The plasmids, primers, and proteins used in this study are shown in Table 1; Supplementary Tables S1, S2, respectively. The synthesized initiator tRNA genes, including the ProK promoter and terminator (Mj-itRNA-1 and Mj-itRNA-2; the DNA sequences are shown in Supplementary Table S3) were cloned into the modified pEVOL (Young et al., 2010) plasmid, in which one copy of the azidophenylalaninyl-tRNA synthetase gene (AzF-RS) is located under the AraBAD promoter (Lee et al., 2015), using the ApaLI and XhoI sites. The resulting plasmids were named pSEPL773 for Mj-itRNA-1 and pSPEL541 for Mj-itRNA-2. The methionyl-tRNA transformylase (MTF) gene was amplified (primers 1 and 6) from the chromosome of E. coli DH10β and then cloned into the plasmid containing itRNA-1 (pSPEL773) or itRNA-2 (pSPEL541) using the NdeI and PstI sites. The two PstI sites present in the MTF gene were removed by assembly PCR using primer pairs (primers 2 and 3 for the first PstI site; primers 4 and 5 for the second PstI site). The MTF gene was located under the glnS promoter. The resulting plasmids were named pSPEL527 (MTF/Mj-itRNA-1) and pSPEL528 (MTF/Mj-itRNA-2), respectively. The initiator factor-2 (IF-2) gene was amplified (primers 7 and 14) from the chromosome of E. coli DH10β and then cloned into the plasmid containing itRNA-1 (pSPEL773) or itRNA-2 (pSPEL541) using the NdeI and PstI sites; the three PstI sites present in the IF-2 gene were removed by assembly PCR using two pairs of primers (primers 8 and 9 for the first PstI site; primer 10 and 11 for the second PstI site; primers 12 and 13 for the third PstI site). The resulting plasmids were named pSPEL780 (IF-2/Mj-itRNA-1) and pSPEL562 (IF-2/Mj-itRNA-2), respectively. Plasmids expressing both MTF and IF-2 were constructed by cloning the IF-2 gene with a ribosome-binding sequence (Kinneer et al., 2019) into the PstI site of pSPEL527 or pSPEL528, and the orientation of the RBS-IF-2 genes was confirmed by DNA sequencing. The resulting plasmids were named pSPEL781 (MTF/IF-2/Mj-itRNA-1) and pSPEL563 (MTF/IF-2/Mj-itRNA-2), respectively.

A DNA double helix for a multiple cloning site (NcoI-BamHI-HindIII-NotI-XhoI-His₆ Tag) prepared by annealing primers 15 and 16 was phosphorylated using T4 polynucleotide kinase and then ligated into the pBbE6a (Lee et al., 2011) plasmid digested with NdeI and XhoI. The RBS-GFP gene with the TAG codon at the translation initiation site was generated by PCR amplification using the 11.3.3 GFP gene (Yoo et al., 2007) as a template with primers 17 and 18, and the product was cloned into the modified pBbE6a mentioned above using the EcoRI and HindIII sites, resulting in pSPEL530. The synthesized gene of the Z domain (the amino acid sequence in Supplementary Table S2) with an amber (TAG) codon at the translation initiation site was cloned into pQE-80L (Qiagen) using the EcoRI and HindIII sites, resulting in pSEPL542. The Z domain gene was amplified using pSPEL542 as a template with primers 19 and 20; the product was cloned into pGEX-4T-1 (Amersham Bioscience) EcoRI and XhoI sites. To introduce an amber (TAG) codon between the glutathione-S-transferase (GST) and Z domain, an annealed DNA double helix of primers 21 and 22 was phosphorylated using T4 polynucleotide kinase and then cloned into the pGEX-4T-1 with the Z domain gene using the BamHI and EcoRI sites, resulting in pSPEL236.

The plasmid coding Z domain (pSPEL542) or GFP (pSPEL530) with an amber (TAG) codon at the translation initiation position was co-transformed with one of the pEVOL derivatives (pSPEL731, 541, 527, 528, 780, or 562) and pBbS2K-ProRS (Lee et al., 2015) encoding E. coli prolyl-tRNA synthetase under the tetracycline-inducible promoter into E. coli DH10β. The cells were grown in 2xYT media containing 200 μg/ml ampicillin, 34 μg/ml chloramphenicol, and 35 μg/ml kanamycin at 37°C until the OD₆₀₀ reached 0.5. To induce AzF-RS and prolyl-tRNA synthetase expression, 0.2% L-arabinose and 50 nM anhydrotetracycline were added to the culture. When OD₆₀₀ reached 1.0, 1 mM isopropyl β-D-thiogalatoside and 1 mM ncAA were added to express the model protein. After 3 h, the cells were harvested by centrifugation at 14,000 × g at 4 C for 20 min, and the cell pellets were stored at −20°C until use.

The cell pellets were resuspended in 1% SDS solution and heated at 95°C for 5 min for lysis. The supernatant obtained after centrifugation at 14,000 × g for 5 min was subjected to a copper (I)-catalyzed click reaction with biotin-PEG₃-azide (Click Chemistry Tools) following the method described (Hong et al., 2009). The reaction mixtures were analyzed by western blotting using a streptavidin-horseradish peroxidase (HRP) conjugate (Click Chemistry Tools) and an anti-His₅-HRP conjugate (Sigma-Aldrich, St. Louis, MO).

The Z-domain protein with an N-terminal His6 tag was purified using a Ni-immobilized resin (Clontech, Mountain View, CA) under native conditions, following the manufacturer’s instructions. The cell pellets stored at −20 C were resuspended in lysis buffer (50 mM sodium phosphate, 300 mM sodium chloride, 20 mM imidazole, pH 7.4) and incubated on ice for 30 min with 50 μg/ml lysozyme. After sonication, the lysed cells were centrifuged at 9,300 × g at 4°C for 1 h. The supernatant was incubated at 4°C for 1 h with Ni resin, pre-equilibrated with lysis buffer. The resin was loaded onto a gravity flow column (Thermo Fisher Scientific, Waltham, MA) and washed three times with washing buffer (50 mM sodium phosphate, 300 mM sodium chloride, 40 mM imidazole, pH 7.4). The protein was eluted using elution buffer (50 mM sodium phosphate, 300 mM sodium chloride, 300 mM imidazole, pH 7.4). The protein solutions were buffer-exchanged with phosphate buffer saline (PBS) solution (10 mM KH2PO4, 150 mM NaCl, pH 7.4) using a centrifugal filter unit (Millipore, 3000 MWCO).

The intact masses of the proteins were analyzed using a Waters ACQUITY I class UPLC system (Milford, MA) with an ACQUITY UPLC Protein BEH C4 column (2.1 mm × 100 mm, 1.7 μm particle size; Waters). The mobile phases were 0.1% formic acid in water (eluent A) and 0.1% formic acid in acetonitrile (eluent B). The gradient applied was: 0–3 min, 5% eluent B; 3–13 min, linear increase to 50% eluent B at 0.2 ml/min. The eluent was injected into a Thermo Orbitrap Elite (Thermo Fisher Scientific, Waltham, MA) and ionized with an electrospray source. MS spectra were acquired in the mass range of 400–2,000 m/z and 120,000 resolution at m/z 200. The deconvoluted mass spectra were generated using Protein Deconvolution 2.0 (Thermo Fisher Scientific, Waltham, MA).

The purified Z-domain protein was transferred to a PVDF membrane (0.45 μm pore size; Pall), and the membrane was washed with ultrapure water and dried in air. The membrane was subjected to an N-terminal sequencing analysis using Procise® LC492 Protein Sequencing System (Applied Biosystems, Waltham, MA). The retention times of phenylthiohydantoin (PTH)-amino acids were compared with those of the standards (Tokyo Chemical Industry, Tokyo). PHT-OpgY was synthesized following a previously published method (Steiman et al., 1985).

Even though Mj-itRNA-2 enabled the initiation of protein synthesis at the amber codon, its interactions with the factors involved in forming the 30S initiation complex might be relatively weak compared with those of Ec-tRNA^fMet. An approach to address this would be to increase the concentrations of these factors instead of engineering the tRNA further. The three interactions of MTF, 30S ribosome, and IF-2 with Mj-itRNA-2 aminoacylated with OpgY could be considered to increase translation initiation with OpgY. We decided to explore the overexpression of MTF or IF-2; increasing the concentration of the 30S ribosome could affect various aspects of physiology and thus was not tested in this study. The genes for MTF or IF-2 were cloned under the constitutive Gln promoter of pEVOL-AzF; a biscistronic gene was used to express MTF and IF-2 simultaneously. Additional expression of MTF significantly improved the incorporation of OpgY into the amber codon at the initiation position of GFP, while the effect of overexpressing IF-2 was marginal (Figure 3C). This observation suggested that the formylation of the α-amine group of OpgY-Mj-itRNA-2 was the rate-determining step in translation initiation with the TAG codon. Expressing IF-2 in addition to MTF via the biscistronic construct decreased the signal compared with the case of expressing only MTF. This might be because the expression of IF-2 decreased the level of MTF due to the limited resources for transcription and translation in expressing multiple genes. The aminoacyl-tRNA synthetase of AzF-RS was originally engineered to activate AzF (Chin et al., 2002) and incorporate AzF into the initiation TAG codon was attempted using the same E. coli strain. However, unlike OpgY, Mj-itRNA-2 did not support protein translation initiation with AzF (data not shown). In the case of Mj-itRNA-1, additional expression of MTF or IF-2 did not result in detectable incorporation of OpgY into the initiation TAG codon (data not shown).

A small protein (Z domain (Nilsson et al., 1987)) was used to evaluate OpgY incorporation using a mass spectrometry. The start codon of Z domain was changed to TAG codon, and the resulting gene was cloned into pQE-80 L. The Z domain was expressed using E. coli cells expressing Mj-itRNA-2, AzF-RS, MTF, and ProRS in the presence of OpgY. The purified protein was analyzed by liquid chromatography-mass spectrometry. The calculated mass of the Z domain with OpgY at its N-terminus was 8,364.06 Da, and a mass of 8,364.17 was detected (Figure 3D). This result, along with the western blot results (Figures 3B,C), indicated that OpgY was incorporated into the TAG codon of the translation initiation position. However, another mass peak (8,291.15 Da) was observed in the Z-domain sample. Since the mass of the wild-type Z domain (Met at its N-terminus) is 8,294.03 Da (Supplementary Figure S3: the observed mass was 8,294.12), it was suspected that a canonical amino acid besides Met was incorporated into the TAG codon. The N-terminal residue of the Z domain was determined by the Edman degradation analysis, and two residues, OpgY and Gln, were detected (Figure 3E). The peaks for Gln and OpgY were integrated (597 for Gln and 1477 for OpgY), and the ratio of OpgY to Gln incorporated into the TAG position was calculated as 2.55. The mass of 8,291.15 Da was consistent with the calculated mass (8,290.96 Da) when Gln was incorporated into the TAG codon.

Next, we examined whether Mj-itRNA-2 could function as an elongator tRNA. The two main elements of Ec-tRNA^fMet, the C1:A72 mismatch, and the three consecutive G:C pairs in the anticodon stem, are known to distinguish the initiator tRNA from elongator tRNAs and to prevent it from being involved in the translation elongation process (Laursen et al., 2005). In addition to the two elements, Mj-itRNA-2 had mutations in the acceptor and the D stem; the T stem has also been reported to interact with E. coli elongation factor Tu (Guo et al., 2009). A plasmid was constructed by introducing the TAG codon between GST and the Z domain, and the protein had a C-terminal His₆-tag. The GST-TAG-Z protein was expressed with either Mj-tRNA^Tyr (Figure 2B) or Mj-itRNA-2, and the cell lysates were analyzed by western blotting after the Cu(I)-catalyzed click reaction with biotin-azide. While OpgY was efficiently incorporated into the TAG codon with the elongator Mj-tRNA^Tyr, Mj-itRNA-2 could not suppress the internal nonsense codon (Figure 4). These results indicate that Mj-itRNA-2 1) is charged with ncAA, 2) initiates protein translation at the TAG codon, and 3) is not active in the elongation process of protein synthesis. Mj-itRNA-2 is the first reported tRNA enabling the orthogonal incorporation of ncAA into the amber codon only at the translation initiation position.