Author response: Intrinsic sequence specificity of the Cas1 integrase directs new spacer acquisition

Full text Figures and data Side by side Abstract eLife digest Introduction Results Discussion Materials and methods References Decision letter Author response Article and author information Metrics Abstract The adaptive prokaryotic immune system CRISPR-Cas provides RNA-mediated protection from invading genetic elements. The fundamental basis of the system is the ability to capture small pieces of foreign DNA for incorporation into the genome at the CRISPR locus, a process known as Adaptation, which is dependent on the Cas1 and Cas2 proteins. We demonstrate that Cas1 catalyses an efficient trans-esterification reaction on branched DNA substrates, which represents the reverse- or disintegration reaction. Cas1 from both Escherichia coli and Sulfolobus solfataricus display sequence specific activity, with a clear preference for the nucleotides flanking the integration site at the leader-repeat 1 boundary of the CRISPR locus. Cas2 is not required for this activity and does not influence the specificity. This suggests that the inherent sequence specificity of Cas1 is a major determinant of the adaptation process. https://doi.org/10.7554/eLife.08716.001 eLife digest In most animals, the adaptive immune system creates specialized cells that adapt to efficiently fight off any viruses or other pathogens that have invaded. Bacteria (and another group of single-celled organisms called archaea) also have an adaptive immune system, known as CRISPR-Cas, that combats viral invaders. This system is based on sections of the microbes' DNA called CRISPRs, which contain repetitive DNA sequences that are separated by short segments of ‘spacer’ DNA. When a virus invades the cell, some viral DNA is incorporated into the CRISPR as a spacer. This process is known as adaptation. CRISPR-associated proteins (or ‘Cas’ proteins) then use this spacer to recognize and mount an attack on any matching invader DNA that is later encountered. Exactly how a spacer is inserted into the correct position in the CRISPR array during adaptation remains poorly understood. However, it is known that two CRISPR proteins called Cas1 and Cas2 play essential roles in this process. Rollie et al. took Cas1 proteins from a bacterial cell (Escherichia coli) and an archaeal species (Sulfolobus solfataricus) and added them to branched DNA structures in the laboratory. These experiments revealed that Cas1 from both organisms can break the DNA down into smaller pieces. Cas2, on the other hand, is not required for this process. This ‘disintegration’ reaction is the reverse process of the ‘integration’ step of adaptation where the CRISPR proteins insert the invader DNA into the CRISPR array. Rollie et al. also found that the disintegration reaction performed by Cas1 takes place on specific DNA sequences, which are also the sites where Cas1 inserts the spacer DNA during adaptation. Therefore, by examining the disintegration reaction, many of the details of the integration step can be deduced. Overall, Rollie et al. show that selection by Cas1 plays an important role in restricting the adaptation process to particular DNA sites. The next step will be to use the disintegration reaction to examine the DNA binding and manipulation steps performed by Cas1 as part of its role in the adaptation of the CRISPR system. https://doi.org/10.7554/eLife.08716.002 Introduction The CRISPR-Cas system is an adaptive immune system present in many archaeal and bacterial species. It provides immunity against viruses and other mobile genetic elements mediated through sequence homology-directed detection and destruction of foreign nucleic acid species (reviewed in Sorek et al., 2013; Barrangou and Marraffini, 2014; van der Oost et al., 2014). Organisms with an active CRISPR-Cas system encode one or more CRISPR loci in their genomes. These comprise a leader sequence followed by an array of short, direct, often palindromic repeats interspersed by short ‘spacer’ sequences, together with a variable number of CRISPR associated (cas) genes. Spacers are derived from mobile genetic elements and constitute a ‘memory’ of past infections. The CRISPR locus is transcribed from the leader, generating pre-CRISPR RNA (pre-crRNA) that is subsequently processed into unit length crRNA species and loaded into large ribonucleoprotein complexes. These complexes, known as ‘Interference’, ‘Effector’ or ‘Surveillance’ complexes, utilize crRNA to detect and degrade cognate invading genetic entities, providing immunity against previously encountered viruses and plasmids. The process of foreign DNA capture and integration into the CRISPR locus is known as ‘Acquisition’ or ‘Adaptation’ (reviewed in Fineran and Charpentier, 2012; Heler et al., 2014). This process underpins the whole CRISPR-Cas system, but is the least well understood aspect. Fundamentally, acquisition can be separated into two steps: the capture of new DNA sequences from mobile genetic elements, followed by the integration of that DNA into the host genome. New spacers are incorporated proximal to the leader sequence and result in the duplication of the first repeat (Goren et al., 2012; Yosef et al., 2012). The leader sequence proximal to the repeat is important for integration, but transcription of the locus is not required (Yosef et al., 2012). New spacers are frequently incorporated in both possible orientations (Shmakov et al., 2014). The integration process in Escherichia coli results in staggered cleavage of the first CRISPR repeat, where the 3′-end of one strand of the incoming DNA is joined to the end of the CRISPR repeat in a trans-esterification (TES) reaction, with another TES reaction occurring on the other strand (Diez-Villasenor et al., 2013) (Figure 1, numbered yellow arrows). Intermediates in this reaction have been observed in E. coli, and the sequence of the leader-repeat1 junction is important for the activity (Arslan et al., 2014). The order of the two integration steps shown in Figure 1 is not yet known, although it has been suggested that the reaction on the minus strand (site 2) occurs first in E. coli (Nuñez et al., 2015). The sequence at site 2 is flanked by the end of the first repeat and the first spacer, and is therefore inherently less well conserved. In E. coli, the last nucleotide of the newly copied repeat is derived from the first nucleotide of the incoming spacer, which imposes further sequence requirements on that system (Swarts et al., 2012). Figure 1 Download asset Open asset CRISPR spacer acquisition and Cas1. (A, 1) The 3′-end of an incoming protospacer attacks the chromosomal CRISPR locus at the boundary between the leader sequence and repeat 1. A trans-esterification (TES) reaction (yellow arrow 1) catalyzed by Cas1 joins the protospacer to the 5′ end of repeat 1. For many integrases a (reverse) disintegration reaction can be observed in vitro. (2) Another TES reaction (yellow arrow 2) joins the other strand of the protospacer to the 5′ end of repeat 1 on the bottom (minus) strand, resulting in the formation of a gapped DNA duplex. (3) The gapped duplex is repaired by the host cell DNA replication machinery, resulting in the addition of a new spacer at position 1 and replication of CRISPR repeat 1. (B) Sequences flanking the two TES reaction sites at repeat 1 in Sulfolobus solfataricus and Escherichia coli are shown. The leader is in blue, repeat in black and spacer 1 in teal. The number of central nucleotides of the repeat omitted from the sequence is shown in parentheses. (C) Structure of Cas1 from Pyrococcus horikoshii (PDB 4WJ0) with subunits coloured blue and cyan, showing the dimeric ‘butterfly’ conformation with the active site residues highlighted in green. https://doi.org/10.7554/eLife.08716.003 Although shown in Figure 1 as fully double-stranded, the incoming spacer DNA could have a partially single-stranded character. Recent work by Sorek and colleagues has shown that the E. coli RecBCD helicase–nuclease complex, which processes DNA double-strand breaks for recombination and degrades foreign DNA, is implicated in the generation of viral DNA fragments captured by Cas1 and incorporated into the CRISPR locus as new spacers (Levy et al., 2015). This confirms previous observations linking Cas1 with RecBCD (Babu et al., 2011) and raises some intriguing questions as RecBCD generates ssDNA fragments asymmetrically, generating fragments tens to hundreds of nucleotides long from the 3′ terminated strand and much longer fragments from the 5′ terminated strand (reviewed in Dillingham and Kowalczykowski, 2008). The Cas4 enzyme, which is a 5′ to 3′ ssDNA exonuclease (Zhang et al., 2012; Lemak et al., 2013), may provide ssDNA fragments for Cas1 in systems lacking RecBCD. However, it is difficult to see how two integration reactions could occur without two 3′ hydroxyl termini (Figure 1) and half-site integration is not observed with a ssDNA substrate (Nuñez et al., 2015). Potentially, the ssDNA fragments generated by these nucleases may re-anneal and experience further processing to generate partially duplex molecules of defined size prior to integration by Cas1. Adaptation requires the products of the cas1 and cas2 genes in vivo and these are the most universally conserved of all the cas genes (Makarova et al., 2006). Cas1 is a homodimeric enzyme with a two-domain structure and a canonical metal dependent nuclease active site in the large domain formed by a cluster of highly conserved residues (Wiedenheft et al., 2009) (Figure 1C). E. coli Cas1 has nuclease activity in vitro, with activity observed against double- and single-stranded DNA and RNA substrates (Babu et al., 2011). Some specificity was observed for branched DNA substrates, in particular for DNA constructs resembling replication forks (Babu et al., 2011). Initial biochemical analyses of a panel of archaeal Cas2 enzymes revealed an endonucleolytic activity against ssRNA substrates that could be abrogated by mutation of conserved residues (Beloglazova et al., 2008). In contrast, Cas2 from Bacillus halodurans has been shown to be specific for cleavage of dsDNA substrates (Nam et al., 2012). Recently however, Doudna and colleagues demonstrated that E. coli Cas1 and Cas2 form a stable complex in vitro and that the ‘active site’ of Cas2 was not required for spacer acquisition, suggesting that Cas2 may not have a catalytic role in spacer acquisition in vivo (Nuñez et al., 2014). It is probable that Cas2 acts as an adaptor protein, either bringing two Cas1 dimers together or mediating interactions with other components necessary for spacer acquisition. Recently, Nunez et al. demonstrated that E. coli Cas1 can integrate a protospacer into a supercoiled plasmid in vitro, in a reaction stimulated by Cas2. Integration events were observed at the boundaries of most CRISPR repeats and in other areas of the DNA close to palindromic regions, implicating a role for palindromic DNA structure in the adaptation process (Nuñez et al., 2015). In order to further mechanistic understanding of the spacer acquisition process, we have undertaken a systematic analysis of the underlying biochemistry. We demonstrate that Cas1 catalyses TES of branched DNA substrates efficiently in vitro in a reaction that represents the reverse- or disintegration of an incoming spacer from the CRISPR locus. The disintegration reactions catalysed by diverse integrases have proven a powerful model system for the understanding of the mechanism of integration. For Cas1, the reaction is strongly sequence dependent with the specificity matching the predicted integration site 1 for both E. coli and Sulfolobus solfataricus Cas1, and does not require Cas2. Results Cas1 catalyzes a TES reaction on branched DNA substrates The Cas1 and Cas2 proteins from S. solfataricus (CRISPR-Cas subtype IA) and E. coli (CRISPR-Cas subtype IE) were expressed in E. coli with N-terminal polyhistidine tags and purified to homogeneity by metal affinity and gel filtration chromatography. Previously, it was demonstrated that E. coli Cas1 (EcoCas1) cleaved branched DNA substrates preferentially (Babu et al., 2011). We investigated the activity of S. solfataricus Cas1 (SsoCas1) against a DNA substrate with a 5′-flap structure (Figure 2). By labelling the single-stranded 5′-flap of the downstream duplex at the 5′-end with radioactive phosphorus, we observed cleavage of the flap by SsoCas1, releasing an 18 nt product (Figure 2B). A variant of SsoCas1 where the active site metal ligand glutamic acid 142 was changed to an alanine (E142A) was inactive, implicating the canonical nuclease active site of Cas1 in the reaction. This result appeared consistent with the earlier studies for EcoCas1 (Babu et al., 2011) and suggested that the ssDNA flap was removed at or close to the branch point. The activity was independent of the presence or absence of SsoCas2, suggesting that Cas2 is not involved in this nuclease activity in vitro. Figure 2 Download asset Open asset Disintegration of a branched DNA substrate by SsoCas1. Denaturing gel electrophoresis was used to analyse the products generated by SsoCas1 with a branched DNA substrate (Substrate 1). The 5′ flap (18 nt) was released when the phosphodiester backbone was attacked by the 3′-hydroxyl group at the branch point. The reaction required active Cas1 and was independent of Cas2. DNA lengths are shown in blue (nt). The TES site is indicated with a yellow arrow and the labelled strand with a red star. (A) Shows reactions with the continuous strand (black) labelled; (B) with the flap (grey) strand labelled and (C) with the upstream (green) strand labelled, each on the 5′ end. Lanes: 1, control with no added protein; 2, WT Cas1; 3, Cas2; 4, Cas1 + Cas2; 5, Cas1 E142A variant; 6, E142A Cas1 + Cas2. (D) The 5′-flap strand was labelled on the 3′ end with a fluorescein moiety, and the flap reduced to 14 nt (Substrate 1-FAM). Cas1 catalyses the TES reaction generating a 53 nt labelled strand. Lane C: control incubation in absence of Cas1. (E) TES reactions were carried out with SsoCas1, or the E142A active site mutant, on a fork substrate containing a nicked SacI restriction site spanning the branch point (SacI substrate). A TES product of 53 nucleotides is visible in lane 2 containing Cas1. The right-hand panel shows the effect of adding SacI restriction enzyme after the TES reaction. The TES product is no longer visible, but a shorter product of 25 nucleotides is present indicating regeneration of the SacI recognition sequence by the TES reaction. https://doi.org/10.7554/eLife.08716.004 We proceeded to label the other strands of the substrate to follow the reaction products in more detail. The continuous (black) strand was not cleaved by SsoCas1 (Figure 2A). However, when the 23 nt (green) strand of the upstream duplex presenting a 3′-hydroxyl end at the junction point was labelled we observed the formation of a new, larger DNA species (Figure 2C). This observation was consistent with a joining or TES of the upstream DNA to the downstream DNA strand by Cas1. By switching to a label at the 3′ end of the downstream duplex we confirmed that the reaction catalyzed by Cas1 involved the joining of the upstream and downstream DNA duplexes with concomitant loss of the 5′-flap (Figure 2D). The lack of evidence for any shorter DNA species in Figure 2D was consistent with the conclusion that we were observing a TES rather than a nuclease reaction. Again, the TES reaction was unaffected by the presence or absence of Cas2 and was dependent on the wild-type active site of Cas1, as the E142A variant was inactive. In order to define the product of the TES reaction in more detail, we modified the sequence of the branch point to introduce an interrupted site for the restriction enzyme SacI across the junction. A precise TES reaction at the branch point to generate duplex DNA would result in the ‘repair’ of the restriction site, generating a substrate for SacI. SsoCas1 processed this substrate generating the 53 bp TES reaction product. On addition of the SacI restriction enzyme, the 53 bp species was converted to a 25 bp product (Figure 2E). This confirmed that the Cas1-mediated reaction resulted in the formation of a functional SacI site in vitro, consistent with a precise TES reaction at the branch point. It is likely that the TES reaction catalyzed by Cas1 with branched DNA substrates in vitro represents the reverse or disintegration of an integration intermediate, as observed recently for EcoCas1 (Nuñez et al., 2015). We therefore tested EcoCas1 with the same set of branched substrates, showing that manganese, magnesium and cobalt all supported the same robust disintegration activity in the absence of Cas2, with no nuclease activity observed (Figure 3A). Given that Eco and SsoCas1 are divergent members of the Cas1 family, this suggests that the disintegration activity is likely to be a general property of Cas1 enzymes. Experiments where the concentration of Cas1 was titrated against a fixed concentration of substrate (50 nM), showed that maximal activity plateaued above 250 nM for EcoCas1 and was maximal from 100 to 500 nM for SsoCas1 (Figure 3B,C). Figure 3 Download asset Open asset TES activity of E. coli and S. solfataricus Cas1. (A) E. coli Cas1 also catalyses an efficient metal dependent disintegration reaction. TES reactions were carried out under standard conditions, using Substrate 3 and varying the divalent metal ion as indicated. EcoCas1 showed robust TES activity in the presence of cobalt, magnesium and manganese. Each of the three strands of the substrate was labelled individually as for Figure 2 (5′ label indicated by a star). Lanes were: c, substrate alone; substrate incubated with Cas1 and 5 mM of E, EDTA; Co, cobalt chloride; Ca, calcium chloride; Mg, magnesium chloride; Mn, manganese chloride for 30 min at 37°C. (B) Concentration dependence of Cas1 TES activity. Substrate 3 (50 nM) was incubated with the indicated concentration of Sso or EcoCas1 for 30 min under standard assay conditions and the reactants were analysed by denaturing gel electrophoresis and phosphorimaging. SsoCas1 showed maximal activity at 250 nM, representing a fivefold molar excess of enzyme over substrate, with a decline in activity above 500 nM enzyme. EcoCas1 had maximal activity that plateaued above 250 nM enzyme. (C) Quantification of the data (raw data provided in Figure 3—source data 1). These data are representative of duplicate experiments. https://doi.org/10.7554/eLife.08716.005 Figure 3—source data 1 Concentration Cas1. https://doi.org/10.7554/eLife.08716.006 Download elife-08716-fig3-data1-v2.xlsx To characterise the specificity of the disintegration reaction in more detail, we examined SsoCas1 activity for a variety of substrates differing in the nature of the 5′-flap or duplex arm released by the reaction (Figure 4). SsoCas1 was indifferent to the presence of duplex or single-stranded DNA in the 5′-flap, processing a nicked 3-way junction with a similar efficiency to the 5′-flap substrate. The disintegration reaction required the presence of the 3′-hydroxyl moiety at the branch point as a three-way (or Y) junction was not a substrate for Cas1. To confirm this observation we studied a 5′-flap substrate with a phosphate moiety at the 3′ end of the upstream strand in place of a hydroxyl group. As expected, this substrate did not support disintegration activity for either Sso or EcoCas1, with no larger TES product detected (right hand lanes). Tellingly, neither enzyme cleaved the 5′-flap of this substrate to generate shorter products (left hand lanes), confirming that Cas1 catalyses a concerted TES reaction rather than a sequential ‘cut and join’ activity. Figure 4 Download asset Open asset Importance of flap and 3′ terminus structure. The importance of the released 25 nt 5′-flap structure was investigated by varying the length of duplex DNA in that arm from 0 (canonical 5′-flap) to a full 25 bp (nicked 3-way junction) (left hand panel, all based on substrate 19). All supported robust disintegration activity by SsoCas1. An intact Y- junction did not support TES activity. Lanes: C, substrate alone (1, 5, 9, 13, 17); E, SsoCas1 E142A variant 30 min incubation (2, 6, 10, 14, 18); incubation with wild-type SsoCas1 for 10 and 30 min (other lanes). The right hand panel shows the effect of replacing the attacking 3′-hydroxyl moiety at the branch point with a phosphate group (3′ phos substrate) no TES or nuclease activity was observed for either Sso or EcoCas1. C, substrate alone; E, SsoCas1 E142A variant. https://doi.org/10.7554/eLife.08716.007 Sequence specificity of the disintegration reaction Disintegration reactions are commonly seen for integrases and transposases, and represent a very useful means to study the underlying integration mechanism (Chow et al., 1992; Delelis et al., 2008) as the local arrangement of the DNA in the integrase active site is typically the same for the two reactions (Gerton et al., 1999). One prediction of this hypothesis is that the disintegration reaction could demonstrate some sequence preference if integration, which must happen at a specific, defined site in the CRISPR locus, is partly driven by the sequence specificity of Cas1. We therefore designed a family of related disintegration substrates by systematically varying the nucleotides flanking the TES site and tested how efficiently Cas1 could disintegrate these substrates. Having demonstrated conclusively that we could follow the progress of the disintegration reaction by monitoring the liberation of a displaced DNA strand from a 5′-flap substrate, we used this assay for all subsequent investigations. The +1 position We first tested the importance of the nucleotide acting as an acceptor for the attacking 3′-hydroxyl of the incoming DNA strand (the +1 position) by varying the nucleotide at this position in the model substrates (Figure 5). In vivo, this acceptor nucleotide should represent the position in the CRISPR locus where new spacers are joined to the end of the repeat. The S. solfataricus CRISPR repeat has a guanine at one end and a cytosine at the other, each of which are predicted to act as acceptors for a new phosphodiester bond formed with the incoming spacer (Figure 1B). Consistent with this, we observed the most efficient disintegration activity with SsoCas1 when substrates had a guanine at the +1 position (Figure 5A). Assays were carried out in triplicate and the reaction products quantified, confirming the qualitative observation of a preference for guanine, followed by cytosine, adenine and thymine, with rates of 0.06, 0.013, 0.0058 and 0.0009 min−1, respectively (Figure 5B). Figure 5 Download asset Open asset Sequence specificity of the disintegration reaction at the +1 position. The nucleotide at the acceptor (+1) position was varied systematically to assess the sequence dependence of the disintegration reaction carried out by Cas1 from S. solfataricus (A, B) and E. coli (C, D) (Substrates 3, 6, 7, 8). In the gels on the left (A, C) each substrate was incubated with Cas1 for 1, 2, 3, 5, 10, 15, 20 and 30 min in reaction buffer prior to electrophoresis to separate the cleaved 5′-flap from the intact substrate. C–control with no Cas1 added. The plots on the right (B, D) show quantification of these assays. Data points represent the means of triplicate experiments with standard errors shown (raw data provided in Figure 5—source data 1 and Figure 5—​source data 2). The data were fitted to an exponential equation, as described in the ‘Materials and methods’, and for EcoCas1 a variable floating end point was included to allow fitting as the reaction did not go to completion. The effect of Cas2 (150 nM) on EcoCas1 (150 nM) sequence specificity for substrates (50 nM) varying at position +1 (substrates 3, 6, 7, 8) was also tested (E). The second panel from the right is a composite image from two phosphorimages of the same time course as indicated by a black line. https://doi.org/10.7554/eLife.08716.008 Figure 5—source data 1 Nucleotide at +1 position. https://doi.org/10.7554/eLife.08716.009 Download elife-08716-fig5-data1-v2.xlsx Figure 5—source data 2 Nucleotide at +1 position. https://doi.org/10.7554/eLife.08716.010 Download elife-08716-fig5-data2-v2.xlsx For E. coli, the first nucleotide of the repeat is a guanine. Although the corresponding position at the other end of the repeat on the minus strand is a cytosine, it has been demonstrated that the new spacer joins at the penultimate residue, which is also a guanine (Swarts et al., 2012). EcoCas1 displayed a striking preference for a guanine at the +1 position for the disintegration reaction, with all three alternative nucleotides strongly disfavoured at this position (Figure 5C), in good agreement with the prediction based on the repeat sequence. For EcoCas1 the reaction did not go to completion and accordingly we fitted the data with a variable end-point as described in the ‘Materials and methods’ (Figure 5D). Although the reaction rates could not be determined accurately, guanine at position +1 supported rates at least 10-fold higher than any other nucleotide. We also tested the effect of inclusion of Cas2 on the sequence specificity of EcoCas1, and observed that Cas2 had no effect, with strong preference for a guanine at +1 still observed (Figure 5E). The −1 position We proceeded to investigate the sequence dependence of the nucleotide at the 3′-end of the attacking DNA strand (the −1 position) in the disintegration reactions. For the first integration site, this should correspond to the last nucleotide of the leader sequence, which is an adenine in both S. solfataricus and E. coli. Although both Cas1 enzymes catalyzed the disintegration of substrates with an adenine at this position, clear sequence preference was not apparent (Figure 6). For S. solfataricus, the −1′ position on the minus strand is variable in sequence. However, in E. coli, site 2 occurs at the penultimate nucleotide of the repeat and therefore has a cytosine at the −1′ position (Swarts et al., 2012). In this situation, the incoming 3′ terminal nucleotide of the spacer has to be a cytosine in order to complete the original repeat sequence. To mimic the TES substrate at this site more closely, we tested substrates where the first nucleotide of the 5′ flap, equivalent to the incoming nucleotide in the forward reaction, was a cytosine, but a cytosine at position −1 was still not favoured by EcoCas1 (Figure 6C). This may suggest that the disintegration reaction is not a good model for integration at site 2, which is further discussed later. Figure 6 Download asset Open asset Sequence specificity of the disintegration reaction at the −1 position. The nucleotides participating in the disintegration reaction were varied systematically at the −1 position (substrates 3, 9, 10, 11). For SsoCas1 (A) there was some preference for adenine at this position, consistent with integration site 1. For EcoCas1 (B, C), a cytosine at position −1 was disfavoured over all other possibilities, even when the residue equivalent to the ‘incoming’ nucleotide was also a cytosine (substrates 15, 16, 17, 18). Each substrate was incubated with Cas1 for 5, 10 and 30 min in reaction buffer prior to electrophoresis. C–control with no Cas1 added. https://doi.org/10.7554/eLife.08716.011 The −2 position The nucleotide at the −2 position, which is part of the conserved leader sequence for integration site 1, is also a potential determinant of integration specificity. Accordingly, we varied this residue systematically and investigated the efficiency of the disintegration reaction for both Cas1 enzymes (Figure 7). In S. solfataricus, the −2 position in the leader is a cytosine, which supported the strongest disintegration activity (Figure 7A). In E. coli, the −2 position in the leader is a guanine. A clear preference for guanine over all other nucleotides was observed for EcoCas1 (Figure 7B), confirmed by a more detailed kinetic analysis (Figure 7C) which was fitted as for Figure 5D. These data are consistent with a role for sequence discrimination at the −2 position by both enzymes, which is relevant for integration site 1 but not site 2, where this position varies depending on the sequence of the last spacer inserted. Figure 7 Download asset Open asset Sequence specificity of the disintegration reaction at the −2 position. The nucleotides participating in the disintegration reaction were varied systematically at the −2 position, which is a cytosine (Sso) or guanine (Eco) at integration site 1, and variable at integration site 2 (substrates 10, 12, 13, 14). (A) SsoCas1; (B) EcoCas1. Each substrate was incubated with Cas1 for 5, 10 and 30 min in reaction buffer prior to electrophoresis. C–control with no Cas1 added. (C) For EcoCas1, the clear preference for guanine at position −2 was confirmed by more detailed kinetic analysis (raw data provided in Figure 7—source data 1) as described for Figure 5. https://doi.org/10.7554/eLife.08716.012 Figure 7—source data 1 Nucleotide at −2 position. https://doi.org/10.7554/eLife.08716.013 Download elife-08716-fig7-data1-v2.xlsx