Tratopedia
Side N繁中
Settings

Text size

Language

Theme

High contrast

Version

v1.230.2

The release this page was built from. It is what the service worker caches under.

Reverse transcriptase · Central dogma · Research briefScience, 16 April 2026; record to 29 October 2026

Stanford, the Alex Gao lab · bacterial anti-phage defence

A Way of Making DNA in Which the Protein Is the Blueprint

A team led by Alex Gao at Stanford described a bacterial defence enzyme, Drt3b, that builds a DNA strand with no nucleic acid anywhere in its active site: two amino acid side chains do the work a template strand normally does. The structure is resolved to 2.6 ångström and the enzyme makes one thing, an alternating repeat. A separate group at Columbia and the University of Tokyo has since shown why a cell would want it, and the argument over what it means for the central dogma is still running.

  • 2 RTDRT3 pairs a conventional Drt3a with an unconventional Drt3b
  • 2.6 Åresolution of the cryo-EM structures of the Drt3a:Drt3b:ncRNA complex
  • 6:6:6stoichiometry of that complex: six copies of each component
  • 20bacterial phyla across which DRT3 relatives are scattered, at least

How Firmly Each Part Stands Two peer-reviewed papers, and one question still open.

How firmWhatWho says so
ConfirmedDRT3 is a ribonucleoprotein complex of Drt3a, Drt3b and a non-coding RNA that makes alternating poly(GT/AC) double-stranded DNA; cryo-EM resolves a D3-symmetric 6:6:6 assembly at 2.6 ångströmScience, peer-reviewed, read in full (PMID 41990131)
ConfirmedTwo conserved residues in Drt3b, Glu26 and Arg253, select the incoming base; a tyrosine at the protein’s own end, Tyr650, starts the strand. Change Glu26 and the product truncates and misincorporates; change Tyr650 and defence stopsScience, peer-reviewed, read in full
ConfirmedThe host complex RecBCD keeps the DNA product below a toxic level; the phage protein Gam, which disables RecBCD, lets it accumulate and the cell arrests. Losing any RecBCD subunit does the same thing as Gam doesCell, peer-reviewed, read in full (PMID 42520802)
ConfirmedA single phage gene decides whether the system fires: every one of five phage T1 mutants that escaped this DRT3 had broken the same gene, ST61Science, peer-reviewed, read in full
UnconfirmedWhat the DNA itself does once it has built up. A “molecular sponge” that soaks up phage proteins is the leading idea, and whether the repeat folds into left-handed Z-DNA, and whether that matters, is written down as unresolvedGao, interviewed, calls the sponge his leading hypothesis and says he is open to other models; the Cell paper sets out the Z-DNA question as one to resolve

Every Publication With a Date Fifteen dated entries, from April to October 2026

  1. 16 Apr 2026Science publishes Deng, Lee, Armijo, Wang and Gao on DRT3: the two reverse transcriptases, the non-coding RNA, the 2.6 ångström structures and the phage T1 trigger ST61. A news article in the same issue carries the finding to a wider readership.
  2. 21 Apr 2026ScienceAlert and the CRISPeR FRENZY blog publish on the same day. The first explains the mechanism; the second reads the paper against the headline and finds the enzyme assembles two letters and repeats them.
  3. 24 Apr 2026Popular Mechanics, syndicated by AOL, carries Nikolai Slavov of Northeastern University calling Drt3b a “stuttering” machine rather than a protein reading itself.
  4. 30 Apr 2026A funder of the work publishes its own account of it, and points past the enzyme to the microbial “dark matter” it was found in.
  5. 20 May 2026Synthetic and Systems Biotechnology publishes Yaojun Tong’s commentary placing DRT3 inside a class of defence reverse transcriptases that write unconventional DNA, and warning that calling it a rewriting of the central dogma would be too strong.
  6. 1 Jun 2026A broadcaster’s science desk interviews Gao, Philip Kranzusch of Harvard Medical School and Rafael Pinilla-Redondo of the University of Copenhagen. Kranzusch calls the research groundbreaking; Pinilla-Redondo says the dogma has not been broken.
  7. 10 Jun 2026Nature Structural & Molecular Biology reports DRT1, a single-gene relative that makes semi-random DNA and switches itself off by forming filaments — the same family, a different output.
  8. 16 Jun 2026Frontiers in Microbiology publishes Sarfaraz Niazi’s review, which states the limits plainly: structural control of sequence without programmability, and no validated application in a mammalian system.
  9. 2 Jul 2026Molecular Cell publishes a comment by Yang, Zhang and Yang on the April paper.
  10. 28 Jul 2026Cell publishes Wang and colleagues, at Columbia and the University of Tokyo: the product is held down by RecBCD, released by the phage protein Gam, and the cell then shuts itself down. A screen also finds the host factor GreA is needed.
  11. 6 Aug 2026Cell publishes Figiel and Nowotny’s preview of that paper, under a title saying reverse transcriptases challenge the central dogma again.
  12. Aug 2026Microbial Biotechnology runs an editorial surveying the anti-phage systems described lately, DRT3 among them, and notes that about 1.5% of bacterial genes are predicted to belong to such systems.
  13. 2 Sep 2026Cell Host & Microbe reports that a cousin system, DRT2, has been given a rewritten RNA template and made to produce a user-defined double-stranded DNA inside living cells.
  14. 1 Oct 2026Biosystems publishes a quantitative model of Drt3b’s base alternation, fitted to the published mutant data and offering predictions for mutants nobody has made yet.
  15. 12 Oct 2026Nature reports DRT10, which adds DNA repeats the way telomerase does, and places telomerase in one clade with the defence reverse transcriptases — a common origin for the chemistry.

Does It Touch the Central Dogma? Five published positions, and they do not agree.

PositionArgument
A challenge, and not the first
Figiel and Nowotny, in Cell
Their preview is titled for the claim that reverse transcriptases challenge the central dogma again, and calls the way the two enzymes build the double strand an unprecedented process.
An expansion, not a rewriting
Yaojun Tong
It would be tempting, he writes, to frame these studies as a rewriting of the central dogma, and that would be too strong: no direct flow from a protein’s sequence to a DNA sequence is demonstrated. What is shown is that proteins can decide how a DNA product is made.
No dogma broken
Rafael Pinilla-Redondo (Copenhagen)
“No, I would not say the central dogma has been broken,” he told an interviewer: what the study shows is a protein helping to build a short, repetitive sequence in a very specific context. The exciting part, he says, is that evolution found an unexpected way to build a DNA molecule.
A rigid, single-purpose machine
Nikolai Slavov (Northeastern)
Drt3b is a highly specialised structural constraint, he wrote in a public comment: a “stuttering” machine physically keyed to produce one simple repetitive sequence, not a protein reading itself to create a complex message.
The dogma was never the point
Anna Meldolesi
The central dogma is by now more a rhetorical device than a law, she argues, and the headline missed the detail that decides the question: all the enzyme does is assemble two letters and repeat them.
What the paper itself establishesThe product is one alternating repeat, and the authors write that the six bases nearest the end of the strand all touch amino acid side chains — which, they say, suggests Drt3b cannot easily be reprogrammed for another sequence without losing efficiency.

How Reverse Transcriptase Keeps Being Repurposed One enzyme fold, a family of different outputs.

  • Retroviruses: in HIV and its relatives the enzyme copies an RNA genome into DNA. The template is RNA rather than DNA, but it is still a nucleic acid — which is the comparison the rest of this list is measured against.
  • Bacterial defence: reverse transcriptases turn up as a rapidly expanding set of clades inside bacterial immunity, sitting alongside CRISPR-associated ones in the same defence islands. CRISPR itself began as a bacterial defence system before it became a laboratory tool.
  • Building a gene that is not in the genome: the DRT2 system reverse-transcribes a non-coding RNA round and round to assemble a toxic protein the chromosome does not encode. Given a rewritten template it has been made to produce a chosen double-stranded DNA in living cells.
  • Other outputs again: DRT1 makes semi-random single-stranded DNA and shuts itself off in filaments; DRT9 makes long runs of a single letter; DRT10 adds repeats the way telomerase does, and the phylogeny puts telomerase in one clade with the family.
  • Drt3b: it goes further than the rest of the class in one respect — the sequence is set by the protein and not by any nucleic acid. Work since suggests this is not confined to DRT3: a preprint describes a fused enzyme, DRT7, that also writes DNA on a protein template.

The Defence Question, Answered in Part A second group, and a trigger on each side.

Wang and colleagues at Columbia and the University of Tokyo, a separate group from the one that first described DRT3, report that the system is poised rather than idle. RecBCD — the host’s own exonuclease-helicase, which processes double-stranded DNA ends during recombination and chews up incoming phage DNA — degrades the DRT3 product continuously, keeping it below the level that would stop the cell. The phage protein Gam disables RecBCD; the product then builds up, and the cell arrests itself, which ends the infection with it. Knocking out any RecBCD subunit does the same thing as Gam. Two other explanations were tested and set aside: the system does not work by exhausting the cell’s nucleotides, and it still defends in cells that cannot mount the SOS response. A mutagenesis screen turned up one host factor the defence needs, the transcription factor GreA — which is interesting because the phage trigger found in the original paper, T1’s ST61, is predicted to fold like GreA does. Philip Kranzusch of Harvard Medical School, who was not involved, calls the research groundbreaking.

  • RecBCDdegrades the product until a phage protein stops it
  • Gamthe phage protein whose attack on RecBCD sets the defence off
  • ST61the phage T1 gene every escape mutant had broken

Not the Only One How much of bacterial immunity is still unopened.

Tong’s commentary reads the defence reverse transcriptases as a class of DNA-writing modules rather than as odd copying enzymes, and DRT3 as the member that adds sequence control to the set. An editorial surveying the field this August puts a number on how much is left: around 1.5% of a bacterium’s genes are predicted to belong to anti-phage systems — about thirty genes in an Escherichia coli genome — and most of those systems have never been described. The October report of DRT10 pushes the point the other way, towards shared ancestry: the enzymes that add repeats to a bacterial defence DNA and the telomerase that caps a human chromosome fall in one clade. Gao himself, speaking of where DRT3 came from, called it a pointer to a vast reservoir of uncharacterised biology within microbial dark matter.

  • 1.5%of a bacterium’s genes are predicted to be anti-phage systems
  • DRT10adds repeats as telomerase does, and shares a clade with it

What Has Not Been Shown The limits, as the published reviews state them.

  • Drt3b has not been made to produce any sequence other than its own alternating repeat. The review that states this calls it structural control of specificity without programmability.
  • No application has been validated in a mammalian system or model.
  • A researcher asked directly whether it could be engineered answered that we do not yet know whether it can be reprogrammed or engineered in a useful way.
  • Any medical use would first have to settle how long the enzyme persists, what accumulating DNA does inside a cell, whether it sets off the innate immune system, and whether it damages the genome.

What to Take Away A mechanism, a function, and an argument that is not settled.

The mechanism is settled; the meaning is not

That a protein’s own side chains can fix the sequence of a DNA strand is peer-reviewed, resolved by cryo-EM and repeated by a second laboratory. What it means for the central dogma is an argument among named scientists, with a Cell preview on one side and the commentary and the interviews on the other, and this page carries both rather than picking.

The defence answer came from elsewhere

The account of what the DNA is for — held down by RecBCD, released when a phage attacks RecBCD, and the cell then stopping itself — comes from a group at Columbia and the University of Tokyo, not from the laboratory that first described DRT3. What the DNA actually does once it accumulates is still the open part.

Early days for engineering

Drt3b still makes one repeat and nothing else, and the contrast within its own family is instructive: a cousin system whose specificity lives in an RNA template has been reprogrammed to make a chosen sequence, while this one’s specificity is built into the protein, which is exactly what makes it hard to move.

What could not be verified

Unconfirmed report — Adi Millman, of the Massachusetts Institute of Technology: that using a protein as the template for DNA synthesis is a significant conceptual shift away from the classical central dogma. The comment is attributed to her in a news article whose own page could not be opened, and no other publication reproduces it.
Unconfirmed report — Aude Bernheim, of the Institut Pasteur: that the prospect of many unannotated bacterial defence systems encoding exotic biochemical functions like this one is exciting. Attributed in the same news article, and not found anywhere that could be read in full.

Quick notes

  1. Reverse transcriptase · Central dogma · Science, 16 April 2026

    A Way of Making DNA in Which the Protein Is the Blueprint

    A bacterial defence enzyme builds DNA with its own amino acids standing in for a template, and makes one alternating repeat.

    DRT3 — protein-templated DNA1 / 15

  2. What happened

    The Finding in Four Numbers

    • 2 RTDRT3 pairs a conventional Drt3a with an unconventional Drt3b
    • 2.6 Åresolution of the cryo-EM structures of the Drt3a:Drt3b:ncRNA complex
    • 6:6:6stoichiometry of that complex: six copies of each component
    • 20bacterial phyla across which DRT3 relatives are scattered, at least

    DRT3 — protein-templated DNA2 / 15

  3. What happened · the mechanism

    A Conventional Drt3a, an Unconventional Drt3b

    • Drt3a copies a short motif in the system’s own non-coding RNA to make one strand of the repeat.
    • Drt3b makes the complementary strand with no nucleic acid in its active site at all. Two side chains, Glu26 and Arg253, pick each base in turn.
    • The two strands anneal, and the cell ends up with a double-stranded repeat whose sequence a protein decided.

    DRT3 — protein-templated DNA3 / 15

  4. What happened · the structure

    Peer-Reviewed, and Resolved by Cryo-EM

    How firmWhatWho says so
    ConfirmedDRT3 is a ribonucleoprotein complex of Drt3a, Drt3b and a non-coding RNA that makes alternating poly(GT/AC) double-stranded DNA; cryo-EM resolves a D3-symmetric 6:6:6 assembly at 2.6 ångströmScience, peer-reviewed, read in full (PMID 41990131)
    ConfirmedTwo conserved residues in Drt3b, Glu26 and Arg253, select the incoming base; a tyrosine at the protein’s own end, Tyr650, starts the strand. Change Glu26 and the product truncates and misincorporates; change Tyr650 and defence stopsScience, peer-reviewed, read in full

    DRT3 — protein-templated DNA4 / 15

  5. What others add · the defence

    The Defence Question, Answered in Part

    • RecBCD, the host’s own exonuclease-helicase, degrades the DNA product continuously and keeps it below a toxic level.
    • Gam, a phage protein, disables RecBCD. The product then builds up and the cell arrests itself, ending the infection with it.
    • ST61, a gene of phage T1, is what every escape mutant in the original paper had broken.

    DRT3 — protein-templated DNA5 / 15

  6. What others add · the defence

    Two Explanations Tested and Set Aside

    • It does not work by exhausting the cell’s nucleotides: profiling found no depletion, infected or not.
    • It is not the SOS damage response: cells that cannot mount one still defend themselves.
    • What the accumulated DNA actually does — sponge, signal, scaffold or poison — is still the open question, and whether the repeat folds into left-handed Z-DNA is written down as unresolved.

    DRT3 — protein-templated DNA6 / 15

  7. The argument

    Does It Touch the Central Dogma?

    PositionArgument
    A challenge, and not the first
    Figiel and Nowotny, in Cell
    Their preview is titled for the claim that reverse transcriptases challenge the central dogma again, and calls the way the two enzymes build the double strand an unprecedented process.
    An expansion, not a rewriting
    Yaojun Tong
    It would be tempting, he writes, to frame these studies as a rewriting of the central dogma, and that would be too strong: no direct flow from a protein’s sequence to a DNA sequence is demonstrated. What is shown is that proteins can decide how a DNA product is made.
    No dogma broken
    Rafael Pinilla-Redondo (Copenhagen)
    “No, I would not say the central dogma has been broken,” he told an interviewer: what the study shows is a protein helping to build a short, repetitive sequence in a very specific context. The exciting part, he says, is that evolution found an unexpected way to build a DNA molecule.

    DRT3 — protein-templated DNA7 / 15

  8. The argument

    And What the Paper Itself Says

    PositionArgument
    A rigid, single-purpose machine
    Nikolai Slavov (Northeastern)
    Drt3b is a highly specialised structural constraint, he wrote in a public comment: a “stuttering” machine physically keyed to produce one simple repetitive sequence, not a protein reading itself to create a complex message.
    The dogma was never the point
    Anna Meldolesi
    The central dogma is by now more a rhetorical device than a law, she argues, and the headline missed the detail that decides the question: all the enzyme does is assemble two letters and repeat them.
    What the paper itself establishesThe product is one alternating repeat, and the authors write that the six bases nearest the end of the strand all touch amino acid side chains — which, they say, suggests Drt3b cannot easily be reprogrammed for another sequence without losing efficiency.

    DRT3 — protein-templated DNA8 / 15

  9. What others add

    One Enzyme Fold, a Family of Outputs

    • Bacterial defence: reverse transcriptases turn up as a rapidly expanding set of clades inside bacterial immunity, sitting alongside CRISPR-associated ones in the same defence islands. CRISPR itself began as a bacterial defence system before it became a laboratory tool.
    • Building a gene that is not in the genome: the DRT2 system reverse-transcribes a non-coding RNA round and round to assemble a toxic protein the chromosome does not encode. Given a rewritten template it has been made to produce a chosen double-stranded DNA in living cells.
    • Other outputs again: DRT1 makes semi-random single-stranded DNA and shuts itself off in filaments; DRT9 makes long runs of a single letter; DRT10 adds repeats the way telomerase does, and the phylogeny puts telomerase in one clade with the family.
    • Drt3b: it goes further than the rest of the class in one respect — the sequence is set by the protein and not by any nucleic acid. Work since suggests this is not confined to DRT3: a preprint describes a fused enzyme, DRT7, that also writes DNA on a protein template.

    DRT3 — protein-templated DNA9 / 15

  10. What others add

    Not the Only One

    • 1.5%of a bacterium’s genes are predicted to be anti-phage systems
    • DRT10adds repeats as telomerase does, and shares a clade with it
    • 20bacterial phyla hold a relative of DRT3, at least

    DRT3 — protein-templated DNA10 / 15

  11. What others add · the limits

    What Has Not Been Shown

    • Drt3b has not been made to produce any sequence other than its own alternating repeat. The review that states this calls it structural control of specificity without programmability.
    • No application has been validated in a mammalian system or model.
    • A researcher asked directly whether it could be engineered answered that we do not yet know whether it can be reprogrammed or engineered in a useful way.
    • Any medical use would first have to settle how long the enzyme persists, what accumulating DNA does inside a cell, whether it sets off the innate immune system, and whether it damages the genome.

    DRT3 — protein-templated DNA11 / 15

  12. Timeline

    April to June 2026

    1. 16 Apr 2026Science publishes Deng, Lee, Armijo, Wang and Gao on DRT3: the two reverse transcriptases, the non-coding RNA, the 2.6 ångström structures and the phage T1 trigger ST61. A news article in the same issue carries the finding to a wider readership.
    2. 20 May 2026Synthetic and Systems Biotechnology publishes Yaojun Tong’s commentary placing DRT3 inside a class of defence reverse transcriptases that write unconventional DNA, and warning that calling it a rewriting of the central dogma would be too strong.
    3. 1 Jun 2026A broadcaster’s science desk interviews Gao, Philip Kranzusch of Harvard Medical School and Rafael Pinilla-Redondo of the University of Copenhagen. Kranzusch calls the research groundbreaking; Pinilla-Redondo says the dogma has not been broken.

    DRT3 — protein-templated DNA12 / 15

  13. Timeline

    July to October 2026

    1. 16 Jun 2026Frontiers in Microbiology publishes Sarfaraz Niazi’s review, which states the limits plainly: structural control of sequence without programmability, and no validated application in a mammalian system.
    2. 28 Jul 2026Cell publishes Wang and colleagues, at Columbia and the University of Tokyo: the product is held down by RecBCD, released by the phage protein Gam, and the cell then shuts itself down. A screen also finds the host factor GreA is needed.
    3. 2 Sep 2026Cell Host & Microbe reports that a cousin system, DRT2, has been given a rewritten RNA template and made to produce a user-defined double-stranded DNA inside living cells.
    4. 12 Oct 2026Nature reports DRT10, which adds DNA repeats the way telomerase does, and places telomerase in one clade with the defence reverse transcriptases — a common origin for the chemistry.

    DRT3 — protein-templated DNA13 / 15

  14. Conclusion

    What to Take Away

    The mechanism is settled; the meaning is not

    That a protein’s own side chains can fix the sequence of a DNA strand is peer-reviewed, resolved by cryo-EM and repeated by a second laboratory. What it means for the central dogma is an argument among named scientists, with a Cell preview on one side and the commentary and the interviews on the other, and this page carries both rather than picking.

    The defence answer came from elsewhere

    The account of what the DNA is for — held down by RecBCD, released when a phage attacks RecBCD, and the cell then stopping itself — comes from a group at Columbia and the University of Tokyo, not from the laboratory that first described DRT3. What the DNA actually does once it accumulates is still the open part.

    DRT3 — protein-templated DNA14 / 15

  15. Conclusion

    And What Would Have to Change

    Early days for engineering

    Drt3b still makes one repeat and nothing else, and the contrast within its own family is instructive: a cousin system whose specificity lives in an RNA template has been reprogrammed to make a chosen sequence, while this one’s specificity is built into the protein, which is exactly what makes it hard to move.

    DRT3 — protein-templated DNA15 / 15

  • Deng, Lee, Armijo, Wang and Gao, Science, 16 Apr 2026, DOI 10.1126/science.aed1656 (PMID 41990131)
  • Wang et al., Cell, 28 Jul 2026 (PMID 42520802)
  • Tong, Synthetic and Systems Biotechnology, 20 May 2026 (PMID 42212300)
  • Niazi, Frontiers in Microbiology, 16 Jun 2026 (PMID 42382356)
  • Yang, Zhang and Yang, Molecular Cell, 2 Jul 2026 (PMID 42392043)
  • Figiel and Nowotny, Cell, 6 Aug 2026 (PMID 42561911)
  • Tang et al., Nature, 12 Oct 2026 (PMID 42863825)
  • Xia et al., Cell Host & Microbe, 2 Sep 2026 (PMID 42685694)
  • Neville et al., Nature Structural & Molecular Biology, 10 Jun 2026 (PMID 42271041)
  • Brüssow, Microbial Biotechnology, Aug 2026 (PMID 42548023)
  • Huang, Liu and Huang, Biosystems, 1 Oct 2026 (PMID 42822822)
  • Nield, ScienceAlert, 21 Apr 2026
  • Meldolesi, CRISPeR FRENZY, 21 Apr 2026
  • Orf, Popular Mechanics via AOL, 24 Apr 2026
  • DW via An-Nahar, 1 Jun 2026
  • The Mathers Foundation, 30 Apr 2026

Reliable

Reliable, but with an interestWorth believing and reliableWeak on both, and used with careHonest, but thin

Worth believing

Each point is one source used, placed by two judgements on a six-step scale. Points carry no names: the chart shows the shape of the evidence, not a verdict on any publisher.

Versions

This document is rewritten when what it says has to change. Every version stays published at its own address.

  1. v0003 current
  2. v0002 superseded
  3. v0001 superseded

The current version is also at latest/.