A leading genome-editing researcher is urging extra caution as drug companies race to turn the landmark technology he helped create into human medicine.
In a paper published today in Nature Medicine, Feng Zhang of the Broad Institute of MIT and Harvard and colleague David Scott argue that researchers should analyze the DNA of patients before giving them experimental medicines that alter their genes with the breakthrough technology CRISPR. The suggestion, among others in the paper, stems from a deeper look at the wide array of subtle differences in human DNA.
Zhang is a key inventor of CRISPR-Cas9, which describes a two-part biological system that slips into the nucleus of cells and irreversibly alters DNA. One part is an enzyme, nature’s molecular scissors, which cuts DNA. The second part is a string of ribonucleic acid (RNA) that guides the enzyme to the proper spot. In five years since its invention, CRISPR-Cas9 has become a mainstay of biological research, and researchers including Zhang (pictured above) have moved quickly to improve upon its components. His work is at the center of a long-running patent battle to determine ownership of the technology.
Zhang and Scott’s recommendation taps into a long-running debate in the gene-editing field about off-target effects—the fear of misplaced cuts causing unintended harm. Most recently, the FDA took up a similar issue at a meeting to assess a type of cell therapy, known as CAR-T, for kids with leukemia. The FDA highlighted the risk that the cells, which have certain genes edited to make them better cancer fighters, may cause secondary cancers long after a patient’s leukemia has been cured. (FDA advisors unanimously endorsed the therapy’s approval nonetheless.)
Some researchers say there should be near certainty that gene altering techniques won’t go awry before testing in humans, caution that stems in part from gene therapy experiments in the U.S. and Europe nearly 20 years ago that killed an American teenager and triggered leukemia in several European boys.
While no medicine is risk-free, other researchers say the tools to gauge risk have improved.
Andy May, senior director of genome engineering at the Chan Zuckerberg Biohub in San Francisco, calls Zhang and Scott’s recommendation for patient prescreening “a good discussion point,” but “the danger is someone will pick up on this and say you can’t push forward [with a CRISPR drug] until everyone is sequenced.”
“It’s an extremely conservative path to take,” says May, who until recently was the chief scientific officer at Caribou Biosciences, a Berkeley, CA-based firm in charge of turning the discoveries of UC Berkeley’s Jennifer Doudna and her colleagues into commercial technology. (May was also a board member of Cambridge, MA-based Intellia Therapeutics (NASDAQ: NTLA), which has exclusive license to use Caribou’s technology in human therapeutics.)
Berkeley is leading the challenge to Zhang’s CRISPR patents and last week filed the first details in its appeal of a recent court decision in favor of Zhang and the Broad Institute.
Zhang sees prescreening as a form of companion diagnostic, which drug companies frequently use to identify the right patients for a study. A whole genome sequence—which costs about $1,000—could filter out patients unlikely to benefit from a treatment or at higher risk of unintended consequences, such as cancer. In the long run, it could also encourage developers to create more variations of a treatment “to make genome-editing based therapeutics as broadly available as possible,” said Zhang.
It’s well known that human genetic variation is a hurdle in the quest to treat genetic diseases either by knocking out disease-causing genes or replacing them with healthy versions. But Zhang and Scott use newly available genetic information to deepen that understanding. In one Broad Institute database with genetic information from more than 60,000 people, they find one genetic variation for every eight letters, or nucleotides, in the exome—that is, the sections of DNA that contain instructions to make proteins. (There are 6 billion nucleotides in each of our cells.) The wide menu of differences is, in effect, an open door to misplaced cuts that CRISPR’s enzymes might be prone to.
Zhang and others are working on many kinds of enzymes, from variations on the workhorse Cas9, to new ones entirely. He and Scott found that the deep pool of genetic variation makes some forms of the Cas enzyme more likely than others to go awry, depending on the three-nucleotide sequence they lock onto in the targeted DNA.
Zhang and Scott write that CRISPR drug developers should avoid trying to edit DNA strings that are likely to have “high variation.” In their paper, they examine 12 disease-causing genes. While more common diseases, such as those related to high cholesterol, will contain higher genetic variation because of the broader affected population, every gene, common or not, contains regions of high and low variation. Zhang and Scott say developers can build strategies around the gene regions they are targeting.
For example, going after a more common disease might require a wider variety of product candidates, akin to a plumber bringing an extra-large set of wrenches, with finer gradations between each wrench, to a job site with an unpredictable range of pipe sizes.
CRISPR companies say they are doing just that. “We have always made specificity a fundamental part of our program,” says Editas Medicine CEO Katrine Bosley. Zhang is a founder of Editas (NASDAQ: EDIT), which has exclusive license to the Broad’s CRISPR-Cas9 patents and some subsequent work. “Certainly, if there is not a strong belief that a given therapy has the potential to be effective for a given patient,” Bosley says, “it would not be appropriate to include that patient in a trial.”
Bosley did not say specifically whether a whole genome sequence of every potential patient in a clinical study, as Zhang and Scott suggest, is part of Editas’s preclinical work. Editas recently announced that its lead program, a treatment for a rare genetic eye disease, would likely enter clinical studies in 2018 instead of 2017.
Just as a conventional drug developer will often start with a vast library of chemical compounds that might be effective against a biological target, CRISPR companies start with thousands of versions of RNA guides. Those guides are matched up against an array of potential off-target sites, says CRISPR Therapeutics (NASDAQ: CRSP) CEO Rodger Novak. If one version of a guide appears to cut at an off-target site, the guide is tossed out, says Novak. CRISPR Therapeutics uses whole genome sequencing in its preclinical work, but not necessarily for each potential patient in a clinical study.
Novak calls the Zhang and Scott paper “interesting… but not necessarily surprising or a fundamental change as to how we look at targeting specific genes.”
CRISPR Therapeutics has entered into collaborations with Bayer HealthCare and Vertex Pharmaceuticals (NASDAQ: VRTX). With Vertex, it might push its first program for the blood disorder beta-thalassemia into clinical studies next year.
In the case of CRISPR, the critical tool in question is the RNA guide that can be customized to match a string of the patient’s DNA, usually about 20 nucleotides long. The guide and scissors travel into the cell’s nucleus together. When the guide finds its match in, say, a string of DNA that is causing disease, the scissors make their cut. CRISPR is also capable of replacing a gene—a cut-and-paste operation, or “correction,” instead of just a cut, or a “knockout”—and Zhang says his recommendations for better, safer CRISPR therapeutics apply to “various modes of gene correction.”
Correcting genes in humans, if at all possible, will be farther down the road, however. The nearer-term goals for CRISPR drug companies are medicines that simply knock out a disease-causing gene, with no replacement necessary.
The new paper comes a few days after a report that researchers at the Oregon Health and Science University used CRISPR to alter genes in human embryos, which would be a first in the U.S. Their work, which used embryos that were never meant to be implanted, has not been published; MIT Technology Review broke the story, citing at least one researcher associated with the work.
The possibility of editing human embryos, sperm, or eggs—together known as the germline, because the altered traits could be passed down through generations—is a far more controversial topic than the one-off, or “somatic” changes that companies and researchers are racing to turn into human therapies. Germline editing is likely necessary to address some diseases. But the ethical roadblocks, including the specter of making aesthetic and other non-medical changes, will keep a lid on real-world applications, at least in the U.S., for some time. Doudna and other prominent scientists called in 2015 for a moratorium on germline editing aimed at in vitro fertilization and other applications.
Zhang photo by Keith Spiro.