Cleavage of a complementary RNA in vivo

Very few RNAs were detected that were transcribed from the opposite (non-leader) ends of the CRISPR loci in P. furiosus; however, we did detect significant antisense transcription from an apparently fortuitous promoter within CRISPR locus 1. A BRE/TATA promoter sequence is present in the antisense orientation within the second guide sequence of CRISPR locus 1, and a significant number of antisense RNAs that begin distinctly 21 nt downstream of the promoter are present in the sequence from the total RNA sample (Figure S1, blue reads). The number of antisense RNA reads is approximately 1/3 the number of reads observed for the sense strand crRNA reads (crRNA 1.01; Figure S1, red reads). Northern analysis reveals a series of antisense RNAs up to ~140 nt in length in total P. furiosus RNA (Figure 4A, 1.01 antisense, total (T) lane).

The antisense CRISPR transcripts found in P. furiosus would be predicted to be recognized and targeted by Cmr complexes containing the crRNA encoded on the opposite strand of CRISPR locus 1 (crRNA 1.01). Previous in vitro studies established that the Cmr complex cleaves complementary RNAs 14 nt upstream of the 3’ ends of the 39- and 45-nt crRNAs (Hale et al., 2008). Accordingly, in this case, the 1.01 crRNAs would be predicted to generate 5’ products of 45 and 39 nt (as well as 3’ products of various lengths) upon cleavage of the antisense transcripts. And indeed, distinct antisense RNAs of 45 and 39 nt were observed in Northern analysis of the total RNA (with a probe to the 5’ region of the antisense transcript; Figure 4A, 1.01 antisense, total (T) lane), suggesting that the Cmr complex targets and cleaves the endogenous CRISPR1 antisense transcript in P. furiosus.

We also found in both RNA sequencing and Northern analysis that the 45- and 39-nt putative 5’ cleavage products are specifically co-immunopurified with the Cmr complex (Figure 4A, 1.01 antisense, immune (I) vs. preimmune (PI) lanes; and Figure 4B, blue reads vs. Figure S1, blue reads), suggesting that the products remain associated with the complex to some extent following cleavage. The 45- and 39-nt antisense RNAs do not possess the 5’ repeat tag sequence, which we hypothesize is important for recognition and function of crRNAs with the Cmr complex. However, because of the coincidence in the size of these RNAs with Cmr complex crRNA species, we tested the possibility that the antisense RNAs are functional guide RNAs. We found that the immunopurified Cmr complexes do not cleave a target RNA complementary to the antisense RNAs, indicating that the antisense RNAs do not guide cleavage by the Cmr complex (Figure 4C, antisense RNA target). At the same time, the immunopurified complexes do cleave an RNA corresponding to the ~140-nt antisense RNA transcript, generating 5’ products of the same sizes observed in vivo (45 and 39 nt; Figure 4C, crRNA target). Together, the findings indicate that the Cmr complex recognizes and cleaves endogenous complementary RNAs in vivo.

The CRISPR repeat tag sequence is required for functional Cmr complex crRNAs

Its conservation suggests that the 5’ CRISPR repeat tag is an important element of functional Cmr complex crRNAs. We investigated the importance of the tag by testing several tag sequence mutants for the ability to guide cleavage by reconstituted Cmr complexes. Removal of the repeat tag sequence from crRNA 7.01 eliminates cleavage of a complementary target RNA (Figure 5A). Substitution of the entire tag sequence to its complement (AUUGAAAG → UAACUUUC) or of the first two nucleotides of the tag sequence (AUUGAAAG → GAUGAAAG) also prevents function of the crRNA, as does the addition of a G at the 5’ end of the tag (Figure 5B).

Open in a separate window

Figure 5

The 5’ repeat tag is required for formation of functional Cmr complexes

A) The 5’ repeat tag is required for Cmr complex function. Reconstitutions were performed with recombinant Cmr1-6 proteins and the 45-nt. wildtype 7.01 crRNA (AUUGAAAG tag) or crRNA 7.01 lacking the tag (−), and activity was tested against the radiolabeled 7.01 target RNA. RNA marker sizes are indicated. The cleavage product observed with the wildtype crRNA is indicated with an asterisk. B) Mutations to the tag sequence prevent function. Reconstitutions were performed with the 45-nucleotide wildtype 7.01 crRNA (AUUGAAAG tag) or crRNA 7.01 with the indicated mutations in the tag sequence (mutated or inserted nucleotides are highlighted in red) and tested in cleavage assays as described in A. The cleavage product observed with the wildtype crRNA is indicated with an asterisk. C) crRNAs with variant tag sequences are underrepresented in the Cmr complex. Deep sequencing profiles of Cmr-associated crRNAs in the region encoding crRNAs 8.08 – 8.11 are shown (see also Figure 2A). The positions of CRISPR repeats (black) and guide sequences (green) are indicated. The 5’ tag sequence is shown below for crRNAs 8.08 - 8.11. An additional nucleotide present in the wildtype tag of crRNA 8.11 is highlighted in red D) Complementarity to the tag does not prevent function. Cmr complexes reconstituted with wildtype crRNA 7.01 were tested for the ability to cleave the 7.01 target RNA or a target with complementarity to the tag sequence (as illustrated). Cleavage products are indicated with asterisks. Non-contiguous lanes from the same gel are indicated with a dashed line.

Analysis of CRISPR locus 8 provided further insight into the basis for the importance of the 5’ repeat tag in vivo. crRNAs from locus 8 were significantly underrepresented in the Cmr complex (Figures 2A and S2). The reduced presence of CRISPR locus 8 RNAs in the Cmr complex correlates with a difference in the 5’ repeat tag found on these RNAs. Most of the crRNAs from locus 8 possess a 7-nt AUUGAAG (rather than 8-nt AUUGAAAG) repeat tag due to a variation in the repeat sequences within this CRISPR locus (Figure 5C). The exception is crRNA 8.11 (the last crRNA encoded by locus 8), which retains the canonical 8-nt repeat tag sequence, and this crRNA is selectively enriched in the Cmr complex relative to the other crRNAs from CRISPR locus 8 (Figures 2A and ​and5C).5C). Together, these results indicate that the CRISPR repeat tag is critical for the formation of functional Cmr complexes by crRNAs.

For CRISPR-Cas systems that target DNA, the 5’ repeat tag on the crRNAs provides a key function in distinguishing self from non-self. In organisms with these systems, it is important for crRNAs to avoid targeting the complementary DNA sequence found in the host CRISPR locus itself, and evidence indicates that, in these systems, potential targets with complementarity to the tag sequence (found in the CRISPR locus but not in invader targets) are not silenced (Marraffini and Sontheimer, 2010). We tested the effect of target complementarity to the tag sequence on RNA cleavage by the Cmr complex, and observed no significant reduction in activity (Figure 5D). The antisense transcript that is cleaved in vivo by the Cmr complex also is complementary to the tag of crRNA 1.01 (Figure 4). Thus, RNA targeting by the Cmr complex does not require a lack of complementarity between the target and the crRNA repeat tag.

Anatomy of the Cmr complex guide RNAs

The crRNAs associated with the Cmr complex are of two sizes, 39 and 45 nt in length, both with an 8-nt CRISPR repeat sequence at the 5’ end (5’ tag) followed by either 31 or 39 nt of invader guide sequence (Figures 1 and ​and3,3, and (Hale et al., 2009)). Unlike other crRNAs found in P. furiosus, the Cmr crRNAs contain little or no CRISPR repeat sequence at the 3’ end (Figure 3). Our studies revealed that the crRNAs present in Cmr complexes are essentially restricted to an absolute length of 39 or 45 nt, which may be defined by a mechanism that measures from the 5’ end of the crRNA or from the 5’ tag sequence, but does not depend on the position of the guide sequence-3’ repeat sequence boundary (Figure 3). Both the 39- and 45-nt crRNAs have 5’ hydroxyl and 3’ phosphate end groups (Figure 1D). Overall, the 39- and 45-nt forms are found in roughly equal proportion in the purified Cmr complexes (Figure 3A); however, the proportion of the two forms for an individual crRNA can vary significantly (see crRNAs 4.01 and 4.05, Figure 3B), and the basis for this variation is unclear. It is not yet known whether individual Cmr complexes contain one or more crRNAs, however, functional Cmr complexes can be reconstituted with either the 39- or 45-nt crRNA species, indicating that both forms are not simultaneously required for cleavage activity and hinting that each Cmr complex may harbor a single crRNA.

While the 5’ ends of the Cmr complex crRNAs are presumably generated by cleavage of CRISPR transcripts by the site-specific riboendonuclease Cas6 (Carte et al., 2010; Carte et al., 2008; Wang et al., 2011), we do not yet know how the 3’ ends of mature crRNAs are generated or how the 39- and 45-nt crRNAs are selectively incorporated into Cmr complexes. This exquisite selectivity may simply be because the Cmr complex proteins preferentially bind crRNAs of these lengths. Alternatively, the proteins may associate with longer crRNA species (e.g., the 1X intermediate generated by Cas6 cleavage that includes an additional 22 nt of the repeat at the 3’ end) and the 3’ ends may be generated following formation of the complexes. The nuclease(s) responsible for crRNA 3’ end trimming could be an intrinsic subunit of the Cmr complex or an extrinsic enzyme that may or may not be specific for crRNA processing. The precisely measured 3’ ends of the Cmr complex crRNAs could result, for example, from a combination of non-specific nuclease activity and specific RNA protection by Cmr proteins. The 45-nt crRNA and the other crRNA size forms found in P. furiosus (which contain longer 3’ ends; see Figures 1 and ​and3A)3A) are components of other RNP complexes currently under investigation ((Hale et al., 2008), and our unpublished data). The P. furiosus genome encodes 21 Cas proteins in addition to the 6 Cmr proteins, including Csa and Cst subtype proteins that are also expected to associate with crRNAs to form additional effector complexes. Future work will elucidate the mechanisms by which distinct crRNA species are sorted among CRISPR-Cas complexes within a cell.

The 5’ tag is critical for Cmr complex crRNA function

The results presented here indicate that the 5’ tag sequence is necessary for crRNA function in Cmr complexes (Figure 5). We found that the sequence, and perhaps size, of the 5’ tag is important for Cmr complex crRNA function; even subtle mutations are detrimental to Cmr activity. We hypothesize that the 5’ tag is important as a binding site for one or more Cmr proteins. Consistent with this idea, we found that crRNAs from CRISPR locus 8 with variant 5’ repeat tags are preferentially excluded from Cmr complexes in vivo (Figure 5C). As discussed above, the basis for exclusion of longer crRNA species, which possess the 5’ tag, from Cmr complexes in P. furiosus is not yet clear. The presence of a 5’ tag sequence is a conserved feature of crRNAs in many CRISPR-Cas systems (Brouns et al., 2008; Hale et al., 2008; Hale et al., 2009; Haurwitz et al., 2010; Lintner et al., 2011; Marraffini and Sontheimer, 2008). A correlation exists between specific CRISPR-Cas modules and the CRISPR repeat sequences found together in various organisms, suggesting that the different systems may recognize characteristic 5’ tags. However, in P. furiosus the crRNAs encoded by the 7 CRISPR loci generally share the same 5’ tag sequence, which evidently functions with the Csa and Cst as well as Cmr modules. We hypothesize that the 5’ tag is also important in effector complex function in the other systems where it is found. Our results reveal the functional importance of the 5’ tag in Cmr effector complex function.

Directing Cmr complexes to cleave RNAs of choice

In this work, we have demonstrated that crRNAs can be designed to target the Cmr complex to cleave novel RNA substrates (Figure 6). crRNAs comprised of a 37-nt complementary guide sequence and 5’-tag were used to direct specific cleavage of RNAs including the β-lactamase mRNA in vitro. Cleavage product sizes indicate that the target was predictably cut 14 nt downstream of the 3’ end of the engineered crRNA. These findings will enable efforts to exploit the site-specific cleavage function of Cmr-crRNA complexes for in vivo gene knockdown experiments. Our studies have defined the minimal components for Cmr complex function (Cmr1-6 and a crRNA; Figure 6 and (Hale et al., 2009)), making it conceivable to express functional Cmr complexes in eukaryotes as well as additional prokaryotes to provide a novel method of RNA silencing. Our results indicate that Cmr-crRNA complexes have the potential to target destruction of mRNAs encoding any gene of interest and to become important research tools with industrial and biomedical applications.

Co-IP/Western analysis

Western blotting of co-IP material was performed by standard procedures.

Co-IP/Northern analysis

Trizol LS reagent (Invitrogen) was used to isolate RNAs from the co-IP samples and P. furiosus extract (total RNA). Northern analysis was performed as described previously (Hale et al., 2008) with modifications (see Supplemental Experimental Procedures). Probe sequences can be found in Supplemental Table 1.

Co-IP/Activity assays

Co-IP beads were washed into 40 mM Hepes pH 7.0, 500 mM KCl and added to a 20 μl cleavage reaction containing radiolabeled target RNA (5,000 cpm). Reactions were incubated for 3 hours at 70°C with mixing. RNAs were isolated by proteinase K treatment, phenol extraction and ethanol precipitation and separated on 7M urea 15% polyacrylamide gels and visualized by phosphor imaging.

Co-IP/End analysis

Trizol LS was used to isolate small RNAs from co-IP samples. The RNAs were treated with thermostable shrimp alkaline phosphatase (Promega). Treated and non-treated RNAs was subject to either: 1) 5’ radiolabeling with γ³²P-ATP (MP Biomedicals) via polynucleotide kinase (Ambion), or 2) 3’ radiolabeling via T4 ssRNA Ligase 1 (NEB) with cytidine-3’,5’-bisphosphate [5’-³²P] (3,000 ci/mmol, MP Biomedicals). Both labeling reactions were carried out for 1 hour at 37°C and the RNAs were isolated, separated by denaturing gel electrophoresis, and visualized by phosphor imaging.

Small RNA library preparation

For preparation of total and Cmr-associated small RNA libraries, Trizol was used to extract RNAs from P. furiosus cells or from a co-IP reaction, respectively. 50 μg of total RNA was used for total small RNA library preparation. Briefly, the RNAs were phosphatase treated to remove 3’ phosphates, followed by 3’ ligation to an adaptor. These products were treated with T4 polynucleotide kinase before 5’ ligation to an additional adaptor. The 5’ and 3’ ligated products were subject to reverse transcription, RNase H digestion and PCR.

Illumina sequencing and analysis

cDNA libraries were visualized by electrophoresis, quantitated by Agilent Bioanalyzer and Nano-drop, diluted to 2 pM, and subject to 76 cycles of sequencing on the Illumina Genome Analyzer IIx. Sequence reads were trimmed of the 3’ linker and reads 18-76 nt in length were aligned to the P. furiosus 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 available upon request.

Recombinant Cmr protein production and activity assays

Expression and purification of recombinant Cmr proteins was performed as described (Hale et al., 2009) with modifications (See Supplemental Experimental Procedures). RNA cleavage assays were performed as described (Hale et al., 2009). Briefly, all six recombinant Cmr proteins (500 nM each) were pre-incubated with unlabeled crRNA (0.05 pmol) prior to addition of 5’ radiolabeled target RNAs (5,000 cpm). Reactions were incubated for 1-3 hours at 70°C and isolated RNAs analyzed by denaturing gel electrophoresis and phosphor imaging.

We thank the UCHC Translational Genomics Core Facility for use of the Illumina Genome Analyzer IIx. This work was supported by NIH grants RO1GM54682 (including ARRA funds) to M.P.T. and R.M.T. and R01GM062516 to B.R.G.