A Complex of Cas Proteins 5, 6, and 7 Is Required for the Biogenesis and Stability of Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR)-derived RNAs (crRNAs) in Haloferax volcanii^*

Strains

H. volcanii strains H119Δcas6 (ΔpyrE2, ΔleuB, ΔtrpA, Δcas6) (this study) and H26ΔcasCluster28 (ΔpyrE2, ΔleuB, ΔtrpA, ΔpHV4:207.288–218.340) (42) were grown aerobically at 45 °C in Hv-YPC medium (42). H. volcanii strain H119Δcas6 containing plasmids with mutated cas6 genes were grown in Hv-Ca medium supplemented with 0.25 mm tryptophan. Escherichia coli strains DH5α (Invitrogen) and GM121 (43) were grown aerobically at 37 °C in 2YT medium (44). Halorubrum lacusprofundi DL12 (45) was grown aerobically in Hv-YPC medium at 37 °C.

Construction of Plasmids and Transformation of H. volcanii

The 3×FLAG tag sequence was amplified by PCR using the oligonucleotides FLAGN and FLAG2 (46). The PCR fragment was digested with NdeI and HindIII and cloned into pTA927 (43), yielding pTA927-FLAG. The cas6 gene was amplified by PCR from genomic DNA using the oligonucleotides Cas6FLAGNup and Cas6FLAGNdo (primer sequences are listed in supplemental Table 1). The resulting fragment was cloned into pBluescriptII (Stratagene) linearized with EcoRV, yielding pBlue-NFLAGcas6 (supplemental Table 2). This construct was used as a template to introduce the mutations, either by the QuikChange II site-directed mutagenesis kit (Agilent Technologies) or by inverse PCR using 5′-phosphorylated oligonucleotides followed by religation of the PCR fragment, using oligonucleotides carrying the desired mutation in both cases (supplemental Table 1). The resulting pBluescriptII constructs carrying the wild type cas6 gene and mutated cas6 genes were digested with HindIII and EcoRI, and the cas6 gene variants were subcloned into the plasmid pTA927-FLAG, yielding the constructs pTA927-cas6 and pTA927-cas6mutX, respectively (supplemental Table 2). The following constructs were generated to complement the strain H26ΔcasCluster28 with different sets of cas genes expressed from H. volcanii vector pTA927: pTA927-Cas6875, pTA927-Cas685, pTA927-Cas687, and pTA927-Cas675. pTA927-Cas6875 was generated by PCR amplification of cas6, cas8, cas7, and cas5 from genomic DNA of the wild type strain H119 with oligonucleotides Cas6FLAGCup and 3-Cas5-BamHI, followed by digestion of the PCR fragment with NdeI and BamHI and its subsequent cloning into pTA927. pTA927-Cas685 was generated by PCR amplification using oligonucleotides Cas6FLAGCup and 3-Cas5-BamHI and genomic DNA from an H119Δcas7 strain as a template, followed by digestion of the PCR fragment with NdeI and BamHI and its subsequent cloning into pTA927. pTA928-Cas687 was generated by PCR amplification using oligonucleotides Cas6FLAGCup and 3-Csh2-HindIII and genomic DNA of the wild type strain H119, followed by digestion of the PCR fragment with NdeI and HindIII and its subsequent cloning into pTA927. pTA927-Cas675 was generated by PCR amplification using oligonucleotides Cas6FLAGCup and 3-Cas5-BamHI and genomic DNA from the H119Δcas8 strain as a template, followed by digestion of the PCR fragment with NdeI and BamHI and its subsequent cloning into pTA927. All plasmids were passaged through E. coli GM121 cells to avoid methylation and then introduced into H. volcanii strains H119Δcas6 or H26ΔcasCluster, respectively, using the polyethylene glycol method (47, 48). Transformants were selected on Hv-Ca plates without uracil.

Northern Blot Hybridization

Total RNA was isolated from exponentially growing H. volcanii cells as described (49) or isolated from the FLAG-Cas7 purified protein fraction as described below. After separation of 10 μg of RNA (total RNA) or 600 ng (FLAG-Cas7 fraction) on 8% denaturing gels, RNA molecules were transferred to nylon membranes (Hybond-N⁺, GE Healthcare) and incubated with oligonucleotides against the different spacers (primer P1.1, which is complementary to the sequence of the first spacer of locus P1; primer P1.2, which is complementary to the sequence of the second spacer of locus P1; primer P2.1, which is complementary to the sequence of the first spacer of locus P2; and primer C1, which is complementary to the sequence of the first spacer of locus C); as an RNA loading control, a hybridization with oligonucleotide 5 S, which binds to the 5 S rRNA was performed. All primers were radioactively labeled at the 5′-end with [γ-³²P]ATP. To quantify the amount of mature crRNA, the membranes were exposed to imaging plates (BAS-MS, Fujifilm) and analyzed using the FLA-3000 scanner (software BASreader version 3.14). The intensity of the signals was measured with ImageJ. The signals of the P1.1 detection were put into relation to the 5 S rRNA signal, which was used as RNA loading control. To obtain the percentage of mature crRNA in the mutants, the signal of complementation of the cas6 deletion strain with the wild type cas6 gene was set to 100%, and the crRNA amounts of the cas6 mutant strains were set in relation to these data. All Northern analyses were performed at least three times. In the analysis of the crRNA amounts with the ΔcasCluster strain, the crRNA signals in the wild type strain were set to 100%, and the ΔcasCluster strains transformed with the different sets of cas genes were set in relation to the wild type crRNA amounts.

RT-PCR and 5′-RACE

RNA was isolated as described above, and 1 μg was subjected to several reverse transcription reactions with primers Csh1#2, Csh2#2, Cas5#2, Cas3#5, Cas4#2, Cas1#5, and Cas2#2. The resulting cDNA was amplified with primers used for RT-PCR and primers Cas6#4, Csh1#3, Csh2#3, Cas5#3, Cas4#4, and Cas1#4. PCR fragments were cloned into pBluescriptII (digested with SmaI), and resulting clones were sequenced. For the 5′-RACE, 1 μg of RNA treated with 5′-phosphate-dependent exonuclease (Epicenter) to enrich primary transcripts was reverse transcribed with primer Cas6#3 using the MINT universal kit (Evrogen) according to the manufacturer's instructions. For the PCR, primer M1 from the MINT kit and primer Cas6#2 were used. PCR fragments were cloned into pBluescriptII (digested with SmaI), and resulting clones were sequenced.

Expression of cas6 in E. coli and H. volcanii

To express the Haloferax cas6 in E. coli, the gene was amplified by PCR using the oligonucleotides 5-Cas6-NcoI and 3-Cas6-NotI. The fragment was digested with NcoI and NotI and cloned into the vectors pET28a and pET32a (Novagen), yielding the constructs pET28a-cas6 (N-terminal His₆ tag) and pET32a-cas6 (N-terminal His₆, S, and Trx tag). For the construct pET30a-cas6-FLAGN, the cas6 gene together with the N-terminal 3×FLAG tag was removed from the construct pTA927-cas6 by digestion with NdeI and EcoRI and cloned into the vector pET30a (digested with NdeI and EcoRI). To obtain a construct with a C-terminal tag, a PCR was performed using the oligonucleotides 5-Cas6-NcoI and 3-Cas6-HindIII, followed by digestion of the resulting fragment with NcoI and HindIII and cloning into the vector pET30a, yielding pET30a-cas6o.stop (C-terminal His₆ tag). To improve the expression, the cas6 gene was optimized to the E. coli codon usage by GeneArt (Invitrogen). The optimized gene was isolated from the original plasmid by digestion with NcoI and XhoI and cloned into different pET vectors, yielding the constructs pET28a-cas6h and pET30a-cas6h. E. coli strains BL21AI and BL21 codon plus (Novagen) were transformed with the respective constructs, and the protein was expressed according to the manufacturer's protocol. To improve the solubility of the recombinant protein, several different expression conditions (growth temperature and the amount of isopropyl 1-thio-β-d-galactopyranoside and/or arabinose for induction) were tested. To express the cas6 gene in Haloferax cells, the strain H119Δcas6 was transformed with the construct pTA927-cas6 (see above). To obtain a construct with His₆ tag, the cas6 gene was amplified by PCR, using the oligonucleotides 5-Cas6-PciI and 3-Cas6-NotI. The fragment was digested with PciI and NotI and cloned into the vector pTA963 (43), yielding the construct pTA963-cas6. The Haloferax strain H1424 (50) was transformed with this construct. To express the tagged Cas6 protein, the respective strains were grown in Hv-Ca medium containing 0.25 mm tryptophan. At exponential growth phase, tryptophan was added to a final concentration of 3 mm, and the culture was grown for an additional 2–16 h. To prepare soluble extracts, the cells were harvested and washed in enriched PBS buffer (2.5 m NaCl, 150 mm MgCl₂, 1× PBS (137 mm NaCl, 2.7 mm KCl, 8 mm Na₂HPO₄, 2 mm K₂HPO₄, pH 7.4)). The cells were resuspended in lysis buffer (1 m NaCl, 100 mm Tris/HCl, pH 7.5, 1 mm EDTA, 10 mm MgCl₂, 1 mm CaCl₂, 8 units/μl DNase RQ1 (Promega), 13 μl/ml protease inhibitor mixture (Sigma)) and lysed by ultrasonification. Insoluble cell debris was removed by centrifugation at 48,000 × g. For affinity purification via the 3×FLAG tag, anti-FLAG M2 affinity gel (Sigma) was equilibrated with ice-cold washing buffer (50 mm Tris/HCl, pH 7.5, 1 m NaCl) and added to the extract. After incubation overnight at 4 °C, the affinity gel was washed extensively with washing buffer, and the protein was eluted by washing buffer to which 3×FLAG peptide (Sigma) was added to a final concentration of 150 ng/μl. His purification was performed according to the manufacturer's protocol with Protino nickel-nitrilotriacetic acid-agarose (Macherey-Nagel), using buffers with different concentrations of NaCl from 300 mm to 2 m.

Purification of a FLAG-tagged Cas7 Protein

To obtain the construct pTA927-Cas68[HisFLAG]75, which contains all four cas genes cas5–8 with the cas7 gene fused to cDNAs for a His and a FLAG tag, first the 3×FLAG sequence was amplified by PCR using the primers FLAGSnaBI and FLAG2. The product was digested with SnaBI and HindIII and cloned into the vector pCDF1b (Novagen) digested with PmlI and HindIII, resulting in the construct pCDF1b-FLAG. This construct was digested with XhoI, treated with Pfu polymerase to generate blunt ends, and then digested with HindIII. The genes cas7 and cas5 were amplified from genomic Haloferax DNA using primers 5-Csh2-HindIII and 3-Cas5-BamHI, and the resulting fragment was digested with HindIII and cloned into the pCDF1b-FLAG vector, yielding the construct pCDF1b-75HisFLAG. The genes for Cas7 and Cas5 together with the N-terminal His₆ and 3×FLAG tag were amplified from this construct using the primers HisFLAGBamHI and 3-Cas5-XbaI and digested with BamHI and XbaI. The genes cas6 and cas8b were amplified by PCR from genomic Haloferax DNA with the primers Cas6FLAGCup and 3-Cas8-NcoIBamHI, digested with NdeI and BamHI, and cloned into the vector pTA927, digested with the same enzymes. This construct was digested with BamHI and XbaI, and the insert Cas75HisFLAG was ligated into it, yielding the construct pTA927-Cas68[HisFLAG]75. The sequence was verified by sequencing, and subsequently Haloferax strain H26casCluster28 was transformed with the plasmid. Cells were grown to exponential phase in HvCa medium containing 0.25 mm tryptophan. Expression of the cas genes and FLAG purification were performed as described above for cas6 gene expression in Haloferax. The resulting purified protein fraction was analyzed using SDS-PAGE and mass spectrometry.

Isolation of RNA from the FLAG-Cas7 Purified Fraction

The FLAG-Cas7 purified fraction was incubated with 20 μg of proteinase K for 30 min at 37 °C in 100 μl of buffer (100 mm Tris/HCl, pH 7.5, 12.5 mm EDTA, 150 mm NaCl, 0.2% SDS); subsequently, the solution was extracted with phenol/chloroform/isoamylalcohol. RNA was precipitated from the aqueous phase, and the resulting pellet was dissolved in water. The RNA fraction was analyzed using a Northern blot as described above.

Mass Spectrometry

For mass spectrometric analysis, proteins were in-gel-digested with trypsin as described previously (51). Extracted peptides were analyzed by liquid chromatography tandem mass spectrometry (LC-MS/MS) on an Orbitrap XL instrument (Thermo Fisher Scientific) under standard conditions. Peptide fragment spectra were searched against a target decoy database for H. volcanii (52), using MASCOT as a search engine. Peptides with a peptide score lower than 20 were not considered specific.

Quantification of Cascade Subunits Using Intensity-based Absolute Quantification (iBAQ)

0.25 μg of the proteins co-purified with FLAG-Cas7 as described above were dried in an Eppendorf tube and dissolved in 10 μl of aqueous solution of Universal Proteomics Standard 2 (UPS2; Sigma-Aldrich) in 1% Rapigest (Waters) to obtain a nominal protein concentration ratio of 0.25:1 μg of total (Cas proteins/UPS2). The sample was reduced with dithiothreitol (10 μl, 50 mm in 100 mm triethylammonium bicarbonate, 37 °C, 1 h), alkylated with iodacetamide (10 μl, 100 mm in 100 mm triethylammonium bicarbonate, 37 °C, 1 h), diluted with 70 μl of 100 mm triethylammonium bicarbonate, and digested with porcine trypsin (Promega; 1:20, 37 °C, 16 h). Following digestion, the Rapigest detergent was cleaved by acidifying the solution with trifluoroacetic acid (20 μl, 5%, 37 °C, 2 h) and pelleting the released fatty acids by centrifugation (13,000 rpm, 30 min, room temperature). The supernatant containing tryptic peptides was transferred to a fresh tube, taken to dryness in a SpeedVac, and dissolved in LC-MS loading buffer (30 μl, 5% acetonitrile, 0.1% formic acid). Samples were analyzed in triplicate (3 × 5 μl) on a nanoflow liquid chromatography system (1100 series, Agilent) coupled to an LTQ-Orbitrap Velos mass spectrometer in a vented column setup at an analytical flow rate of 300 nl/min achieved through passive splitting. Samples were desalted on an RP-C18 precolumn (20 mm, 0.15-mm inner diameter, ReproSil-Pur C18-AQ 5 μm; Dr. Maisch). Separation was achieved on an RP-C18 column (150 mm, 0.075-mm inner diameter, ReproSil-Pur C18-AQ 3 μm; Dr. Maisch) packed into a SilicaTip emitter (FS360-75-10-N, New Objective) using a 50-min linear gradient of 3–36% acetonitrile with 0.1% formic acid as modifier). MS data were acquired using a Top8 method with CID fragmentation, using a normalized collision energy of 45%.

Results were analyzed using MaxQuant software version 1.2.7.4 (53). The MS data were matched against an H. volcanii-filtered UniProt/trEMBL protein sequence database (version 2013-11) supplemented with the sequences of the 48 proteins contained in the UPS2 standard. iBAQ values from replicate values were averaged, and the S.D. value was calculated to judge the precision of analysis (supplemental Table 3). A total of 20 UPS2 standard proteins were observed in three of three replicate analyses and used for calibration. A calibration curve was obtained by linear regression from a double logarithmic plot (log(iBAQ) versus log(amount); supplemental Fig. 1). The calibration function was then used to calculate the respective Cas5-7 amounts in the co-purified fraction.

Generation of a cas6 Gene Deletion Strain

The deletion of the cas6 gene was achieved by using the pop-in/pop-out method as described previously (43, 44, 46). The cas6 gene was PCR-amplified with flanking regions (∼500 base pairs each) from the chromosomal DNA of H. volcanii strain H119 using primers Cas6KOup and Cas6KOdo. The resulting 1915-nt PCR fragment was subsequently cloned into the vector pTA131 (EcoRV), yielding pTA131-cas6geneupdo. To remove the cas6 gene, an inverse PCR was performed on pTA131-cas6geneupdo with primers IPCas6up and IPCas6do2; the resulting linear PCR product was ligated, yielding pTA131-cas6updo. Plasmids were passaged through E. coli GM121 to prevent methylation, and H. volcanii strain H119 was subsequently transformed with this construct to allow integration (pop-in) of the plasmid into the genome. The subsequent selection for loss of the pyrE2 marker by plating on 5-fluoroorotic acid revealed pop-out mutants. Chromosomal DNA was isolated from the wild type and potential cas6 deletion mutants. Southern blot hybridization was performed as described (47), with the following modifications: 10 μg of SalI-digested DNA was separated on a 0.8% agarose gel and transferred to a nylon membrane (Hybond^TM-N, GE Healthcare). The 500-base pair cas6 downstream region was amplified using primers IPCas6KOdo2 and Cas6KOdo, labeled using the PCR DIG Probe Synthesis Kit (Roche Applied Science), and used as a hybridization probe. Hybridization and detection were performed according to the DIG manual (DIG Luminescent Detection Kit, Roche Applied Science).

Cloning of the Heterologous cas6 Genes

The cas6 gene from Methanosarcina mazei was isolated from the original plasmid pRS714 (13) by PCR using primers Cas6Mmazfwd and Cas6Mmazrev. After digestions with NdeI and EcoRI, the gene was cloned into pTA927 (digested with NdeI and EcoRI), yielding pTA927-Mmazcas6. The cas6 gene from M. maripaludis was isolated from the original plasmid pHR6 (16) by digestion with NdeI and HindIII and was cloned into the pTA927 vector, which was previously digested with NdeI and HindIII, thus yielding pTA927-Mmarcas6. The cas6 genes Hlac3333 and Hlac3572 from H. lacusprofundi were ordered from Geneart (Invitrogen). Clones for Hlac3333 and Hlac3572 were digested with NcoI and BamHI and with NcoI and EcoRI, respectively, and resulting fragments were then cloned into vector pTA1228, yielding pTA1228-Hlac3333 and pTA1228-Hlac3572. pTA1228 is an improved version of the pTA963 overexpression plasmid (43). It features a unique NspI site in the polylinker (downstream of the His₆ tag). This NspI site is compatible with SphI ends and allows for the cloning of genes where the second codon starts with C. pTA1228 was derived from pTA963 by the removal of two existing NspI sites. The NspI site (bp 1033) located between the pBluescript backbone and the pHV2 origin was inactivated by cutting with NsiI and filling in with Klenow. The NspI site (bp 7165) located in the hdrB selectable marker was inactivated by replacement of a 762-bp BspEI-NcoI fragment (comprising the 5′ part of hdrB and the 3′ part of pyrE2) with a nearly identical 762-bp BspEI-NcoI PCR fragment that features a synonymous Gly → Ala mutation (GAC → GAT) at the NspI site. The internal hdrB-dNsp primers with the Gly → A mutation were hdrB-dNsp F and hdrB-dNsp R. The sequence and map of pTA1228 are available upon request.

Modeling of cas6 Three-dimensional Structure

To obtain a prediction of the tertiary structure of the Haloferax Cas6 protein, the amino acid sequence was submitted to the Phyre server (protein homology/analogy recognition engine) (54). According to the Phyre database, the Pryrococcus Cas6 protein showed the closest related structure. The three-dimensional structure was modeled with the help of the graphical software PyMOL.

Interference Tests

H. volcanii strain H119Δcas6 was transformed with the plasmids containing the cas6 wild type gene (pTA927-cas6) or the cas6 mutants (pTA927-cas6mutX). In a second transformation, these strains were transformed with the invader plasmid pTA352-PAM3 or pTA352-PAM9 (supplemental Table 2) (42). Transformants were selected on Hv-Min plates without leucine and uracil. As a positive control, the respective strains were transformed with the vector pTA352. To confirm a successful interference reaction, H. volcanii cells were transformed at least three times with one of the plasmid invader constructs. As has been observed in similar studies (42, 55), it is difficult to accurately determine transformation rates; therefore, we defined only those sequences that led to at least a 100-fold reduction in transformation rates (reduced by a factor of 0.01) in this plasmid assay as a successful interference reaction.

The Family of Cas6 Proteins and the Haloferax Cas6 Protein

The Haloferax Cas6 protein shows very little sequence similarity to other Cas6 proteins from type I systems and to most systems in general (Fig. 2). As visible in the alignment, only three positions are conserved in all Cas6 proteins analyzed (Gly-122, Gly-258, and Gly-258). Only one amino acid, shown to be important for catalytic activity in other organisms, is conserved in the Haloferax Cas6 protein (His-41). To analyze the family of Cas6 proteins and to determine Cas6 protein conservation, we first performed a BLAST search with the H. volcanii Cas6 (YP_003533663.1) and subsequently quantified similarity via pairwise global sequence alignments (using the NCBI implementation of the Needleman-Wunsch algorithm (56)). Close homologs of the Haloferax Cas6 protein were only found in closely related Haloarchaea with global percentage identities between 41 and 44% and BLAST E-values of <1 × 10⁻⁶⁰, including Haloferax mediterranei (ATCC 33500), Haloarcula marismortui (ATCC 43049), Halorhabdus utahensis (DSM 12940), H. lacusprofundi (ATCC 49239), Natrinema (sp. J7-2), and Halomicrobium mukohataei (DSM 12286). Apart from the Haloarchaea, the similarity to other Cas6 proteins is very low, with pairwise similarities of only 13–22% to other Cas6-like proteins (Fig. 3). Even the Cas6 protein from P. furiosus, which of all of the Cas6 proteins investigated is the closest to H. volcanii, has an equally low similarity of 18%. The Cas6 proteins from other systems show comparably low levels of sequence similarities with each other.

Conservation of the H. volcanii repeats is analogous to the conservation of the Cas6 protein; the Haloferax repeat sequences are very different from other published systems, except in Haloarchaea. Repeats from haloarchaeal CRISPR-Cas systems are, in fact, well conserved and can form a minimal hairpin structure with three base pairs (23). When using the Web server CRISPRmap (39), which compares all available repeats according to sequence and structure similarity, the haloarchaeal CRISPR repeats cluster together into the sequence family F19. The uniqueness of the haloarchaeal family F19 is highlighted by the fact that it is located on an isolated branch of the hierarchical CRISPRmap tree of all repeat sequences (Fig. 3). Thus, in the large family of Cas6-like proteins, not only is the sequence similarity low, but the individual proteins are highly specific to their associated repeat sequences and have different functional mechanisms. Obviously, the full spectrum of this family has not yet been explored. Therefore, we analyzed the Haloferax Cas6 protein in detail to provide insights into this extremely diverse family.

cas Genes Show a Low Level of Expression

Earlier studies showed that the CRISPR RNAs are constitutively expressed in Haloferax (42). To analyze whether the cas genes are likewise constitutively expressed, we used Northern blot analyses to determine the amount and the length of the cas mRNAs. No signals were detected for the cas mRNAs on Northern blots with RNA isolated from cells grown under different conditions (standard conditions, low and high temperatures, low and high salt concentrations) (data not shown). Therefore, we used RT-PCR to determine if the genes are transcribed at all, revealing that the mRNA for all cas genes is present (data not shown). Employing 5′-RACE, we determined the 5′-end of the cas6 mRNA, which is located 39 nucleotides upstream of the Cas6 protein start codon. Expression of the cas genes was confirmed in a parallel study in which the Haloferax proteome was investigated.³ Here, it was shown that all eight Cas proteins were present in the proteome in both the exponential and stationary phases. Taken together, we could show that the cas mRNA is present, albeit in low amounts.

Cas6 Is Necessary for crRNA Production

To investigate the biochemical characteristics of the Cas6 protein, we expressed recombinant Cas6 protein in E. coli and H. volcanii using different approaches (for details, see “Experimental Procedures”). The cas6 gene was cloned into different expression vectors, and Haloferax and different E. coli strains were transformed with these constructs. All attempts to produce an active recombinant Cas6 protein failed; therefore, we next used an in vivo approach to investigate the Cas6 protein. To determine the in vivo function of Cas6 in Haloferax, we generated a cas6 deletion strain. Because the cas6 gene overlaps with one nucleotide of the downstream cas8b gene (Fig. 1), we deleted the complete reading frame with the exception of the last nucleotide that is shared with the downstream cas8b gene. The cas6 deletion strain showed no visible phenotype in comparisons with the respective wild type strain when grown under different conditions, such as different temperatures and salt concentrations (data not shown). To investigate whether Haloferax is able to produce mature crRNAs without the Cas6 protein, we isolated RNA from wild type and cas6 deletion strains and analyzed the amount of crRNAs by Northern blot. Hybridization with probes against spacer 1 from locus P1 (Fig. 4) and spacers from the other CRISPR loci (P1.2, P2.1, and C1) (data not shown) revealed that the wild type strain contains mature crRNAs, but the deletion strain does not. To confirm that the failure to produce or stabilize mature crRNAs is due to the missing Cas6 protein and not to some effect on the downstream-encoded cas genes, we complemented the cas6 deletion strain with the FLAG-Cas6 overexpression construct (described under “Experimental Procedures”) (Fig. 4). Complementation restores the wild type amount of crRNAs, verifying that Cas6 is required for efficient crRNA production. To investigate whether cas6 genes from other organisms would be able to complement the deletion mutant, we transformed the deletion mutant with the cas6 genes from another halophilic archaeon, H. lacusprofundi. Northern analysis showed that these genes are not able to complement the deletion mutant strain (data not shown). Furthermore, we complemented the cas6 gene deletion mutant with type I-B cas6 genes from two non-halophilic archaea, M. maripaludis (16) and M. mazei (13). Neither of the two heterologous cas6 genes was able to complement the Haloferax cas6 gene deletion mutant (data not shown). The repeat sequences of the CRISPR RNAs from the four different organisms show almost no sequence similarities, which might be the reason that the heterologous Cas6 proteins cannot process the Haloferax repeat. Taken together, we could show that without the cas6 gene, crRNAs cannot be generated or maintained in vivo in Haloferax.

Single Amino Acid Mutations in Cas6 Reveal both a Loss and a Gain in crRNA Levels

To identify amino acids necessary for generating or stabilizing crRNAs, we introduced 21 point mutations in the cas6 gene to change the original amino acid to alanine (Fig. 2). We have chosen amino acids for mutation that, according to the alignment in Fig. 2, either directly align with amino acids shown in other Cas6 proteins to be important for function or are in the vicinity of such amino acids. In addition, amino acids that were shown to be conserved in at least three Cas6 proteins were chosen. The Haloferax H119Δcas6 strain was transformed with the 21 cas6 mutant constructs, and RNA was subsequently isolated for Northern analyses. Hybridization with a probe against spacer 1 from the CRISPR locus P1 showed that most of the mutants were still able to generate the crRNA (Fig. 5 and Table 1). Only three mutations resulted in reduced (less than 50% of wild type amount) crRNA amounts. Changing the amino acid His-41 to Ala reduced the amount of crRNA to 37% of the wild type amount, whereas changing Gly-256 or Gly-258 reduced it to 12 and 0% of the wild type amount, respectively. Interestingly, we also observed a gain of function in two mutants (mutants Ser-115 and Ser-224) that resulted in higher crRNA amounts of 129 and 164%, respectively. To investigate whether the severely reduced crRNA levels in mutants Gly-256 and Gly-258 are due to parts of the protein being in the insoluble fraction due to improper folding, we isolated whole cell extract from the transformed cells and analyzed the insoluble and soluble protein fraction for the presence of the Cas6 protein mutants. Both Cas6 protein mutants were detectable in the soluble fraction, and the amount of protein did not change significantly (data not shown). Therefore, the reduced activity is not due to production of an insoluble protein.

Cas6 Mutations Can Abolish the Interference Reaction

Type I Cas6 proteins have been shown to be involved in the generation of crRNAs. They have also been shown to stay attached to the crRNA after processing. To determine whether an intact Cas6 is important for the interference reaction and which amino acids are important for this activity, we subjected all cas6 mutants to an interference test, which we previously established in Haloferax by using a plasmid-based invader (42). The invader plasmid contains a protospacer that is recognized by one of the Haloferax crRNAs and PAM motifs that are recognized by the Haloferax CRISPR-Cas system and trigger the defense reaction. Plasmid invaders were introduced into a Haloferax strain that cannot grow without uracil because the strain lacks the pyrE2 gene. Transformants were selected by uracil prototrophy conferred by the pyrE2 gene on the plasmid invader. Plasmid elimination via the CRISPR-Cas defense of the host cell were detected by a drastically reduced transformation rate. Cells with an active interference system degrade the invader plasmid efficiently, reducing the transformation rate by at least a factor of 0.01. Transformation of the cas6 deletion strain with the invader plasmid resulted in normal transformation rates, showing that deletion of the cas6 gene abolishes the CRISPR interference reaction of the cells (Table 2). If the cas6 deletion strain was complemented with the cas6 wild type gene and then transformed with the invader plasmid, only a few transformants were obtained (the transformation rate was reduced by a factor of 0.01) (Table 2), showing that the interference was working again. Next, the cas6 gene deletion strain was transformed, first with the mutated cas6 genes and then with the invader plasmid, to investigate which Cas6 protein mutations would interfere with the defense reaction. Three mutants showed reduced or no interference reactions: His-41, Gly-256, and Gly-258 (Table 2). Transformation of the cas6 deletion strain with the His-41 mutant and the invader plasmid resulted in reduction of the transformation rate by only a factor of 0.2; the plasmids were not degraded very efficiently. Mutants Gly-256 and Gly-258 showed a normal transformation rate; thus, the interference reaction was not active and was not able to degrade the plasmid. The increased crRNA amounts in the Ser mutant strains did not change the behavior in the interference reaction. The interference reaction was working efficiently, reducing the transformation rate by a factor of at least 0.001 (Ser-224) and 0.002 (Ser-115) (Table 2). Taken together, we could show that cas6 mutations, which reduce the crRNA amounts, likewise reduce the effectiveness of the interference reaction.

A Strain with Only cas6 Is Not Active in crRNA Production

Next, we investigated whether Cas6 alone is sufficient for crRNA maturation. We used a Haloferax strain in which all cas genes were deleted (H26ΔcasCluster28 (42)) for transformation with the wild type cas6 gene to investigate whether the Cas6 protein alone is sufficient for pre-crRNA processing and stabilization. Northern analysis of this strain showed no crRNA maturation, suggesting that the cas6 gene is necessary for crRNA production (because deletion of the cas6 gene results in loss of crRNA processing) but not sufficient for crRNA maintenance. Therefore, one or more of the other Cas proteins are also required. The Cas proteins that are potentially also involved in crRNA maintenance are Cas5, Cas7, and Cas8b, which are encoded together with Cas6 at the 5′-end of the cas gene locus (Fig. 1). To analyze whether the presence of all four of these genes rescues crRNA generation, we transformed strain H26ΔcasCluster28 with the complete 5′ part of the cas gene cluster (encoding the Cas5–8b proteins) (Fig. 6A). RNA from this strain was analyzed with Northern blots for crRNA generation, revealing that crRNAs were generated and stably maintained (Table 3 and Fig. 6B). To determine if one of the four Cas proteins can be omitted, we next transformed the H26ΔcasCluster28 strain with the genes for 1) Cas5, Cas6, and Cas7; 2) Cas5, Cas6, and Cas8b; and 3) Cas6, Cas7, and Cas8b. The combination of Cas6, Cas7, and Cas8b was not able to stably maintain crRNAs, whereas the combination of Cas5, Cas6, and Cas8b resulted in some amounts of crRNA (Table 3 and Fig. 6B). Only the combination of Cas5, Cas6, and Cas7 yielded normal crRNA amounts. We also tested all dimeric combinations of the Cas proteins 5–8b, but none of these were active in crRNA protection (Fig. 6B) (Table 3). Taken together, these results indicate that the Cas6 protein is necessary for generating crRNAs but not sufficient for its maintenance. In vivo, at least Cas5 and Cas7 are additionally required for crRNA generation or stabilization.

A Cascade-like Complex in Haloferax

To confirm an association of the Cas5, Cas6, and Cas7 proteins, we expressed a FLAG-Cas7 fusion protein and purified this protein together with all potential interaction partners (Fig. 7). The SDS gel of the purified protein fraction shows co-purification of two additional proteins. Mass spectrometry analyses of the proteins showed that these proteins are Cas5 and Cas6. To further investigate the stoichiometry of Cas5, Cas6, and Cas7 in the complex by mass spectrometry, we used the iBAQ quantification approach (53, 57). The isolated complex was spiked into a mixture of quantified standard proteins (UPS2), which spanned a concentration range of 5 orders of magnitude, and the complete protein mixture was then digested in solution. The iBAQ values obtained for the Cas proteins were calibrated using a linear function derived from the iBAQ values for the standard proteins. The derived protein concentrations indicate a Cas5/Cas6/Cas7 stoichiometry of 1.7:1:8.5. This low/low/high type stoichiometry is generally in agreement with stoichiometries previously observed for Cascade-type protein complexes in E. coli (58) and P. aeruginosa (24). It should be noted that bottom-up mass spectrometric methods like iBAQ show limited accuracy for determining high stoichiometries in protein complexes because the ratio accuracy is largely limited by the accuracy of determining the higher protein copy numbers in complexes.

To investigate whether crRNAs are likewise part of the complex, we isolated RNA from the FLAG-Cas7 purified protein fraction and determined the nature of the RNA using a Northern blot (Fig. 7C). Hybridization with a probe against spacer 1 from CRISPR locus P1 showed that the RNA is indeed crRNA. In addition to the crRNA of about 68 nucleotides in length, a shorter crRNA of about 51 nucleotides is detected, suggesting a secondary processing event. These shorter crRNAs are also visible in other Northern analyses (Fig. 6) (42) but not as prominent as in the FLAG-Cas7 purified fraction. In summary, we could show that Cas5, Cas6, and Cas7, as well as crRNA, co-purify and that this Cascade-like complex has a similar composition as the E. coli Cascade complex.

The Cascade I-B Complex Requires Cas5, Cas6, and Cas7 for Maintaining a Stable crRNA Population

A Haloferax strain in which all cas genes, with the exception of cas6, have been removed is not able to stably maintain crRNAs. crRNAs can only be stably maintained if Cas5, Cas6, and Cas7 are present. This finding suggests the presence of a Cascade-like complex for crRNA processing and stabilization in the H. volcanii type I-B system, similar to the type I-A (Cascade I-A) and I-E (Cascade I-E) systems. The Cascade I-B complex clearly requires the Cas proteins 5–7 for maintaining a stable crRNA population. This hypothesis is confirmed by the co-purification of Cas5, Cas6, and crRNA molecules with a FLAG-tagged Cas7.

Interestingly, the combination of Cas proteins Cas5, Cas6, and Cas8b results in some amount of crRNAs (32%, Table 3). Thus, the crRNA is generated, but the amount of crRNA is not as high as it is in wild type strains. This result might suggest that the missing Cas7 protein is required for efficient stabilization of the crRNA. In addition, this observation suggests that Cas8b, the large subunit of the type I-B system (59), can protect the crRNA to some extent. The large subunit of the E. coli type I-E system (Cse1) was shown to be located close to the crRNA 5′-end, interacting with the PAM sequence of the target DNA (60, 61). Further, it was suggested that Cse1 interacts with two bases in the 5′-handle of the crRNA (62). Thus, the Haloferax Cas8b might protect the crRNA from degradation by binding to its 5′-end. Taken together, this suggests that the large subunits of the I-B and the I-E system have similar roles in binding the crRNA and PAM recognition.

A comparison of the Cascade complexes from the different type I systems (IA (22), IE (58, 63, 64), and IF (24, 63)) shows that they all contain multiple copies of the Cas7 protein (in type I-F, this is probably the Csy3 (59, 63)) and that the Cas7 protein forms the backbone of the complex (14). In addition, type I-A and I-E complexes contain the Cas5 protein that interacts with Cas7. The Cas6 protein is also associated with the type I systems either as an integral part (I-E and I-F) or transiently (I-A) (14). Furthermore, these complexes can contain a small subunit (Csa5 in I-A and Cse2 in I-E) and/or a large subunit (Cas8 in type I-A, Cse1 in type I-E, and Csy1 in type I-F) (14). The data presented here for the Haloferax type I-B Cascade complex fit very well to these general observations for the type I Cascade complexes. Cas5, Cas6, and Cas7 co-purify, showing that the Haloferax complex consists of these subunits. Furthermore, the three Cas proteins are present in almost the same ratio as in the E. coli Cascade complex. The crRNA obviously needs Cas5–7 proteins to be generated and to be stably maintained. Structural analysis will show whether the Cas7 protein likewise forms the backbone of the complex binding and protecting the crRNA and whether the Cas8b protein represents the large subunit of the complex.

In addition, these data clearly show that the Cas1–4 proteins are not involved in this reaction. This is in contrast to the results observed with the type I-B system in H. mediterranei, in which the deletion of the cas1, cas3, and cas4 genes resulted in lower levels of crRNA (12).