When the Enzyme Is Its Own Blueprint: Protein-Templated DNA Synthesis in DRT3

Protein as blueprint: Drt3b synthesizes sequence-specific poly(AC) DNA using active site residues as a chemical template, with no nucleic acid required.
Cryo-EM was the decisive experiment: The 2.6 Å structure of the 6:6:6 complex was the only way to demonstrate the absence of a nucleic acid template in the Drt3b active site.
DRTs are a deep reservoir of novel biochemistry: DRT2, DRT4, DRT7, and DRT9 each use distinct DNA synthesis strategies, suggesting many more mechanisms remain uncharacterized.
Biotechnological potential: Engineering Drt3b to accept new sequence specificities could open routes to template-free synthesis of defined DNA sequences for biomaterials and beyond.

Protein-templated synthesis of dinucleotide repeat DNA by an antiphage reverse transcriptase: Deng et al., Science, 2026
Diverse enzymatic activities mediate antiviral immunity in prokaryotes: Gao et al., Science, 2020
De novo gene synthesis by an antiviral reverse transcriptase: Tang et al., Science, 2024
Protein-primed homopolymer synthesis by an antiviral reverse transcriptase: Tang et al., Nature, 2025
Bacterial reverse transcriptase synthesizes long poly(A)-rich cDNA for antiphage defense: Song et al., Science, 2025

A bacterial defense system has revealed that a protein’s active site geometry alone can direct sequence-specific DNA synthesis, extending the known boundaries of biological information flow

The central dogma of molecular biology defines a constrained information flow: DNA into RNA into protein, with reverse transcriptases permitting RNA to be read back into DNA. Even that exception requires a nucleic acid template. The assumption that sequence-specific DNA synthesis always depends on a nucleic acid blueprint has stood largely unchallenged for five decades. A paper by Deng et al., published in Science in April 2026, changes that, and the mechanism at its heart was made visible only through high-resolution cryo-electron microscopy.

The story originates in bacterial immunity. Defense-associated reverse transcriptases (DRTs) are a broad class of anti-phage systems first systematically identified by Gao et al. in 2020. In contrast to CRISPR, which targets foreign DNA for cleavage, DRTs respond to phage attack by synthesizing it. DRT2 systems produce coding genes de novo through rolling-circle reverse transcription of a noncoding RNA, generating an effector protein that aborts infection. Two independent groups found that DRT9 synthesizes poly(dA) DNA homopolymers to disrupt viral replication, with a parallel study showing the same system responds to phage-driven dATP accumulation. The DRT family was already emerging as a collection of biochemical surprises.

DRT3 is the most striking variant yet. Alex Gao’s group at Stanford cloned the DRT3 system from E. coli and found it contains two distinct reverse transcriptases, Drt3a and Drt3b, and a noncoding RNA scaffold. Drt3a synthesizes poly(GT) DNA conventionally, reading an ACACAC template encoded in the ncRNA. Drt3b does something entirely new: it generates the complementary poly(AC) strand in the complete absence of any nucleic acid template. Conserved active site residues mimic a template strand in chemical terms, enforcing strict alternating base incorporation through protein geometry alone. Cryo-EM structures at 2.6 Å resolution resolved the complete assembly as a D3-symmetric 6:6:6 ribonucleoprotein complex, with both enzymes engaged simultaneously in a single coordinated machine.

“These findings expand the functional landscape of nucleic acid polymerases, revealing a protein-templated mechanism for sequence-specific DNA synthesis.”

The structural data were decisive. Without the atomic-resolution cryo-EM model, it would have been impossible to demonstrate conclusively that no nucleic acid strand occupies the Drt3b active site: the protein is both enzyme and blueprint. Whether DRT3 formally violates the central dogma is debated; the poly(GT/AC) product is repetitive, non-coding, and does not appear to enter the bacterial genome. But it extends the known modes of biological information transfer in an unambiguous way. How the duplex DNA product confers anti-phage immunity remains open, with sequestration of viral replication factors and formation of unusual DNA secondary structures both proposed. More broadly, the finding reinforces a point that often needs making: in mechanistic enzymology, atomic structures are not confirmatory data. They are the experiment.