhibition by NU7441 sensitizes breast cancer cells to ionizing radiation and doxorubicin

Dynamin impacts homology-directed repair and breast cancer response to chemotherapy

Rad51 protein expression and survival in patients with glioblastoma multiforme

Augmented HR Repair Mediates Acquired Temozolomide Resistance in Glioblastoma

Ionizing radiation in glioblastoma initiating cells

Chemoresistance and chemotherapy targeting stem-like cells in malignant glioma

A small molecule inhibitor of human RAD51 potentiates breast cancer cell killing by therapeutic agents in mouse xenografts

A highly sensitive assay for monitoring the secretory pathway and ER stress

Na+, K+-ATPase inhibitor, ouabain accentuates irradiation damage in human tumour cell lines

Ouabain sensitizes tumor cells but not normal cells to radiation

Cardiac glycosides suppress the maintenance of stemness and malignancy via inhibiting HIF-1alpha in human glioma stem cells

Systems biology-based drug replicates). (F) TIDER analysis of D (Supplementary Figure 7A, B) reveals a linear correlation between BLRR ratio and HDR%/NHEJ% (R 2 = 0.7411)

We are grateful for Dr. Hiroaki Wakimoto for providing Primary GBM cells used in this study. We thank Dr. Bakhos Tannous for providing some of the reagents used in our study and for his valuable input. We acknowledge the MGH Vector Core for producing the viral vector supported by NIH/NINDS P30NS04776. We are grateful for Dr. Mei-Yeh Lu for consultation on NGS sequencing, and service from NGS core at BRCAS in Academia Sinica. We would like to acknowledge the service provided by the DNA Sequencing Core of the Centre for Biotechnology, National Taiwan University.

The authors declare that they have no competing interests.