The Gene-Synthesis Revolution
Researchers can now design and mass-produce genetic material — a technique that helped build the mRNA vaccines. What could it give us next?,
The Tech and Design Issue
The Gene-Synthesis Revolution
Researchers can now design and mass-produce genetic material — a technique that helped build the mRNA vaccines. What could it give us next?
Credit…Illustration by Jaedoo Lee
By Yiren Lu
- Nov. 24, 2021, 5:00 a.m. ET
Ten years ago, when Emily Leproust was a director of research at the life-sciences giant Agilent, a pair of scientist-engineers in their 50s — Bill Banyai and Bill Peck — came to her with an idea for a company. The Bills, as they were later dubbed, were biotech veterans. Peck was a mechanical engineer by training with a specialty in fluid mechanics; Banyai was a semiconductor expert who had worked in genomics since the mid-2000s, facilitating the transition from old-school Sanger sequencing, which processes a single DNA fragment at a time, to next-generation sequencing, which works through millions of fragments simultaneously. When the chemistry was miniaturized and put on a silicon chip, reading DNA became fast, cheap and widespread. The Bills, who met when Banyai hired Peck to work on a genomics project, realized that there was an opportunity to do something analogous for writing DNA — to make the process of making synthetic genes more scalable and cost-effective.
At the time, DNA synthesis was a slow and difficult process. The reagents — those famous bases (A’s, T’s, C’s and G’s) that make up DNA — were pipetted onto a plastic plate with 96 pits, or wells, each of which held roughly 50 microliters, equivalent to one eyedropper drop of liquid. “In a 96-well plate, conceptually what you have to do is you put liquid in, you mix, you wait, maybe you apply some heat and then take the liquid out,” Leproust says. The Bills proposed to put this same process on a silicon chip that, with the same footprint as a 96-well plate, would be able to hold a million tiny wells, each with a volume of 10 picoliters, or less than one-millionth the size of a 50-microliter well.
Because the wells were so small, they couldn’t simply pipette liquids into them. Instead, they used what was essentially an inkjet printer to fill them, distributing A’s, T’s, C’s and G’s rather than pigmented inks. A catalyst called tetrazole was added to bind bases into a single-strand sequence of DNA; advanced optics made perfect alignment possible. The upshot was that instead of producing 96 pieces of DNA at the same time, they could now print millions.
The concept was simple, but, Leproust says, “the engineering was hard.” When you synthesize DNA, she explains, the yield, or success rate, goes down with every base added. A’s and T’s bond together more weakly than G’s and C’s, so DNA sequences with large numbers of consecutive A’s and T’s are often unstable. In general, the longer your strand of DNA, the greater the likelihood of errors. Twist Bioscience, the company that Leproust and the Bills founded, currently synthesizes the longest DNA snippets in the industry, up to 300 base pairs. Called oligos, they can then be joined together to form genes.
Today Twist charges nine cents a base pair for DNA, a nearly tenfold decrease from the industry standard a decade ago. As a customer, you can visit the Twist website, upload a spreadsheet with the DNA sequence that you want, select a quantity and pay for it with a credit card. After a few days, the DNA is delivered to your laboratory door. At that point, you can insert the synthetic DNA into cells and get them to begin making — hopefully — the target molecules that the DNA is coded to produce. These molecules eventually become the basis for new drugs, food flavorings, fake meat, next-gen fertilizers, industrial products for the petroleum industry. Twist is one of a number of companies selling synthetic genes, betting on a future filled with bioengineered products with DNA as their building blocks.
In a way, that future has arrived. Gene synthesis is behind two of the biggest “products” of the past year: the mRNA vaccines from Pfizer and Moderna. Almost as soon as the Chinese C.D.C. first released the genomic sequence of SARS-CoV-2 to public databases in January 2020, the two pharmaceutical companies were able to synthesize the DNA that corresponds to a particular antigen on the virus, called the spike protein. This meant that their vaccines — unlike traditional analogues, which teach the immune system to recognize a virus by introducing a weakened version of it — could deliver genetic instructions prompting the body to create just the spike protein, so it will be recognized and attacked during an actual viral infection.
As recently as 10 years ago, this would have been barely feasible. It would have been challenging for researchers to synthesize a DNA sequence long enough to encode the full spike protein. But technical advances in the last few years allowed the vaccine developers to synthesize much longer pieces of DNA and RNA at much lower cost, more rapidly. We had vaccine prototypes within weeks and shots in arms within the year.
Now companies and scientists look toward a post-Covid future when gene synthesis will be deployed to tackle a variety of other problems. If the first phase of the genomics revolution focused on reading genes through gene sequencing, the second phase is about writing genes. Crispr, the gene-editing technology whose inventors won a Nobel Prize last year, has received far more attention, but the rise of gene synthesis promises to be an equally powerful development. Crispr is like editing an article, allowing us to make precise changes to the text at specific spots; gene synthesis is like writing the article from scratch.
Like many technologies in their infancy, gene synthesis (along with the field it has enabled, synthetic biology) has sparked a good deal of speculation and start-up activity. Most of the companies — excepting those working on the coronavirus — are in experimental phases; their applications have yet to return conclusive results. Still, the possibilities captivate both investors and scientists, whether they are fabricating microorganisms to produce industrial chemicals or engineering human cells to treat medical disorders. If even a small percentage of these efforts succeed, they could lead to trillion-dollar markets. The analogy frequently used by biotech venture capitalists is that we are in the Apple II days of synthetic biology, with the equivalent of iMacs and iPhones still to come. It’s a grandiose claim — but not implausible, especially now that Covid has battle-tested some of the underlying technologies. Personal computing created our digital lives; reading and writing DNA could mean control over our physical ones.
Among the aphorisms of synthetic biology is this: Nature is the best innovator. For example, CaS-9, the “cutting” enzyme used in Crispr, was originally a defense that bacteria evolved to fight off viruses. But the aphorism glides over the fact that for most of human history, nature has also been opaque, requiring that humanity stumble upon its inventions entirely by chance. Penicillin, quinine — many of our medicine-cabinet staples have been discovered from leaving food out for too long or by finding the active ingredients in herbal remedies. Only since the advent of modern chemistry have we been able to write down the sort of formulas that are common in physics and math.
Then came the genomics revolution. The first phase, marked by milestones like the sequencing of the human genome and by the emergence of companies like 23andMe, focused on reading genes. The second phase, just underway, is about writing genes. It is now possible to take our understanding of molecular biology — how DNA specifies the sequence of RNA, which in turn specifies the production of proteins — and use Crispr and DNA synthesis to devise genetic recipes that produce the outputs we want. So what does this look like in practice?
One of Twist’s biggest clients is Ginkgo Bioworks, a cell-engineering company that went public to much fanfare in September and by mid-November was valued at $25 billion. Ginkgo’s main offices occupy a converted warehouse in Boston’s seaport district. When I visited a few months ago, Patrick Boyle, a Ginkgo executive, walked me through their five “foundries” — so named after microchip fabrication plants. We passed one machine that uses microfluidics technology to mix reagents and cells and another that uses mass spectrometry to rapidly analyze the chemical composition of liquids.
For decades, the fundamental labor unit of biological research has been the lowly grad student, who toils away pipetting liquids, taking measurements, looking through results and, if lucky, maybe running a few experiments a month. Ginkgo, in contrast, has brought an assembly line’s efficiency to the lab, utilizing machines that can pipette, mix and assay with far more precision than any human ever could, therefore making it possible to run thousands of different experiments at the same time.
Ginkgo is a “platform” company — instead of producing end products for itself, it engineers cells for its clients. The process goes roughly like this: A client calls up Ginkgo and says, “We’re looking to produce a rose scent for our perfumes that’s cheaper than distilling it from flowers.” Ginkgo’s designers comb through a library of genes and pick out those that are known from previous observation or sequencing to produce the characteristics of rose oil. After these sequences are laid out on a computer, Ginkgo orders the DNA from Twist or other providers, who do much of the synthesizing of the base pairs.
At Ginkgo, the synthesized DNA is then inserted into a host cell, perhaps yeast, which starts producing enzymes and peptides. Trial and error follow. Maybe the outputs from the first gene sequence are too floral, not spicy enough; maybe the ones from the second gene sequence have the right scent, but the cells don’t produce enough of it. Once an effective prototype is found, Ginkgo increases its production by growing the yeast in large vats and streamlining a process for extracting the desired molecules from the soup. What Ginkgo delivers is a recipe and ingredients — the winning genetic code, the host cell and the conditions in which the cells have to be nurtured — which the client can then use on its own.
Ginkgo’s platform first attracted customers in the fragrance industry, but in the last two years it has been partnering with pharmaceutical companies to search for new therapeutics. One such project is seeking to discover the next generation of antibiotics, in order to counter antibiotic resistance. Lucy Foulston, whose background is in molecular microbiology, is leading the effort; Tom Keating, a chemist, is working with her. Together, they highlighted for me a beautiful, twisted paradox — most antibiotics, and most antibiotic resistance, come from bacteria themselves. Bacteria carry genetic snippets with instructions to produce antimicrobial molecules that kill other bacteria. Typically they also have a capacity for self-resistance, so that the bacteria making a particular antibiotic avoid killing themselves, but this resistance can be transferred among bacteria, so that it becomes widespread.
Historically, two paths have been taken to come up with new antibiotics. The first, celebrated in stories of Alexander Fleming and moldy bread, is to seek them in the natural world: Scientists go out, obtain a little bit of soil from a geyser or coral reef, put what they find in a petri dish and see whether it kills any interesting bacteria. The second approach is to comb through chemical libraries in search of molecules that show antibacterial activity. Together, these two approaches gave us a steady supply of new antibiotics up until the 1980s and ’90s, when discoveries began to dry up.
“There was a lot of speculation,” Keating says. “Did we find all the useful ones? Did we find everything that was easy to find? Did we run into bacteria that are now so difficult to kill that the new ones we find don’t really work on them?” Whatever the reason, the reality is that we’ve been running out of new antibiotics in the face of growing antibiotic resistance.
‘I think what we’re just scratching the surface of is, can we program biology to do what chemists have traditionally done.’
The antibiotics project at Ginkgo is looking through bacterial genomes for segments encoded to generate novel antimicrobials. The sequencing efforts of the ’90s and 2000s yielded large databases of bacterial genomes, both public and private, that have given scientists an increasingly sophisticated understanding of which genes produce which molecules. And scientists have also developed the necessary techniques to, as Foulston says, “take these genes out, put them in another bacterial strain” — one they know how to work with — “and then coax that particular strain to produce the molecule of interest.”
Keating continues: “We don’t need the organism anymore. We don’t need it to be growing on a plate. We don’t need it to be killing anything else. All we need is the code.”
No matter how many programming metaphors you use, DNA is messier than code. If you type “print ‘hello world,'” you expect the computer to return “hello world.” If you synthesize a DNA sequence, ACTCAG, and put it in a cell, you might be able to predict with some confidence what comes out of the cell, but you never really know.
Nevertheless, biotech has arrived at a singular new moment — one in which software, hardware, data science and lab science are all finally mature enough to work together and reinforce one another. mRNA vaccines, which had not been approved by the Food and Drug Administration before the pandemic, are a prime example; Ginkgo’s antibiotics project is another. And further advances in machine learning and computer modeling will only multiply the possibilities. The same goes for semiconductors: As small as one of Twist’s 10-picoliter wells might seem, Leproust points out that from the perspective of the 21st-century semiconductor industry, it’s “a Grand Canyon, almost like being in the Stone Age.” Already, the company is experimenting with chips whose wells are more than 300 times smaller, with diameters of 150 nanometers. (For reference, Intel is now fabricating seven-nanometer silicon chips for computers.) It’s a progression that promises to lower the cost of gene synthesis a millionfold and make it accessible to ever more researchers and useful in ever more experiments and applications.
For synthetic biology, the next frontier is to go where even nature hasn’t gone. Instead of trying to replicate the scent of a rose, can we combine genes to produce even more intoxicating aromas? Can we turn DNA into circuits that enable cells to act as living computers? “So far, we’re just taking what nature has already invented, copying it, maybe optimizing it,” Keating says. But he aspires to the sort of command and creative power now enjoyed by chemists, who can synthesize whatever can be diagrammed. “I think what we’re just scratching the surface of is, can we program biology to do what chemists have traditionally done,” he says. “If you can draw a molecule on a piece of paper, can we engineer an organism to produce that molecule, even if it’s something that nature has never seen before? We’re nowhere near that — but, you know, baby steps.”
Yiren Lu is a writer and software engineer based in New York. She last wrote for the magazine about start-ups trying to fix virtual meetings.