In vitro amplification of whole large plasmids via transposon-mediated oriC insertion

There is growing demand for methods to produce large DNA molecules. In the field of synthetic biology, the entire genome of a Mycoplasma bacterium (1 Mb) can be synthesized artificially using yeast as a host organism [1,2]. In the DNA sequencing field, long-read sequencing technology such as MinION nanopore sequencing (Oxford Nanopore Technologies, Oxford, UK) makes it possible to obtain long read lengths (>100 kb) [3,4]. Because of the size limitation of in vitro DNA amplification techniques such as PCR, preparation of large-sized DNA molecules from a small amount of DNA samples has relied on DNA cloning in yeast or Escherichia coli. However, cell-based cloning is labor-intensive and time-consuming. Moreover, cloning of some specific DNA sequences is difficult; for example, due to cell toxicity.

Recently, an in vitro method for amplification of large circular DNA, the replication cycle reaction (RCR) approach, was developed [5,6]. RCR consists of 26 purified proteins that mediate continuous repetition of the E. coli chromosomal replication cycle, thus enabling exponential amplification of circular DNA, even starting from a single molecule, in an isothermal reaction. RCR can amplify high-molecular-weight DNA up to 1 Mb as intact circular molecules [7], and the amplification fidelity is extremely high (1.2 × 10^-8 errors per base per cycle). Because the initiation reaction for replication by RCR occurs at a replication origin from the E. coli chromosome, oriC, an essential requirement of the circular DNA template for RCR amplification is that it must contain an oriC sequence, which has a minimum size of 245 bp [8]. Despite the powerful amplification potential of RCR, the requirement for oriC has remained as a bottleneck for amplification of various circular DNAs such as plasmids, bacterial chromosomes or organelle DNA.

The plasmid pET-His-Tnp for overexpression of hyperactive Tn5 transposase (E54K/L372P) was constructed by cloning the Tn5 transposase gene into the pET-His vector using a seamless cloning method, as described previously [5]. The Tn5 transposase gene was prepared by PCR using the primers SUE829 and SUE830 from pTXB1-Tn5 [9], which was a gift from R Sandberg (Addgene plasmid # 60240; [10]). The plasmid pRpoABCDZ was constructed by insertion of the rpoDZ gene from MG1655 into pGEMABC [11], a gift from Katsu Murakami (Addgene plasmid # 45398; [12]). The pTT8 plasmid was purified from Thermus thermophilus HB8 (obtained from RIKEN BRC, Tsukuba, Japan). F-plasmid was purified from laboratory stock of E. coli JM109 (K-12 recA1 endA1 gyrA96 thi hsdR17 supE44 relA1 Δ(lac-proAB) (F′ traD36 proAB lacI^qZΔM15)) using NucleoBond Xtra BAC (TAKARA, Kusatsu, Japan), followed by exonuclease treatment.

The plasmid pRpoABCDZ::Km-oriC was prepared by transposon-mediated oriC insertion reaction using a Km-oriC transposon, followed by isolation of a kanamycin (Km)-resistant and ampicillin (Amp)-sensitive transformant.

An OriC-1 transposon was prepared by PCR using primers SUE996 and SUE997 with 5′-phosphate from a synthetic DNA fragment, OriC-1 (Eurofins Genomics, Tokyo, Japan). The OriC-1 transposon sequence was 5′-CTGTCTCTTATACACATCTgaagatccggcagaagaatggagtatgttgtaactaaagataacttcgtataatgtatgctatacgaagttatacagatcgtgcgatctactgtggataactctgtcaggaagcttggatcaaccggtagttatccaaagaacaactgttgttcagtttttgagttgtgtataacccctcattctgatcccagcttatacggtccaggatcaccgatcattcacagttaatgatcctttccaggttgttgatcttaaaagccggatccttgttatccacagggcagtgcgatcctaataagagatcacaatagaacagatctctaaataaatagatcttctttttaatacccaggatccatctatgtcgggtgcggagaaagaggtaatgaaatggctttagttacaacatactcaggtctttctcaagccgacAGATGTGTATAAGAGACAG-3′; capital letters indicate 19-bp mosaic ends (MEs), while underlining indicates the 245-bp minimal oriC. An OriC-2 transposon was similarly prepared by PCR, except that an OL (OriC-loxP) cassette [6] was used as the template. An oriC transposon including a kanamycin-resistance gene (Km-oriC transposon) was prepared by PCR from pPKOZ [5] using primers SUE814 and SUE815 with 5′-phosphate ends. The sequences of the primers are listed in Supplementary Table 1.

The hyperactive Tn5 transposase (55 kDa) was overexpressed in BL21 (DE3) cells harboring pET-His-Tnp. The cells were grown to an OD₆₀₀ of ∼0.75 in LB medium containing 50 μg/ml carbenicillin at 37°C, and further incubated for 4 h at 23°C in the presence of 0.4 mM IPTG (isopropyl β-D-thiogalactopyranoside). The cell lysate, prepared using the lysozyme method, was then subjected to ammonium sulfate (0.28 g/ml) precipitation. The precipitate was resuspended in His-column buffer (20 mM Tris-HCl, pH 7.5, 10% glycerol, 500 mM NaCl, 2 mM 2-mercaptoethanol) including 10 mM imidazole, and applied to a HisTrap HP column (GE Healthcare, IL, USA). After the wash step, peak fractions were eluted with a gradient of 130–230 mM imidazole in His-column buffer and pooled. The pooled fraction was diluted with heparin-column buffer (20 mM Tris-HCl, pH 8.0, 20% glycerol, 2 mM DTT, 1 mM EDTA, 0.1% Triton™ X-100) and subjected to a HiTrap 1-ml Heparin HP column (GE Healthcare) equilibrated with heparin-column buffer including 300 mM NaCl. Peak fractions of Tn5 transposase eluted with a gradient of 490–620 mM NaCl in heparin-column buffer were pooled and stored at -80°C.

For the Tn-oriC insertion reaction, a hyperactive mutant form of Tn5 transposase (E54K/L372P, 116 nM) and an OriC-1 transposon (144 nM) were incubated at 30°C for 30 min in 10 μl of Tn-formation buffer (20 mM Tris-acetate pH 7.5, 30% glycerol, 100 mM potassium glutamate, 2 mM DTT, 0.2 mM EDTA). A portion (0.5 μl) of the reaction was then mixed with plasmid in 5 μl of Tn-insertion buffer (20 mM Tris-HCl, pH 7.5, 150 mM potassium glutamate, 10 mM magnesium acetate, 50 ng/μl tRNA) and incubated at 37°C for 15 min, followed by heat inactivation at 70°C for 5 min. For the emulsion system, the insertion reaction, including 3% PEG, was added to 250 μl mineral oil including 2% ABIL^® EM90 and 0.05% Triton X-100, then mixed by vortexing for 1 min. The average volume of the emulsion droplets was estimated from microscopy images. After elimination of the oil phase by centrifugation and chloroform extraction, a portion of the aqueous phase was used for RCR.

RCR was performed essentially as described previously [5,6]; RCR reagents are available from OriCiro Genomics, Inc. The RCR mixture (3.6 μl) was preincubated at 30°C for 15 min before adding 0.4 μl of the Tn-oriC insertion reaction. Isothermal RCR was performed at 30°C for the indicated periods. If indicated, cycler RCR was performed using 30 or 40 cycles of 37°C for 1 min and 24°C for 30 min, followed by a final reaction in which the reactions were diluted tenfold with RCR buffer and further incubated at 30°C for 30 min to complete the replication reaction. Portions (1 μl) of each of the RCR reactions were mixed with 4 μl of stop buffer (25 mM Tris-HCl, pH 8.0, 25 mM EDTA, 0.1% sodium dodecyl sulfate, 0.05 mg/ml proteinase K, 5% glycerol, 0.1% bromophenol blue) and analyzed by 0.5% agarose gel electrophoresis, followed by SYBR Green I staining (Molecular Probes, OR, USA). Images were acquired with a Typhoon FLA 9500 scanner (GE Healthcare). For amplification of F-plasmid, 60 nM RecG, 0.5 U/μl RecJ, 60 mU/μl Exonuclease III and 3 ng/μl lambda phage DNA were included in the RCR mixture. Lambda phage DNA can stabilize amplification of very large DNA by blocking endonuclease activity caused by slight contamination in in RCR enzymes.

For scarless removal of the OriC transposon, the indicated concentration of Tn5 transposase and 10 ng pRpoABCDZ::Km-oriC were incubated in 5 μl of Tn-insertion buffer for 16 h at 37°C, followed by a further 5-min incubation at 70°C. A portion (1 μl) of the reaction was added into 5 μl of Exonuclease III digestion mixture (50 mM Tris-HCl, pH 8.0, 5 mM MgCl₂, 1 mM DTT, 20 mU/μl Exonuclease III) and then incubated for 10 min at 30°C. Subsequently, a self-annealing reaction was carried out by further incubation for 5 min at 65°C and gradual cooling of the mixture down to 12°C. The excision efficiency and the self-annealing efficiency were analyzed by E. coli (DH5α) transformation assay using an LB agar plate containing 50 μg/ml kanamycin or 100 μg/ml ampicillin.

In order to insert oriC into plasmid DNA in vitro, we used a Tn5 transposition reaction [9,13]. An OriC transposon is a ∼0.5-kb DNA fragment with a 19-bp ME on either side of the oriC sequence (Figure 1). An OriC transpososome (Tn-oriC) was then formed by combining a hyperactive mutant version of the Tn5 transposase with the OriC transposon. The insertion efficiency of the resultant Tn-oriC was verified using an E. coli transformation assay with an OriC transposon that includes a Km-resistance gene (Figure 2). To assess Tn-oriC-dependent plasmid amplification in RCR (Tn-RCR), we used a 15-kb plasmid (pRpoABCDZ). For the insertion reaction, a mixture including pRpoABCDZ and the Tn-oriC complex was incubated at 37°C for 15 min and then the transposase was inactivated at 70°C for 5 min (Figure 1). A subsequent RCR amplification showed that pRpoABCDZ was successfully amplified as a monomer in supercoiled form (Figure 3A). No amplification products were detected in the absence of a Tn-oriC insertion. A concatemer (formed due to rolling-cycle replication) and a slight open circular form (a replication intermediate) were also observed, as has been described previously [5].

We next examined how small a number of plasmid molecules can be used as starting material for successful amplification by Tn-RCR. To do this, we developed a cycler-RCR method in which amplification was performed using 30 cycles at 37°C for 1 min and 24°C for 30 min, an approach that should reduce size bias during amplification. The initiation step in RCR requires a temperature of >28°C, whereas the elongation step proceeds at a lower temperature [14]. The temperature cycling therefore allows synchronization of the replication cycle by arrest of initiation at 24°C while completion of replication is ongoing, followed by release of initiation at 37°C. Because the elongation rate of the single fork is about 250 bp/sec at 23°C [15], the 30-min incubation period at 24°C could allow replication of up to 900 kb of DNA by two bidirectional forks, and was a sufficient period to replicate the 15-kb plasmid in this experiment. Even when present in a very small amount, a smaller-sized plasmid contaminant in the reaction was amplified more easily, due to the rapid replication cycles in isothermal RCR, particularly when the amount of the target plasmid was decreased (Supplementary Figure 1). However, cycler-RCR was able to repress amplification of a smaller-sized plasmid contaminant (Figure 3B). Using cycler-RCR, we detected amplification even when only 500 fg plasmid was used in the insertion reaction. DNA sequencing analysis of the smaller-sized contaminant product revealed that the product was derived from a non-oriC plasmid contaminated very slightly in a purified protein preparation; this observation highlights the amplification sensitivity of Tn-RCR.

In order to amplify a further small amount of plasmid such as environmental DNA, we developed an emulsion system. As a result, we succeeded in decreasing the minimum necessary amount of starting plasmid in the insertion reaction to 50 fg (3000 molecules) (Figure 3C). The emulsion system encapsulates Tn-oriC and plasmid in a small-volume emulsion droplet (∼100 fl), allowing for efficient Tn-oriC insertion even if the amount of plasmid present is very small.

Amplification of GC-rich DNA by PCR is difficult due to the formation of secondary structures [16]. To test Tn-RCR amplification of a GC-rich plasmid, we used T. thermophilus pTT8 plasmid (9.3 kb), which has a GC content of 69%. The results showed efficient amplification of pTT8 plasmid that was dependent on Tn-oriC insertion (Figure 3D & Supplementary Figure 2), which highlights the ability of RCR to amplify GC-rich DNA. We also tested Tn-RCR amplification of very large DNA molecules using the E. coli F-plasmid (232 kb). Although a portion of the linear form was generated during the amplification reaction due to double-strand breaks, the supercoiled form of the F-plasmid was successfully amplified using Tn-RCR (Figure 3E & Supplementary Figure 3).

The amplified product generated using Tn-RCR has an OriC-transposon insertion at a random plasmid position. We next developed a system that removes this OriC-transposon scarlessly from the amplified products. Tn5 transposase has an activity to excise a DNA insertion flanking two ME sequences from a plasmid, and in the case of a transposon-inserted plasmid, the Tn5 excision leaves 9-bp overlap sequences at each end of the excision sites due to the 9-bp duplication generated during the transposon-mediated insertion [13]. We utilized this mechanism for excision of the OriC-transposon insertion and scarless self-circularization via the 9-bp overlapping ends (Figure 4A). To induce the self-circularization reaction, the transposon-excised plasmid DNA was treated with Exonuclease III, and the resultant 5′ overhangs were annealed via the 9-bp cohesive ends. To monitor whether a gene that was disrupted by the OriC-transposon insertion can be recovered by this Tn5 excision and self-circularization system, we used pRpoABCDZ::Km-oriC, which has a Km-oriC transposon, inserted into the Amp-resistance gene by the Tn5 transposition reaction. A mixture including pRpoABCDZ::Km-oriC and Tn5 transposase was incubated at 37°C for 16 h and then the transposase was heat inactivated. After Exonuclease III treatment and annealing, the product was used for E. coli transformation assay. Figure 4B shows a Tn5 transposase-dependent decrease in Km-resistant plasmids, indicating that the Km-oriC transposon was successfully removed. Amp-resistant plasmids were correspondingly recovered, particularly when Exonuclease III was present in the self-circularization reaction (Figure 4C). The results demonstrate that the transposon-mediated oriC insertion can be removed scarlessly through the Tn5 excision and self-circularization. We also confirmed the scarless circularization using Sanger sequencing of the plasmids isolated from the Amp-resistant colonies. In agarose gel electrophoresis analyses of the transposon excision products, almost all of the supercoiled DNA became the linear form using 100 nM Tn5 transposase, and the amount of linear DNA was comparable to that of the reaction using 300 nM Tn5 transposase (Figure 4B, insert). On the other hand, the colony-forming units of either Km-resistant or Amp-resistant E. coli decreased when 300 nM Tn5 transposase was used, probably due to the detrimental effect of excess transposase on the E. coli transformation.

Tn-RCR allows in vitro amplification of various circular DNAs, including GC-rich or large plasmids of up to 200 kb. Because the Tn5 transposition reaction does not require that any sequence information be available for the template plasmid, Tn-RCR will be useful to prepare or identify large circular DNAs such as plasmids, bacterial chromosomes or organelle DNA present in a variety of environments. Furthermore, in human cancer, extrachromosomal DNA carrying oncogenes is known to exist as large circular DNA [17]. Tn-RCR would provide a useful tool for amplification and analysis of such extrachromosomal circular DNA.