Published April 2026 | Biology | Molecular Science
For as long as biology has been a serious science, one rule has held firm: to make DNA, you need a template. You need something to read from. An existing strand of DNA or RNA that a protein enzyme can follow, base by base, like a typist copying a manuscript.
That rule just got its first real exception.
A Stanford team discovered a bacterial enzyme that can build DNA without reading any DNA or RNA template.
Instead, the enzyme’s own protein structure acts as the blueprint.
The study was published in the journal Science in April 2026, and the scientific community has been buzzing about it since.
This is not a minor footnote. This is a genuinely new category of biological information transfer, one that nobody had seen before.
Let’s break it down.
First, a Quick Refresher on How DNA Is Normally Made
Every cell in your body carries DNA, and every time a cell divides, that DNA has to be copied. The way this happens has been understood for decades.
An enzyme called a polymerase reads one strand of the DNA double helix and builds a matching complementary strand alongside it. Adenine pairs with thymine. Guanine pairs with cytosine. The sequence of the new strand is entirely determined by the sequence of the old one. No template, no sequence. This is the fundamental logic of genetic copying.
There is one known exception to the DNA-as-template rule: reverse transcription. Certain viruses, including HIV, carry an enzyme called reverse transcriptase that can read an RNA strand and write a DNA copy of it. This surprised biologists when it was first discovered in the 1970s, because it seemed to run the usual information flow backwards. But even here, the principle holds: there is still a nucleic acid template being read. The enzyme is still following instructions written in a molecular language it can decipher through base pairing.
Key context
Even reverse transcription still depends on a template. The new discovery goes beyond that boundary.
The new discovery blows past even that boundary.
Meet DRT3, a Bacterial Immune System With a Strange Trick
Bacteria have been at war with viruses, called bacteriophages or just phages, for billions of years. To survive, bacteria have evolved a remarkable variety of defense systems. CRISPR, the gene-editing technology that has transformed medicine, is one of them. It was discovered as a bacterial immune memory system before scientists figured out how to repurpose it.
DRT3 is another bacterial defense system, part of a family called defense-associated reverse transcriptases (DRTs). Researchers have been studying these systems for a few years now, finding increasingly strange biochemistry inside them. DRT3 turned out to be the strangest yet.
The DRT3 system has three working parts: two enzymes called Drt3a and Drt3b, and a small piece of non-coding RNA. Together, they produce a long, repetitive strand of double-stranded DNA, a sequence that alternates between GT and AC units, over and over, for kilobases of length.
The Stanford team, led by biochemist Alex Gao, used cryo-electron microscopy to visualize this machine at near-atomic resolution. What they found in Drt3b stopped them cold.
Why this is already interesting
This is not ordinary DNA replication. It is part of a bacterial antiviral defense system, the same kind of biological arms race that gave us CRISPR.
The Part That Should Not Be Possible
Drt3a, the first enzyme, behaves normally. It reads a short sequence within the non-coding RNA, a repeating ACACAC motif, and uses it as a template to build a poly-GT DNA strand. Classic reverse transcription. Nothing new there.
Drt3b is different.
Drt3b builds the complementary poly-AC strand, but when the researchers looked at its active site under the microscope, there was no RNA there. No DNA either. The channel where a template strand would normally sit was physically blocked by the enzyme’s own structure.
The central shock
Drt3b was building sequence-specific DNA with no nucleic acid template at all.
Instead of reading an external strand, Drt3b uses specific amino acid residues in its active site as the instructions. Two residues do most of the work: one called Glu26, which specifically stabilizes the incorporation of adenine, and another called Arg253, which enforces cytosine addition at alternating positions. The protein’s own shape, the arrangement of its side chains and the geometry of its active site, dictates exactly which nucleotide gets added at each step.
When the researchers mutated these two residues, the sequence fidelity collapsed and the bacteria lost their ability to fight off phage infections. This was not a biochemical curiosity. The protein-templated synthesis was real, precise, and biologically essential.
“The protein itself serves as the blueprint for the DNA sequence. That was quite a surprise.”
Gao called it “a fundamentally new way that life produces DNA.”
What This Does and Does Not Mean for the Central Dogma
At this point, it is worth being precise, because this discovery has already been mischaracterized in some corners of the internet.
The central dogma of molecular biology, as Francis Crick originally framed it, describes the flow of sequence information: DNA to RNA to protein. It does not say that proteins cannot have unusual enzymatic activities. And crucially, the central dogma has already accommodated reverse transcription for decades.
What DRT3 adds is a new arrow: protein to DNA. But with an important caveat that Gao himself is careful to state: Drt3b only produces one specific repetitive sequence. It does not represent a general mechanism by which proteins can write arbitrary genetic code. Other biologists have also pointed out that the DNA produced by DRT3 does not get incorporated into the bacterium’s genome. It is not rewriting the cell’s hereditary information.
Important clarification
This does not mean proteins generally write genes back into the genome. It means one protein machine can dictate one fixed DNA pattern for a specific biological purpose.
That said, the conceptual shift is real. A protein can now be understood as a direct template for DNA synthesis in a way that has never been documented before at this length, this precision, and with this mechanistic clarity.
Why Scientists Are Genuinely Excited
There are a few reasons this discovery has generated real excitement, beyond the headline.
1. It expands what we think polymerases can do
For a long time, polymerases were thought to be fundamentally dependent on nucleic acid templates. The discovery of telomerase showed some flexibility, and enzymes like CCA-adding enzyme can add a few nucleotides without reading a strand. But Drt3b synthesizes kilobases of sequence-specific DNA using only protein logic. That is a qualitative leap.
2. It raises questions about the origins of life
The finding hints that early biological systems may have used protein-like structures or scaffolds to guide the synthesis of nucleic acids before modern template-based replication evolved. Whether this mechanism is ancient or more recent remains open, but it gives researchers a new hypothesis to explore.
3. The biotech potential is real, even if distant
CRISPR became revolutionary because scientists learned to reprogram a bacterial immune system. DRT3 raises a similar possibility: if Drt3b can be engineered so that different active-site arrangements produce different DNA sequences, it could become a tool for template-free, programmable DNA synthesis. Gao’s team has suggested applications including custom DNA hydrogels and novel biomaterials.
Why this feels big
It is not just a weird enzyme. It is a proof that biology can encode DNA-building instructions in protein geometry itself.
The Bigger Picture
Biology keeps surprising us. Every few years, something turns up in a bacterium that rewrites a chapter of the textbook, usually in a defense system, usually in the molecular machinery of an arms race that has been running for billions of years.
CRISPR came from bacterial immunity. The mechanisms of RNA interference were first understood in non-human systems. And now DRT3, hiding in ordinary Escherichia coli, has shown us that the rule we thought was absolute, that DNA synthesis always needs a nucleic acid template, is not absolute after all.
The universe of possible biochemistry is larger than we thought. And somewhere inside that extra space, there may be tools we have not yet imagined.
The takeaway
One bacterial enzyme has shown that DNA can be built not by reading nucleic acids, but by following the logic embedded in a protein’s own structure.
Primary source: Deng et al., “Protein-templated synthesis of dinucleotide repeat DNA by an antiphage reverse transcriptase,” Science, April 2026. DOI: 10.1126/science.aed1656
0 Comments