Bacteria make DNA without templates

DNA polymerases are supposed to follow a script. They usually copy DNA from DNA, or make DNA from RNA, or else they add kind of sloppy untemplated stretches that are simple and repetitive. This new bacterial result is weird because one enzyme seems to do something in between — not copying a nucleic-acid template, but still making a specific sequence pattern. That matters because “how DNA gets made” is one of those rules biology treats as bedrock, and this finding adds a new category rather than just a footnote. ### What actually got discovered? A Stanford team studied a bacterial anti-virus system called DRT3 — short for a defense-associated reverse transcriptase system. It has two enzymes, Drt3a and Drt3b, plus a noncoding RNA. Together they make long double-stranded DNA with alternating GT and AC repeats. One strand is poly(GT). The other is poly(AC). The paper appeared in *Science* in April 2026. (science.org) ### Why is that surprising? Because only half of the job follows the normal playbook. Drt3a uses a conserved ACACAC sequence embedded in the system’s noncoding RNA as a template, then makes the matching poly(GT) strand. But Drt3b makes the complementary poly(AC) strand “in the complete absence of a nucleic acid template.” That is the part that breaks the usual rule. It is not just random nucleotide dumping — the sequence alternates in a controlled way. (science.org) ### So what is Drt3b using instead? Basically, the enzyme seems to use its own protein structure as the guide. The *Science* paper describes this as a protein-templated mechanism for sequence-specific DNA synthesis. Cryo-EM structures at 2.6 Å resolution showed a D3-symmetric 6:6:6 complex of Drt3a, Drt3b, and the ncRNA, and the authors argue that conserved residues in Drt3b enforce the A-C-A-C pattern. Think of it less like copying text from a page and more like a machine whose grooves only let parts snap together in one repeating order. (science.org) ### Is this totally out of nowhere? Not exactly. Bacterial defense-associated reverse transcriptases were already looking strange. Other systems had been shown to make random DNA, long poly(A) tracts, or bizarre repetitive products used in anti-phage defense. A 2025 *Nature Communications* paper on DRT4, for example, showed template-independent synthesis too — but there the products were random single-stranded DNA, not a specific sequence pattern. (science.org) The new DRT3 result is the sharper claim because it is both template-free and ordered. ### What is the bacteria using this for? To fight phages — viruses that infect bacteria. In DRT systems, the DNA product is usually not the point by itself. It is part of a defense response that gets triggered during infection and ends up poisoning the infection cycle, often by shutting down the host cell before the virus can spread. That is why these systems keep turning up as a source of biochemical oddities — evolution is willing to invent very strange chemistry if it helps bacteria stop phages. (nature.com) ### Does this rewrite the central dogma? No — but it does widen the catalog of what polymerases can do. DNA replication in cells is still overwhelmingly template-directed. The catch is that textbooks often present the alternatives as either template copying or template-independent but low-information addition. DRT3 suggests there is a third bucket where a protein’s shape can encode enough rules to generate a defined repeating sequence. (phys.org) That is a smaller claim than “all DNA can be made this way,” but it is still a big mechanistic shift. ### Why do origins-of-life and biotech people care? Because this hints that sequence information might be imposed by structures other than nucleic acids. That is interesting for origins-of-life models, where people ask how ordered polymers emerged before modern replication systems existed. It is also interesting for synthetic biology and enzymatic DNA manufacturing, where researchers want new ways to build DNA without traditional templates. (science.org) The immediate use case is not obvious yet — DRT3 makes a very specific repetitive product, not arbitrary designer DNA — but the concept is the real news. ### Bottom line? The cleanest way to say it is this: a bacterial virus-defense enzyme appears to make a specific DNA strand without reading DNA or RNA. If that mechanism generalizes, one of molecular biology’s oldest categories just got an extra branch. (science.org)