David Liu's office on the eighth floor of the Broad Institute in Cambridge, Massachusetts is designed to quiet the mind. A museum-grade gemstone collection lines the walls, interspersed with blue-tinged photos Liu has taken of inspiring science-on-location scenes—the concrete corners of the Salk Institute, a sunset through the Scripps pier, the lights of Durango, Colorado where Darpa often meets. (Liu is a member of Jason, an elite group of scientists that advises the US government on next-generation technologies.) The only thing out of place in the 45-year-old's chemist's office is a three-foot-high perfect replica of Iron Man standing atop his Hulkbuster armored suit.
"It weighs 30 pounds," says Liu, who was on a waitlist for the toy for months. "You should have seen me try to bring it past lobby security. There was a lot of head scratching." It was worth the effort to him, though, to have the daily reminder of the kind of lateral thinking the zilionaire Tony Stark used to beat the biggest, angriest problems.
Because even though Liu isn't fighting green-skinned, gamma-rayed humanoids, he is going after mutants. Specifically, the mutations that cause the 6,000 known human genetic diseases. In the past few years, Liu's become one of the most brightly-shining luminaries in the rapidly advancing field of gene editing. Since 2013, he's published paper after paper in Science and Nature and founded three companies based on his transformative tech, with two more on the way. For any other chemist, a rise to the upper ranks of the biological revolution ignited by Crispr would be beyond improbable.
But not for Liu, who's spent the last two decades harnessing the Darwinian ruthlessness of natural selection to create totally novel molecules. Now he's setting his custom-built evolution engines loose on the molecular machines that cut, paste, erase, and edit DNA. His goal is to create a massive library of disease-targeting tools—so that one day when scientists want to make a genetic fix, they can just pull whichever one they need off the shelf.
It was December 1990, and EJ Corey had just given the most consequential lecture of his career. Addressing a room full of scientists in Stockholm, the organic chemist had explained the work for which he was accepting a Nobel prize. Now he stood to the side of the stage answering questions from a delegation of young Japanese students. One young man at the back of the pack asked him how he'd managed to take an insect hormone, chock full of carbon-carbon double bonds, and turn just one of them into an epoxide. Before answering, Corey remarked how good the young man's accent was.
Liu smiled, and explained that he was actually a freshman at Harvard, where Corey taught organic chemistry and led a world-renowned research laboratory. Born to Chinese parents but raised entirely in California, Liu said—in perfect English—that he wanted to join Corey's lab. The newly-minted Nobel laureate told the 17-year-old to come back when he'd learned some organic chemistry.
True to his word, Liu showed up in Corey's office spring semester, after having completed the intro course. This time Corey caved.
"What you have to understand about David is that he is fearless," says Corey, recalling Liu's penchant as an undergrad for pulling all-nighters to get experiments to work. "In the lab that means taking on experiments that are total long-shots. He sees beyond his contemporaries in sensing important new challenges and attacking them, even as they appear to be quite formidable."
After Liu completed his undergraduate work in Corey's lab, he moved out to Berkeley for his PhD, where he invented new methods for incorporating synthetic amino acids—beyond the 21 that occur naturally—into proteins. Corey had told him not to spend more time in graduate school than necessary. But he was still surprised when his chemistry colleagues at Harvard offered the 25-year-old Liu a job after hearing him give just one talk about his dissertation. Today he holds joint appointments at Harvard, the Broad, and the Howard Hughes Medical Institute. But in the fall of 1999, Liu was a first-time professor half as young as his faculty peers, starting a lab in a totally new field.
"I had no idea what I was doing," says Liu, who was often chastised for not calling his former teachers by their first names now that they were colleagues. "In retrospect my ignorance should have been a cause for alarm. But I think it also gave me a sense that I could explore any kind of problem because I wasn't worried about the feasibility of it working out."
He set his lab on a course to explore how you could apply the principles of evolution on a molecular scale. It didn't go well at first; the NIH rejected all his proposals and journal editors wouldn't even look at his papers. But then he hit his first big invention: DNA templated synthesis, today known as DNA-encoded libraries.
Liu figured out that if you attached chemicals to DNA strands, you could change their final product. Instead of proteins, you could use DNA to code for small, man-made molecules, aka drugs. By hacking biology's natural laws of attraction, you could make lots of combinations of new drugs, very quickly. Today, the technique for making vast libraries of molecules is a standard tool of the pharmaceutical industry.
But Liu wanted to go bigger. And for an organic chemist, that meant bringing the process of unnatural selection to proteins, to give the workhorses of the biology world functions never before seen in nature. His students had already been doing it manually—making lots of colonies of bacteria, mutating their genes, and selecting for the properties they wanted. But it sometimes took generations to stumble on to the desired effect, and each cycle took about a week to complete and months to analyze. Then Kevin Esvelt walked in Liu's door.
The scientist now best known for introducing the world to Crispr-based gene-drives was at the time, in 2004, a brand-new grad student. Esvelt asked Liu to give him the most difficult project he had. Alright, said Liu: Figure out how to make proteins evolve on their own.
Esvelt imagined mapping natural selection onto the 10-minute life cycle of bacteriophages—viruses that attack bacteria—so that proteins capable of totally novel chemical reactions would mutate in vivo inside a hot, soupy medium-filled vessel they affectionately referred to as "the lagoon." It took Esvelt five and a half years before he got the system, dubbed phage-assisted continuous evolution, or PACE, to work.
"That allowed us to evolve molecules at speeds up to 50 generations per 24 hours instead of one per week," says Liu. His students have since used PACE to fashion enzymes that outperform their natural counterparts, like resistance-proof insecticidal proteins (which Monsanto promptly licensed). But nothing's garnered more excitement than using it on Crispr.
The gene-editing tool is a combination of a DNA-snipping enzyme called Cas9, and little snippets of RNA that guide it to a specific spot in the genome. But Cas9 can't bind just anywhere—it needs a certain sequence to grab hold, a sequence that only occurs in about 6 percent of the human genome. And it's also not very good at swapping DNA sequences, because it relies on the cell's own machinery for repair. Because DNA breaks are scary business, some cells go into first-aid mode, rejecting Crispr edits. And as scientists reported this week, getting around that cellular stubbornness could make Crispr'd cells more vulnerable to becoming cancerous.
So scientists have been racking their brains trying every which way to add to Crispr's utility while also making it safer. Some are out scouring the globe looking for novel Crispr-associated proteins in rare, unsequenced bacteria. Others are manually tinkering with the enzyme's structure. Liu's lab—housed in the Broad's Crispr crucible alongside other pioneers like George Church and Feng Zhang—is evolving the next generation of genome manipulation tools instead.
Nicole Gaudelli first heard about Liu's evolution workshop during a talk he gave at Johns Hopkins in 2013, where she was doing her PhD at the time. The moment it was over she marched up a flight of stairs to her advisor's office, closed the door, and told him that she was going to do her postdoc with Liu or she wasn't doing one at all. By February of the next year she was in Cambridge using PACE to make new kinds of antibiotics. Then the postdoc got caught up in the Crispr craze.
One of her colleagues, Alexis Komor, had recently published something Liu's lab was calling a "base editor," a modified Cas9 enzyme that didn't cut DNA. Instead, it worked more like a pencil, rewriting single nucleotides to convert C:G base pairs to T:A. That kind of a fix could cure about 15 percent of the 32,000 single base errors that cause genetic disease. Gaudelli wanted to go after a bigger piece of the pie. If she could make an editor that flipped G:C to A:T, that would address half those diseases.
It was theoretically possible—if she could retool an existing enzyme that made the swap in RNA. PACE wouldn't work for her particular project; Gaudelli would have to go back to evolving by hand. Her choice made her the first person in 19 years to break Liu's only rule: "If step one is to evolve your starting material, pick a different project." Because even if step one worked, Gaudelli would still have to Frankenstein it together with the remaining components of her base editor—a high risk bet that could leave her with nothing to show for her postdoc.
Liu let her because she was up for the challenge. But also perhaps because he saw something of himself in her. "My background wasn't in this, so I didn't perceive it as an impossibly risky thing to do," says Gaudelli. "And the environment David has created is 180 degrees from the culture in most chemistry labs. It's so nurturing that it removes the barriers like fear of failure. He just makes you feel invincible."
Seven rounds and two grueling years later she had her new base editor. Liu submitted their paper describing a way to fix half the disease-causing single base snafus to Nature the Thursday before Columbus Day 2017. The paper was online 16 days later—a record for his lab. It was a surprising turnaround, but not something Liu says he's hanging his hat on. "Taxpayers don't support our research so that we can just publish more papers," he says. "We have an obligation to return these technologies to the public for the benefit of society."
To that end, Liu's become a bit of a serial entrepreneur. In 2013, he signed on as a scientific cofounder of Editas Medicine, one of the first three big Crispr human therapeutics companies, alongside Broad colleagues Church and Zhang. In March, he unveiled Pairwise Plants, a Monsanto-backed startup aimed at engineering fruits and vegetables. In May, he and Zhang launched Beam Therapeutics, to turn base editing into treatments for genetic diseases. Gaudelli, who had offers to start her own lab at more than one top 10 school, opted instead to take a research job at Beam. She wants to be there to take the base editor she birthed out of bacterial evolution and figure out how to get it into patient's hands.
Patients are on Liu's mind too, even as he teaches undergrads about biological chemistry and pushes forward with new ways to gain access to all 3 billion bits of the human genome. His lab recently came up with an evolution-accelerating PACE system for base-editing enzymes. The work is still unpublished, but it means a student won't have to go through what Gaudelii did to evolve each of the four remaining kinds of base editors. In a drawer behind his desk Liu keeps letters he receives from the parents of children with genetic diseases, who've read about his work and want to know when it might become available to help their kids.
One mom in Seattle recently sent him a painting her seven-year-old daughter made of red flowers blooming out of long green stems. Her Dravet Syndrome—which causes severe seizures—is caused by a single T to G mutation. It's one that Liu's group hasn't yet figured out how to fix. Tony Stark's Hulkbuster might help Liu see that there's always a way to crack those still unsolved problems. But it's the letters that remind him why they're worth solving.