Skip to main content
Premium Trial:

Request an Annual Quote

As CRISPR Tools Improve, Scientists Mull How to Calibrate Safety, Efficacy in Clinical Applications


This is the first article in an in-depth series on the impact of CRISPR/Cas9 genome editing technology on basic and clinical research, the biotechnology industry, and the world at large. GenomeWeb will be running the rest of the multi-part series over the next several months.

WASHINGTON, DC (GenomeWeb) – Scientists are contemplating how to bring CRISPR/Cas9 genome editing tools out of the research lab and into the wider world, but they face a paradox: even as tools are improving, there is still no standard way of comparing them or even measuring how well they edit their targets, or how poorly they stay on target.

The risk of off-target activity, a known occurrence in CRISPR/Cas9 editing, is just one of the unintended consequences being considered here at this week's International Summit on Human Gene Editing. It's a small and technical matter in comparison to some of the other issues, such as morals and ethics and social equity, but it offers a concrete example of how a community could come together to address a risk. By the same measure, it also offers an example of the difficulties of agreeing on a path forward, even when there's consensus that something must be done. 

"One of the major challenges for genome editing, especially using the CRISPR/Cas9 system, is the target specificity," Feng Zhang, of the Massachusetts Institute of Technology and the Broad Institute, told GenomeWeb. "We've been trying to think about ways to improve specificity for a long time now."  

Zhang's lab, among others, has developed several tools that feature reduced off-target editing, such as a mutant Cas9 that makes a single-strand nick in DNA, rather than a double-strand break, using two "nickases" to make it harder for stray nuclease activity to cause a break. But all other methods have their limits, he said.

When Zhang's lab discovered a new CRISPR enzyme that has potential for use in human genome editing, called Cpf1, they applied a systemic approach looking at genomic databases for sequences that looked similar to the functional domains in the Cas9 enzyme. Zhang applied that same, rational approach to interrogate the crystal structure of the gold-standard Cas9 from Streptococcus pyogenes, SpCas9, and engineer a variant that improves its specificity, which he calls enhanced SpCas9, or eSpCas9. He and his co-authors published their study Tuesday in Science

Zhang said the new eSpCas9 enzyme works because it reduces the affinity for the Cas9 protein to bind to the non-target strand of DNA, allowing a greater chance of re-hybridization with the target strand.

"We found a region on the protein where the DNA strand that's not base pairing with RNA is highly positively charged," Zhang said. By substituting alanine at various locations, Zhang and his colleagues found mutants that were neutrally charged and didn't bind as well, offering an increase in specificity. For some of the mutants, they couldn't detect any off-target activity.

"It started with a hypothesis we had when looking at the structure," he said. "The underlying thing is, we wanted to see if there are generalizable principles to see how these RNA-guided enzymes work."

While the crystal structure study was done with Cas9, Zhang said the same approach could also be used with Cpf1 or some of the other alternative CRISPR systems he's discovered. "This generalizable mechanism is probably going to manifest as well," he said. "Cpf1 is a RNA-guided nuclease. There's some region on the protein that has to displace the strand of DNA that is not pairing with RNA."

To show that the eSpCas9 had less off-target activity, Zhang used a method called BLESS, developed by MIT's Nicola Crosetto, combined with targeted deep sequencing. But that's not the only way to determine specificity. There are several strategies available to researchers.

"They're all finding off-targets but they're not all finding the same ones," Jennifer Doudna, of the University of California, Berkeley, told GenomeWeb. Because of this, she and other leaders in the CRISPR field have called for standardization of the methods used to assess both on- and off-target activity of CRISPR systems.

So far there's been no direct head-to-head comparison of even two of the half-dozen methods for assessing off-target activity, Harvard University and Massachusetts General Hospital researcher Keith Joung said. Joung's lab developed GUIDE-seq, one of the methods used to find off-targets.

"Right now, it's hard to know if the differences [in off targets revealed by the different methods] are due to the assay or different cell types or even the different labs. It's going to take someone to invest time to learn how to do the different methods and to do a comparison side by side," he said.

That's easier said than done, in part because the field of CRISPR genome editing is moving so fast. "Even in our own GUIDE-seq work we haven't always done multiple replicates for our guide [RNAs]," Joung said.

Even as there's agreement that the field needs standardization on how to measure off-target activity, there's no firm consensus on how much off-target activity in clinical applications of the technology would be too much.

In presentations on Tuesday, Harvard professor George Church raised the issue of the background mutation rate and Jin-Soo Kim of Seoul National University, who has developed an in vitro assay to asses off-target activity, noted that conventional drugs that have already been approved by the FDA also have effects on the genome.

Both Doudna and Joung participated in a breakout session at this week's summit to discuss specificity and efficacy of CRISPR/Cas9. Some scientists didn't think off-target effects were a concern. Joung was more cautious.  "Could [off-target effects] be important? I think yes, they could be," he said, but added, "nobody has ever shown [that CRISPR/Cas9] off-target effects cause a deleterious effect in the transformation of a human cell."

If CRISPR/Cas9 genome editing is to make it into clinical use, regulatory bodies will have to be comfortable that, like any therapy, it is safe and effective.

Sangamo Biosciences, which uses zinc-finger nucleases as a genome editing technology, has already cleared the US Food and Drug Administration's bar for preclinical studies. The firm has completed a Phase I clinical trial for editing the CCR5 protein in T cells to treat HIV and has several open investigative new drug applications for other genome-editing therapies, Sangamo Senior Scientist Fyodor Urnov said. "I can't speak for the FDA, but I'm hopeful that as other technologies advance to the clinic they look at what we've done and learn what they can," he said.

One of the companies looking to follow Sangamo is Intellia Therapeutics, a company Doudna co-founded. Intellia CMO John Leonard said that avoiding off-target activity can be dealt with by selecting the right guide RNAs to target the genome. "One can find promiscuous guides and one can find optimized guides," he said. Finding appropriate guides is part of the work the firm is doing to bring CRISPR/Cas9 into the clinic. "One of the incredibly exciting things about SpCas9 is there are so many possibilities to choose from. One has many shots to find a very selective guide."

FDA officials at the summit declined to discuss specifics of what the agency might be looking for with regards to on- and off-target activity of CRISPR/Cas9 in human clinical studies. 

"I think these regulatory agencies are grappling with the same issues being discussed at this meeting," Doudna said. "The FDA is in the same boat as the rest of us in figuring out what they need to be concerned about."

She added that she had heard from colleagues that they were in discussion with the FDA. "I think a combination of more data and better models are going to provide the kind of baseline tools we need to have," she said, but warned that the field was at least several months away from generating those data.

As the CRISPR scientists try to decide on standard practices, Zhang said he'll keep looking to add to the genome editing toolkit.

"Nature has had many years to evolve new things," he said. "Looking at what nature has already invented is one approach; at the same time, taking a rational engineering approach like using crystal structures to guide a modification or mutation [in a nuclease] is complementary to exploring natural diversity. We take both approaches."

"We're all working very hard on developing these various systems, not just my lab, there's a whole field of talented researchers," Zhang said. "It's possible we'll have a toolbox of enzymes. So, depending on the specific genetic change we want to make, we'll go to the toolbox and pick the best one."