Skip to main content
Premium Trial:

Request an Annual Quote

UC Berkeley Team Develops Enzymatic DNA Synthesis Method


NEW YORK (GenomeWeb) – Researchers at the University of California, Berkeley have developed an enzyme-based DNA synthesis method that promises to improve on the decades-old chemical synthesis approach for making oligonucleotides.

In a proof-of-concept study published in Nature Biotechnology this week, the team, led by Jay Keasling at the Department of Bioengineering at UC Berkeley and the Joint BioEnergy Institute, demonstrated that it could enzymatically synthesize 10-mer oligonucleotides.

With further optimization, the approach could potentially allow scientists to make 1,000-mer oligos in a single shot within a short timeframe. Two of the UC Berkeley researchers have filed a patent application on the technology and have founded a startup, Ansa Biotechnologies, that hopes to develop it commercially.

Dan Arlow, a doctoral student in the biophysics graduate program at UC Berkeley and one of the lead authors of the study, said he was originally interested in synthetic biology for engineering microbes to produce useful chemicals. However, he and his research partner and co-first author Sebastian Palluk, a doctoral student at TU Darmstadt in Germany, were getting frustrated by how long it took to get oligos manufactured and by the fact that some types of sequences just could not be made. "Both of us independently started thinking about, 'Are there better ways you could manufacture DNA?'" Arlow said.

Currently, he explained, DNA is manufactured chemically, using the nucleoside phosphoramidite method that was developed more than 35 years ago. "Judged by organic chemistry standards, it's essentially perfect," he said, but the technology, which involves three or four steps per nucleotide addition, has a number of drawbacks.

For example, because the yield per cycle is only around 99.5 percent, oligos are limited in length to about 200 to 300 nucleotides. Also, chemical side reactions can damage the DNA, and the reactions take place in organic solvents, requiring anhydrous conditions, which is cumbersome and means that most labs outsource oligo synthesis to commercial manufacturers.

Finally, longer DNA sequences need to be stitched together from shorter oligos, which doesn't work for all types of sequences, such as repetitive or self-complementary ones. "I have always found this very frustrating," Arlow said. "You have on your computer screen the sequence of the molecule you want, and you need it to be manufactured and shipped to you so you can do your experiment, but it's not always possible."

Making DNA enzymatically instead could increase the yield per cycle, resulting in longer DNA molecules. "The promise of enzymatics is that the reactions could be so clean that you could essentially get to almost 100 percent yield per step," he said.

Scientists have already known a suitable enzyme, which doesn't require a template to build DNA, for decades: terminal deoxynucleotidyl transferase (TdT), a vertebrate polymerase with a role in creating antibody diversity that adds single nucleotides to the 3'-end of single-stranded DNA.

In order to be harnessed for in vitro DNA synthesis, the enzyme would need to add a 3'-blocked nucleotide triphosphate, followed by an unblocking step, similar in principle to the sequencing-by-synthesis approach used by next-generation sequencing technologies from Illumina and others.

However, TdT doesn't work efficiently with 3'-blocked nucleotides, also known as reversible terminators, Arlow said. Others, including a company called Molecular Assemblies, have been trying to use nucleotides modified in other places than the 3'-end that still block DNA elongation, so-called virtual terminators, but so far, they have not reported any success, he said.

Arlow and Palluk came up with a different approach, tethering a single nucleotide to a TdT enzyme through a molecular linker. After the enzyme adds the nucleotide to a DNA molecule, it remains covalently attached to the DNA, thus preventing any further additions. After cleaving the enzyme off, the next nucleotide can be added.

The process leaves a tiny molecular "scar" on each base of the newly synthesized DNA, so it is not completely natural, but Arlow said that for many applications, where the DNA is then PCR-amplified with natural nucleotides, that is not a problem. Also, it might be possible to use alternative linker chemistries that either reduce or eliminate the scar.

In general, the approach differs from others because it treats the enzyme like a disposable reagent. "It's counterintuitive because you think 'Oh, you are throwing away the enzyme,'" Arlow said. "But we believe that the economics of this are actually plausible."

While the cost of an optimized process is difficult to predict, he said, for certain applications that only require a tiny amount of DNA to be made, the technology will almost certainly be cost effective. And a back-of-the-envelope calculation showed that it should even be feasible to make nanomole amounts of DNA, for example PCR primers.

For their proof-of-concept project, Arlow and his colleagues synthesized a 10-mer oligonucleotide, using coupling times between 1.5 and 3 minutes. They estimated the reaction yield for each step, which ranged from 93.4 percent to 99.5 percent, with an average of 97.7 percent.

The next step is to investigate the source of this variability in yield and to increase it. "Our dream is to make a DNA synthesizer that's so perfect it can directly synthesize 1,000-mers," Arlow said, which would require a stepwise yield of about 99.9 percent. "We have a ways to go, but we have a lot of ideas about how the process can be improved, and we are pretty confident that we will be able to get there," he said.

For example, each step in their experiment had about a 1 percent insertion error, where two nucleotides instead of one were incorporated. A possible explanation for these doublets is that some enzymes accidentally had two instead of one nucleotide attached to them. The researchers have tried to mitigate that by adjusting the conditions of the labeling reaction, Arlow said, and are thinking about changing the linker chemistry.

Another problem is that some DNA strands form secondary structures, such as hairpins, during the synthesis, which may prevent TdT from extending them further because it does not like blunt or recessed ends. These secondary structures could be disrupted by temporarily modifying the bases or by raising the temperature or using solvents and modifying the enzyme accordingly.

Finally, to develop the method commercially, the DNA would need to be attached to some kind of solid support, and the fluidics would need to be automated, Arlow said.

Twist Bioscience CEO Emily LeProust, who Arlow has already talked to informally, agreed that enzymatic DNA synthesis could be faster and create longer DNA strands than current chemical synthesis, potentially even at lower costs. However, the established phosphoramidite chemistry has been commercialized for decades, she said, so there are no supply chain issues, for example regarding the quality or quantity of the required reagents and materials. "For commercialization, the supply chain needs to be set up at scale and at cost," she said.

LeProust said that several companies are developing enzymatic DNA synthesis technology, though none she knows of have commercialized it yet. Twist, she said, would be very interested in adopting these technologies. "When an enzymatic approach works and is able to be scaled efficiently, we will be the first customer in line to apply it to our silicon platform that spatially controls the synthesis of DNA," she said, adding that Twist already has a robust infrastructure in place. "To change the chemistry component is relatively straightforward, as it would fit directly into our workflow."

Arlow is the CEO, and Palluk the CTO of newly-formed Ansa Biotechnologies, which is part of the QB3 incubator and located in the San Francisco Bay Area, but Arlow said it is too early to provide further information about the firm.