Multiplexed epigenetic memory editing using CRISPRoff sensitizes glioblastoma to chemotherapy

Abstract

Background: Glioblastoma (GBM) carries a poor prognosis, and new therapeutic strategies are necessary to improve outcomes for patients with this disease. Alkylating chemotherapies including temozolomide (TMZ) and lomustine (CCNU) are critical for treating GBM, but resistance mechanisms, including hypomethylation of O6-methylguanine-DNA methyltransferase (MGMT) promoter, undermine treatment. CRISPRoff is a programmable epigenetic memory editor that can induce stable and heritable gene silencing after transient delivery, and we hypothesize that CRISPRoff could potentiate the activity of TMZ and CCNU through long-term suppression of target genes.

Methods: We transiently delivered CRISPRoff mRNA along with sgRNAs against target genes using both electroporation and lipid nanoparticles (LNPs) into established GBM cell lines, patient-derived primary GBM cultures, and orthotopic GBM xenografts. Gene repression, specificity, and stability were measured by RT-qPCR, Western blot, bisulfite sequencing, and RNA sequencing. Sensitivity to chemotherapies was measured by cell viability dose-response, microscopy, and bioluminescence imaging. Genome-wide mapping of CCNU sensitizers was performed using CRISPRi screens.

Results: CRISPRoff induced complete suppression of MGMT and sensitization to TMZ that was stable for over 8 months of continuous cell propagation. GBM orthotopic tumors treated with CRISPRoff against MGMT demonstrated sensitivity to TMZ in vivo, and CRISPRoff delivery resulted in chemosensitivity in patient-derived primary GBM. Genome-wide CRISPRi screens identified combinatorial genetic vulnerabilities (BRIP1, FANCE) that were targetable by multiplexed CRISPRoff to achieve sensitization to CCNU.

Conclusion: Transient delivery of a site-specific epigenetic memory can induce stable, complete, and multiplexed suppression of target genes for therapeutic application in GBM.

Keywords: CCNU; CRISPR; epigenetic editing; glioblastoma; temozolomide.

Figure 1.

CRISPRoff potentially and specifically represses MGMT in GBM cells. (A) Schematic of CRISPRoff mRNA and sgRNA delivery using electroporation into GBM cells, leading to gene repression through DNMT3A/3L and KRAB domains. (B) Genomic locus of the MGMT promoter with protospacer locations of the 3 independent sgRNAs targeting the CpG island, as well as 4 CpG sites significantly associated with a hypermutated subtype of glioma with improved survival reported in. (C) RT-qPCR of MGMT mRNA normalized to RPLP0 housekeeping gene, in polyclonal and monoclonal LN18 cells at the indicated number of days following delivery of CRISPRoff and the indicated sgRNA. Primer Set K and O represent 2 independent qPCR primer pairs from (Kreth et al. 2011) and Origene, respectively. (D) RT-qPCR of MGMT mRNA as in (C) for LN18 cells delivered with a pool of all 3 sgRNAs against MGMT. (E) RT-qPCR of MGMT mRNA in polyclonal and monoclonal T98G cells following delivery of CRISPRoff and indicated sgRNA(s). (F) Western blot against MGMT protein in polyclonal and monoclonal LN18 cells at the indicated number of days following delivery of CRISPRoff and the indicated sgRNA. (G) Western blot against MGMT protein as in (F) for LN18 cells delivered with a pool of all 3 sgRNAs against MGMT. (H) Western blot against MGMT protein as in (F) for T98G cells. (I) Targeted bisulfite sequencing of the MGMT promoter, assessing 17 CpG sites in LN18 monoclonal cells following delivery of CRISPRoff and the indicated sgRNA. (J) RNA-seq volcano plots comparing CRISPRoff targeting of MGMT promoter across 2 independent sgRNAs (sgMGMT-1B, -3B) in intermediate (40 – 64d) and late (134–209d) time points.

Figure 2.

CRISPRoff repression of MGMT exhibits long-term stability. (A) RT-qPCR of MGMT mRNA normalized to RPLP0 housekeeping gene, in monoclonal LN18 cells at the indicated number of days following delivery of CRISPRoff and the indicated sgRNA(s). (B) RT-qPCR of MGMT mRNA as in (A) for LN18 cells continuously passaged long term. (C-D) RT-qPCR of MGMT mRNA for T98G monoclonal populations continuously passaged long term following delivery of CRISPRoff and the indicated sgRNA(s). (E) Western blot against MGMT protein in monoclonal LN18 cells at the indicated number of days following delivery of CRISPRoff and the indicated sgRNA. (F) Western blot against MGMT protein as in (E) for LN18 cells delivered with a pool of all 3 sgRNAs against MGMT. (G) Western blot against MGMT protein in monoclonal LN18 cells continuously passaged long term following delivery of CRISPRoff and the indicated sgRNA. (H-I) Western blot against MGMT protein in T98G monoclonal populations continuously passaged long term following delivery of CRISPRoff and the indicated sgRNA(s).

Figure 3.

CRISPRoff repression of MGMT sensitizes GBM cells to TMZ long term. (A) Cell viability dose–response curves for temozolomide (TMZ) sensitivity in LN18 (left) and T98G (right) polyclonal populations at the indicated number of days following delivery of CRISPRoff and the indicated sgRNA(s). IC50 is calculated for each condition. (B) Cell viability dose–response curves for temozolomide (TMZ) sensitivity in LN18 (left) and T98G (right) intermediate and long-term monoclonal populations at the indicated number of days following delivery of CRISPRoff and the indicated sgRNA(s). IC50 is calculated for each condition using 3 parameter nonlinear fit on [inhibitor] versus response.

Figure 4.

CRISPRoff repression of MGMT sensitizes GBM xenografts to TMZ in vivo. (A) Schematic of CRISPRoff mRNA and sgRNA delivery using electroporation into GBM cells, followed by intracranial transplantation of monoclonal populations (2 million cells per animal, n = 5 animals per condition) into immunocompromised mice. Animals were treated with TMZ between days 6–10 at 500 µg/kg/day by oral gavage. (B) Bioluminescent imaging (BLI) of orthotopic xenografts of CRISPRoff modified LN18 cells. P value = 2-way ANOVA comparing across all groups at day 55. (C) Representative BLI images for orthotopic xenografts for each CRISPRoff modified LN18 tumor at the specified timepoint post-transplantation. (D) Time to complete tumor regression, as measured by BLI, for orthotopic tumors following 5d TMZ treatment between days 6–10. P value = log-rank test comparing all time-to-event curves across the entire duration of the experiment. (E) H&E and immunohistochemistry against MGMT or cleaved caspase 3 from orthotopic tumors 30 minutes following the final dose of TMZ (day 10). Representative images are shown for n = 3 – 4 tumors per condition. (F) Quantification of IHC against MGMT or cleaved caspase 3. P value = 2 tailed Student’s t-test.

Figure 5.

CRISPRoff targeting of MGMT in primary GBM. (A) Western blots against MGMT protein in monoclonal (SF7996) or polyclonal (SF14259, SF14346, SF14590, SF14599) populations of patient-derived primary GBM cultures at the indicated number of days following delivery of CRISPRoff and the indicated sgRNA(s). (B) Crystal violet assay of cell viability 7 days following cell seeding and TMZ treatment (replenished every 3 days) at the indicated concentrations for SF7996 primary GBM. Quantification of cell growth (right). Scale bar = 1000 µm. (C) Propidium iodide cell cycle analysis of SF7996 cells treated with TMZ. P value = 2 tailed Student’s t-test for S-phase compared to sgScrambled 20 µM TMZ condition, n = 3 biological replicates. (D) Brightfield light micrographs of CRISPRoff-modified primary GBM cultures (SF14259, SF14346, SF14590, SF14599) 15–35 days following cell seeding, corresponding to the time at which vehicle-treated cells reached confluency. TMZ was replenished every 3 days during culture. Integrated intensity from live cells was quantified (right) for each cell type. (E) Apoptosis assay of primary GBM cultures 7 days following cell seeding and TMZ treatment. P value = 2 tailed Student’s t-test compared to cognate sgScrambled.

Figure 6.

Multiplex targeting of CCNU sensitizers identified through Genome-wide CRISPRi screens. (A) Schematic of genome-wide CRISPRi screens in GBM cells stably expressing dCas9-Zim3. Dual sgRNA libraries were transduced via lentivirus. Parallel screens were performed in the presence of vehicle or CCNU in LN18 as well as T98G cells. Phenotypes of CRISPRi suppression of target genes were quantified using targeted sequencing of integrated sgRNA barcodes. (B) Waterfall plot of all gene targets from genome-wide CRISPRi screens in LN18 (left) and T98G (right) ranking the log2 ratio of sgRNA barcodes in the CCNU vs. vehicle conditions at the endpoint of each screen. Phenotypes are the average across 3 replicate screens. Light gray = non-targeting sgRNA. Dark gray = gene targeting sgRNA. (C) Gene set enrichment analyses of CCNU screens, ranked by the topmost negative enrichment terms. (D) RT-qPCR of BRIP1 or MGMT mRNA normalized to RPLP0 housekeeping gene, in polyclonal LN18 cells following delivery of CRISPRoff mRNA and the indicated combination of sgRNAs. (E) as in (D) but for T98G. (F) Cell viability dose–response curves for CCNU sensitivity in LN18 (left) and T98G (G) polyclonal populations following delivery of CRISPRoff and the indicated combinations of sgRNAs. IC50 is calculated for each condition.