Genome Editing in Drug Discovery. Группа авторов
cells are typically monitored by targeted analysis of the genomic loci by a number of assays routinely offered by CROs, including the T7 endonuclease‐based Surveyor assay (Qiu et al. 2004), or the sequencing‐based TIDE assay (Brinkman et al. 2014). A comprehensive assessment of OTEs would require a genome‐wide approach (Zischewski et al. 2017), which are typically resource intensive. Even mapped, it is still unclear to what extent the detected OTEs will impact interpretation of the experimental results. More practically, OTEs can be controlled by analyzing multiple independent clones that are derived either from using the same gRNAs (or a mix of 2–3 gRNAs) or ideally from using different gRNAs. Given that each gRNA has its unique off‐target sites to which the OTEs are most likely to occur, different gRNAs targeting the same gene or locus thus will most likely have different OTE profiles. Further, it is a good practice to also obtain “wild type” (WT) clones for use as controls. These WT clones are derived from the same CRISPR editing reaction yet are wild type at the targeting site, but may carry indels at off‐target sites. Such WT clones thus offer better controls than unedited cells since they have the potential to account for OTEs. The number of independent clones to analyze will vary dependent on a balance of quality, time, and cost. Scientifically, it depends on multiple factors including homogeneity of the parental cells, since clonal heterogeneity can lead to confounding or even invalid results (Ben‐David et al. 2018). Also, editing specificity and efficiency need to be considered. For practicality, it is commonplace to analyze 3–5 independent edited clones along with 3–5 WT clones. Statistically, the more clones analyzed, the higher confidence one will have in the resulting data.
4.5.4 Deciding on Specific Quality Control Experiments on Engineered Cells
A key part of the research contract is to define quality control assays for the editing project, in addition to the prerequisite sequence confirmation of the editing site (on‐target editing). While these assays are not always essential, and will require additional effort and cost, certain assays can provide additional evidence supporting the quality of the resulting edited cells.
4.5.4.1 Confirmation of Gene KO at Protein Level
Ideally, gene KO clones with frameshift mutations should be further confirmed via protein analysis (e.g. Western) for depletion of the protein at the expected protein size, as long as a specific antibody is available. While this is observed for many genes in cellular KO experiments, deviations have been observed. In a recent study by Smits and colleagues, systematic characterization of frameshift KO mutations in HAP1 cell lines revealed that one‐third of the quantified targets were still detected at variable levels from low to original (Smits et al. 2019). Further studies of these frameshift KO clones, as well as other studies on genetically confirmed KO clones, revealed that residual levels of expression are either derived from introducing a premature stop codon, skipping of edited exon due to alternative mRNA splicing or from using an alternative initiation codon (Mou et al. 2017; Sharpe and Cooper 2017; Smits et al. 2019; Tuladhar et al. 2019). Dependent on the particular gene in study, and influenced by the location of the edited exon and the isoform expression profile in the cellular host, such truncated proteins may have full, partial, or even opposite activities (Smits et al. 2019; Tuladhar et al. 2019). Caution should be taken when interpreting KO effects of such clones. A practical approach to mitigate such effects is to analyze multiple independent clones derived from using gRNAs targeting different exons of the gene to obtain a concordant phenotypic observation. With a specific antibody, additional quality control assays can be carried out to analyze expression of the target protein in edited cells, including FACS analysis and cell staining. However, such assays only report on the presence/absence of the domain(s) the antibody recognizes, but not report on the size of the proteins/truncates detected, whereas Western analysis can report on both. Even with Western analysis, the results are impacted by epitope specificity of the antibody used. An N‐terminal‐specific antibody detects the full length and truncated proteins resulting from premature translation terminations; whereas a C‐terminal‐specific antibody enables detection of the full length and truncates resulted from in‐frame exon skipping as well as alternative translation initiations. In cases where a heterologous reporter gene (i.e. GFP, Luc) or an epitope tag is inserted in frame with an endogenous gene via knock‐in reactions, well‐validated reporter/epitope‐specific antibodies can be readily used in QC assays at protein levels.
4.5.4.2 Confirmation of Genetic Manipulation at RNA Level
Gene KO clones can also be QC analyzed at the mRNA level via RT‐PCR or RT‐qPCR. This is based on the assumption that mRNAs with frameshift indels are noncoding and are subjected to nonsense mediated decay (NMD) (Popp and Maquat 2016). While this is the case for many genes with frameshift indels in defined cells, some edited RNAs have been observed to escape NMD (Smits et al. 2019). Further analyses revealed that these NMD‐independent RNAs persist in cells via mechanisms involving alternative translation initiation, in‐frame exon skipping, or simply location of indel‐derived premature termination codon (Popp and Maquat 2016; Tuladhar et al. 2019). As such, RT‐qPCR is not commonly used as a quality control assay in gene KO experiments. Nevertheless, targeted RT‐PCR followed by gel electrophoresis or DNA sequencing analysis can reveal detailed information about variant RNA species transcribed from the edited gene, which can facilitate picking of appropriate cell clones for downstream functional analysis. It should be noted that RNA‐based QC assays may become essential when the edited site/sequence involves expression or function of a noncoding RNA (i.e. microRNA, lncRNA).
In designing HDR‐based knock‐in experiments for nucleotide mutation generation, a frequently raised question is whether to incorporate a silent codon mutation corresponding to the PAM sequence in the donor sequence. Such “PAM‐silent” mutations, once incorporated into the target site, will block re‐cleavage by the CRISPR nuclease. This has become a common practice for SNP/mutation generation to increase KI efficiencies especially when it is low (i.e. <5%). However, this would inevitably introduce a nucleotide mutation in the edited site, albeit as a silent codon change. This silent codon change may have an impact on mRNA (stability, secondary structure, or translation efficiency) and further confound the effect of the targeted nucleotide SNP/mutation on mRNA. It is therefore preferred not to adopt the PAM‐silent strategy, especially if the editing efficiencies are greater than 10%. If used, a QC assay at the mRNA level (i.e. RT‐qPCR) will need to be applied to assess the effect of the mutations (the on‐target nucleotide change plus PAM‐silent mutation) on mRNA among isogenic lines before functional assays are applied.
4.6 Summary
In this chapter, we reviewed important design considerations for CRISPR experiments and the major providers of reagents, and discussed ways of working with CROs for genome editing projects. Many of the considerations are equally applicable to projects carried out in‐house. Regardless of where the genome editing projects are to be conducted, we suggest applying the formula Model x Assay x Perturbation, as well as the triple constraint principle – quality, time, and cost – in designing and managing execution of each project. We hope that the guidelines discussed here will help researchers strike an appropriate balance of these principles to maximize scientific impact of this exciting technology within a given budget or resource.
Acknowledgments
We would like to thank Gerard Drewes and Michelle Kimberland for critical reading of the manuscript.
References
1 Acharya, S., Mishra, A., Paul, D. et al. (2019). Francisella novicida Cas9 interrogates genomic DNA with very high specificity and can be used for mammalian genome editing. Proc. Natl. Acad. Sci. U. S. A. 116: 20959–20968.
2 Aguirre, A.J., Meyers, R.M., Weir, B.A. et al. (2016). Genomic copy number dictates a gene‐independent cell response to CRISPR/Cas9 targeting. Cancer Discov. 6: 914–929.
3 Behan, F.M., Iorio, F., Picco, G. et al. (2019). Prioritization of cancer therapeutic targets using CRISPR‐Cas9