Scientists Discover a Bizarre New Way Life Can Make DNA... Without DNA

For life to reproduce information, the usual routes are relatively straightforward.

DNA is normally copied by using DNA as a template. Cells generally make RNA from a DNA template, and RNA then supplies the instructions needed to produce proteins.

A surprising route to making DNA

A bacterium has now exposed another route, however, and it is profoundly unexpected.

In one part of this newly identified system, a protein serves as a molecular template that builds a particular repeating DNA sequence. In essence, information encoded in amino acids determines information encoded in DNA.

That reverses the expected direction – rather like using a cake as the model for writing a recipe, instead of using a recipe to make a cake.

"Rather than copying one DNA strand to make the other, as happens in normal DNA replication, the system uses two different reverse transcriptases to independently generate the two complementary strands," biochemist Samuel Sternberg of Columbia University told ScienceAlert.

"One enzyme copies an RNA template, while the other uses amino acids in the protein itself to determine which DNA building blocks get added. The two resulting DNA strands then find each other and pair up to produce double-stranded DNA, which ultimately helps the bacterium defend itself against viruses."

The structure of DNA, showing how its two strands are joined by complementary base pairs: A with T, and C with G. (National Human Genome Research Institute)

DNA is, in effect, life’s system for storing information. Its well-known double-helix form comprises two strands built from four chemical bases: adenine (A), cytosine (C), guanine (G), and thymine (T). Together, these four components constitute the genetic material of every living thing.

They cannot be combined at random. Instead, they follow fixed pairing rules: A pairs with T, while C pairs with G. As a result, the sequence on either strand holds the information required to construct the other.

When a cell duplicates DNA, it separates the two strands like a zip. Each individual strand then becomes the template for the new strand that wraps around it. For instance, a strand with the sequence A-G-C-T will clearly need the sequence T-C-G-A.

In certain circumstances, enzymes known as reverse transcriptases instead produce DNA from an RNA template; retroviruses use this approach, for example. Either way, the process involves a nucleic acid.

How the DRT3 bacterial defence system works

This is where bacteria become somewhat unusual, as Sternberg and his colleagues discovered.

These organisms are known to employ a system called defence-associated reverse transcriptases (DRT) within their immune response to viruses, although precisely how the system functioned had not been clear.

The bacterium Escherichia coli, which uses an arsenal of molecular defences against invading viruses. (Science Photo Library/Getty Images)

Using a range of experiments, Sternberg and his colleagues investigated one such system, DRT3, in Escherichia coli to establish the roles of its two reverse transcriptases.

They found something unexpected when they studied the structure of one of these enzymes, DRT3b, with cryo-electron microscopy.

"We knew from our experiments that DRT3b could make this very specific repetitive DNA sequence without the RNA that the other reverse transcriptase requires. But we didn't know how it was doing that," Sternberg said.

"The real 'aha' moment came from biochemical experiments and the cryo-EM structure. Hiroshi Nishimasu and his team could actually see the DNA product sitting in the enzyme, and the structure put these two amino acids in exactly the right positions to explain the alternating sequence.

"That immediately suggested a pretty wild possibility: the protein itself was specifying the sequence."

The structure on its own did not demonstrate this conclusively. The researchers therefore changed the two amino acids they believed were responsible for specifying the sequence.

After doing so, both DNA synthesis and the bacterium’s antiviral defence failed, confirming that these amino acids were essential to the process.

Even more strikingly, DRT3b accomplished this without any nucleic acid template. Rather, two amino acids located beside the enzyme’s active site effectively functioned as the template: one favoured the addition of A and the other C.

The repeating DNA strand produced by DRT3b, as revealed by cryo-electron microscopy. The enzyme uses amino acids in the protein itself to determine the alternating C-A-C-A sequence. (Wang et al., Cell, 2026)

This produces an exceptionally simple DNA strand made up of the repeated sequence C-A-C-A-C-A.

"So in a very literal sense, information in the protein's amino-acid sequence is being translated into a DNA sequence," Sternberg said. "That's a very different flow of biological information from what we normally think about."

Yet this represents only one half of DRT3’s activity.

The DNA strand templated by DRT3b’s protein is one of two strands produced by the system. Independently, a second reverse transcriptase, DRT3a, makes another repeating DNA strand – though it does so conventionally, using RNA as its template.

The two strands are complementary. After they are made, they naturally locate one another and pair, creating the familiar double helix of double-stranded DNA.

This also runs counter to the usual process. Ordinarily, DNA’s two strands are not manufactured independently: an existing strand acts as the template for its new partner.

In this case, neither newly produced strand supplies the instructions for the other. One obtains its sequence information from RNA, the other from protein, and they join only afterwards.

"One additional aspect of the study that I find especially fascinating is that it challenges the usual picture of how double-stranded DNA gets made," Sternberg said.

"Normally, we think of one strand as carrying the information needed to make the other. Here, the two strands are synthesized independently, using two completely different kinds of templates – RNA for one and protein for the other – and only afterward come together to form a double helix.

"To me, that's a pretty remarkable example of how inventive evolution can be with very familiar molecular building blocks."

DRT3 turns viral defences against viruses

The DRT3 defence system in action. (Wang et al., Cell, 2026)

DRT3 does not carry out this elaborate molecular manoeuvre merely for show. It is a defence system, and the unusual DNA seems to operate as a type of booby trap for incoming viruses.

In normal conditions, an enzyme called RecBCD continuously breaks down DNA made by DRT3, stopping it from accumulating. RecBCD is itself part of the bacterium’s defences, helping to destroy invading viral DNA. As part of their attack machinery, some viruses make proteins that disable RecBCD.

"The virus is trying to disable RecBCD because RecBCD is itself an important antiviral defense," Sternberg explained. "But by disabling it, the virus inadvertently removes the brake on DRT3."

Related: Your DNA Isn't Just a Double Helix. Scientists Just Found What Else It's Hiding

As DNA produced by DRT3 accumulates, the bacterium ceases growing. This, in turn, prevents the virus from replicating efficiently. The bacterial cell is sacrificed so that the virus cannot spread through the surrounding population.

More research is required to establish how DRT3 prevents bacterial growth. The researchers also want to examine whether comparable systems can write other DNA sequences.

"My suspicion is that this is the tip of the iceberg," Sternberg said.

"I doubt that DRT3b is the only enzyme capable of this kind of unconventional DNA synthesis. Whether there are many other proteins that use exactly this amino-acid-templating mechanism remains to be seen, but I'd be surprised if nature only invented it once.

"More broadly, I think there is a huge amount of unexplored enzymology hiding in bacterial immune systems."

The findings were published in Cell.

This article was fact-checked by Fiona MacDonald and edited by Fiona MacDonald. While we pride ourselves on our process, we are only human. If you spot a mistake, please let us know.