Protospacer selection, self/non-self discrimination and autoimmunity issues

The subjects of protospacer selection from the invader genome for incorporation into CRISPR loci and discrimination between the exogenous and the host DNA are two of the most important unanswered questions in CRISPR functioning. An increasing number of studies suggest that short sequence motifs adjacent to the protospacer are implicated in both processes, albeit with a still unknown mechanism.

Deveau et al. (2008) and Horvath et al. (2008) first identified a highly conserved

tetranucleotide motif located two nucleotides downstream of phage protospacer

sequences matching CRISPR spacers in S. thermophilus. Mutations in this motif

enabled phages to escape CRISPR immunity, suggesting a role in target recognition.

These motifs were identified as a universal CRISPR characteristic by Mojica et al.

(2009), who found that strict motif conservation was limited to 2-3 nt located one position after the 3’ end of the protospacer and introduced the term Protospacer Adjacent Motifs (PAMs). The consensus motif sequences depend on the CRISPR

repeat group, as assigned by Kunin et al. (2007). Moreover, the PAMs appear to

determine the spacer orientation in relation to the protospacer, as the spacer end that corresponds to the PAM - proximal side of the protospacer is always oriented towards

the leader sequence (figure 1.18, Mojica et al. 2009). This conserved orientation along with the fact that novel spacers are always added at the leader-proximal end of a CRISPR locus indicates a potential role in the selection of protospacer sequences and

integration procedure, presumably as a binding sequence for Cas proteins (Mojica et

al. 2009).

Figure 1.18: Orientation of

protospacers in regard to their PAM

Protospacers 1 and 2 are located in opposite strands, but are always incorporated with the PAM-proximal side towards the leader (adapted from Mojica et al. 2009).

Subsequent studies confirmed the role of the PAMs in interference efficiency, as any mutation within the motif prevented CRISPR interference and led to successful infection of the virus/plasmid carrying the respective mutation, while an intact PAM

motif was required for interference (Semenova et al. 2009; Marraffini and Sontheimer,

2010; Gudbergsdottir et al. 2010). However, studies in P. furiosus (type III-A) and S.

solfataricus (types I-A and III-B) did not detect a role for the PAM in CRISPR

interference (Hale et al. 2009; Manica et al. 2011), indicating that the importance of

this motif is still elusive.

The mechanism by which the CRISPR system distinguishes the cognate

CRISPR spacers from the invader protospacer sequences (both of which would exhibit perfect complementarity to the respective crRNA and would therefore constitute interference targets in the absence of such a mechanism) was elucidated by Marraffini and Sontheimer (2010) using engineered conjugative plasmids of Staphylococcus epidermidis (type III-A). The authors demonstrated that the differential complementarity between the generally conserved 5’ handle of crRNA and the 3’ region downstream of the target protospacer (that is, beyond the region of complete basepairing between the crRNA spacer and the target protospacer) is responsible for discrimination between self and non-self sequences and subsequent interference. This region is fully complementary only to the endogenous CRISPR repeat sequences, and if basepairing occurs (especially in positions -2, -3 and -4 of the 5’ handle), then the target DNA is protected. If no complementarity is found between the 5’ handle and

the downstream region of the protospacer, then the system proceeds with target cleavage (figure 1.19). Although the region screened for complementarity on the invader DNA corresponds to the PAM location, involvement of a specific motif was not observed in this mechanism, supporting the putative role of PAM in protospacer selection.

Figure 1.19: Model for discriminating between self and non-self DNA during CRISPR target recognition

Complementarity between the repeat-derived 5’ and 3’ handles of the crRNA and the target DNA ensures protection of the endogenous CRISPR locus (adapted from Marraffini and Sontheimer, 2010).

The deleterious consequences of autoimmunity are well known for all existing immunity systems, whether prokaryotic or eukaryotic. In the CRISPR/Cas system, this risk is manifested with the incorporation of a spacer into a CRISPR array that targets a cognate sequence. Although not abundant, self-targeting spacers have been detected

in CRISPR loci (Mojica et al. 2005; Bolotin et al. 2005; Horvath et al. 2008, 2009; Shah

et al. 2009), leading to the suggestion that, in analogy to the eukaryotic RNAi, CRISPR

arrays could also have a regulatory role within the organism (van der Oost et al. 2009).

An extensive case analysis by Stern et al. (2010) disproved this theory by

demonstrating that, although self-targeting spacers were present in 18% of the CRISPR-encoding strains analysed in this study (330 total), they do not exhibit conservation across species (as would be expected for a successful regulatory element) and are always accompanied by one of the following adaptations: deletion of the respective spacer, loss of the Cas operon, inactivation of the CRISPR locus, degeneration of the flanking repeats, mutation of the target self-protospacer, or

inferred deletion of the whole CRISPR array (Stern et al. 2010). These observations

confirm that the accidental incorporation of a self-targeting spacer leads to negative autoimmunity effects potentially lethal for the host, and various adaptations have been observed to ensure inactivation of the specific spacer.

This procedure was documented in vivo by Manica et al. (2011), when a

endogenous host gene was introduced in S. solfataricus. The attack was lethal, and the few surviving cells had either lost the plasmid, or recombination had occurred between the plasmid and the genomic CRISPR loci in order to replace the self- targeting spacer.