You'd Prefer An Argonaute

CRISPR RNA maturation by trans-encoded small RNA and host factor RNase III

Elitza Deltcheva, Krzysztof Chylinski, Cynthia M. Sharma, Karine Gonzales, Yanjie Chao, Zaid A. Pirzada, Maria R. Eckert, Jörg Vogel & Emmanuelle Charpentier

Nature 471, 602–607, 31 March 2011.
doi:10.1038/nature09886

This week’s superb summary and analysis by Josien van Wolfswinkel:

Over the last couple of decades CRISPR has become known as the prokaryote version of an adaptive immune response against viruses. In contrast to the vertebrate version of immunity, this system functions not on a protein level, but via recognition of the DNA sequence of the invader by crRNAs.

The CRISPR system consists of two units that together form the CRISPR locus: the “clustered, regularly interspaced short palindromic repeats” (CRISPRs), and the CRISPR associated (cas) proteins which are encoded as a group directly adjacent to the CRISPRs. A CRISPR sequence consists of ~30nt long repeats which are interrupted by ~40nt sequences that can be derived from phages. The whole sequence is initially transcribed as a long single-stranded pre-crRNA, from which mature crRNAs consisting of a single spacer sequence with some surrounding repeat sequence are processed by the cas proteins. Based on the combination of proteins encoded in the cluster, 8 types of CRISPR clusters can be distinguished. Most of them contain an enzyme that is known to perform the RNA processing reaction, however in two types of clusters (Nmeni and Dvulg) no processing enzyme had been found so far.

In this paper Delcheva et al. describe a novel pathway for the processing of mature crRNA from the pre-crRNA primary transcript. First, the authors established that many clinical isolates of Streptococcus pyogenes contain two types of CRISPR clusters (Nmeni and Dvulg), of which only the first type is expressed. The S. pyogenes Nmeni cluster produces mature crRNAs with a 5′ monoP, suggesting that these are not primary transcripts, yet none of the known CRISPR processing enzymes is encoded among the cas genes of this cluster. The authors identified a locus adjacent to the CRISPR cluster, which produces two primary transcripts and one processed RNA species at high levels, and named this the trans-activating crRNA (tracrRNA). Closer inspection of the locus sequence revealed a 25 nucleotide stretch present in both primary tracrRNA transcripts that has almost perfect complementarity to part of the repeat in the pre-crRNA primary transcript. Notably, the 5′ end of the processed tracrRNA, as well as the 3′ end of mature crRNAs are located within this basepairing region.

The authors used deletions of both tracrRNA and pre-crRNA loci to show that the production of mature crRNA depends on co-processing with the tracrRNA. The positioning of the cleavage sites on the pre-crRNA/tracrRNA duplex shows a 2nt 3′ overhang, suggestive of processing by an RNase III type enzyme. Indeed deletion of the S. pyogenes RNase III gene rnc abolished the co-processing, and recombinant Rnc was sufficient to drive co-processing of pre-crRNA and tracrRNA in vitro. In vivo however, the cas gene csn1 (but none of the other cas genes) was also required.

The authors then asked whether this CRISPR cluster can effectively confer resistance to invading phages or plasmids. They created a plasmid containing a sequence identical to one of the spacers in the CRISPR locus and found that this plasmid cannot be transfected into wildtype S. pyogenes, but is accepted by mutants in pre-crRNA, tracrRNA, rnc, or csn1.

Finally the authors identified tracrRNA loci in other species carrying the Nmeni type CRISPR cluster, and show that RNA from these loci is expressed and processed with similar dynamics as in S. pyogenes. Therefore, the mechanism described in this paper may well be a general mechanism for processing crRNA from Nmeni type clusters.

Technically, the paper is solid, but it is the conceptual aspects of it that make it remarkable. First, there have been many indications that CRISPR clusters have been transmitted between bacteria by horizontal gene transfer, and so far the clusters seemed to function as autonomous entities, which are independent of the rest of the bacterial genome. The mechanism for crRNA processing described in this paper is the first report of CRISPR dependency on unlinked loci (i.e. the Rnc that is required for the processing is not present in the CRISPR cluster). It is unclear whether this is due to a loss of independence of the Nmeni cluster, or whether the Nmeni cluster actually represents an ancestral minimal version of the CRISPR system. Second, these Nmeni crRNAs lack the 8nt 5′ repeat-derived tag–defined by positioning of the processing enzyme– characteristic of previously studied crRNAs in other species. The presence of these repeat-derived tags has also been shown to be important for the discrimination between self and non-self. In contrast, in the Nmeni-specific mechanism described in this paper, the repeat-derived tag is on the 3′ end, and therefore it is this end of the mature crRNA that is precisely defined (in this case by the Rnc and the base-pairing with the tracrRNA). This suggests that the mechanism of self versus non-self discrimination could be similar even between classes of crRNAs that differ in the positioning of their repeat-derived tags at opposite ends.

Finally, there is the very tempting parallel between this type of crRNA processing in prokaryotes and the diverse regulatory small RNA pathways in eukaryotes. Both systems use the functionality of an RNase III type enzyme to create small RNAs that are used in a silencing response. The remaining parts of the biogenesis pathways and the modes of silencing differ substantially, but nevertheless, it is interesting that the use of RNase III for regulatory small RNA processing is so widespread.