Active in vivo translocation of the Methanosarcina mazei Gö1 Casposon

Abstract Casposons are transposable elements containing the CRISPR associated gene Cas1solo. Identified in many archaeal genomes, casposons are discussed as the origin of CRISPR-Cas systems due to their proposed Cas1solo-dependent translocation. However, apart from bioinformatic approaches and the demonstration of Cas1solo integrase and endonuclease activity in vitro, casposon transposition has not yet been shown in vivo. Here, we report on active casposon translocations in Methanosarcina mazei Gö1 using two independent experimental approaches. First, mini-casposons, consisting of a R6Kγ origin and two antibiotic resistance cassettes, flanked by target site duplications (TSDs) and terminal inverted repeats (TIRs), were generated, and shown to actively translocate from a suicide plasmid and integrate into the chromosomal MetMaz-C1 TSD IS1a. Second, casposon excision activity was confirmed in a long-term evolution experiment using a Cas1solo overexpression strain in comparison to an empty vector control under four different treatments (native, high temperature, high salt, mitomycin C) to study stress-induced translocation. Analysis of genomic DNA using a nested qPCR approach provided clear evidence of casposon activity in single cells and revealed significantly different casposon excision frequencies between treatments and strains. Our results, providing the first experimental evidence for in vivo casposon activity are summarized in a modified hypothetical translocation model.


INTRODUCTION
Transposons are mobile genetic elements (MGEs), found in the genomes of organisms of all domains of life. They occur in high quantities in plants, but are also present in animals, bacteria, and archaea (1)(2)(3)(4). In general, MGEs are one major factor of evolution and adaptation, since their translocation within chromosomal and extra-chromosomal DNA can result in all kinds of mutations from large scale chromosomal rearrangements to single nucleotide polymorphisms (SNPs), often leading to gene silencing or generation of new coding genes ( 5 , 6 ). MGEs include all kinds of elements that can change their genomic position within a host genome or between host cells, including viruses, plasmids and transposons. Transposons are classified by their translocation mechanism -Class I comprising retrotransposons and Class II DNA transposons -and are further distinguished by their length and gene content (7)(8)(9). Class II transposons, which translocate primarily via a cut-and-paste mechanism ( 8 , 9 ), include both short insertion sequence (IS) elements, which often encode for only one transposase gene (e.g. Tc1 mariner superfamily elements) and large elements longer than 10 kb (e.g. Mavericks / Polintons) ( 8 , 10-12 ).
The distribution of MGEs in the archaeal domain is v ery une v en, since most of the known transposable elements ar e r estricted to the orders of Halobacteriales , Sulfolobales , Thermoplasmatales and Methanosarcinales ( 3 ). Methanosar cinales , particularly Methanosar cina mazei, have been shown to carry a large number of transposable elements r epr esented by 102 identified transposase encoding genes (mostly IS elements) ( 13 ). Within the last two decades, transposable elements were shown to be connected to adapti v e immunity of eukaryotes and prokaryotes. In eukaryotes the key enzyme of V(D)J recombination (RAG1) was found to be deri v ed from a T r ansib transposon ( 14 , 15 ). In contrast to this, clustered regularly interspaced short palindromic repeats (CRISPR) with CRISPR associated (Cas) proteins and complexes are representati v es of an adapti v e defense system in prokaryotes, with the ability to incorporate so-called spacer sequences deri v ed from invader genomes into CRISPR arrays for future defense of the same invader by targeted degradation (re vie wed in ( 16 )). CRISPR deri v ed immunity, with its three phases of targeted defense -adaptation, expression and interference -is completely dependent on expression of Cas proteins (re vie wed in ( 16 )). With the discovery of many different CRISPR-Cas systems showing clear homologies in functions and mechanisms, the question regarding the evolutionary origin of those systems has become a focus of attention. Solitary family members of the key CRISPR adaptation protein Cas1 were found on transposon-like elements, hence designated casposons, which led to the proposal that transposons contributed to the origin of CRISPR-Cas systems ( 17 ). While transposases are essential for the activity of typical transposable elements, this type of enzyme has not been found to be encoded in casposons. The only encoded enzyme common to all discovered casposons is Cas1solo or the so-called casposase, which was proposed to mediate the casposon translocation ( 17 ). This new class of DNA transposons has been proposed to be the first self-synthesizing member of DNA transposons in prokaryotes and bears some similarity to Polintons / Mavericks, due to the common type B polymerases in addition to their enormous size of se v eral kilobase pairs ( 10 , 11 , 17 ). To date, four distinct casposon families are described and classified mainly by the domain structure of their Cas1solo enzyme ( 18 , 19 ). Krupovic and colleagues proposed a casposon translocation model close to the transposition of Polintons / Mavericks ( 11 , 17 ). They predicted a TIR mediated looping out of the casposon from the lagging strand during cellular DNA replication, followed by Cas1solo mediated excision, polB replication and Cas1solo media ted reintegra tion into the genome ( 17 , 20 , 21 ). The reintegr ation by gener ation of target site duplications was proposed to be very similar to the functions of the spacer adaptation module of the CRISPR-systems ( 17 ). Krupovic et al. wer e pr edicting a persistence of the target site duplication post excision ( 17 ). Although casposons were discovered almost ten years ago, final proof of principle is still lacking. Characterizations of casposon functions have been limited to in vitro assays, mainly based on heterolo gousl y expressed Cas1solo proteins analyzed for specific activities (22)(23)(24)(25).
Extensi v e studies of the casposon identified in Acidulipro-fundum boonei , especially biochemically characterizing its Cas1solo variant, gave first insights into the translocation mechanism in vitro and suggested casposon activity in vivo (23)(24)(25). Various experiments with purified heterolo gousl y expressed Cas1solo of A. boonei and M. mazei showed that the enzymes were able to integrate substrates like casposon derivates or short synthetic oligonucleotides in a sitespecific process into different target sites present on pUC19 deri v ed acceptor plasmids ( 19 , 22-24 ). The site-specific integration showed indeed similarities to the integration of new spacer sequences into CRISPR arrays during the adaptation step of the CRISPR-Cas immunity process ( 17 , 22 ). To date, all studies conducted on casposons have lacked in vivo proof of principle. This is due to the fact that convenient model systems are exceedingly rare, as the number of casposon carrying host organisms, mostly archaeal species, which are culturable in laboratories is still very low as is the number of genetically tractable systems ( 17 , 19 ). Consequently, the aim of the here presented study was to detect in vivo activity of the M. mazei casposon MetMaz-C1 first described by Krupovic and colleagues ( 19 ). MetMaz-C1 has a size of a pproximatel y 10 kb, encodes eight genes and is flanked by a TIR of 31 nt (GGGA TA TAGGTAA CTCAAAAAAA CG-CAACGC) and TSD of 14 nt (5-A TAA TCTTAA TGCG-3) ( ( 17 , 19 ); Figure 1 A).
Her e, we r eport on the casposon MetMaz-C1 being an acti v e MGE. We were able to characterize and quantify its excision frequency in vivo . For investigation of the casposon activity we established and used an optimized in vivo minicasposon assay in M. mazei G ö1 based on preceding in vitro studies (22)(23)(24)(25). Our results show that MetMaz-C1 can excise by leaving an empty target site and acti v ely integrates into new genomic loci or forms tandem structures by integrating into the original target sites. Based on our findings, we suggest to modify the previous model for translocation predicted by Krupovic and colleagues ( 17 ).

Description of plasmids used in this study
All cloning strategies in this study were based on restriction cloning, for which all major restriction enzymes were purchased from New England Biolabs (NEB, Ipswich, MA, USA). In addition, all chemicals used in this study were purchased from Carl Roth GmbH + Co . KG (Karlsruhe , Germany) unless otherwise stated.

Cloning of mini-casposon and cas1solo o ver expr ession plasmid
The mini-casposon suicide vector (pRS1520) was generated based on the gene synthesis of the construct TSD-TIR-NdeI-ApaI-R6K ␥ origin-TIR-TSD inserted in a pEX-A256 backbone obtained from Eurofins Genomics Ger many GmbH (Ebersberg, Ger many). The kanamycin resistance cassette was cloned into the ApaI site. The mini-casposon sequence was amplified by PCR using LongAmp ® Taq DN A Pol ymerase (NEB) according to manufacturer's protocol for deletion of the irrelevant pEX-A256 backbone. Both primer sequences Minicas SDM for and Minicas SDM re v (Supplementary Tab le S1) including  ( 17 , 19 ). MetMaz-C1 encodes 8 genes, highlighted are the genes encoding the key enzyme Cas1solo ( MM RS16885 ; blue) and the type B Polymerase ( MM RS16905 ; orange). ( B ) pRS1520 (mini-casposon on suicide plasmid) consisting of the MetMaz-C1 target site duplications (TSDs; red; 5-A TAA TCTTAA TGCG-3) and terminal inverted repeats (TIRs; light gray; 5-GGGA TA TAGGTAA CTCAAAAAAA CGCAA CGC-3), flanking a R6K ␥ origin (gray striped), a kanamycin (gray) and neomycin resistance cassette (dark gray). Sequencing primers mcs1 and mcs2 used for testing of mini-casposon integration based on rescue cloning are depicted (black arrows). ( C ) pRS1270, Cas1solo ov ere xpression plasmid with the nati v e cas1solo gene ( MM RS16885 ) under the control of the constituti v e promotor mcrB (pmcrB) and the respecti v e ribosome binding site (RBS) in a pWM321 backbone ( 26 ). Sequencing primers (M13 for / rev; light blue) were used for cross contamination tests in the long-term evolution experiment. ( D ) pRS1511, based on the pCRII vector (TOPO ™ TA Cloning ™ Kit, Thermo Fisher Scientific, Waltham, MA, USA), was used as a control and for the normalization in casposon excision frequency (CEF) determination by nested qPCRs. The upstr eam (5 -r egion) and downstr eam r egions of the nati v e casposon MetMaz-C1 (3 -r egion) wer e PCR amplified and cloned into pCRII by restriction (using BamHI / XbaI). Primer sequences for the nested qPCRs are depicted as follows: primer set A (magenta; 1 st PCR) and primer set A* (light gray; 2 nd PCR). ( E ) pRS1713 used to determine the genome copy numbers for normalization of casposon excisions per genome. pRS1713 was generated by TOPO-TA cloning of MM RS06290 (bifunctional hexulose-6-phosphate synthase) into pCRII (TOPO ™ TA Cloning ™ Kit, Thermo Fisher Scientific, Waltham, MA, USA). qPCR primers, 1713 for and 1713 rev, targeting the introduced gene are depicted in yellow.
terminal NheI restriction sites were binding to the plasmid upstr eam or downstr eam of the mini-casposon. Primers were annealed to the template at 68 • C for 30 s, followed by elongation of the PCR products at 65 • C for 5 min and a total of 35 cycles. E. coli DH5 α cells were transformed with the resulting 2 kb PCR products, which were self-ligated post NheI restriction. The neomycin resistance cassette was PCR amplified from pRS830 using the primer set NeoR for and NeoR rev (Supplementary Table S1) and cloned into the NdeI site resulting in the final mini-casposon suicide vector pRS1520 (Figure 1 B). Two oligonucleotides were designed to verify positive translocation into the M. mazei genome and to exclude integration of the complete plasmid into the chromosome: mini-casposon sequencing primers mcs1 and mcs2 (Figure 1 B; Supplementary Table S1).
Constituti v e Cas1solo ov ere xpression plasmid (pRS1270, Cas1solo OP, Figure 1 C) was generated based on PCR amplification of MM RS16885 ( cas1solo ) using the primer set MM RS16885 for and MM RS16885 rev (Supplementary Table S1), and genomic DN A (gDN A) of M. mazei as PCR template. The PCR product (1236 bp) was cloned via NcoI and KpnI restriction into pRS1248 (pDRIVE backbone) under the control of the mcrB promotor (pmcrB) and with the addition of the mcrB ribosome binding site (RBS). Constructed fragments of pmcrB -RBS -cas1solo were cloned into the shuttle vector pWM321 ( 26 ) using KpnI and SacI, resulting in pRS1270 (Figure 1 C). During this study M. mazei mutants were generated following the liposome-mediated transformation protocol of Ehlers and colleagues ( 27 ). Mutant strains were stably transformed with either pRS1270 (Mm Mut 208) or the empty pWM321 vector (Mm Mut 203b) as a control. Both strains were used side by side in a long-term e volution e xperiment, which r equir ed checking for cross contaminations between the strains. For this purpose, PCR contamination tests of subsamples with M13 primers (for / r ev) wer e performed (Figur es 1 C and 2 B). The PCR products were clearly separatable by gel electrophoresis. PCR fragments deri v ed from the empty v ector control were approx. 450 bp in size, while the PCR product of pRS1270 was roughly 1800 bp.

Cloning of quantitative (q) PCR control plasmids pRS1511 and pRS1713
In the current study optimized nested PCR and qPCR protocols were established using plasmids as PCR controls or in case of qPCRs a plasmid normalization for absolute quantification. For this purpose, two different control plasmids were generated. pRS1511 was designed as a PCR control and for normalization of nested qPCRs to determine the casposon excision frequency (CEF). Two PCR products with terminal restriction sites and a length of 400 bp each of the upstream and downstream regions of the nati v e casposon MetMaz-C1 were generated. The first 400 bp fragment was amplified by using the primers US BamHI for and US SpeI rev (Supplementary Table S1), and M. mazei gDNA. The second PCR fragment was generated by using the same template with primers DS SpeI for and DS XbaI rev (Supplementary Table S1). The first PCR fragment was digested by BamHI and SpeI and the second fragment by SpeI and XbaI. Both fragments were cloned to BamHI and XbaI linearized pCRII vector (Figure 1 D). The final plasmid (pRS1511) was used in the nested qPCRs for normalization by standar d curv e. The plasmid was targeted with the two primer sets A 1 :A 2 and A 1 *:A 2 * (Figure 1 D; Supplementary Table S1).
The second qPCR normalization plasmid (pRS1713) was also based on the pCRII-TOPO vector but was used for the estimation of genome copy numbers for normalization to genome numbers in qPCR reactions. The house-keeping gene MM RS06290 encoding the bifunctional hexulose-6-phosphate synthase was PCR amplified from M. mazei gDNA using MM RS06290 for and MM RS06290 rev (Supplementary Table S1) and TA cloned to pCRII ( Table S1.

Mini-casposon in vivo assay and rescue cloning
For investigation of the mini-casposon translocation in a plasmid-based assay, M. mazei G ö1 wildtype (DSM3647) was transformed with the generated pRS1520 suicide vector carrying the mini-casposon by liposome-mediated transforma tion (Figure 1 B) ( 27 ). Optimiza tions, dif fering from the published methods of Ehlers and colleagues ( 27 ) are highlighted in the following. 50 ml log phase growing M. mazei culture was harvested by centrifugation for 10 min at 3.000 x g in oxygen free atmosphere and carefully resuspended in sucrose buffer (10 mM MES, 0.15 M sucrose, 6.3 pH). 1 g pRS1520 was diluted in 50 l sucrose buffer and added to mixture of 30 l DOTAP (Roche Holding AG, Basel, Switzerland) and 70 l sucrose buffer, followed by an incubation for > 30 min a t 37 • C . 990 l of resuspended cells were added to sucrose-DN A-DO TAP mixture. Liposome-media ted transforma tion reactions were incubated for 5 h under anaerobic conditions and supplemented with 5 mM MgCl 2 and 5 mM MnCl 2 after 3 h. The cells were split to roughly 500 l per reaction and transferred to 5 ml complemented minimal medium containing additionally 0.1 M sucrose. Cells were incubated over night at 37 • C. New media containing neomycin (22 M) were inoculated by transferring 500 l overnight cultures. Cultur es wer e incubated f or 14-21 da ys until cell growth could be detected. Growing populations were transferred to new media immediately and reinoculated se v eral times until gDNA for mini-casposon translocation analysis was isolated using Wizard Kit (Promega, Madison, WI, USA) as recommended. gDNA was dissolved in RNAse free water. Purified gDNAs were first analyzed by PCR using mcs1 and mcs2 primers to exclude potential false positi v es resulting from single-crossover events of pRS1520 (Figure 1 B) with M. mazei genome e.g. by homologous sequences (e.g. IS-elements).
Populations with no detectable PCR fragments from the vector backbone were further analyzed for the presence of the mini-casposon by rescue cloning. For this purpose, 1 g of isolated gDNA was restriction digested by AccI to Figure 2. Set-up of the long-term evolution experiment. ( A ) Pr epar ed minimal medium was split and complemented according to the respective treatment (500 mM NaCl, 0.5 g / ml mitomycin C, 40 • C, untreated controls (nati v e; +N)). 1.4 ml respecti v e medium was filled in used wells (gray) of 96-deep-well plates according to depicted pipetting scheme to reduce contaminations between treatments. Strains Mm Mut 203b, carrying an empty pWM321 vector, and Mm Mut 208, carrying the Cas1solo ov ere xpression plasmid pRS1270, were cultivated under anaerobic conditions on separate deep-well plates, to exclude cross contamination. Medium was inoculated by transferring 50-200 l of pr eceding cultur es under consideration of roughly estimated cell densities. Plates were incubated for 4-5 days under appropriate treatment conditions (37 • C or 40 • C) and used for inoculations of the next round of plates. Cells were harvested by centrifugation for DNA isolation or were mixed with glycerol f or cry o-storage at − 80 • C. The long-term evolution experiment was running for 472 days r epr esenting a minimum of a pproximatel y 600 generations. ( B ) Post inoculation of new cultures, genomic DNA was isolated from the remaining cultures using the Wizard ® SV 96 Genomic DNA Purification System (Promega, Madison, WI, USA) as described in the method section. DNA quality and quantity were verified by agarose gel electrophoresis. Casposon activity based on casposon excision frequency (CEF) was tested by nested qPCR reactions targeting flanking chromosome regions. PCR fragments were only generated in case of an empty casposon locus, due to short elongation times (0.5-1 min) and MetMaz-C1 size (approx. 10 kb) resulting in a final PCR product of 239 bp. Primer set A was used for the 1 st PCR, Primer set A* for the 2nd PCR ( Figures 1 D and 3 ). Quantification was performed under consideration of genome copy numbers in qPCR reactions. completeness in a total volume of 50 l using 20 U enzyme at 37 • C for 2 h, followed by heat inactivation at 80 • C for 20 min. Reactions were incubated on ice for 5 min and split into three T4 ligation reactions. T4 ligation reactions had a final volume of 20 l with 0.1 U / l ligase and were incubated for 1 h at room temperature (RT) prior transformation in chemical competent One Shot ™ PIR1 (Thermo Fisher Scientific, Waltham, MA, USA) and selection for kanamycin resistance. Single clones were selected and sequenced via Sanger sequencing technology using mcs1 and mcs2 for sequencing out of the mini-casposon part into the potential genomic insertion site.

Long-term evolution experiment -strain generation, growth conditions and setup
Single clones of M. mazei transformed with pRS1270 (Cas1solo OP, Figure 1 C) or pWM321 (empty vector control, ( 26 )) were generated like described ( 27 ). A single colony of M. mazei carrying pRS1270 was named Mm Mut 208, whereas empty vector control (pWM321) was designated Mm Mut 203b. Both strains were grown in an oxygen fr ee atmospher e in 96-deep-well plates with 1.4 ml minimal medium per well (see Figure 2 A) according to the methods adapted from ( 28 , 29 ). Sterile deep-well plates were stored in a sterile anaerobic chamber for three weeks prior usage in the experiment and were further 'pr e-r educed' with 1 ml of sterile pr e-r educing solution (PRS: 6.3 mM Na 2 S, 18.6 mM cysteine, 7.5 mM NaOH) per well for 2 h, followed by decanting and drying for 30 min. Minimal medium was flushed with sterile N 2 :CO 2 (80:20) for 10 min and supplemented with methanol (150 mM), acetate (40 mM), Na 2 S (1 mM), cysteine (2 mM), puromycin dihydrochloride (9 M; Merck, Dar mstadt, Ger many) and ampicillin (0.3 mM) or kanamycin sulfate (52 M) as described before (( 30 )); Figure 2 A). The bacteria-targeting antibiotics -kanamycin and ampicillin -were used alternating e v ery second inoculation. According to the treatments, medium was either complemented with 500 mM NaCl (sodium chloride treatment) or 0.5 g / ml mitomycin C (mitomycin treatment; Fisher Scientific GmbH, Schwerte, Germany), or was incuba ted a t 40 • C (tempera tur e tr eatment) or left untr eated (nati v e; +N) (Figure 2 A). The strains were split to different plates and treatments (+N; NaCl treatment; mitomycin tr eatment) wer e separated by two empty rows (Figur e 2 A). Temperatur e tr eated cultur es wer e inocula ted on a separa te 96-deep-well plate. 12 replicates per strain and treatment were inoculated with 50-200 l corresponding preceding cultur e (Figur e 2 A). The cultur e plates wer e closed with needle perforated silicon covers and custom-made autoclavable aluminum pr essur e plates (AAPP) to avoid evaporation and contamination. Pr epar ed cultur es wer e incubated for 4-5 days at 37 • C (+N; NaCl; mitomycin) in an anaerobic chamber or in an anaerobic pot at 40 • C (temperature treatment) with the same gas atmosphere (N 2 :CO 2 / 80:20) until the next set of respective media was inoculated. After inoculating the next cultures, the remaining cultures were used to isolate gDNA and occasionally to generate a cryo-storage by storing mixtures of 60 l of the respecti v e culture and 25 l sterile anaerobic glycerol (86%) in 96-microtiter plates at − 80 • C (Figure 2 A). The long-term evolution experiment ran for 472 days, which corresponds to approximately 600 generations by interpolating the lowest doubling time. The doubling time ranged around 16.5 ± 2.5 h for the different tr eatments (OD 600 measur ements and doubling time calculations can be found in the supplement: Supplementary  Table S2).
For gDNA isolation, all cultur es wer e transferr ed together to one single 96-deep-well plate and harvested by centrifuga tion a t 3220 x g and 4 • C for 20 -30 min (Figure 2 B). Since M. mazei cell pellets are not solid but rather mucoid, the supernatant was carefully discarded by pipetting with a self-de v eloped 3D-printed pellet saving spacer unit (PSSU) (Figure 2 B). Pellets were washed once with 1 ml of phosphate buffered saline (PBS: 0.73 M NaCl, 0.31 M Na 2 HPO 4 , pH 7.5), followed by a second centrifugation step. Supernatant was again discarded by pipetting using the PSSU (Figure 2 B). Pellets were carefully resuspended in the remaining PBS volume of roughly 50 l by shaking at 500 rpm for 20 min and stored at −20 • C in microtiter plates or immediately processed with Wizard ® SV 96 Genomic DNA Purification System (Promega) using Vac-Man ® 96 Vacuum Manifold (Promega). Manufacturer protocol was adjusted according to cell pellet size and cell pellet morphology. Pellets w ere thaw ed for 10 min, car efully r esuspended in 200 l SV-Lysis buffer to avoid foaming and incubated at RT for 10 min. 275 l mixture was transferred to the bind-ing plate. After binding, washing and drying, gDNA was eluted with 175 l water (0.8 % RNAse) and incubated for 10 min at RT. 60 l aliquots of elutions wer e stor ed at 4 • C and −20 • C (Figure 2 B).
For quality assessment culture plates were PCR screened with M13 (for / rev) primers (Supplementary Table S1) targeting the pWM321 as well as the pRS1270 monthly (Figures 1 C and 2 B). Each batch of minimal medium was tested by inoculation of 5 ml complemented medium with M. mazei DSM3647. The pr epar ed mitomycin C stock solutions were tested for its ability to inhibit growth of Mm Mut 203b at 2.5 g / ml with three independent biological replicates. Puromycin stock solution was tested weekly.

Long-term evolution experiment: determination of excision frequency by a nested qPCR approach
In this stud y, dif fer ent PCR protocols wer e set up for different purposes. The primers were designed to target the nati v e MetMaz-C1 locus in the M. mazei genome. Primers belonging to set A were targeting flanking casposon regions, while combinations of set A primers with set B or set C primers were targeting the casposon TSDs IS1a or IS1b (Figure 3 A). All primer sets were designed with two different annealing temperatures and target sequences for setting up PCRs or nested qPCRs, since single PCR reactions were found to be not sensiti v e enough to detect casposon e xcision e v ents on single cell le v el. To detect empty target sites as a hint for positi v e e xcision and for the determination of the casposon e xcision frequency (CEF) primer set A was used (Figure 3 B).
Selected samples were screened for casposon excision e v ents using a nested qPCR approach (Figures 2 B and 3 ). First PCR reaction was set up using 3 l gDNA of culture of interest or dilution of pRS1511 control plasmid. The protocol of the 1 st PCR was based on manufacturer protocol of GoTaq Polymerase (Promega), with a final volume of 25 l and the following PCR protocol: initial dena tura tion a t 95 • C for 3 min, cycle dena tura tion a t 95 • C for 30s, annealing at 47 • C for 15 s, cycle elongation at 72 • C for 1 min, final elonga tion a t 72 • C for 5 min. The primers were used a t final concentrations of 0.4 M per oligonucleotide. Primer set A 1 :A 2 was used to target the upstream and downstream region of the known casposon location in the M. mazei genome (3946587-3956667) ( Figure 3 ). 5 l of the 1 st PCR product was used as template for the 2 nd qPCR with a final volume of 25 l. Plasmid normalized qPCR reactions were set up using PowerUP SYBR Green Mastermix (Thermo Fisher Scientific, Waltham, MA, USA) as recommended. Primers A 1 * and A 2 * were used at a final primer concentration of 0.2 M per oligonucleotide. qPCR program was similar to the first PCR protocol, except that the annealing temperature was 55 • C for 19 s and cycle elongation was set to 30 s. Both PCR protocols were run for 35 cycles. The qPCR protocol included a melt curve analysis using the following steps: 95 • C for 15 s, 60 • C for 60 s and 95 • C for 15 s with temperature gradients at 1.6 • C / s. The quantification of nested qPCRs based on the cycle threshold (ct) values was performed by normalization to a 1:10 serial dilution of previously Qubit measured concentration of pRS1511 control plasmid (1x DNA Broad Range Kit) according to methods described elsewhere ( 31 ). pRS1511 insert consist-  ing out of the upstream and downstream casposon regions was targeted with the same primer sets as described for the casposon locus mentioned above (Figures 1 D and 3 ).
To compar e differ ent biological r eplicates , strains , or time points, normalization to the pRS1713 control plasmid was performed to quantify M. mazei genome numbers (copy numbers) per reaction by qPCR in parallel using the same samples and sample volumes. The qPCR protocol was the same as for the 2 nd qPCR described above, but the normalization plasmid and the primers wer e differ ent. A genome quantification control plasmid (pRS1713) was designed by cloning MM RS06290 (bifunctional hexulose-6-phosphate synthase) into pCRII-TOPO according manufacture protocol (Figure 1 E). Primers, 1713 for and 1713 r ev, wer e targeting the MM RS06290 sequence (Figure 1 E). Serial dilution was set up in similar manner as described for pRS1511.

Statistical analysis of casposon e x cision frequency (CEF) values determined for different samples
For the CEF quantification within ancestor strains, gDNA of freshly inoculated strains with six technical replicates was used, normalizing to three technical replicates of the serial dilutions of pRS1511 and pRS1713. For statistical analysis a two-tailed t-test was performed in GraphPad Prism [v9.3.1 for Mac, GraphP ad Softwar e, San Diego, CA, USA]. For comparisons between the CEF values deter mined for long-ter m e volution e xperiment samples, the same number of technical replicates was used for pRS1511 and pRS1713 dilutions, but with three selected biological replicates for each condition and strain with fiv e technical replicates each. The resulting data was analyzed in Graph-Pad by performing a two-way ANOVA under consideration of multiple comparisons using Š íd ák's correction. All samples for quantification within one sample set normalized to one of the plasmids were run on the same qPCR plate, to exclude run and plate effects.

Sequencing and bioinformatics
Plasmids and PCR products were sequenced via Sanger sequencing at the Institute of Clinical Molecular Biology (IKMB, Kiel, Germany). The Sanger sequencing results were analyzed by BLAST and r efer ence-based alignments using Bowtie2 [v2. 4 ( 38 ). The polished genome assemblies were manually adjusted in Geneious prime by circularization, followed by manually adjusting genome start site by mapping of first 100 bp of M. mazei NC 003901.1 r efer ence to the respecti v e assemb ly. Genomes were re v erse complemented if necessary. The annotation was performed using NCBI Prokaryotic Genome Annotation Pipeline (PGAP) (39)(40)(41).

RESULTS
We obtained two independent lines of evidence that the M. mazei casposon actively translocates in the M. mazei genome: (i) by using a mini-casposon approach and (ii) by evaluating the chromosomal position of the casposon in M. mazei populations during an evolution experiment. The mini-casposon assay was based on a suicide vector carrying a synthetic mini-casposon with a neomycin resistance cassette. During transformation of M. mazei cells, the minicasposon should be able to translocate to the M. mazei chromosome resulting in neomycin resistance. The acti v e translocation was analyzed and verified by a PCR and a rescue-cloning approach, followed by Sanger sequencing.
The second approach was focusing on a potential activity of the nati v e, chromosomal casposon MetMaz-C1. For this purpose , two strains , one empty vector control (Mm Mut 203b) and one Cas1solo ov ere xpression mutant (Mm Mut 208) were cultured under four different conditions (Figure 2 ). The casposon activity was characterized by its casposon excision frequency (CEF), which was determined by nested qPCRs (Figure 3 ). The nested qPCRs were designed to amplify empty target sites sensiti v e enough to detect single excision positive M. mazei genomes. The generation of PCR products at empty target sites was based on nested PCRs with very short elongation times, so that PCR fragments were generated only when the target sites were empty, which was due to the excision of the full-length casposon of 10 kb. First focus was gi v en to the ancestor cultures, which were used to inoculate the long-term evolution experiment, followed by the characterization of samples of the long-term e volution e xperiment. Additionally, generated various PCR products were sequenced to obtain information about the structure of the empty casposon locus and potential new integration sites.

A mini-casposon was actively translocating into the genome of M. mazei in vivo
To track a potential casposon translocation in vivo , a plasmid-based mini-casposon assay was established. The assay is based on the vector pRS1520, constructed during this study, carrying a mini-casposon (TSD-TIR-R6K ␥ origin-Kan R -Neo R -TIR-TSD) (Figure 1 B). M. mazei was transformed with the vector, allowing the mini-casposon if acti v e to translocate from the suicide v ector into the host genome using chromosomally expressed essential proteins encoded in MetMaz-C1 (e.g. Cas1solo) (Figure 4 A). For detection of positi v e mini-casposon integrations, M. mazei populations were selected for the mini-casposon mediated neomycin r esistance (Figur e 4 A). To rule out potential false positi v es due to homologous recombination, resistant populations were further characterized with respect to mini-casposon integration as follows. First, to exclude integration of the complete pRS1520 plasmid via single-crossov er e v ents, PCRs wer e used to scr een isolated gDNA of neomycin resistant populations for remaining suicide plasmids. For this purpose, PCRs were set up with the two primers targeting the mini-casposon in outside direction (mcs1, mcs2). Samples from popula tions tha t generated a PCR product r epr esenting an intact suicide vector backbone were excluded from downstream rescue cloning analysis. In these cases, it was very unlikely that the minicasposon was correctly translocated to the chromosome, but rather the plasmid integrated into the chromosome by a single-crossover. Secondly, selected populations and their gDNA samples were used for the rescue cloning of the casposon with flanking chromosomal DN A fragments, w hich was based on the mini-casposon mediated R6K ␥ origin and the kanamycin resistance cassette, which both were essential for replication and persistence of rescue-cloningderi v ed plasmids in E. coli . E. coli was transformed with the respecti v e circularized chromosomal DNA fragments, selected on kanamycin containing plates and characterized by restriction digestion, PCRs and Sanger sequencing. Sequence analysis of a final selection of plasmids unraveled mini-casposon integration e v ents and sites. The e xperimental procedure is summarized in Figure 4 A.
According to the outlined strategy and analysis pipeline, M. mazei wt was successfully transformed with pRS1520 r esulting in thr ee neomycin r esistant populations (Pop 1-3) (Figure 4 B). gDNA of those populations was first analyzed by PCRs using mcs1 and mcs2 primers targeting the mini-casposon ends (outside direction), amplifying the pRS1520 backbone (approx. 500 bp) (Figures 1 B and 4 B).
No PCR product was expected if the mini-casposon had translocated from the suicide plasmid to the chromosome, which was the case for Pop 1 and Pop 2 (Figure 4 B). In contrast to this, PCRs with gDNA of Pop 3 resulted in a clear PCR fragment indicating remnants of pRS1520, suggesting a persistent suicide plasmid as extra chromosomal DNA fragment or its integration into the chromosome by a single-crossover. To further distinguish between a directional mini-casposon translocation and a potential integration of the mini-casposon sequence by single-crossovers, the structure of the original MetMaz-C1 locus was analyzed by PCR using the described three primer sets A, B and C (Figure 3 A). If the mini-casposon integrated into the original MetMaz-C1 locus, these e v ents should be clearly detectable b y v ariation in the PCR fragment sizes in comparison to the controls. If the mini-casposon would have replaced the whole MetMaz-C1 sequence based on a potential homologous recombination due to TIR and TSD similarity in a doub le-crossov er, the PCR product generated using primer set A 1 :A 2 would have resulted in a PCR product smaller than the expected 10 kb. This was not the case for any of the tested populations (Figure 4 C, first lane for each population). In case that the mini-casposon integrated into one of the original TSDs of MetMaz-C1 generating a tandem structure, changes in size of the PCR products using primer set A 1 :B and C:A 2 should be visible. Comparisons of PCR fragment sizes of PCRs using gDNA of Pop 1 and 3 as templates with the gDNA control indicated no differences in the target sites, due to equal fragment lengths (Figure 4 C, lane 2 in each case). In contrast to these, the PCRs with Pop 2 and primer set A 1 :B were showing a strong difference To verify these results, Pop 1 and Pop 2 were selected for downstream analysis using the outlined rescue cloning approach and the analysis of the presence of the NheI restriction site. After transformation of self-ligation reactions of complete Acc1 digested M. mazei gDNA of Pop 1 and Pop 2 into E. coli , plasmids of single kanamycin-resistant clones were isolated and verified by restriction digestion with NheI. In case of a positi v e mini-casposon translocation into the host genome, the NheI site originally present in the pRS1520 backbone cannot be present in the rescuecloning-deri v ed plasmids (Figures 1 B and 4 D). Analysis of plasmids deri v ed from Pop 1 showed positi v e restriction di-gestion products, due to clear plasmid linearization (Figure 4 D). Pop 2 did not show any linearization, which suggested the absence of a NheI site in the rescue-cloningderi v ed plasmids of Pop 2. These results indica te tha t a true mini-casposon integration e v ent took place in Pop 2, while Pop 1 seemed to be the result of a single-crossover, although only a small fragment with additional low sequence homolo gy mainl y based on the TSDs (2 × 14 nt) and TIRs (2 × 38 nt) between the suicide vector and the M. mazei genome was available for homologous recombination. A potential increase of sequence similarity could be achie v ed by acti vity of a host deri v ed IS-element during the transformation process (Figure 4 D, Pop 1). In case of a translocation of an acti v e IS-element into the fresh introduced plasmid pRS1520, the ability of a homologous recombination of pRS1520 and the M. mazei genome would have been increased. These plasmids would be able to integrate into the chromosome by single-crossover as indicated in Figure 4 D, Pop 1).  Table S3).
To further verify the obtained results, as well as to address the IS-element hypothesis, rescue-cloning-deri v ed plasmids of Pop 1 and Pop 2 were sequenced via Sanger sequencing using mcs1 and mcs2 primers (Figures 1 B and 4 E; Supplementary Table S4, mini-casposon integration). Alignments of Sanger sequencing reads to the M. mazei reference (NC 003901.1) using Bowtie2 in Geneious prime indica ted further dif fer ences between r escue-cloning-deri v ed plasmids from Pop 1 and 2. Three different regions of integration (ROI 1-3) were detected (Figure 4 E). Reads from Pop 1 showed two regions of integration in the M. mazei genome at two different sites (ROI 1 + ROI 2; Figure 4 E gray boxes), but further showed remnants of the suicide vector backbone e.g. the NheI site between the TSDs, which was expected based on the restriction results (Figure 4 D, E). Sequencing r eads dir ected from the R6K ␥ origin showed breakage after the second TSD and showed similarity to M. mazei IS-elements found in ROI 1 and ROI 2 (Figure 4 E, gray boxes). These IS-elements belong to the families IS66 ( MM RS16865 ) and IS1634 ( MM RS15760 ) (Figure 4 E). The integration close to these IS-elements supported the hypothesis of an integration based on IS-element activity during transformation (Figure 4 D, E).
In contrast to the Pop 1 results, the integration into ROI 3 (Figure 4

Demonstrating low frequent in vivo activity of MetMaz-C1 in untreated ancestor cultures
A long-term e volution e xperiment was performed with two different M. mazei strains under various stress conditions to track casposon movement in vivo . The ancestor strains Mm Mut 203b (empty vector control) and the Cas1solo overexpression strain (Mm Mut 208) were sequenced with Illumina and Oxford Nanopore technology prior to the longterm e volution e xperiment to gain a deeper understanding of nati v e MetMaz-C1 acti vity. The basal CEF in both strains without any str ess tr eatment was determined by a nested qPCRs approach (see methods section).
Assembled genomes of both strains, based on the described pipeline, showed already high differences in the casposon sequences. Whereas the Mm Mut 203b casposon was unaltered, the Mm Mut 208 assembly indicates a single or double chromosomal integration of the Cas1solo overexpression plasmid (pRS1270) into the host genome. These integration e v ents would increase the size of MetMaz-C1 by 9-18 kb respecti v ely (Figure 5 A).
Nested qPCR reactions of gDNA extracted from freshly inoculated strains were set up according to the described procedures in the methods section using the pRS1511-based  Table S3). Although PCR products were detected in both strains at the end of the nested qPCR procedure, the calculated CEFs were highly different (Figure 5 D). Unexpectedly, the empty vector control Mm Mut 203b showed a much higher frequency (mean: 1.97 × 10 −7 ± 0.97 × 10 −7 excisions per genome) compared to a mean of 0.19 × 10 −8 ± 0.40 × 10 −8 excisions per genome determined for the Cas1solo OP Mm Mut 208. This difference between the two different strains was highly significant based on a two-tailed t-test (

Verification of casposon e x cision and integration into new genomic loci in vivo via sanger sequencing
For verification of the casposon excision by the nested qPCR approach, the resulting nested PCR products of both strains obtained above were sequenced via Sanger sequencing using the qPCR primers A* (Figure 3 B). Sequence alignments to the r efer ence genome re v ealed empty target sites, without any casposon footprints or TIRs (Figure 6 A). The nati v e casposon MetMaz-C1 was cut out of the genome sequence by leaving a single empty target site (Figure 6 A). These results were verified by sequencing PCR products generated from 85 cultures of the first culture plate of the long-term e volution e xperiment (Supplementary Tab le S4). Based on this finding M. mazei genome loci were checked for potential new integration of the nati v e casposon after 5-6 generations. Potential insertion sites with similarity to already published potential insertion sites ( 19 ) were tested by additional nested PCR reactions. Primers B and B* or C and C* (Figure 3 A) were used in combination with primers upstream the locus of interest (Supplementary Table S1). These nested PCRs re v ealed the presence of MetMaz-C1 TIRs and TSDs in at least one new site which was previously published by Krupovic Table S4). One example site designated 'tLEU2' (193177 -193190), due to its partial similarity to a tRNA-LEU gene in the M. mazei genome was found to contain the MetMaz-C1 TSD and TIR sequence. The PCR fragment deri v ed from site-specific primers and primer set C (Figure 3 A) displayed the sequence of the right TIR and the TSD (IS1b) followed by the sequence of the tRNA-LEU homologous r egion (Figur e 6 B, Supplementary Table S4). The sequence did not show the original tLEU2 site (CGCAcTtAttTTtT), but a re v erse complement the original right TSD of MetMaz-C1 (AtAAtCTtAaTGCG; Figure 6 B), indicating a potential integration of MetMaz-C1 by transferring at least one of its own target sites (Supplementary Table S4), which somehow replaced the original tLEU2 site of the locus with IS1b (AtAAtCTtAaT-GCG). At least one further example was obtained with an additional tRNA-Leu integration site (tLEU1; CGCAT-CAAATTTCT; 181365-181378; see Supplementary Table  S4). These findings suggest a different translocation mechanism than the mechanism originally proposed by Krupovic ( 17 ).

In vivo casposon activity in the long-term evolution experiment influenced by strain and treatments
Due to the influence of cellular stress on transposon activity (42)(43)(44)(45)(46)(47), we assumed that stress might also enhance caspo-son activity. Consequently, the effect on the CEF of four differ ent str ess conditions (500 mM NaCl; 0.5 g / ml mitomycin c; 40 • C; +N (N-sufficiency)) was investigated using 12 biological replicates each of the two M. mazei strains Mm Mut 203b and Mm Mut 208 in a long-term evolution experiment. Estimation of the CEF based on the previously described nested qPCR approach, was established for comparison of casposon activity in both strains and all used trea tments. Dif ferent time points of the long-term evolution experiment were selected to determine the respecti v e CEF. Based on the assumption that selection pr essur e is highest at the first contact of an organism with a changed environment and that this pr essur e decr eases over time by adaptation to this stress condition, the first samples of the longterm experiment was considered first to investigate the effects of different conditions on M. mazei , especially on the activity of transposons or transposon-like elements such as MetMaz-C1. The first sample of the experiment was taken after 4 days of incubation of the two strains under the four different conditions. For monitoring the MetMaz-C1 activity based on excision and integration, gDNA of these samples was isolated and analyzed by nested qPCRs (Figure 7 A; Supplementary Table S4). Quantitative analysis confirmed the results above and showed statistically significant differences of the CEF between the two strains used, moreover it demonstrated differences between the treatments (Figure 7 A).
Nucleic Acids Research, 2023, Vol. 51, No. 13 6939 The CEF values between the two strains cultured under the four conditions still showed the same trend as the ancestors, yet the four cultivation conditions had a large effect on the CEF value determined within the treatments of strain  Table S5).
To follow the CEF under evolutionary adaptation to the treatments, two time points of the long-term evolution experiment were selected and analyzed with the same nested qPCR a pproach. For this anal ysis onl y Mm Mut 203b samples were used because of the detected influence of treatments on the CEF in this strain Figure 7 A. CEF values were determined for samples taken at day 124 in comparison to samples taken at day 316. This time period of 192 days r epr esented r oughly 250 generations of stable gr owth. The comparisons of treatments and time points did not show any significant differences, but a trend to a lower variation of CEF values in long-term cultures was observed. The enrichment of casposon excisions over time was not observed during the long-term evolution experiment, consistent with the expectation of highest selection pr essur e at the beginning of each treatment (Figure 7 B; Supplementary  Table S6).

DISCUSSION
Casposons, described as a new class of transposons in 2014, have been found in se v eral archaea species such as Aciduliprofundum boonei and some M. mazei strains, but their activity has been analyzed or predicted e xclusi v ely by computational analysis or a few in vitro studies using recombinant Cas1solo protein ( 17 , 19 , 22 , 25 ). Demonstration of casposon activity in in vivo assays was so far not possible to detect, due to the low number of available model organisms, which are genetically tractable. Ther efor e, the behavior of the nati v e M. maz ei casposon MetMaz-C1 was char acterized in vi vo in the current study using two different experimental approaches: (i) determination of excision e v ents of the nati v e casposon analyzed by nested qPCRs of M. mazei cultured under different conditions, and (ii) a genetic approach using a mini-casposon clearly demonstrating excision and integration of the casposon in vivo .
For characterization of the casposon translocation mechanism, the key processes of excision and integration are discussed individually, starting with the process of caspo-son excision. During the characterization of the long-term e volution e xperiment samples and its ancestor cultures by nested qPCRs, deeper insights into the excision mechanism of the M. mazei casposon MetMaz-C1 were gained. In this study, the CEF was established as a k ey mark er for casposon activity in vivo , allowing to detect casposon excision on a very low level with 10 −5 -10 −9 excision events per genome. These low values obtained for excision and the necessity of nested PCR approach suggested casposon activity on a very low le v el. Only single cells of the observed populations or e v en single chromosomes of their multiple genome copies showed casposon activity. CEF values determined for the empty vector control strain (Mm Mut 203b) and for the Cas1solo ov ere xpression strain (Mm Mut 208) were significantly different to each other. Unexpectedly, the CEF was significantly higher in Mm Mut 203b than in Mm Mut 208. This strong significant difference between the two strains could have various reasons, a simple explanation might be a casposon size effect. The observed size difference of the two casposon variants of Mm Mut 203b and Mm Mut 208 of 9-18 kb resulted from a single or double integration e v ent of the Cas1solo ov ere xpression plasmid into the M. mazei chromosome. Based on the 100% identical cas1solo genes, encoded on the ov ere xpression plasmid pRS1270 and on the chromosome in the casposon locus, a homologous recombination between these two genes was highly likely. For M. mazei , homologous recombination was shown to depend on similarity between regions of at least 800 bp ( 27 ). One potential indication for a size dependent excision might be identified by comparing CEF values of Mm Mut 203b with reported excision frequency values determined for other IS-elements. A comparable qPCR approach using Taqman probes was established for quantification of excision frequencies of the IS-element IS492 found in Pseudoalteromonas atlantica ( 48 ). Determined excision frequencies of IS492 were at least 100-1000-fold higher compared to the frequencies determined for the two MetMaz-C1 variants in Mm Mut 203b and Mm Mut 208 in the current study. Both transposable elements differ significantly in size, the nati v e MetMaz-C1 is roughly eight times larger than IS492 (1200 bp) ( 19 , 49 ), which might have a significant impact on its translocation frequency. Additionally, a comparable size effect was f ound f or the transposon Tn10, where smaller v ariants were ex cising mor e fr equently ( 50 ). A further explanation might be the genomic environment of MetMaz-C1, which could have inhibitory impact on the excision frequency as discussed for IS492 ( 48 ). Alternati v ely, a decreased CEF in Mm Mut 208 might be based on the strong plasmid deri v ed constituti v e e xpression of Cas1solo, since this kind of expression was shown to frequently cause inhibitory effects on transposable elements (reviewed in ( 51 )).
An alternati v e e xplanation for detected low le v el CEF values of both strains in comparison to IS492 might be a dir ect r eintegration of MetMaz-C1 into the same genome locus. These kind of e v ent would not be detectable with the nested qPCR approach used in this work, which was only targeting empty target sites based on the time point of sampling, and thus might result in underestimation of the CEF. For the IS-element IS492 it was shown that excision e v ents were not crucially coupled with insertion into new sites ( 48 , 52 ). Ther efor e, it is necessary to mention that this kind of explanation of direct reintegration into the same locus might be ruled out for MetMaz-C1 by qRT-PCRs targeting the casposon genes, especially the cas1solo gene encoding the key enzyme, the casposase. c as1solo transcription was slightly higher under Methanosarcina spherical virus (MetSV) challenge, but sho wed lo w expression values in untreated cells, which might explain a very low basal casposon activity ( 31 ). Additional stress conditions used in the long-term evolution experiment had little effect on the CEF, in contrast to reports of se v eral organisms in which transposon activity was increased by stress trea tment. In this stud y mitomycin C trea tment was chosen based on its direct inhibition of DNA synthesis in bacteria ( 53 ), since its ability to crosslink DNA molecules mediating DNA damage (re vie wed in ( 54 )). Mitomy cin C was further shown to positi v ely influence recombination frequency in Drosophila melanogaster and induces the lambda prophage ( 55 , 56 ). The CEF in Mm Mut 203b:mitomycin was significantly higher in comparison to the corresponding untreated Mm Mut 203b:+N. This observation might be interpreted that casposon excision is a recombina tion-rela ted process since an increase in recombination frequency was observed in Drosophila under mitomycin influence. Further, the comparisons of Mm Mut 203b:NaCl to Mm Mut 203b:mitomycin and Mm Mut 203b:40 • C were significant as well. NaCl treatment was used in the experimental design to allow a higher expression of the Cas1solo, which has been already shown for CRISPR locus associated proteins in M. mazei ( 57 ), and thus to positively influence the casposon activity in the same way. Low Mm Mut 203b:NaCl CEF values, not significant in the comparison to untreated cells (Mm Mut 203b:+N), might potentially imply a higher genome stability in the casposon region during high salt str ess. Mor eo ver, Cas1solo o verproduction did not increase casposon activity in Mm Mut 208 (Figure 7 ). Therefore, an effect of NaCl stress on Cas1solo-mediated casposon activity seemed to be very unlikely or was completely masked by the casposon size ef fects. Varia tion of cultiva tion temperature was further shown to affect transposon activity in many different organisms but had only a slight effect on casposon activity in the current study. Tn3, characterized in E. coli C600, or transposons found in the archaeon Halobacterium halobium can serve as examples for cold stress induced transposons, whereas many transposons found in plants such as Arabidopsis sp. showed heat induction ( 44 , 46 , 47 ).
Sequencing of PCR products re v ealed that only a single empty target site remained in the genome after casposon excision (Figures 6 and 8 ; Supplementary Table S4). This finding is in disagreement with the model proposed by Krupovic and colleagues ( 17 ), which predicted a persistent duplication of the target site, potentially separated by common transposon footprints ( 17 ). Howe v er, for the MetMaz-C1 excision, a model close to the translocation of IS492 is more likely ( 48 , 49 ). IS492 was shown to generate a single precise empty target site lacking persistent duplication post excision and forming a circular transposition intermediate ( 48 , 49 ). Consequently, based on our findings we hypothesize a circular MetMaz-C1 intermediate during the excision process (Figure 8 ), although we were not able to detect it during the current study. Howe v er, based on the overall very low CEF values determined for MetMaz-C1 compared to the frequencies determined for IS492 ( 48 ), it is hardly surprising that this intermediate was not detected. In case of IS492 (with high excision frequency) the circular intermediate was not persistent over extended periods of time either and could only be detected in special treated cultures at low amounts ( 49 ). Leaving one single empty target site during casposon translocation further suggests a staggered cut within the target site (Figure 8 ), like it was first discussed by Krupovic and colleagues ( 17 ). Studies of Polinton / Maverick transposons carrying capsid proteins in rare cases and descriptions of virus encoded Cas genes ( 58 , 59 ), combined with a potential circular casposon intermediate could be a potential hint for a plasmid or circular virus deri v ed origin of casposons.
To detect potential reintegration into predicted target sites ( 17 , 19 ), additional nested PCRs of these sites were performed ( Figure 6 ; Supplementary Table S4). Integration of the nati v e casposon in the described tLEU2 site (CGCAcTtAttTTtT) was detected via Sanger sequencing of PCR products. PCR products suggested integration in the tLEU2 locus by inversion of the casposon sequence and integrating the original target site. This kind of integration is the second hint to a circular intermediate carrying its o wn tar get site during translocation ( Figure 8 ). Integration by homologous recombination is very unlikely, due to very short overlap of sequences by the described GCGA-motif ( 17 , 19 , 22 ). The process of this 'replacement' of the original tLEU2 site is unknown and may also be a kind of artifact, because all our analyses were based on nested PCRs with gDNA of whole populations. Ther efor e, it could not be excluded that the casposon locus in question, which served as the basis for the PCR product discussed here and was possibly present in this form in only one genome, already carried a mutated version of the tLEU2 site before a potential casposon integration.
Casposon activity in vivo , its excision and integration, was further verified during the second genetic approach, the plasmid based mini-casposon assa y. Man y preceding studies focusing on MGEs used plasmid-based transposition assays to characterize transposons or integrating phages e.g. Mu-phage ( 60 ). Many of these systems use at least two separated plasmids, where one plasmid carries the engineered transposon and helper plasmids carry transpositiondependent genes e.g. transposases. These systems were often optimized for generation of mutant strains by introduction of genes of interest into target genomes ( 61 ). The in vivo mini-casposon assay used in the current study is completely independent from other components, since its potentially relevant proteins are chromosomally expressed by the MetMaz-C1 locus (e.g. Cas1solo). A cross element / cross system activity between the casposon and the M. mazei CRISPR-Cas systems regarding Cas1 was unlikely, since M. mazei CRISPR-Cas systems were found to be inacti v e ( 57 ). The mini-casposon designed on the suicide vector pRS1520 consisted of MetMaz-C1 deri v ed TSDs and TIRs flanking a kanamycin resistance cassette, a neomycin resistant cassette and a R6K ␥ origin, and was ther efor e in analogy of various preceding studies with transposons ( 23-25 , 62-64 ). If this engineered casposon was acti v e, it would result in integration of the mini-casposon into the M. mazei genome at its specific target sites mediating a neomycin resistance. Analysis of neomycin-resistant M. mazei populations by PCRs and a rescue cloning approach adapted from Quin and colleagues ( 62 ), lead to the conclusion that the mini-casposon had acti v ely integrated into the host genome. Sequenc-ing of positi v e rescue-cloning-deri v ed plasmids showed correct mini-casposon translocation to the left TSD (IS1a) of MetMaz-C1 in at least one case. Further incorrect integration e v ents of the suicide plasmids based on singlecrossov er e v ents were observ ed in close association with M. maz ei deri v ed IS-elements, w hich impl y activity of these elements as well. We hypothesize IS-element translocation to the freshly introduced suicide plasmid (pRS1520) by accident, followed by homologous recombination into another copy of the host chromosome via the IS-element (see Figure 4 D).
The obtained results and arguments discussed within this study allowed first deeper insight in the casposon activity in vivo . Data suggested low le v el casposon acti vity in single cells of a M. mazei population , which is neither beneficial nor lethal in most cases for the respecti v e host cells and thus will not be enriched. MetMaz-C1 and the used minicasposon showed strong site-specific integration as it was already concluded for different casposons by characterization of their key enzymes in vitro (22)(23)(24). MetMaz-C1 appeared to excise and integrate by potentially transferring the original target site, suggesting a staggered cut within the target sites ( Figure 8 ). Based on our data, we hypothesize, that the casposon is forming a circular intermediate, although w e w ere not able to detect such an intermediate ( Figure 8 ). The mini-casposon assay re v ealed integration in the same MetMaz-C1 target site resulting in tandem structures, w hich potentiall y counts as a further argument for its evolutionary role in the generation of CRISPR like arrays. Such tandem structures were reported for a huge variety of IS-elements, transposons and e v en for some casposon relati v es in different Methanosarcina species (re vie wed in ( 6 , 19 ). Overall, the current study not only gave first experimental evidence for in vivo activity of MetMaz-C1 but in addition also indications for revising the previous translocation model.

DA T A A V AILABILITY
Assemblies of Mm Mut 203b and Mm Mut 208 as well as belonging raw data are provided under the following Bioproject IDs: PRJNA929716 and PRJNA929891. Supplementary Table S1: List of all used primers. Supplementary