Impact of Different Target Sequences on Type III CRISPR-Cas Immunity

ABSTRACT

Clustered regularly interspaced short palindromic repeat (CRISPR) loci encode an adaptive immune system of prokaryotes. Within these loci, sequences intercalated between repeats known as “spacers” specify the targets of CRISPR immunity. The majority of spacers match sequences present in phages and plasmids; however, it is not known whether there are differences in the immunity provided against these diverse invaders. We studied this issue using the Staphylococcus epidermidis CRISPR system, which harbors spacers matching both phages and plasmids. We determined that this CRISPR system provides similar levels of defense against the conjugative plasmid pG0400 and the bacteriophage CNPX. However, whereas antiplasmid immunity was very sensitive to the introduction of mismatches in the target sequence, mutations in the phage target were largely tolerated. Placing the phage and plasmid targets into a vector that can be both conjugated and transduced, we demonstrated that the route of entry of the target has no impact on the effect of the mismatches on immunity. Instead, we established that the specific sequences of each spacer/target determine the susceptibility of the S. epidermidis CRISPR system to mutations. Therefore, spacers that are more resistant to mismatches would provide long-term immunity against phages and plasmids that otherwise would escape CRISPR targeting through the accumulation of mutations in the target sequence. These results uncover an unexpected complexity in the arms race between CRISPR-Cas systems and prokaryotic infectious genetic elements.

IMPORTANCE CRISPR-Cas loci protect bacteria and archaea from both phage infection and plasmid invasion. These loci harbor short sequences of phage and plasmid origin known as “spacers” that specify the targets of CRISPR-Cas immunity. The presence of a spacer sequence matching a phage or plasmid ensures host immunity against infection by these genetic elements. In turn, phages and plasmids constantly mutate their targets to avoid recognition by the spacers of the CRISPR-Cas immune system. In this study, we demonstrated that different spacer sequences vary in their ability to tolerate target mutations that allow phages and plasmids to escape from CRISPR-Cas immunity. These results uncover an unexpected complexity in the arms race between CRISPR-Cas systems and prokaryotic infectious genetic elements.

INTRODUCTION

Clustered regularly interspaced short palindromic repeat (CRISPR) loci and their CRISPR-associated (Cas) proteins encode an adaptive immune system that defends prokaryotes from bacteriophage (1) and plasmid (2) infection. Within these loci, sequences intercalated between repeats known as “spacers” specify the targets of CRISPR immunity. Spacer sequences are acquired during infection (1), when a subset of Cas proteins insert a short sequence of the invader's genome upstream of the first repeat of the CRISPR cluster (3). Spacers are transcribed and processed into short antisense RNAs, the CRISPR RNAs (crRNAs), that are used by RNA-guided Cas nucleases to recognize and cleave a matching sequence (also known as the protospacer) in the invader phage or plasmid (4). The presence of mutants in the phage or plasmid population harboring mismatches with the spacer sequence allows these genetic elements to escape CRISPR-Cas immunity (5). CRISPR-Cas loci are very diverse and differ in the number and nature of the associated cas genes, the repeat sequence, and the number of repeat-spacer units. Based on their cas gene content, CRISPR-Cas systems have been classified into five major types, I to V, with different subtypes within each group (6). Of these different groups, only types I, II, and III have been studied in detail.

Staphylococcus epidermidis RP62A harbors a type III-A CRISPR-Cas locus with three spacers, the first one (spc1) matching a region of the nickase (nes) gene present in most staphylococcal conjugative plasmids (2), the second (spc2) matching a sequence of the cn20 gene of the staphylococcal phage CNPH82 (7), and the third without homology to GenBank sequences (Fig. 1A). The molecular mechanism of immunity provided by the S. epidermidis type III-A CRISPR-Cas system is highly elaborate and differs significantly from the mechanisms underlying type I and II CRISPR immunity. For example, target transcription is required for defense (8). This is a result of the biochemical activity of the Cas10-Csm complex encoded by this locus, which performs cotranscriptional crRNA-guided nucleolytic cleavage of the target DNA and its transcripts (9). The target sequence requirements of type III CRISPR immunity are less understood than those of the other systems. To ensure immunity, type I and II CRISPR-Cas systems require a short sequence known as the protospacer-adjacent motif (PAM) located immediately upstream and downstream of the target, respectively (5, 10–13). In addition, both type I and type II systems require a perfect match between the spacer sequence and a “seed” region within the target. In type II systems, the first 6 to 8 nucleotides (nt) upstream of the PAM compose the seed (5, 11), whereas in type I systems it is composed of nt 1 to 5 and 7 immediately downstream of the PAM (10, 13, 14). This is because the interaction of crRNA with single-stranded DNA (ssDNA) at position 6 forms a noncanonical ribbon structure in which every sixth nucleotide is rotated out of the double helix and is therefore not engaged in the formation of a base pair (15–17). In contrast, type III immunity does not seem to require a PAM. Instead, effective immunity demands the presence of substantial mismatches between the 8 nt of repeat sequence present at the 5′ end of the crRNA (the crRNA tag) and the flanking sequence of the target (the antitag) (18). A seed sequence has not been described for type III immunity. It was recently shown that the Cas10-Csm complex of Thermus thermophilus can perform crRNA-guided RNA cleavage in vitro in spite of the presence of mutations that create mismatches with the crRNA guide (19), but whether this is the case for crRNA-guided DNA cleavage is not known. In vivo, results from our work suggest that there are differences in the effects of mutations for different targets. On the one hand, the introduction of nine mutations into the nes target of pG0400, targeted by spc1 of the S. epidermidis type III-A system, completely abrogates immunity (2). On the other, the presence of five mismatches between the target on gene gp32 of the ΦNM1 staphylococcal phage and a spacer found in the type III-A CRISPR-Cas system of S. aureus MSHR1132 does not affect immunity to this phage (8).

FIG 1.

The CRISPR-Cas system of S. epidermidis and its targets. (A) Organization of the type III-A CRISPR system in S. epidermidis RP62A. This system contains 9 CRISPR-associated (cas and csm) genes, 4 direct repeats (blue boxes), and 3 spacers (yellow, green, and purple boxes). The first spacer targets the nickase gene present in staphylococcal conjugative plasmids; the second matches the cn20 gene of the CNPH82 S. epidermidis phage. “L” indicates the leader sequence, which contains the promoter elements for expression of the repeat-spacer region. DB5 is a derivative of S. epidermidis RP62A that lacks the repeat-spacer array but that harbors the leader and CRISPR-associated genes. To restore CRISPR-Cas immunity in this strain, the full repeat-spacer array of S. epidermidis RP62a was cloned into the staphylococcal plasmid pC194, generating the complementing plasmid pCRISPR. WT, wild type. (B) The cn20 target contains a single mismatch in position 24, which was eliminated by introducing a compensatory mutation in the spc2 sequence of pCRISPR.

In this work, we investigated the effects of mismatches between the crRNA guide and targets with different sequences and on different mobile genetic elements using the type III-A CRISPR-Cas system of S. epidermidis. We analyzed the immunity provided by spc1 and spc2 against the pG0400 conjugative plasmid and the CNPX phage (a derivative of CNPH82), respectively. We found that, while a full spacer:target match provides efficient immunity against both pG0400 and CNPX, the introduction of mismatches results in contrasting effects for antiplasmid and antiphage immunity. Whereas mismatches in similar positions of the spacer-target interaction abrogated CRISPR targeting of the plasmid, immunity against phage infection remained intact. We explored whether differences in the entry of the genetic material and/or replication of these elements could be responsible for the dissimilar results. Placing both spc1 and spc2 targets into pC221 (20), a conjugative plasmid that can also be transduced by the CPNX phage, we determined that the effects of the mismatches are not contingent on the mode of entry of the target but on the intrinsic sequence and location (lagging strand versus leading strand) of each target. These results reveal that different spacer sequences have different degrees of tolerance of target mismatches and predict that the more resistant spacers would provide long-term immunity against phages and plasmids that would otherwise escape CRISPR targeting through the accumulation of target mutations. Our findings indicate that the intrinsic sequence of the acquired spacers could have important effects in the arms race between CRISPR-Cas systems and prokaryotic infectious genetic elements.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

S. epidermidis RP62a (21), LM1680 (22), and DB5 and S. aureus RN4220 (23) and OS2 (24) strains were grown in brain heart infusion (BHI) medium (Difco). When required, the medium was supplemented with the following antibiotics: neomycin (15 μg/ml) for selection of S. epidermidis strains, chloramphenicol (10 μg/ml) for selection of pCRISPR-based plasmids, mupirocin (5 μg/ml) for selection of pG0400, gentamicin (5 μg/ml) for selection of pGO1, and tetracycline (5 μg/ml) for selection of pC221-based plasmids.

Plasmid construction.

pCRISPR-based plasmids are derivatives of pCRISPR (18) that were produced using primers containing the desired mutation. PCR products were 5′ phosphorylated using T4 polynucleotide kinase (New England BioLabs) and then circularized using T4 DNA ligase (New England BioLabs). The reaction mixture was used to transform S. aureus OS2 by electroporation; ligated plasmids were isolated to corroborate the presence of the introduced mutations by Sanger sequencing and then used to transform S. epidermidis DB5 using a transformation protocol previously described (2). Primer sequences and their use are shown in Table 1. The construction of plasmids corresponding to fusion between pC221 (20) and pT181 (25) was performed using Gibson assembly (26) of two amplification products to exchange the chloramphenicol resistance gene from pC221 with the tetracycline resistance gene from pT181. At the same time, the spc1 or spc2 target sequences and 10 bp of their respective flanking sequences were cloned in either the leading strand or the lagging strand of the pC221-pT181 fusion plasmid. This generated four plasmids: pTgt1^Lead, pTgt1^Lag, pTgt2^Lead, and pTgt2^Lag ( for all four plasmids, primers I91 and I92 were used as reverse primers for pC221 and pT181, respectively). The assembled products were used to transform S. aureus RN4220 harboring the pGO1 mobilization plasmid (27). Plasmids were isolated and sequenced to corroborate their insertions. For transduction purposes, each of the four pTgt plasmids was used to transform S. epidermidis LM1680.

TABLE 1.

Oligonucleotides used as primers in this study

Primer	Sequence (5′–3′)	Template	Purpose(s)
I54	CCTTTTCTTCGGGGTGGGTATCGATCCGATACTTTAAC		Reverse primer used to create all spc1 mutants
I53	GGACGAGAACAGCAATGCCGAAGTATATAAATCATCAGTACAAAG	pNF03	Inverse PCR: spc1(2–4)
I68	GGACGAGAACTGCATACGCGAAGTATATAAATCATCAGTACAAAG	pNF03	Inverse PCR: spc1(1–8)
I77	GGACGAGAACACGTATGCGCTTCATAATAAATCATCAGTAC	pNF03	Inverse PCR: spc1(9–16)
I75	GGACGAGAACACGTATGCCGATCATATATAATCATCAGTAC	pNF03	Inverse PCR: spc1(12–19)
I88	GGACGAGAACACGTATGCCGAAGTATATATTAGTAGTGTACAAAG	pNF03	Inverse PCR: spc1(20–27)
I71	GGACGAGAACACGTATGCCGAAGTATATAAATCATCACATGTTTCGATCG	pNF03	Inverse PCR: spc1(28–35)
I67	CATCAGTACAAAGGATCGATACCCACCCCGAAGAAAAGG		Reverse primer used to create all spc2 mutants
I66	GGACGAGAACTTCAAATAATTGTCATTTGCATACGTTACATCG	pNF03	Inverse PCR: spc2(2–4)
I69	GGACGAGAACATCATTATATTGTCATTTGCATACGTTACATCG	pNF03	Inverse PCR: spc2(1–8)
I76	GGACGAGAACTAGTAATATAACAGTATTGCATACGTTACATCG	pNF03	Inverse PCR: spc2(9–16)
I74	GGACGAGAACTAGTAATAATTCAGTAAACCATACGTTACATCG	pNF03	Inverse PCR: spc2(12–19)
I72	GGACGAGAACTAGTAATAATTGTCATTTGGTATGCAAACATCG	pNF03	Inverse PCR: spc2(20–27)
I70	GGACGAGAACTAGTAATAATTGTCATTTGCATACGTTTGTAGCTAGATCG	pNF03	Inverse PCR: spc2(28–35)
I79	GGACGAGAACATCATTATTAACAGTAATGCATACGTTACATCG	pNF03	Inverse PCR: spc2(1–17)
I80	GGACGAGAACTAGTAATAATTGTCATTACGTATGCAATGTAGCTAGATCGATAC	pNF03	Inverse PCR: spc2(18–35)
L86	CATATAGTTTTATGCCTAAAAACC		PCR confirmation of spc1/2 mutants
I116	TATTTAGAGAACGTATGCCGAAGTATATAAATCATCAGTACAAAGGTAAGAATCAGTTGCTTTGCAGCTTCATCAT	pC221	Gibson assembly: pTgt1Lag
I91	ATAAAGATTTAAAATTTAGGAGGAAATTATA	pC221	Gibson assembly: all pTgt
I117	TTATTCCATTGTTAGAGTTTCTATTTAGAGAACGTATGCCGAAGTATATAAATCATCAGTACAAAGGTAAGAATCA	pT181	Gibson assembly: pTgt1Lag
I92	ATAAAGATTTAAAATTTAGGAGGAAATTATAATGTTATTGAGTGATGAAGAAGAATG	pT181	Gibson assembly: all pTgt
I118	TGATTCTTACCTTTGTACTGATGATTTATATACTTCGGCATACGTTCTCTAAATAGTTGCTTTGCAGCTTCATCAT	pC221	Gibson assembly: pTgt1Lead
I119	TTATTCCATTGTTAGAGTTTCTGATTCTTACCTTTGTACTGATGATTTATATACTTCGGCATACGTTCTCTAAATA	pT181	Gibson assembly: pTgt1Lead
I112	ATCGCCTTCGTAGTAATAATTGTCATTTGCATATGTTACATCGATACAGTGTAGTGTTGCTTTGCAGCTTCATCAT	pC221	Gibson assembly: pTgt2Lag
I113	TTATTCCATTGTTAGAGTTTCATCGCCTTCGTAGTAATAATTGTCATTTGCATATGTTACATCGATACAGTGTAGT	pT181	Gibson assembly: pTgt2Lag
I114	ACTACACTGTATCGATGTAACATATGCAAATGACAATTATTACTACGAAGGCGATGTTGCTTTGCAGCTTCATCAT	pC221	Gibson assembly: pTgt2Lead
I115	TTATTCCATTGTTAGAGTTTCACTACACTGTATCGATGTAACATATGCAAATGACAATTATTACTACGAAGGCGAT	pT181	Gibson assembly: pTgt2Lead
I94	CTTGTAGCCATGGCTATCCTTAT		PCR confirmation and sequencing of all pTtg, RT-PCR
I95	GGAGTGAATTTAATTTTATACACGC		PCR confirmation and sequencing of all pTgt, RT-PCR

Isolation of CNPX phage.

S. epidermidis RP62a is resistant to infection by phage CNPH82 (7) through a defense mechanism that does not depend on its CRISPR-Cas locus. CNPX was isolated by repeated infection of S. epidermidis LAM1680 (22). This phage can infect strain DB5, a derivative of RP62a that lacks a novel phage-defense system based on a serine/threonine kinase encoded by SERP2479 and the CRISPR array of repeats and spacers but that contains an intact set of cas genes (F. Depardieu, A. Bernheim, A. Sherlock, H. Molina, and D. Bikard, manuscript in preparation). The sequence of CNPX was shown to contain the same spc2 target as that of its CNPH82 parent phage.

Conjugation of pG0400 and pTgt plasmids into S. epidermidis DB5.

Conjugal transfer of pG0400 was carried out by filter mating as previously described (2). Transfer of the different pTgt plasmids was performed under similar conditions but selecting transconjugants with tetracycline. Corroborating the presence of the desired plasmid in transconjugants and recipients was achieved by extracting DNA of at least two colonies and using it as the template for the amplification of key elements of the plasmid/CRISPR locus and sequencing the resulting PCR product.

Phage propagation assay.

Phage propagation was determined using a plaque assay described by Budzik et al. (28). Briefly, strains to be tested for phage infection were cultured in BHI medium with necessary antibiotics at 37°C overnight. Cells (200 μl) were added to 4 ml of soft heart infusion agar (HIA) (20 g/liter) containing 5 mM CaCl₂ and poured over a prewarmed HIA plate containing the necessary antibiotics and 5 mM CaCl₂. After solidification, 10 μl of 10-fold serial dilutions of CNPX stock was spotted onto the top agar lawns and incubated at 37°C overnight. Plaques were counted and measured as the number of PFU per milliliter.

Pairwise competition assay.

This assay was performed as previously described (29). Briefly, overnight cultures of wild-type S. epidermidis RP62a and a competing transconjugant strain (harboring mupirocin-resistant plasmid pG0400) were mixed at a ratio of 1:1. The cultures were grown at 37°C with shaking for 24 h and subjected to vigorous vortex mixing, and 100-μl aliquots were transferred to fresh flasks containing 10 ml of fresh BHI medium. This serial transfer process was repeated daily three times. The total concentration of cells in these cultures and the densities of plasmid-bearing cells were estimated at each transfer by seeding serial dilutions on BHI agar plates with and without mupirocin, respectively. The relative frequency of plasmid-bearing cells was calculated as the ratio between the CFU levels enumerated in the presence and absence of mupirocin.

Transduction assays.

Overnight cultures of S. epidermidis LM1680 harboring the different pTgt plasmids were grown at 37°C. The following day, 20 μl of cells was added to 2 ml of fresh BHI medium supplemented with 5 mM CaCl₂ and tetracycline. Two samples were prepared; one was supplemented with 1 μl of CPNX stock, while the other was left untreated. Cultures were grown for 3 h at 37°C with shaking until complete lysis was observed. The lysates were inoculated with an additional 1 ml of exponentially growing S. epidermidis LM1680 cells with the desired plasmids (optical density at 600 nm [OD₆₀₀] of 1), and samples were incubated under the same conditions until complete lysis was achieved again. Upon lysis, cultures were spun down and filtered with a 0.45-μM-pore-size filter, and the resulting transducing stocks were stored at 4°C. Recipient strains were grown exponentially at 37°C in the presence of 5 mM CaCl₂ and proper antibiotics. At an OD₆₀₀ of 1.0, 1 ml of the recipient culture was mixed with 100 μl of lysate and incubated for 20 min at 37°C with shaking. The samples were spun down and washed twice with BHI medium containing 100 mM sodium citrate to prevent further phage infection, incubated for 2 h with the same medium, and seeded on BHI agar plates containing tetracycline to enumerate the number of transductants as the CFU counts per milliliter of phage lysate.

Efficiency of COI assay.

Assays of the centers of infection (COI) were conducted using exponentially growing cells (OD₆₀₀ = 0.8) as described previously (30). Bacterial cultures (150 μl) and CNPX phage lysates were mixed to reach a multiplicity of infection (MOI) of 5 phages per bacterial cell. Phage were allowed to adsorb for 5 min at 37°C. Cells and adsorbed phage were washed twice in 150 μl of medium (BHI) to remove free phage and then diluted 10-fold, 100-fold, and 1,000-fold to a final volume of 20 μl. A 2-μl volume was plated onto a lawn of the phage-sensitive host (the DB5/pΔCRISPR strain) prepared by adding 200 μl of overnight cultures to 4 ml of top HIA containing 5 mM CaCl₂. Enumeration of the plaques defined the number of COI formed on the test strain. Quantification of the efficiency at which COI formed (ECOI) was obtained by dividing the number of COI from the test strain by the number of COI from the sensitive host.

RT-PCR.

For detection of the target transcripts, total RNA was extracted as described previously (31), and 1 μg of total RNA was subjected to reverse transcription (RT) into cDNA by using Moloney murine leukemia virus (M-MLV) reverse transcriptase (Invitrogen) with primer I94 (Table 1). Amplification of cDNA was performed with primers I94 and I95.

RESULTS

The type III-A CRISPR-Cas system of S. epidermidis provides immunity against transfer by plasmid conjugation and phage infection.

S. epidermidis RP62a harbors a type III-A CRISPR-Cas system with three spacers, two of which have matches in GenBank (Fig. 1A). spc2 has homology to the cn20 gene of the S. epidermidis CNPH82 phage (7), but its ability to protect cells from phage infection has not been determined. To test this, we created strain DB5, a mutant lacking the CRISPR array of repeats and spacers from S. epidermidis RP62a, and isolated CNPX, a derivative of CNPH82 that is capable of infecting this strain. We used pCRISPR, a plasmid harboring the CRISPR array (31), to restore CRISPR-Cas immunity in strain DB5. Because spc2 contains a single mismatch with the cn20 target sequence of CNPH82 and CNPX, we modified pCRISPR to create a fully complementary spacer and eliminate the mismatch variable from subsequent experiments (Fig. 1B). We tested the immunity provided by spc2 by enumerating the PFU obtained by spotting serial dilutions of the CNPX phage stock on lawns of S. epidermidis DB5 harboring different complementing pCRISPR plasmids (Fig. 2A). While a pCRISPR plasmid with a deletion of the CRISPR cluster of repeats and spacers (pΔCRISPR) could not prevent the formation of plaques in DB5, pCRISPR prevented the propagation of CNPX in this strain. This result demonstrates that spc2 provides efficient antiphage immunity in S. epidermidis.

FIG 2.

Tolerance of crRNA:target mismatches during antiplasmid and antiphage type III-A CRISPR-Cas immunity. (A) Mismatch mutations in spc2 (mutations in red) within different versions of the pCRISPR complementing plasmid are shown. Each of these plasmids was introduced into S. epidermidis DB5 to generate different hosts for CNPX infection. Phage propagation in each host is measured as the number of PFU per milliliter obtained after plaquing serial dilutions of the phage stock into lawns of host bacteria. Three independent infection experiments were performed (means ± standard deviations [SD] are reported). The limit of detection of this assay is 100 PFU/ml. (B) Mismatch mutations in spc1 (mutations in red) within different versions of the pCRISPR complementing plasmid are shown. Each of these plasmids was introduced into S. epidermidis DB5 to generate different recipient strains for conjugation assay testing of the transfer of pG0400. Conjugation is measured as the number of CFU per milliliter of recipients and transconjugants obtained after three independent filter matings of donor and recipient strains (means ± SD are reported). The limit of detection of this assay is 100 CFU/ml.

spc1 matches a sequence present in the nickase gene (nes) of most staphylococcal conjugative plasmids and was shown to provide immunity against the transfer of the pG0400 target plasmid in wild-type S. epidermidis RP62a (2). We corroborated this result for strain DB5 by performing filter mating assays between the S. aureus RN4220/pG0400 donor strain and the S. epidermidis DB5 recipient strain containing different pCRISPR plasmids and enumerating the transconjugant CFU after selection with appropriate antibiotics (Fig. 2B). Similarly to the results obtained for spc2-mediated immunity against CNPX, recipients harboring pΔCRISPR but not pCRISPR enabled the transfer of pG0400, corroborating spc1-mediated antiplasmid immunity. Altogether, these results indicate that, as expected, both spc1 and spc2 are functional spacers that mediate robust CRISPR-Cas immunity in S. epidermidis.

Target-crRNA mismatches have different effects on antiplasmid and antiphage immunity.

Type I and II CRISPR-Cas immunity requires the presence of a “seed” region within the target sequence. Complementarity between the seed and the spacer sequence in the crRNA guide is essential for immunity (5, 10, 11, 13, 14). To investigate such a requirement in the type III-A CRISPR-Cas system of S. epidermidis, we introduced mutations in spc1 and spc2 to create mismatches within the target-crRNA interaction. We mutated different positions of the spacer sequence (positions 2 to 4, 1 to 8, 9 to 16, 12 to 19, 20 to 27, and 28 to 35) by replacing the wild-type nucleotides with their Watson-Crick complementary base (Fig. 2). With the exception of the substitution of nucleotides in positions 28 to 35, all mutations in spc1 resulted in a high number of transconjugant colonies, i.e., disrupted immunity against pG0400 (Fig. 2A). Surprisingly, a nearly opposite result was obtained when we introduced mutations in spc2. None of the mismatches disrupted immunity against CNPX infection, as measured by the inability of CNPX to form plaques in the hosts carrying different pCRISPR plasmids harboring mutations in spc2 (Fig. 2B). Additional mutations that eliminated target complementarity with half of the spacer sequence, either in the 5′ end (1–17) or in the 3′ end (18–35), were necessary to abrogate CRISPR defense (Fig. 2B).

We explored whether differences inherent in the assays used to measure plasmid conjugation and phage infection could be responsible for these disparate results. If a spacer mutation were to provide partial immunity, this would most likely result in the formation of a small colony or plaque in the conjugation or phage infection assays, respectively. Because small colonies can be easily detected but small plaques cannot, results based on counting colonies or plaques can seem diametrically opposed but in truth represent similar phenotypes. To solve this inconsistency, we performed more-sensitive assays to determine if spacer mutations confer partial immunity against either plasmid conjugation or phage infection. Partial immunity against pG0400 is expected to reduce the plasmid copy number within the population of transconjugants over time, resulting in the progressive loss of the mupirocin resistance provided by this plasmid. To test this, we carried out a pairwise competition experiment (29) to evaluate the fitness of different transconjugants in the presence of mupirocin relative to that of wild-type S. epidermidis RP62a (Fig. 3A). As expected due to the lack of CRISPR immunity, transconjugants harboring the pΔCRISPR plasmid showed no significant fitness disadvantage. A similar result was obtained with transconjugants carrying pCRISPR modified using mutations of positions 2 to 4 of spc1 [pCRISPR^spc1(2–4)], demonstrating that this mutation indeed abrogates CRISPR immunity. In contrast, transconjugants harboring pCRISPR^{spc1(12–19)} [as well as those with pCRISPR^{spc1(20–27)} (data not shown)] showed a significant fitness reduction, most likely due to the presence of partial CRISPR-Cas targeting of pG0400 in the population.

FIG 3.

Mismatches in positions 2 to 4 affect CRISPR immunity against conjugation and but not against phage infection. (A) Fitness of different transconjugants was measured by pairwise competition experiments. The change in the relative frequency of plasmid-bearing cells (y axis) is plotted against the number of transfers (one transfer per day; x axis). The top left panel shows the predicted changes in frequency for different selection coefficients, s. These are calculated from the equation dq/dt = −q(1 − q)s, where q is the relative frequency of the plasmid bearing cells and s is the selection coefficient (an s value of >0 indicates that the plasmid-bearing cells are at a disadvantage and an s value of <0 that the plasmid-bearing cells have an advantage). We are assuming 1/100 dilutions or t = 6.64 generations in each transfer. In all panels, the growth of transconjugants carrying pG0400 and pΔCRISPR, pCRISPR^spc1(2–4), or pCRISPR^{spc1(12–19)} was compared to growth of wild-type S. epidermidis RP62a. The black line and circles indicate the average change in relative frequency (the values for each of three independent experiments are shown as colored symbols). (B) Analysis of phage propagation measured as the efficiency of infection (ECOI). The ECOI value obtained for the propagation in sensitive cells (carrying pΔCRISPR) was set as 100% and was used to determine the value for the immune cells carrying versions of pCRISPR harboring different mutations in spc2. Experiments were performed in triplicate, and means ± SD are reported.

To evaluate immunity against phage infection more precisely, we performed assays to determine efficiency of the center of infection (ECOI). These experiments measure the number of infected cells that release at least one functional particle (28) and offer a more accurate measurement of the effect of CRISPR-Cas immunity on phage propagation than the plaque assay whose results are presented in Fig. 2B. The number of phages (measured as the number of PFU obtained after spotting an infected culture into indicator plates, i.e., containing DB5 cells completely sensitive to the phage) was set to 100% ECOI for infection of nonimmune S. epidermidis DB5/pΔCRISPR cells. In the presence of the pCRISPR plasmid containing a fully complementary spc2 sequence, the ECOI decreased to ∼1%, reflecting the restoration of immunity in DB5 cells by this plasmid (Fig. 3B). Mutations in positions 2 to 4, 12 to 19, 20 to 27, and 28 to 35 of spc2 sequence also resulted in levels of ECOI similar to those seen with the strain containing a fully complementary spc2 sequence, corroborating the results obtained with the plaque formation assay whose results are presented in Fig. 2B. However, mutations in positions 1 to 8 and 9 to 16 resulted in ∼100% ECOI, suggesting that these mutations significantly reduce CRISPR-Cas immunity against CNPX, an effect that was not detected with the less-sensitive plaque formation assay of Fig. 2B. Together, the results shown in Fig. 2 and 3 reveal that both the antiplasmid immunity and the antiphage immunity mediated by spc1 and spc2, respectively, are sensitive to the introduction of 8 consecutive mismatches in the first half of the target-crRNA at positions 1 to 8 and 9 to 16. However, we detected a striking difference for the effect of mismatches in position 2 to 4, which completely abrogated spc1-mediated immunity against pG0400 but did not affect spc2-mediated defense against CNPX infection.

The intrinsic spacer sequence determines the impact of mismatches on CRIPSR-Cas immunity.

The different immune responses generated by the mismatches at positions 2 to 4 could be the consequence of a number of variables differently affecting each target sequence, including (i) its mode of entry (ssDNA for pG0400 and dsDNA for CNPX), (ii) its intrinsic sequence, and (iii) its location within each genetic element, which could affect target transcription (found to be essential for S. epidermidis CRISPR-Cas targeting [8, 9]) and orientation with respect to the origin of replication (i.e., whether the target is in the leading or lagging strand of the plasmid or phage). To investigate these possibilities, we developed a mobile genetic element that could be manipulated to introduce each target in any strand and that could be transferred to S. epidermidis by both conjugation (mimicking pG0400 entry) and CNPX-mediated transduction (mimicking CNPX entry). We used the small staphylococcal plasmid pC221 (20), which can be conjugated from donor cells containing the pGO1 mobilization plasmid and which is small enough to be transduced with high efficiency by CNPX viral particles (Fig. 4A). Because both pC221 and the pCRISPR plasmids are resistant to chloramphenicol, we fused the genes encoding the origin of transfer protein (oriT) and the mobilization proteins (orfA and orfB) of pC221 with the genes encoding the origin of replication (ori), replication (repC), and recombination (pre) proteins and the gene encoding tetracycline resistance (tet) carried by pT181 (25), obtaining a small conjugative plasmid compatible with our pCRISPR plasmids containing spacer mutations (Fig. 4B). We then located the nes and cn20 target sequences (containing 10 bp of flanking sequences on each side) from pG0400 and CNPX, respectively, in the noncoding region between the pre gene and the ori^T site. Each target was placed with the sequence complementary to the crRNA in either the leading strand or the lagging strand with respect to the origin of replication (25), thus generating four pTgt plasmids: pTgt1^Lead, pTgt1^Lag, pTgt2^Lead, and pTgt2^Lag (Fig. 4C).

FIG 4.

Design of a plasmid that can be transferred to cells via conjugation or CNPX-mediated transduction. (A) Schematic of transduction and conjugation of pC221-derived plasmids into a host carrying a targeting CRISPR-Cas system. (B) Mobilizable, tetracycline-resistant pC221-pT181 hybrid plasmid used in this study. It was created by fusion of the mobilization genes (orfA and orfB) and origin of transfer (ori^T) of pC221 with the tetracycline resistance (Tet^r) gene, replication gene (repC), and recombination gene (pre) of pT181, as well as its origin of replication gene (ori). The site of target insertion is shown. (C) Schematic of the pTgt plasmid series showing the insertion of the nes or cn20 targets in the leading or lagging strand, according to replication from the ori site of pT181.

Using this system, we tested the immunity provided by spc1 and spc2 when their respective targets were in the lagging strand (Fig. 5). In agreement with the results obtained for pG0400, conjugation of pTgt1^Lag was prevented by the wild-type spc1 gene but not in the presence of mismatches in positions 2 to 4 (Fig. 5A). As expected, mutations in spc2 did not affect immunity against pTgt1^Lag. In contrast, conjugation of pTgt2^Lag was prevented both by the wild-type strain and by the spc2 mutant with mismatches in positions 2 to 4. Similarly, when both plasmids were transduced using CNPX, mismatches in positions 2 to 4 allowed transfer of the pTgt1^Lag plasmid but not of the pTgt2^Lag plasmid (Fig. 5B). Transduction of pTgt2^Lag was ∼1 to ∼2 orders of magnitude higher than pTgt1^Lag transduction, presumably due to the presence of a CNPX sequence (the cn20 target) in pTgt2^Lag. This has been observed in other experiments measuring phage-mediated plasmid transduction (32), and it is believed to be a consequence of high levels of recombination between the phage genome and the plasmid that facilitate the packaging of the latter. This note aside, these results indicate that the effect of the mismatches in positions 2 to 4 of the target-crRNA interaction does not depend on the route by which each target is transferred into CRISPR-Cas immune S. epidermidis cells. Rather, each intrinsic sequence has different degrees of tolerance of the presence of mismatches in these positions.

FIG 5.

Spacer-target mismatches eliminate immunity against the nes but not the cn20 target regardless of the mode of entry when the target is in the lagging strand. (A) Conjugation of pTgt1^Lag and pTgt2^Lag into recipient strains carrying mismatches in positions 2 to 4 within spc1 or spc2. CFU counts per milliliter of recipients and transconjugants were determined in three independent experiments, and means ± SD are reported. The limit of detection of this assay is 100 CFU/ml. (B) CNPX-mediated transduction of pTgt1^Lag or pTgt2^Lag into cells carrying mismatches in positions 2 to 4 within spc1 or spc2. CFU counts per milliliter of transductants were determined in three independent experiments, and means ± SD are reported. The limit of detection of this assay is 100 CFU/ml.

The location of the target sequence affects mismatch tolerance.

We then tested the immunity provided by spc1 and spc2 when their respective targets were in the leading strand (Fig. 6). Both the conjugative transfer and CNPX-mediated transduction of the pTgt2^Lead plasmid tolerated the mismatches at positions 2 to 4. Unexpectedly, immunity against pTgt1^Lead, both when transferred by conjugation and when transferred by transduction, was not affected by the mismatches in positions 2 to 4. Although these mismatches prevent the transfer of this plasmid when the target is in the lagging strand (Fig. 5), spc1-mediated CRISPR-Cas immunity against the plasmid worked in spite of the mismatches when the target was in the leading strand (Fig. 6).

FIG 6.

Spacer-target mismatches eliminate immunity against both the nes target and the cn20 target regardless of the mode of entry when the target is in the leading strand. (A) Conjugation of pTgt1^Lead and pTgt2^Lead into recipient strains carrying mismatches in positions 2 to 4 within spc1 or spc2. CFU counts per milliliter of recipients and transconjugants were determined in three independent experiments, and means ± SD are reported. The limit of detection of this assay is 100 CFU/ml. (B) CNPX-mediated transduction of pTgt1^Lead or pTgt2^Lead into cells carrying mismatches in positions 2 to 4 within spc1 or spc2. CFU counts per milliliter of transductants were determined in three independent experiments, and means ± SD are reported. The limit of detection of this assay is 100 CFU/ml. (C) RT-PCR was performed on RNA samples extracted from S. aureus RN4220 cells harboring pTgt1^Lead or pTgt1^Lag plasmids using primers that detect transcription of either the leading strand or lagging strand. PCRs of samples where reverse transcriptase was omitted are shown as controls.

Recently, we found that type III-A CRISPR-Cas immunity requires transcription of the target sequence (8, 9). All four plasmids generated in this study are targeted by either spc1 or spc2 crRNAs; therefore, both the nes and cn20 targets must be transcribed. This is in agreement with our observation that small staphylococcal plasmids display transcription of most of their genome in both directions (8). However, the opposite effects of mismatches in spc1 positions 2 to 4 in immunity against pTgt1^Lead and pTgt1^Lag open the possibility that differences in transcription of the leading and lagging strands across the nes target affect mismatch tolerance. Admittedly, this is unlikely since such a transcriptional effect should be in principle also be detected for the cn20 target. To corroborate this, we performed RT-PCR to detect bidirectional transcription across the target in the pTgt1^Lead and pTgt1^Lag plasmids (Fig. 6C). We found that the two targets were similarly transcribed. Altogether, these results indicate that, in addition to the intrinsic target sequence, the location of this sequence with respect to the origin of replication is another variable that affects mismatch tolerance during CRISPR-Cas immunity.

DISCUSSION

We explored the effects of mismatches in the interaction between two different targets and their respective crRNAs of the type III-A CRISPR-Cas system of S. epidermidis. In this system, spc1 targets the nes gene of the conjugative plasmid pG0400, whereas spc2 matches the cn20 gene of the CNPX bacteriophage. We found that for both targets, 8-nt-long mismatches were largely tolerated in the 3′ half but not the 5′ half of the spacer sequence. This is consistent with the observation that the crRNAs produced by S. epidermidis (and other type III systems) undergo a process of maturation by which the 3′ end sequences are trimmed in increasing intervals of 6 nt (22). The disruption caused by the presence of mismatches in the first 8 nt of the crRNA:target sequence could be due to the presence of a “seed” region located in the 5′ end of the target, similarly to the case of type I and II CRISPR-Cas targets (5, 10, 11, 13, 14); more experiments will be needed to test this hypothesis. The elaborate targeting mechanism of type III-A CRISPR-Cas immunity, which cleaves both DNA and RNA (transcript) targets (9), adds another layer of complexity to the analysis of the effects of the mismatches in a putative seed region, since they could affect immunity by preventing DNA and/or RNA cleavage. Additional work testing the effect of single- or double-nucleotide mismatches, as well as testing their effects on DNA and RNA cleavage, will be required to understand seed sequence requirements for type III targets.

Interestingly, we found that mismatches in positions 2 to 4 of these targets abrogated immunity against pG0400, but not against CNPX, in an orientation-dependent manner. To eliminate the possibility that these opposite results are a consequence of differences in the replication and/or mode of entry of these foreign genetic elements, we placed each target on both the leading and lagging strands of a small plasmid that could be transferred into S. epidermidis via both conjugation and CNPX-mediated transduction. We found that the effects of mismatches in positions 2 to 4 do not change when the target sequence complementary to the crRNA is on the leading strand regardless of the mode of entry of the target sequence. This result indicates that the ability of a spacer sequence to retain its immunity properties after the introduction of mismatches is influenced by its intrinsic sequence. One explanation for this observation is that the sequence context of the mismatches could be responsible for how well they are tolerated. For example, the strength of the base pair interactions flanking the mismatched sequences could determine how well the mismatches are tolerated, with stronger interactions retaining the immunity properties of the spacer. A similar situation seems to occur during target recognition by the crRNA guides of the type II Cas9 nuclease during genome editing experiments. Many studies that looked at the off-target effects generated by the imperfect pairing of a crRNA guide to its target (33, 34) have shown that the position of mismatches cannot always predict the off-target sites for different crRNAs. One possibility is that the GC content of the spacer sequence is important to determine the effects of the mismatches. spc1 and spc2 have 28% and 23% GC content, respectively, a difference that does not seem significant enough to account for the differences observed between these spacers in response to the introduction of the 2-to-4 mismatch. It is also possible that the specific mutations introduced in the spacer sequence produce different types of mismatches, with different effects on the target-crRNA interaction.

In contrast, when the complementary sequence is located in the lagging strand, the negative effects of the mismatches are eliminated for the nes sequence; i.e., it is effectively targeted in the presence of the 2-to-4 mismatches. We speculate that the differences in targeting of the leading strand versus the lagging strand may be a result of the exposure of the target sequence during replication. The pTgt plasmids are derivatives of the pT181 staphylococcal vector and replicate through asymmetric rolling circle replication (35). In this mode of replication, the lagging strand is released as ssDNA before being copied to dsDNA (36). In contrast, the leading strand maintains a dsDNA structure. The interactions of the Cas10-Csm complex with a target presented as ssDNA (at least temporarily) or dsDNA could be different and could thus result in different tolerances of mismatches. This could also apply to the targets in pG0400 and CNPX, as both undergo rolling circle replication. While this mode of replication has been demonstrated for pG0400 (27), CNPX shares the genetic organization of the lambda phage, with extensive homology in the region encoding genes involved in DNA replication (7). Therefore, we believe that CNPX most likely has a replication cycle similar to the lambda replication cycle, which undergoes about six rounds of theta replication before switching to rolling circle replication (37).

How different spacer sequences tolerate mismatches is an important aspect of CRISPR-Cas immunity, since the main route for viral escape is contingent of the presence of target mutations that prevent the recognition and/or cleavage of the invading DNA. This has been well documented for the escape from type I (13) and type II (5) CRISPR-Cas immunity. Not much is known of how phages escape type III CRISPR-Cas targeting or of how mismatches affect it. One study performed in Sulfolobus solfataricus, an archaeon that carries a type III CRISPR-Cas system, reported that up to 15 mismatches between the pNOB-ORF406 target and the 3′ half of the A53 spacer were tolerated by the system, still triggering 50% of DNA degradation. In contrast, mutations at the 5′ end of the A53 spacer are essential in target recognition (38). Our results are in agreement with that previous report but show in addition that the intrinsic sequence and location of the target influence the effect of the mismatches. Hence, they suggest that, in a situation where two target sequences present in the phage have similar probabilities of accepting mutations, the rates of escape could be dramatically different. Whereas one spacer could tolerate the mismatches generated by the mutations and provide efficient immunity against both wild-type and mutant viruses, the immunity provided by the other spacer could be abrogated by the mismatches, thus allowing the mutant viruses to escape. Spacer sequences are acquired upon infection from regions of the invading genome that are flanked by appropriate sequences (the PAM for type I and II immunity and the absence of an antitag sequence for type III). These constraints are not exceedingly stringent, and there are usually thousands of sequences that can be acquired as spacers from a given phage after one round of infection (39, 40). It is believed that the benefit of such a mechanism is the creation of a host population with highly heterogeneous spacer content that would be resistant to all the possible escape mutations that are present in the phage population (41). Our results suggest that not only the presence of such escape mutations but also the intrinsic sequence of the acquired spacers, some of which, depending on their sequence composition and/or the strand that they target, could still provide immunity against many of the multiple genotypes present in the infecting population, is important in the CRISPR-mediated arms race between prokaryotes and their viruses.