Aminoacyl-tRNA Specificity of a Ligase Catalyzing Non-ribosomal Peptide Extension

Abstract

Peptide aminoacyl-transfer ribonucleic acid ligases (PEARLs) are amide-bond-forming enzymes that extend the main chain of peptides by using aminoacyl-tRNA (aa-tRNA) as a substrate. In this study, we investigated the substrate specificity of the PEARL BhaB_C ^Ala from Bacillus halodurans, which utilizes Ala-tRNA^Ala. By leveraging flexizyme, a ribozyme capable of charging diverse acids onto a desired tRNA, we generated an array of aa-tRNAs in which we varied both the amino acid and the tRNA to dissect the substrate scope of BhaB_C ^Ala. We demonstrate that BhaB_C ^Ala catalyzes peptide extension with noncognate proteinogenic and noncanonical amino acids, hydroxy acids, and mercaptocarboxylic acids when attached to tRNA^Ala. For most of these, the efficiency was considerably reduced compared to Ala, indicating that the enzyme recognizes the amino acid. By variation of the different parts of the tRNA, enzyme specificity was shown to also depend on the acceptor stem and the anticodon arm of the tRNA. These findings establish the molecular determinants of PEARL specificity and provide a foundation for engineering these enzymes for broader applications in peptide synthesis.

Amide bond formation is a critical process for the preparation of peptide- and protein-based therapeutics. − Accessing these structures usually relies on solid-phase/liquid phase peptide synthesis or ribosomal translation. − The recent increased interest in using biocatalysis presents opportunities to incorporate enzymes that form amide bonds into synthetic processes. − Enzyme-based synthesis potentially offers scalability with high efficiency, selectivity, and waste minimization. − One class of amide-bond-forming enzymes are the peptide aminoacyl-transfer ribonucleic acid ligases (PEARLs). − PEARLs utilize aminoacyl-transfer ribonucleic acid (aa-tRNA) to form a new peptide bond at the C-terminus of a peptide (Figure A). , Whereas the specificity for the peptide substrate has been investigated, the factors that determine the specificity for the aminoacyl-tRNA substrate are currently unresolved.

1.

(A) The proposed mechanism of amide-bond formation catalyzed by PEARLs, in this case BhaB_C ^Ala. Here, the cosubstrate Ala-tRNA^Ala is formed by AlaRS. (B) Examination of the substrate specificity of PEARLs using diverse aa-tRNA chimeras generated by flexizyme, the focus of this work. X-LG here represents 4-chlorobenzyl thioester, 3,5-dinitrobenzylester, or cyanomethyl ester, which are leaving groups recognized by flexizyme (details in Supporting Information and Table S1).

Here we investigate this specificity for a representative PEARL from Bacillus halodurans, BhaB_C ^Ala (NCBI accession identifier: BAB05753.1), which naturally utilizes Ala-tRNA^Ala. Our data show that the combination of the correct amino acid and cognate tRNA ensures fidelity in the cellular context, with the anticodon arm and the acceptor stem predominantly determining the tRNA specificity of BhaB_C ^Ala. The enzyme also recognizes the charged amino acid (i.e., attached to the tRNA), but it exhibits some tolerance that may be explored for engineering purposes.

In previous studies on PEARLs, where product formation was assessed through both in vitro assays and in vivo coexpression in Escherichia coli, the aa-tRNA was generated using aminoacyl-tRNA synthetases (aaRSs). Because aaRSs are highly specific toward their cognate aa-tRNAs, often containing editing domains that hydrolyze mischarged tRNAs, further exploration of the aa-tRNA specificity of BhaB_C ^Ala using aaRSs is limited to alanine and natural tRNA isoforms. Therefore, we first investigated whether both isoforms of tRNA^Ala in E. coli (GGC and UGC anticodons) were competent substrates for the BhaB_C ^Ala-catalyzed addition of Ala to its substrate peptide BhaA (NCBI: WP_010898193.1) in vitro. Analysis by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry indicated that both isoacceptors were substrates (Figure S1). The limitation of aaRSs to investigate substrate scope of PEARLs can be overcome using flexizymes, 45- or 46-nucleotide ribozymes that catalyze the formation of aminoacyl-tRNAs from diverse activated amino acids and tRNA sequences (Figure B). − Using flexizymes, we investigated whether BhaB_C ^Ala can extend the substrate peptide with various acids attached to tRNA^Ala. We also investigated the tolerance of BhaB_C ^Ala toward alanine attached to various noncognate tRNA sequences.

We first conducted BhaB_C ^Ala reactions using Ala-tRNA^Ala(UGC) generated by flexizyme instead of AlaRS. The flexizyme reaction products were partially purified by ethanol precipitation, and the concentrations of aa-tRNA were estimated using a previously reported NaIO₄-RNA extension assay (Figure S2). − The reaction efficiency of BhaB_C ^Ala at different aa-tRNA concentrations was analyzed using liquid chromatography–electrospray ionization mass spectrometry (LC-ESI-MS). For all experiments, we used a truncated version of BhaA lacking the first 19 N-terminal residues (Δ19BhaA), which underwent quantitative modification by BhaB_C ^Ala when using Ala-tRNA^Ala(UGC) generated by AlaRS (Figure S3). The BhaB_C ^Ala reaction efficiency increased with higher concentrations of Ala-tRNA^Ala prepared by flexizyme and reached complete modification. These observations demonstrate that the BhaB_C ^Ala enzymatic assay was compatible with flexizyme-prepared aa-tRNA (Figure S4), even though it contains significant amounts (∼65%, Figure S2) of non-aminoacylated tRNA that could serve as a competitive inhibitor (for a general discussion, see the Supporting Information – Materials and Methods).

Having established the use of flexizyme-prepared aa-tRNA in the BhaB_C ^Ala assay, we next examined whether cognate tRNA^Ala(UGC) charged with different acids would be accepted by BhaB_C ^Ala (Figure ). The selected acids, including proteinogenic and noncanonical amino acids, a hydroxy acid, and a mercaptocarboxylic acid, were first chemically activated (Table S1) and used as flexizyme substrates as previously reported. − As described above for Ala-tRNA^Ala, we first estimated the tRNA acylation efficiency with the various amino acids using NaIO₄-RNA extension assays (Figure S5) and then conducted the BhaB_C ^Ala assays at various estimated aa-tRNA concentrations (Figures S6–S16). The highest observed conversion levels in end-point assays are shown in Figure .

2.

Amino acid specificity of BhaB_C ^Ala. E. coli tRNA^Ala was charged with various acids using flexizyme. The acyl-tRNA^Ala chimeras were then tested for BhaB_C ^Ala-catalyzed ligation with Δ19BhaA (1.8 μM, sequence in Figure S3). Reaction efficiencies were evaluated at multiple acyl-tRNA concentrations using LC-ESI-MS; the values shown represent the highest conversions observed. Products were not detected for Aib and N-Me-Ala. All extracted ion chromatograms (EICs) (with the estimated acyl-tRNA concentrations used to achieve the conversions in this figure in parentheses) are provided in Figures S4 (Ala, 4.7 μM), S6 (Gly, 14 μM), S7 (Gln, 15 μM), S8 (Glu, 22 μM), S9 (Phe, 19 μM), S10 (Trp, 31 μM), S11 (β-Ala, 12 μM), S12 (d-Ala, 41 μM), S13 (Aib, 17 μM), S14 (N-Me-Ala, 40 μM), S15 (Lac), and S16 (ThioGly); Lac and ThioGly were not amenable to the NaIO₄-RNA extension assay to estimate concentration.

BhaB_C ^Ala catalyzed amide formation with all evaluated proteogenic aminoacyl-tRNA^Ala analogs with a wide range of efficiencies. Gly-tRNA^Ala (2) was incorporated with efficiency akin to Ala-tRNA^Ala (1) whereas Phe-tRNA^Ala (5) was utilized with moderate activity. Although other proteinogenic amino acids (3, 4, and 6) attached to tRNA^Ala were accepted with significantly lower activity, the observed substrate tolerance is promising for future engineering applications. We next evaluated noncanonical amino acids attached to tRNA^Ala (Figures S11–S14). While BhaB_C ^Ala catalyzed efficient addition of β-Ala (7), much lower efficiency was observed with d-Ala (8). We did not observe peptide extension with 2-aminoisobutyric acid (Aib, 9) or N-Me-Ala (10). BhaB_C ^Ala also catalyzed C–O and C–S bond formation when lactate (11) and thioglycolic acid (12), respectively, were attached to tRNA^Ala(UGC) (Figures S15 and S16), , although product formation was reduced compared to that observed with l-Ala and Gly. The resulting thioester product formed with (12) successfully underwent native chemical ligation with cysteamine (Figure S17). Collectively, these results show that BhaB_C ^Ala appears to have evolved to recognize l-Ala, but other small amino acids and hydroxy/mercaptocarboxylic acids attached to tRNA^Ala(UGC) are also accepted, showing considerable tolerance with respect to the nucleophile. Directed evolution experiments on the enzyme may be able to increase the observed reaction efficiencies for desired substrates.

The observation that various acyl groups attached to tRNA^Ala(UGC) were substrates for BhaB_c ^Ala in vitro, whereas only extension with Ala is observed in E. coli, suggests that the enzyme also recognizes features of the tRNA. Since Gly-tRNA^Ala was an efficient substrate, we first tested whether Gly-tRNA^Gly(GCC) generated by E. coli GlyRS was accepted by BhaB_C ^Ala. This reaction resulted in only moderate conversion (Figure S18) while the reaction with Gly-tRNA^Ala(UGC) at similar concentration of substrates gave quantitative modification (Figure S6), indicating that tRNA identity contributes to substrate recognition. The results described thus far suggest that BhaB_C ^Ala evolved to recognize Ala-tRNA^Ala in vivo based on the combination of the amino acid and the tRNA component. To further examine this hypothesis, we used flexizyme to charge Ala onto various noncognate tRNAs having different sequences (Figures , S19 and S20, Table S2). The tRNAs for Trp, Glu, and Cys were chosen because they are substrates of previously reported PEARLs or PEARL-like enzymes. ,,, As described above, we first estimated the flexizyme-catalyzed aminoacylation efficiency of each tRNA with Ala using NaIO₄-RNA extension assays (Figure S21) and then conducted the BhaB_C ^Ala assays at various estimated aa-tRNA concentrations (Figures S22–S24). The highest observed conversion levels in the end-point assays are shown in Figure . All Ala-tRNA chimera lead to lower BhaB_C ^Ala reaction efficiency compared to Ala-tRNA^Ala, with Pseudomonas syringae Ala-tRNA^Cys(GCA) (2.6%, Figure S22) and E. coli Ala-tRNA^Trp(CCA) (5.9%, Figure S23) being particularly poor substrates.

3.

tRNA specificity of BhaB_C ^Ala. (A) Various chimeric tRNAs with the D-arm, T-stem, or Anticodon Arm of E. coli tRNA^Ala(UGC) and P. syringae tRNA^Cys (GCA) interchanged were aminoacylated with Ala using flexizyme. These Ala-tRNAs were used in the BhaB_C ^Ala-catalyzed ligation with Δ19BhaA (sequence in Figure S3). (B) Mutations in P. syringae tRNA^Cys(GCA) that considerably improved BhaB_C ^Ala activity. In panels A and B, regions derived from E. coli tRNA^Ala are in sky blue and regions from P. syringae tRNA^Cys are in purple. The cloverleaf structure of each tRNA was predicted using tRNAScanSE and Forna. Reactions were performed using a range of Ala-tRNA concentrations and analyzed using LC-ESI-MS; results are summarized in Table S2; values shown represent the highest estimated efficiencies observed. EICs are provided in Figure S4 (E. coli tRNA^Ala), Figure S22 (P. syringae tRNA^Cys), Figure S25 (Chimera 1), Figure S26 (Chimera 2), Figure S27 (Chimera 3), Figure S29 [P. syringae tRNA^Cys (U71A)], and Figure S30 [P. syringae tRNA^Cys (C34G/A35C/U71A)]. (C) AlphaFold3 model of the complex of BhaB_C ^Ala, BhaA, E. coli tRNA^Ala, Mg²⁺, and ATP. The tRNA is colored by structural region: acceptor stem (orange), D-arm (yellow), anticodon arm (green), variable arm (black), and T-stem (blue). The sequence of Δ19BhaA is colored light blue, with the first 19 residues present in full-length BhaA in black. Predicted local distance difference test (pLDDT) scores and predicted aligned error (pAE) plots for the model are provided in Figure S31 and Supplementary Data 1. The protein image was made using Chimera.

To determine which regions of the tRNA were critical for BhaB_C ^Ala recognition, we used flexizyme to charge Ala onto chimeric tRNAs having individual regions of E. coli tRNA^Ala(UGC) replaced with the corresponding sequences from the poor substrate P. syringae tRNA^Cys(GCA) (Figures A, S20, S21, S25–S27, Table S2). We observed quantitative product formation with the chimera containing the D-arm (Chimera 1, Figure A) and T-stem (Chimera 2, Figure A) from tRNA^Cys, albeit a reduction in efficiency was noted with the D-arm chimera at lower aa-tRNA concentrations (Figures A, S25, S26). In contrast, grafting the anticodon arm from tRNA^Cys(GCA) into tRNA^Ala(UGC) (Chimera 3, Figure A) significantly reduced activity, suggesting that the anticodon arm plays an important role in recognition (Figures A, S27). This conclusion was supported by introducing the anticodon arm from the poor substrate E. coli Ala-tRNA^Trp(CCA) (Figure S23) into tRNA^Ala(UGC), which also markedly decreased BhaB_C ^Ala activity (Chimera 4, Figure S28). These results provide an explanation for the poor acceptance of Ala-tRNA^Trp(CCA) and Ala-tRNA^Cys(GCA), but they do not explain why Thermobispora bispora Ala-tRNA^Glu(CUC) was observed to be a more competent substrate (93%, Figure S24).

We next evaluated the acceptor stem for BhaB_C ^Ala activity, a region of importance for related aa-tRNA utilizing enzymes. A variant of the poor substrate P. syringae Ala-tRNA^Cys carrying a substitution in the acceptor stem [P. syringae tRNA^Cys (U71A), Figure B], introduced to match the discriminator base of E. coli tRNA^Ala, showed significantly improved activity compared to the wild-type Ala-tRNA^Cys(GCA) (Figures B, S29, Table S2). Activity was further enhanced when the second and third positions of the anticodon were mutated [P. syringae tRNA^Cys (C34G/A35C/U71A), Figure B] to match that of E. coli tRNA^Ala (Figures B, S30, Table S2), underscoring the importance of the anticodon. These findings also provide an explanation for Ala-tRNA^Glu(CUC) being a moderately competent substrate. Unlike the poor substrates tRNA^Cys and tRNA^Trp, this tRNA^Glu naturally has the same discriminator base and third base in the anticodon as tRNA^Ala (Figure S20). The collective results also agree well with an AlphaFold3 model of BhaB_C ^Ala, BhaA, and E. coli tRNA^Ala(UGC), in which the enzyme primarily interacts with the acceptor stem and the anticodon arm of the tRNA (Figures C, S31). Similar interactions were previously observed in an Alphafold3 model of the PEARL BhaB_C ^Trp with E. coli tRNA^Trp, and in a model of AmmB₄ ^Arg with tRNA^Arg. In support of these models, mutation of residues of BhaB_C ^Trp predicted to interact with the acceptor stem of the tRNA abolished enzyme activity. The model of AmmB₄ ^Arg with tRNA^Arg (the first PEARL to use an amino acid with a charged side chain) also provided the first insights into the potential recognition of the aminoacyl group by the enzyme. However, high resolution visualization of the molecular interactions between the anticodon arm, acceptor stem, and aminoacyl group of aa-tRNA with residues on PEARLs will require structural biology studies. The AlphaFold3 model of BhaB_C ^Ala also shows that the first 19 residues of the substrate peptide do not interact with the protein, consistent with the findings above that these residues are not essential for BhaB_C ^Ala activity (Figures S3, S31, S32).

In summary, we used flexizyme to generate various noncognate aa-tRNA pairs and examined the substrate specificity of BhaB_C ^Ala, an amide-forming enzyme that naturally uses Ala-tRNA^Ala. Our data show that the observed specificity of BhaB_C ^Ala toward its cognate aa-tRNA in vivo is determined by both the amino acid and the tRNA component. Since all of our tRNA molecules were obtained by in vitro transcription, we currently cannot rule out that post-transcriptional modifications may contribute further to this specificity. Despite this specificity, in vitro BhaB_C ^Ala can extend peptide chains with a range of amino acids and can catalyze bond formation beyond amides, such as the formation of esters and thioesters. Experimental and structural models indicate that BhaB_C ^Ala recognizes the tRNA through both the acceptor stem and the anticodon region with contributions from the discriminator base and the anticodon sequence. These findings lay the groundwork for future potential directed evolution of BhaB_C ^Ala variants capable of efficiently extending peptide chains with diverse amino acids beyond the native substrate.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.5c12610.

• Experimental procedures, Figures S1–S33 showing assay data and AlphaFold models, and Tables S1–S5 with nucleotide sequences, primers, and MS data (PDF)

• Structure of BhaB_C ^Ala–BhaA–tRNA–ATP complex predicted by AlphaFold3 (PDB)

#.

Department of Pharmacy and Pharmaceutical Sciences, Faculty of Science, National University of Singapore, Singapore 117544

¶.

Department of Chemistry, Graduate School of Science, Kyoto University, Kitashirakawa-Oiwakecho, Sakyo, Kyoto 606-8502, Japan.

This work was supported in part by grants from the National Institutes of Health (GM058822 to W.A.vdD. and GM123998 to D.A.M.), by KAKENHI (JP23H04546, JP24K01634, JP24H01754, and JP24K21760 to Y.G.; JP22H02218 to A.V.; and JP20H05618 to H.S.) from the Japan Society for the Promotion of Science, and by the Adopting Sustainable Partnerships for Innovative Research Ecosystem (JPMJAP2418) from the Japan Science and Technology Agency to Y.G. W.A.vdD. is an Investigator of the Howard Hughes Medical Institute.

The authors declare no competing financial interest.