Optimized metrics for orthogonal combinatorial CRISPR screens

CRISPR screening has become a powerful technology to identify genetic dependencies with single-gene resolution. Genomic codependencies can be extracted with CRISPR perturbation screens with the integration of additional data layers, such as genomic and transcriptional alterations, and with that, genetic interactions can be identified. Even though these codependencies are correlated with their biological relevance, they are limited to the sample size and mutagenic spectrum of the additionally used data layers. Combinatorial CRISPR screens have been developed to accelerate the identification of biologically relevant codependencies and specifically to test for direct genetic relationships, map genetic interactions, identify multigenic dependencies, and explore complex biological questions. There are two different types of combinatorial CRISPR screening methodologies available to date: single and orthogonal screening approaches. While single refers to the use of one type of Cas nuclease, orthogonal refers to the use of different CRISPR-Cas modalities within the same experiment. Orthogonal approaches have the advantage of combining different gene perturbation applications compared to single approaches, such as double-strand breaks (knockouts), gene expression inhibition or activation (CRISPRi, CRISPRa), and theoretically enable innovative combinations with base- or prime-editing. Several combinatorial screening approaches that support single and orthogonal applications have been studied, with orthogonal approaches being established for SpCas9:SaCas9 (Big Papi) and SpCas9:Lb/AsCas12a (CHyMErA) formats. However, available data sets from these experiments do not allow side-by-side comparisons and the extraction of critical parameters needed to enable the mapping of combinatorial genotype-to-phenotype associations at scale. Even though combinatorial CRISPR screens are powerful, they suffer from scalability issues related to the large number of to-be-investigated query genes. Besides library diversity, library uniformity is a critical size-determining factor of the experimental scale, with uniform libraries supporting experimental down-scaling and improving scalability and feasibility. Despite library formats and their qualitative parameters (e.g., uniformity, completeness), different Cas nucleases and gRNA-expression systems are available for combinatorial screens, with none of them or combinations thereof yet being robustly established for combinatorial or orthogonal approaches. This is attributed to the overall lack of quantitative data comparing combinatorial CRISPR technologies. In this project, we performed a side-by-side and systematic comparison of combinatorial and orthogonal CRISPR screening approaches. In addition, we identified optimized metrics for combinatorial genotype-to-phenotype associations. Our preliminary analysis of comparing single and orthogonal SpCas9, AsCas12a, and CHyMErA screens revealed SpCas9 to perform most robustly. We demonstrated that the need for RNA processing of AsCas12a gRNAs negatively impacts the induction time and strength of gRNA-associated phenotypes. Moreover, we provide evidence that orthogonal CRISPR screens perform optimally with SpCas9 and RNA-processing-free enAsCas12a gRNAs. Combining this knowledge, we engineered an orthogonal CRISPR screening approach, called the multiplex SpCas9-enAsCas12a (multiSPAS) system, that generates highly active gRNAs for both SpCas9 and enAsCas12a and enables robust and efficient orthogonal genetic perturbations.