Ready, Aim, CRISPR: Will Gene-Editing Tools Hit Their Targets?

Xconomy National — 

Gene therapy is finally upon us. After more than two decades of work, one product is approved in Europe. Others are steadily advancing through clinical trials, and gene therapy companies are doing big deals and raising big sums of cash.

Now here comes a more precise version of gene therapy—gene editing. Instead of trying to insert a correct version of a faulty or missing gene as in gene therapy, the idea behind gene editing is to actually snip out the faulty genes that cause disease, and perhaps even replace them with new, improved versions.

Editing genes, not just adding genes, could be a huge help with the rare inherited Li-Fraumeni Syndrome, says Robert Lufkin, an oncologist in Portland, OR, and advisor to the Li-Fraumeni Syndrome Association. Li-Fraumeni, which causes cancer in childhood or early adulthood, and with multiple types of cancer, stems from a mutation in a gene called p53 that normally suppresses tumor growth. But adding a healthy version of p53 isn’t necessarily the answer, because accumulation of the mutated version has its own problems. “I am hopeful that someday a form of genome editing may be used in severe genetic disorders” like Li-Fraumeni, says Lufkin.

My colleagues and I write frequently about the promising technologies that are making gene editing possible, such as zinc fingers and CRISPR/Cas9.

In particular, CRISPR/Cas9 already has caught on around the world as a research tool to cut out or replace genes in organisms from bacteria to wheat to mice to monkeys. Work in human cells is starting to emerge, too. Derived from a bacterial defense system, CRISPR/Cas9 is a potential Nobel-winning biotech discovery. (CRISPR stands for clustered regularly interspaced short palindromic repeats; Cas9 for CRISPR-associated protein 9.)

A few companies are racing to make gene-editing therapies. With its proprietary zinc finger technology, Richmond, CA-based Sangamo Biosciences (NASDAQ: SGMO)) is the first (and only) company to reach clinical trials: It’s now running a Phase 2 trial in HIV and has just been greenlighted by the FDA to begin clinical trials in beta thalassemia.

(A third gene editing system, known by the acronym TALEN, has yet to generate the widespread use of CRISPR/Cas9 or the clinical progress of zinc fingers, although French TALEN developer Cellectis inked a deal with Pfizer (NYSE: PFE) last December.)

Before it can be a successful treatment or cure, however, gene editing must solve a thorny problem: How to ensure that the edits in DNA are made in the right spots? Go off target, and the genetic manipulation could have serious consequences. The history of gene therapy offers cautionary tales of genetic engineering gone awry, such as the unexpected trigger of leukemia in a French trial for X-linked severe combined immune deficiency disorder (the “bubble boy disease”), and the death of Jesse Gelsinger from an immune system reaction in Philadelphia in 1999.

But now comes progress in the effort to build a system of quality control for gene editing. In the past two months, three academic groups have published papers—all in Nature Biotechnology—describing new ways to measure the frequency and location of off-target cuts in cells’ DNA. A review in this month’s issue covers all three papers, and the reviewers laud them for moving the field forward: “The new studies are a major step toward clinical applications of genome engineering as they show that sensitive, genome-scale detection of nuclease activity is now technically feasible.” (Nucleases are the “scissors” that gene editing systems use to cut DNA.)

That sensitive detection is important, because it’s virtually inevitable that the gene-editing tools will go off target. The goal is to know quickly when, where, and how frequently, and use those data to adjust the technology in certain diseases or patient groups—or steer clear of them entirely.

Take CRISPR/Cas9. The system uses two main components: an enzyme called a nuclease to cut DNA, and a strand of RNA that acts as the nuclease’s guide by matching up with the segment of DNA the enzyme is meant to cut. The beauty of CRISPR/Cas9 is that for most uses, the scissors stay the same. Only the guide needs to be swapped out, a relatively simple exercise in many biomedical labs these days.

That’s why it has caught on so rapidly. “You can make hundreds of these things trivially,” says Jacob Corn, scientific director of the new Innovative Genomics Initiative, jointly run by the University of California, Berkeley, and the University of California, San Francisco. Experiments that used to take years can now be run in weeks, but now researchers need the tools to detect edits just as quickly and easily, Corn says.

(IGI’s executive director is Jennifer Doudna, a UC Berkeley professor whose pioneering work on CRISPR/Cas9 helped turn it into a genome-editing technology. I wrote about her role in an ongoing patent battle here.)

As carefully engineered as the scissors and the guide might be, though, therapeutic uses could end up with millions of copies going into millions of cells per patient. Mistakes will be made, as a politician might say.

But how many? And where in the DNA will they occur? And how are they different from the constant DNA cuts and mutations happening constantly in our cells, which for the most part our cells know how to deal with?

Especially with CRISPR/Cas9, there is a huge gap between the ability to produce new guides (and hit new targets) and the ability to see if—and where—the scissors went awry. “The most pressing need is to know how CRISPR/Cas9 works,” says Corn. “It’s only been around three years now.”

Knowing where and how often gene-editing technology is making those cuts will give therapeutics developers more data points to consider as they move a product toward clinical trials. Keith Joung, a researcher at Massachusetts General Hospital who coauthored one of the new studies, likens the knowledge to the preclinical toxicology tests that developers of traditional chemical drugs use. (Joung is also a scientific founder of Editas Medicine of Cambridge, MA, one of three venture backed startups competing to develop therapies that use CRISPR/Cas9.)

Keith Joung

Keith Joung

How can scientists find the wayward cuts? In their study, Joung, Shengdar Tsai (a postdoc in Joung’s MGH lab), and colleagues took advantage of a well-known trait of cell repair. When a cell fixes broken strands of DNA, other material nearby can get incorporated into the fix. The researchers introduced tiny oligonucleotide tags into cells at high concentrations, which were taken up into the repair sites. Those tags could then be tracked and counted. “The number of times you see a tag ‘hop’ into a particular place would correlate with how frequently the site gets cut,” says Joung.

He calls the process, named GUIDE-Seq, “comprehensive and sensitive,” and says it seems to be picking up mutation-causing edits of extremely rare frequency (“probably much lower” than 0.1 percent).

Some of Joung’s previous work has been licensed to Editas, but it’s unclear if GUIDE-Seq will follow suit. He and his colleagues will continue to refine it in at least two ways.

First, GUIDE-Seq needs to be tested in more “therapeutically relevant” cells, says Joung. The cells he used were old cancer cell lines that are great for research because they are practically immortal, but they don’t resemble anything you’d find in the real world—in part because of the very repair machinery they use to keep themselves going.

Second, GUIDE-Seq needs to be tested with a wider variety of guides and scissors. Right now, CRISPR/Cas9 tools come in mainly one “brand,” so to speak, derived from the bacterium Streptococcus pyogenes. But researchers including Joung are feverishly working to expand that toolkit, using enzymes from other bacteria (S. thermophilus, essential to yogurt and cheese production, is on Joung’s list) and building different kinds of guides that help the enzymes hone in on the right strands of DNA to cut.

Another new method for finding off-target cuts comes from a group of researchers in the Alt Laboratory at Boston Children’s Hospital. They measure instances, called chromosomal translocations, of major pieces of DNA being sheared off by the gene-editing scissors and moved to another part of the genome. That paper caught the attention of Matthew Porteus, a Stanford University researcher and doctor who treats children with hematologic cancers. “It’s as important as Keith (Joung)’s paper, or perhaps more,” says Porteus, who works with gene-editing technology and is helping the venture-backed Crispr Therapeutics, of London, develop a hematological product with CRISPR/Cas9 tools.

The paper suggests that gene-editing tools could create translocations between the target site and random sites in the genome, Porteus explains. “That needs exploration,” he says, because of a potential link between translocations and cancer. “What’s the functional consequence of a low frequency of translocations? Are they associated with cancer or of no consequence?” he asks. “We don’t have enough data.”

Porteus certainly wants better quality control systems. But he also cautions that the perfect should not be the enemy of the good. He counts himself among those who suspect off-target breaks from these new tools are something our bodies—our cells—already know how to deal with. “Cells are accumulating small insertions and deletions all the time,” says Porteus. “Our genome is not stable.” He’s not convinced that these new tools create more random mutations than occur during the normal life of a cell—or that the mutations they might incur will have any bearing on health. He uses Li-Fraumeni Syndrome as an example. Even with a damaged p53 gene that should let tumors proceed unchecked, people born with Li-Fraumeni sometimes go years, even into adolescence, before developing a detectable cancer.

In other words, people working on therapeutics may not always need to know exactly where, and how often, their gene-editing tools are making unintended cuts. There are degrees of understanding, and of risk, in every therapeutic undertaking. Regulators charged with public safety know this. Patients with life-threatening diseases know this. Investors who back cutting-edge biotech startups know this.

As Corn builds out his lab at the new Innovative Genomics Initiative center on the UC Berkeley campus, he says he also wants to foster a “deeper conversation beyond these three papers” in Nature Biotech. (The third study, from scientists in Southern California and China, took a similar approach to Joung and colleagues but instead of an oligonucleotide used a defanged lentivirus as the “tag” that was taken up when the cell repaired its DNA.)

Corn agrees with Porteus that early worries about off-target effects “haven’t played out yet, but I don’t want to get into a situation where we do things too fast in patients.”

Still, every gene therapy or editing program that advances in the clinic under regulatory watch gives everyone more confidence about, in Rumsfeldian terms, the known unknowns. As mentioned previously, Sangamo has received the FDA’s blessing to test two of its therapies in humans. The company gets high praise from its peers for characterizing—understanding and describing—well known off-target effects of its zinc finger technology. “They don’t know the comprehensive spectrum, but they know the critical ones,” says Caribou Biosciences chief scientific officer Andy May, whose company, cofounded by Berkeley’s Doudna, is building a suite of CRISPR/Cas9 tools for human therapeutics, agriculture science, and more. (It has exclusively licensed the therapeutic uses of its tools to Intellia Therapeutics of Cambridge, MA, as I wrote about here.)

CEO Edward Lanphier and other Sangamo officials have described to me in the past how they have developed tools and methods to minimize off-target hits with their zinc-finger therapies. And they have questioned whether CRISPR/Cas9, which uses a less complicated guide to move its nucleases—its enzymatic scissors—into place, will ever reach the same level of specificity. (Sangamo officials did not return requests for comment for this article.)

Will the quality control tools soon be good enough for drug developers to use? Joung says he thinks GUIDE-Seq would help developers do two things: choose the right guide RNAs for their targeted gene, and after the cells are edited, serve as a check for unwanted mutations.

May, who as an Intellia board member was privy to that company’s landmark deal with Novartis (NYSE: NVS) to explore CRISPR/Cas9 in Novartis’s CAR-T immunotherapy program, wants to see more work from GUIDE-Seq and the other two systems. “All have their pros and cons,” he says. “None are unbiased.” The next step, he says, is for the systems to test a wider array of guides and scissors: “To understand the phenomenon, you need to use a much larger set.”