Science: bacteria make DNA without template
- Stanford researchers reported in *Science* that a bacterial antiphage system called DRT3 can build one DNA strand without reading DNA or RNA. - The surprise sits in Drt3b, a reverse transcriptase that makes a precise poly(AC) strand while its partner Drt3a copies an ACACAC RNA motif. - If this holds up broadly, DNA synthesis is more flexible than textbooks suggest — and bacterial immunity just got a lot weirder.
DNA polymerases are supposed to follow a script. They copy DNA from DNA, or they copy DNA from RNA, or — in the weird edge cases — they add a few loose nucleotides without much precision. This new bacterial result matters because it points to a fourth mode: an enzyme that writes a specific DNA pattern without reading a nucleic-acid template at all. The team, led at Stanford and published in *Science* in April 2026, says part of a bacterial antiviral system does exactly that. ### What actually is the thing here? The system is called DRT3. It shows up in bacteria as an antiphage defense — basically a molecular booby trap against viruses that infect bacteria. DRT3 has three parts: one reverse transcriptase called Drt3a, a second one called Drt3b, and a noncoding RNA. Together they make a double-stranded DNA product with a repeating GT/AC pattern. (science.org) ### Why is that surprising? Because one half of the job looks normal enough, and the other half really doesn’t. Drt3a follows the usual logic. It reads a conserved ACACAC sequence embedded in the system’s RNA and uses that as a template to make the poly(GT) strand. That part still fits the standard base-pairing story. ### So what does Drt3b do instead? Drt3b makes the complementary poly(AC) strand, but the claim is that it does this in the complete absence of a nucleic-acid template. (science.org) The enzyme is “protein-primed,” and the shape and chemistry of its own active site appear to force the alternating pattern. In plain English — the protein itself acts like the guide. Not by spelling out a long arbitrary genome, but by enforcing a very specific repeating sequence. ### Is this totally unprecedented? Not exactly. Biology already had some oddball template-independent polymerases. But those usually make simple homopolymers, near-random tails, or very short motifs. The reason this paper got attention is that Drt3b seems to land in a new middle ground — still not making arbitrary long code, but making a defined, sequence-specific DNA product through a protein-templated mechanism. That is the conceptual jump. (science.org) ### How did they figure that out? The paper combines genetics, biochemical assays, sequencing, and cryo-EM structures. The structure showed a 6:6:6 complex of Drt3a, Drt3b, and the ncRNA at 2.6 Å resolution. The authors say that in Drt3b, the usual channel where a template strand would sit is blocked, while conserved residues in the active site help dictate which nucleotide gets added next. That structural piece is doing a lot of the work here. (science.org) ### What is the bacteria using this for? Defense. The team put DRT3 into *E. coli* and saw protection against phages. The system was not just constantly on — it was triggered by a specific phage protein called ST61. Once activated, it generated those long repetitive DNA products as part of the antiviral response. The exact downstream kill mechanism is still the murkier part for a general reader, but the broad role is bacterial immunity. (science.org) ### Does this break the central dogma? Not really, or at least not in the internet-hype way. The central dogma is about the flow of sequence information from nucleic acid to nucleic acid to protein. This result does not show proteins generally writing arbitrary genomes. It shows one bacterial enzyme using protein structure to enforce a specific repeating DNA sequence in a defense context. That is still a big deal — just a narrower and more precise one than “biology was wrong.” This is partly an inference from what the paper actually demonstrates and from how the mechanism is bounded. (phys.org) ### Why does anyone outside microbiology care? Because enzymes that make nucleic acids are some of the most foundational machines in biology. If one branch of them can encode sequence choice through protein architecture, that expands what people will look for in evolution, bacterial immunity, and maybe biotechnology. The bottom line is simple: bacteria seem to have been hiding a stranger way to make DNA than most people thought existed. (science.org)
