The Ribosome and the Origin of Translation

The ribosome's core has a derivable history, and the inferred proto-ribosome has been built and makes peptide bonds. What no laboratory has produced is a process that assigns amino acids to codons. Michael Lynch calls the origin of translation the most enigmatic unsolved problem in the origin of life, and Eugene Koonin reaches for the phrase irreducible complexity to describe it.

Draft:  September 2026

"Of the multiple unsolved problems, the origin of the translation system … remains the most enigmatic." — Michael Lynch and Andrew Ellington, 2026.1

"The origin of cells engenders the ominous specter of irreducible complexity." — Eugene Koonin, 2014, using the apologists' own phrase and then declining their conclusion.2

Summary

The ribosome is the machine that reads the genetic code.  It is the strongest single argument for design in the whole origin-of-life literature, and this article both makes it and cuts it down to what the evidence supports.

  • The ribosome's large subunit has a derivable history, and this must be conceded. Konstantin Bokov and Sergey Steinberg used a directional dependency in the structure to order the additions and strip the molecule back to a small core.3 It is a structural, testable argument, not a story.
  • The inferred proto-ribosome has been built, and it works. Ada Yonath's laboratory made the proposed ancestor and showed several constructs "are indeed capable of mediating peptide-bond formation."4 Three groups have now done it.5
  • What it cannot do is the thing translation requires. The same paper: it "did not possess the ability to support substrate motion from one site of the symmetrical pocket to the other, as required in the contemporary mode of protein biosynthesis."4 A peptide bond is not a coded chain.
  • The root of the model is contested in print, by people who accept the enterprise. Gustavo Caetano-Anollés re-analysed the data and found "an incorrect branch-to-trunk insertion assignment," concluding that "unwarranted assumptions, dismissals of conflicting data, structural insertion ambiguities, and lack of phylogenetic information compromise" the model.6
  • Half of the famous circularity is broken. Ribozymes charge transfer RNA. Suga's flexizyme is 45 nucleotides, gives more than 105 rate enhancement and more than ten-thousand-fold discrimination against non-cognate amino acids.7 "Only a protein can charge a tRNA" has been false since 1995.
  • The half that matters is not broken. Nothing demonstrated decides which amino acid goes on which transfer RNA. A flexizyme is promiscuous by design, which is the opposite of a code. Sixty years of work has produced no physical process that yields a rule connecting a triplet to an amino acid.
  • And the fidelity gap is a number, not an adjective. Modern polymerases misincorporate at 10-6 to 10-4 per site. Non-enzymatic templated replication runs at 0.01 to 0.1.1 Two to four orders of magnitude separate demonstrated prebiotic copying from what a translation system needs.

The verdict.  This is the one argument in the origin-of-life debate that survives the audit in something like the form apologists state it.  The machinery can be taken apart, its core rebuilt, its charging step run without proteins, and its table edited.  What nobody has produced is a process that generates the assignments.

The ribosome can be taken apart

Start with the concession, because it is large and it is usually left out.

The large ribosomal subunit is a 2,900-nucleotide RNA wrapped in proteins.  It looks like the least decomposable object in biology.  In 2009 Bokov and Steinberg showed it is not.

The method is elegant and it is worth understanding rather than summarising.  Ribosomal RNA is full of A-minor motifs, where a stack of adenosines packs into the groove of a double helix elsewhere in the molecule.  These contacts have a direction of dependency.  As Petrov and colleagues restate it, "the donating adenosine in an A-minor interaction is dependent on the accepting double helix, whereas the accepting helix is less dependent on the adenosine.  If the conformation and structure of the rRNA was maintained throughout its evolution, the donating adenosine is the more recent addition."8

Directions chain into an order.  Bokov and Steinberg used that order to remove outermost elements one at a time, each removal leaving a structure that still stands.  What survives is a small core centred on the peptidyl transferase centre, the part that actually makes the peptide bond.  Their abstract:  "a detailed analysis of the structure of the modern ribosome reveals a concerted and modular scheme of its early evolution."3

This is a real result and an argument from design has to answer it.  It is not a narrative.  It is a derivation from physical contacts visible in a crystal structure, and independent groups working from different data and assumptions have converged on broadly the same picture.

Then the model was tested.  Yonath's group built the proposed proto-ribosome:  "after constructing such pockets, we show… that several protoribosome constructs are indeed capable of mediating peptide-bond formation."4  A 2024 review reports that three separate groups have now made such constructs, which "spontaneously fold, dimerize, and catalyze the formation of peptide bonds and of short peptides."5

What the model does not reach

Four limits, three of them stated by the people who built it.

It cannot translocate.  The constructed proto-ribosome makes a peptide bond and cannot move the substrate from one site to the other, "as required in the contemporary mode of protein biosynthesis."4  Bond formation is chemistry.  Translation is bond formation plus movement plus reading, and only the first has been rebuilt.

It covers the wrong subunit.  The model is of the large subunit.  The codon is read in the small subunit, which comes out of these analyses as a late addition.  The part that does the coding is the part the reconstruction does not reach.

It presupposes the transfer RNAs.  A proto-ribosome that joins two amino acids delivered on tRNAs explains nothing about how the amino acids got onto the right tRNAs.

And the root is contested.  Caetano-Anollés re-analysed the insertion data and argues the ribosome grew from the decoding and translocation machinery rather than from the peptidyl transferase centre.  His verdict on the accretion model is blunt:  "Unwarranted assumptions, dismissals of conflicting data, structural insertion ambiguities, and lack of phylogenetic information compromise the construction of an unequivocal insertion-based model."6  Report that fairly.  It is an argument about which end the ribosome grew from, between groups who both accept it grew by accretion.  It is not a challenge to the enterprise.  But "the ribosome has been traced back to a proto-ribosome" should not be printed as settled fact without it.

The circularity, stated precisely

This is the argument that survives, so it is worth stating in a form a chemist would accept rather than in the form that circulates.

What the argument is not.  It is not that the ribosome cannot work without proteins;  its catalytic core is RNA and the proteins in the large subunit are peripheral.  It is not that no molecule can do two jobs.  Anyone stating it loosely will be corrected, and rightly.

What it is.  The mapping from codon to amino acid is currently carried by twenty protein enzymes, the aminoacyl-tRNA synthetases, whose own sequences are specified by that mapping.  The system is needed to build the system.

Half of it is broken

Ribozymes charge transfer RNA, and have since 1995.  Hiroaki Suga's flexizyme is the clean case.  It is "a 45-nucleotide ribozyme capable of charging tRNA in trans with various activated l-phenylalanine derivatives.  In addition to a more than 10(5) rate enhancement and more than 10(4)-fold discrimination against some non-cognate amino acids, this ribozyme achieves good regioselectivity:  of all the hydroxyl groups of a tRNA, it exclusively aminoacylates the terminal 3′-OH."7

That is a real answer to a real objection.  The chemistry of attaching an amino acid to an RNA has been solved without proteins.

And a 2025 result goes further, in a direction that should be conceded.  Szostak's group built "an autocatalytic assembly pathway that generates a chimeric, amino acid–bridged aminoacyl-RNA synthetase ribozyme," which sustains itself over indefinite cycles of serial dilution and also assembles ribozymes distinct from itself.9  A charging catalyst that makes more of itself is a long way from nothing.

The half that matters is not

A flexizyme is promiscuous by design.  Its selling point in synthetic biology is that it will charge almost any transfer RNA with almost any activated amino acid.  That is the opposite of what a genetic code needs.  A code requires that this amino acid goes on that transfer RNA and no other.  The flexizyme's specificity is for the chemical leaving group, not for the pairing of a side chain with an anticodon.

So the honest sentence is:  the chemistry of attaching an amino acid to an RNA has been solved without proteins;  the chemistry of deciding which amino acid goes on which RNA has not.

Sixty years of work has produced ribozymes that acylate, RNA binding sites statistically enriched in their ligand's codons, and models that reproduce the code table from assumed rules.  It has not produced a physical process that, run on plausible starting materials, yields a rule connecting a triplet to an amino acid.

Unlike most arguments in this area, this is not an argument from how little has been tried.  It is a specific, named interdependence that the field has worked on since the 1960s, with money and attention, and the people working on it say it is unsolved.

A serious minority denies the premise

The circularity assumes an RNA world came first.  Charles Carter's laboratory denies it, and the argument deserves reporting because it is the most substantive alternative on offer.

Both classes of aminoacyl-tRNA synthetase cut down to roughly 125-residue fragments, which Carter calls urzymes.  These "both acylate tRNAs ∼106 times faster than the uncatalyzed peptide synthesis rate," and "have only 15–25% of the mass but exhibit 61% of the catalytic proficiency of contemporary aaRSs."10  And the two classes may descend from opposite strands of a single ancestral gene, which would make the two halves of the code a single invention rather than two.11

Carter and Peter Wills draw the conclusion:  "Bootstrapping of increasingly detailed coding is thus intrinsic to polypeptide aaRS, but impossible in an RNA world."12

What that does and does not do.  It removes the circularity by pushing the problem back:  translation began in a peptide-and-RNA world, so the code did not have to wait for protein synthesis.  What it does not do is explain where the first specific peptides came from, and Carter's own paper concedes the point:  urzymes "appear highly evolved, implying that they had even simpler ancestors."

The fidelity gap, in numbers

The most useful recent statement of the problem is from Michael Lynch, one of the most cited evolutionary geneticists alive, writing with Andrew Ellington in 2026.  Their opening:  "Of the multiple unsolved problems, the origin of the translation system (the means by which the genetic code inscribed on chromosomes is converted into reliable protein sequences) remains the most enigmatic."1

They state the chicken-and-egg problem in their own words, and then the compounding one:  "if one assumes that the protoribosomal function was translation, this conundrum is exacerbated by the fact that other translation-associated molecules and reactions, including tRNAs and their synthetases, would need to have been available or invented around the same time."1

And they supply the number that matters more than any of this.  Modern RNA polymerases misincorporate "in the range of 10-6 to 10-4 per nucleotide site per replication," while the earliest life "likely relied upon nonenzymatic templated replication, which has in vitro error rates of 0.01 to 0.1 per nucleotide site.  Thus, understanding how replication error rates evolved to low enough levels to enable sufficient genome size expansion to support a translation system remains one of the central mysteries in the origin of life."1

A translation system needs a genome big enough to encode it, and a genome that size needs copying accuracy two to four orders of magnitude better than anything demonstrated prebiotically.  That is the same error-threshold constraint that limits the RNA world, arriving from a different direction.

Be fair about what Lynch and Ellington are doing.  They are not conceding defeat.  They propose that the proto-ribosome began as a molecular parasite of an RNA-world host, with translation emerging from that mutual dependence, and they say the key steps "can be evaluated in the laboratory."1  They also concede reach:  "short of a time machine, we may never know with certainty how the pre-LUCA endowments of life became established."

Koonin's phrase

The field's most-cited reviewer reaches for the apologists' own vocabulary to describe this problem, and the full quotation is more interesting than the half that circulates.

Eugene Koonin, 2014:  "Given the intricacy of the molecular machinery of even the simplest cells, the lack of any observations on cell origin, except from other cells, and the lack of any apparent intermediate stages in the evolution of cells, the origin of cells engenders the ominous specter of irreducible complexity."2

His next sentence is the qualifier, and this article prints it because its rules require it:  "Of course, this only means that we might still lack the essential data and theoretical models, let alone experimental approaches, to reconstruct the early stages of cellular evolution."2

Koonin is not conceding design.  He is saying the problem has the shape apologists say it has, and then declining to draw their conclusion.  That is the honest state of this argument, and it is why the audit below rates it sound while refusing to make the inference.

The claims, audited

Claim Verdict
"The ribosome cannot have evolved" Not supported. The large subunit's core has a derivable history and has been rebuilt in a test tube
"The ribosome needs proteins to work" False. The catalytic core is RNA; the proteins are peripheral
"Only a protein can charge a transfer RNA" False since 1995. A 45-nucleotide ribozyme does it with 105 rate enhancement
"The proto-ribosome model is settled" Overstated. The root is contested in print and the construct cannot translocate
"Nothing produces a codon assignment" Sound. Sixty years, and no process yields a rule connecting a triplet to an amino acid
"Irreducible complexity applies here" Sound as a description, and Koonin uses the phrase. It is not an inference to design, and he says so in the next sentence
"Prebiotic copying is accurate enough to support translation" False by two to four orders of magnitude

Objections

1. This is god-of-the-gaps

The article makes no inference from the gap.  It reports what has been built, what has been derived, and what has not.  The difference from a gaps argument is that the missing item is named and specific:  a demonstrated process that assigns amino acids to codons.  If one is published, this article is wrong in a way anyone can check.  The parent article on the origin of life states the position in full.

2. The proto-ribosome has been built, so the argument is over

It makes a peptide bond.  It cannot translocate, it is the large subunit only, and it presupposes charged transfer RNAs.  Those three limits are from the builders' own paper and the reviews of it.  What has been shown is that the bond-forming core is decomposable, which is worth conceding and is not the same as explaining translation.

3. Ribozymes charge transfer RNA, so the circularity is broken

Half of it is, and this article says so before arguing.  What a flexizyme does not do is decide which amino acid belongs on which transfer RNA, and it is promiscuous by design, which is the opposite of a code.  The unbroken half is the half that is the genetic code.

4. Carter's urzymes dissolve the problem

They relocate it.  If the synthetases began as 125-residue peptides that acylate a million times faster than nothing, the code did not need a full RNA world first.  But those urzymes "appear highly evolved," in Carter's own words, and where the first specific peptides came from is now the question.  A genuine alternative, and not a solution.

5. You are quoting Koonin against his own position

Deliberately, and with his next sentence attached.  He uses the phrase "irreducible complexity" and then says it only means the data may be missing.  Quoting the first half without the second is the practice this project audits in others.  What his sentence establishes is the shape of the problem, from someone with no motive to state it that way.

6. Translation had to be preceded by simpler coding, so this is a false dichotomy

That is the mainstream position and it is reasonable.  The evidence for it is real:  reconstructed pre-LUCA synthetase ancestors could not tell leucine from isoleucine from valine, and one lacked tryptophan entirely.13  So a simpler, sloppier code did precede this one.  The question the article asks is unchanged by that:  what physical process assigned the amino acids, sloppily or otherwise.

What this article claims, and what it does not

It claims that the large ribosomal subunit has a derivable history, that the inferred proto-ribosome has been built and forms peptide bonds, that ribozymes charge transfer RNA without proteins, and that all of this should be conceded.  It claims that the constructed proto-ribosome cannot translocate, that the model does not reach the subunit which reads the codon, that its root is disputed in print, that nothing demonstrated assigns an amino acid to a codon, and that prebiotic copying accuracy falls two to four orders of magnitude short of what a translation system needs.

It does not claim the ribosome could not have evolved, and it does not infer design from the gap.  Bokov and Steinberg's paper is closed access and was not opened;  the account of its method here is second-hand from three papers that were.  One correction is recorded rather than buried:  this project's own research file attributed a sentence about the models being "in essential agreement" to Petrov et al. 2015;  it is not in the copy on disk, and the article makes the convergence claim in its own words instead.  Those gaps are listed in the literature review.

Sources

  1. 1.

    Lynch, Michael, and Andrew Ellington.  “A symbiotic origin of the ribosome?”  PNAS Nexus 5:pgag019.  2026.  A Perspective article by Michael Lynch of Arizona State University and Andrew Ellington of the University of Texas at Austin.  It argues that the origin of the translation system, the means by which the genetic code is turned into reliable proteins, remains the most enigmatic of the unsolved problems in the origin of life, and it proposes that the protoribosome began as a parasite that coevolved with its host.  The article cites it for its statement of the problem and its numbers, not as a solution, because the authors offer the hypothesis as testable rather than established. ↩ ↩

    1. 1a.

      “Although the RNA polymerases in today's organisms have misincorporation rates in the range of 10-6 to 10-4 per nucleotide site per replication (37, 38), the earliest stages of life likely relied upon nonenzymatic templated replication, which has in vitro error rates of 0.01 to 0.1 per nucleotide site (35, 39–41).  Thus, understanding how replication error rates evolved to low enough levels to enable sufficient genome size expansion to support a translation system remains one of the central mysteries in the origin of life.”  This is the number that carries the argument. ↩ ↩

    2. 1b.

      “Ribosomes are so entrenched in the biology of today's organisms that even their transient loss is equivalent to a death sentence.  Their sole role is to make proteins.  Yet, a large fraction of ribosomal biomass and that of their interacting partners consists of dozens of proteins, raising in some minds a 'chicken or egg' type of problem.”  “if one assumes that the protoribosomal function was translation, this conundrum is exacerbated by the fact that other translation-associated molecules and reactions, including tRNAs and their synthetases, would need to have been available or invented around the same time.” ↩

    3. 1c.

      The paper proposes that the proto-ribosome began as a parasite of an RNA-world host and that translation emerged from host-parasite mutual dependence.  The authors state it as testable and concede its reach.  “In principle, however, the conditions under which the key steps underlying our hypothesis might proceed can be evaluated in the laboratory”.  “short of a time machine, we may never know with certainty how the pre-LUCA endowments of life became established.”  It is cited together with the problem statement, because the paper is misused when its problem is quoted without its proposed solution. ↩

  2. 2.

    Koonin, Eugene V.  “The origins of cellular life.”  Antonie van Leeuwenhoek 106(1):27–41.  2014.  Section Introduction:  Two sentences earlier, Koonin quotes Woese:  “the greatest of evolutionary problems”.  “all of biological evolution that occurred after the origin of cells … is routine compared to that momentous breakthrough.”  A theory article by Eugene Koonin of the US National Center for Biotechnology Information on how cells arose.  It sets a protocellular scenario, in which membranes enclosed RNA segments, against a non-cellular one, in which membranes evolved inside virus-like elements.  Koonin notes that Woese called the origin of cells the greatest of evolutionary problems, and he says the lack of observations and of intermediate stages gives the origin of cells the look of irreducible complexity.  The article cites him for that apologetic phrase and for the sentence after it that declines the conclusion, since the phrase usually circulates without that qualifier. ↩ ↩

    1. 2a.

      Section Introduction:  The sentence immediately following, in the same paragraph, says explicitly that this shape of the problem may reflect missing data rather than a missing mechanism.  He uses 'irreducible complexity' descriptively, of the shape of the problem, and not as a conclusion.  Quoting only the first sentence would misrepresent him, and that is the quote-mine this article warns against, and it is how the passage usually circulates in apologetics. ↩

  3. 3.

    Bokov, Konstantin, and Sergey V. Steinberg.  “A hierarchical model for evolution of 23S ribosomal RNA.”  Nature 457:977–980.  2009.  Section Abstract:  “Surprisingly, a detailed analysis of the structure of the modern ribosome reveals a concerted and modular scheme of its early evolution.”  Petrov and colleagues restate the ordering logic in their own words:  “Bokov and Steinberg (2) reasoned that the donating adenosine in an A-minor interaction is dependent on the accepting double helix, whereas the accepting helix is less dependent on the adenosine.  If the conformation and structure of the rRNA was maintained throughout its evolution, the donating adenosine is the more recent addition.”  Fox, 2010, gives the first limit, that the primacy of the PTC is an input to the model rather than something it shows:  “The core areas of the structure are again the PTC by assumption”.  Fox, 2010, gives the second limit, that the small subunit, which holds the decoding centre, is treated as late:  “Bokov and Steinberg argue that the acquisition of the GTPase center and L1 protuberance and even the addition of the 30S subunit are relatively late additions to the ribosome.”  The root is also contested in print by Caetano-Anollés, Gustavo, in Journal of Molecular Evolution 80 (2015): 162–165, in a paper titled:  “Ancestral Insertions and Expansions of rRNA do not Support an Origin of the Ribosome in Its Peptidyl Transferase Center”.  The paper argues:  “Unwarranted assumptions, dismissals of conflicting data, structural insertion ambiguities, and lack of phylogenetic information compromise the construction of an unequivocal insertion-based model of macromolecular accretion.”  That dispute is about which end the ribosome grew from, between two groups who both accept that it grew by accretion and that the history is recoverable.  A structural analysis of 23S ribosomal RNA, the main RNA of the large ribosomal subunit, which orders its parts by the direction of their contacts and proposes that the molecule was built up in stages.  The article cites it for the claim that this build-up can be tested, because the order lets the molecule be stripped back to a small core around the peptide-bond centre. ↩ ↩

  4. 4.

    Bose, Tanaya et al.  “Origin of life: protoribosome forms peptide bonds and links RNA and protein dominated worlds.”  Nucleic Acids Research 50(4):1815–1828.  2022.  Section Abstract:  “Herein, after constructing such pockets, we show using the 'fragment reaction' and its analyses by MALDI-TOF and LC–MS mass spectrometry techniques, that several protoribosome constructs are indeed capable of mediating peptide-bond formation.”  Section Discussion:  “This molecular machine, however, did not possess the ability to support substrate motion from one site of the symmetrical pocket to the other, as required in the contemporary mode of protein biosynthesis.”  The replication by three groups is reported in a review by Agmon, Ilana, titled:  “On the Re-Creation of Protoribosome Analogues in the Lab”.  published in International Journal of Molecular Sciences 25 (2024): 4960, which says:  “Recently, three research groups experimentally demonstrated that several distinct dimeric constructs of protoribosome analogues… possess the ability to spontaneously fold, dimerize, and catalyze the formation of peptide bonds and of short peptides.”  An experimental study from the laboratory of Ada Yonath, the ribosome crystallographer who shared the 2009 Nobel Prize in Chemistry.  The team built the proposed protoribosome, a small RNA pocket from the ribosome's active site, and used a 'fragment reaction' checked by mass spectrometry to show that several constructs make peptide bonds, but the construct does not translocate, so it cannot make a chain processively, and it is not reading anything.  The article cites it as the laboratory test of that idea, and for the limit the authors state themselves: the pocket cannot move the substrate from one site to the other. ↩ ↩

    1. 4a.

      Section Discussion:  “This molecular machine, however, did not possess the ability to support substrate motion from one site of the symmetrical pocket to the other, as required in the contemporary mode of protein biosynthesis.”  The limitation is the builders' own, stated in the Discussion.  A correction was issued on this paper, Nucleic Acids Research 50 (2022): 11993, DOI 10.1093/nar/gkac1097.  It adds a missing reference and two 'Adapted with permission' credits, one to a table and one to a supplementary figure, so it is an attribution fix and not a scientific retraction. ↩ ↩

  5. 5.

    Agmon, Ilana.  “On the Re-Creation of Protoribosome Analogues in the Lab.”  International Journal of Molecular Sciences 25:4960.  2024.  Section Abstract:  “Recently, three research groups experimentally demonstrated that several distinct dimeric constructs of protoribosome analogues, derived predicated on the approximate 2-fold rotational symmetry inherent in the PTC region, possess the ability to spontaneously fold, dimerize, and catalyze the formation of peptide bonds and of short peptides.”  A review by Ilana Agmon of the Technion, who is also a co-author of the 2022 Yonath study.  It reports that three separate research groups built small protoribosome analogues, made from the two-fold symmetric part of the active site, that fold, pair up and make peptide bonds, and it suggests that a credible model should be a heterodimer with two different halves.  The article cites it as the report that the protoribosome has been rebuilt by three groups, not by one laboratory alone. ↩ ↩

  6. 6.

    Caetano-Anollés, Gustavo.  “Ancestral Insertions and Expansions of rRNA do not Support an Origin of the Ribosome in Its Peptidyl Transferase Center.”  Journal of Molecular Evolution 80:162–165.  2015.  Section Abstract:  “Analysis of a crucial three-way junction connecting the long-helical coaxial branch that supports the PTC to the L1 stalk and its translocation functions reveals an incorrect branch-to-trunk insertion assignment that is in conflict with…”  “Unwarranted assumptions, dismissals of conflicting data, structural insertion ambiguities, and lack of phylogenetic information compromise the construction of an unequivocal insertion-based model.”  A short letter to the editor by Gustavo Caetano-Anollés, an evolutionary bioinformatician at the University of Illinois, who re-analyses the rRNA insertion data behind the accretion model of Petrov et al. (2014).  He argues that one insertion is assigned to the wrong branch-to-trunk direction, and that AES39 is the older trunk rather than AES1, so the origin of the ribosome in the peptidyl transferase centre is not supported; his verdict on the model is quoted below.  The article cites it as a dispute between researchers who both accept that the ribosome grew by accretion and that its history is recoverable, about which end it grew from, not as a challenge to the enterprise. ↩ ↩

  7. 7.

    Xiao, Hong et al.  “Structural basis of specific tRNA aminoacylation by a small in vitro selected ribozyme.”  Nature 454:358–361.  2008.  Section Abstract:  “Flexizyme is a 45-nucleotide ribozyme capable of charging tRNA in trans with various activated l-phenylalanine derivatives.  In addition to a more than 10(5) rate enhancement and more than 10(4)-fold discrimination against some non-cognate amino acids, this ribozyme achieves good regioselectivity:  of all the hydroxyl groups of a tRNA, it exclusively aminoacylates the terminal 3′-OH.”  The flexizyme system was introduced in Murakami, Hiroshi, Atsushi Ohta, Hikaru Ashigai, and Hiroaki Suga, A highly flexible tRNA acylation method for non-natural polypeptide synthesis, Nature Methods 3 (2006): 357–359, and its precedent is Illangasekare, Mali, Gregory Sanchez, Timothy Nickles, and Michael Yarus, Aminoacyl-RNA synthesis catalyzed by an RNA, Science 267 (1995): 643–647.  The state of the art, as summarised by Miyachi and Ichihashi 2026:  “although protein synthesis using ribozyme-based aminoacylation is still a challenge (Chen et al., 2021).”  A structural study by Hiroaki Suga, Adrian Ferré-D'Amaré and their co-authors of how a small ribozyme (an RNA enzyme), selected in the laboratory, charges tRNA with an activated amino acid.  The flexizyme is promiscuous by design: it charges almost any tRNA with almost any activated acyl donor, which is the opposite of what a genetic code requires, and its specificity is for the leaving group on the activated amino acid rather than for pairing a side chain with an anticodon.  The article cites it as the clean case of ribozymes charging tRNA, while noting that a charging catalyst is not a coding catalyst. ↩ ↩

  8. 8.

    Petrov, Anton S. et al.  “History of the ribosome and the origin of translation.”  Proceedings of the National Academy of Sciences 112(50):15396–15401.  2015.  The restatement of the Bokov and Steinberg logic reads:  “Bokov and Steinberg (2) reasoned that the donating adenosine in an A-minor interaction is dependent on the accepting double helix, whereas the accepting helix is less dependent on the adenosine.  If the conformation and structure of the rRNA was maintained throughout its evolution, the donating adenosine is the more recent addition.”  A research article from Georgia Institute of Technology, led by Anton Petrov and Loren Dean Williams, that compares ribosome structures from across the tree of life.  It proposes an 'accretion model' in which the ribosome grew by recursively adding expansion segments of RNA, and it places the evolution of prokaryotic ribosomes in six phases.  The article uses it to explain how the direction of A-minor contacts dates each part of the molecule. ↩

  9. 9.

    Radakovic, Aleksandar et al.  “Autocatalytic assembly of a chimeric aminoacyl-RNA synthetase ribozyme.”  Science Advances 11(14).  2025.  Section Abstract:  “Here we report an autocatalytic assembly pathway that generates a chimeric, amino acid–bridged aminoacyl-RNA synthetase ribozyme.  The noncovalent complex of ribozyme fragments initiates low-level aminoacylation of one of the fragments, which, after loop-closing ligation, generates a highly active covalently linked chimeric ribozyme.  The generation of this ribozyme is increasingly efficient over time due to the autocatalytic assembly cycle that sustains the ribozyme over indefinite cycles of serial dilution.  Because of its trans activity, this ribozyme also assembles ribozymes distinct from itself, such as the hammerhead.”  Note what it still does not do: it aminoacylates, which is the step flexizymes already solve, and nothing in it assigns a particular amino acid to a particular anticodon.  A laboratory study from Jack Szostak's group at the University of Chicago, which builds an RNA enzyme from pieces and follows it through repeated rounds of serial dilution.  The assembly is autocatalytic: each copy of the ribozyme charges more of its pieces, which ligate into more copies, and the product can also assemble a different ribozyme, the hammerhead.  The article cites it as a 2025 result that goes further toward RNA doing the charging, and it concedes that point while noting what the work does not show. ↩

  10. 10.

    Li, Li et al.  “Aminoacylating Urzymes Challenge the RNA World Hypothesis.”  Journal of Biological Chemistry 288(37):26856–26863.  2013.  The earlier citation gave this title, which the paper does not use:  “Histidyl-tRNA synthetase urzymes:  Class I and II aminoacyl-tRNA synthetase urzymes have comparable catalytic activities for cognate amino acid activation.”  Its background line states the motive:  “RNA World scenarios require high initial fidelity, greatly slowing lift-off.”  The paper's conclusion, which matters for how far the result reaches:  “Urzymes appear highly evolved, implying that they had even simpler ancestors.”  A research paper from the laboratory of Charles Carter at the University of North Carolina, written with Christopher Francklyn of the University of Vermont.  It cut the two classes of aminoacyl-tRNA synthetase down to 120 to 130 residue cores, which the authors call urzymes, and measured how fast each core charges its tRNA, finding about 10^6 times the uncatalysed peptide-synthesis rate and about 60% of modern enzymes' catalytic power.  The article cites it for the claim that these cores are already highly evolved, so simpler ancestors must have come before them, and that the RNA-world model struggles with the accuracy it needs to get started. ↩

  11. 11.

    Martinez-Rodriguez, Luis et al.  “Functional Class I and II Amino Acid-activating Enzymes Can Be Coded by Opposite Strands of the Same Gene.”  Journal of Biological Chemistry 290(32):19710–19725.  2015.  From the paper's summary block:  “Two distinct aminoacyl-tRNA synthetase classes may have descended from opposite strands of a single gene…  Both products from a designed gene encoding ATP-binding sites from Class I and II synthetases catalyze amino acid activation…  The unique information in a gene can be interpreted two alternative ways, both of which accelerate similar chemistry.”  A research paper from Carter's laboratory at the University of North Carolina, which tests whether the two synthetase classes could come from opposite strands of one gene.  The authors had a gene designed in Rosetta to encode the ATP-binding sites on opposite strands, and both peptide products accelerated amino acid activation by about 750,000 to 1,300,000 times the uncatalysed rate.  The article cites it for the claim that the two halves of the code may be one invention, and the gene is a designed construct rather than a natural one, so the result shows chemical possibility, not history. ↩

  12. 12.

    Carter, Charles W. Jr., and Peter R. Wills.  “Interdependence, Reflexivity, Fidelity, Impedance Matching, and the Evolution of Genetic Coding.”  Molecular Biology and Evolution 35(2):269–286.  2018.  A paper by Charles Carter of the University of North Carolina and Peter Wills of the University of Auckland, who argue that the two classes of tRNA-charging enzymes arose from opposite strands of one gene and that genetic coding emerged together with them.  They argue that translation began in a peptide-RNA world, which removes the RNA-world circularity by denying that an RNA world came first.  The article quotes their conclusion.  It is a minority position inside the mainstream rather than a creationist one, and it relocates the question of where the first specific catalysts came from rather than removing it. ↩

  13. 13.

    Goldman, Aaron D. et al.  “Universal paralogs provide a window into evolution before the last universal common ancestor.”  Cell Genomics 6:101140.  2026.  A Perspective article that reports two earlier reconstructions of ancient aminoacyl-tRNA synthetases, the enzymes that attach each amino acid to its tRNA, from before the last universal common ancestor, and reports no new results of its own.  One reconstructed ancestor was shown to contain sites specific for leucine, isoleucine and valine, and another was devoid of tryptophan, which the authors take to suggest later neofunctionalization.  The article cites it to show that a simpler, sloppier code came before the modern one, which leaves its question about the physical process unchanged.  Its two primary studies are its references 75 and 70. ↩