Viral vectors, such as adeno-associated viruses, have been studied extensively for their potential as delivery vehicles to carry CRISPR reagents (Cas endonucleases and gRNA) to cancer cells in animal models in vivo. These vectors present dual problems: they cause toxicity and/or immunogenicity, and they also display a lack of specific tissue targeting. To diminish the toxicity/immunogenicity issues seen with viral vectors, researchers have alternatively looked to lipid nanoparticles (LNPs) as delivery agents. However, like viral vectors, most LNPs are not tissue-specific, and they tend to accumulate in the liver in animal models. For targeting cancers outside of the liver, neither viral nor LNP delivery methods for CRISPR have demonstrated much success in animals such as mice and rats.
Experiment and results
In this collaborative study with IDT scientists, Rosenblum et al.  started by developing a novel LNP system by screening lipids for their ability to encapsulate Cas9 mRNA and sgRNA (highly chemically modified by IDT) simultaneously. Lipid 8 (L8) in the screen formed LNPs similar in size to another LNP (DLin-MC3-DMA) frequently used for siRNA delivery. L8 showed similar biophysical properties to DLin-MC3-DMA as well. LNPs made from either L8 or DLin-MC3-DMA were tested for their ability to deliver GFP-targeted Cas9 mRNA and sgRNA to cells stably expressing GFP in culture. Although both LNP types delivered the CRISPR reagents (as demonstrated by flow cytometry), only LNPs from L8 were able to cause knockout of GFP. After demonstrating that GFP could be disrupted in many GFP-expressing cancer cell lines in culture, the researchers investigated knockout of PLK1, a kinase essential for mitosis. Without PLK1, dividing cells (such as cancer cells in vitro) will die, as they are arrested at the G2-M phase of the cell cycle. Using the newly developed kind of LNPs to target PLK1 instead of GFP was successful in cell culture and caused cell cycle inhibition and cell death as shown by XTT and DAPI/Annexin V assays, respectively. This approach was successful with two cancer cell lines, GBM 005 (a glioblastoma multiforme cell line) and OV8 (an ovarian cancer cell line) in the laboratory condition.
The researchers’ next goal was to induce cancer in mice using GBM 005 and OV8 cells, and then treat these mice with LNPs targeting the glioblastoma and ovarian cancer, to determine the therapeutic effect of such LNPs in mice. However, before proceeding with this set of experiments, the researchers first decided to establish whether their new CRISPR-delivering LNPs (cLNPs) caused any toxicity or immunogenicity in healthy mice. They administered sgGFP-cLNPs intravenously into mice and examined the effects on liver enzymes, blood counts, and cytokines. No toxic or immune responses occurred.
Having demonstrated the safety of cLNPs in mice, the researchers proceeded to create the mouse glioblastoma model by injecting GBM 005 cells into the hippocampus of mice. Tumors were allowed to grow for 10 days. Then, sgPLK1-cLNPs were injected directly into the tumors. Editing was seen in 68% of the PLK1 loci within the first two days. As the researchers observed the mice over the next few weeks, all of the control (PBS-treated) mice died by day 40, whereas 30% of the sgPLK1-cLNP treated mice were still alive at day 60 (when the experiment concluded). This successful result was obtained even though the mice had received only one injection of sgPLK1-cLNP. Additionally, bioluminescence imaging demonstrated that the tumors in the treated mice were noticeably smaller than in the control mice.
The team next investigated the therapeutic effect of cLNPs in an ovarian cancer model, which, unlike the glioblastoma model, is characterized primarily by a propensity to metastasize. OV8 cells were injected intraperitoneally into mice, and cancers were allowed to grow and spread. Since these cancers metastasize widely in mice, it would not be possible to treat them with a single intratumoral injection as the researchers had done for the glioblastoma mouse model. Therefore, a molecularly targeted approach was used instead. The researchers used antibodies against epidermal growth factor receptor (EGFR) to coat the LNPs. EGFR was chosen as the antibody target because OV8 cells are known to express this receptor at high levels. On day 10 after injection of the cancer cells, the mice were intraperitoneally injected with anti-EGFR antibody-coated sgPLK1-cLNPs, or they received control treatments. This injection was repeated on day 17. Survival increased by 80% in sgPLK1-cLNP treated mice compared to controls. The tumors were much smaller in mice that received sgPLK1-cLNP treatment as well.
Rosenblum and colleagues concluded that their LNP system was an effective and safe way to deliver CRISPR Cas9 mRNA and sgRNA in vivo in mice. They pointed out that this is the first system able to target CRISPR reagents to metastasized mouse tumors, and that flexibility to target specific cancer cell types could be achieved by using different antibodies. They also stated that because flexibility in the choice of CRISPR target is easily attainable, this kind of approach could be taken in future research studies to target animal model cancers caused by fusion genes such as BCR-ABL, as well as cancers caused by mutations in oncogenes such as RAS. The researchers also suggested that CRISPR Cas endonucleases other than Cas9 should be studied for their efficacy with L8 LNPs as well, since other Cas enzymes could address concerns about off-target editing or improve rates of homologous recombination. Finally, the authors suggested that many more research studies, including safety studies, should be done in the future.