SpyCas9 ancestors’ optimisation for CRISPR-Cas toolbox refinement using in vivo models

CRISPR-Cas systems represent a breakthrough technology able to facilitate genome editing in a simple and versatile manner. Among them, class II CRISPR-Cas systems are the simplest approaches based on a protein (Cas) and a guide RNA (gRNA). First, Cas protein recognizes a sequence in the genome, called protospacer adjacent motif (PAM), and cleavages only when the sgRNA sequence specifically binds to DNA target. The most used and efficient system is CRISPR-SpyCas9 from Streptococcus Pyogenes which PAM is 5’NGG. Despite the potential of CRISPR-Cas systems, some constraints can mitigate the use of this technology. For example, PAM restriction, which does not allow free access to all sequences in the genome, together with immune response activation caused by Cas proteins in eukaryotic cells, can limit the efficiency and safeness of this technology for potential biomedical application. To circumvent these barriers, different approaches have been developed such as the discovery of new natural CRISPR-Cas systems, the modification of known Cas or, more recently, the implementation of synthetic ancestral Cas (anCas). AnCas derived from an ancestral sequence reconstruction (ASR) technique. This approach is based on computational models able to infer extinct Cas proteins’ sequences from current CRISPR-Cas systems as SpyCas9. In particular, the two eldest predicted Cas predicted from CRISPR-SpyCas9 system, Firmicutes common ancestor (FCA) and Bacilli common ancestor (BCA) proved to show almost PAM-less behaviour towards the target, potentially increasing the accessability to many more sites in the genome. Moreover, the origin far back in time of these Cas proteins justify the low presence of anCas specific antibodies potentially recognised by the human immune system. Despite the potential of FCA and BCA, all the current results derive from in vitro experiments without any systematic in vivo implementation using transient formulations frequently employed in biomedical approaches such as mRNA-gRNA or ribonucleoprotein (RNP) delivery. Here, we optimised FCA and BCA activity in vivo using zebrafish (Danio rerio) as vertebrate model. First, we verified FCA and BCA behaviour on genome targets containing a 5’NGG sequence which allowed us to compare the efficiency of these proteins to SpyCas9 on these canonical targets. Second, we verified the PAM-less potential of anCas (FCA and BCA) in non-canonical sites, both 5‘NA/GN and 5‘NC/TN and compared them to SpRY, another near PAM-less system, with a preference on 5‘NA/GN-PAM targets, but which also cuts on 5’NC/TN-PAM sites. Beyond, and since it has been observed that both FCA and BCA have a strong single strand break (nicking) activity, we also tested a double nicking approach in all three PAM conditions. Therefore, to potentially enhance global editing efficacy, we targeted two regions within the same gene using gRNAs in different DNA strands to ultimately allow a double strand break. The results showed lower activity of FCA and BCA in 5’NGG PAM sites compared to SpyCas9, however, introducing two gRNAs increased their efficacy confirming their potential as nickases. Although the anCas proteins didn’t overcome SpRY in 5‘NA/GN targets, we could detect a higher efficiency of FCA in less SpRY-accesible 5’NC/TN-PAM sites. Altogether, our results show that FCA and BCA are complementary systems within the genome editing field pointing out possible biomedical applications in context where a relaxed PAM, nickase activity or low immune response are needed.