E.coli DH5α, Top10, XL10-Gold, DH10β, and Rosetta blue(DE3) for gene cloning were stored in our lab. The expression vectors pET23a and pET28a were obtained from Novagen (United States). Plasmid pUC57 was obtained from Invitrogen (United States). Plasmids pET28a-gfp, pET28a-mCherry, and pET28a-Kan for multi-segment assembly were stored in our lab. Luria–Bertani (LB) and SOC media for the cultivation of E. coli were prepared as described in the Manual of Molecular Cloning. Appropriate antibiotics were added whenever necessary.

PrimeSTAR® Max DNA polymerase was purchased from TaKaRa (Japan). T5 exonuclease and all restricted enzymes were purchased from New England Biolabs (NEB, United States). The Gibson Assembly® Cloning Kit was purchased from NEB (United States). DNA purification kits were purchased from Omega (United States). All oligonucleotides were synthesized by Shanghai Sangon Biological Engineering Technology & Services Co., Ltd. (Shanghai, China). All other chemicals were analytical reagents.

The vector backbones were obtained through either restriction enzyme digestion or PCR. Digestion of the vectors was performed as described in the instructions given by the New England Biolabs (United Kingdom). PCR was performed as described in the manual for PrimeSTAR® max DNA polymerase (TaKaRa, Japan). The PCR products were digested with DpnI to remove the plasmid template, followed by resolving with agarose gel electrophoresis and purifying with the gel purification kit (Omega, United States).

Inserts were gained by PCR with two homologous ends introduced to the PCR products via primer pairs. All the primers used for the amplification of the inserts are listed in Supplementary Table S1.

The T5 exonuclease was 10-fold diluted with dH₂O and stored at 4°C. A 5 µL reaction system was included—3.5 μL of the mixture of the linear vector and insert at a molar ratio of 1:3, 0.5 μL of NEBuffer 4, and 1 μL of diluted T5 exonuclease (0.5 U). The mixture was incubated on ice for 5 min, followed by the addition of 50 μL of E. coli DH5α competent cells for transformation.

E.coli DH5α competent cells were prepared according to the previous report (Green and Sambrook, 2018). The transformation efficiency was approximately 3 × 10⁸ cfu/μg. The competent cells (50 μL) were thawed on ice and added to the DNA sample prepared by the TLTC method. The mixture was incubated on ice for 30 min, followed by heat shock at 42°C for 45 s. After cooling the ice for 2 min, 900 μL of the SOC medium was added, followed by shaking at 37°C for 1 h. The samples were diluted and spread on LB agar plates supplemented with the appropriate antibiotics. The plates were incubated at 37°C overnight. The transformation efficiency is defined by the number of colonies, and the recombinant efficiency is evaluated by the percentage of positive colonies against total colonies.

According to a previous report, the optimal reaction temperature of T5 exonuclease is 37°C, and it has high activity under this condition. To generate short sticky ends, the sample has be treated for an extremely short time, which is challenging in practical application and lacking of generality. Therefore, we chose to decrease the activity of T5 exonuclease by decreasing the reaction temperature and the amount of enzyme molecules in the reaction system. Therefore, T5 exonuclease was diluted to 1 U/µL, and 1 µL enzyme was added to pUC57 linearized with EcoRV (100 ng). The samples were incubated at 0°C for 3–5 min, followed by the addition of Klenow to remove 3′ overhangs formed by T5 exonuclease. Then, the fragments were cyclized with T4 DNA ligase and transformed into E. coli (Figure 2A). Colonies were selected randomly for DNA sequencing. The 127-bp sequence around the EcoRV site was aligned, and the result indicated that the length of the sticky ends generated by T5 exonuclease increased with the extension of treatment. The average 10-nt ends were formed in 3 min and average 15-nt ends were formed in 5 min (Table 1). This result proves that T5 exonuclease retained a weak activity at 0°C and the short sticky ends could be generated under this condition.

Next, we established the TLTC method based on this condition. To test this method, a gfp fragment was amplified and cloned into pUC57 plasmid. The result indicated that the transformation efficiency of the cloning increased as the temperature decreased and reached a maximum at 4°C (Figure 2B). Interestingly, the result at 0°C was similar to that at 4°C, which indicated that the T5 exonuclease retained activity at 0°C. Since 0°C is easy to gain with ice water mixture, this temperature was used for all the following experiments. The optimal concentration of T5 exonuclease used for the TLTC method was investigated. The result indicated that only 0.5 U of the enzymes was required for a 5 µL reaction system (Figures 2C, D). We also investigated the optimal digestion time of T5 exonuclease. The result indicated that the transformation efficiency reached the maximum after 4–6 min of digestion, and the recombination efficiency was >99% (Figures 2E, F).

The effect of the length of the homologous ends flanking the DNA fragments to the recombination efficiency was tested. First, we mixed the DNA fragments and vector backbones baring the homologous ends of different lengths. The mixture was directly transformed into E. coli DH5α. The result indicated that the transformation and recombination efficiencies increased as the compatible ends extended and a minimum 30-nt homologous region was necessary for high recombination efficiency of >95% (Supplementary Figures S1A, B). Meanwhile, we treated the DNA mixture with T5 exonuclease before the transformation. In comparison to the sample without treatment, the number of colonies on the plates increased 10-fold and recombination efficiencies reached >99% with a homologous region of only 15 nt (Figures 2G, H). These results propose that the single-stranded DNA facilitated DNA cloning in TLTC, and a single-stranded homologous region of 15–20 nt was enough for efficient cloning. The results are consistent with previous reports.

The optimal molar ratio of the inserts and vector backbones was investigated. The concentration of the vectors was fixed at 30 fmol as the amount of inserts increased. The result indicated that TLTC worked at a wide range of molar ratios and the best ratio was 1:3 (Supplementary Figure S2), which is consistent with the conventional gene cloning methods.

To evaluate the host strain on TLTC efficiency, E. coli DH5α, XL10-gold, Top10, DH10β, and Rosetta blue(DE3) were used to test their compatibility with TLTC. The transformation efficiency of these competent cells was approximately 1 × 10⁷ cfu/μg. As shown in Supplementary Figure S3, DH5α, XL10-gold, and Top10 showed higher efficiency for cloning than DH10β and Rosetta blue(DE3).

Multi-segment assembly was carried out using the TLTC method. All fragments were mixed and treated with T5 exonuclease at 0°C for 5 min, followed by transformation of E. coli. For two-fragment assembly, pUC57 provided the vector backbone, while the plasmid pET28a-gfp provided the gfp fragment as the insert. The positive colonies were green under blue light. For the three-fragment assembly, pUC57 provided the vector backbone, while the plasmids pET28a-gfp and pET28a-mCherry provided gfp and mCherry fragments as the inserts. The positive colonies could express gfp and mCherry simultaneously and showed orange fluorescence under blue light. For four-fragment assembly, pET28a-Kan provided the kanamycin-resistant cassette as the third insert. The positive colonies were orange and represented the resistance to kanamycin. The results showed that the transformation efficiency decreased obviously as the number of fragments increased (Figure 3A). On the other hand, the recombinant efficiency still remained high, which was approximately 80% for the four-fragment assembly (Figure 3B).

TLTC is a simplified and versatile system derived from the Gibson assembly. The methods were compared for efficiency with the two-fragment or four-fragment assembly (Supplementary Figure S4). In the two-fragment assembly, when the homologous region was 10 bp, TLTC had greater efficiency than the Gibson assembly, and when the homologous region was extended to 15 bp or 20 bp, TLTC was comparable to the Gibson assembly (Supplementary Figure S4A). In the four-fragment assembly, the Gibson assembly provided greater efficiency (Supplementary Figure S4B).

Promoter, ribosome-binding site (RBS), and terminator strengths determine protein mass fractions and synthesis rates. It is necessary to construct from scratch a full bacterial transcription unit when screening for functional regulatory elements or engineering host cells for improved target protein production. In this research, three DNA fragments (Supplementary Figure S5A): TF1 (containing T7 promoter and RBS), TF2 [containing coding sequence of superfolder GFP (sfGFP)], and TF3 (containing T7 terminator) were assembled into PUC57 with TLTC to construct an entire transcription unit. The TLTC reaction mixture was transformed into Rosetta blue(DE3) competent cells and induced using IPTG. As shown in Supplementary Figure S5B, most of the colonies (>80%) exhibited green fluorescence after 18 h of incubation.

A previous report indicated that the short branch was formed if the annealing took place between 3′-ends of the inserts and the distal sequence several nucleotides away from the 3′-ends of the vector backbone (Figure 4A). The branch was later removed by the E. coli DNA-repairing system. Hence, short oligonucleotides at the ends of the vector backbone can be eliminated through this asymmetric homologous recombination. In the present study, we tested the efficiency of TLTC in removing the extra nucleotides at the ends of the vector backbones. The result indicated that non-homologous sequences of less than 12 nt were eliminated through TLTC if both ends were not fully compatible with the insert, while non-homologous sequences reaching 30 nt were removed if only one end was not fully compatible (Figure 4B).