Article activity feed

  1. Author Response:

    Reviewer #1:

    The authors constructed synthetic tRNAs with different 4-base anticodons and recorded their efficiency in decoding a series of quadruplet codons in Escherichia coli. Phage based library generation and selection was used to survey several quadruplet codons for their potential to incorporate each of the 20 standard amino acids. Additional library-based mutagenesis was used to identify optimal bases as positions surrounding the anticodon. Finally, mass spectrometry was used to identify tRNAs that appear to enable selective or ambiguous decoding of different quadruplet codons. Overall, the manuscript provides exciting new data and represents an important exploration of the potential and limitations of a 4-base genetic code.

    The manuscript should be revised as some statements are not supported by the data. For example, the concluding remark "our deliberate exploration of the evolution of functional quadruplet translation will launch synthetic efforts to assemble a 256-amino acid genetic code." While a complete 4-base genetic code would have 256 codons, the authors have neglected to discuss the potential for degeneracy in that code. Limitations of quadruplet decoding resulting from competition with normal 3-base decoding is not clearly addressed.

    We thank the reviewer for this comment. We have expanded on our discussion of triplet codon competition in the discussion.

    Reviewer #2:

    The manuscript of DeBenedictus et al describes careful and comprehensive investigation of the requirements for translation by tRNAs decoding 4-base codons. There is considerable interest in engineering organisms to use 4 base codons, as it would allow 256 codons to recode to alternative amino acids etc in synthetic biology. Here the authors tested whether there is a fundamental limitation in using natural tRNAs as scaffolds for 4 base codon reading by engineering their anticodon loops. They tested 57 of the possible 256 codon-anticodon pairs, and all twenty isoacceptors. They applied a combination of simple luciferase assays for readthrough of a single 4-base codon with an expressed tRNA mutant, in parallel measuring the growth defects of the different tRNA mutants. Their initial results focused on 4 anticodons, where the last base is repeated twice, to attempt to ensure efficient aminoacylation by codon-recognizing aminoacyl tRNA synthetases. Overall, the efficiencies of the tRNAs for suppression are poor, leading to 1-2% of protein production compared to wild-type triplet decoding in their reporter system, at best. They apply molecular evolution techniques to attempt to optimize the 4 base anticodon context, and show improvement for a tRNAser scaffold by changes in positions 32, 37 and 38 flanking the anticodon. Finally they tested the amino acids incorporated at the 4 base codon position in a test protein, and found that incorporation was homogeneous, with mainly one amino acid incorporated and often it was that encoded by the quadruplet tRNA. However, in several instances, arginine was incorporated irregardless of the identity of the tRNA. This was explained by the low specificity for ArgRS for the anticodon and has been observed previously.

    Overall, the manuscript tackles a complex, important problem in synthetic biology in a comprehensive fashion. The efficiencies of 4 base decoding are exceptionally low, but there is hope presented here that through evolution approaches such efficiencies can be improved.

    We thank the reviewer for these comments.

    I personally would have liked to see both deeper mechanistic and biochemical questions probed here-are the tRNAs modified and where, what are their aminoacylation efficiencies, what indeed are the problems with translational efficiencies? Yet the authors are frank that their purpose is elsewhere, and more a proof of concept that 4 base codons can work comprehensively without crosstalk.

    We thank the reviewer for these comments. Our interest is elsewhere, but we have expanded the discussion of possible sources of limited translational efficiency.

    The writing and precise experiments are often confusing. For example, the nature of the pili selection experiment is not well characterized by the figures.

    Thank you for letting us know this was unclear. We apologize for the brevity and have expanded on how this selection works in the text. It now reads, "We applied an equivalent approach based on the use of a M13 bacteriophage tail fiber pIII as a selection marker. In this selection scheme, a qtRNA is encoded on the genome of a ΔpIII M13 bacteriophage. Phage are challenged to infect bacteria bearing a plasmid that encodes pIII containing a quadruplet codon at permissive residue 29. Functional qtRNAs are capable of producing full- length pIII and thus phage progeny, while non-functional qtRNAs result in production of truncated pIII and thus no further phage."

    The authors should also engage in deeper discussion of what worked-why were certain tRNA anticodons more amenable to decoding than others. Even some speculation would be useful to the community to deepen these studies.

    Thank you for this comment. We have added a paragraph in “Compiled trends in nascent qtRNA evolution” that discusses insights into trends we see in Figure 6A.

    Reviewer #3:

    This manuscript would be appealing to a broad audience, subject to the following revisions:

    1. It would be helpful to explain the criteria that were used to select the 21 E. coli tRNAs (including f-Met) that were the starting point for this study. Given a choice, it seems the authors preferred either G or C at the third codon position. They do not appear to have taken into account codon usage in E. coli or an effort to maximize orthogonality among the chosen codons.

    Thank you for this comment. We have added a section to the methods, “tRNA scaffold selection.” It reads, “We used the first scaffold listed in each isoacceptor class, based on ecogene.org listing for E. coli K12 circa January 2019.” Yes, you are correct, we did not take codon usage into account.

    1. Do the various reporter proteins (luciferase, pIII, and sfGFP) tolerate deletion of the amino acid that is encoded by the mutated codon?

    We thank the reviewer for this comment. We have added citations throughout to studies that demonstrate that luxAB-357, pIII-29 and sfGFP-151 are all permissive residues. In the methods section, we have noted the identity of the original residue.

    If so, what is the possibility of a ribosomal frameshift to skip this position? In the mass spec analyses, did the authors seek to detect a tryptic fragment corresponding to deletion of Tyr151? For each reporter protein, it should be noted what is the wild-type amino acid encoded by the mutated codon.

    Thank you for this comment. Unfortunately, the mass spectrometry software we used is not able to search for deletions in the same way as altered residues, and for that reason we did not analyze ribosome skipping in this study. The possibility that quadruplet codons in transcripts are skipped is an interesting one that we would be interested to investigate in the future.

    The Addgene links provided in the Methods section are broken.

    These have been fixed, thank you.

    1. The mass spec analyses are a critical component of this study, but are not mentioned until near the end of the manuscript. The fact that such analyses confirmed the incorporating of the quadruplet-coded amino acid should be stated in the abstract and in the last paragraph of the introduction. Otherwise many readers (including me) will be carrying doubt until those data are presented.

    Thank you for this comment. We have added a mention of mass spectrometry in the introduction, “Many of these selectively incorporate a single amino acid in response to a specified four-base codon, as confirmed with mass spectrometry,” as well as in the introduction, “... we found that 12/20 isoacceptor classes of tRNAs can be readily converted to selectively charged qtRNAs, as confirmed with mass spectrometry. The efficiency is often low, but can often be improved by ...”

    The tryptic digest fragment depicted in Figure 4A appears incorrect. Cleavage would be expected following Lys140 and Lys156, generating a product two residues shorter than what is shown.

    Thank you for this comment; we have corrected the figure. In the data, we do sometimes observe peptides that contain the additional Q157 K158 due to incomplete tryptic digest.

    1. The last sub-section of the Results belongs in the Discussion. In that sub-section the authors discuss the prospects for the combined use of multiple quadruplet codons. It needs to be stated clearly that this has not been done in the present study, although in the Introduction the authors reference prior studies where up to four unique quadruplets were co-translated from a common transcript. Nor does the present study investigate the possibility of multiple occurrences of the same quadruplet codon within one transcript. Based on the reported results with a single occurrence, the effect of multiple occurrences on translation efficiency is likely to be severe. Neither of these qualifiers diminish the significance of the present study.

    We have adjusted the sub-section “Trends in nascent qtRNA evolution” to discuss only the data presented in Figure 6A (compiled results of all qtRNAs tested with luxAB and sfGFP reporters) and 6B (orthogonality measurement). Additional text has been moved to the discussion.

    1. In the discussion regarding the tolerance of aminoacyl tRNA synthetases to altered codon size, the authors make the excellent suggestion that synthetases that arose later during evolution may be more precisely tuned to triple anticodon recognition. It would also be worth noting that the decoding site of the ribosome is likely to have become more precise over the course of evolution. As the proteome expanded, there would have been strong selection pressure favoring increased fidelity of translation, whereas during the early history of life, especially if there were fewer amino acids to distinguish, the entire translation apparatus is likely to have been more permissive.

    This is an interesting point as well, we have added this to the discussion.

    Read the original source
    Was this evaluation helpful?
  2. Evaluation Summary:

    Using a phage-based library generation and selection, the authors generated a suite of 4-base decoding tRNAs with improved efficiency in quadruplet decoding. The data represent an important step toward enhancing protein synthesis with 4-base codons. Overall, the approach to generate many tRNA variants with quadruplet anticodons is intriguing and provides a wealth of valuable information to the field. The results, once some of the reviewer concerns have been addressed, should become foundational for the field of synthetic biology.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #3 agreed to share their name with the authors.)

    Read the original source
    Was this evaluation helpful?
  3. Reviewer #1 (Public Review):

    The authors constructed synthetic tRNAs with different 4-base anticodons and recorded their efficiency in decoding a series of quadruplet codons in Escherichia coli. Phage based library generation and selection was used to survey several quadruplet codons for their potential to incorporate each of the 20 standard amino acids. Additional library-based mutagenesis was used to identify optimal bases as positions surrounding the anticodon. Finally, mass spectrometry was used to identify tRNAs that appear to enable selective or ambiguous decoding of different quadruplet codons. Overall, the manuscript provides exciting new data and represents an important exploration of the potential and limitations of a 4-base genetic code.

    The manuscript should be revised as some statements are not supported by the data. For example, the concluding remark "our deliberate exploration of the evolution of functional quadruplet translation will launch synthetic efforts to assemble a 256-amino acid genetic code." While a complete 4-base genetic code would have 256 codons, the authors have neglected to discuss the potential for degeneracy in that code. Limitations of quadruplet decoding resulting from competition with normal 3-base decoding is not clearly addressed.

    Read the original source
    Was this evaluation helpful?
  4. Reviewer #2 (Public Review):

    The manuscript of DeBenedictus et al describes careful and comprehensive investigation of the requirements for translation by tRNAs decoding 4-base codons. There is considerable interest in engineering organisms to use 4 base codons, as it would allow 256 codons to recode to alternative amino acids etc in synthetic biology. Here the authors tested whether there is a fundamental limitation in using natural tRNAs as scaffolds for 4 base codon reading by engineering their anticodon loops. They tested 57 of the possible 256 codon-anticodon pairs, and all twenty isoacceptors. They applied a combination of simple luciferase assays for readthrough of a single 4-base codon with an expressed tRNA mutant, in parallel measuring the growth defects of the different tRNA mutants. Their initial results focused on 4 anticodons, where the last base is repeated twice, to attempt to ensure efficient aminoacylation by codon-recognizing aminoacyl tRNA synthetases. Overall, the efficiencies of the tRNAs for suppression are poor, leading to 1-2% of protein production compared to wild-type triplet decoding in their reporter system, at best. They apply molecular evolution techniques to attempt to optimize the 4 base anticodon context, and show improvement for a tRNAser scaffold by changes in positions 32, 37 and 38 flanking the anticodon. Finally they tested the amino acids incorporated at the 4 base codon position in a test protein, and found that incorporation was homogeneous, with mainly one amino acid incorporated and often it was that encoded by the quadruplet tRNA. However, in several instances, arginine was incorporated irregardless of the identity of the tRNA. This was explained by the low specificity for ArgRS for the anticodon and has been observed previously.

    Overall, the manuscript tackles a complex, important problem in synthetic biology in a comprehensive fashion. The efficiencies of 4 base decoding are exceptionally low, but there is hope presented here that through evolution approaches such efficiencies can be improved. I personally would have liked to see both deeper mechanistic and biochemical questions probed here-are the tRNAs modified and where, what are their aminoacylation efficiencies, what indeed are the problems with translational efficiencies? Yet the authors are frank that their purpose is elsewhere, and more a proof of concept that 4 base codons can work comprehensively without crosstalk. The writing and precise experiments are often confusing. For example, the nature of the pili selection experiment is not well characterized by the figures. The authors should also engage in deeper discussion of what worked-why were certain tRNA anticodons more amenable to decoding than others. Even some speculation would be useful to the community to deepen these studies.

    Read the original source
    Was this evaluation helpful?
  5. Reviewer #3 (Public Review):

    This manuscript would be appealing to a broad audience, subject to the following revisions:

    1. It would be helpful to explain the criteria that were used to select the 21 E. coli tRNAs (including f-Met) that were the starting point for this study. Given a choice, it seems the authors preferred either G or C at the third codon position. They do not appear to have taken into account codon usage in E. coli or an effort to maximize orthogonality among the chosen codons.

    2. Do the various reporter proteins (luciferase, pIII, and sfGFP) tolerate deletion of the amino acid that is encoded by the mutated codon? If so, what is the possibility of a ribosomal frameshift to skip this position? In the mass spec analyses, did the authors seek to detect a tryptic fragment corresponding to deletion of Tyr151? For each reporter protein, it should be noted what is the wild-type amino acid encoded by the mutated codon. The Addgene links provided in the Methods section are broken.

    3. The mass spec analyses are a critical component of this study, but are not mentioned until near the end of the manuscript. The fact that such analyses confirmed the incorporating of the quadruplet-coded amino acid should be stated in the abstract and in the last paragraph of the introduction. Otherwise many readers (including me) will be carrying doubt until those data are presented. The tryptic digest fragment depicted in Figure 4A appears incorrect. Cleavage would be expected following Lys140 and Lys156, generating a product two residues shorter than what is shown.

    4. The last sub-section of the Results belongs in the Discussion. In that sub-section the authors discuss the prospects for the combined use of multiple quadruplet codons. It needs to be stated clearly that this has not been done in the present study, although in the Introduction the authors reference prior studies where up to four unique quadruplets were co-translated from a common transcript. Nor does the present study investigate the possibility of multiple occurrences of the same quadruplet codon within one transcript. Based on the reported results with a single occurrence, the effect of multiple occurrences on translation efficiency is likely to be severe. Neither of these qualifiers diminish the significance of the present study.

    5. In the discussion regarding the tolerance of aminoacyl tRNA synthetases to altered codon size, the authors make the excellent suggestion that synthetases that arose later during evolution may be more precisely tuned to triple anticodon recognition. It would also be worth noting that the decoding site of the ribosome is likely to have become more precise over the course of evolution. As the proteome expanded, there would have been strong selection pressure favoring increased fidelity of translation, whereas during the early history of life, especially if there were fewer amino acids to distinguish, the entire translation apparatus is likely to have been more permissive.

    Read the original source
    Was this evaluation helpful?