A decade ago, Ying Zhang found herself on the cusp of a medical revolution.

CRISPR crops are coming to Europe — why bolder would be better

Zhang was one of the first scientists to be hired by CRISPR Therapeutics in Zug, Switzerland. The biotechnology firm was formed in 2013 to treat disease using CRISPR–Cas9 gene editing. The technique had only recently been published, and Zhang and her colleagues were dreaming up ways to put the innovative tool to use.

But she soon grew frustrated. Many of her plans would require replacing an entire faulty gene with a healthy copy — and that, she quickly learnt, can be a challenge with CRISPR–Cas9. “When we first got there, we were full of excitement,” Zhang says. “And when we actually did all the work, we found it’s more complicated than we thought.”

Today, it’s easier. Now at Wuhan University in China, Zhang and others in her field are developing a fast-growing toolkit for rewriting genomes, whether for therapeutics or basic research. Some methods build on CRISPR–Cas9; others have broken free of it entirely. “It’s been an insane explosion of new technologies,” says Shannon Miller, a bioengineer at Scripps Research in La Jolla, California.

It is still early days for these techniques, and each has its pros and cons. None has demonstrated the simplicity and versatility that CRISPR–Cas9 brought to more-modest DNA-editing tasks. Which method to choose depends on the size of DNA that needs to be inserted, where in the genome it needs to go and which cell type is being modified (see ‘A short guide to big gene editing’). “There are so many different flavours,” says Amy Pooler, head of neurotherapeutics at Regeneron in Tarrytown, New York. “It’s not going to be a one-size-fits-all approach.”

Choosing the right tool for the job

Convenient and versatile, CRISPR–Cas9 gene-editing technologies truly shine when they are used to make small changes in the DNA sequence to abolish a gene’s function. A related and more precise technique, called base editing, harnesses the CRISPR machinery to change individual DNA ‘letters’.

Both techniques have been used to treat disease in people, but they have limitations. CRISPR–Cas9 gene-editing tools must break DNA before they can alter it. The Cas9 enzyme slices both DNA strands at the target site and researchers rely on a cell’s natural — and error-prone — repair machinery to seal the gap. The outcomes of these edits can be unpredictable and often disable the gene.

And base editing must be tailored to an individual’s DNA, making it too precise to be practical for some diseases. Stargardt’s disease, which is a heritable form of vision loss, for instance, can be caused by any of more than 1,200 variants of the ABCA4 gene, about 750 of which could be corrected using base editing. But hundreds of distinct base-editing therapies would be needed to treat them all. “It’s not realistic to develop a drug that only has one patient,” says Zhang. “If you can install the entire healthy gene, you can at least cover one genetic disorder with one treatment.”

Epigenetic editing makes its mark

Some researchers hope to use large gene insertions to endow cells with fresh functions. The technique could, for instance, give chimeric antigen receptor (CAR) T-cell therapies the ability to produce immune-stimulating proteins or genetic switches that activate the cells only when they are near a tumour.

Moreover, scientists doing basic research can manipulate large DNA segments to create improved models of human diseases, reproduce metabolic pathways, probe genomic structure–function relationships and build synthetic chromosomes.

However, CRISPR–Cas9 isn’t always the right tool to accomplish these goals. It can insert DNA, but its efficiency decreases as the segment gets larger so researchers typically use it with DNA segments that are shorter than 2,000 base pairs. The technique can also be difficult to use in cells that are not actively dividing, such as neurons. Another CRISPR-based technique, called prime editing, can insert sequences without breaking both DNA strands but it can accommodate only a few hundred bases. “This really is a bottleneck for the field,” says Bin Liu, who develops gene-editing techniques at The Ohio State University in Columbus.

A small package

To resolve that bottleneck, researchers have been combining elements of CRISPR gene-editing tools with other enzymes capable of suturing large DNA segments into genomes. For example, genome engineers Omar Abudayyeh and Jonathan Gootenberg at Harvard Medical School in Boston, Massachusetts, have developed a method that harnesses an enzyme capable of stitching tens of thousands of DNA bases into a genome¹.

That enzyme, called a large serine integrase, inserts DNA at specific sites without the need for breaking the two DNA strands. Abudayyeh and Gootenberg use an integrase called Bxb1. They use prime editing to introduce the sites (called landing pads) and a Cas9 protein coupled to Bxb1 to insert the DNA. The team named the technique PASTE.

Researchers at the now-closed company Tome Biosciences in Watertown, Massachusetts, which was co-founded by Abudayyeh and Gootenberg, used a variation of PASTE in cynomolgus monkeys (Macaca fascicularis) to replace a gene that has been implicated in haemophilia B, a human blood disease². About half the animals’ liver cells contained the corrected version after a single treatment, the researchers reported.

But PASTE requires several elements: a modified Cas9 enzyme that cuts, or nicks, one DNA strand at the target site; the landing-pad sequence; a reverse transcriptase that copies the landing pad into the genome; the integrase; a guide RNA; and donor DNA.

To get all of those elements into the cells in the monkey experiments, the researchers used two delivery methods: lipid nanoparticles carried messenger RNA encoding the enzymes and the guide RNA; and a virus called an adeno-associated virus or AAV contained the donor DNA and landing pad sequence. “Those complexities add up,” says Abudayyeh. “We got 50% in the liver, but targeting any other organ becomes quite challenging.”

Other methods are even more complex. Take, for instance, mobile DNA elements called transposons. With help of enzymes called transposases, many transposons can insert themselves into DNA with seeming randomness. But CRISPR-associated transposases (CASTs) can target a specific site in the genome if provided with a guide RNA and donor DNA that has transposon sequences on either end.

CRISPR’s next act: the companies editing the epigenome to treat disease

Sam Sternberg, a biochemist at Columbia University in New York City and chemical biologist David Liu at the Broad Institute of MIT and Harvard in Cambridge, Massachusetts, have optimized one such system for use in human cells. The system, called evoCAST, can insert up to 15,000 DNA bases into human cells, including at several therapeutically relevant sites in the genome³.

But some evoCAST designs involve six or seven proteins — a heavy lift for current delivery methods. “One big focus has been streamlining the number of components into as few molecules as possible, and working on different ways to deliver them,” says Sternberg.

By contrast, Miller is turning to ‘recombineering’ — a technique based on genetic recombination used to rewrite the genomes of bacteria. With about 1,500 base pairs (not counting the donor DNA), her system can be packed into a single AAV genome, depending on the size of the donor DNA⁴. Miller and her colleagues are trying to boost the system’s efficiency in human cells.

Another class of mobile genetic elements, called R2 retrotransposons, can be delivered entirely as RNA molecules packaged inside a lipid nanoparticle, which can carry much larger cargoes than AAVs can.

But targeting retrotransposons to distinct genome regions requires re-engineering the retrotransposon’s DNA-binding domain — a much more challenging task than altering a guide RNA. As a result, researchers usually stitch the donor DNA into a predetermined ‘safe harbour’ region of the genome where it is unlikely to cause harm.