Gilmore’s deduction was that “the repeated introduction of various antibiotics since the ’40s not only selected for resistance to those antibiotics, but it selected for the ability to acquire [drug] resistance with enhanced facility.” In other words, bacteria were gaining a selective advantage by losing their CRISPR defense system. That might leave them vulnerable to phages, but it would make it easier for them to acquire new genetic elements that could confer antibiotic resistance (via a mechanism called horizontal gene transfer). The overuse of antibiotics has resulted in bacteria effectively lowering their genome defenses to make it easier to acquire resistance from other microbes.9 The genomes of clinical isolates of some bacteria might be 25 percent larger than those of commensal strains.10
Zhang’s ears pricked up as Gilmore casually mentioned that bacteria contain CRISPR and its attendant nucleases. The acronym had appeared in several top journals by this time, but Zhang had to google the term to learn more about it. The next day, Zhang flew to Miami to attend a conference,IV but hunkered down in his hotel room instead, devouring the CRISPR literature. The more he read, the harder it was to contain his excitement. He read a CRISPR review article by Horvath and Barrangou in Science11 and the “amazing” 2010 Nature paper by Moineau’s group that showed that the type II CRISPR-Cas system cleaves phage DNA.12 But Zhang wasn’t interested in bacterial immunity or how to make a cheesier pizza. His focus was genome editing—a technique that could work in animal models and ultimately in humans. If, as Moineau’s paper suggested, you could use RNA and CRISPR to target DNA, this would be much easier than either ZFNs or TALENs.
On Saturday February 5, Zhang fired off an email to Le Cong: “Take a look at this,” he wrote, including a link to the Horvath-Barrangou review. “Maybe we can test in mammalian system.” Le Cong replied, “It should be very cool to test in mammalian systems.” Two days later, Zhang emailed: “Hey let’s keep this confidential. This can completely replace any kind of [zinc finger] system. I ordered the Cas genes for synthesis. We should be able to test them… I’ve done a patent search.”
On February 13, Zhang filed a “memorandum of invention,” an internal Broad Institute document that summarized his new invention: multiplexed genome engineering. The idea, which he said originated just nine days earlier at Gilmore’s lecture, was a clear statement that Zhang thought CRISPR might complement, or even replace, the current methods of gene editing, ZFNs and TALENs.13 Zhang judged that CRISPR had the makings of a programmable gene-editing technology that could be targeted to almost any DNA sequence.
In a short McGovern Institute video filmed at the end of 2011, Zhang discussed his research goals. Coming from an engineering background, he said, “I think about how to take things apart, put them together, and then try to fix it.” Using the same approach, Zhang hoped “to understand disease mechanisms and be able to fix the brain.” The main tools in his arsenal were the TALEN proteins.14 He exuded a quiet fearlessness, as if nothing was beyond his reach. But there was no mention yet of CRISPR or the potential to edit human DNA.
Zhang and Le Cong’s early efforts using Cas9 did not work as planned. Two key modifications were required to get Cas9 to work in human cells, as I mentioned earlier: one was codon optimization to make the gene appear less foreign to a human cell. The other was to add a nuclear localization signal, a motif that helps ferry DNA into the cell nucleus (obviously not an issue in bacteria as they lack a nucleus). But for some reason, the S. thermophilus Cas9 wasn’t behaving nicely. Zhang needed a new system and a Cas expert. The man he was looking for—who had almost been a colleague of Gilmore’s—was just down the road in New York City.
Shortly before 10:00 P.M. on January 2, 2012, Zhang sent a short email to Marraffini at the Rockefeller University. The Argentine had received flattering faculty offers from MIT and Yale, but the prospect of living in cosmopolitan New York was irresistible. And Rockefeller’s rich tradition in microbiology and genetics proved a perfect fit. Zhang didn’t beat about the bush:
Dear Luciano,
Happy new year! My name is Feng Zhang and I am a research [sic] at MIT. I read many of your papers on the Staphylococcus CRISPR system with great interest and I was wondering if you would be interested in collaborating to develop the CRISPR system for applications in mammalian cells.
Would it be possible to schedule a phone call with you in the next few days?15
Marraffini had never heard of Zhang but a quick Google search left him suitably impressed. He replied ninety minutes later and said yes. “We have been working on a ‘minimal’ CRISPR system that could be useful… Happy 2012!” They sealed the collaboration in a phone call the next day. A week later, after receiving a nudge from Zhang, Marraffini emailed an eight-page document that highlighted DNA sequences and other useful information about CRISPR in S. pyogenes. He closed with a five-step plan to produce the “minimal” Cas system that he believed could be used to edit human genes. The relevant materials, including the gene template for Cas9 and the tracrRNA, followed by mail.
Zhang was a man in a hurry. He was working in the trenches with his team, his weapons of choice a row of personalized pipettors with FENG taped across each one. On January 12, he told Marraffini he’d already identified a pair of target sites in a human gene (AAVS1) and was preparing to express the bacterial genes in mammalian cells. Around this time, he was included in an $11 million NIH grant application filed by the Broad Institute’s then deputy director, David Altshuler.16 The main idea was to use genome editing to engineer stem cell models for type 2 diabetes and other diseases. Zhang proposed using the four components described by Charpentier—the CRISPR array, Cas9, the tracrRNA, together with another enzyme (RNase III)—to reconstitute the active CRISPR complex in human cells.
Within a few months, Zhang had enough data to show he could deliver the CRISPR-Cas9 machinery into the nucleus of mouse or human cells (by flanking Cas9 with nuclear localization signals) and target a gene of choice. Marraffini also had preliminary results showing that Cas9 could target a human gene sequence. They also found an isoform of the tracrRNA that was expressed in human cells. Zhang toyed with the idea of writing up these preliminary results—potentially the first demonstration of CRISPR gene editing in animal cells—but decided against it. “I [wanted] to wait until we have a paper that can make a significant difference, not just to be first with something,” he said.17
In late June, Zhang saw the names of Doudna and Charpentier on a CRISPR paper in Science. The consensus was that this was a landmark paper, but Zhang was uncharacteristically dismissive. “I didn’t feel anything,” he told WIRED. “Our goal was to do genome editing, and this paper didn’t do it.”18 But the report impacted Zhang’s team in at least two important respects. First, it showed that competition in CRISPR was heating up. “We didn’t think we got scooped, but we knew people would jump on the topic so we had to speed up,” Le Cong told me.19 Second, it introduced the idea of the single-guide RNA (sgRNA). Zhang and Le Cong didn’t hesitate to put this idea into action with the help of a new recruit with something to prove.