Enzymatic Supercoiling of Bacterial Chromosomes Facilitates Genome Manipulation.

The physical stability of bacterial chromosomes is important for their in vitro manipulation, while genetic stability is important in vivo. However, extracted naked chromosomes in the open circular form are fragile due to nicks and gaps. Using a nick/gap repair and negative supercoiling reaction (named SCR), we first achieved the negative supercoiling of the whole genomes extracted from Escherichia coli and Vibrio natriegens cells. Supercoiled chromosomes of 0.2-4.6 megabase (Mb) were separated by size using a conventional agarose gel electrophoresis and served as DNA size markers. We also achieved the enzymatic replication of 1-2 Mb chromosomes using the reconstituted E. coli replication-cycle reaction (RCR). Electroporation-ready 1 Mb chromosomes were prepared by a modified SCR performed at a low salt concentration (L-SCR) and directly introduced into commercial electrocompetent E. coli cells. Since successful electroporation relies on the genetic stability of a chromosome in cells, genetically stable 1 Mb chromosomes were developed according to a portable chromosome format (PCF). Using physically and genetically stabilized chromosomes, the democratization of genome synthetic biology will be greatly accelerated.

In the era of genome synthetic biology, there is a need for technologies to routinely manipulate and analyze bacterial chromosomes in vitro.[1−4] However, unlike small plasmids, bacterial chromosomes are fragile and difficult to handle when extracted from cells.[4] In the most successful approach,[5−8] a bacterial chromosome was cloned in yeast cells, extracted from the yeast cells in an agarose plug, and transplanted to Mycoplasma capricolum cells to boot up the engineered or heterologous genome. M. capricolum has an exceptionally high competency to uptake a chromosome as large as 1.2 megabase (Mb).[5,8] Because M. capricolum cannot be converted to evolutionally distant bacteria,[8] the phylogenetic range of competent host bacteria must be expanded. Recently, we succeeded in the transformation of Escherichia coli via electroporation with a 1.0 Mb chromosome purified using a bacterial artificial chromosome (BAC) purification kit.[9] A population of negatively supercoiled 1 Mb chromosomes survived the purification and electroporation processes, probably owing to the compacted structure.[10] Furthermore, these supercoiled chromosomes were separated by size using conventional agarose gel electrophoresis.[9] These findings suggested that even larger chromosomes might be manipulatable when covalently closed and negatively supercoiled.

There are two traditional methods for estimating the sizes of bacterial chromosomes and megabase-sized plasmids using agarose gel electrophoresis: pulsed-field gel electrophoresis (PFGE)[11,12] and modified Eckhardt gel electrophoresis.[13−16] In PFGE, small agarose plugs containing naked chromosomes are embedded in a large agarose gel as samples for electrophoresis. Double-stranded DNA (dsDNA) molecules physically or enzymatically linearized are resolved by size, while large circular dsDNA molecules in the open circular form are trapped near the start.[6,17,18] In successful cases, circular dsDNA molecules, probably in the supercoiled form, appear in the gel.[19] However, there is no established way of estimating the size of supercoiled megabase-sized dsDNA molecules by PFGE.[20] On the other hand, the modified Eckhardt method employs a normal agarose gel electrophoresis and resolves supercoiled dsDNA molecules. Bacterial cells are gently lysed in wells and directly used as the electrophoresis samples. Intact supercoiled chromosomes and plasmids are run on a 0.8–0.9% agarose gel for a half day to a few days. Up to 2.1 Mb plasmids can be detected and separated by size.[15,16] The supercoiled genomic chromosome might appear in the gel only in a successful case.[16] Although the two agarose gel electrophoresis methods are authentic,[21] it is ideal if supercoiled chromosomes could be prepared in a PCR tube, directly applied to wells, and separated by size using a conventional agarose gel electrophoresis method within a few hours at room temperature.

Establishment of genome-reduced strains and genome-split strains of E. coli facilitates whole genome engineering.[9,22−24] In nature, a considerable portion of bacteria, but not E. coli, have a multipartite genome.[25] For example, most Vibrio species have two chromosomes (Chr1 and Chr2).[19,26,27] Since Vibrio is closely related to E. coli,[28] the chromosome replication origin and partitioning system of Vibrio Chr2 has been used to develop a secondary chromosome in E. coli.[29−31] The partitioning system of the E. coli F plasmid (sopABC) has also been employed to maintain a sub-megabase-sized chromosome in E. coli.[22,31−33] By combining all of these efforts, we recently established a tripartite-genome strain RGF1389 from a genome-reduced (4.65 → 2.98 Mb) E. coli strain DGF-298W.[24] RGF138 looked normally and grew twice slowly as the parent. Among the three split-chromosomes (1.12, 0.84, and 1.02 Mb), the second and third ones were individually introduced via electroporation into an E. coli cloning strain HST08 as an extra chromosome.[9] This meant that the two chromosomes are “portable.” In contrast, the first one was not portable, probably due to the burden of the duplication of the genomic core region or due to the lack of any additional chromosome partitioning system and segregation system.[9] In this study, we show that enzymatically supercoiled bacterial chromosomes can be handled in vitro and are useful for genome analysis and chromosome implantation. Also, we show that the genetic stability of chromosomes can be improved according to a portable chromosome format.

Results and Discussion

Enzymatic Supercoiling of Bacterial Chromosomes

Plasmid’s topology can be converted from open circular to highly supercoiled using E. coli enzyme mixtures composed of DNA polymerase I (PolI), DNA ligase (LigA), Exonuclease III (ExoIII), and DNA gyrase (GyrAB). A plasmid supercoiling reagent was commercially available from a company. The nicks and gaps on open circular plasmids are filled and fused by PolI and ligase,[34] while ExoIII makes easy-to-repair gaps.[35,36] Gyrase helps in the supercoiling of scarless plasmids in an ATP-dependent manner.[37,38] Since these reactions are involved in the chromosome replication cycle of E. coli, these enzymes are included in the reconstituted E. coli replication-cycle reaction (RCR).[31,39] Because RCR produced supercoiled plasmids of up to 1.1 Mb in size in a PCR tube,[31] we assumed that enzymatic supercoiling of open circular bacterial chromosomes could be mediated by a subset of RCR enzymes and buffers. This dedicated reaction was named “SuperCoiling and Repair Reaction” (SCR). SCR reagents were established using a 200 kb plasmid as a model DNA template. By heat treatment which introduces nicks, supercoiled plasmids were converted to be open circular. As expected, a significant proportion of the open circular plasmids were supercoiled using a complete reaction mixture with the four enzymes in the RCR buffer (Figure A). Remember that using the conventional agarose gel electrophoresis, open circular large plasmids are trapped in the wells, and large linear dsDNAs migrate to the compressed zone, while the supercoiled large plasmids are separated by size.[39,40] Omission of any of PolI, LigA, and GyrAB fully impaired the reaction, while ExoIII contributed to the maximum reaction efficiency (Figure A), probably by eliminating damaged 3′-OH ends[35] and apurinic/apyrimidinic sites.[36] To evaluate the maximum potential of SCR, larger bacterial chromosomes were prepared using the conventional agarose plug method from several E. coli strains and Vibrio natriegens Vmax Express (Codex DNA, Inc.). Without SCR, supercoiled chromosomes were hardly detectable using conventional agarose gel electrophoresis (Figure B), which is consistent with the recent findings that the genomic chromosome in E. coli is often nicked and gapped.[41,42] Furthermore, chromosomes may be damaged during the gel extraction step using β-agarase at 65 °C for 10 min. In contrast, the SCR products of bacterial chromosomes ranging from 1.0 to 4.6 Mb in size were detected and separated by size in a 0.3% SeaKem Gold Agarose gel[43] by the conventional agarose gel electrophoresis performed for 1–1.5 h at room temperature (Figure B,C). The overnight SCR protocol was 24 °C for >10 h (up to 4–5 Mb), while the quick protocol was 30 °C for 30 min (for 200 kb), 2 h (for 3 Mb), and 3 h (for 4–5 Mb), depending on the total length of chromosomes (Table ). Thus, even the wild-type genomic chromosomes of E. coli (4.6 Mb) and V. natriegens (3.2 Mb + 1.9 Mb) were supercoiled and manipulated for the subsequent analyses. In summary, we noticed that open circular bacterial chromosomes can be supercoiled in vitro in a similar fashion as for small plasmids.

Figure 1

Supercoiling and repair reaction (SCR) supercoils open circular forms of bacterial chromosomes. (A) Agarose gel electrophoresis analysis (0.5% gel) of 200 kb plasmids incubated with indicated enzymes. The 200 kb plasmid had been heat-treated to introduce nicks (70 °C for 50 min). The open circular plasmids were trapped in the wells, while the supercoiled plasmids migrated into the gel. All four enzymes were essential or important for SCR. (B) Agarose gel electrophoresis analysis (0.3% SeaKem Gold gel, 0.5 × Tris borate-EDTA (TBE), 40 V 60 min) of the SCR products of the two chromosomes of E. coli HF053 cells extracted in an agarose plug. Because megabase-sized chromosome samples as well as their SCR products are often highly viscous, a few replicates of each sample were applied and analyzed throughout this study. (C) Agarose gel electrophoresis analysis (0.3% SeaKem Gold gel, 0.5 × TBE, 40 V 90 min) of the SCR products of bacterial chromosomes ranging from 1.0 to 4.6 Mb in size.

Table 1

SCR/L-SCR and RCR Protocols

SCR/L-SCR: 37 °C for a few hours or 24 °C overnight
10× SCR/L-SCR buffer 1	1 μL
10× SCR/L-SCR buffer 2	1 μL
10× SCR/L-SCR enzyme mix	1 μL
template DNA	1 μL (up to 2 μL)
Milli-Q water	6 μL (or 5 μL)
total volume in the PCR tube	10 μL
RCR: 20 cycles at 37 °C for 30 s and 22 °C for 60 min
10× RCR buffer 1	1 μL
10× RCR buffer 2	1 μL
10× RCR enzyme mix	0.5 μL
20× RecGJ ExoIII	0.25 μL
60 ng/μL λ phage DNA	0.5 μL
2% DMSO	0.5 μL
40% dextran	0.625 μL
template DNA	1 μL
Milli-Q water	4.625 μL
total volume in the PCR tube	10 μL

Size Separation of Supercoiled Bacterial Chromosomes

To evaluate the resolution of the conventional agarose gel electrophoresis using 0.3% gels, we used E. coli chromosomes of various sizes: 0.84, 1.02, 1.12, 1.31, 1.67, 1.86, 1.96, and 2.14 Mb. These chromosomes were prepared from several bipartite-genome strains of E. coli. The pairs of 1.02 Mb + 1.96 Mb and 1.12 Mb + 1.86 Mb were from HF053 and HF054, respectively,[9] while the pairs of 0.84 Mb + 2.14 Mb and 1.31 Mb + 1.67 Mb were from newly developed bipartite-genome strains YGF017 and YGF012, respectively. The two chromosomes of YGF017 and YGF012 were constructed by the dual Flp-POP cloning method (Figure A), a modified Flp-POP cloning method[31] developed for cloning two genomic regions into a single BAC vector. Upon a Flp-POP cloning, a genomic region flanked by FRT sites is flipped out by flippase as an origin-less circular DNA and integrated into a BAC vector by HK022 phage integrase.[31] This Flp-FRT recombination combines the N-terminal and C-terminal halves of an antibiotic selection marker gene intervened by the FRT sites.[31] The 1.31 and 1.67 Mb chromosomes were confirmed by PFGE analysis (Figure B), while the 0.84 and 2.14 Mb chromosomes were confirmed using SCR (Figure S1A). These eight split-chromosomes that ran for 2 h on a 0.3% agarose gel were separated by size (Figures C and S1B). Two chromosomes of very similar sizes may not be resolved in the same lane. We believe that the conventional agarose gel electrophoresis provides acceptable resolution for supercoiled circular chromosomes in the range of about 1–2 Mb. The dual Flp-POP cloning method is useful for constructing large chromosomes of any size. The combination of SCR and the conventional agarose gel electrophoresis may greatly facilitate experiments using megabase-sized circular chromosomes.

Figure 2

Evaluating the resolution of the conventional agarose gel electrophoresis method for supercoiled circular chromosomes within the range of about 1–2 Mb. (A) Scheme for the dual Flp-POP cloning method to develop a bipartite-genome E. coli strain YGF012. Two genomic regions were marked by two FRT sequences and two FRT3 sequences via λ-red recombination, together with four kinds of the indicated antibiotic marker genes. Abbreviations: hyg for hygromycin, tet for tetracycline, bsd for Blasticidin S, zeo for Zeocin, cat for chloramphenicol, spec for spectinomycin, and kan for kanamycin. Upon induction of the flippase, the two FRT/FRT3-flanked regions were popped-out to develop two circular DNA molecules, while the main chromosome gained the full-length cat and spec marker genes carrying an FRT/FRT3 sequence inside their open reading frames which had been separated by FRT/FRT3 sequences. Upon induction of the HK022 phage integrase, the two circular DNAs were integrated into the two attBHK022 sites in the BAC vector via the attB-attP recombination to develop a new 1.31 Mb chromosome. (B) PFGE analysis (1% PrimeGel Agarose GOLD 3-40K, 0.5 × TBE, 24 h, 6 V/cm, 120°, 60–120 s switch time ramp) of the chromosomes of six colonies showing resistance to spectinomycin and chloramphenicol. Five clones have the bipartite-genome configuration, while one has a single chromosome. Clone #5 was renamed YGF012. The marker chromosomes from yeast cells did not migrate in a fully size-dependent manner. (C) Agarose gel electrophoresis analysis (0.3% SeaKem Gold gel, 0.5 × TBE, 40 V 120 min) of the SCR products of bacterial chromosomes ranging from 0.84 to 2.14 Mb in size.

Enzymatic Replication of Bacterial Chromosomes

RCR is another way to produce supercoiled chromosomes. Our previous study suggests that RCR has the potential to replicate oriC-containing chromosomes of up to 1.1 Mb in size.[31] Although the agarose gel electrophoresis analysis is a direct way of detecting supercoiled molecules produced by RCR, it was difficult to properly estimate the size and yield of these molecules. Using SCR products as supercoiled DNA size marker, we examined the maximum potential of RCR. RCR is composed of about 26 kinds of recombinant proteins and amplifies oriC-containing plasmids/chromosomes in an exponential manner.[39] One of the main differences between this reconstituted system and the native cellular system is the lack of the replication restart mechanism from the site where a replisome dropped off.[44] Another difference is the lack of mismatch repair systems. Thus, the processivity of the replisome limits the maximum size of RCR-amplifiable chromosomes. To prepare 1.12 and 1.96 Mb chromosomes containing an oriC locus, the chromosomes of bipartite-genome E. coli strains HF054 (1.12 Mb + 1.86 Mb) and HF053 (1.96 Mb + 1.02 Mb) were extracted in agarose plugs. Since RCR includes the SCR enzymes, nonreplicated but repaired and supercoiled products should be discriminated from RCR-replicated chromosomes. In principle, SCR products of E. coli cell-derived chromosomes are methylated and enzymatically digestible with DpnI, while RCR-replicated chromosomes are unmethylated and thus resistant to DpnI. This is the same as PCR products amplified from E. coli plasmids.[45]

First, we performed a preliminary experiment to find an RCR protocol for the 1.12 Mb chromosome. Agarose plugs were made using either 108 cells or 109 cells. RCR reaction mixtures were incubated for 10–40 cycles at 37 °C for 30 s (for oriC firing) and 22 °C for 60 min (for extension). The SCR product from a 109 plug yielded a clear band for 1.12 Mb, while the 1.12 Mb band from a 108 plug was faint (Figure A). As expected, DpnI-digested SCR products lack bands for any large DNA (Figure A). When using a 108 plug, 1.12 Mb bands were visible on the gel after >20 cycles of RCR amplification and became clear after 40 cycles. These supercoiled 1.12 Mb chromosomes remained after DpnI treatment (Figure A), indicating that this oriC-containing chromosome was replicated by RCR. Using a 109 plug, clear bands for DpnI-resistant chromosomes were observed even after 10 cycles (Figure A). Longer incubation rather produced a bunch of byproducts which made the reaction solution highly viscous. Then, the 1.96 Mb chromosome was amplified by RCR using this protocol (109 plug and 10 cycles) and was shown as resistant to DpnI (Figure B). Next-generation sequencing (NGS) analysis of chromosomes before and after RCR was performed. The RCR-amplification rates of the 1.12 and 1.96 Mb circular chromosomes were estimated as 9 times and 4 times, respectively, by roughly comparing the read depths (Figure C). We also observed a heap of reads around the oriC locus, indicating that the linear RCR byproducts were caused by the spontaneous drop-off of the replisomes and/or double-strand break during the reaction.

Figure 3

Evaluating the maximum size of bacterial chromosomes replicable by replication-cycle reaction (RCR). (A) Agarose gel electrophoresis analysis (0.3% SeaKem Gold gel, 0.5 × TBE, 40 V 60 min) of the SCR and RCR products of the HF054 chromosomes. The SCR and RCR products were incubated or not incubated with DpnI prior to the electrophoresis. The number of RCR thermal cycle steps was indicated. (B) Agarose gel electrophoresis analysis of the SCR and RCR products of the HF053 chromosomes. The condition was fixed to an agarose plug with 109 cells and 10 RCR thermal cycle steps. (C) NGS analyses of extracted chromosomes before and after RCR with the alignments to the HF033 reference genome (single chromosome). The relative average read depth of the oriC-containing chromosome compared to the oriC-less chromosome increased by about 9 and 4 times for HF054 and HF053, respectively. As for the after-RCR samples, many short reads were aligned and heaped on/around the oriC locus on the reference genome, indicating frequent failures during chromosome replication from oriC. (D) Agarose gel electrophoresis analysis of the SCR and RCR products of the HF053 and HF054 chromosomes. The usual amount of RCR and half the amounts of RCR and SCR enzymes were used, as indicated. The RCR thermal cycle steps were 20.

We then examined RCR mixtures with half the amount of enzymes. We assumed that the usual amount of enzymes was excess for a small number of megabase-sized chromosome molecules as the original RCR mixture was developed for the exponential amplification of small plasmids. After 20 cycles of RCR amplification, clear bands of DpnI-resistant 1.12 and 1.96 Mb supercoiled chromosomes were detected (Figure D). Thus, we employ the half enzyme RCR protocol for >1 Mb chromosomes (Table ). It is also indicated that the processivity of the reconstituted E. coli replisome in the bidirectional replication mode is up to 1.0 Mb. RCR replication of a 3 Mb chromosome has not yet been achieved in a reproducible manner. We think RCR is potentially very useful if the replisome processivity and the replication fidelity (reportedly ∼1.2 × 10–8 per base per replication cycle)[39] would be greatly improved. On one hand, SCR is very handy and oriC-independent.

Electroporation-Ready Chromosomes Prepared by SCR

A supercoiled 1 Mb chromosome can be implanted into E. coli via electroporation if the chromosome is genetically stable inside the host cells.[9] In contrast, stretched 1 Mb DNA molecules are not only fragile but also even longer than a bacterial cell. Although purified and desalted DNA samples are usually used for electroporation, it is ideal if SCR products could be directly used for electroporation to avoid any physical damages of purification. We modified the buffer composition of SCR and found potassium acetate (KOAc) and dithiothreitol (DTT) unnecessary in the SCR buffer. Thus, we omitted KOAc and DTT. The final salt concentration was reduced from about 200 mM to about <50 mM (10 mM Mg(OAc)2, 20 mM Tris–HCl, 10 mM ammonium sulfate, and minor ingredients). This Low Salt Concentration SCR was named L-SCR (Table ). L-SCR successfully produced supercoiled 1, 2, and 3 Mb chromosomes (Figure A). To examine the transferability of L-SCR products via electroporation, we prepared a 1.02 Mb chromosome carrying a kan gene and a spectinomycin resistance gene (spec) at the opposite poles to facilitate antibiotic selection. Two microliter aliquots of the L-SCR solution of the dual-marker 1.02 Mb chromosome were mixed with a vial of HST08 competent cells (from Takara Bio, Inc.) and electroporated without arcing. One transformed colony (in reality, three neighboring colonies) was obtained using five vials of the competent cells. The agarose gel electrophoresis analysis of its L-SCR product detected the 1.02 Mb chromosome (Figure B). The NGS analysis of the transformed cells confirmed the SCR result (Figure C) and detected no spontaneous mutation. When aligning against the host genome, greater read depth was observed on the overlapping regions of the host genome and the 1.02 Mb chromosome. Because the 1.02 Mb region is derived from a genome-reduced E. coli strain, discontinuous regions of a total 1.02 Mb overlap over the 1.66 Mb region of the wild-type HST08 genome (Figure C). An important tip is that chromosomes and competent cells should be well mixed by gentle pipetting using a wide bore tip, as reported.[9]

Figure 4

Electroporation-ready chromosomes can be prepared by low salt concentration SCR (L-SCR). (A) Agarose gel electrophoresis analysis (0.3% SeaKem Gold gel, 0.5 × TBE, 40 V 90 min) of the L-SCR products of bacterial chromosomes ranging from 1.02 to 3.2 Mb. (B) L-SCR products of the 1.02 Mb chromosome of AO19 were introduced into E. coli HST08 cells via electroporation. One hit colony (one of the three neighboring colonies) was analyzed by L-SCR. Agarose gel electrophoresis analysis (0.3% SeaKem Gold gel, 0.5 × TBE, 40 V 70 min) of the L-SCR products from AO19 and the transformed HST08 clone was performed. (C) NGS analysis of the HST08 clone carrying the 1.02 Mb chromosome for confirming the full-length 1.02 Mb chromosome in HST08. Mapping to the HST08 reference genome showed discontinuous heaps of the duplicated regions (in total 1.02 Mb). The ori-kan and dif-spec loci and the two junctions (indicated with arrows) used for colony PCR are shown on the maps. (D) Genetic modification of the 1.02 Mb chromosome in HST08 to remove the AvrII restriction site and the SCR incubation of the original and modified 1.02 Mb chromosome with or without AvrII (0.1 U/μL). Agarose gel electrophoresis analysis (0.3% SeaKem Gold gel, 0.5 × TBE, 40 V 60 min) of the SCR products was performed to examine their AvrII sensitivity and resistance.

To confirm the genetic stability in vivo as well as the physical stability in vitro, we removed the sole AvrII restriction site of the 1.02 Mb chromosome in the transformed strain by inserting a chloramphenicol resistance gene (cat) via λ-red recombination (Figure D). The 1.02 Mb chromosome remained episomal because it was detected from the modified strain (Figure D). The original and modified 1.02 Mb chromosomes were incubated in SCR mixtures with or without AvrII. While the original chromosome disappeared in the presence of AvrII, the modified ΔAvrII chromosome was supercoiled regardless of AvrII (Figure D).

We repeated the 1 Mb L-SCR and electroporation experiments. To increase the DNA amount, we prepared agarose plugs using 4 × 109 cells. In the first trial, we got 9 hit colonies using 5 vials of competent cells. All colonies were confirmed to have the dual selection markers at the top and bottom and two PCR-checkable regions on the left and right sides of the 1.02 Mb chromosome (Figure C). The next trial using the same plugs yielded 4 hit colonies using 5 vials (colony number: 1, 1, 1, 1, 0). We changed the L-SCR protocol from 30 °C for 3–5 h to 24 °C for 14 h because the agarose plugs contain 4× more DNA than normal plugs. The last trial using the same plugs yielded 34 hit colonies using 6 vials (colony number: 18, 0, 7, 7, 1, 1). All hit colonies were positive by PCR. We newly prepared another plug and repeated the experiment (Figure S2). When L-SCR is successful, the L-SCR reaction solution is less viscous. The photo of 19 transformed colonies on a single selection agar plate and the result of their colony PCR check were shown (Figure S2). Thus, 1 Mb electroporation was reproducibly and routinely achieved with the optimized protocol.

L-SCR is superior to the anion-exchange column purification method[9] in some points. Agarose plugs can be stored for months and is usable for L-SCR on demand, whereas the purified DNA solution should be prepared at the time of use (empirically, within a week). Also, purified DNA solutions are usually highly viscous. While a large volume of fresh liquid culture (750 mL) is required in the BAC purification protocol, agarose plugs can be made by scraping cells grown on agar plates. Furthermore, the BAC purification method was limited to up to 1.12 Mb chromosomes. To test even larger chromosomes, we needed a >1 Mb chromosome which can be maintained as an extra chromosome in E. coli cloning cells. Otherwise, it is difficult to find transformed colonies. In the next section, we sought a way to improve the genetic stability of the 1.12 Mb chromosome.

Constructing a Portable Megabase-Sized Chromosome

As mentioned in the introduction, the 1.12 Mb chromosome of the three-partite genome E. coli strain (1.12, 0.84, and 1.02 Mb) was not portable by electroporation nor bacterial conjugation.[9] In what format should a portable megabase-sized chromosome be constructed? We invented a portable chromosome format (PCF) (Figure A). A portable chromosome should have proper replication origin(s) and chromosome partitioning system(s), at least two selection marker genes at the opposite poles, and a chromosome terminal (Ter)[46] region located at the opposite side of the origin(s). To validate our idea, we tried to clone of the 1.12 Mb region into a BAC vector via the oriT-POP cloning method that was employed to assemble a conjugally transferable 1 Mb chromosome.[31]

Figure 5

Portable chromosome for genome swap. (A) Portable chromosome format. (B) Procedure for the assembly of a portable 1.12 Mb chromosome in HST08 ΔattB using the pGCORE-star BAC vector via the two-step oriT-POP cloning from HF033-derived strains. The oriT-mediated transfer and circularization of the left and the right halves of the 1.12 Mb region are indicated with blue arrows and dotted lines, respectively. The HK022 integrase-mediated attB-attP integration is indicated with green arrows. The attBv and attPv pair is a variant of the wild-type pair attB and attP. The developed chromosome confers resistance to kanamycin, spectinomycin, zeocin, and tetracycline. (C) Procedure for the genome swap via the conjugation-based methods. Because RecA-deficient strains of E. coli were used, circular chromosomes were transferred as a circular chromosome upon conjugation. The whole genome sequences of HF054, BG8, and BG123 strains were verified by NGS. (D) L-SCR products of the 1.12 Mb chromosome of BG8 were introduced into E. coli HST08 cells via electroporation. Agarose gel electrophoresis analysis (0.3% SeaKem Gold gel, 0.5 × TBE, 40 V 90 min) of the SCR products from BG8 and the transformed HST08 clones was performed to detect the intact 1.12 Mb chromosome in the transformed HST08 clones. (E) NGS analysis of one the HST08 clones carrying the 1.12 Mb chromosome for confirming a full-length 1.12 Mb chromosome in HST08. Mapping to the HST08 reference genome showed discontinuous heaps of the duplicated regions (in total 1.12 Mb). The oriC, kan, and tet loci and the colony PCR site (indicated with arrows) are shown on the maps.

The 1.12 Mb region carries the oriC region and five ribosomal RNA operons (rrnABCDE) as well as many other housekeeping genes (Figure B). A new 1.12 Mb chromosome was developed by assembling three pieces via oriT-POP cloning in HST08 ΔattBHK022.[31] The 30 kb core region containing the oriC and the rrnC operon constitutes the BAC vector together with the sopABC genes for chromosome partitioning, a kanamycin resistance gene (kan) for selection, two terB sequences[47] to prevent rolling circle replication, two attB sequences[31] as cloning sites, and an artificial mini-Ter macrodomain[46] containing matS and KOPS sequences[48] inserted in the 8 kb region around the dif sequence of pRahaq201 derived from Rahnella aquatilis.[31,49] The developed 42 kb minichromosome, pGCORE-star (Figure B), was stable in HST08 and had a spontaneous mutation in the oriC (oriC*). In the oriC* variant, the GATC methylation site #2[50] is changed to TATC, which sometimes enriches during RCR amplification. The left region (from after phoU until after lysS) and then the right region (from after maoSP until after polB) of the 1.12 Mb region were cloned into pGCORE-star from the 2.98 Mb genome strain HF033 (endA::tet) (Figures B and S3A,B). To facilitate the NGS analysis of the reconstituted 1.12 Mb chromosome (Figure S3B), this chromosome was transferred to a bipartite-genome strain HF054 (1.12 Mb + 1.86 Mb) via conjugation (Figure S4A). We obtained a new bipartite-genome strain BG8 that is controlled by the reconstituted and sequence-verified 1.12 Mb chromosome and the original 1.86 Mb chromosome (Figures C and S4B–E). The NGS result is summarized in Figure S4, while chromosome map and sequence data are provided in the Supporting ZIP file. Since the tetracycline resistance gene (tet) in the left region is located at the opposite side of the kan gene on the 1.12 Mb chromosome, it obeys our PCF (Figure A). To allow antibiotic selection at high concentrations, these marker genes were expressed from strong artificial promoters.[31]

The new 1.12 Mb chromosome from BG8 cells was supercoiled by L-SCR. After electroporation using HST08 electrocompetent cells, six transformants were obtained from one vial of electrocompetent cells, while no transformant was obtained from the other four cuvettes. Interestingly, this hit cuvette was once arced at 1600 V without sparks and used for the second electroporation at 1400 V without arcing. The agarose gel electrophoresis analysis of SCR products detected the 1.12 chromosome in the six transformants (Figures D and S5). The NGS analysis of one of the six transformants confirmed the SCR result (Figure E). Thus, the new 1.12 Mb chromosome is portable. It also meant that a full set of portable split-chromosomes was obtained. We then prepared L-SCR mixtures using 2 μL aliquots of DNA solution extracted from agarose plugs made of 4 × 109 BG8 cells to increase the DNA concentration (Table ). In one experiment, 10 μL aliquots of the L-SCR solutions were mixed with 250 μL of competent cells, and 9 hit colonies were obtained using 10 vials (10 × 50 μL) of competent cells (Figure S6A). All hit colonies were positive by PCR and by dual antibiotic selection (Figure S6A). Four additional colonies that appeared after 2 days of incubation were also positive (Figure S6A). In another experiment, we obtained 49 hit colonies using 18 vials of competent cells (Figure S6B). Again, 26 additional colonies appeared after 2 days of incubation, and they all were positive (Figure S6B). We obtained only two false-positive colonies, indicating that the dual antibiotic selection was highly effective. Among the 75 clones, 48 were derived from three selective agar plates. For them, 2 μL aliquots of the L-SCR solutions were mixed with each vial of competent cells. The way of mixing DNA and cells seemed important. Under the optimal condition, four colonies per vial were obtained on average.

As for the 1.86 Mb chromosome of BG8 cells, we obtained no transformant. On the other hand, via conjugation, the 1.86 Mb chromosome of BG8 was replaced by the 1.86 Mb chromosome of RGF123SH, another bipartite-genome strain (Figure C). The obtained bipartite-genome strain BG123 (Figures C and S7A–C) has experienced a two-step complete genome swap from HF054. Future studies would increase the electroporation efficiency and stabilize a 2 Mb extra chromosome in E. coli cells.

Materials and Methods

Agarose Plugs

Agarose plugs were prepared using a new protocol slightly modified from the previous one.[9] In principle, cells were incubated on LB agar plates and harvested. A cell pellet of 1010E. coli cells was suspended in 0.5 mL of 50 mM ethylenediaminetetraacetic acid (EDTA) (pH 8.0) and well-vortexed. As for V. natriegens Vmax (Codex DNA, Inc.) cells, the EDTA solution was diluted with the same volume of 2× V2 salt (final 204 mM NaCl, 4.2 mM KCl, 23.14 mM MgCl2) to avoid cell lysis. Low melting point agarose solutions (1.5%) prepared using PrimeGel Agarose LMT 1–20 K (Takara Bio, Inc.) and 50 mM EDTA (pH 8.0) were preheated at 45 °C. Briefly, 0.5 mL aliquots of the cell solution and the agarose solution were mixed by vortexing and immediately poured into 10 molds. Agarose plugs were cooled for >1 h in the refrigerator. One agarose plug gel is about 100 μL in volume and contains about 109 cells. Each agarose plug gel was incubated with lysozyme in 1 mL of 50 mM Tris–HCl (pH 7.5) in a 2 mL tube for a few hours at 35 °C with shaking at 300 rpm using Eppendorf ThermoMixer C. The plugs were then incubated in 0.3 mL of a lysis buffer [25 U/mL Proteinase K, 1% sodium lauryl sarcosine, 0.5 M EDTA (pH 9.0)] overnight (for 16–20 h) at 50 °C with shaking at 300 rpm. The plugs were then washed with 1 mL of a wash buffer [50 mM EDTA (pH 9.5)] for 1 h on ice and with 1 mL of the wash buffer supplemented with phenylmethylsulfonyl fluoride (PMSF) (1 mM final) for another hour on ice. PMSF was removed by washing the plugs with 1 mL of Tris–EDTA (pH 8.0). The washed plugs were preserved in 1 mL of Tris–EDTA (pH 8.0) in the refrigerator for up to several months. A small block of an agarose plug (about 108 cells) was digested using 0.5 μL of Thermostable β-agarase (Nippon Gene) at 65 °C for 10 min to elute the DNA molecules and was then cooled on ice.

Replication-Cycle Reaction (RCR)

The RCR reagents were modified from the original receipt[39] for replicating megabase-sized chromosomes. In addition to the supplementation of RecG, RecJ, ExoIII, and λ phage DNA to the reaction,[31] dimethyl sulfoxide (DMSO) and dextran were added at the final concentrations of 0.1 and 2.5%, respectively. The new lot of RCR enzyme mix (0004Y) contains 2.8× LigA compared to the original one.[39] Reaction mixtures were incubated for 10–40 cycles at 37 °C for 30 s and 22 °C for 60 min. After the RCR incubation, the reaction mixtures with 1× enzymes were diluted by five times with 1.25× RCR buffer with or without DpnI (final 0.4 U/μL) and incubated for an additional hour at 30 °C. The reaction mixtures with 0.5× enzymes were diluted by two times with 1.4× RCR buffer with/without DpnI and incubated at 22 °C. This second incubation step finalizes the last chromosome replication cycle. DpnI cleaves methylated and hemi-methylated GATC sites during the second incubation step.

Supercoiling and Repair Reaction (SCR)

The 10× SCR mix includes 1 mg/mL bovine serum albumin (BSA) (Roche), 140 nM LigA (Takara Bio or NIPPON GENE), 170 nM of PolI (Takara Bio), 200 U/μL ExoIII (Takara Bio), and 250 nM of GyrAB (Takara Bio) and was diluted using a dilution buffer [final concentrations: 10% glycerol, 20 mM Tris–HCl (pH 7.5), 0.1 mg/mL BSA, 8 mM DTT, 10 mM Mg(OAc)2, and 125 mM KOAc]. The RCR buffer[39] is the SCR buffer. The 10× L-SCR mix was prepared by omitting KOAc and DTT from the dilution buffer. The L-SCR buffer is also devoid of KOAc and DTT. Usually, 1 μL aliquot of eluted DNA solutions was incubated in a 10 μL reaction volume in a PCR tube using ProFlex PCR System (Applied Biosystems). The DpnI reaction of SCR products was performed in the same manner as for RCR products. Electroporation was performed as described[9] using ELEPO21 (Nepa Gene).

Agarose Gel Electrophoresis

The conventional agarose gel electrophoresis analyses were performed essentially as reported.[31] For megabase-sized chromosomes, 0.3% gels were made using SeaKem Gold agarose (Lonza)[42] and run for 1–2 h at a 40 V constant voltage in 0.5× TBE buffer. The PFGE analysis was performed as reported.[9] DNA ladder markers were Nippon Gene Marker 3 and Gene Ladder 100, and ExcelBand 1 KB (0.25–10 kb) DNA Ladder DM3100 (SMO). The gels were stained with dsGreen (Funakoshi) and scanned by a Typhoon FLA 9500 (GE Healthcare). Images were developed and edited using NIH ImageJ.

Plasmids

Plasmids were constructed essentially in the same manner as described.[9,31] For constructing pGCORE-star, we used Q5 High-fidelity DNA polymerase (NEB) and RA-RCR[39] (equivalent to the OriCiro Cell-Free Cloning System of OriCiro Genomics, Inc.). pBAD-traRP4min-pac was developed by inserting a puromycin N-acetyltransferase gene (pac) under control of the EM7 promoter into the PstI site inside the ampicillin resistance gene (bla).

Chromosome and Cell Engineering

Chromosomes were modified via λ-red recombination using PCR products and several helper plasmids (Table ), as reported previously.[9] Since the chromosomal integration of an oriT cassette and large DNA cassettes was very inefficient in DGF-298W ΔrecAX strains, we expressed the native recAX operon from a helper plasmid. New chromosomes were assembled via dual Flp-POP cloning method or via oriT-POP cloning method[31] using a new helper plasmid pKD46-int. The dual Flp-POP cloning method was updated from Flp-POP cloning method[31] for the nearly simultaneous pop-out and cloning of two genomic regions. To achieve this, expanded combinations of antibiotic resistance genes were employed. After induction of the flippase and HK022 integrase from pMW118gent-flp-int,[9] the two genomic regions were each circularized and cloned into the two HK022 attB sites (attB and attBv) in the BAC vector pVtu9xV2, while the main chromosome gained the full-length cat and spec genes. Spectinomycin resistance colonies were subjected to chloramphenicol selection, and vice versa, to finally obtain bipartite-genome strains. The portable 1.12 Mb chromosome was constructed in HST08 ΔattB carrying pGCORE-star via the two-step oriT-POP cloning[31] of the left and right halves of the 1.12 Mb region. Because the insertion of the original attPv-zeo-oriT cassette[31] between the polB and leuD genes were not successful, we added an rrnC terminator in front of the attPv sequence. pGCORE-star has two cloning sites (attB and attBv).[31] We realized that HF033 as well as DGF-298W had experienced an occasional genomic inversion between the rrnD and rrnE operons[51] during the establishment of DGF-298W from W3110. Unlike W3110, DGF-298W and its descendants have a typical configuration of the ribosomal RNA operons in the genome.[51] The left half was first cloned into attB site to develop the half-sized chromosome. Then the right half was cloned into the attBv site of the half-sized chromosome to develop the full-length chromosome. In addition to the combination of pGCORE-star and HST08 ΔattB, we also examined pGCORE-wt carrying the wild-type oriC sequence and HST08 ΔattB Δtus deficient in the replication fork trap system, as demonstrated in Supporting Figure S3. We chose the original combination because two of the authors individually succeeded in the construction. Bacterial conjugation was performed as described[9,31] but with a choice of two helper plasmids carrying either of the ampicillin and puromycin selection markers. By an unknown reason, DGF-298W derivatives exhibited a native resistance to puromycin. A strain list is provided (Table ). The sequence information of DNA cassettes and homology arms are provided (Supporting Tables S1 and S2).

Table 2

List of Helper Plasmids and BAC Vectorsa

name	description
pKD46[52]	ts, λ-red recombination
pMW118-gba[31]	λ-red recombination
pMW118-g’ba[31]	λ-red recombination, Δgam, “curable”
pMW118-recAX-g’ba	λ-red recombination, Δgam, E. colirecAX expression, hardly “curable”
pMW118gent	gent, “curable”
pKD46-int	ts, HK022 integrase expression
pMW118gent-flp-int[9]	flippase and integrase expression, “curable”
pBAD-traRP4min[31]	RP4 tra expression, ΔoriT
pBAD-traRP4min-pac	RP4 tra expression, ΔoriT, bla::pac
pOri2spec-recAX[31]	E. colirecAX expression, “curable”
pVtu9xT[31]	V. tubiashiiori2-based BAC vector with oriT
pVtu9xF[9]	V. tubiashiiori2-based BAC vector without oriT
pVtu9xV2	V. tubiashiiori2-based BAC vector, pVtudif2[31] derivative lacking crtS and 1× parS2
pGCORE-star	BAC vector for developing a portable 1.12 Mb chromosome
pGCORE-wt	pGCORE-star variant with the wild-type oriC

ts indicates that the plasmid is temperature sensitive. “Curable” indicates that the indicated helper plasmid is easily curable in DGF-298W-derived strains.

Table 3

List of E. coli and V. natriegens Strainsa

name	notes
DGF-298W[24]	E. coli K-12 W3110 derivative DGF-298WΔ100:revΔ234:SC
RGF008C[9]	DGF-298W ΔsacB-cat ΔterFIJ ΔrecAX
RGF123SH	RGF123-derived,[9] [ChrOri and ChrLR+Ter (pVtu9xT) with RCRori-kan::spec-hyg] with pBAD-traRP4min
HF014	DGF-298W ΔsacB-cat ΔterFIJ Δtus kan-oriC@ybgL-gltA hyg-oriC@glpC-menF endA::tet
HF025	HST08 tus::zeo with a 530 kb plasmid[31]
HF033[9]	RGF008C derivative with endA::tet
HF053[9]	RGF008C-derived, endA::tet, [ChrOri+LR and ChrTer (pVtu9xF)]
HF054[9]	RGF008C-derived, endA::tet, [ChrOri and ChrLR+Ter (pVtu9xF)]
HF055	HF033 derivative with kan@ybgL-gltA
AO13	HST08 with pMSR227 (205 kb)[39]
AO19	HF053-derived, [ChrOri+LR and ChrTer with spec@dif]
HST08	F–, endA1, supE44, thi-1, recA1, relA1, gyrA96, phoA, Φ80dlacZΔM15, Δ(lacZYA-argF)U169, Δ(mrr-hsdRMS-mcrBC), ΔmcrA, λ–
HST08 ΔattB[31]	no attBHK022 site
HST08 ΔattB Δtus[31]	used for cloning genome regions with terB sequences
YST01[9]	HST08 carrying ChrLR originated from RGF152[9]
YST03[9]	HST08 carrying ChrTer originated from HF053
SKN112	HST08 ΔattB carrying portable ChrOri, pKD46-int, pBAD-traRP4min-pac, and pGCORE-star
BG5 (no. 1)	HF054-derived, [portable ChrOri and pGCORE-star from SKN112 and ChrLR+Ter]
BG7	BG5-derived, [portable ChrOri and pGCORE-star and ChrLR+Ter with cat@murJ-rne]
BG8	BG7-derived, [portable ChrOri and ChrLR+Ter with cat@murJ-rne]
BG123	BG8-derived, [portable ChrOri and ChrLR+Ter from RGF123SH]
YGF012	RGF008C-derived, [Chr (1.67 Mb) and Chr (1.31 Mb using pVtu9xV2)]
YGF017	RGF008C-derived, [Chr (2.14 Mb) and Chr (0.84 Mb using pVtu9xV2)]
V. natriegens[53] Vmax Express	a clean genome strain of V. natriegens (Codex DNA, Inc.)

Abbreviations of split-chromosomes according to the previous study:[9] ChrOri, ChrLR, ChrTer, ChrOri+LR, and ChrLR+Ter denote 1.12, 0.84, 1.02, 1.96, and 1.86 Mb chromosomes, respectively.

NGS

The NGS analyses of some genomes, chromosomes, and plasmids were performed using iSeq 100 (Illumina) and NEBNext Ultra II FS DNA Library Prep with Sample Purification Beads and Multiplex Oligos for Illumina (NEB). The FASTQ data were analyzed using Geneious Prime (Biomatters Ltd.). DNA sequence and map data were edited using SnapGene (GSL Biotech LLC) and provided as a Supporting ZIP file. The draft genome sequences of HST08 and DGF-298W were assembled previously.[31]