|Home | About | Journals | Submit | Contact Us | Français|
CRISPR loci consist of an array of short repeats separated by spacer sequences that match the genome of viruses and plasmids that infect prokaryotes. Transcription of the CRISPR array generates small antisense RNAs that mediate immunity against these invaders. In recent years, there has been a notable increase in the investigation of CRISPR immunity, but studies have been restricted to organisms in which genetic manipulations are possible. Therefore, there is a need for the development of simple genetic tools that facilitate the study of this important pathway. Here we describe the use of CRISPR decoys, plasmids containing a non-transcribed repeat-spacer unit that disrupt CRISPR immunity. We show that decoys abrogate immunity against conjugation in S. epidermidis to levels comparable to a CRISPR deletion mutant. This technique can be used to generate full or spacer-specific CRISPR knockdowns in organisms in which decoy plasmids can be introduced.
CRISPR (clustered regularly interspaced short palindromic repeats) loci are composed of a series of repeats separated by “spacer” sequences that match the genomes of prokaryotic viruses (phages) and other mobile genetic elements.1-4 The repeat-spacer array is transcribed as a long precursor and processed within repeat sequences to liberate small crRNAs5-10 that specify the targets of CRISPR immunity, also known as proto-spacers. CRISPR-associated (cas) genes usually flank the repeat-spacer array and encode the enzymatic machinery responsible for crRNA biogenesis and targeting. Remarkably, CRISPR-Cas loci encode for an adaptive immune system, and upon phage or plasmid infection, they provide the machinery to acquire new spacer sequences from the invader.11-13 New spacers added to the CRISPR array are then used to prevent further infection, making CRISPR loci analogous to an “immunization record” of the host. In addition to the targeting of plasmids and phages, CRISPR-Cas loci can also prevent DNA transformation14 and, therefore, constitute a general barrier against horizontal gene transfer. These unique properties of CRISPR immunity have drawn much attention and following the first experimental demonstration of CRISPR activity11 there has been an exponential increase in the number of studies of this pathway. Many of these studies are aimed at investigating the genetic requirements of CRISPR immunity and rely on the construction of mutants lacking vital elements of CRISPR-Cas loci, such as the repeat-spacer array. In spite of the wide distribution of CRISPR-Cas loci in bacteria and archaea, genetic studies have been limited to genetically tractable organisms such as Escherichia coli,12,15 Streptococcus thermophilus,11 Staphylococcus epidermidis,16 Pseudomonas aeruginosa,17 Pectobacterium atrosepticum,18 Streptococcus pyogenes8 and Sulfolobus islandicus.19 Therefore, there is a need for the development of simple genetic tools to facilitate the study of this important pathway.
CRISPR immunity of prokaryotes and RNA interference in eukaryotes are both genetic interference pathways mediated by small antisense RNAs, the crRNAs and siRNAs and microRNAs, respectively.20 One fundamental difference is that, while siRNAs and microRNAs target messenger transcripts, crRNAs target the genomic DNA of the invader (with one notable exception, the CRISPR-Cas system of P. furiosus, which destroys RNA21). In particular, microRNAs bind to the 3′UTR of target mRNAs and mediate post-transcriptional silencing.22 MicroRNA loss-of-function studies have been significantly advanced by the use of “microRNA sponges,” artificial transcripts engineered to have high levels of expression and multiple binding sites for a microRNA of interest that titrate out the microRNA pool.23 Based on this concept, we sought to develop a similar technology for the competitive inhibition of crRNAs.
S. epidermidis RP62a harbors a CRISPR-Cas system containing a spacer (spc1) that targets the nickase gene of staphylococcal conjugative plasmids and prevents their transfer.16 We decided to develop decoys that will trap spc1 crRNAs and allow the transfer of the staphylococcal conjugative plasmid pG040024 [renamed pG0(wt) in this work to distinguish it from a mutant version, see below] into this strain; i.e., knockdown spc1-mediated CRISPR immunity against conjugation. As with microRNA sponges, a requisite for a crRNA decoy is that it should provide a large number of alternative targets. Since spc1 crRNAs target DNA, we decided to introduce the proto-spacer sequence in a high copy number staphylococcal plasmid, pC194.25 However, as opposed to microRNAs, the crRNA:proto-spacer interaction leads to target destruction.21,26 Therefore, the decoy has to carry a target able to compete for crRNAs without being degraded. Previous work demonstrated that the presence of repeat sequences upstream of the nickase protospacer abrogate spc1-mediated CRISPR immunity.27 We hypothesized that crRNAs could still interact with a target flanked by repeats and, therefore, we decided to use the S. epidermidis CRISPR array as our decoy.
We first introduced pCRISPR9 into S. epidermidis RP62a. This plasmid contains the CRISPR array preceded by the leader sequence, an A/T rich sequence containing the promoter for the expression of the array (Fig. 1A). When we tested the transfer of pG0400 into this strain, we did not detect any reduction in CRISPR interference against conjugation (Fig. 1B). This is an expected result, since transcription of the CRISPR array provides additional spc1 crRNAs, the opposite effect of the decoys we sought to develop. To prevent transcription of the array, we eliminated the leader sequence and transformed the resulting plasmid, pCRISPR(dec), into the RP62a strain. When used as a recipient for pG0400 conjugation, this strain lacked CRISPR immunity against this plasmid and many transconjugants were obtained (Fig. 1B and Table 1). The disruption of CRISPR interference was equivalent to that observed for the transfer of a version of pG0400 with mutations in the nickase proto-spacer [pG0(mut)] and to the transfer of the wild-type plasmid into a Δcrispr recipient lacking the array (Fig. 1B and Table 1). The only difference we observed was the appearance of a mixture of big and small colonies (1/3‒1/8 of the colonies were smaller, depending on the experiment), a suggestion that some level of CRISPR interference is still present. We genotyped two colonies of each type in four independent experiments and found that all contained intact CRISPR-Cas loci, pG0400 and decoy sequences. These results indicate that we were able to successfully introduce a CRISPR decoy able to compete for crRNAs and disrupt CRISPR immunity.
Ideally, a decoy should be able to disrupt CRISPR immunity in a spacer-specific fashion. The pCRISPR(dec) plasmid, however, contains all spacers of the array and, therefore, it should provide a general knockdown of CRISPR interference. To construct a spc1-specific decoy, we eliminated the entire array sequence downstream of this spacer in pCRISPR(dec), generating pDR1-Spc1(wt) (Fig. 1A). This plasmid proved to be an efficient CRISPR decoy in S. epidermidis RP62a (Fig. 1B and Table 1); providing a similar disruption of CRISPR immunity as pCRISPR(dec). Finally, our model predicts that decoys sequester crRNAs through base-pair interactions. To test this, we altered the spacer sequence of the spc1 decoy, generating pDR1-Spc1(mut) (Fig. 1A). We introduced a set of nine nucleotide substitutions that are responsible for the lack of CRISPR immunity against pG0(mut).16 The altered decoy no longer disrupted CRISPR immunity toward pG0400 (Fig. 1B and Table 1), demonstrating the requirement of complementarity between the crRNA and proto-spacer sequences.
This base-pair complementarity requirement made us hypothesize that if we could disrupt the decoy-crRNA interaction in a controlled fashion, we would be able to induce CRISPR immunity, thus adding a useful feature to our decoys. We hypothesized that high temperatures could dissociate the decoy-crRNA base-pair interaction and liberate the crRNAs for interference against pG0400. To test this, we performed conjugation assays at 42°C (Fig. 2A). First we checked that the CRISPR immune response was not altered at this temperature. RP62a recipients containing either pCRISPR or pΔcrispr plasmids showed the same anti-plasmid CRISPR activity at both 42°C and 30°C (Fig. 2A). In contrast, both the full array and spc1-specific decoys lost their ability to disrupt CRISPR immunity at the higher temperature. We hypothesize that this thermosensitive phenotype may be a consequence of the presence of licensing flaking sequences in the pG0400 target, which allow a tighter binding of the target by a type III-A crRNA ribonucleoprotein targeting complex. To test if this temperature-sensitive response could be exploited to induce CRISPR targeting by a temperature shift, we grew transconjugants at 30°C for 90 min and then elevated the temperature to 42°C (Fig. 2B). Strains containing the pCRISPR(dec) or pDR1-Spc1 decoys and the pG0(wt) plasmid showed a decrease in growth rate when compared with strains containing the non-targeted plasmid pG0(mut). This indicates that the difference in growth is due to CRISPR immunity against pG0(wt), most likely due to the dissociation of the crRNA from the decoy at the higher temperature. Interestingly, cells continued growing at the pre-shift rate for an extra 90 min and then showed an inflexion in the growth curve at 180 min (Fig. 2B, open triangle). Altogether, these results show that decoy activity is temperature-sensitive, a property that can be exploited to induce CRISPR immunity.
The CRISPR pathway has three distinct phases: adaptation, crRNA biogenesis and targeting.28 In order to understand which of these phases is affected by the decoys, we measured the accumulation of spc1 crRNAs by primer extension (Fig. 2C). We found that the presence of pCRISPR(dec) does not affect the levels of spc1 crRNAs regardless of the temperature and of whether a target is present [pG0(wt) or pG0(mut), respectively]. Since the adaptation phase is not involved in the mechanism of action of decoys, these data suggest that the targeting phase is compromised.
In summary, here we present a new technology for the genetic study of CRISPR-Cas systems that we termed “CRISPR decoys.” These consist of a silent CRISPR array cloned into a high copy number plasmid but can also be designed to be spacer-specific by cloning only a single repeat-spacer sequence. While we don’t completely understand the mechanism of action of these constructs, our preliminary characterization strongly suggests that the presence of a non-transcribed, non-targeted array still allows the binding of the crRNA ribonucleoprotein complex to the decoy, effectively reducing their availability to attack the target (Fig. 3). This binding most likely occurs in the chromosomal CRISPR locus as well, but we believe that active transcription through the array may displace the crRNA ribonucleoprotein complexes and liberate them for targeting. Alternatively, there may be proteins dedicated to prevent the binding of the crRNA ribonucleoprotein complex to the chromosomal CRISPR array. It has been reported that the SSO454 protein of the Sulfolubus plasmid pNOB8 can bind CRISPR repeats.29 In addition to the molecular mechanism by which decoys prevent CRISPR immunity, it remains to be determined how broadly this technology can be applied. Depending on the cas gene content and repeat sequence, CRISPR-Cas systems can be classified into type I, II or III, each group containing different subtypes.28 S. epidermidis contains a type III-A system and it is unknown if CRISPR decoys work effectively in types I and II systems. If they do, it is also expected that the spacer-specific decoys may have a different composition, as the minimal functional repeat-spacer sequences for these types is yet to be defined. Finally, fundamentally different decoys will have to be developed for type III-B systems targeting RNA molecules,21 possibly more similar to the microRNA sponges.23
We believe that CRISPR decoys can be useful for the study of different aspects of CRISPR immunity. For example, most spacers with matches on genebank target phages and in principle CRISPR immunity should prevent infection by the targeted phage. Even if phage resistance is demonstrated experimentally, the anti-phage response of bacteria and archaea is very diverse and generally does not solely rely on CRISPR-Cas loci.30 Therefore, CRISPR-mediated defense needs to be corroborated by constructing an isogenic CRISPR mutant strain that becomes susceptible to phage attack. CRISPR decoys could provide this possibility for organisms where the generation of isogenic mutants is difficult or impossible but small plasmids can be introduced. In addition, our results indicate that a simple temperature shift can deactivate the decoys to induce CRISPR immunity. This is an invaluable tool for the study of the targeting step of the CRISPR pathway.
S. epidermidis RP62a31 and S. aureus RN422032 or OS233 cells were grown in Brain-Heart Infusion (BHI) media. The medium was supplemented with antibiotics as follows: neomycin (15 µg/ml) for selection of S. epidermidis; chloramphenicol (10 µg/ml) for selection of pC194-based plasmids and mupirocin (5 µg/ml) for selection of pG0400 plasmids.
Conjugation was performed by filter mating as described elsewhere,24 with the following changes: donor (S. aureus RN4220/pG0400 or pGOmut) and recipient (S. epidermidis RP62a) cells were cultured in BHI medium with the necessary antibiotics at 37°C overnight. Donors and recipients were equally mixed in 5 ml of fresh BHI medium and vacuum-filtered through 0.45 µm filters (Millipore). Filters were cut by half, laid on BHI agar plates and incubated overnight as follows: one half at 30°C and the other at 42°C. The next day, filters were resuspended in 3 ml of fresh BHI. Serial dilutions were then plated on BHI agar containing the appropriate antibiotics for the enumeration of recipients or transconjugants and incubated at 30°C or 42°C for 48 or 24h, respectively. Corroboration of the presence of the desired plasmid in transconjugants was achieved by extracting DNA of at least two colonies, performing PCR with primers that amplify the construct (L86-f, 5′- CATATAGTTTTATGCCTAAAAACC-3′; L87-r, 5′-ATATATTTATTTGGCTCATATTTGC-3′) or pG0400 (L70-f, 5′-AAAAAAGCTTCAAGAATCCAATGAAGTAGGGG-3′; L71-r, 5′-AAAAAAGCTTCTAAATTAGAACATGATACTAACG-3′) and sequencing the resulting PCR product.
Construction of pCRISPR and pΔcrispr was previously described9 and were transformed into S. epidermidis Δcrispr16 for this study as described previously.16 Construction of plasmid pCRISPR(dec) was described in27 as pDR1(wt); we changed the name for clarity. Plasmid pDR1-Spc1(wt) was generated by PCR using primers L55-f (5′-TAAATCTAACAACACTCTAA-3′) and A11-r (5′-ACAAATGCCATCACAACTATATTTCAAGCATC-3′) and pCRISPR(dec) as template. The PCR product was 5′-phosphorylated using T4 Polynucleotide Kinase (New England Biolabs) and then circularized using T4 DNA ligase (New England Biolabs). The reaction mix was transformed into S. aureus OS2. Plasmids from individual colonies were isolated, their mutations were corroborated by sequencing and transformed into S. epidermidis RP62a as described previously.16 Plasmid pDR1-Spc1(mut) was generated by PCR using primers L107-f (5′-CGCATGCCAAAATACATTAATCACCAATATAAGGG-3′) and L108-r (5′- CCCTTATATTGGTGATTAATGTATTTTGGCATGCG-3′) and pDR1-Spc1(wt) as template. The PCR product was cut with SphI (New England Biolabs) and then circularized using T4 DNA ligase (New England Biolabs), followed by transformation of the ligation products into S. aureus OS2.
Transconjugants containing decoy plasmids obtained at 30°C were grown overnight, diluted 1/100 and incubated at the same temperature for 90 min. At this time, temperature was shifted to 42°C for an extra 180 min. Cultures were grown in triplicates on 96-well plates using a Tecan Infinite 200 Pro plate reader to measure OD600 every 15 min.
Primers P1 (5′-TTTGTACTGATGATTTATATAC-3′) and PrrfA (5′-GTGACCTCCTTGCCATTGTC-3′) were used for spc1 crRNA and 5S rRNA primer extension, respectively, using a protocol previously described.16 The expected lengths of primer extension products are 26 (5S rRNA) and 42 (spc1 crRNA) nucleotides.
We would like to thank David Bikard for critical discussion of the experiments. L.A.M. is supported by the Searle Scholars Program, the Rita Allen Scholars Program, an Irma T. Hirschl Award and a NIH Director’s New Inovator Award (1DP2AI104556-01).
No potential conflicts of interest were disclosed.
Previously published online: www.landesbioscience.com/journals/rnabiology/article/24287