CRISPR loci deletions are not dependent on protospacer transcription

Each of the above changes observed in the CRISPR loci will prevent both the CRISPR/Cas and CRISPR/Cmr systems from targeting the cloned viral genes or their transcripts respectively (Marraffini and Sontheimer, 2008; Hale et al., 2009; Shah et al., 2009). In order to inhibit transcription and remove the possibility of viral gene transcripts feeding back onto the CRISPR loci and inducing spacer deletions, vectors were constructed for the genes of ATV145 and ATV618 where no transcription occurred. Promoter regions were replaced by an elongated archaeal T-rich terminator region (see Experimental procedures). For both these constructs, again transformation efficiency was very low and two transformants were analysed from each construct. For the ATV145 construct, CRISPR locus D deletions of 4.4 and 4.6 kb were estimated from PCR reactions (Fig. 3C) and for the ATV618 constructs, CRISPR locus D deletions of 3.3 and 3.4 kb were detected. Subsequent sequencing showed that all four lacked matching spacers 28 or 29 respectively (Table 1). For ATV145, the crRNA is complementary to the reverse transcript and it was established by Northern blottings that there was no transcription on either strand across the protospacer (Fig. 4B). For ATV618, where the crRNA is complementary to the mRNA, using the same vector construct, no transcription was observed in Northern blots (data not shown). This result eliminates the possibility that the deletions occurring within the CRISPR loci resulted from a feedback mechanism involving transcripts from either strand of the gene.

Some CRISPR/Cas modules may be inactive

CRISPR locus F has been shown, exceptionally, to be identical in sequence between S. solfataricus strains P1 and P2, and to lack a leader region and some cas genes (Lillestøl et al., 2006; 2009;). This suggested that its capacity to add new spacers had been impaired. Therefore, we investigated whether CRISPR locus F was affected by vector-borne matching protospacers. Spacer 67 of CRISPR locus F (Fig. 2) matches imperfectly (three mismatches) the gene of a putative partitioning protein NOB8-315 encoded by the Sulfolobus conjugative plasmid pNOB8 (She et al., 1998). The gene with three mismatches was inserted into pEXA2 cloning site in addition to a mutated gene with a perfect match. Constructs were also prepared with and without the family I CC PAM motif but each construct produced transformation efficiencies similar to the empty vector [on average 10⁴ cfu (µg DNA)⁻¹] and no deletions of spacer 67 were detected in the sequenced PCR products (Table 1). Therefore, we infer that cluster F is inactive and this is compatible with the observation that the RNA transcript from cluster F is incompletely processed such that the active crRNA components, in the size range 35–45 nt, are not formed (Fig. S1).

Protospacers carrying mismatches also affect CRISPR loci in Sulfolobus islandicus REY15A and S. solfataricus P2

Given that the CRISPR loci changes were observed independently of whether viral genes carrying matching protospacers were transcribed, we inferred that it was sufficient to perform further experiments with vector-cloned protospacer DNAs carrying appropriate PAM motifs (Lillestøl et al., 2009). Moreover, since a reliable and stable genetic system has recently been developed for Sulfolobus islandicus REY15A (Deng et al., 2009), which also carries a much simpler CRISPR/Cas system with only a tandem pair of CRISPR loci 1 and 2 (and two cmr cassettes) (Fig. 2B), we performed a series of parallel experiments in which we synthesized and cloned protospacers matching to selected CRISPR spacers, or combinations of spacers, for both S. solfataricus P2 and S. islandicus REY15A (Fig. 2).

Several vector constructs were prepared each carrying a protospacer showing increasing degrees of sequence mismatches (listed in Table 2) to spacer 45 of family I CRISPR locus 2 of S. islandicus REY15A and to spacer 28 of family II CRISPR locus A of S. solfataricus P2 (Fig. 2) and the constructs carried the protospacer PAM motifs CC for the former and CT for the latter (Table 2). Both spacers were 38 bp in length and constructs were prepared with perfect matches, one mismatch at spacer positions 1, 19 or 38, two mismatches exclusively for S. solfataricus at positions 1 and 38, and additional constructs were made with four mismatches (positions 34–38) and 8 or 10 mismatches (positions 29–38) all at or near the 3′ end (Table 2). For each construct, including the 8–10 mismatches at the 3′ end, low transformational efficiencies were observed compared with the empty vector (Table 3). Furthermore, for all except the latter constructs, some deletions were observed within CRISPR loci covering the matching spacer (Table 2).

PCR products across the deleted regions of locus 2 of S. islandicus are shown in Fig. 5A for transformants arising from the perfectly matching protospacers, with control PCR reactions provided for locus 1 showing no changes except for sample 8 where both CRISPR loci had been deleted. These PCR products are indicated in a summary of all the results listed for S. islandicus in Table 3. Similarly, for the S. solfataricus constructs, examples of PCR products from constructs carrying both perfectly matching, and mismatching, protospacers are given in Fig. 5B, including an example, in sample P3, of an ISC1359 insertion. All the results are summarized in Table 3.

Open in a separate window

Fig. 5

Agarose gels showing examples of PCR products obtained from CRISPR loci of transformants carrying protospacers and CC and CT PAM motifs cloned into pDL1 and pEXA2 respectively.

A. S. islandicus REY15A where both locus 2 carrying the matching spacer and, as a control, locus 1 were amplified.

B. S. solfataricus P2. P1 to P7 refer to the experiments listed in Table 3.

M indicates DNA size markers. Bracketed spacer numbers denote the presence of the matching spacer. All the illustrated samples carrying deletions are marked with asterisks in Table 3.

In a control experiment to confirm that the PCR products reflected the predicted and predominant genomic changes, and not some minor smaller products, Southern blot experiments were performed on the S. islandicus CRISPR loci. Results for five of the transformants in Fig. 5A which produced major genomic changes (samples 2, 3, 4, 6 and 9) were tested. The results presented in Fig. 6A all show fragment sizes that are consistent with the sizes of the PCR reactions (Fig. 5A; Table 3)

Open in a separate window

Fig. 6

Southern blot analysis of CRISPR locus deletions of S. islandicus REY15A transformants where sample C denotes a control sample of S. islandicus REY15A. The deleted regions determined by sequencing are indicated above each well.

A. Analysis of deletions in locus 2 determined from the PCR products shown Fig. 5A where the numbers in the brackets correspond to the sample numbers on agarose gel.

B. Samples correspond to the PCR products illustrated in Fig. 7A and B. Tracks 1 and 2 carry transformants 1 and 2 while tracks 3 and 4 carry transformants with protospacers matching both spacer 20 of locus 1 and spacer 45 of locus 2 (Table 5).

The lower diagram shows the restriction sites in the CRISPR region where the expected DNA fragment sizes are indicated.

There was an important difference between the two sets of results. Whereas changes in the CRISPR locus 2 of S. islandicus exclusively involved deletions, insertions of the transposable element ISC1359 were observed within matching spacer 28 for some of the S. solfataricus transformants, after its target site ACGT (Redder and Garrett, 2006) (Table 3; Fig. 5B). This result is consistent with the relatively high transpositional activity found only in the latter strain (Martusewitsch et al., 2000; Redder and Garrett, 2006; Deng et al., 2009).

Testing for the significance of the family I CC PAM motif

A CC PAM motif was predicted to be important for family I crRNA targeting in Sulfolobus species, with a low tolerance for a T at either position (Lillestøl et al., 2009; Shah et al., 2009). Therefore, using the S. islandicus genetic system, we prepared a series of constructs with protospacers matching spacer 45 of CRISPR locus 2 (Fig. 2) with variants of the CC motif. High transformation levels (see Experimental procedures) were observed with motifs GG, GA and TT, and there was no evidence of deletions in CRISPR loci among the transformants examined (Table 4). The results obtained with CT and TC were less clear. The former yielded relatively high transformation efficiency but produced a single large deletion including the matching spacer, for one of six transformants examined, while the TC motif produced few transformants but no evidence of deletions in the four transformants examined. Only CC yielded low transformation efficiencies and high levels of deletions, for 12 of the 17 transformants examined (Table 3). These experimental results confirm that the predicted CC PAM motif is required for targeting foreign DNA in vivo for the family I CRISPR/Cas system of Sulfolobus (Lillestøl et al., 2009) thereby facilitating the distinction of foreign protospacer sequences from host CRISPR spacer sequences which carry the equivalent motif AA in their repeat.

Strains and vector construction

The S. solfataricus P2 vector host S. solfataricus P2-InF1 carried a pyrF gene inactivated by a single copy of ISC1225 and the S. solfataricus vector carried Sulfolobus acidocaldarius pyrE/F genes. For S. islandicus, the double-deletion mutant S. islandicus E233S was used, which carried a large pyrEF gene deletion and a complete lacS gene deletion (Deng et al., 2009).

pDL1 was constructed by self-ligation of the PvuII- and ZraI-digested pZC1 large fragment (Peng et al., 2009). pEXA was produced by the triple ligation of products of: SphI- and AvaI-digested pDL1; SphI- and NdeI-digested araS promoter fragment, and NdeI- and SphI-digested multiple cloning site (MCS) fragment. The araS promoter fragment was amplified by primer araSF (5′-gcGCATGCTTTTTTTTAGAAAAACATCCAATATGTTAAC-3′) and araSR (5′-ttttttttCATATGCTCGGGTACTTTTATGACCTAAC-3′) using total S. solfataricus P2 DNA as template (underlined nucleotides indicate restriction digestion sites and small letters constitute protection nucleotides for restriction sites).

The MCS fragment was synthesized by primers MCS1, MCS2, MCS3, MCS4, MCS5 and MCS6 (MCS1-5′ ttttttttCATATGCATCATCACCATCATCATAGTAGTGGTTTAGT; MCS2-5′CATCGATTCGCGATCCCCTTGGTACTAAACCACTACTATGATGATGGTGA; MCS3-5′AAGGGGATCGCGAATCGATGCTAGCTACGCGTCTCCGGATGTACAAAGGC; MCS4-5′ATCAGCGTCGACCTTATCGTCATCATCAGGCCTTTGTACATCCGGAGACG; MCS5-5′GACGATAAGGTCGACGCTGATCAAGCGGCCGCACACCATCATCATCACCA; MCS6-5′aaCCCGGGAAAAAAAAGATTTTGCTTAGTGGTGATGATGATGGTGTGCG) as follows: all primers were added to the final concentration of 0.2 µM in a standard 50 µl PCR reaction: 94°C for 20 s, 65°C for 20 s and 72°C for 20 s over 15 cycles. One microlitre of the product was then added as DNA template to a standard PCR reaction using MCS1 and MCS6 as primers. After 20 cycles of 94°C for 20 s, 65°C for 20 s and 72°C for 20 s, the MCS fragment was synthesized.

To construct protospacer-containing vectors, SOE-PCR was performed to generate protospacer fragments. Briefly, primers were designed with about 18 nt overlap at their 3′ ends and were added to a standard PCR reaction, run at 95°C for 3 min, 30 cycles at 95°C for 15 s, 40°C for 15 s and 68°C for 5 s. The product was then purified by a Nucleotide removal kit (Qiagen, Westburg, Germany), digested overnight and repurified using this kit. Ligation was performed by T4 ligase with the corresponding digested vector at 22°C for 1 h. The ligation product was transformed into E. coli DH5α directly by heat shock following a standard protocol (Sambrook and Russell, 2001) and transformants were screened by colony PCR and verified by sequencing.

For S. solfataricus experiments, the S. solfataricus pyrEF genes in pEXA were replaced by pyrEF genes from S. acidocaldarius to minimize the possibility of homologous recombination. The pEXA vector was fully digested with Cfr91, destroying one of three KpnI sites on the vector and then partially digested with KpnI to remove the pyrEF genes and the cut vector was purified with a QIAquick gel extraction kit (Qiagen). The S. acidocaldarius pyrEF genes were amplified with primers S.aci1: 5′-tcccCCCGGGAGAAAAAAAAGCTATGGATATTGTCTTACCAC-3′ and S.aci2: 5′-cgg GGTACCagaaaaaaaatctgttgtgggaacttcac-3′, purified and cut with Cfr91 and KpnI and then inserted into the linear pEXA to produce pEXA2. Bold nucleotides represent mismatches to the genomic sequence to enable transcriptional terminators to be inserted into the vector on both sides of the pyrEF genes. The integrity of the whole construct was confirmed by DNA sequencing.

To remove the araS promoter from pEXA2, the plasmid was digested with SphI and NdeI. To inhibit transcription, a terminator sequence was generated using the primer set Term/SphI: 5′-acatGCATGCTTCTTTTTCTTTCCCTTCTTTTTTTACC-3′ and Term/NdeI: 5′-ggaattcCATATGAAAAAGAGGTAAAAAAAGAAGGG-3′ and inserted into the cut vector.

Gene constructs

pEXA2 was cut with MluI and EagI and purified with QIAquick gel extraction kit (Qiagen, Westberg, Germany). Primers with overhanging restriction sites were used to amplify the genes from an in-house libraries of the ATV virus and pNOB8. PCR fragments were purified using either QIAquick PCR purification kit or QIAquick gel extraction kit (Qiagen, Westberg, Germany), before cutting with MluI and EagI and ligating into the cut pEXA2 vector. Correct inserts were verified by restriction analyses and DNA sequencing.

Site-directed mutagenesis was accomplished using mutagenic primers as described by Heckman and Pease (2007). Internal mutagenic primers are designed overlapping one another and used with the original gene construct primers to create two intermediate PCR products. These PCR products were then used as template DNA for a second round of PCR to generate the mutated gene.

Determination of transformation efficiencies

Transformation efficiency measurements for the two Sulfolobus strains used are not quantitative for two main reasons. First, the plating material Gel-rite generally contains low and variable levels of pyrimidine compounds, some of which survive autoclaving, and therefore some non-transformants always appear on plates. Moreover, as discussed earlier for the two-layer Gel-rite plating system, some non-transformants appear on plates because of pyrimidines liberated from lysis of dead cells during the long (7–21 days) incubations (Deng et al., 2009). We invariably distinguished transformants from non-transformants by growing them in uracil-deficient media after plating, and for all the experiments where CRISPR loci underwent deletions there were very few viable transformants. Experiments included in Tables 1 and ​and33 were performed over a 3-year period using different batches of Gel-rite, and competent cells, and therefore we only give the approximate low percentage range of the control transformation efficiencies. Experiments involving altering the PAM motif (Tables 4 and ​and5)5) were performed over a shorter time period with one batch of Gel-rite and competent cells. The transformation efficiencies given are averaged and approximate.

Localizing CRISPR locus deletions

PCR reactions were performed with either Taq or Pfu polymerase. Annealing temperatures used were 5°C lower than the estimated Tm of the primer. Elongation times allowed for about 1 min per kilobase pair of the expected product. For the experiment in Fig. S2 elongation times were reduced from 7 min to 2.5 min to enhance the yields of the smaller products. Amplified DNA was purified and generally ran on 0.8% agarose gels with DNA size markers (GeneRuler™ DNA Ladder Mix, 0.1–10 kb, Fermentas).

PCR products were obtained across the chromosomal regions of S. solfataricus P2 CRISPR loci A to D and of S. islandicus REY15A CRISPR loci 1 and 2 using pre-mixed Ex Taq according to the manufacturer's protocol with 75–100 ng of genome DNA in a 10 µl reaction. For S. solfataricus, primer pairs AF: 5′-TGGCGGTTATTAATTGGGA-3′ and AR: 5′-TTGCGGATTCTTGACGTG-3′ and DF: 5′-GCACGCTTCCTACCTTCATTTCCACTAC-3′ and DR: 5′-CATCCCTCTAACCCTTCCCAACCTCATA-3′ were used to amplify the CRISPR loci A and D respectively. For S. islandicus C1F: 5′-AGCTTGCTTACCTCAAGGTACTTTACGT-3′ and C1R: 5′-TTAATAAACGACGATTTTCCTCTTGAT-3′, C2F: 5′-AGGATAGCGAAGTCGTAG AGTTTGGAT-3′ and C2R: 5′-TAACGCACGGTATTGAAACTTCTCATC-3′ were used to amplify CRISPR loci 1 and 2 respectively. Purified PCR products were sequenced directly (Eurofins MWG Operon, Ebersberg, Germany) or first cloned using CloneJET™ PCR Cloning Kit (Fermentas, St. Leon-Rot, Germany) and then sequenced.

RNA preparation and Northern blotting

Total RNA was prepared using Trizol (Invitrogen, Paisley, UK) according to the Invitrogen protocol essentially as used for extracting plant si-RNAs (Sunkar et al., 2005). For Northern blotting of small RNAs, 20 µg of RNA was mixed with 10 µl of Gel Loading Buffer II (Applied Biosystems/Ambion, Austin, USA) and fractionated in a 6–10% polyacrylamide gel containing 7 M urea, 90 mM Tris, 90 mM boric acid, 2 mM EDTA, pH 8.3, together with a 10–150 nt ladder (Decade Marker System, Ambion, Huntingdon, UK) or a 0.1–2.0 kb RNA ladder (Invitrogen). Procedures for transferring, and immobilizing RNA on nylon membranes, pre-hybridizing, end-labelling of complementary nucleotides, hybridization and film exposure followed the protocol described earlier (Lillestøl et al., 2009). The following oligonucleotides probes were used for Northern blotting: 5′-TCGGTTCCACATCTAGTTATCAAA-3′ to detect transcription from spacer 29 of CRISPR locus D of S. solfataricus P2; 5′-CCAATCCCAAGATACATCATCG-3′ to detect reverse transcription from the ATV145 gene, and 5′-CGATGATGTATCTTGGGATTGG-3′ to determine forward transcription from ATV145 gene. The latter unlabelled probe was also used as positive control sample for the detection of ATV145 reverse transcription.

We appreciate Hien Phan's assistance with DNA analyses. Shiraz A. Shah provided useful bioinformatic insights and discussions. Reidun K. Lillestøl and Xu Peng are thanked for their helpful advice and Reidun K. Lillestøl ran the Northern analysis in Fig. S1. The work was supported by grants from the Danish Natural Science Research Council (Grant No. 272-08-0391), the Danish Research Council for Technology and Production (Grant No. 274-07-0116) and the Danish National Research Foundation.