Simple and Rapid Site-Specific Integration of Multiple Heterologous DNAs into the Escherichia coli Chromosome

ABSTRACT Escherichia coli is the most studied and well understood microorganism, but research in this system can still be limited by available genetic tools, including the ability to rapidly integrate multiple DNA constructs efficiently into the chromosome. Site-specific, large serine-recombinases can be useful tools, catalyzing a single, unidirectional recombination event between 2 specific DNA sequences, attB and attP, without requiring host proteins for functionality. Using these recombinases, we have developed a system to integrate up to 12 genetic constructs sequentially and stably into in the E. coli chromosome. A cassette of attB sites was inserted into the chromosome and the corresponding recombinases were cloned onto temperature sensitive plasmids to mediate recombination between a non-replicating, attP-containing “cargo” plasmid and the corresponding attB site on the chromosome. The efficiency of DNA insertion into the E. coli chromosome was approximately 107 CFU/μg DNA for six of the recombinases when the competent cells already contained the recombinase-expressing plasmid and approximately 105 CFU/μg DNA or higher when the recombinase-expressing plasmid and “cargo” plasmid were co-transformed. The “cargo” plasmid contains ΦC31 recombination sites flanking the antibiotic gene, allowing for resistance markers to be removed and reused following transient expression of the ΦC31 recombinase. As an example of the utility of this system, eight DNA methyltransferases from Clostridium clariflavum 4-2a were inserted into the E. coli chromosome to methylate plasmid DNA for evasion of the C. clariflavum restriction systems, enabling the first demonstration of transformation of this cellulose-degrading species. IMPORTANCE More rapid genetic tools can help accelerate strain engineering, even in advanced hosts like Escherichia coli. Here, we adapt a suite of site-specific recombinases to enable simple, rapid, and highly efficient site-specific integration of heterologous DNA into the chromosome. This utility of this system was demonstrated by sequential insertion of eight DNA methyltransferases into the E. coli chromosome, allowing plasmid DNA to be protected from restriction in Clostridium clariflavum and enabling genetic transformation of this organism. This integration system should also be highly portable into non-model organisms.

C. clariflavum RM systems likely requires the expression of more than 5 DNA methyltransferases in E. coli.
We sought to expand upon the CRIM system by leveraging large serine recombinases to develop a system for efficiently inserting heterologous DNA into the E. coli chromosome. By designing the system to allow for simple removal of the selectable marker and vector backbone, our system enables successive integration of up to 12 constructs without the accumulation of repetitive DNA sequences and selectable markers. As a demonstration of this system in E. coli, we expressed 8 methyltransferases encoded in the C. clariflavum 4-2a genome to methylate DNA and enable transformation of C. clariflavum for the first time.

RESULTS
Design of the integration system. A heterologous DNA integration system using serine recombinases was created by first integrating a cluster of attB sites into the E. coli chromosome at the HK022 phage site using the CRIM system (Fig. 1). The resulting strain contains 14 orthogonal attB sites stably inserted into the chromosome: U370, UBT1, R4, BxB1, TP901-1, RV, Spbc, TG1, UC1, MR11, UK38, A118, Wb, and BL3.
A non-replicating integration vector, which carries the "cargo" DNA to be inserted, contains the kanamycin resistance marker, the conditional oriR6K origin of replication FIG 1 Overview of the site-specific recombinase system for DNA integration into the E. coli chromosome. (A) 14 attB sites were added to chromosome at the HK022 phage site (AG2005, AG3525, and AG4277) and the 14 corresponding serine recombinases were cloned onto temperature sensitive plasmids to mediate integration between a plasmid containing an attP site (e.g., pLAR067) and the corresponding attB site on the chromosome. In this example, the attP plasmid encodes lacZA, and the lacZA-attP cassette is flanked by NdeI and BamHI unique restriction sites to facilitate future cloning. (B) The kanamycin resistance marker and plasmid origin of replication can then be removed by introducing the UC31 recombinase on a temperature sensitive plasmid (pLAR047). (C) Chromosomal structure of the final strain after removal of the kanamycin resistance marker.

Rapid Chromosomal Integration Tool in E. coli
Journal of Bacteriology that only replicates in an E. coli strain that expresses the pir gene (26), and 1 of the corresponding attP sites. The oriR6K-kanR cassette is flanked by UC31 attB and attP sites, which facilitates marker removal (see below). Outside of the UC31 sites is an arabinose-inducible promoter driving lacZa in plasmid pLAR067. This serves as a convenient site for cloning of the intended cargo, with NdeI and BamHI sites flanking lacZa. Each of the corresponding large serine recombinases were cloned into a temperature sensitive helper plasmid, pInt-Ts (8), replacing the l integrase. These genes are under the control of the l pR promoter and cI857, a heat-inducible promoter system. This enables transient expression of the recombinase to mediate integration and, after raising the temperature, subsequent loss of the recombinase plasmid. The recombinase helper plasmid mediates DNA recombination between an attP site from the nonreplicating "cargo" plasmid and its associated attB on the E. coli chromosome, resulting in the integration of the entire non-replicating plasmid into the chromosome. Recombination creates 2 new attachment sites called attL and attR that flank the newly integrated DNA (Fig. 1B). Next, the oriR6K-kanR cassette can be removed by transient expression of the UC31 integrase. This results in recombination between the UC31 attB and attP sites, leaving behind a single UC31 attR site and any associated genetic cargo ( Fig. 1B and C). To consistently test each recombinase, a single integration plasmid was constructed that contains each of the corresponding attP sites (poly-attP cassette; pLAR31).
Testing integration efficiency of the recombinase helper vectors. To test integration efficiency of the recombinases, the temperature sensitive helper plasmids encoding recombinases were each transformed into E. coli strain Top10 Ddcm::frt, which contains the poly-attB cassette on the chromosome. Next, a non-replicating plasmid containing the poly-attP cassette (pLAR031) was transformed into each strain expressing a recombinase (Fig. 2). Integration was observed with 12 of the recombinases, of which 6 (A118, TG1, BL3, Wb, TP901, and Spb) had an efficiency of approximately 10 7 CFU/mg of DNA. Another 4 showed recombination frequencies in the range of 10 5 to 10 6 CFU/mg of DNA. UC1 enabled the lowest recombination frequency of an average of 200 CFU/mg of DNA, while 2 recombinases, UBT and RV, did not enable any integration.
Integration confirmation. Serine recombinases have been shown to integrate DNA into naturally occurring secondary attB sites, called pseudo-attB sites, that have similarity to the native sequences (27). To determine if the non-replicating plasmid integrated into the intended locus, rather than a pseudo-att site, colonies from the resulting transformations were screened using a four-primer multiplexed PCR screen, with primers flanking the 2 att sites (P1, P2, P3, and P4 indicated in Fig. 1A), similar to the screen used for the CRIM system (8). The 4 primer system gives distinct band sizes to distinguish between the attB (P1 and P4), attP (P2 and P3), attL (P1 and P3), and attR (P2 and P4) sites to evaluate if integration occurred at the intended locus. Upon correct insertion, bands consistent with the attL and attR will be observed. However, if the poly-attP cassette plasmid integrated into a pseudo-attB site, a single band, indicative of the parent strain, will form. Ten colonies from each transformation were screened and all colonies showed the correct integration bands for all 10 colonies except MR11 and UC1, where 7 of 10 and 4 of 10 integrated at the correct locus, respectively (Fig. S1).
Streamlined integration system via co-transformation. To determine if the process could be further streamlined to accelerate strain construction, we explored whether co-transformation of the recombinase helper plasmid and the integrating plasmid could allow for DNA insertion into the chromosome. In this scenario, the recombinase is transiently expressed, but without selection for the helper plasmid. If successful, this would eliminate one round of competent cell preparation, transformation, and plasmid curing. When using the temperature sensitive helper plasmids described above in combination with integrating poly-attP cassette plasmid at a nonpermissive temperature, no kanamycin-resistant colonies were observed, indicating no recombination. Expression of the recombinases could have been too low to enable DNA integration; therefore, the temperature sensitive recombinase helper plasmids were reconstructed to use the Clostridium thermocellum enolase promoter (P Ct_eno ), which has been shown to be highly active in E. coli (28) and is presumably constitutively expressed. Of the 12 recombinases shown to be functional in Fig. 2, the promoter was successfully replaced in all except Spbc and Wb, where the cloning was not successful.
To test the efficiency of these new recombinase expression plasmids, each were cotransformed with the poly-attP cassette vector into the E. coli Top10 Ddcm::frt poly-attB cassette strain and recovered at a temperature too high to allow replication of the helper plasmid. Each of the 10 enabled integration in the 1 step integration process (Fig. 2). Five of the recombinases had an insertional efficiency of 10 5 CFU/mg of DNA or higher: BxB1, TG1, BL3, U370, and R4. The UC1 and TP901 recombinases also enabled DNA integration in a single step but had low efficiencies of less than 100 CFU/mg of DNA.
Backbone removal of oriR6K-kanR. Some applications will require the removal of the antibiotic resistance gene, for instance to enable further genetic modification. Therefore, we tested the UC31 recombinase-based marker removal system described above using a strain in which the attP cassette plasmid integrated into the R4 attB locus ( Fig. 1B and C). After integration into the R4 attB site was confirmed, a temperature-sensitive plasmid encoding UC31 Int was transformed into the strain at 30°C. After further incubation at a 37°C to the cure the UC31 helper plasmid, 12 colonies were screened for successful recombination between the UC31 att sites and loss of the helper plasmid by patching single colonies on each antibiotic used in the process (kanamycin and carbenicillin). Ten colonies (83%) did not grow on either antibiotic and were PCR screened for removal of the backbone using primers P5 and P6 (Fig. 1C) that flank the UC31 sites. Of the colonies that were sensitive to both antibiotics, 100% were confirmed to have lost the oriR6K-kanR cassette by PCR (Fig. S2).
Identification and expression of methyltransferases for C. clariflavum 4-2a. A major barrier to the genetic transformation of non-model microorganisms is the presence of RM systems that degrade foreign DNA. One approach to evading these RM systems is to express the target organism's DNA methyltransferases in E. coli. Therefore, as a demonstration of the utility of the above integration system, we targeted heterologous expression of C. clariflavum DNA methyltransferases to methylate plasmids prior to transformation.
To engineer E. coli to mimic the methylome of C. clariflavum to evade restriction, 8 of the identified methyltransferase were expressed from an arabinose-inducible promoter and inserted sequentially into the E. coli chromosome into different attB sites. CloclaDRAFT 1996 was predicted to target G(m 6 A)TC and was not included because E. coli natively methylates this sequence. The methyltransferases were cloned into the non-replicating integration vector with a single attP site. The methyltransferases were sequentially integrated into an E. coli strain that natively encodes Dam methyltransferase to methylate G(m 6 A)TC but lacks dcm because C(m 5 C)WGG is not methylated by C. clariflavum, resulting in strain AG5645 (Table S1). The plasmid backbone of each integrated vector was removed as described above using UC31 Int, allowing repeated use of the kanamycin resistance gene. Methylome analysis was performed to examine the functionality of each methyltransferase in E. coli (Table 2). Six motifs were identified; of these, 4 showed complete methylation, and 2 showed partial methylation. No methylation was detected for ATGCAT and CAGAAG, suggesting that some methyltransferases were not functioning in E. coli.
Transformation of C. clariflavum 4-2a. Even though methylation in E. coli was incomplete, we attempted to transform C. clariflavum with plasmid DNA methylated  To determine the impact of DNA methylation on transformation of C. clariflavum, plasmid pMTL83151 was isolated out of the arabinose-induced E. coli methylation strain AG5477 and used to transform C. clariflavum. As a control, unmethylated plasmid was isolated out of Top10 Ddcm::frt and used to transform the strain. Unmethylated plasmid never yielded colonies, while the methylated plasmid yielded 9 6 3 CFU per microgram of DNA (CFU/mg).

DISCUSSION
The DNA integration system developed here expands the E. coli genetic toolbox to provide a quick and simple method to integrate up to 12 heterologous DNA constructs stably and sequentially into the chromosome of E. coli. In combination with the CRIM system, this enables insertion of up to 16 plasmids. Integration by our system can occur via a single co-transformation of 2 plasmids, including 1 recombinase-expression plasmid that is common for all insertions into that attB site and 1 that can be customized to carry the desired cargo. Site-specific recombination inserts DNA into the chromosome more quickly than classic techniques like homologous recombination. Furthermore, an additional site-specific recombination step can be utilized to remove vector backbones and resistance markers, which avoids spontaneous resistant mutants than can hinder sacB and other counter-selection systems. Using this system enables stable chromosomal integration in 1 day and unmarked insertions in 3 days, while tools such as homologous recombination often requires several steps and at least 5 days.
Another characteristic of the system is the ability of the recombinases to recombine DNA efficiently and reliably. They function without a significant decrease in efficiency as the size of the insert increases; in our hands, cargo sizes upwards of 7 kb have been integrated, and no decrease in recombination efficiency was seen (unpublished data). Size-based limitations on DNA insertion should be primarily dependent on the diminishing efficiency of DNA entry into the cell as the DNA size increases. This overcomes some of the size-associated challenges with other genetic tools, like use of the lambda red recombinase system (3) for heterologous DNA insertion. The ability to rapidly integrate constructs, at single copy, will help to increase the speed of engineering in E. coli, especially for high throughput, single copy library evaluation. The DNA integration system described here contains other useful features. Like the CRIM system (8), it uses the conditional oriR6K origin of replication, which is only able to replicate in E. coli strains engineered to express the pir gene. This enables the integration vectors to act as a suicide vector in most E. coli strains, including the one containing the poly-attB cassette. Next, the vector backbone can easily be removed after DNA integration because the oriR6K origin and the antibiotic resistance marker are flanked by UC31att sites. Removal of the backbone allows for reuse of the antibiotic resistance marker and reduces repetitive DNA, diminishing the likelihood of homologous recombination events when stacking multiple genes into the poly-attB cassette. Another feature of this system is the unidirectional nature of the serine recombinases. Integration creates new attachment sites, attL and attR, and these new att sites are not recognized by the recombinases. This is especially beneficial when integrating multiple constructs in close proximity on the genome. Unlike the Flp-frt and Cre-lox systems, where the resulting scars are still substrates for the recombinase, UC31 does not act on the remaining UC31 attR sites after the marker is removed. Because these resulting attR scars cannot recombine with each other, there is no genome instability if 2 scars are near each other. Of note for DNA methyltransferase expression in particular, the dam-dcmstrain is necessarily recA1 because recA and dam are synthetically lethal (31), so the chance of homologous recombination between repetitive regions, like the P BAD promoter, should be considered in this strain. Other strains in this study (Top10/ DH10B and its derivatives) are recA-, which should reduce the probability of recombination. While most researchers can work in recAgenetic backgrounds, in the future, the poly-attB cassette can be divided into multiple, smaller clusters and distributed to distant sites in the chromosome to further reduce the possibility of recombination between identical parts of inserted plasmids.
Initially, the first set of recombinase helper plasmids was constructed that allowed for integration in 2 steps. Using this approach, 12 of the recombinases were functional, but 2 were not. For unknown reasons, the BT and RV recombinases did not function as expected; this could be further explored in the future to increase the number of available sites for DNA insertion. To decrease the number of steps in this process, the ability to transform both plasmids in 1 event was desirable. The second set of helper plasmids enabled 1-step integration, though 2 of the recombinases were not successfully cloned. We speculate that high expression of the Wb and Spb recombinases was toxic for E. coli. However, the availability of the first set of plasmids still allows the use of the Wb and Spb recombinases via the 2-step process. Of the recombinases that are functional in the one-step process, R4, BxB1, and TG1 are highly efficient and reliable, so these 3 insertion sites should be targeted first when using this integration system. The UC1 and MR11 recombinases are least reliable, commonly integrating into pseudo-att sites and with lower efficiencies; therefore, these should be the last 2 sites to be considered. There is often a decrease in transformation efficiency when co-transforming with the strong promoter helper plasmids, and A118 and TP901-1 had especially large drops in efficiency for unknown reasons. However, the speed of the 1-step process makes them more useful when speed is desired. The 2-step approach, on the other hand will be most useful when high integration efficiency is needed, such as for the creation of large libraries.
Technologies to transform diverse non-model microorganisms are desperately needed, and mimicking the methylation patterns of the target organism to avoid restriction is an important approach to enable transformation. To this end, and as a demonstration of the utility of this tool, we engineered E. coli to express C. clariflavum methyltransferases to protect plasmid DNA and enable the genetic transformation of C. clariflavum. Transformation efficiency in this strain is still low, and further optimization will likely increase the efficiency. Two of the methylated motifs found in C. clariflavum were not found in the E. coli methylation strain, and 2 other motifs were methylated less than 100% in the E. coli genome, which likely impacts the transformation efficiency. Because C. clariflavum is a thermophilic organism, the activity of the methyltransferases at E. coli growth temperatures could decrease functionality. Future utilization of mesophilic methyltransferases that target the same motifs or engineered versions of the thermophilic enzymes to increase activity at mesophilic temperatures could help to overcome this challenge. Even with the incomplete methylation, low transformation efficiency was enabled, which allows for further genetic manipulation of C. clariflavum for fundamental studies and metabolic engineering related to lignocellulose deconstruction and bioconversion.
Finally, the tools developed here are directly applicable to other organisms, including non-model microbes. Because these recombinases do not require any host factors to function, this system should be adaptable to any strain in which the poly-attB "landing pad" can be inserted into the chromosome. Recently, this has been developed for various Gamma-and Alpha-proteobacteria in a toolset called Serine recombinase-Assisted Genome Engineering (SAGE) (32). Together, these toolsets will help accelerate strain engineering across diverse organisms.

MATERIALS AND METHODS
Strains and culture conditions. E. coli Top 10 Ddcm::frt was used for maintenance of all replicating vectors and integration of the attB cassette (21). PIR-dependent plasmids, containing the oriR6K origin of replication, were propagated in E. coli GT115 (Invivogen). All E. coli strains were grown in LB broth (Miller) with antibiotics, as necessary, for maintenance of vectors and for selection of integrants. Carbenicillin was used at 50 mg/mL, kanamycin was used at 50 mg/mL for replicating vectors and 30 mg/ mL for single copy integrants, and spectinomycin and streptomycin were used together at 100 mg/mL each for replicating vectors and 50 mg/mL each for single copy integrants. All strains were grown at 37°C, unless the strain contains a temperature sensitive recombinase helper plasmid, in which case growth was at 30°C. C. clariflavum 4-2a was grown in a Coy anaerobic chamber (Coy Laboratory Products) in CTFUD medium (21) at 50°C. CTFUD medium is comprised of 3 g/L sodium citrate tribasic dehydrate, 1.3 g/L ammonium sulfate, 1.43 g/L potassium phosphate monobasic, 1.8 g/L potassium phosphate dibasic trihydrate, 0.5 g/L L-Cysteine hydrochloride, 10.5 g/L MOPS sodium salt, 6 g/L Rapid Chromosomal Integration Tool in E. coli Journal of Bacteriology glycerol-2-phosphate disodium, 5 g/L cellobioase, 4.5 g/L yeast extract, 0.13 g/L calcium chloride dehydrate, 2.6 g/L magnesium chloride hexahydrate, 0.1 mg/L ferrous sulfate heptahydrate, and 0.5 mL/L 0.2% (wt/vol) resazurin. The CTFUD pH was adjusted to 7.0 with 45% (wt/vol) potassium hydroxide and was supplemented with 5 mg/mL thiamphenicol when needed. Plasmid construction. Annotated plasmid maps for all plasmids used in this study are provided in the Supplemental Material, and GenBank-style files are available upon request. All plasmids and strains used in this study are listed in Table S1. The first set of recombinase helper plasmids were constructed by GenScript, where the gene encoding l Int in pINT-ts (8) was replaced with the synthesized serine recombinases that were previously described (7), resulting in plasmids pGSs037, 038, 040, 041, 042, 043, 044, 045, 046, 047, 048, 050, 053, 054, and 082 (Table S1). Recombinases were codon optimized for expression in E. coli.
For the second set of helper plasmids, recombinases were cloned under the C. thermocellum DSM 1313 enolase promoter (28) with a ribosome binding site modified to AGGAGGA. First, UC31 was synthesized with the enolase promoter into a pUC replicating vector. The promoter and recombinase DNA were then amplified by PCR using Phusion High Fidelity master mix (Thermo Scientific) and inserted using Gibson Assembly (New England Biolabs) into pINT-ts, replacing the native promoter and lambda recombinase, resulting in pLAR047. Using pLAR047 as a backbone, the remaining serine recombinases were cloned and inserted, replacing UC31, using Gibson Assembly, resulting in plasmids pLAR052-62 and 074 (Table S1).
The base integration plasmid, pGSs009, was synthesized and constructed by GenScript. It was constructed from pAH55 (8) and a synthesized insert containing UC31 attB and attP sites flanking the T5lac promoter and lacZa. The promoter in pGSs009 was replaced with P BAD from pLA2 (8) to create pLAR067. For recombinase efficiency experiments, a synthesized cassette containing 13 attP sites, excluding SpbC, was first inserted into the BamHI and NdeI sites of pGSs009 using Gibson Assembly. Next, oligonucleotides with the Spb attP and homology to pGSs009 were inserted into the EcoO1091 site, creating pLAR031.
Testing integration efficiencies in 2 steps. To test integration efficiency of the first set of recombinase helper plasmids, each plasmid was transformed into the poly-attB cassette strain, AG2005. Batches of electrocompetent cells were made in duplicate for each strain, and 200 ng of pLAR031 was electroporated into 25 mL of each competent cell batch, each in duplicate (n = 4). Cells were resuspended in 250 mL super optimal broth with catabolite repression (SOC) and recovered at 37°C for 1 h and 42°C for 30 min. The recovery was plated on LB with kanamycin and incubated at 37°C overnight. Colonies were picked into 5 mL LB with kanamycin and PCR screened after growth. PCR was performed on strains using primers P1 to 4 (all screening primers are listed in Table S3) to screen for the corresponding recombination event using Quick-Load taq master mix (NEB).
Testing integration efficiencies in 1 step. To test integration efficiency of co-transformation of the putative constitutively expressed recombinase helper plasmids and cargo plasmid, batches of electrocompetent cells were prepared of the poly-attB cassette strain (E. coli strain AG2005) in duplicate. Both the helper plasmid and pLAR031 were co-transformed in duplicate for each batch of competent cells (n = 4), using 200 ng of each plasmid and 25 mL of electrocompetent cells. Transformants were resuspended in 250 mL of SOC medium and recovered at 37°C for 1 h. The recovery was plated on LB 1 kanamycin plates and incubated at 37°C overnight.
Removal of plasmid backbone. To remove the plasmid backbone, electrocompetent cells were made from the confirmed integration strain and transformed with the UC31 helper plasmid. Cells were recovered at 30°C for 1 h and plated on LB 1 carbenicillin plates. Colonies were picked into LB (with no antibiotics) and grown at 37°C for 8 h to allow plasmid curing. Then, a 5 mL aliquot of the outgrowth was streaked on LB plates and grown overnight. Resulting colonies were then patched on LB, LB 1 carbenicillin, and LB 1 kanamycin plates to screen for loss of the kanamycin resistance marker and loss of the helper plasmid.
Methyltransferase expression and methylome analysis. Methyltransferase expression plasmids were sequentially integrated into E. coli strain AG4277 as described above, ultimately yielding strain AG5645. Strain AG5645 was grown with 1 mM arabinose to induce methyltransferase expression. Genomic DNA was isolated from C. clariflavum and AG5645 using the Qiagen Genomic-tip kit (Qiagen). Methylome analysis was performed on these strains using PacBio SMRT sequencing (35) and WGBS as previously described (21).
Transformation of C. clariflavum 4-2a. Electroporation was performed similarly to existing protocols for C. thermocellum (36). Two 5 mL cultures were inoculated with C. clariflavum 4-2a and grown overnight. The next day, two 200 mL cultures were inoculated with a 1% inoculum and grown until an optical density at 600 nm (OD 600 ) of 0.9. Once the cultures were grown, they were centrifuged in 50 mL conical tubes at room temperature at 6,000 Â g for 15 min. The supernatant was decanted, and 25 mL of electroporation buffer (250 mM sucrose, 10% glycerol) was added to the tube without disrupting the cell pellet. Cells were centrifuged again and washed twice more in the same way. After the last spin, the cell pellets were resuspended in ; 100 mL electroporation buffer and transferred to a microcentrifuge tube.
Using fresh electrocompetent cells, 20 mL of cells were transformed with 1 mg of pMTL83151 (37). Cells were electroporated in a 1 mm cuvette with a square wave using a Bio-Rad Gene Pulser Xcell Electroporation System set at 1200 V with a 1.5 msec pulse. Cells were then resuspended in 1 mL CTFUD medium and incubated for 3 h at 50°C to recover. After recovery, transformants were plated within CTFUD with 1.5% agar and 5 mg/mL thiamphenicol and incubated for 4 days at 50°C. Colonies from the plates were picked into liquid CTFUD medium. Liquid culture was screened by PCR for the presence of the cat gene and further confirmed via 16S rRNA gene sequencing to verify the culture. Two batches of C. clariflavum competent cells were transformed with plasmid from each methylation state, each time in duplicate (n = 4 total).
Data availability. Data are available at SRA under accession numbers SRX2120272, SRX2120271, and SRX2120270.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 2.5 MB.