New Cas9 Molecule Points the Way to Viral Delivery of CRISPR Systems

By Aaron Krol 

April 1, 2015 | Startup biotechs and major pharmaceutical companies alike are working overtime to apply the recently discovered CRISPR genome editing technology to treating or even curing human diseases. Derived from a natural defense mechanism of bacteria, the CRISPR system uses a class of proteins called Cas9 to engineer precise cuts in cellular DNA, making it possible for scientists to flexibly delete or insert genes nearly anywhere in the genome.

There are obstacles to using CRISPR in living organisms, however, and perhaps the biggest is the problem of delivering Cas9 to cells in the body. The most popular vector for gene therapies, in which new DNA sequences are inserted into the human genome to combat the effects of disease-causing mutations, has long been the adeno associated virus (AAV), which can deliver a genetic payload to a single, well-defined region of the genome without noticeable side effects. However, AAV vectors have been largely unavailable to CRISPR scientists, because these viruses can only carry payloads up to a certain size — and a standard Cas9 system, including the Cas9 molecule itself and multiple guide RNA sequences to direct Cas9 to its DNA target, comes in over the limit.

Today, a paper in Nature demonstrates a way around this limitation, successfully engineering genetic changes in mice using an AAV-delivered CRISPR system. The work comes from the lab of Feng Zhang, of the Broad Institute of MIT and Harvard, who first demonstrated that CRISPR genome editing could be performed in mammalian cells in 2013.

Almost all scientists using CRISPR today choose to use a Cas9 molecule derived from the bacterium Streptococcus pyogenes. “The reason the S. pyogenes has been used as the standard is because it was the only Cas9 that was very reliable, in terms of efficiency, and also had a broad targeting range,” explains Le Cong, co-lead author of the new paper and a postdoctoral researcher at the Broad Institute. (His fellow lead author, Fei Ann Ran, presented some results from this paper at the Molecular Medicine Tri Conference this February.)

The efficiency refers to how reliably the Cas9 molecule cuts both strands of DNA at a targeted site, while the targeting range refers to how many potential regions of the genome the molecule can interact with. All Cas9 molecules need to make cuts next to specific PAM (protospacer-associated motif) sequences, but the S. pyogenes Cas9 uses the PAM sequence NGG, which is so common across the genome that it can in effect be targeted almost anywhere.

However, S. pyogenes Cas9 (or spCas9) was not always the default choice. Cong, who received his PhD under Zhang and was a lead author of the original paper demonstrating CRISPR genome editing in mammalian cells, recalls that this early work used Cas9 derived from Streptococcus thermophilus instead. Cong and his colleagues reasoned that another Cas9 molecule from another well-characterized species might be suitable to use with an AAV vector. “We wanted to find a new Cas9 that is of equal efficiency and targeting range [to spCas9], but smaller,” he says.

To find candidate molecules, the lab partnered with Eugene Koonin of the National Center for Biotechnology Information, who studies the evolution of Cas9 molecules across species of bacteria. Koonin, in collaboration with the Zhang lab, had built phylogenic trees tracing the relationships between distantly related Cas9 molecules. From a set of over 600 molecules, the lab ultimately chose six new ones for detailed screening.

“The rationale was, if we could only sample a few Cas9s, we wanted to sample representative ones across the evolutionary space,” says Cong. After initial in vitro tests of these molecules in human kidney cells, the authors found that one of them, derived from Staphylococcus aureus, was comparably efficient to spCas9 This new saCas9 was a promising candidate for in vivo experiments: at just over 1,000 amino acids in length, it is short enough to include in an AAV cartridge, and its PAM sequence, NNGRRT, is fairly well represented across the genome.

To test the system in animals, the lab engineered AAV vectors with saCas9 systems to remove two different genes from two different groups of mice. One experiment targeted the APOB gene, which encodes a protein involved in lipid transport, while the other targeted PCSK9, a gene with major therapeutic implications because people who do not express this gene have markedly low levels of LDL cholesterol and lowered risks for cardiovascular disease. (Both of these genes are especially active in the liver, important because AAV is relatively easy to deliver to liver cells.)

Both experiments were highly successful: in the PCSK9 knockout mice, over 40% of liver cells showed deletions of the targeted gene within one week of a single treatment. Even more significantly, in both populations of mice, the researchers were able to detect phenotypic changes as a result of their treatments.

“To achieve phenotypic changes, and demonstrate a physiological change in the animal, you have to be able to hit many cells,” says Cong. “It requires a very efficient and scalable system.” The AAV-delivered saCas9 system appears to meet this high bar: in the APOB knockout mice, telltale accumulations of lipids could be found in the liver, while the PCSK9 knockout mice had blood cholesterol levels 40% lower than wildtype mice and almost nonexistent blood levels of the Pcsk9 protein.

One surprising result of these experiments was that the DNA of the affected mouse cells appeared to contain very few off-target cuts, where Cas9 sheared the genome in places other than the targeted region, even weeks after treatment. “We didn’t observe any off-target cleavage above the level of detection of our assay,” notes Cong, who says this result was unexpected. Earlier in vitro experiments had suggested that saCas9, like spCas9, would cause low but detectable numbers of off-target cuts, which could be a serious concern in human therapies, potentially leading to difficult-to-predict side effects.

Cong speculates that the AAV vector, which by design delivers only a limited payload of Cas9, may have contributed to the apparent absence of off-target effects. However, he cautions that more research is needed before anyone can rule out off-targets with this method. “We think in vivo specificity is a very complicated issue,” he says. “Longer-term study is necessary to really assess the long-term toxicity, or off-target effects of the system.” Importantly, cells that have received AAV payloads will continue to produce Cas9 throughout their lifespans, meaning there is a potential for off-target cuts to accumulate over time.

SaCas9 is a promising tool for therapies based on CRISPR — so much so that Editas Medicine, a therapeutics company to which Zhang serves as a scientific founder and adviser, is already experimenting with this new molecule. However, Cong notes that this paper only scratches the surface of Cas9 diversity in the natural world.

“This is a first step, but there are so many other Cas9s out there in the metagenomic space, we expect the diversity of the Cas9 system to have even more versatile and potentially better Cas9 systems,” he says. “And we can also rationally design a Cas9 system, now that we have knowledge of multiple Cas9 systems instead of just one. Metagenome mining and rational design are really important now to engineer and improve Cas9.”