DMS3-42: the Secret to CRISPR-Dependent Biofilm Inhibition in Pseudomonas aeruginosa

TEXT

Clustered, regularly interspaced short palindromic repeats (CRISPR) are a topic of intense interest in microbiology. Described en masse for sequenced prokaryotic genomes in 2000 and 2002 (7, 11), a functional role for CRISPR in Streptococcus thermophilus phage defense was reported in 2007 (2). That study and subsequent studies of S. thermophilus, Staphylococcus epidermidis, and Escherichia coli contribute to the commonly reported model of CRISPR as a prokaryotic immune system (1, 6, 10). In a collection of three reports (4, 5, 13), the O'Toole group has sought to vet this model in Pseudomonas aeruginosa by studying infections with the temperate phage DMS3. Instead of conclusive evidence supporting a role for P. aeruginosa CRISPR in phage defense, the work has nicely shown that CRISPR modulate group behaviors—biofilm formation and swarming motility—in DMS3 lysogens. In this issue of the Journal of Bacteriology, Cady and O'Toole (4) present a tour de force analysis of the host and phage requirements for CRISPR-dependent biofilm inhibition and in the process raise a number of intriguing questions about the role of CRISPR in P. aeruginosa.

P. aeruginosa is an opportunistic pathogen associated with acute and chronic lung infections of cystic fibrosis (CF) patients (9), among other human infections. This bacterium has been studied extensively as a model organism for biofilm formation and quorum sensing, with each being prokaryotic group behaviors (8). In early studies, the O'Toole group recovered the temperate phage DMS3 from a clinical P. aeruginosa isolate, showing that the phage could lysogenize several P. aeruginosa laboratory and clinical strains (3). Notably, biofilm formation and swarming motility were inhibited in P. aeruginosa strain PA14 DMS3 lysogens, and this biofilm inhibition was reversed in a lysogen with a transposon insertion in the CRISPR-associated gene csy3 (13). The CRISPR region of PA14 was found to possess the CRISPR core genes cas1 and cas3 and Ypest CRISPR subtype genes csy1 to csy4 flanked by two CRISPR spacer arrays, CRISPR1 and CRISPR2. Interestingly, it was noted that CRISPR2 possesses a spacer sequence with 100% identity to a sequence within DMS3 gene 24 (DMS3-24) (13). This spacer sequence is now referred to as spacer 20 (4). Deletion of CRISPR2, but not CRISPR1, restored biofilm formation to DMS3 lysogens, as did deletion of csy3 or csy4 or transposon insertion in cas3, csy1, or csy2 (13). These results strongly supported the conclusion that the P. aeruginosa PA14 Ypest subtype CRISPR locus regulates biofilm formation in DMS3 lysogens.

In a subsequent study, Cady et al. (5) profiled Ypest and Ecoli subtype CRISPR content in a collection of P. aeruginosa clinical isolates, detecting these subtypes in 40 and 7 of 122 isolates, respectively. Based on the identity of P. aeruginosa CRISPR spacers with known prophage, temperate phage, and pathogenicity island sequences, but not with lytic phage or plasmid sequences, those authors proposed that P. aeruginosa CRISPR interacts only with chromosomally integrated mobile elements, not providing defense from infection by these elements. In support of this hypothesis, the authors generated deletions of the entire Ypest CRISPR regions in 3 P. aeruginosa strains, including PA14, and demonstrated that phage susceptibilities in these strains and in their CRISPR-positive parents possessing spacers with 100% identity to DMS3 or the temperate phage MP22 were comparable.

In work reported in this issue of the Journal of Bacteriology, Cady and O'Toole (4) build on this concept, generating an impressive repertoire of genetic constructs and bacterial strains to define specific interactions between the host CRISPR region and DMS3 prophage. The requirement for CRISPR-associated genes in the biofilm inhibition observed for DMS3 lysogens was confirmed using clean deletion strains and genetic complementation, and catalytic sites within the Csy4 and Cas3 proteins were queried. In a series of elegant experiments, Cady and O'Toole (4) demonstrated the following (figures refer to their work): (i) CRISPR2 spacer 1, and not spacer 20, is required for CRISPR-dependent biofilm inhibition (Fig. 6); (ii) specific residues within spacer 1 are required for this inhibition (Fig. 6); (iii) spacer 1 shares imperfect identity with DMS3 gene 42 (DMS3-42) (Fig. 7A); (iv) deletion of DMS3-42 in a DMS3 lysogen restores biofilm formation to CRISPR-positive cells (Fig. 7B); and (v) specific nucleotides within DMS3-42 with complementarity to spacer 1 are required for biofilm inhibition (Fig. 8). The implications of this work are impressive, providing further evidence that P. aeruginosa CRISPR interacts with a chromosomally integrated mobile element, an elucidation of the specific target within DMS3, and the revelation that this interaction is mediated by a spacer that does not share full identity with DMS3 (spacer 1) rather than by a spacer that does (spacer 20). This exciting work generates a number of interesting questions.

What is the nature of the interaction between spacer 1 CRISPR RNA (crRNA) and DMS3-42? Cady and O'Toole (4) do not favor a crRNA-DNA interaction, noting that degradation of lysogen DNA or of replicating DMS3 phage would lead to decreased cell or phage titers, respectively, in CRISPR-positive cells, which has not been observed. If the CRISPR2 array is transcribed, as predicted by the direction of the new spacer addition (5), then the spacer 1 crRNA will be oriented antisense to the DMS3-42 transcript. Could interaction of spacer 1 crRNA and the DMS3-42 transcript result in a knockdown of DMS3-42 transcript levels? This crRNA-directed, posttranscriptional regulation of prophage gene expression would add an exciting twist to CRISPR biology. However, certain data from this study seem to conflict with a role for spacer 1 crRNA in DMS3-42 transcript silencing. A DMS3-42 knockout strain is rendered defective for phage replication, suggesting that decreased DMS3-42 transcript levels resulting from crRNA interference would lead to decreased phage replication. The observation that CRISPR-positive DMS3 lysogens are not compromised for phage production (13) leads us to question a silencing model, although it is possible that the levels of transcript required for phage replication and for biofilm inhibition differ (see below). It will be interesting to determine whether DMS3-42 transcript levels vary in the presence and absence of spacer 1.

Questions also remain as to the specific effector(s) of biofilm inhibition. Cady and O'Toole (4) provide evidence that the DMS3-42 protein itself is not required. A start codon mutation in DMS3-42 abolishes phage production but does not reverse biofilm inhibition, which would be expected if the DMS3-42 protein were important for this phenotype. DMS3-42 occurs within a predicted transcriptional unit encompassing genes DMS3-32 to DMS3-52 (13), and the spacer 1 crRNA targets inside the DMS3-42 coding region (4). Therefore, if spacer 1 crRNA targets this transcript, it is possible that any of these 21 genes are involved in the biofilm inhibition observed for DMS3 lysogens.

Cady and O'Toole (4) provide evidence that spacer 1 crRNA mediates a nonidentity interaction with DMS3-42. What is the origin of spacer 1? It is thought that spacers are sampled from invading mobile elements, as has been observed for S. thermophilus (2); however, spacer 1 does not share full identity with DMS3 or other known sequences. From where did P. aeruginosa acquire it? Is it significant that this spacer may have been acquired after spacer 20, which does share 100% identity with the DMS3 sequence? Further, is there a biological role for spacer 20 in DMS3 lysogens? Based on data presented for spacer 1, can we speculate that spacer 20 regulates some other DMS3-dependent phenotype?

Finally, what is the clinical significance of CRISPR-dependent biofilm inhibition? Biofilm formation is thought to be an important contributor to the progressive decline in lung function that occurs in CF patients as a result of chronic P. aeruginosa infection (9). Similarly, phage targeting P. aeruginosa have been isolated from CF lung sputa (12). If a P. aeruginosa population possessing spacer 1 becomes lysogenized by a DMS3-like phage in situ, does this impact CF disease? Interestingly, at a low frequency, DMS3 lysogens arise in vitro that are not inhibited for biofilm formation and that have apparently lost CRISPR sequence (13). This suggests that a lysogenized population heterogeneous for CRISPR and biofilm formation might arise in the CF lung as a result of phage infection. Can we detect this population heterogeneity in in situ P. aeruginosa, and is it relevant to disease? While perhaps difficult to experimentally address, these concepts are important to consider. Regardless, it is clear that phage infection and CRISPR act together to impact clinically relevant P. aeruginosa behaviors, and we anticipate that this interaction will be the focus of much exciting future work.