Published online 16 March 2005 | Nature | doi:10.1038/news050314-8


DNA gets a fake fifth base

Artificial sequences could one day answer questions about evolution.

DNA is changing.DNA is changing.© SPL

One of the first things any biology student learns is that DNA, the recipe for life, is written with four letters. But what if you could add extra ones? Researchers who have managed to build, and replicate, DNA with an ersatz fifth letter are on their way to finding out.

Getting the modified DNA to work requires the team to answer all sorts of basic biochemistry questions. But the ultimate hope is that a few of these artificial letters could be sprinkled into the genome of a living microbe, to track its adaptation and evolution.

The four letters that occur naturally in DNA are chemical bases called adenine (A), guanine (G), cytosine (C) and thymine (T). The sequences that code for the different proteins to be built within a cell are spelled out using these four letters.

Floyd Romesberg and his colleagues at The Scripps Research Institute in La Jolla, California, have developed a fifth base, called 3-fluorobenzene, or 3FB. The bases in a zipper of DNA pair up across the molecule's two strands: A always pairs with T, and G always pairs with C. The 3FB pairs with itself, forming a completely new base pair.

Oily solution

The base had to be designed so that it would fit into the DNA strand without disturbing its structure. But a tougher challenge was to find something that DNA polymerase would recognize. This is the enzyme that replicates DNA by zipping along one strand while assembling its opposite, and the team needed it to be able to incorporate the new base.

Romesberg and his team initially designed molecules that had large flat surfaces so they would pack well in the DNA strand, a bit like pushing a pile of poker chips into a neat stack. But the large flat areas overlapped each other as they bonded, distorting the base pair so the polymerase was unable to get past and extend the strand. "We overcompensated out of fear that they wouldn't be stable," he says.

The researchers then came up with 3FB. The base is hydrophobic, that is, it has oily properties, and this turned out to be enough to pair up two bases in a strand of DNA. They described their results on 14 March at a meeting of the American Chemical Society in San Diego.

As the DNA strand replicates, the new base gets picked up and matched against another 3FB without problems about 100 times less often than the standard base pairs. But this is still fairly good: it means just one mistake per 1,000 base pairs. "I was happy with that," Romesberg says.

Enzyme search

Romesberg and his team are now looking at the problem from the other end. They are trying to evolve polymerases that recognize the fake base pairs and work with them more efficiently.

So far the new polymerases are not only better at working with Romesberg's base pairs, they are better at matching the natural pairs too. "A lot of what we have evolved might be the ability to grab the DNA more tightly," he guesses.


So far all this work has been carried out using chemicals in tubes, but Romesberg's ultimate goal is to incorporate his base pairs into a living microbe, "to have increased genetic information in a genome that's evolvable". He thinks that we'll be looking at colonies of bacteria decked out with "unnatural DNA" in a decade or less.

He emphasizes that there's no need to worry about the modified microbes going out of control: they need a steady supply of the fake base to reproduce. "Unless they start making it themselves, which won't happen this millennium."