Cas9-mediated genome editing in the methanogenicarchaeon Methanosarcina acetivoransDipti D. Nayaka and William W. Metcalfb,1

aCarl R. Woese Institute for Genomic Biology, University of Illinois at Urbana–Champaign, Urbana, IL 61801; and bDepartment of Microbiology, University ofIllinois at Urbana–Champaign, Urbana, IL 61801

Edited by Mary E. Lidstrom, University of Washington, Seattle, WA, and approved February 1, 2017 (received for review November 9, 2016)

Although Cas9-mediated genome editing has proven to be a pow-erful genetic tool in eukaryotes, its application in Bacteria hasbeen limited because of inefficient targeting or repair; and itsapplication to Archaea has yet to be reported. Here we describethe development of a Cas9-mediated genome-editing tool thatallows facile genetic manipulation of the slow-growing methano-genic archaeon Methanosarcina acetivorans. Introduction of bothinsertions and deletions by homology-directed repair was remark-ably efficient and precise, occurring at a frequency of approxi-mately 20% relative to the transformation efficiency, with thedesired mutation being found in essentially all transformants ex-amined. Off-target activity was not observed. We also observedthat multiple single-guide RNAs could be expressed in the sametranscript, reducing the size of mutagenic plasmids and simulta-neously simplifying their design. Cas9-mediated genome editingreduces the time needed to construct mutants by more than half (3vs. 8 wk) and allows simultaneous construction of double mutantswith high efficiency, exponentially decreasing the time needed forcomplex strain constructions. Furthermore, coexpression the non-homologous end-joining (NHEJ) machinery from the closely relatedarchaeon, Methanocella paludicola, allowed efficient Cas9-mediatedgenome editing without the need for a repair template. The NHEJ-dependent mutations included deletions ranging from 75 to 2.7 kb inlength, most of which appear to have occurred at regions of nat-urally occurring microhomology. The combination of homology-directed repair-dependent and NHEJ-dependent genome-editingtools comprises a powerful genetic system that enables facile in-sertion and deletion of genes, rational modification of gene ex-pression, and testing of gene essentiality.

Cas9 | Archaea | methanogens | Methanosarcina | genetics

The CRISPR (clustered regularly interspaced palindromicrepeats) array and associated cas genes are widespread in

microbial genomes (1), where they confer acquired immunity tophage and foreign DNA elements (2). The type IIA system fromStreptococcus pyogenes is especially well characterized and hasbeen widely applied as a remarkably effective genome-editingtool (3). During genome editing, heterologous expression of theRNA-guided DNA endonuclease Cas9 and a chimeric single-guide (sg) RNA, comprised of a 20-bp spacer that targets thechromosome and an 80-bp scaffold that binds Cas9, leads to alethal double-strand break (DSB) at all target sites within thegenome that are flanked by a 3′ NGG protospacer adjacent motif(PAM) (4) (Fig. S1). In eukaryotes, the nonhomologous end-joining (NHEJ) repair pathway can mend the DSB by generatingsimple insertions or deletions at the sgRNA target site, thus pre-venting additional rounds of Cas9-mediated cleavage (3, 5). Alter-natively, the native homology-dependent repair (HDR) pathwaycan repair the fatal DSB, so long as a repair template that modifiesor removes the sgRNA target site is provided, again preventingadditional rounds of Cas9-mediated cleavage (3, 5) (Fig. S1). Ap-propriately designed repair templates allow recovery of strains withprecise insertions and deletions, allowing unprecedented ability tomanipulate the genomes of these diploid (or polyploid) organisms(6). Although Cas9-mediated genome editing has been successfully

and broadly implemented in eukaryotes (3), similar progress hasnot been achieved in prokaryotes, with Cas9-mediated genomeediting having been demonstrated in only 10 bacterial genera(7–10); to our knowledge, it has not been applied in archaea.Archaea have been recognized as a phylogentically distinct

group since the 1990s (11) and it is now well-established that theyare prevalent in many environments, often providing keystoneecosystem functions (12, 13). As a result, archaea play a major rolein the biogeochemical cycling of nitrogen, sulfur, and carbon (13).Methanogenic archaea are particularly noteworthy from thisstandpoint. These microorganisms are widely distributed in strictlyanaerobic environments, such as waterlogged rice paddies, sewagetreatment plants, and the digestive systems of numerous animals(14), where they generate the overwhelming majority of methanereleased in the atmosphere. As such, it is not surprising that theyhave a significant impact on climate change and the global carboncycle. Members of the genus Methanosarcina are among the mostabundant and metabolically versatile methanogens known (15).They are also genetically tractable (16) and have emerged as im-portant model organisms for genetic analysis of methanogen bi-ology. Although the range of genetic techniques available for usein Methanosarcina is fairly comprehensive (16, 17), slow-growthand fastidious cultivation requirements have dramatically affectedthe pace of genetic studies within this genus.With this in mind, we explored whether the Cas9-mediated

genome-editing technique could increase the efficiency, efficacy,and speed of genetic analysis in Methanosarcina. Our results showthat Cas9-mediated editing driven by the native HDR machinery in

Significance

Methanogenic archaea play a central role in the global carboncycle, with profound implications for climate change, yet ourknowledge regarding the biology of these important organ-isms leaves much to be desired. A key bottleneck that hindersthe study of methanogenic archaea, especially those within thegenus Methanosarcina, results from the time-consuming andoften cumbersome tools that are currently available for geneticanalysis of these microbes. The Cas9-mediated genome-editingapproach for Methanosarcina acetivorans described in thisstudy addresses this major constraint by streamlining the mu-tagenic process and enabling simultaneous introduction ofmultiple mutations. This work also sheds light on the distinctproperties of homology-dependent repair and nonhomologousend-joining machinery in Archaea.

Author contributions: D.D.N. and W.W.M. designed research; D.D.N. performed research;D.D.N. and W.W.M. analyzed data; and D.D.N. and W.W.M. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The sequences reported in this paper have been deposited in the Gen-Bank database (accession no. KY436376) and the Sequenced Reads Archive (SRA) in theNational Center for Biotechnology Information (accession no. PRJNA352863).1To whom correspondence should be addressed. Email: metcalf@illinois.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1618596114/-/DCSupplemental.

2976–2981 | PNAS | March 14, 2017 | vol. 114 | no. 11 www.pnas.org/cgi/doi/10.1073/pnas.1618596114

Dow

nloa

ded

by g

uest

on

Aug

ust 2

5, 2

020

this archaeon is extremely rapid and efficient, even when mul-tiple mutations are simultaneously introduced. Furthermore,although Methanosarcina species do not encode a native NHEJpathway, coexpression of the NHEJ machinery from the closelyrelated archaeon, Methanocella paludicola, along with the Cas9–sgRNA complex, allowed robust template-independent repair.

ResultsDevelopment of a Cas9-Dependent Genome-Editing System forMethanosarcina acetivorans. To determine whether the appropri-ate components for genome editing from S. pyogenes are func-tional in M. acetivorans, we constructed a Methanosarcina/Escherichia coli shuttle vector that expresses the S. pyogenes Cas9ORF from a tetracycline inducible Methanosarcina promoter(Fig. 1A). We also constructed a derivative of this plasmid thatemploys a methanol-inducible promoter to express an sgRNAthat targets Cas9 to ssuC, a gene required for uptake of themethanogenesis inhibitor bromoethane sulfonic acid (BES) (18)(Fig. 1 A and B). M. acetivorans was readily transformed with theCas9-only plasmid pDN206 (78,900 ± 9,940 PurR transfor-mants); however, pDN208, which contains the ssuC-targeting

sgRNA in addition to Cas9, produced only 4 ± 3 PurR trans-formants. This difference in plating efficiency of more than fourorders-of-magnitude strongly suggests that Cas9 is not toxic byitself, but that the Cas9–sgRNA complex from S. pyogenes iscapable of generating a lethal DSB in M. acetivorans. Similarresults were obtained with and without the inducers methanoland tetracycline.Next, we determined the ability of the native HDR machinery

in M. acetivorans to repair the lethal DSB generated by thesgRNA–Cas9 complex. To this end, repair templates of varyingsize were added to the ssuC-targeting vector. These repairtemplates generate a 34-bp deletion/frameshift mutation withinssuC that removes the targeting site while simultaneously in-troducing a diagnostic NotI restriction endonuclease site (Fig.1B). Addition of repair templates with 1-kb homology arms tothe plasmids relieved the lethal effect of targeting Cas9 to ssuC,generating nearly 20,000 PurR transformants per 2 μg DNA. Asimilar plasmid with 0.5-kbp homology arms generated roughlyhalf as many transformants. Significantly, the 103-fold highertransformation efficiency for pDN211 relative to pDN208 indicatedthat 99.9% of the PurR transformants are likely to be mutants (i.e.,

A

C

B

Fig. 1. Cas9-mediated genome editing in M. acetivorans. (A) Key elements of the pDN_CRISPR plasmid series include the Cas9 ORF from S. pyogenes fused tothe tetracycline-inducible PmcrB(tetO1) promoter (in green), sgRNA(s) fused to the methanol-inducible PmtaCB1 promoter (in pink), a homology repairtemplate (in orange), and the entire pC2A plasmid replicon containing an autonomous Methanosarcina origin of replication (in gray). The puromycintransacetylase (pac) marker enables selection of puromycin resistant (PurR) transformants and the hypoxanthine phosphoribosyltransferase (hpt) markerfacilitates plasmid curing by counter selection on medium containing 8ADP. Note: The E. coli replicon and resistance marker genes have not been shown.(B) Expression of sgRNA with a 20-bp target sequence identical to a region of the WT ssuC locus (in blue) flanked by a 3′ NGG PAM (in red) with Cas9 generatesa DSB at the ssuC locus. A region of the plasmid pDN211 contains a homology repair (HR) template to abolish the target site by generating a 34-bp deletionand simultaneously introducing a diagnostic NotI restriction endonuclease site in the ssuC ORF (in orange). (C) The chromosomal ssuC locus amplified from 20 PurR

transformants containing pDN211 as well as the parent strain (WWM60) and subjected to restriction digest with NotI. Upon digestion, 1.1-kbp and 1.3-kbp fragmentsare observed for all PurR transformants (lanes 2–21), whereas a single 2.4-kbp fragment corresponding to the WT locus is observed for WWM60 (lane 22).

Nayak and Metcalf PNAS | March 14, 2017 | vol. 114 | no. 11 | 2977

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Aug

ust 2

5, 2

020

only 1 of every 1,000 PurR transformants would still contain the WTlocus). To validate this hypothesis, 20 of these transformants weregenotyped by a performing a NotI digest of a PCR amplicon con-taining the edited ssuC locus: all tested positive for the introducedmutation (Fig. 1C). Furthermore, as expected for null mutations inthe ssu locus, all 20 transformants were resistant to 0.4 mM BES,a concentration lethal to the parent strain. Genome editing wasalso observed when plasmids were integrated into the chromo-some using a ΦC31 integrase system (17).The initial gene-edited strains produced in these experiments

retain the targeting machinery; thus, we constructed plasmidderivatives that include a counter selectable marker (hpt) to facili-tate curing of gene-editing vector. This marker confers sensitivity tothe purine analog 8-aza-2,6-diaminopurine (8ADP) in strains thatlack the native hpt gene (19). To validate the plasmid curing system,which has not previously been attempted in Methanosarcina, weselected 8ADPR clones from three independent PurR transformantsconstructed using the counter selectable vectors. All 8ADPR isolatesanalyzed were PurS and also contain the frameshift mutation at thessuC locus (Fig. S2A). PCR-based screening with plasmid-specificprimers showed that the vector was indeed cured from thesestrains (Fig. S2B). These proof-of-principle experiments showthat a Cas9-mediated genome editing technique can be used toeffectively introduce unmarked mutations in M. acetivorans.

Optimization of the Cas9-Dependent Genome-Editing Technique inM. acetivorans. To determine the optimal expression levels for thegenome-editing machinery, we varied the transcription of Cas9by selecting transformants on media with increasing concentra-tions of tetracycline, and of the sgRNA by plating on media witheither methanol (induced) or trimethylamine (TMA; repressed)as growth substrates. Surprisingly, no significant difference ingenome-editing efficiency was observed (Fig. 2A). In fact, thebasal level of transcription provided by the two promoters in the

absence of the inducers was sufficient for effective editing (Fig.2A). A control vector identical to pDN211 but lacking thesgRNA (pDN207) was used to estimate the efficiency of genomeediting. The efficiency of genome editing was measured as theratio of mutant recovery (i.e., plating efficiency of pDN211)relative to the plating efficiency of the control vector and wasestimated on media with either methanol or TMA as growthsubstrates. Significantly, genome editing in these experimentswas particularly efficient, with edited strains being obtained atfrequencies of approximately 20–25% relative to the control (i.e.,one in four cells that receive the plasmid undergo gene conver-sion) (Fig. 2B).To examine the maximum size of deletions that can be reliably

generated by a single sgRNA, we tested repair templates with 1-kbhomology arms placed at varying distance from the sgRNA-directedDSB (Fig. 2C). The transformation efficiency remained steady fortemplates that are ≤250 bp away from each end of the DSB, butdeclined precipitously when the distance increased beyond thispoint (Fig. 2C). Thus, a single sgRNA can be reliably used to deleteup to 0.5 kbp of the chromosome, although larger deletions (upto 1 kbp) can be produced at the expense of efficiency.

Multiplex Expression of sgRNAs inM. acetivorans Enables SimultaneousIntroduction of Multiple Mutations. To explore the possibility ofusing multiple Cas9-mediated DSBs to create larger deletions, orto simultaneously introduce more than one mutation, we testedtwo alternate arrangements for the expression of multiple sgRNAs.In the first arrangement, sgRNAs were expressed individually,whereas in the second they were expressed as a single transcriptseparated by a 30-bp linker sequence (Fig. 3A). Plasmids withsgRNAs in either arrangement were equally efficient in generatingstrains with complete deletions (approximately 2 kbp) of themtmCB1 and mtmCB2 loci, which are highly homologous genesencoding monomethylamine methyltransferase isozymes (Fig. 3Band Fig. S3). Thus, to reduce the size of mutagenic plasmids and

A

C

B

Fig. 2. Optimization of Cas9-mediated genome editing in M. acetivorans.(A) A dose–response curve showing the relative transformation efficiency ofpDN211 for different expression levels of Cas9 and the sgRNA. Transformantswere plated on solid medium containing either TMA hydrochloride (sgRNAuninduced; in blue) or methanol (sgRNA induced; in green) as the growthsubstrate with tetracycline concentrations ranging from 0 to 64 μg/mL,as indicated. (B) Mean transformation efficiencies of pDN211 and pDN207(a control vector that lacks the sgRNA targeting ssuC). (C) Mean trans-formation efficiencies of plasmids containing repair templates placed atvariable distance from the sgRNA-directed DSB for ssuC. Values above eachcolumn represent the fraction of transformants for the correspondingplasmid that tested positive for the desired mutation by a PCR-based screen.The error bars represent one SD of the mean transformation efficiency forthree independent transformation reactions. All transformations wereplated on medium lacking tetracycline with TMA as the growth substrate.

A

C

B

Fig. 3. Simultaneous expression of multiple sgRNAs and generation ofmultiple mutations in M. acetivorans. (A) Two configurations for the ex-pression of multiple sgRNAs were tested: in configuration one each sgRNAcontains an individual promoter, whereas in configuration two a singlepromoter drives the expression of multiple sgRNAs separated by a 30-bplinker sequence. (B) Mean transformation efficiency of plasmids with sgRNAsin configuration one (light gray) or two (dark gray) configurations to deleteeither mtmCB1 or mtmCB2. Note: two independent transformation reac-tions were performed per plasmid. (C) Mean transformation efficiencies ofplasmids to generate either ΔmtmCB1 (green), ΔmtmCB2 (blue), orΔmtmCB1ΔmtmCB2 (purple) simultaneously. The error bars represent oneSD of the mean transformation efficiency for three independent trans-formation reactions. All transformations were plated on medium lackingtetracycline with methanol as the growth substrate.

2978 | www.pnas.org/cgi/doi/10.1073/pnas.1618596114 Nayak and Metcalf

Dow

nloa

ded

by g

uest

on

Aug

ust 2

5, 2

020

simultaneously simplify their design, the placement of sgRNAs ona single transcript was preferred. Subsequently, we generated aplasmid containing all four sgRNAs and each of the correspondingrepair templates to simultaneously delete mtmCB1 and mtmCB2(Fig. S4). Surprisingly, transformants that simultaneously acquiredboth the ΔmtmCB1 and ΔmtmCB2 mutation were obtained at thesame frequency as transformants that acquired only one of twomutations (Fig. 3C). Furthermore, the genomes of two random,independent isolates each for the ΔmtmCB1, ΔmtmCB2, andΔmtmCB1ΔmtmCB2 mutants were completely sequenced and nooff-target activity was detected (Table S1). This finding is espe-cially notable given the high levels of homology between thesgRNA target sites in the two genes (Fig. S5). Hence, Cas9-mediated genome editing is remarkably precise in M. acetivorans.

Insertion of Large DNA Segments via Cas9-Dependent GenomeEditing. To assess the efficacy of gene knockins, plasmids designedto insert either a 3.05-kbp fragment containing the mtmCB1 locusor a 2.53-kbp fragment containing the mtmCB2 locus were con-structed and used to introduce WT copies of each gene into thessuC gene of the ΔmtmCB1ΔmtmCB2 double mutant (Fig. S6).Although 20–60% fewer transformants were observed (relative tothe simple 34-bp deletion mutation described above), all trans-formants screened contain the desired gene insertions.

A Cas9-Dependent Genetic Screen to Test for Gene Essentiality. Twomethods have previously been used to test gene essentiality inMethanosarcina (17, 20); however, both approaches are laboriousand time-consuming. We therefore sought to use the efficientrepair of the DSB generated by the Cas9–sgRNA complex toassay gene essentiality in M. acetivorans. For nonessential genes(ssuC, mtmCB1, mtmCB2), 103- to 104-fold more transformantsare consistently observed when a repair template to generate adeletion is provided in addition to the sgRNA–Cas9 complex.We expected that this would not be true for essential genes,because HDR-directed repair using a deletion cassette wouldalso be lethal. To test this idea, we constructed plasmids, withand without a repair template, that target the previously estab-lished essential genes mcrA and hdrED (17, 21). In contrast tothe results with the nonessential ssuC, mtmCB1 and mtmCB2loci, where we obtained thousands of transformants in thepresence of a repair template, plasmids that targeted the es-sential genes generated fewer than five transformants, regardlessof whether a repair template is present (Table 1). Thus, the ratioof transformants obtained in the presence versus absence of arepair template can be used as a reliable and simple test for geneessentiality in M. acetivorans.

Heterologous Expression of NHEJ Genes Leads to Template-Independent Repair in M. acetivorans. A lethal phenotype forplasmids expressing a Cas9–sgRNA complex in the absence of arepair template was uniformly observed across a wide range of

sgRNAs tested in this study, suggesting that NHEJ does notoccur in M. acetivorans (Table 1). This result is consistent withthe absence of genes related to Ku and LigD in the completelysequenced genome (22). Nevertheless, in some circumstancesHDR-independent gene editing would be very useful. Therefore,we examined whether NHEJ could be established in Meth-anosarcina for use in conjunction with the Cas9–sgRNA com-plex. For this purpose, we chose the NHEJ machinery from theclosely related methanogen M. paludicola, which has previouslybeen reconstituted in vitro (23, 24). An artificial operon encod-ing four M. paludicola NHEJ proteins [DNA ligase (Lig), poly-merase (Pol), phosphoesterase (PE), and Ku] was synthesizedand transcriptionally fused to the moderately expressed serCpromoter (25) to allow transcription in M. acetivorans (Fig. S7).This cassette was then added to the Cas9 ssuC-targeting vectorwithout a repair template. Transformation with this plasmid wasapproximately 100-fold less efficient than the correspondingHDR vector, but approximately 10-fold higher than with plas-mids lacking the NHEJ system (Fig. 4A). Therefore, expressionof the M. paludicola NHEJ machinery overcame the lethal effectof the Cas9 ssuC-targeting vector without a repair template.Molecular analysis of the ssuC locus in these transformantsrevealed deletions ranging from 75 bp to 2.7 kb in length, oftenoccurring at naturally occurring regions of microhomology 6–11 bpin length (Fig. 4B). Thus, the combined Cas9/NHEJ system pro-vides the opportunity to generate a variety of mutations surroundinga single target site. Importantly, these plasmids are much simpler toconstruct, requiring only addition of target-specific sgRNA. We alsotested whether addition of two sgRNAs targeting DNA sequencesapproximately 450 bp apart in conjunction with NHEJ could beused to generate precise deletions without a repair template.Interestingly, attempts to construct ssuC deletions via this methodwere not successful: only a handful of colonies were obtained (6 ± 3per 2 μg plasmid) and none had the precise deletion desired. Weexamined 20 transformants obtained by this method. Two con-tained a 1.3-kb deletion of the ssuC locus, which occurred at aregion of microhomology (Fig. 4B). The remainder had WTcopies of the ssuC gene and, thus, are likely to be so-called es-cape mutants in which the Cas9 gene or sgRNA has mutated onthe targeting plasmid (26).

DiscussionThe Cas9-based tools developed in this study will have a trans-formative impact on the speed, scope, and scale of research that canbe accomplished in the methanogenic archaeon, M. acetivorans.

Table 1. Using Cas9-mediated genome editing as a screen forgene essentiality in M. acetivorans

Gene/operon

Transformation efficiency of plasmids with genomeediting machinery

Repair template absent Repair template present

ssuC 4 ± 3 19,040 ± 4,255mtmCB1 <1 4,033 ± 716mtmCB2 Not tested 5,300 ± 235mcrA 2 ± 1 <1hdrED <1 0

Transformation efficiencies indicate the mean ± 1 SD of puromycin re-sistant colonies for three independent transformations.

1

10

100

1000

10000

HDR NHEJ No Template

Co

lon

y fo

rmin

g u

nit

sp

er 2

g p

lasm

id

B

1/5075 bp -------

1/501080 bp TCCTGC3/50386 bp CCCTGACAA

7/501747 bp GTGGACGAAGC

8/502624 bp CTGAAGAAGCC

FrequencyDeletion

28/501361 bp CCCTCAGCCA

Microhomology

A

Fig. 4. Coexpression of NHEJ genes with the Cas9-sgRNA complex inM. acetivorans. (A) Transformation efficiency of plasmids with a sgRNA tar-geting the ssuC locus containing either a repair template for HDR-mediatedDSB repair, the NHEJ genes, or no repair template. The error bars representone SD of the mean transformation efficiency for three independent trans-formation reactions. All transformations were plated on medium lackingtetracycline with TMA as the growth substrate. (B) Regions of naturally oc-curring microhomology surrounding the ssuC locus at which NHEJ-mediateddeletions were observed in PurR transformants.

Nayak and Metcalf PNAS | March 14, 2017 | vol. 114 | no. 11 | 2979

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Aug

ust 2

5, 2

020

Most notably, multiplexed gene-editing plasmids will enable gen-eration of strains with multiple mutations, ranging from SNPs tolarge indels, in a matter of weeks versus years. These tools willenable researchers to swiftly tag genes at their native loci on thehost chromosome, allowing the study of context-specific gene ex-pression, “pull-down” experiments to establish protein–protein andprotein–DNA interaction networks, and purification of proteinsthat contain unique amino acids (27) or novel posttranslationalmodifications (28). Furthermore, deleting a gene of interest usingthe NHEJ-based technique is very cost-effective, as it simply re-quires the insertion of a commercially synthesized DNA fragmentcontaining the appropriate sgRNAs into pDN243, the vector con-taining Cas9 and the NHEJ machinery. Thus, studies that werepreviously inconceivable, such as constructing a library of strainswith single-gene deletions in every nonessential gene on theM. acetivorans chromosome [as described in Wang et al. (29)],will now become feasible, in terms of both time and cost. Finally, weexpect that minor modifications will enable the application of thisapproach to a broad range of methanogens and other archaea.Certain features of Cas9-mediated genome editing in

M. acetivorans are particularly unique and noteworthy. For exam-ple, unlike eukaryotes (30), targeting of the Cas9–sgRNA complexto a particular chromosomal region inM. acetivorans is remarkablyprecise, as no off-target activity was observed upon resequencingmultiple, independent genome-edited mutants (Table S1). Be-cause the M. acetivorans genome (approximately 5.75 Mbp) is 10-to 100-fold smaller in comparison with eukaryotic genomes, it ispossible that fewer off-target sites are present. However, no off-target activity could be detected, despite our intentional choice ofhighly similar sgRNA targets in the mtmCB isozymes (Fig. S5).Thus, it is likely that properties of the Cas9–sgRNA complex, in-cluding target specificity, vary significantly across domains of life,perhaps because of differences in chromosomal organization andDNA repair machinery. Notably, unlike the Cas9-mediated genomeediting in bacteria (7–9), we observe a high rate of HDR for theCas9-mediated DSB and a very low frequency of “escape” mutants.These key distinctions are likely to stem from evolutionarily distinctHDR machinery. Archaeal DNA repair involves homologs of theeukaryotic proteins Mre11 and Rad50, and two other uniqueproteins HerA and NurA, which perform end-resection after aDSB occurs (31). Subsequently, the RecA orthologs RadA andRadB, again more closely related to recombination proteins ofeukaryotes, mediate strand invasion (31). Finally, Hjc, unrelatedto the RuvABC complex in bacteria (32), is involved in theresolution of the Holliday junction (31). Thus, it is tempting tospeculate that coexpression of archaeal HDR machinery alongwith Cas9 might overcome some of the obstacles that have beenreported in recent bacterial work (7–9).We observed similar host-specific effects upon heterologous

expression of the NHEJ machinery from M. paludicola inM. acetivorans. These archaeal proteins have biochemical activ-ities that are strikingly similar to the well-characterized bacterialKu and LigD of Mycobacterium tuberculosis (33). Thus, we weresomewhat surprised by the robust template-independent repairthey conferred when coexpressed with the Cas9–sgRNA complexin M. acetivorans (Fig. 4A). These data are in sharp contrast toa recent study in which coexpression of Ku and LigD fromM. tuberculosis did not rescue the Cas9-mediated DNA break inE. coli (34). Furthermore, we observed that template-independentDNA repair happens at naturally occurring regions of micro-homology (ranging from 6 to 11 bp), which supports a recenthypothesis that the archaeal NHEJ pathway conduct micro-homology-mediated end joining (MMEJ) in vivo (24). Thus, weexpect that in addition to its application as a means of generatingrandom site-specific mutations in M. acetivorans, this tool canalso be used to dissect the archaeal MMEJ machinery in vivo. Inthis context, we note that no particular sequence pattern or anydistinct signature (GC content, nucleotide frequency) could be

inferred from the regions of microhomology at which repairoccurred (Fig. 4B). Moreover, DNA repair mediated by MMEJis almost completely abolished when two sgRNAs were simul-taneously expressed, suggesting that the repair mechanism hasthe ability to distinguish breaks that occur at discrete loci.Finally, one might ask why we chose to use the well-estab-

lished S. pyogenes Cas9–sgRNA complex for genome-editingpurposes over the native type I or type III CRISPR/Cas systemsthat are commonly found in Methanosarcina spp. (35), as wasdone in Sulfolobus islandicus (36). First, the CRISPR/Cas subtypesvary significantly across the genus Methanosarcina, even withinstrains belonging to the same species (35). Hence a genome-editingtechnique reliant on the native CRISPR/Cas machinery for onestrain might not work in other closely related strains. Recent studiesacross a wide-range of bacteria have revealed that anti-CRISPRproteins to silence the native CRISPR/Cas system are also oftenencoded on the chromosome (37). Although no anti-CRISPRproteins have been detected in Methanosarcina, it is possible thatthey exist and might potentially complicate use of the nativeCRISPR/Cas machinery for genome editing. Finally, tweaking thenative CRISPR/Cas machinery for genome editing purposes is likelyto impact organismal physiology in an unpredictable fashion andskew genetic analyses downstream. Thus, we chose to deploy thesimple, modular Cas9-mediated genome editing machinery on avector that will be transiently maintained in M. acetivorans.

Materials and MethodsStrains, Media, and Growth Conditions. All chemicals were purchased fromSigma-Aldrich unless otherwise specified. M. acetivorans strains were grownin single-cell morphology (38) at 37 °C in bicarbonate-buffered high-salt (HS)liquid medium containing 125 mM methanol or 50 mM TMA hydrochloridein Balch tubes with N2/CO2 (80/20). Plating solid medium was conducted inan anaerobic glove chamber (Coy Laboratory Products) as described pre-viously (25). Solid media plates were incubated in an intrachamber anaer-obic incubator maintained at 37 °C with N2/CO2/H2S (79.9/20/0.1) in theheadspace, as described previously (39). Puromycin (CalBiochem), the purineanalog 8ADP (R. I. Chemicals) and BES were added to a final concentration of2 μg/mL, 20 μg/mL, 0.4 mM, respectively, from sterile, anaerobic stock solu-tions. Anaerobic, sterile stocks of tetracycline hydrochloride in deionizedwater were prepared fresh shortly before use and added to a final con-centration as indicated. E. coli strains were grown in LB broth at 37 °C withstandard antibiotic concentrations. WM4489, a DH10B derivative engi-neered to control copy-number of oriV-based plasmids (40), was used as thehost strain for all plasmids generated in this study (Table S2). Plasmid copynumber was increased by adding sterile rhamnose to a final concentrationof 10 mM.

Plasmids. All plasmids used in this study are listed in Table S2. The plasmidpMJ0806 was obtained from Jennifer Doudna, University of California,Berkeley, CA (Addgene plasmid # 39312). The S. pyogenes (Spy) Cas9 ORFwas fused to the PmcrB(tetO1) promoter in pJK027A (17), and linearizedwith NdeI and HindIII by the Gibson assembly method, as described pre-viously (41). The DNA segments containing sgRNA flanked by putativemtaCB1 promoter and terminator sequences from M. acetivorans were syn-thesized as double-stranded DNA fragments (“gBlocks”) from Integrated DNATechnologies and used for cloning purposes per the manufacturer’s instructions.A 3.25-kbp artificial operon with the NHEJ polymerase (Mcp_2125), DNA ligase(Mcp_2126), phosphoesterase (Mcp_2127), and Ku (Mcp_0581) genes from M.paludicola SANAE fused to the Methanosarcina barkeri Fusaro serC promoterwas ordered from the GeneArt gene synthesis service (Life Technologies). Allsynthetic DNA fragments and repair templates were introduced in the ap-propriate vector backbone linearized with either AscI or PmeI by the Gibsonassembly method, as described previously (41). The entire pC2A plasmid wasintroduced in the appropriate pJK027A-derived vector (carrying the λattBsite) by retrofitting with pAMG40 (carrying the λattP site) using the BPClonase II master mix (Invitrogen) per the manufacturer’s instructions.WM4489 was transformed by electroporation at 1.8 kV using an E. coli GenePulser (Bio-Rad). Standard techniques were used for the isolation and ma-nipulation of plasmid DNA. All pJK027A-derived plasmids were verified bySanger sequencing at the Roy J. Carver Biotechnology Center, University ofIllinois at Urbana–Champaign, and all pAMG40 cointegrates were verified byrestriction endonuclease analysis. Primers used in this study are listed in

2980 | www.pnas.org/cgi/doi/10.1073/pnas.1618596114 Nayak and Metcalf

Dow

nloa

ded

by g

uest

on

Aug

ust 2

5, 2

020

Table S3. The plasmid sequence and annotations for pDN211 have beensubmitted to GenBank (accession no. KY436376).

In Silico Design of Target Sequences.All target sequences used in this study arelisted in Table S4. Target sequences were designed using the CRISPR sitefinder tool in Geneious version R9 (42). The M. acetivorans chromosome andthe plasmid pC2A were used to score off-target binding sites.

Transformation of M. acetivorans. All M. acetivorans strains used in this studyare listed in Table S5. Liposome-mediated transformation was used forM. acetivorans, as described previously(43), and 10 mL of late-exponentialphase culture of M. acetivorans and 2 μg of plasmid DNA were used foreach transformation.

Genome Sequencing and Analysis. Genomic DNA from M. acetivorans wasextracted using a protocol described previously (44). DNA libraries were

prepared with the Hyper Library construction kit (Kapa Biosystems) andquantified using qPCR. All libraries were sequenced on one lane of an Illu-mina MiSeq v2 (Illumina) at the Roy J. Carver Biotechnology Center, Universityof Illinois at Urbana–Champaign using a 500 cycles v2 sequencing kit (Illu-mina). Trimmed, paired end 250-nt reads were mapped to the M. acetivoransreference genome (NC_003552) using default parameters for breseq v0.25(45). Trimmed genome sequencing reads have been deposited in the Se-quenced Reads Archive at the National Center for Biotechnology Informationunder accession no. PRJNA352863.

ACKNOWLEDGMENTS. We thank Dr. Mary Elizabeth Metcalf for technicalassistance. This work was supported in part by the Division of Chemical Sciences,Geosciences, and Biosciences, Office of Basic Energy Sciences of the US De-partment of Energy Grant DE-FG02-02ER15296 (to W.W.M.), and the Carl R.Woese Institute for Genomic Biology postdoctoral fellowship (to D.D.N.). D.D.N.is currently a Simons Foundation fellow of the Life Sciences Research Foundation.

1. Makarova KS, et al. (2015) An updated evolutionary classification of CRISPR-Cas sys-tems. Nat Rev Microbiol 13(11):722–736.

2. Horvath P, Barrangou R (2010) CRISPR/Cas, the immune system of bacteria and ar-chaea. Science 327(5962):167–170.

3. Doudna JA, Charpentier E (2014) Genome editing. The new frontier of genome en-gineering with CRISPR-Cas9. Science 346(6213):1258096.

4. Jinek M, et al. (2012) A programmable dual-RNA-guided DNA endonuclease inadaptive bacterial immunity. Science 337(6096):816–821.

5. Ran FA, et al. (2013) Genome engineering using the CRISPR-Cas9 system. Nat Protoc8(11):2281–2308.

6. Hsu PD, Lander ES, Zhang F (2014) Development and applications of CRISPR-Cas9 forgenome engineering. Cell 157(6):1262–1278.

7. Luo ML, Leenay RT, Beisel CL (2016) Current and future prospects for CRISPR-basedtools in bacteria. Biotechnol Bioeng 113(5):930–943.

8. Barrangou R, van Pijkeren JP (2016) Exploiting CRISPR-Cas immune systems for ge-nome editing in bacteria. Curr Opin Biotechnol 37:61–68.

9. Mougiakos I, Bosma EF, de Vos WM, van Kranenburg R, van der Oost J (2016) Nextgeneration prokaryotic engineering: The CRISPR-Cas Toolkit. Trends Biotechnol 34(7):575–587.

10. Li H, et al. (2016) CRISPR-Cas9 for the genome engineering of cyanobacteria andsuccinate production. Metab Eng 38:293–302.

11. Woese CR, Kandler O, Wheelis ML (1990) Towards a natural system of organisms:Proposal for the domains Archaea, Bacteria, and Eucarya. Proc Natl Acad Sci USA87(12):4576–4579.

12. DeLong EF, Pace NR (2001) Environmental diversity of bacteria and archaea. Syst Biol50(4):470–478.

13. Offre P, Spang A, Schleper C (2013) Archaea in biogeochemical cycles. Annu RevMicrobiol 67(1):437–457.

14. Liu Y, Whitman WB (2008) Metabolic, phylogenetic, and ecological diversity of themethanogenic archaea. Ann N Y Acad Sci 1125(1):171–189.

15. Costa KC, Leigh JA (2014) Metabolic versatility in methanogens. Curr Opin Biotechnol29(1):70–75.

16. Kohler PRA, Metcalf WW (2012) Genetic manipulation of Methanosarcina spp. FrontMicrobiol 3:259.

17. Guss AM, Rother M, Zhang JK, Kulkarni G, Metcalf WW (2008) New methods fortightly regulated gene expression and highly efficient chromosomal integration ofcloned genes for Methanosarcina species. Archaea 2(3):193–203.

18. Guss AM, Mukhopadhyay B, Zhang JK, Metcalf WW (2005) Genetic analysis of mchmutants in two Methanosarcina species demonstrates multiple roles for the meth-anopterin-dependent C-1 oxidation/reduction pathway and differences in H(2) me-tabolism between closely related species. Mol Microbiol 55(6):1671–1680.

19. Pritchett MA, Zhang JK, Metcalf WW (2004) Development of a markerless geneticexchange method for Methanosarcina acetivorans C2A and its use in construction ofnew genetic tools for methanogenic archaea. Appl Environ Microbiol 70(3):1425–1433.

20. Rother M, Boccazzi P, Bose A, Pritchett MA, Metcalf WW (2005) Methanol-dependentgene expression demonstrates that methyl-coenzyme M reductase is essential inMethanosarcina acetivorans C2A and allows isolation of mutants with defects inregulation of the methanol utilization pathway. J Bacteriol 187(16):5552–5559.

21. Buan NR, Metcalf WW (2010) Methanogenesis by Methanosarcina acetivorans in-volves two structurally and functionally distinct classes of heterodisulfide reductase.Mol Microbiol 75(4):843–853.

22. Galagan JE, et al. (2002) The genome of M. acetivorans reveals extensive metabolicand physiological diversity. Genome Res 12(4):532–542.

23. Bartlett EJ, Brissett NC, Doherty AJ (2013) Ribonucleolytic resection is required forrepair of strand displaced nonhomologous end-joining intermediates. Proc Natl AcadSci USA 110(22):E1984–E1991.

24. Bartlett EJ, Brissett NC, Plocinski P, Carlberg T, Doherty AJ (2016) Molecular basis forDNA strand displacement by NHEJ repair polymerases. Nucleic Acids Res 44(5):2173–2186.

25. Metcalf WW, Zhang JK, Shi X, Wolfe RS (1996) Molecular, genetic, and biochemicalcharacterization of the serC gene of Methanosarcina barkeri Fusaro. J Bacteriol178(19):5797–5802.

26. Gomaa AA, et al. (2014) Programmable removal of bacterial strains by use of ge-nome-targeting CRISPR-Cas systems. MBio 5(1):e00928–e13.

27. Srinivasan G, James CM, Krzycki JA (2002) Pyrrolysine encoded by UAG in Archaea:Charging of a UAG-decoding specialized tRNA. Science 296(5572):1459–1462.

28. Kahnt J, et al. (2007) Post-translational modifications in the active site region ofmethyl-coenzyme M reductase from methanogenic and methanotrophic archaea.FEBS J 274(18):4913–4921.

29. Wang T, Wei JJ, Sabatini DM, Lander ES (2014) Genetic screens in human cells usingthe CRISPR-Cas9 system. Science 343(6166):80–84.

30. Fu Y, et al. (2013) High-frequency off-target mutagenesis induced by CRISPR-Casnucleases in human cells. Nat Biotechnol 31(9):822–826.

31. White MF (2011) Homologous recombination in the archaea: The means justify theends. Biochem Soc Trans 39(1):15–19.

32. West SC (1997) Processing of recombination intermediates by the RuvABC proteins.Annu Rev Genet 31(1):213–244.

33. Della M, et al. (2004) Mycobacterial Ku and ligase proteins constitute a two-compo-nent NHEJ repair machine. Science 306(5696):683–685.

34. Cui L, Bikard D (2016) Consequences of Cas9 cleavage in the chromosome of Es-cherichia coli. Nucleic Acids Res 44(9):4243–4251.

35. Youngblut ND, et al. (2015) Genomic and phenotypic differentiation among Meth-anosarcina mazei populations from Columbia River sediment. ISME J 9(10):2191–2205.

36. Li Y, et al. (2016) Harnessing type I and type III CRISPR-Cas systems for genome ed-iting. Nucleic Acids Res 44(4):e34.

37. Pawluk A, et al. (2016) Inactivation of CRISPR-Cas systems by anti-CRISPR proteins indiverse bacterial species. Nat Microbiol 1(8):16085.

38. Sowers KR, Boone JE, Gunsalus RP (1993) Disaggregation of Methanosarcina spp. andgrowth as single cells at elevated osmolarity. Appl Environ Microbiol 59(11):3832–3839.

39. Metcalf WW, Zhang JK, Wolfe RS (1998) An anaerobic, intrachamber incubator forgrowth of Methanosarcina spp. on methanol-containing solid media. Appl EnvironMicrobiol 64(2):768–770.

40. Kim SY, et al. (2012) Different biosynthetic pathways to fosfomycin in Pseudomonassyringae and Streptomyces species. Antimicrob Agents Chemother 56(8):4175–4183.

41. Gibson DG, et al. (2009) Enzymatic assembly of DNA molecules up to several hundredkilobases. Nat Methods 6(5):343–345.

42. Kearse M, et al. (2012) Geneious Basic: An integrated and extendable desktop soft-ware platform for the organization and analysis of sequence data. Bioinformatics28(12):1647–1649.

43. Metcalf WW, Zhang J, Apolinario E, Sowers KR, Wolfe RS (1997) A genetic system forArchaea of the genus Methanosarcina: Liposome-mediated transformation andconstruction of shuttle vectors. Proc Natl Acad Sci USA 94(6):2626–2631.

44. Boccazzi P, Zhang JK, Metcalf WW (2000) Generation of dominant selectable markersfor resistance to pseudomonic acid by cloning and mutagenesis of the ileS gene fromthe archaeon Methanosarcina barkeri fusaro. J Bacteriol 182(9):2611–2618.

45. Barrick JE, et al. (2014) Identifying structural variation in haploid microbial genomesfrom short-read resequencing data using breseq. BMC Genomics 15(1):1039.

Nayak and Metcalf PNAS | March 14, 2017 | vol. 114 | no. 11 | 2981

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Aug

ust 2

5, 2

020