NEW YORK (GenomeWeb) – Every new study that is published about CRISPR-Cas-based genome editing seems to add yet another possible use to the growing list of what could be done with the technology: therapeutics for a variety of diseases, administered both in vivo and ex vivo; gene drives to control the spread of disease; agricultural applications to increase crop yields and create healthier animals, and so on.
But there's always one question that lingers, especially when it comes to using CRISPR for medicinal purposes — how do you shut it off?
The answer to that question lies in the same biological chess game being played between prokaryotes and viruses that resulted in the development of the CRISPR-Cas immune system in the first place. Viral phages have developed anti-CRISPR proteins in an evolutionary countermove to defend themselves against destruction by CRISPR systems when invading bacteria.
Your move, phages
Though some phages can develop mutations to get around bacterial CRISPR systems, bacteria evolve rather quickly to acquire new spacer sequences, which then allow them to retarget the mutant phages. Rather than just conferring mutations onto phages to allow them to escape CRISPR-mediated destruction, anti-CRISPR proteins work in a variety of ways to stop CRISPR's genome editing function altogether, either by disrupting DNA binding or by inhibiting the cleavage of target sequences.
The first anti-CRISPR proteins were published in 2013 in Nature by the University of Toronto's Alan Davidson and his team. Since then, dozens more have been found, leading Davidson and his former postdoc Joseph Bondy-Denomy to start a database in order to keep track of the proteins' names, which CRISPR systems they inhibit, and what their sequences are.
Bondy-Denomy, who is now a principal investigator at the University of California, San Francisco, said that the database has grown to include almost 50 different anti-CRISPR proteins.
The first CRISPR inhibitors that were discovered were for the Type IF system, which includes the Csy 1, Csy 2, and Csy 3 CRISPR nucleases. But as those proteins aren't widely used (other than by Pseudomonas aeruginosa), researchers quickly moved on to see if they could find inhibitors for more commonly utilized enzymes such as Cas9 and Cas12.
Indeed, Bondy-Denomy said, a number of different research teams have found Cas9 anti-CRISPRs, resulting in about 11 known inhibitors for Streptococcus pyogenes Cas9 (SpCas9), one of the most commonly used orthologs of Cas9. The most recent anti-CRISPR was discovered by a team from the Fred Hutchinson Cancer Research Center, which used a functional metagenomics-guided discovery approach and recovered acrIIA11 from a Lachnospiraceae phage, among the strongest known SpyCas9 inhibitors to date. Further, according to Bondy-Denomy, five inhibitors of Type IIC Cas9 enzymes have been found so far. This enzyme is also used in gene editing.
And in 2018, Bondy-Denomy's team, along with researchers from Jennifer Doudna's lab at the University of California, Berkeley, reported in Science their discovery of 15 new CRISPR inhibitors, including the first inhibitors for the Cas12a enzyme.
Interestingly, some research teams have also recently endeavored to create synthetic anti-CRISPRs or to search for small molecules that achieve the same purpose as the naturally occurring inhibitors. In December 2018, Ohio State University researchers developed synthetic oligonucleotides that inhibited the ability of Cas12a to induce double-strand breaks in DNA in a time-dependent and sequence-independent manner. As they wrote in Cell Reports, the OSU investigators constructed CRISPR RNA (crRNA) variants consisting of chemically modified DNA-crRNA and RNA-crRNA duplexes and found that phosphorothioate-modified DNA-crRNA duplex completely blocked the function of Cas12a.
And last week, a team led by Broad Institute Associate Member and Brigham and Women's Hospital Associate Biologist Amit Choudhary reported in Cell its development of a high-throughput platform for identifying small molecules that can disrupt the genome editing activity of CRISPR-Cas nucleases.
The researchers developed a suite of high-throughput assays to measure SpCas9 functions, including an assay that screens for SpCas9 binding to the protospacer adjacent motif (PAM), and used them to screen a structurally diverse collection of natural-product-like small molecules. The ultimate goal was to identify compounds that disrupt the interaction between SpCas9 and DNA.
Indeed, they identified the first synthetic small-molecule inhibitors of SpCas9. The inhibitors weigh less than 500 daltons — about 500 hydrogen atoms, in essence — and are cell permeable, reversible, and stable under physiological conditions, the researchers noted.
How do they work?
But finding the inhibitors is only a small part of the problem. Researchers are also trying to figure out exactly how they work to inhibit Cas-mediated gene editing, how those mechanisms could be harnessed to counteract CRISPR-based technology, and whether the anti-CRISPRs themselves could have any deleterious effects akin to the off-target effects of Cas enzymes.
In the simplest terms, the inhibitors that correspond to specific CRISPR enzymes cut off the editing activity of those enzymes. That, in turn, contributes to a curtailing of the off-targeting activity of the Cas enzymes themselves, as lingering editing activity tends to contribute to a higher off-target rate.
"A simple way to look at it is that usually when you have an enzyme in a substrate pair, the substrate is present in much larger quantities than the enzyme. But when you are doing genome editing, it is flipped and you have very few copies of the substrate," Choudhary said. "When you have that kind of situation, that leads to these off-target effects where the enzymes have found the substrate but there's still a lot of enzyme hanging around and it's not a happy situation." Anti-CRISPR molecules could be used to ameliorate those situations.
But there's still research to be done in the lab to understand the underlying biological mechanisms of how these inhibitors function.
"There's a lot of mechanistic work going on behind the scenes to understand how they all work," Bondy-Denomy said. "There haven't really been any sort of omics studies on anti-CRISPRs in human or bacterial cells so we're doing a lot of pull-downs now in the lab to see how they might bind to bacteria."
For example, he and his team have found that anti-CRISPRs can multiplex themselves to create inhibitory complexes that fight CRISPR systems in multiple ways within the same organism. In the same Science paper where he and his colleagues reported the discovery of Cas12a inhibitors, Bondy-Denomy also found single prophages encoding multiple types of anti-CRISPR in response to the multiple types of CRISPR within the Bordetella bacterium.
"I actually think multiplexing is happening in nature. For example, in Bordetella prophages, we found Type IC, Type IF, and Type VA inhibitors in a related prophage family. So, these prophages are encoding multiple inhibitors of these distinct systems," he said. "I think that's an example of a natural multiplexing of both CRISPR and anti-CRISPR."
Researchers have also found that anti-CRISPRs are roughly 10 times smaller than the Cas protein they inhibit. So, if there's an effective strategy to deliver a CRISPR-Cas editing system, along with its guide RNA, to cells in the lab or cells in an animal, the same strategy could be used to deliver anti-CRISPR proteins five or six different anti-CRISPR proteins, according to Bondy-Denomy.
"You could really dampen down a Cas protein's activity by blocking it in two or three different ways," by adding multiple anti-CRISPRs to the compound, he added. And if several Cas enzymes are multiplexed together to create a specific reaction, their inhibitors could also theoretically be multiplexed into the CRISPR-Cas-gRNA compound to stop the reaction when needed.
So far, he noted, his research hasn't indicated that anti-CRISPR proteins would do anything other than inhibit CRISPR-mediated gene editing. "Of course, we've been very focused on its role for inhibiting CRISPR so I wouldn't say that that's been comprehensively addressed," he added. "But the off-target activity of CRISPR nucleases is many-fold because it's driven by the fact that they bind to guide RNAs, that they cut DNA, and that they have a base-pairing potential with the risk of [causing] point mutations. Anti-CRISPRs have none of these problems."
That could be because anti-CRISPRs aren't nucleases, for the most part, and they bind to the Cas protein with very high affinity, according to Bondy-Denomy. So, if the Cas protein target is present in the cell, the inhibitor will likely bind to it "and not get too distracted," he said, adding, "Of course, biology can always throw new things your way. So, we'll wait and see what comes out of some unbiased approaches to see what anti-CRISPRs bind."
Choudhary and his team are also looking into whether the small molecules they've found could have unintended effects if used as inhibitors in human or animal cells. Broadly speaking, he noted, the small molecule he described in his paper was very specific and only inhibited the editing activity of SpCas9. In fact, when added to a Cas12 compound, it didn't inhibit that nuclease's editing activity.
"I think we do need to start evaluating all of these molecules and their ability to bind to other nucleases or other molecules," he added, whether those are naturally occurring anti-CRISPRs or synthetic inhibitors. "If you look at anti-CRISPR proteins, they are viral proteins. So, of course, there's a concern of immunogenicity. But also, these proteins may bind to other nucleotides in the cell, and so a systematic study of off-target analysis has to be done for all of these anti-CRISPR molecules."
The 'off' switch
But even before the mechanistic work is complete on these inhibitors, there's already a discussion of how they could be used in therapeutic CRISPR applications. They could be administered to a patient at a certain time point after the application of a CRISPR therapeutic to stop the Cas enzyme's editing action. They could even be attached to the CRISPR-Cas compound directly and developed to act in a time-release fashion after a certain amount of time has passed, though that would require more extensive protein engineering.
Either way, Bondy-Denomy said, the need for such an inhibitory mechanism for therapeutic purposes has come into sharp focus in recent months. In November 2018, Editas Medicine announced that the US Food and Drug Administration had accepted the company's Investigational New Drug application for EDIT-101, an experimental CRISPR genome editing medicine being investigated for the treatment of Leber congenital amaurosis type 10 (LCA10).
LCA10 is a disease of the eye caused by mutations in the CEP290 gene. In January, the company published a study in Nature Medicine showing the effect of EDIT-101 in humanized mice. Importantly, the data showed that a single administration of the Cas9 compound led to Cas9-guide RNA and mRNA expression of the Cas9 protein up until the last timepoint recorded in the study, which was 40 weeks after the mice were injected.
"In the eye this is probably OK, because it's sort of an isolated site. But imagine if we're going to be administering an AAV-encoding Cas9 systemically to people. These AAVs are meant to pump out gene products for a really long time because they were developed for gene therapy. So, if you have a Cas9 protein that's being over-expressed for a long time, even if you try to dial down the promoters, you need to have a mechanism, I think, post-translationally to prevent cutting of any off-target DNA," Bondy-Denomy cautioned. "And I think I think that's something that's overlooked in this field. All the off-target edit work is usually a short timeframe."
He also noted that recent research has shown that the most widely used orthologs of Cas9, which are derived from Staphylococcus aureus and S. pyogenes, may provoke immune responses in some people given that these two bacterial species infect humans at high frequencies. In a study in Nature Medicine in January, Stanford University's Matthew Porteus and his colleagues demonstrated that there are preexisting humoral and cell-mediated adaptive immune responses to Cas9 in humans, which they said should be taken into account as CRISPR-based therapeutics move toward clinical trials.
"We're starting to appreciate how much of an immune response there is to Cas9, so having an inhibitor that could turn it off could really be an ideal reagent," Bondy-Denomy said.
There's also potential for small molecule inhibitors of Cas enzymes to be used as dose regulators in CRISPR-based therapeutics. "One of the hallmarks of a therapeutic agent is dose control — dose makes the poison," Choudhary said. "It's a very fundamental attribute of a therapeutic agent. So, if we can develop molecules that are very good at providing that dose control, then I think those molecules will have utility."
He further noted that the small molecule inhibitors could reduce the threat of immunogenicity because they can be used to create proteolysis-targeting chimeras (PROTACs) — essentially, Cas9 shredders.
PROTACs are heterobifunctional small molecules that contain a target-protein binder and a ubiquitin-ligase binder joined by a linker, Choudhary and his colleagues wrote in their paper. They noted that they could form a PROTAC by joining their SpCas9 small molecule inhibitor to the ubiquitin-ligase binder, and that this PROTAC should theoretically recruit ubiquitin ligase to SpCas9, promoting ubiquitination and proteasomal degradation of SpCas9.
"Once you have identified inhibitors, you can start applying these really cutting- edge chemical biology principles and convert those inhibitors to shredders," Choudhary said. "The shredding can be very useful, particularly if you have an adverse immune response and you really want to get rid of the protein as soon as possible." Such a mechanism could act as a safety trigger for a possible therapeutic if a patient had an adverse reaction.
Anti-CRISPRs could even be used to control the spread of gene drives, though that mechanism would likely work in a different way than in a therapeutic, according to Bondy-Denomy.
Rather than delivering a therapy to a human who would not be encoding Cas9 or an anti-CRISPR in their genome, gene drive technology is based on the idea of transgenic animals — mosquitos, for example —encoding Cas9 in their germline genomes and then passing those changes down to their progeny.
In order for anti-CRISPRs to be deployed as an anti-gene drive mechanism, they would likely have to be released a parallel or reactionary measure to a gene drive already taking place, "where we're post-releasing transgenic animals like transgenic mosquitos that encode anti-CRISPRs in their genomes to protect them from the gene drive," Bondy-Denomy speculated.
"I think that's a very cool area, the potential for inhibitors to block an existing gene drive because we might not have the ability to tinker with the promoter. If we have a gene drive happening and we need to stop it, that's where these post-translational inhibitors really could come in handy. But I think the only way that would work is that we're not spraying anti-CRISPR on mosquitos, but we're releasing mosquitos that have the ability to produce that gene from their own genome."
Naturally occurring vs. man-made
As the research progresses on the various classes of inhibitors, however, some may question whether there are advantages to using naturally occurring anti-CRISPRs over small molecules, or whether small molecules are more suited to certain purposes than naturally occurring anti-CRISPRs.
As far as Bondy-Denomy is concerned, anti-CRISPR proteins may have the upper hand as modulators of CRISPR-based therapeutics because they're certain to turn the Cas enzyme off.
"I think that the ability to genetically encode inhibitors is a big advantage [in therapeutic applications] because we can program where it goes, which cell types it's expressed in, and where it's used. And we can do something like gene drive inhibition where we can make transgenic animals," he said. "I think that anti-CRISPRs have been fine-tuned over millions of years to harness specificity."
He also noted that a deeper understanding of naturally occurring anti-CRISPRs could lead to researchers being able to create better synthetic CRISPR inhibitors. "For us, we're really focused on the natural biology of CRISPR. We're not focused on inhibiting CRISPR per se — we're really focused on how phages inhibit CRISPR," he said. "But we are very keen to understand where anti-CRISPRs have evolved from and I think there's a really important role there in asking if we can create them. Can we create proteins that inhibit CRISPR and how can that help teach us where they might have come from or how they might have evolved? That's something I'm very interested in."
For Choudhary's part, he does believe that the small molecule inhibitors have several advantages over the phage anti-CRISPRs. For one thing, there's their size. The SpCas9 inhibitor that his team's study focused on weighed in at 453 daltons. That's roughly 400 times smaller than Cas9. And given that most anti-CRISPR proteins weigh in at 10 to 15 kilodaltons, the small molecules have a significant size reduction advantage over the already-tiny anti-CRISPRs.
The small molecules are also able to diffuse into the cells without the need for delivery. "So, you just add these compounds to the media and they automatically get into the cells. And so, another advantage is that when you remove the media, you can essentially reverse it," he said. "If you have an anti-CRISPR protein, first you'd have to deliver a Cas9, and that does stress the cells a little bit. And then you had to deliver the anti-CRISPR again. So it adds to the complexity of two deliveries: one to activate Cas9, the other to control Cas9."
One disadvantage to the small molecules may be their relative lack of potency. Anti-CRISPR proteins tend to be highly potent, binding to the target enzyme and shutting it off. The small molecules have more of a gradient effect, according to Choudhary, tempering the effect of the Cas enzyme rather than switching it off completely.
But that could also be an advantage rather than a fault. Certain therapeutics may require dose modulation rather than total cessation, in which case, a small molecule inhibitor could be preferable to an anti-CRISPR. A lot of additional research needs to be done before any of these questions can definitely be answered, Choudhary said.
Move-countermove
Despite their seemingly brilliant tactic to stave off destruction, viral phages can't quite declare checkmate against prokaryotic CRISPR systems. Bacteria have evolved over millennia to survive just about anything that nature throws at them — some bacterial CRISPR systems have started to develop what Bondy-Denomy has loosely termed anti-anti-CRISPR systems to counteract the anti-CRISPRs from viral phages.
In research he recently conducted with Toronto's Davidson, he characterized phage and host proteins that antagonize anti-CRISPR deployment. The collaborators have submitted one paper on the subject and are currently authoring a second.
"They're not anti-anti-CRISPRs in the sense that they are proteins that bind to the anti-CRISPR. Anti-CRISPR proteins are so diverse that it actually would be really challenging for bacteria to have a whole arsenal of inhibitors for the inhibitors. What we found instead is that they basically utilize transcriptional regulators to inhibit the deployment of the anti-CRISPRs because the promoter sequences that drive their expression are so highly conserved," Bondy-Denomy explained. "So, you could think of that as the Achilles heel of the anti-CRISPR system, that they have these absurd regulatory regions. And so, the strategy seems to be preventing transcription of the anti-CRISPR region."
For his part, Choudhary is working not only on fully characterizing the small molecule he and his team have already identified, but also on validating additional small molecule inhibitors identified by his high-throughput platform. He noted that his colleagues will publish additional papers in short order on Cas-specific small molecules that they have identified, adding that they are also developing broad-spectrum inhibitors that could work on a variety of enzymes at once.
In the viral anti-CRISPR world, there are still a lot of empty spaces in Bondy-Denomy's database, which he hopes other CRISPR researchers will help to fill out. And although it might be possible that there are CRISPR enzymes that don't have a corresponding anti-CRISPR, he doesn't think it's very likely.
"It's a possibility of course — anything is possible in biology. I doubt it, though. These are just proteins. They're proteins that bind to a guide, and then they bind to a target nucleic acid. There's nothing magical about them," he said. "And now, because we've found so many as a community, my default assumption is that they exist for every system. It would take some sort of extravagant explanation about why they might not exist. If there's something special about the mechanism of the CRISPR system, or maybe it doesn't target phages and it does something else, then maybe there aren't inhibitors for it because it performs a different function. But if we're talking about phage-targeting CRISPR systems, then there are definitely inhibitors for them."