The three major types of CRISPR-Cas systems function independently in CRISPR RNA biogenesis in Streptococcus thermophilus

SUMMARY

CRISPR-Cas systems are small RNA-based immune systems that protect prokaryotes from invaders such as viruses and plasmids. We have investigated the features and biogenesis of the CRISPR (cr)RNAs in Streptococcus thermophilus (Sth) strain DGCC7710, which possesses four different CRISPR-Cas systems including representatives from the three major types of CRISPR-Cas systems. Our results indicate that the crRNAs from each CRISPR locus are specifically processed into divergent crRNA species by Cas proteins (and non-coding RNAs) associated with the respective locus. We find that the Csm Type III-A and Cse Type I-E crRNAs are specifically processed by Cas6 and Cse3 (Cas6e), respectively, and retain an 8-nucleotide CRISPR repeat sequence tag 5′ of the invader-targeting sequence. The Cse Type I-E crRNAs also retain a 21-nucleotide 3′ repeat tag. The crRNAs from the two Csn Type II-A systems in Sth consist of a 5′-truncated targeting sequence and a 3′ tag; however these are distinct in size between the two. Moreover, the Csn1 (Cas9) protein associated with one Csn locus functions specifically in the production of crRNAs from that locus. Our findings indicate that multiple CRISPR-Cas systems can function independently in crRNA biogenesis within a given organism – an important consideration in engineering co-existing CRISPR-Cas pathways.

INTRODUCTION

The CRISPR-Cas systems are adaptive RNA-directed immune systems present in many bacteria and most archaea. The CRISPR loci contain short sequences acquired from various invaders, separated by copies of a characteristic short direct repeat sequence (24–37 bp in length) (Grissa et al., 2007). CRISPR locus transcripts are processed to generate multiple CRISPR (cr)RNAs that each contain an individual invader-derived sequence and common crRNA tag sequences derived from the repeat element. The crRNAs guide effector complexes formed with Cas (CRISPR-associated) proteins to cleave complementary foreign DNA or RNA (for reviews see (Wiedenheft et al., 2012, Terns & Terns, 2011, Barrangou & Horvath, 2012, Jore et al., 2012, Sorek et al., 2013)).

There are multiple, diverse CRISPR-Cas systems defined by distinct sets of Cas proteins and crRNA species. Each of the 11 identified systems includes the universal Cas1 and Cas2 proteins along with a module of subtype-specific Cas proteins named for an organism where the system is found (e.g. the Cas subtype E. coli or Cse proteins) (Haft et al., 2005, Makarova et al., 2006, Makarova et al., 2011). Distant relationships have been identified between some components of different CRISPR-Cas systems, allowing the definition of broad superfamilies of Cas proteins and the classification of the systems into three Types: I - III (Makarova et al., 2011). The Cse system, for example, is the E subtype of the Type I systems, or Type I-E system. The 11 distinct CRISPR-Cas systems generally use different components and mechanisms for the essential functions of the immune systems.

CRISPR-Cas defense requires three essential steps. Adaptation entails acquisition and integration of a short segment of invader sequence into the microbe’s CRISPR array. Evidence indicates that the common Cas1 and Cas2 proteins function in adaptation (Yosef et al., 2012, Datsenko et al., 2012). crRNA biogenesis produces the panel of crRNAs that target the individual collected invader sequences. The CRISPR locus transcript is processed within the repeat elements to release crRNAs that retain part of the common repeat sequence as well as an invader-specific guide sequence (Richter et al., 2012, Hale et al., 2008, Hale et al., 2009, Deltcheva et al., 2011, Hatoum-Aslan et al., 2011, Nam et al., 2012, Brouns et al., 2008, Carte et al., 2008). Interference occurs via mature crRNAs guiding effector Cas proteins to destroy invader DNA or RNA (Marraffini & Sontheimer, 2008, Hale et al., 2009, Garneau et al., 2010, Magadan et al., 2012, Jinek et al., 2012, Gasiunas et al., 2012, Sinkunas et al., 2013, Zhang et al., 2012).

crRNA biogenesis occurs by diverse mechanisms in the distinct CRISPR-Cas systems. crRNAs are typically released from the large CRISPR transcript by endonucleolytic cleavage within the conserved repeat elements, generating RNAs with repeat sequence flanking the guide sequence at both ends. In many cases, these 1X unit crRNAs are further processed as noted below. The initial repeat cleavage is catalyzed by “Cas6 superfamily” endoribonucleases in multiple Type I and III CRISPR-Cas systems: Cas6 in Csa Type I-A, Cst Type I-B, Csc Type I-D, Csm Type III-A, and Cmr Type III-B systems, Cse3 (Cas6e) in the Cse Type I-E system and Csy4 (Cas6f) in the Csy Type I-F system (Brouns et al., 2008, Carte et al., 2008, Richter et al., 2012, Haurwitz et al., 2010, Nam et al., 2012, Scholz et al., 2013, Lintner et al., 2011). The 8-nucleotide repeat sequence generated at the 5′ end of the crRNAs by the Cas6 superfamily endoribonucleases is retained in mature crRNA species (Hale et al., 2008, Brouns et al., 2008, Hatoum-Aslan et al., 2011, Lintner et al., 2011, Richter et al., 2012, Scholz et al., 2013) and has been shown to be critical for function (Hale et al., 2012, Zhang et al., 2012). Two of the Cas6 superfamily enzymes, Cse3 (Cas6e) and Csy4 (Cas6f), recognize and cleave downstream of a stem-loop structure in associated CRISPR repeats, and that repeat stem-loop is retained at the 3′ end of the Cse Type I-E and Csy Type I-F crRNAs (Brouns et al., 2008, Haurwitz et al., 2010, Richter et al., 2012, Nam et al., 2012, Sashital et al., 2011). The 3′ ends of crRNAs generated by Cas6 (Csa Type I-A, Cst Type I-B, Csm Type III-A and Cmr Type III-B systems) are processed to remove some or all of the 3′ repeat sequence, and sometimes a small portion of the guide sequence (Hale et al., 2008, Richter et al., 2012, Hale et al., 2012, Hatoum-Aslan et al., 2011, Zhang et al., 2012) likely via trimming by non-Cas (exo)ribonucleases (Hatoum-Aslan et al., 2013). In one of the Type I systems, the Csd Type I-C system, CRISPR transcripts are processed by a Cas5 superfamily protein, Cas5d, which generates an 11-nt 5′ repeat tag (Garside et al., 2012, Nam et al., 2012). (Other Cas5 superfamily proteins do not function in crRNA biogenesis but instead are required components of CRISPR-Cas defense complexes (Brouns et al., 2008, Hale et al., 2009, Makarova et al., 2011)).

In one of the Type II systems, the Csn Type II-A system, base-pairing of a small non-coding RNA containing complementarity to the CRISPR repeat, termed the trans-activating crRNA or tracrRNA, triggers cleavage of CRISPR locus transcripts by the non-Cas protein RNase III (Deltcheva et al., 2011). The tracrRNA is encoded within the Csn Type II-A CRISPR-Cas module. The 5′ ends of Csn Type II-A crRNAs undergo further processing that removes the 5′ repeat fragment and ~10 nucleotides of the guide sequence (Deltcheva et al., 2011). The Csn Type II-A crRNAs retain a 19–22-nucleotide 3′ repeat tag (Deltcheva et al., 2011).

Streptococcus thermophilus DGCC7710 (Sth) has four CRISPR-Cas modules (CC1-CC4), each containing a cluster of cas genes and a CRISPR array ((Horvath & Barrangou, 2010) and Figure 1). Two of the CRISPR-Cas modules (CC1 and CC3) are Csn Type II-A systems, and both of these are known to be active in both adaptation and interference in Sth as they develop phage resistance (Barrangou et al., 2007, Garneau et al., 2010, Magadan et al., 2012, Deveau et al., 2010, Levin et al., 2013). The effector complex of the Csn Type II-A system, comprised of Csn1 (Cas9) and associated RNAs, was recently successfully adapted as a promising tool for genome editing and gene expression control in human cells as well as multiple model organisms (reviewed in (Pennisi, 2013, Mali et al., 2013a, Terns & Terns, 2014)). Sth also contains a Csm Type III-B and a Cse Type I-E CRISPR-Cas system (CC2 and CC4, respectively). Here, we have investigated the composition and biogenesis of crRNAs from the four coexisting CRISPR-Cas systems in Streptococcus thermophilus.

Figure 1.

Overview of the CRISPR-Cas loci found in S. thermophilus DGCC7710 (Sth). CRISPR arrays are composed of short direct repeats (black rectangles) interspaced with unique invader-derived spacer sequences (colored rectangles). The number of spacers is indicated for each locus. Transcription of CRISPR arrays is initiated within upstream leader sequences (“L”), except perhaps for CC4 (see text). Each CRISPR array contains an adjacent subset of cas genes (shown as colored boxes). Genes involved, or predicted to be involved in crRNA biogenesis are indicated with an asterisk. Note that cas2-4 is a cas2-dnaQ fusion (Horvath & Barrangou, 2010). Diagram is not drawn to scale.

RESULTS

crRNAs associated with 4 CRISPR-Cas systems in Sth

To characterize the population of crRNA molecules generated by the CRISPR-Cas systems in Sth, we deep sequenced total Sth RNA and analyzed the RNAs derived from each of the CRISPR loci. Prior to construction of cDNA libraries, the RNAs were phosphatase and kinase treated to allow sequencing of RNAs containing diverse 5′ and 3′ chemical end groups (see Materials and Methods). We found that all four CRISPR arrays in the Sth genome are expressed and give rise to multiple small crRNAs that each contain an invader targeting sequence (Figure 2A). Potential transcription promoters recognized by the Sth sigma-70 family housekeeping polymerase, SigA, were identifiable within the leader regions of CRISPRs 1, 2 and 3 in Sth (at approximately -35 and -10 relative to the 5′ ends observed in sequenced RNAs from the region) (Figure S1) and confirmed as likely transcription start sites through RNA sequencing analysis of transcripts at the leader regions of these CRISPR loci (Figure S2). However for CRISPR 4, the translational stop codon of the cas2 gene is located just 20 nts upstream of the first repeat element of CRISPR locus 4 (Figure S1), and the locus may be transcribed as part of the upstream cas gene operon (Figures S2 and S3). The greatest numbers of crRNAs were from the loci with the strongest homology to the σ70 promoter consensus sequence – Csn Type II-A CRISPRs 1 and 3 (Figures 2A and S1).

Figure 2.

RNA sequencing profiles of total crRNAs from Sth. (A) Deep sequencing reads from Sth total RNA map to each of the four CRISPR loci. The number of unique reads for a given nucleotide position are indicated on the Y axis in thousands. CRISPR repeats are shown as black boxes below each graph. For CC3, spacers 4 and 11, as well spacers 5 and 12 are identical, and thus do not map uniquely to the genome. In both cases the total numbers of reads have been divided equally between each position (shown as dotted boxes). (B) Mapping crRNA 5′ and 3′ termini. The graphs show the percentages of sequenced RNAs from Sth total RNA that map to the indicated position relative to the repeat-guide junction. Below each graph is an illustration of the mature crRNAs from each CRISPR locus.

Additionally, cas gene expression flanking each of the four CRISPR loci was observed (Fig. S3), consistent with proteomic studies performed using this same Sth strain where certain Cas proteins were detected for each of the loci (Young et al., 2012). Collectively, the results show that both crRNAs and Cas proteins are constitutively expressed from each of the four CRISPR-Cas loci in this organism.

The sequencing revealed that the crRNAs produced from each of the four CRISPR loci in Sth are distinct; the RNAs from each locus are comprised of a characteristic and specific combination of invader-targeting sequence and repeat sequence tag elements. The distinctive patterns can be seen by mapping the ends of the sequenced RNAs from each locus relative to the repeat and guide element boundaries (Figure 2B).

The RNAs from the two Csn Type II-A systems in Sth (CC1 and CC3) are similar in overall structure, but nonetheless distinct. The crRNAs from both of the Csn-associated loci contain substantial 3′ tags comprised of the conserved repeat sequence (Figure 2B). However, the 3′ repeat tags of the RNAs from the two loci are distinct in length as well as sequence. crRNAs from CRISPR 1 primarily have a 17-nucleotide 3′ repeat (guuuuuguacucucaag), while CRISPR 3 crRNAs include a 22-nucleotide 3′ repeat that differs significantly in sequence beyond the first 5 nts (guuuuagagcuguguuguuucg). In addition, the Csn crRNAs from both loci only include ~2/3 of the guide sequence encoded in the genome. The crRNAs from CRISPR loci 1 and 3 lack the first 9 and 10 nucleotides of the guide sequence (which is 30 nts in total average length for both loci), respectively. Thus, the mature crRNAs derived from CC1 generally have a size of 38 nts (21 nts of guide and 17 nts of repeat sequence) while the crRNAs from CC3 are 42 nts in length (20 nts of guide and 22 nts of repeat sequence). Indeed, the most abundant RNAs detected from these CRISPR loci in Northern analysis (Figure S4) correspond in size with the sequenced RNA species of 38 and 42 nucleotides, respectively (Figure 2B). The overall features of the Csn crRNAs in Sth are similar to those described for Streptococcus pyogenes crRNAs, which consist of a 20-nucleotide guide sequence and 19–22 nucleotide 3′ repeat tag (Deltcheva et al., 2011). Our results indicate that the intriguing absence of ~1/3 of the encoded guide sequence derived from the spacer is a common feature of Csn Type II-A system crRNAs.

The Csm Type III-A system crRNAs in Sth (CC2) are very different from the Csn Type II-A (CC1 and CC3) crRNAs, but are similar to those described for other CRISPR-Cas systems. The Csm Type III-A crRNAs include an 8-nucleotide 5′ repeat tag upstream of the guide sequence (Figure 2B), as do crRNAs associated with Cmr, the other Type III system, and with most characterized Type I systems including Csa (I-A), Cst (I-B), Csc (I-D), Cse (I-E), and Csy (I-F) (Hale et al., 2009, Hale et al., 2008, Lintner et al., 2011, Brouns et al., 2008, Nam et al., 2012, Haurwitz et al., 2010, Richter et al., 2012, Zhang et al., 2012, Scholz et al., 2013). In addition, similar to crRNAs associated with the Cmr Type III-B effector complex (Hale et al., 2009), Csm Type III-A crRNAs in Sth are uniform in length independent of the variable length of the guide region encoded in the CRISPR locus. The guide sequences of the three Csm Type III-A crRNAs encoded in CC2 are 36, 40 and 39 nt in length, however, the guide sequences are trimmed to 35 nts in the crRNAs (Figure S5). crRNAs of 43 nucleotides total length (8-nucleotide repeat tag and 35 nt guide) are the most abundant species of Csm Type III-A crRNAs detected by RNA sequencing and Northern analysis (Figures 2B and S4). The levels of the crRNAs decline significantly with distance along the locus (Figure S5), with the highest numbers closest to the 5′ region. In fact, crRNA 2.03 was not even detected in Northern analysis (Figure S4). Similar processing has been reported for Csm Type III-A crRNAs expressed in Staphylococcus epidermidis (Hatoum-Aslan et al., 2011). In that study of S. epidermidis strains overexpressing crRNAs from a plasmid construct, an additional, much less abundant, 37 nt species was also noted (Hatoum-Aslan et al., 2011, Hatoum-Aslan et al., 2013), and a similar minor species can be observed in sequence reads and Northern analysis of Sth crRNA 2.01 (Figures S4 and S5). Each crRNA associated with the related Cmr system is found in two size forms, 39 and 45 nt, that are often nearly equally abundant (Hale et al., 2009, Hale et al., 2012). Thus, the crRNAs associated with the two Type III CRISPR-Cas systems, Csm (I-A) and Cmr (I-B), share common features including an 8 nt 5′ repeat tag, constant RNA lengths independent of the encoded guide sequence length, and two size forms, suggesting similarities in the biogenesis pathways.

The crRNAs that we identified from the locus associated with the Cse Type I-E system in Sth closely resemble those characterized in E. coli (Brouns et al., 2008, Jore et al., 2011). The RNAs possess an 8-nt repeat sequence tag upstream of the guide sequence and the remaining repeat sequence (21 nt) at the 3′ end (Figure 2B).

Together, our RNA sequencing and Northern findings indicate that the crRNAs that arise from each of the four CRISPR-Cas systems present in Sth have a characteristic composition that is different from that of the crRNAs from the other loci, reflecting the extensive diversity of CRISPR-Cas systems within S. thermophilus.

5′ and 3′ termini of crRNAs

We also investigated the end groups of the crRNAs from the four Sth CRISPRs. To determine whether crRNAs possess a 5′ phosphate end, total Sth RNA was treated with a 5′ phosphate-dependent exonuclease (TEX) and the sensitivity of the crRNA to the exonuclease was assessed by Northern analysis (Figure 3A). Similarly, the presence of a 3′ hydroxyl group was probed by treatment with poly(A) polymerase, which catalyzes the transfer of AMP to the 3′ termini of RNAs that contain a 3′ hydroxyl (Sippel, 1973), and Northern analysis (Figure 3B). Mature crRNA species from Csn Type II-A CRISPR loci 1 and 3 were lost upon treatment with 5′ phosphate-dependent exonuclease (Figure 3A), indicating that these crRNAs possess a phosphate group at the 5′ end. The TEX-resistant crRNAs from CRISPR loci 2 and 4 likely have 5′ hydroxyl end groups. Treatment with poly(A) polymerase results in a reduction in the gel mobility of mature crRNAs from both Csn Type II-A locus 1 and 3 as well as Csm Type III-A locus 2 (Figure 3B), indicating the presence of a 3′ hydroxyl end group. RNAs that are resistant to extension by poly(A) polymerase may have 3′ phosphate or cyclic phosphate end groups. We found that pre-treatment with polynucleotide kinase (PNK), which removes 2′,3′ cyclic phosphate groups (Becker & Hurwitz, 1967), resulted in accessibility of the majority of the Cse Type I-E locus 4 crRNAs to poly(A) polymerase (Figure 3B). Pre-treatment with calf intestinal phosphatase (CIP), which removes 3′ phosphates (but not 2′,3′ cyclic phosphates), allowed polyadenylation of a smaller fraction of the crRNAs indicating that a small fraction of the apparently full-length crRNAs harbor 3′ phosphate termini (Figure 3B). The results suggest that crRNAs from Cse Type I-E locus 4 primarily possess cyclic phosphate groups at the 3′ end. Similar results were described for Cse Type I-E system-associated crRNAs from E. coli (Jore et al., 2011). The distinct end groups found on the crRNAs from each type of CRISPR-Cas system in Sth are indicated in Figure 3.

Figure 3.

Sth crRNA 5′ and 3′ chemical end groups. (A) 5′ end analysis was performed by incubating total RNA from Sth in the absence or presence of Terminator 5′-Phosphate-Dependent Exonuclease (TEX) followed by Northern blotting using probes for the RNAs indicated below each panel. 1.01 is the leader proximal crRNA from CRISPR 1, 2.01 is the leader proximal crRNA from CRISPR 2, etc. (B) For 3′ chemical end analysis of crRNAs from CRISPRs 1–3, gel extracted sRNAs from Sth were incubated in the absence or presence of E. coli poly(A) polymerase (PAP) followed by Northern blotting. The chemical end groups present on crRNAs from CRISPR4 were determined by Northern analysis of Sth total RNA following combinations of treatments with Thermosensitive Alkaline Phosphatase (TSAP), T4 polynucleotide kinase (PNK), and E. coli poly(A) polymerase (PAP). Mature crRNAs are indicated by an asterisk. A very minor ~37 nt form detected for 2.01 crRNA discussed in the text is indicated by an arrow.

Mechanisms of crRNA biogenesis in Sth

Candidate biogenesis factors have been identified for each of the three types of CRISPR-Cas systems present in Sth. In the Csn Type II-A system characterized in S. pyogenes, crRNA biogenesis was found to involve a short RNA species termed a tracrRNA that contains a region of complementarity to the CRISPR repeat sequence, and the non-Cas protein RNase III (Deltcheva et al., 2011). We observed expression of tracrRNAs from the two Sth Csn Type II-A systems in our deep RNA sequencing (Figure 4B and 4C, blue reads) and in Northern analysis (Figure 5E and 5F). The complementarity between the tracrRNA and the CRISPR repeat sequence can stimulate cleavage of the base-paired CRISPR transcript and tracrRNA by RNase III (Deltcheva et al., 2011). As illustrated in Figure 4A, tracrRNA-stimulated cleavages within the repeats of a CRISPR transcript generate 1X crRNA units with repeat sequence flanking the guide sequence at both ends (as well as cleaved tracrRNAs) (Deltcheva et al., 2011).

Figure 4.

crRNA biogenesis from Sth CC1 and CC3. (A) Model depicting tracrRNA-mediated crRNA biogenesis. A duplex composed of a tracrRNA and the CRISPR primary transcript is cleaved by RNase III (green double arrow lines) to generate a 1X intermediate composed of a single invader-targeting sequence with fragments of the CRISPR repeat at either end. The 1X intermediate is then cleaved (by an unknown mechanism) to produce mature crRNAs containing ~20 nucleotides of guide sequence and ~20 nucleotides of CRISPR repeat sequence at the 3′ end. (B) and (C) Alignment of observed crRNA (red) and tracrRNA (blue) profiles with the respective sequences in the region of duplex formation. Deep sequencing profiles of crRNAs 1.01 and 3.01 were aligned with the profiles of their respective tracrRNAs and were used to predict the positions of cleavage by RNase III (green double arrow lines). RNase III cleavages typically occur 2 nt apart on each strand of an RNA duplex. In the case of CRISPR 1 (B), the 3′ end of the most abundant crRNA species and 5′ end of the most abundant tracrRNA species are not found in locations consistent with RNase III cleavage, so cleavage is proposed to occur at a site marked by the ends of less abundant species (green double arrow lines), and end trimming is proposed (black horizontal arrows). The locations of the ends of the most abundant RNAs associated with CRISPR 3 (C) are consistent with RNase III cleavage, but in addition, there is evidence of similar trimming of a fraction of the crRNA and tracrRNA ends (3′ and 5′ respectively; dashed horizontal arrows).

Figure 5.

Csn1 (Cas9) from CC1 is required for processing of crRNAs and tracrRNAs from CRISPR 1, but not other CRISPRs. Northern analysis was performed using total RNA extracted from wild-type Sth and strains in which csn1-1 (cas9) or csn2-1 (from CC1) were inactivated. Blots were probed for the RNA indicated below each panel. Asterisks indicate the location of mature crRNAs. For crRNA 1.01, the positions of the 1X intermediate and the 3′-trimmed 1X intermediate are indicated by an arrow and a double arrow, respectively. Diagrams illustrating predicted tracrRNA primary transcripts and processing intermediates, as well as the positions of the probes used for Northern analysis are shown to the right of each blot.

The 5′ region of the 1X crRNAs - the repeat sequence and a portion of the guide sequence - is not found in the mature crRNAs from Sth CRISPRs 1 and 3 (Figures 2B, and 4B and 4C) or in S. pyogenes (Deltcheva et al., 2011), indicating that the 1X RNAs are processed at the 5′ end. In addition, our data suggest that in some cases these same crRNAs can undergo further processing at the 3′ end (and that the tracrRNA is also further processed at the 5′ end in parallel). On one hand, the most commonly observed ends of the tracrRNA and the crRNAs from Sth CRISPR 3 are located at a potential RNase III cleavage site, 2 nucleotides apart within the region of tracrRNA-crRNA base-pairing (Figure 4C, green double arrow line). However, the observed ends of the most abundant tracrRNA and crRNA species from CRISPR 1 are located at the edge of the region of contiguous base-pairing and are separated by 4 nt (Figure 4B, solid black lines), making it unlikely that these ends are generated by RNase III cleavage. We propose that RNase III cleavage of RNAs from Sth CRISPR 1 occurs at another site within the region of complementarity marked by less abundant crRNA and tracrRNA species (cleavage occurring with the expected 2 nucleotide offset (Figure 4B, green double arrow line)) and that the mature crRNAs are generated by 3′ end processing (7 nt; Figure 4B, solid black arrow line). The tracrRNA product of RNase III cleavage appears to undergo parallel 5′ end processing (9 nucleotides; Figure 4B, solid black arrow line). The initial product of cleavage of the tracrRNA by RNase III is detectable by Northern analysis as well as sequencing (Figure 5E, middle band). Interestingly, there is also evidence of trimming of a fraction of the RNase III cleavage products from locus CRISPR 3. CRISPR 3 crRNA species trimmed (~6–8 nucleotides) at the 3′ end and tracrRNA species trimmed (~8 nucleotides) at the 5′ end are also detectable by Northern and deep sequencing analysis (Figure 4C, dashed line and Figure 5F, bottom band).

In S. pyogenes, crRNA production also depends on Csn1 (Cas9) (Deltcheva et al., 2011). The two Csn Type II-A systems in Sth each includes a csn1 (cas9) and csn2 gene (Figure 1). We examined the impact of loss of both Csn1 (Cas9) and Csn2 from CC1 on crRNA biogenesis from all four CRISPR loci in Sth by Northern analysis of the two mutant strains. The profiles and levels of crRNAs from CRISPRs 2–4 were not altered in strains in which the csn1 (cas9) or csn2 gene associated with CRISPR locus 1 was deleted (Figure 5, B–D). Furthermore, production of crRNAs from CC1 was not altered by deletion of the CC1 csn2 gene (Figure 5A). However, deletion of the CC1 csn1 (cas9) gene did specifically affect biogenesis of crRNAs from CRISPR 1. Neither mature crRNAs nor processed tracrRNAs were detected in the absence of Csn1 (Cas9) (Figure 5, A and E). Levels of an RNA corresponding to the 3′ trimmed 1X crRNA were also significantly reduced in the absence of Csn1 (Cas9) (59 nucleotides; Figure 5A, double arrowhead); however, levels of the 1X crRNA were notably not reduced (66 nucleotides; Figure 5A, single arrowhead), suggesting that tracrRNA-stimulated production of 1X crRNAs does not require Csn1 (Cas9). The tracrRNA product that accompanies 1X crRNA production is not observed (Figure 5E), which could simply reflect lower overall levels of both crRNAs and tracrRNAs and reduced traffic through the processing pathway. Processing of the tracrRNA from CRISPR 3 is unaffected by the absence of Csn1 (Cas9) associated with CRISPR 1 (Figure 5F). Our results indicate that Csn1 (Cas9) functions specifically in the production or stability of 5′ and/or 3′ processed 1X crRNAs. Moreover, our findings reveal that the Csn1 (Cas9) protein from CRISPR-Cas system 1 functions specifically in the production of mature crRNAs from the associated CRISPR locus in Sth.

The Csm Type III-A CRISPR-Cas system at CRISPR locus 2 in Sth includes a gene encoding Cas6, an endoribonuclease that we previously found binds to and cleaves within the repeat regions of crRNA precursors associated with Csa Type I-A, Cst Type I-B, and Cmr Type III-B systems in Pyrococcus furiosus (Carte et al., 2008, Carte et al., 2010). Cas6 cleaves 8 nt upstream of the guide region, generating the 5′ repeat sequence tag that is important for crRNA function with the Cmr complex in P. furiosus (Hale et al., 2012). Deletion of the cas6 gene resulted in loss of all species of the Csm crRNAs in S. epidermidis (Hatoum-Aslan et al., 2011). To determine whether Cas6 is involved in the biogenesis of crRNAs in Sth, we tested the ability of the Sth protein to bind and cleave repeat RNAs from each of the four Sth CRISPR loci in vitro. In gel shift analysis, Sth Cas6 demonstrated significant binding to RNAs comprised of the CRISPR 2 repeat sequence relative to repeat sequences from the other three loci (Figure 6A). In addition, Cas6 specifically cleaved RNA containing the CRISPR 2 repeat sequence, generating a product of ~32 nt, consistent with cleavage approximately 8 nt upstream of the guide sequence (Figure 6B). The previously determined structure of P. furiosus Cas6 (Carte et al., 2008) revealed a cluster of three conserved amino acids positioned similarly to a catalytic triad found in archaeal tRNA splicing endonuclease, which we found is critical for CRISPR repeat RNA cleavage, but not binding, in P. furiosus (Carte et al., 2010). We generated a mutation in one of the equivalent conserved amino acids in Sth Cas6 (H39A) and tested the activity of the protein with CRISPR 2 repeat RNAs. The mutation did not appreciably affect the affinity of the protein for CRISPR 2 repeat RNA in gel shift assays; however, cleavage activity was significantly reduced (Figure 6C and 6D). Our findings indicate that Cas6 specifically binds and cleaves RNAs from the Csm-associated CRISPR locus, and not from the other CRISPR loci in Sth.

Figure 6.

Cas6 from CC2 selectively binds and cleaves CRISPR 2 repeat RNA (A) and (B). ³²P-labeled RNAs composed of each of the four CRISPR repeats were incubated in the absence or presence of purified Sth Cas6 (0.2 μM). RNA binding was assessed by loading half of each reaction on a native 8% polyacrylamide gel (A) and RNA cleavage was evaluated by separation of the RNAs on a 15% denaturing (7 M urea) polyacrylamide gel (B). His39 of Sth Cas6 plays a critical role in catalysis but not RNA binding (C) and (D). ³²P-labeled CRISPR 2 repeat RNA was incubated with various concentrations (indicated in μM) of purified wild-type or H39A mutant Cas6 protein. RNA binding and cleavage were assessed by native gel mobility shift (C) and denaturing PAGE (D) as in (A) and (B). Specific RNA-protein interactions (A and C) and cleavage products (B and D) are indicated with asterisks.

The Cse3 (Cas6e) protein has been found to be essential for crRNA processing, and to bind and cleave crRNA repeat sequences in organisms with a lone Cse Type I-E system (Brouns et al., 2008, Gesner et al., 2011). We tested the activity of Sth Cse3 (Cas6e) with RNAs containing repeat sequences from each of the 4 CRISPRs in Sth. Cse3 (Cas6e) specifically bound and cleaved RNAs containing the repeat sequence from the Cse Type I-E system at CRISPR locus 4, generating a product of ~21 nucleotides, but not the other three loci (Figure 7A and 7B). Mutation in a conserved residue of E. coli Cse3 (Cas6e) (H20A) was previously found to disrupt crRNA biogenesis (Brouns et al., 2008). Mutation of an equivalent residue in Sth Cse3 (Cas6e) (H20A) prevents cleavage (Figure 7D), but our gel shift analysis indicates that it also disrupts interaction of Cse3 (Cas6e) with the CRISPR repeat RNA (Figure 7C). Our findings confirm that Cse3 (Cas6e) is the crRNA endoribonuclease of the Cse CRISPR-Cas system, but indicate that conserved residue H20 is required for RNA binding, making it more difficult to ascribe a role for this amino acid in catalysis based on existing information.

Figure 7.

Cse3 (Cas6e) from CC4 selectively binds and cleaves CRISPR 4 repeat RNA (A) and (B). ³²P-labeled RNAs composed of each of the four CRISPR repeats were incubated in the absence or presence of purified Sth Cse3 (Cas6e) (1 μM). RNA binding was assessed by loading half of each reaction on a native 8% polyacrylamide gel (A) and RNA cleavage was evaluated by separation of the RNAs on a 15% denaturing (7 M urea) polyacrylamide gel (B). His20 of Sth Cse3 (Cas6e) is indispensable for function (C) and (D). ³²P-labeled CRISPR 4 repeat RNA was incubated with various concentrations (indicated in μM) of purified wild-type or H20A mutant Cse3 (Cas6e) protein. RNA binding and cleavage were assessed by native gel mobility shift (C) and denaturing PAGE (D) as in panels (A) and (B). Specific RNA-protein interactions (A and C) and cleavage products (B and D) are indicated with asterisks.

DISCUSSION

CRISPR-Cas systems were identified as CRISPR arrays and associated modules of Cas protein-coding genes that segregate between prokaryotic genomes as intact units; multiple such CRISPR-Cas systems have now been identified. Some organisms naturally possess more than one CRISPR-Cas system while the introduction of a CRISPR-Cas system into organisms with existing CRISPR-Cas systems is an emerging goal in strain engineering. While the systems are generally capable of functioning independently, it is not known whether the systems share some components or processes, or alternatively may interfere with one another, when present in a common environment. Here we have examined four co-existing crRNA biogenesis pathways: Cse Type I-E and Csm Type III-A pathways that utilize Cas6 superfamily proteins, and two distinct Csn Type II-A pathways. Our findings indicate that all four of these CRISPR-Cas systems, representing the three major types of known CRISPR-Cas systems, function independently in crRNA biogenesis in Sth.

Independent crRNA processing pathways and unique crRNA species

The Cas6 superfamily proteins found in the Sth Csm Type III-A and Cse Type I-E systems – Cas6 and Cse3 (Cas6e) – specifically recognize and cleave only RNAs from the directly associated CRISPR locus (Figures 6 and 7). As summarized in Figure 8, cleavage by Sth Cas6 or Cse3 (Cas6e) generates an 8 nt repeat tag sequence and 5′ hydroxyl group that is retained in both the Csm Type III-A and Cse Type I-E crRNAs. Mature Cse Type I-E crRNAs are not further processed and retain a 21 nt repeat tag and a 2′-3′ cyclic phosphate at the 3′ end. On the other hand, the products of the Csm Type III-A system Cas6 cleavage are subject to 3′ end processing by an unknown mechanism that generates RNAs of a fixed length that have a 3′ hydroxyl group and lack 3′ repeat sequence.

Figure 8.

Summary of crRNA biogenesis pathways in Sth. CRISPR locus transcripts are cleaved within the repeat elements; cleavage of transcripts from each locus is mediated by distinct and specific factors as indicated. (The tracrRNAs that guide RNase III cleavage of Csn Type II-A transcripts from CRISPRs 1 and 3 contain sequences specific for the corresponding CRISPR repeat sequence.) The 1X RNAs from the Csn Type II-A and Csm Type III-A systems (CRISPRs 1-3) are futher processed. The Csn1 (Cas9) protein from the Csn Type II-A system at CRISPR 1 is specifically required for the accumulation of mature crRNAs from CRISPR 1. The four independent crRNA biogenesis pathways in Sth produce crRNAs with distinct combinations of repeat tag sequence, guide region length and end groups.

Co-existing CRISPR-Cas systems with Cas6 superfamily crRNA nucleases likely process crRNAs from a specific associated CRISPR in many cases (such as those documented here in Sth), but not in all cases. In Synechocystis, knockouts of Cas6 proteins from Csc Type I-D and Csm Type III-A systems affect production of crRNAs from the directly associated CRISPR, consistent with independently functioning coexisting Cas6 pathways (though the effect of the Cas6 protein knockouts on crRNA production from the other locus was not determined in that study) (Scholz et al., 2013). At the same time however, it is likely that one or both of these Cas6 proteins also contribute to processing of crRNAs from a third CRISPR present in Synechocystis that is associated with a Cmr Type III-B module. Moreover, evidence indicates that a single Cas6 crRNA processing enzyme generates crRNAs for both the Csa Type I-A and Cmr Type III-B systems present in a Sulfolobus islandicus strain (Deng et al., 2013).

We found that the Csn1 (Cas9) protein from Sth CC1 is required only for the production of crRNAs from the directly associated CRISPR, and in particular, is not required for biogenesis of crRNAs from the other Csn Type II-A system in the organism (Figure 5). In addition, the two Sth Csn Type II-A systems produce distinct crRNAs (Figure 2B). The crRNAs produced from both systems include ~2/3 of the encoded guide sequence and a 3′ repeat tag sequence (see Figure 8). However, while very similar in architecture, the major species of crRNAs from the two Csn Type II-A loci have guide sequences of slightly different lengths (20 and 21 nt) and distinct 3′ tags. Processing at the 5′ end of the 1X intermediate RNAs (generated by tracrRNA-stimulated RNase III cleavage) removes the 5′ repeat sequence and either 9 or 10 nt of the guide sequence (from CRISPR 1 and CRISPR 3 RNAs, respectively). The initial RNase III cleavage of CRISPR 1 RNAs proposed here (Figure 4B) would generate crRNAs with a 24 nt 3′ repeat sequence (similar to the 22 nt 3′ tag found on the crRNAs from CRISPR 3). However, the primary CRISPR 1 species found in Sth has a shorter 17 nt repeat tag. Our observations of RNAs from both CRISPRs 1 and 3 indicate that Csn Type II-A crRNAs may be subject to 3′ end processing as well as 5′ end processing (Figure 4). tracrRNAs appear to undergo parallel 5′ end processing (Figure 4). The major Csn Type II-A crRNA species retain 5′ monophosphate and 3′ hydroxyl end groups (Figure 3).

Our findings indicate that Csn1 (Cas9) may play a role in the 5′ or 3′ processing of Csn Type II-A crRNAs. In the absence of Csn1 (Cas9), RNase III production of the 1X intermediate does not appear to be reduced, but mature crRNA and tracrRNA species are lost (Figure 5, A and E). Csn1 (Cas9) has two nuclease active sites (RuvC and HNH type) (Makarova et al., 2011) that are both involved in silencing, each cleaving one strand of the DNA target (Jinek et al., 2012, Gasiunas et al., 2012). One of these sites (or a yet unidentified site) may be involved in crRNA processing; however, the role of Csn1 (Cas9) in crRNA biogenesis may also be indirect (e.g. protection of the regions found in the mature crRNA and also tracrRNA). The Csn1 (Cas9) protein associated with CRISPR 1 is essential for production or accumulation of mature CRISPR 1 RNAs but not crRNAs from the three other loci (Figure 5). The molecular basis for the interaction between Csn1 (Cas9) and crRNAs – which is likely essential for both crRNA processing and crRNA-guided DNA cleavage – is not known. Our results indicate that Csn1 (Cas9) discriminates between RNAs from the four CRISPRs in Sth.

Effective Cas9-based genome editing and gene regulation applications

Csn1 (Cas9) is the effector nuclease of the Csn Type II-A system, cleaving DNA targets recognized by the crRNAs (Garneau et al. 2010, Magadan et al, 2012), and has recently been co-opted as a powerful tool for genome engineering and modification of gene expression (Jinek et al., 2012, Gasiunas et al., 2012, Mali et al., 2013b, Cong et al., 2013, Cho et al., 2013, Hwang et al., 2013, Jiang et al., 2013, Qi et al., 2013, Chang et al., 2013, Dicarlo et al., 2013, Jinek et al., 2013, Terns & Terns, 2014). Csn1 (Cas9) DNA activity requires both a mature crRNA and tracrRNA (Karvelis et al., 2013, Gasiunas et al., 2012, Jinek et al., 2012) and understanding the precise nature of the functional forms of these RNAs will aid in optimization of Csn1 (Cas9) activity for the various emerging gene engineering applications. The 3′-end processing of Csn Type II-A crRNAs and 5′-end processing of tracrRNAs observed here in vivo (Figure 4) may yield more effective crRNA and tracrRNA species for (some or all) Csn1 (Cas9) proteins and applications. Detailed knowledge of Csn1 Type II-A systems capable of functioning independently within the same cellular environment, like the two distinct Csn Type II-A systems characterized here, will facilitate deployment of multiplexed approaches such as concurrent genome editing and modification of gene expression recently achieved with distinct Csn1 (Cas9) orthologs co-expressed in human cells (Esvelt et al., 2013).

Additional perspectives

Our finding that crRNA biogenesis occurs independently to generate distinct crRNA species in Sth suggests that these systems also likely function independently in silencing and adaptation in Sth, and other findings support this paradigm. The two Csn Type II-A systems require different PAM sequences adjacent to target sequences for both adaptation of the target sequence into CRISPRs and silencing (Garneau et al., 2010, Magadan et al., 2012, Deveau et al., 2008, Horvath et al., 2008). In addition, a recent study identified a distinct PAM-like requirement for silencing by the Cse Type I-E system in Sth (Sinkunas et al., 2013). We predict that the distinct crRNAs produced from each locus function exclusively with their corresponding silencing complexes in this organism.

The functional significance for co-occurrence of four distinct CRISPR-Cas loci in this bacterial strain remains to be further investigated, however the available evidence indicates a concept of redundant function of the modules as opposed to specialized function of a given module toward different types of invaders. There is evidence that each of the CRISPR-Cas systems is targeting phages and plasmids and that different CRISPR-Cas systems can target the same invaders (Barrangou et al., 2007, Garneau et al., 2010, Magadan et al., 2012, Sinkunas et al., 2013). CRISPR 1 and 3 (both Type II-A systems) can acquire spacers (that overlap in some cases) from the same phage. (CRISPR 2 and 4 do not acquire spacers under the same laboratory conditions that trigger adaptation by CRISPR 1 and 3.) Numerous Streptococcus thermophilus strains lack one or more of these four modules.

Our understanding of the co-function of specific CRISPR-Cas systems provides insight into the mechanisms of prokaryotic defense and is also important for predictable application of CRISPR-Cas technologies. Our findings indicate that Csn Type II-A, Cse Type I-E and Csm Type III-A CRISPR-Cas systems - and even two Csn Type II-A systems - can function independently within one environment to produce distinct crRNA species. Other co-existing CRISPR-Cas systems will share components (e.g. crRNAs) or processes (e.g. crRNA biogenesis) (Deng et al., 2013), and additional studies will be needed to determine the factors that restrict or allow cross-talk between systems.

MATERIALS AND METHODS

RNA isolation and RNA library preparation

S. thermophilus DGCC7710 (Sth) cells were grown to mid-log phase in M17 media (Oxoid) supplemented with 0.5% lactose (LM17). The csn1-1 (cas9) and csn2-1 gene disruption strains were generated in a previous study (Barrangou et al., 2007). The cells were harvested by centrifugation at 10,000 × g for 10 minutes and lysed by bead-beating with a Mini-Beadbeater (Biospec Products). Trizol LS (Invitrogen) was used to extract total RNA. RNA libraries used to examine crRNA expression were produced as described previously (Hale et al., 2009) except that no size-selection was done prior to RNA manipulation. Briefly, 10 μg of Sth total RNA was treated with Thermosensitive Alkaline Phosphatase (TSAP) (Promega) followed by 3′ ligation to a 5′ adenylated adaptor (see Table S1 for sequences) with T4 RNA ligase 2, truncated (NEB). The 3′ ligated RNAs were gel purified away from free adaptor and treated with T4 polynucleotide kinase (Ambion) before 5′ ligation to an additional adaptor with T4 ssRNA Ligase 1 (NEB). The 5′ and 3′ ligated RNAs were reverse transcribed with SuperScript II reverse transcriptase (Invitrogen), digested with RNase H (Promega), and PCR amplified with Crimson Taq DNA polymerase (NEB). For assessing cas gene mRNA expression profiles associated with each of the four CRISPR-Cas loci, including possible readthrough of Cas gene expression into each CRISPR locus (Figures S2 and S3), strand-specific RNA-seq libraries were prepared using prerelease Directional RNA-Seq Library Kits (Illumina). Briefly, 10 μg of Sth total RNA was depleted of ribosomal RNA (MICROBExpress kit, Ambion) and fragmented (RNA fragmentation reagent, Ambion) to an average size of ~200 nts. All other steps for library construction were done essentially as described above.

Illumina sequencing and analysis

RNA libraries were subject to 76 cycles of sequencing on an Illumina Genome Analyzer IIx. Sequence reads were trimmed of the 3′ linker, and reads 18–76 nt in length were aligned to the Sth genome using Bowtie (Langmead et al., 2009). The 5′ and 3′ ends of reads mapping to the CRISPR loci were calculated using a custom PERL script as described (Elmore et al., 2013). For the main RNA library used in this study (All figures except supplemetal figures S2 and S3), we obtained a total of 3,661,517 reads and had 3,484,419 reads that were 16–76 nt after trimming the adapters. Of this, 253,332 or 7.27% of the reads mapped uniquely to the genome. Of these reads, 33,863 or 13.37% represent crRNAs that map to the CRISPR loci. Thus, crRNA reads represent approximately 1% of the sequenced reads. For the rRNA-depleted library that specifically enabled analysis of cas gene expression (See supplemental figure S3), we obtained a total of 19,684,176 reads and had 17,538,483 reads that were 16–76 nt after trimming the adapters. Of this, 4,531,544 or 25.84% of the reads mapped uniquely to the genome.

TracrRNA/crRNA alignment (mapping putative RNase III cleavage sites)

Sth CRISPR 1 and 3 tracrRNAs were identified via homology search of the Sth genome with the CRISPR 1 and 3 repeat sequences, respectively, using the BLAT (Blast-like alignment tool) (Kent, 2002) function of the Archael Genome Browser; and tracrRNA/crRNA base-pairing was predicted by RNA-fold analysis (http://rna.tbi.univie.ac.at/cgi-bin/RNAfold.cgipair) of a tandem tracrRNA/crRNA fusion sequence (5′-tracrRNA-crRNArepeat-3′). The probable site of cleavage by RNase III was determined by mapping prominent tracrRNA and crRNA processing sites onto the predicted secondary structure. Paired cleavage sites located in a region of perfect homology and yielding a 2-base 3′ overhang (McRae and Doudna 2007) were observed for both CRISPR 1 and 3. These sites exihibit structural and (in the case of Sth CRISPR 3 significant sequence similarity) to the site previously defined for S. pygenes tracrRNA/crRNA cleavage (Deltcheva et al. 2011).

Chemical end group analysis of crRNAs

Chemical end groups of crRNAs from Sth were analyzed by enzymatic treatment followed by Northern analysis. To map 5′ chemical end groups, reactions containing 10 μg of Sth total RNA were treated with Terminator 5′ phosphate-dependent exonuclease (Epicentre) according to the manufacturer’s protocol. The reactions were terminated by phenol/chloroform/isoamyl alcohol (PCI) extraction followed by ethanol precipitation. Half of each reaction was separated on 7M urea TBE 15% polyacrylamide gels for Northern analysis, probing for the first spacer sequence from each CRISPR locus (Probe sequences can be found in Table S1). Northern blotting was performed as described previously (Hale et al., 2009).

For 3′ chemical end group analysis of crRNAs from CRISPR loci 1–3, 100 μg of Sth total RNA was separated on a 7M urea TBE 15% acrylamide gel and RNAs between ~30 and ~65 nucleotides were gel purified as described previously (Hale et al., 2009). Gel-purified RNAs were treated with E. coli poly(A) polymerase (NEB) in reactions containing 10% of the gel-purified RNA, 1X poly(A) reaction buffer, and 1 mM ATP. Reactions were incubated at 37° C for 20 minutes followed by PCI extraction and ethanol precipitation. Northern analysis was performed with half of each reaction. Chemical end groups of crRNAs from CRISPR locus 4 were analyzed using combinations of poly(A) polymerase, TSAP, calf intestinal phosphatase (Promega) and T4 polynucleotide kinase. Sth total RNA (10 μg) was treated with either TSAP or T4 polynucleotide kinase, or untreated. Reactions were stopped by PCI extraction and ethanol precipitation. RNAs were then treated with poly(A) polymerase as described above and Northern analysis was performed following PCI extraction and ethanol precipitation.

Cloning and mutagenesis of Sth cas6 and cse3 (cas6e)

Genomic DNA was isolated from Sth as described previously (Hill et al., 1991) with minor modifications. Cells were not treated with proteinase K, and the DNA was precipitated with isopropanol/ammonium acetate prior to phenol/chloroform extraction. DNA primers specific to Sth cas6 and cse3 (cas6e) genes (listed in Table S1) were designed and ordered from Eurofins MWG Operon. The primers were used to amplify cas6 and cse3 (cas6e) from Sth genomic DNA using the Expand High Fidelity PCR system (Roche) according to the manufacturer’s protocol. The genes were cloned into the pET24d plasmid using the In Fusion PCR cloning system (Clontech) according to the manufacturer’s protocol. Colonies were screened by PCR, and nucleotide sequences of positive clones were confirmed by DNA sequencing. These constructs were used to generate mutant cas6 and cse3 (cas6e) constructs using specific primers (listed in table S1) and the QuikChange site-directed mutagenesis kit (Stratagene). The nucleotide sequences of mutant constructs were confirmed by DNA sequencing.

Production of recombinant Sth Cas6 and Cse3 (Cas6e) proteins

Sth Cas6 and Cse3 (Cas6e) (both 6x N-terminally histidine-tagged) were expressed and purified from E. coli as described previously (Hale et al., 2009) with the following modifications. The cells (from a 100 mL culture) expressing Sth Cas6 were resuspended in 20 mM sodium phosphate (pH 7.6, buffer A), 0.1 mM phenylmethysulfonyl fluoride (PMSF), and 5 mM imidazole. Cells expressing Sth Cse3 (Cas6e) were resuspended in buffer A supplemented with 50 mM NaCl and 20 mM β-mercaptoethanol. The cells were lysed by sonication and centrifuged at 14,000 rpm for 10 minutes. Proteins were isolated from the cleared lysate by batch purification using 50 μL Ni-NTA agarose beads (Qiagen). Proteins were bound by end-over-end rotation for one hour at room temperature. The beads were washed three times with buffer A and once with buffer A supplemented with 20 mM imidazole. Proteins were eluted from the beads in buffer A plus 300 mM imidazole. Elutions containing Sth Cas6 were dialyzed into 50 mM Tris-HCl (pH 7.0) using Slide-a-lyzer MINI dialysis cassettes (ThermoScientific). Elutions containing Sth Cse3 (Cas6e) were dialyzed into the same buffer supplemented with 50 mM NaCl and 20 mM β-mercaptoethanol. Glycerol was added to 50%,% and the proteins were stored at −20° C until use.

RNA binding and cleavage reactions

Radiolabeled crRNA repeat substrates were generated by in vitro transcription as described previously ((Carte et al., 2008), see Table S1 for oligo sequences). RNA binding and cleavage assays were carried out as described previously (Carte et al., 2008) with the following modifications. Sth Cas6 was incubated with 5,000 cpm of uniformly ³²P-labeled RNA substrate in 25 mM Tris-HCl (pH 7.0), 0.75 mM DTT, 1.5 mM MgCl₂, 5 μg E. coli tRNA, and 10% glycerol. The reaction conditions for Sth Cse3 (Cas6e) were identical except that DTT was replaced with 5 mM β-mercaptoethanol and NaCl was added to 50 mM. Reactions were incubated at 37° C for 30 minutes prior to electrophoretic separation on both native TBE 8% polyacrylamide (RNA binding) and 7M urea TBE 15% acrylamide (RNA cleavage) gels. The gels were dried and RNAs detected by phosphor imaging (Amersham).

Abbreviations

CRISPR

clustered regularly interspaced short palindromic repeats

Cas

CRISPR-associated

crRNA

CRISPR RNA

Sth

Streptococcus thermophilus

tracrRNA

trans-activating CRISPR RNA