|Home | About | Journals | Submit | Contact Us | Français|
The CRISPR arrays found in many bacteria and most archaea are transcribed into a long precursor RNA that is processed into small clustered regularly interspaced short palindromic repeats (CRISPR) RNAs (crRNAs). These RNA molecules can contain fragments of viral genomes and mediate, together with a set of CRISPR-associated (Cas) proteins, the prokaryotic immunity against viral attacks. CRISPR/Cas systems are diverse and the Cas6 enzymes that process crRNAs vary between different subtypes. We analysed CRISPR/Cas subtype I-B and present the identification of novel Cas6 enzymes from the bacterial and archaeal model organisms Clostridium thermocellum and Methanococcus maripaludis C5. Methanococcus maripaludis Cas6b in vitro activity and specificity was determined. Two complementary catalytic histidine residues were identified. RNA-Seq analyses revealed in vivo crRNA processing sites, crRNA abundance and orientation of CRISPR transcription within these two organisms. Individual spacer sequences were identified with strong effects on transcription and processing patterns of a CRISPR cluster. These effects will need to be considered for the application of CRISPR clusters that are designed to produce synthetic crRNAs.
Clustered regularly interspaced short palindromic repeats (CRISPR) and CRISPR-associated (cas) genes define an anti-viral defence system in Archaea and Bacteria. CRISPR loci are composed of repeat sequences with an average length of 24–47nt, which alternate with unique spacer sequences derived from previous encounters with foreign nucleic acids (i.e. viruses, plasmids) (1–4). CRISPR loci are transcribed and processed to generate the small interfering crRNAs. Diverse sets of cas genes are often found adjacent to a CRISPR locus and encode proteins that are involved in the three phases of CRISPR/Cas activity: acquisition of new spacers, processing of crRNAs and interference with foreign nucleic acid (5–9). Although there is little information available for the process of new spacer acquisition, recent progress has led to a better understanding of the other two phases. The maturation of precursor crRNA into small crRNAs is performed by diverse Cas endonucleases that belong to a protein family termed Cas6 (10–16). In CRISPR/Cas Type-I the interference step is mediated by a complex of different Cas proteins (Cas complex for antiviral defence: Cascade) bound to crRNAs that target the invading nucleic acid through base complementarity which ultimately results in the inactivation or degradation of foreign DNA by Cas3 (17–24). Type-II CRISPR/Cas systems use the single Cas9 protein for interference (25) and Type-III systems use a multi Cas protein complex that is distinct from Cascade (26,27).
Computational analyses of these defence systems identified a surprising diversity of different CRISPR/Cas types and subtypes, which are spread throughout archaeal and bacterial kingdoms.
This classification has defined three major types which can be further divided into at least 10 CRISPR/Cas subtypes (28). The subtype I-B, found, e.g. in Clostridia, methanogens and halophiles, is defined by the subtype-specific protein Cas8b. In Clostridium thermocellum and Methanococcus maripaludis the minimal subtype I-B Cas protein organization consists of the universal Cas1, Cas2 and Cas4 proteins that are proposed to mediate the integration of spacers as well as Cas3, Cas5, Cas7 and Cas8b which are proposed to form the Cascade complex of this subtype. Finally, a Cas6 protein is required for the processing of crRNA (10–16).
A Cas6 protein was first described for CRISPR/Cas subtype III-B in Pyrococcus furiosus as a metal-independent endonuclease involved in the processing of precursor crRNA into mature crRNA (10,11,14,15). Cas6 enzymes were also characterized for CRISPR/Cas subtype I-F in Pseudomonas aeruginosa (Cas6f, also termed Csy4) (13) and CRISPR/Cas subtype I-E in Thermus thermophilus and Escherichia coli (Cas6e, also termed Cse3) (12,16). The amino acid sequence similarity of these Cas6 proteins is limited, yet they share ferredoxin-like folds and perform analogous reactions in the different CRISPR/Cas systems. These Cas6 proteins do not only differ in substrate specificity, but also in the composition of their active sites. For example P. furiosus Cas6 (Pf Cas6) interacts with single-stranded RNA while Cas6e and Cas6f seem to specifically bind to hairpin structures formed by the repeats (10–16). Further differences can be found in the catalytic site of the Cas6 proteins. Pf Cas6 uses a catalytic triad composed of tyrosine, histidine and lysine residues (10,14), while in Cas6f a catalytic dyad of a histidine and a serine residue proved to be important for protein activity (13,29). Activity of Cas6e relies on a tyrosine and a histidine residue (12,16). Although there are variations in their active site composition and the recognition of RNA substrates, the different Cas6 cleavage reactions always generate crRNAs that consist of a spacer unit that is flanked by 8nt of the repeat sequence as a 5′-terminal tag and a 3′-terminal repeat tag (11–13). Finally, Cas6 was shown to deliver the mature crRNA to the Cascade complex (18,30).
In this study, we provide the first analysis of crRNA processing for CRISPR/Cas subtype I-B for one bacterial model organism, C. thermocellum and one archaeal model organism, M. maripaludis (detailed information of CRISPR loci and gene organization can be found in Supplementary Figure S1). The abundance and processing of crRNAs were analysed in vivo by RNA-Seq methodology. In addition, the Cas6 enzymes of this CRISPR/Cas subtype (termed Cas6b) were identified and M. maripaludis Cas6b (Mm Cas6b) was analysed for crRNA processing in vitro.
Methanococcus maripaludis C5 cells were a kind gift of W.B. Whitman (Georgia). Clostridium thermocellum (DSM1237) cells were obtained from DSMZ (German collection of micro-organisms and cell cultures). All E. coli cells were grown in LB-media with appropriate antibiotics at 37°C and shaking at 200rpm.
Methanococcus maripaludis C5 was grown at 37°C in complex medium for methanococci (McC) (31) with H2/CO2 atmosphere (80%/20%) and one bar (15psi) overpressure. Clostridium thermocellum cells were incubated in complex medium (32) at 60°C with an anaerobic atmosphere (N2).
The cas6 genes MmarC5_0767, Cthe_3205 and Cthe_2303 were amplified from genomic DNA of M. maripaludis C5 or C. thermocellum ATCC 27405 and cloned into the vector pET-20b to facilitate protein expression with a C-terminal His-tag. Oligonucleotides for site-directed mutagenesis were designed using Agilents QuickChange Primer Design tool and cas6 mutants were created using the QuickChange site-directed mutagenesis (Stratagene) according to the manufacturer’s instructions. Mutations were confirmed by sequencing (MWG Eurofins).
All Cas6 variants were produced in E. coli (Rosetta2 DE3) cells. Induction of protein expression was performed by addition of isopropylthio-β-d-galactoside (IPTG) to a final concentration of 0.5mM after growing the cells to an OD578 of 0.6. Four hours after induction the cells were harvested, the pelleted cells re-suspended in lysis buffer (10mM Tris–HCl [pH8.0], 300mM NaCl, 10% glycerol and 0.5mM DTT) and incubated on ice with lysozyme (1mg/g cell pellet) for 30min. Cell disruption was performed using sonication (8×30s; Branson Sonifier 250). Clearing of the lysate was achieved by centrifugation (20000rpm, 30min, 4°C) and the supernatant was applied to a Ni–NTA–Sepharose Column (GE-Healthcare) and purified using a FPLC Äkta-Purification system (GE-Healthcare). Elution of the proteins was performed by a linear imidazole gradient (0–500mM). Purity of the proteins was determined by sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS–PAGE) and Coomassie Blue staining. The protein was dialysed into lysis buffer and the protein concentration was determined by Bradford Assay (BioRad).
The spacer2–repeat–spacer3 and repeat–spacer27–repeat RNA substrates were generated by in vitro run-off transcription using T7 RNA polymerase and internally labelled using [α-32P] adenosine triphosphate (ATP) (5000ci/mmol, Hartman Analytic) (33). The repeat RNAs and repeat RNAs with a substitution of the first unprocessed nucleotide against a dexoy nucleotide were synthesized by Eurofins MWG Operon. End labelling of these substrates was performed using T4 polynucleotide kinase (Ambion) and [γ-32P] ATP (5000ci/mmol) according to the manufacturer’s instructions.
Templates for in vitro transcription were obtained by cloning of the pre-crRNA sequences with an upstream T7 RNA polymerase promotor sequence into pUC19 vector. After linearization of the plasmid with HindIII, in vitro transcription was performed in a final volume of 20μl [40mM HEPES–KOH (pH8.0); 22mM MgCl2; 5mM DTT; 1mM spermidine; 4mM UTP, CTP, GTP and 2mM ATP; 20U RNase Inhibitor; 1µg T7 RNA polymerase; 1μg linearized plasmid] at 37°C for 1h. End labelling of synthesized RNA was done in a 20µl reaction volume: 10µl of the RNA was labelled using 2µl T4 Polynucleotide Kinase (PNK) buffer (New England Biolabs (NEB)) and 25 U T4 PNK (Ambion) at 37°C for 30min.
The RNAs were separated by denaturing PAGE (8M urea; 1× TBE; 10% polyacrylamide), and afterwards respective bands were cut out using sterile scalpels in reference to brief autoradiographic exposure. The RNA was eluted from the gel piece using 500μl RNA elution buffer [250mM NaOAc, 20mM Tris–HCl (pH 7.5), 1mM ethylenediaminetetraacetic acid (EDTA) (pH8.0), 0.25% SDS] and overnight incubation on ice. Precipitation of RNA was performed by adding two volumes EtOH (100%; ice cold) and 1/100 glycogen for 1h at −20°C and subsequent washing with 70% EtOH of pelleted RNA.
Different indicated concentrations of purified Cas6 enzyme were incubated with radio labelled RNA substrates and buffer [250mM KCl, 1.875mM MgCl2, 1mM DTT, 20mM HEPES–KOH (pH 8.0)]. The reaction mix was incubated for 10min at 37°C and then immediately mixed with 2× formamide buffer [95% formamide; 5mM EDTA (pH 8.0), 2.5mg bromophenol blue, 2.5mg xylene cyanol) and incubated at 95°C for 5min to stop the cleavage reaction. The reaction was applied to a denaturing 12–15% polyacrylamide gel running in 1× TBE with 12W for 1.5h. Visualization was achieved by phosphorimaging.
RNA and DNA were extracted from cell lysates with phenol/chloroform (1:1; phenol pH5 for RNA and pH8 for DNA) (34). A Proteinase K and 55°C heat shock treatment preceded the phenol/chloroform step. Small RNAs (<200nt) were purified from total RNA using the mirVana RNA extraction kit (Ambion). Three micrograms of isolated small RNA from either M. maripaludis C5 or C. thermocellum were treated with T4 PNK to ensure proper termini for ligation. A protocol for the dephosphorylation of 2′-, 3′-cyclic phosphate termini was modified from (35): 1μg of RNA was incubated at 37°C for 6h with 10U T4 PNK and 10μl 5× T4PNK buffer (NEB) in a total volume of 50μl. Subsequently, 1mM ATP was added and the reaction mixture was incubated for 1h at 37°C to generate monophosphorylated 5′-termini. RNA libraries were prepared with an Illumina TruSeq RNA Sample Prep Kit and sequencing on an Illumina HiSeq2000 sequencer was performed at the Max-Planck Genomecentre Cologne.
Sequencing reads were trimmed [(i) removal of Illumina TruSeq linkers and poly-A tails and (ii) removal of sequences using a quality score limit of 0.05] and mapped to the reference genomes (GenBank: CP000568 and CP000609) with CLC Genomics Workbench 5.0 (CLC Bio, Aarhus, Denmark). The following mapping parameters were used (mismatch cost: 2, insertion cost: 3, deletion cost: 3, length fraction: 0.5, similarity: 0.8). Reads <15nt were removed. Initial crRNA identification was obtained from crisprdb (36) and gene annotations were obtained from Genbank.
A model of the Mm Cas6b (MmarC5_0767) protein structure was built with the I-TASSER platform (37). The program identified P. furiosus Cas6 (pdb ID 3PKM) as the top template for structure prediction. The protein model was compared with the Pf Cas6 crystal structure using the program DaliLite (38) and their alignment revealed two homologous structures (Z-score 19.7, RMSD 2.5Å). Cas6b sequences were aligned with ClustalW2 (39).
The processing of crRNAs of the CRISPR/Cas subtype I-B was analysed by RNA-Seq for M. maripaludis C5 and for C. thermocellum ATCC 27405. The isolated total small RNAs were modified with T4 polynucleotide kinase to allow proper adapter ligation and were sequenced through Illumina HiSeq2000 RNA-Seq methodology. Over 14 million individual sequence reads were mapped to the corresponding reference genomes and elucidated the abundance and processing patterns of the CRISPR arrays of these two organisms. M. maripaludis C5 possesses a single CRISPR cluster with 28 repeats of 37-nt length that are interspersed by 27 unique spacers. The CRISPR region is constitutively transcribed and processed into small crRNAs (Figure 1). The crRNAs contain a clearly defined 5′-terminal 8-nt tag with the sequence 5′-AUUGAAAC-3′. The 3′-termini are gradually shortened and most often contain a minimal 2-nt tag with the repeat nucleotides 5′-CU-3′. The abundance of crRNAs declines gradually from the leader proximal to the leader distant region with the crRNA containing the highly AT-rich (30A or T out of 34nt) spacer 3 being underrepresented.
RNA-Seq analysis of the small RNAs of C. thermocellum revealed five constitutively transcribed and processed CRISPR clusters. Two of these CRISPR/Cas subtype I-B systems are very similar to the one found in M. maripaludis and contain 37-nt repeat elements. The other three CRISPR clusters have 30-nt repeat sequences. Processing of both C. thermocellum CRISPR repeat sequences into mature crRNAs yields the same 5′-terminal 8-nt (5′- AUUGAAAC-3′) tag that is also found for M. maripaludis crRNAs (Figure 2 and Supplementary Figure S2). The 3′-termini are trimmed leaving mostly short tags. The abundance of crRNAs follows the pattern found in M. maripaludis and described for other CRISPR/Cas subtypes with one notable exception. The CRISPR locus 3 contains an internal signal to promote crRNA transcription within the CRISPR array (Figure 2A and Supplementary Table S1). The overall crRNA abundance gradually declines from Spacer 1 to Spacer 103 before crRNA production peaks again starting with the crRNA containing spacer number 104. Interestingly, the 8-nt repeat tags are not identical for the crRNAs from this CRISPR locus 3 as at Spacer 116 the final U base of the 5′-terminal tag is changed to the commonly found base C (Figure 2A and Supplementary Table S1). Close analysis of this sudden spike of internal crRNA abundance revealed a transcription start site at the A residue at Position 29 within Spacer 103. Our data suggest that this spacer is sufficient to promote transcription within the CRISPR region and that the 28-nt upstream of the transcription start within the spacer provide the necessary promoter elements in the context of the flanking repeats. Although it is difficult to pinpoint the pribnow box, the extreme AT-richness of the spacer (26 out of 28nt upstream of the transcription start site are A and T residues) suggests relaxed strand separation. DNA sequencing of the genomic region upstream of Spacer 104 excluded errors in the initial genome assembly during whole-genome sequencing.
In addition to internal promotion, we observed several cases of bidirectional transcript production for the CRISPR arrays. Anti-crRNA transcripts can start at the region opposite of the leader (CRISPR loci 1,2,5) or internally (CRISPR locus 4) (Figure 2B and Supplementary Figure S2). Although the amount of these anti-crRNA transcripts is usually very small in comparison to the abundance of crRNAs, individual anti-crRNAs show a conserved processing pattern within the repeats that yields RNAs with complete reverse complementary spacer sequences. These anti-crRNAs usually contain 18-nt 5′-tags and 15-nt 3′-tags for CRISPR loci 1 and 2 and 22-nt 5′-tags for CRISPR loci 4 and 5 (Supplementary Figure S3). The presence of processed anti-crRNAs can correlate with the reduced abundance of the respective sense crRNA (Figure 2B and Supplementary Figure S2).
To identify the enzyme that generates crRNAs for CRISPR/Cas subtype I-B, we analysed the cas genes of M. maripaludis and C. thermocellum. A set of only eight cas genes was identified in the genome of M. maripaludis C5. One of these potential cas gene products (MmarC5_0767, Mm Cas6b) showed 12% amino acid identity to Pf Cas6 which identified it as a Cas6b candidate for CRISPR/Cas subtype I-B. As the protein shares limited sequence identity with Cas6 proteins of other CRISPR/Cas subtypes (11–13), the structure of Mm Cas6b was modelled with I-Tasser (40). Pf Cas6 was identified as the closest structural homologue and shares a very similar overall architecture [Dali-Lite Z-score 19.7, RMSD 2.5Å, (38)] (Figure 3). The structural alignment of the Mm Cas6b model and Pf Cas6 also reveals a conserved histidine residue of Mm Cas6b in close proximity to the catalytic histidine of Pf Cas6. The comparison of different Cas6b homologues of CRISPR subtype I-B (Figure 3) indicates high sequence similarity and conserved residues. Clostridium thermocellum contains one Cas6b homologue (Cthe_3205) associated with the 37-nt repeat sequences and a potential second Cas6 enzyme (Cthe_2303) associated with the 30-nt repeat sequences. Classification of Cthe_2303 is not unambiguously possible as the neighbouring cas genes do not clearly fit into the commonly used 10 CRISPR/Cas subtypes.
The genes for Cas6 enzymes from M. mariplaudis and C. thermocellum were cloned and the recombinant enzymes were produced. Clostridium thermocellum Cas6b (Cthe_3205) production yielded only insoluble protein, but the two other Cas6 proteins were obtained in soluble form and allowed the analysis of their involvement in crRNA processing. Size exclusion separation of Mm Cas6b revealed a monomeric structure of the protein. The purified Cas6 candidates (Mm Cas6b and Cthe_2303) showed in vitro endonuclease activity and processed repeat RNA sequences from M. maripaludis and C. thermocellum, respectively (Figure 4). Therefore, the unclassified Cthe_2303 provided a good control for crRNA processing specificity of Mm Cas6b, as both enzymes specifically recognized only the repeats from their associated CRISPR clusters. Addition of bivalent metal ions did not influence the activity of Mm Cas6b. To define the cleavage site, a repeat RNA was synthesized with a deoxy nucleotide substitution at the proposed processing position that generates the 8-nt crRNA 5′-tag observed in vivo. This substitution resulted in the loss of Mm Cas6b and Cthe_2303 processing.
To test the influence of different RNA substrates (Supplementary Table S2) for Mm Cas6b activity, cleavage assays were performed with (i) repeat RNA, (ii) repeat–spacer–repeat RNA and (iii) spacer–repeat–spacer RNA using repeat and spacer sequences of the M. maripaludis CRISPR. For all three substrates, product formation was observed that corresponds with Mm Cas6b processing at the cleavage site determined by the deoxy nucleotide substitution within the repeat (Figure 4). This cleavage site determines that the conversion of the repeat (37nt) results in a 29-nt fragment, while the repeat–spacer27–repeat structure (131nt) is processed into three fragments (74, 38 and 19nt) and the spacer2–repeat–spacer3 substrate (110nt) is cleaved into two fragments (67 and 43nt) (Figure 5). Since all used substrates were cleaved in similar efficiency, the repeat RNA was used for further analysis of the catalytic site of Mm Cas6b. In order to test the influence of the computationally predicted [RNAfold (41)] short hairpin structure of the repeat (Figure 1), the mutation G16C was introduced that disrupts a G–C base pair within this hairpin. The mutated repeat was cleaved less effectively than wild-type repeats (Supplementary Figure S4).
To deduce catalytic residues, potentially important amino acids were identified based on the structural model of Mm Cas6b and the observed conservation of amino acids in the alignment of Cas6b homologues (Figure 3). Cas6 proteins of other CRISPR/Cas subtypes contain a single catalytic histidine residue. In Cas6b, there are two conserved histidine residues (H38 and H40), separated only by a single amino acid that could potentially fulfil this role. A set of Mm Cas6b mutants was produced (Supplementary Figure S5) of which two mutants (Y49A and Y47A/Y49A) yielded insoluble proteins. The other mutants were used in endonucleolytic cleavage assays testing the processing of the repeat RNA substrate in comparison to wild-type Mm Cas6b (Figure 6). The single histidine mutant H38A and the tyrosine mutant Y47A showed reduced processing activity compared with wild-type Mm Cas6b. The H40A mutation reduced Mm Cas6b activity by >50%. Surprisingly, both single histidine mutants retained considerable cleavage activity. However, the mutation of both histidine residues into alanine (H38A/H40A) resulted in a drastic loss of substrate processing. Mutation of the lysines at Position 29 or 30 did not show any notable effect on endonucleolytic activity.
The observed crRNA processing and abundance patterns of CRISPR/Cas subtype I-B are in good agreement with crRNA maturation previously analysed for other CRISPR/Cas subtypes and substantiate that 8-nt 5′-terminal and trimmed 3′-terminal crRNA tags are an universal feature of many CRISPR/Cas subtypes. The surprising observation of a spacer sequence that promotes crRNA production internally in C. thermocellum exemplifies the effect that an individual spacer can have for the abundance of mature crRNAs and subsequently the efficiency of entire CRISPR regions. The exchange of 1nt of the otherwise universal 5′-terminal 8-nt tag of the crRNAs in the vicinity of the internal CRISPR transcription start site opens the possibility that two CRISPR elements might have been fused and subsequently portions of a leader element might have been incorporated into the repeat–spacer–repeat pattern. In addition, spacer elements were shown to promote transcripts of the reverse orientation. These anti-crRNAs were first described in Sulfolobus (42) and were also found in P. furiosus (43). However, they appear to be absent in most organisms. The occurrence of specific processing patterns for anti-crRNAs from different repeats of C. thermocellum CRISPR and the absence of anti-crRNAs in M. maripaludis indicate that this phenomenon might be specific for organisms with relaxed transcription start site definition rather than for the CRISPR/Cas subtype. Individual anti-crRNAs appear to be better suited for a conserved maturation process at their termini by a currently unknown mechanism. Reverse transcripts can form double-stranded RNA duplexes that might reduce the abundance and efficiency of crRNAs. Taken together these results highlight the effects that individual spacer sequences within a CRISPR region can have in both forward and reverse direction. These strong effects will need to be taken into consideration in the anticipated and proposed design of synthetic CRISPR regions for biotechnologically or medically important processes.
We identified the Cas6b endonuclease responsible for crRNA maturation in the CRISPR/Cas subtype I-B. Cas6 enzymes are among the most diverse members in the sets of Cas protein of the different CRISPR/Cas subtypes and can be used to classify CRISPR/Cas systems. The similarity of repeat sequences and Cas6b enzymes between Bacteria (i.e. Clostridia) and Archaea (i.e. methanogens) hints at a horizontal transfer event for these CRISPR/Cas systems. Cas6 proteins might show this remarkable degree of divergence due to their individual adaptation to the given repeat sequence and/or structure. Evidence for this can also be found in the different principles of recognition for Pf Cas6, Cas6e and Cas6f. While Pf Cas6 binds to unstructured RNA, both Cas6e and Cas6f need a secondary structured RNA to bind and process in vitro (13,14,16). For Type II CRISPR systems, no Cas6 activity was reported. In these systems, the presence of a guide RNA (tracrRNA) recruits RNase III for the processing of crRNAs (44). These pathways exemplify the differences in crRNA maturation among organisms and CRISPR subtypes.
The Cas6 enzymes Pf Cas6, Cas6e and Cas6f of different CRISPR/Cas subtypes were all shown to require a single conserved histidine residue for catalysis (10,12–14,16,29). In this study not one but two conserved histidine residues were identified for Mm Cas6b. Only the simultaneous mutation of both histidine residues resulted in a drastic loss of endonuclease activity. This implies that Cas6b exhibits the first example of a more flexible catalytic core in which both histidine residues are potentially representing the catalytic histidine and able to complement the loss of the other residue. Why did Cas6b evolve two catalytic histidine residues where this function can be fulfiled by a single histidine in other Cas6 enzymes? One possible explanation is the advantage such setup would have in coping with different substrates, e.g. with crRNA precursors that contain spacer of different length or structure. In Pf Cas6 a catalytic triad was described that provides a catalytic site for general acid–base catalysis (10,14). Mm Cas6b does not contain an identical catalytic triad but our observation of the importance of tyrosine 47 for Mm Cas6b activity and the occurrence of clustered amino acids that could provide general bases and acids might indicate more flexible active site architecture.
In conclusion, we provide the first description of crRNA processing in vivo and in vitro for CRISPR/Cas subtype I-B. These analyses of Cas6b in a bacterial and an archaeal model organism highlight the similarities between different CRISPR/Cas subtypes and the differences in crRNA processing. Two interchangeable catalytic histidine residues in Cas6b and internal promotion of crRNA production in C. thermocellum exemplify two new concepts that were found for CRISPR/Cas I-B systems.
Supplementary Data are available at NAR Online: Supplementary Tables 1 and 2 and Supplementary Figures 1–5.
The Deutsche Forschungsgemeinschaft [DFG, FOR1680] and the Max Planck Society. Funding for open access charge: Max Planck Society.
Conflict of interest statement. None declared.
The authors thank Andreas Su for technical help and André Plagens for advice and discussions The authors are very grateful to Michael Rother, Rolf Thauer and William B. Whitman for their help with the handling of Methanococcus maripaludis.