The Dawn of Cheap and Easy DNA Writing

These startups are developing new ways to reprogram biology by producing custom genes from scratch.

The 4,000-square-foot suite, tucked into a small office park in the heart of San Diego’s biotech corridor, is about as small and unassuming as a university biology-lab classroom. But the 16 people who work here at Molecular Assemblies are chasing a goal that could revolutionize synthetic biology: the ability to write DNA molecules using enzymes.

The field of synthetic biology has, for years, been promising the ability to custom-design organisms that serve as everything from new antibiotics to plastic eaters. But despite substantial advances in both sequencing and editing DNA, one of the big holdups in fulfilling this potential is that whipping up DNA to order is slow and—especially for long molecules—expensive. In truth, the way scientists write DNA hasn’t changed much since the process was first developed in 1981. It’s slow, laborious, and environmentally hazardous.

But several startup companies believe they can do it cheaper, faster, and more accurately with a new method. They believe that using enzymes, which is how DNA is written in nature, is the way to go.

Molecular Assemblies recently raised $12.2 million in an initial round of funding, though it’s far from alone in its quest to use enzymes for building DNA. At least seven startups are trying to do it. Researchers at the University of California at Berkeley, who have published the only paper so far to describe a successful enzymatic approach, founded Ansa Biotechnologies to commercialize it. Ansa, like Molecular Assemblies, is building its business model around providing DNA to customers who send in orders. But another contender, DNA Script, recently announced $38.5 million in a second round of funding for a benchtop machine that would enable labs and hospitals to build DNA themselves.

Today, commercially available DNA synthesis uses a process called phosphoramidite chemistry, which relies on toxic, flammable reagents that create hazardous waste. The process has important limitations—most companies that build DNA this way top off at lengths of about 100 to 150 base pairs, which isn’t even as long as many genes, and the harsh chemistry involved in adding each A, G, C, or T can start to degrade the already-written part as the molecule grows longer. This method also produces molecules that aren’t compatible with water-based biology. That’s acceptable if you want to use DNA as a form of data storage, but to be useful in biological applications, DNA made through phosphoramidite chemistry has to be put through additional processing, which increases the cost.

“It’s amazing that they get chemical synthesis of DNA to work,” says Andrew Hessel, president of Humane Genomics, which is developing new cancer therapies by reprogramming viruses, and co-founder of Genome Project-write. GP-write, as it’s known, an international effort to explore the prospects of redesigning human cells, just concluded its annual meeting in New York City. “The reality is that nature uses enzymes to write DNA, and that is an incredibly complex process. Every time a cell in your body divides, it has to write a whole human genome perfectly without any additional modifications.”

As synthetic biology advances, researchers want longer and longer segments of DNA—ideally, at least the length of genes. The longer the DNA molecule, the fewer the segments that scientists have to stitch together to make a desired sequence, which should reduce the cost and the chances for errors to be introduced.

Those involved in Hessel’s GP-write project have their sights set on writing full genomes, which would allow them to engineer human cells (and other organisms’ cells) so as to better understand health and disease. For example, some scientists involved in GP-write are exploring ways of making cells resistant to viruses. Others are investigating how cells could produce essential nutrients that people now have to derive from food. But making genome-length DNA—even bacterial genomes—using chemical synthesis is currently cost-prohibitive.  

William Efcavitch, chief science officer and co-founder of Molecular Assemblies, helped lead the development and commercialization of the original phosphoramidite method in the early 1980s, but now he says it’s clear a better approach is required. “Rather than trying to push 35-year-old chemistry to make longer strands, we said: Let’s start with an enzymatic process that can already make long strands and teach it to do it in a user-friendly fashion,” says Efcavitch.

“You have to control the enzyme and tell it what to write. And that’s tricky.”

Andrew Hessel

The challenge is that in their natural habitat within a cell, enzymes don’t create DNA from scratch. Instead, they duplicate a pre-existing strand by pulling nucleic acids, one by one, to the growing molecule. So Molecular Assemblies and most of the other companies have turned to the only enzyme known to build DNA without a template. This DNA-creating enzyme, or polymerase, is called terminal deoxynucleotidyl transferase (TdT). Typically found in vertebrate immune cells, it is responsible for building the new and ever-changing antigen receptors a cell needs to fight unfamiliar viruses and bacteria.

TdT evolved to make long strands of DNA in a random fashion, but the new breed of DNA-writing startups think they can program it. All of them, however, are still working to figure out exactly how. “The challenge with enzymatic synthesis from scratch is that you have to control the enzyme and tell it what to write,” Hessel says. “And that’s tricky.”

Chemical synthesis uses a computer to control a system that adds A, G, C, or T—one drop at a time—in a four-step process: The DNA molecule is extended by one nucleotide held in place with an unstable bond; then the incomplete end is capped off; then the newly linked nucleotide is stabilized; and then the molecule is prepared for the next addition. Enzymatic synthesis eliminates two of those steps: the polymerase just needs to be stopped and started for each additional nucleotide. “Right now,” Efcavitch says, “we’re trying to optimize those two steps.”

The enzymatic synthesis startups have shown modest success. Ansa has built short DNA fragments called oligonucleotides (or oligos) of 50 base pairs. DNA Script has hit 200, and another company—Camena Bioscience—recently announced it had reached 300. Molecular Assemblies won’t specify how long its oligos have gotten other than to say they haven’t yet reached 150.

The companies’ claims remain largely untested by the synthetic biology community.

“There have been almost no publications,” says Calin Plesa, a synthetic biology researcher at the University of Oregon. “It’s been very difficult to know what’s been going on inside these companies.”  

Plesa himself is a heavy user of synthetic DNA for building DNA libraries, as is Sri Kosuri, a synthetic biologist at the University of California, Los Angeles, and co-founder of a startup called Octant. Kosuri describes himself as a “synthetic DNA addict” whose lab consumes large amounts of oligonucleotides to explore the relationship between DNA sequences and their functions. He appreciates how the companies pursuing enzymatic DNA synthesis are trying to improve the accuracy of the technology. “Accuracy is an issue. It’s what limits even our own work,” Kosuri says. But he adds that it doesn’t yet appear that the DNA-writing startups have gotten the enzymatic process near the accuracy of phosphoramidite chemistry.

George Church, a geneticist at Harvard University who is a cofounder of both the Human Genome Project and GP-write, says chemical DNA synthesis methods generally induce an error every 1 in 300 bases. Error-correction methods can improve the figure to 1 in 10,000. When enzymes naturally copy a strand of DNA in cells, however, the error rate is close to one in a billion. But he agrees with Kosuri that no enyzmatic synthesis company has come even close to such low error rates. “Right now, there’s no evidence than enzymatics is more accurate [than chemical synthesis]. I think it’s likely but not proven,” Church says.

Today, the longest oligonucleotides being produced are coming out of South San Francisco-based Twist Bioscience, which has miniaturized the chemistry using a silicon chip with thousands of tiny wells, creating a platform that that can make one million oligos simultaneously. They are used for screening, diagnostics, therapeutics, and genetic research. Twist can now make oligos up to 300 base pairs long in these wells, more than twice what most enzymatic companies are capable of at the moment.

But Twist Bioscience CEO Emily Leproust says that if a better method of synthesizing DNA presents itself, Twist’s method can accommodate it. “We don’t really have a dog in the fight,” she says. “If there is better [synthesis through] enzyme chemistry, I’ll be the first customer.” Once the approach reaches one of any number of milestones—longer, fewer errors, or faster production—she’d be on board. “I’ll take cheaper but frankly I’ll pay more if it’s faster or better or longer.”

She’s confident that one or more of the companies pursing the enzymatic approach will hit the target eventually. “I don’t think they have to break any rule of physics to get there—I think it’s just engineering,” she says. “It’s a question of how much money do you need, and how much time do you need, and can you recoup that investment in commercialization.”

Church and Hessel both agree that enzymatic synthesis will start to gain traction soon. “I fully expect that bacterial-scale genomes will be within anyone’s reach within the next 10 years,” Hessel says. And that would be just the start. “I don’t think we’ve started to unlock the possibilities here. I can’t wait to see how these tools and technologies change the world.”

Go Deeper