Genome editing is awash in acronyms. Revolutionary genomic tools allow scientists to sequence genes (NGS) and then to edit them (CRISPR, TALEN and ZFN) in many organisms by carving through DNA strands to delete a targeted sequence of the two base pairs: adenine (A) with thymine (T) and guanine (G) with cytosine (C). Cellular processes mend these DNA strands using the HDR pathway (low efficiency — less than 10% — but precise) or the NHEJ pathway (about five times as efficient but imprecise). Judicious genomic editing can disable trouble-making genes and proteins responsible for illnesses like sickle cell anemia or cystic fibrosis, or actively insert or enhance genes in humans and other organisms. A few drawbacks of current gene editing approaches? Cost. Off-target effects within cells, which are typically benign but can be dangerous. And cell death, particularly when many base sequences and genes are targeted through an exciting new frontier known as multiplexing.

Uber-innovator and genomic scientist George Church, PhD, is a professor of Genetics at Harvard Medical School and a professor of Health Sciences and Technology at Harvard and MIT. He helped launch the Human Genome Project in 1984, pioneered the field of gene editing in 1997 and founded the Personal Genome Project in 2005. Dr. Church answers our questions about genome editing and applications for organoids in biopharma.

Interview edited and condensed for clarity

What are the relative strengths and weaknesses of CRISPR-Cas9 versus newer approaches, and what do you anticipate that we will see in the biotech space?

Some of the newer approaches are variations. For example, you can strap a deaminase on to Cas9. It’s hard to say whether that’s Cas-9 or not at that point. It’s definitely not a nuclease anymore. It’s really radically different. With a nuclease, you’re hoping the cells will clean up the mess. With deaminases, you’re turning a single C base into a T, or A into G, but can make no other precise changes, so far.

What else is out there? There are so-called integrases and recombinases. These are used extensively in microbes and stand a good chance of replacing nucleases if they can be optimized for mammalian cells and be efficiently reprogrammed — thereby enabling precise swapping of your favorite gene sequence.

We’re trying to perfect methods to make lots of changes at once. You might want to make multiple changes to test polygenic risk — some human diseases involve dozens, maybe hundreds of genes — or try to slow down aging, or some effects of aging. We got started on this because we wanted to change the pig genome in dozens of places to make them more compatible to humans — better for transplants — but also to eliminate retroviruses built into the pig genome that would pose a health threat to humans if transplanted. When we eliminated the DNA nicking activity common to most genomic tools, then empirically tried different targets within repetitive elements, we found different cells had different outcomes, but some cells had as many as 13,000 edits.

As genome editing moves closer to the clinic, what kinds of safety studies will be needed?

Pretty much the same FDA approval process as other gene therapies and therapies in general: preclinical animal trials, then human phase I, II and III trials. There’s something to be said for a human cell or organoid phase, as well. I don’t know how this is playing out in clinical trials yet, but I can see it coming. Organoids are getting better fast; they can simulate almost any tissue. You can imagine this has some of the advantages of animal models and some of human trials without putting any humans at risk. Organoids and human-animal chimeras might be useful to get a better feeling for off-target effects specific to the human genome.

Can you share your thoughts about the promise for applications of organoids in biopharma and explain which aspects of the drug discovery and development process they might be useful for?

You can get tissues of a single cell type, you can get organoids with multiple tissue types — that’s next — and then you start getting systems of multiple organs. You could conceivably get a complex system that includes neuroendocrine cells, immune cells, liver enzymes, all this in one set of organoids.

Maybe you want to screen a library of drugs for a disease specific to particular tissue. So, you make a surrogate for that tissue, and you screen the library. You may want to do a tox screen, particularly on cardiac tissue. You might also want to see if the liver metabolizes the drug being tested into another version of the drug that could be toxic.

Or take a late-onset disease like Alzheimer’s disease. To perfectly reproduce it, you’d get patient cells, reprogram them and wait 70 years. But, there are subtle effects you don’t see in the clinic that you can see in the cells. Working with the Yankner lab, we found you can detect pathological effects as the organoid is developing in two to four weeks. That applies to Alzheimer’s disease and two other diseases that show late onset in humans and yet very rapid onset in organoids.

Continue the conversation with us @HMS_ExecEd or with Dr. Church @geochurch.

— Francesca Coltrera