This year has seen significant advances in the search for human gene editing of Mendelian disease. Back in April, a team from three major institutions in Seoul reported in Nature Biotechnology on the use of a recent CRISPR variation that does single-base-pair editing. Their proof-of-concept was the “Himalayan mutation“, an A-to-G switch in the tyrosinase gene that affects mouse pigmentation, and they got exactly the color mice that they were expecting. With this result in hand, they went on to a mutation that produces Duchenne muscular dystrophy: a premature stop codon in the Dmd gene was converted back to the codon for glutamine, allowing the gene to code for the missing dystrophin protein again. Intramuscular injections of the gene-editing system in AAV vectors (which had to be split due to the size of the needed proteins) caused about 17% of the muscle fibers to produce dystrophin, well above the level believed needed for therapeutic benefit.
And this week two papers have come out in Nature Medicine with further results. One is from Zürich (the university and the ETH) on correction of phenylketonuria (the well-known PKU) in adult mice. The second is from a team at Penn and Children’s Hospital in Philadelphia, also in mice, but targeting two different genes in utero. Both of these groups appear to have been very successful, and these results make it more likely than ever that we’re going to be able to reverse single-nucleotide genetic diseases in human beings.
The Zürich group also (like the Korean one) used an AAV viral vector targeting a base switch in the phenylalanine hydroxlase enzyme, again with an intein-split base editor system to get around the cargo capacity of that viral technique. (Split inteins allow you to “reassemble” a longer protein from two pieces; there are quite a few variations on the idea, and it’s a whole topic in itself). Intravenous treatment of the mutant mice with this system appears to have worked: the mice show reduced levels of phenylalanine in the blood, and mRNA from the liver shows corrected transcripts (in a dose-dependent fashion). The rate of correction increases over the weeks after injection – the highest dose group, at the 14-week time point, shows 63% corrected mRNA. The mice themselves start to put on weight normally and to grow normally pigmented fur (reversing two other consequences of the PKU mutation). A check for ten closely homologous DNA regions showed that none of them had been inappropriately edited.
Meanwhile, the Penn paper also edited hepatocytes, but this time before the mice had even been born. They used Ad adenoviral vectors rather than AAV (they didn’t go to the trouble of working out the split-intein system, and the Ad vector can carry more freight), so this is a delivery system may have to be modified for human use. But the proof-of-concept in mice certainly worked: they introduced a mutation into PCSK9 to try to bring on the low-cholesterol phenotype that’s noted in natural mutations in this gene. And that’s exactly what they got – interestingly, with this vector the prenatal-treated mice showed lasting effects, whereas a postnatal treatment group showed gradually decreasing ones. But the gene editing again seems to have been specific, and no effects were seen in the mothers, either. They also tried correcting hereditary tyrosinemia type 1 (a mutation in the Fah gene), and succeeded in this as well: normal mice resulted, with increasing correction over time as the normalized cells expanded compared to the impaired ones, and no evidence of the other predicted likely off-target edits.
Taken together, these studies have been extremely successful, from what I can see. The mouse systems studied here are directly analogous to the human diseases; you could hardly ask for stronger animal-model evidence that these techniques should be applicable in human patients. And the selectivity and overall lack of side effects are very impressive. There are a great many genetic disorders that fall into just the category that has been addressed here: point mutations that can be corrected by bringing in the guide-RNA targeted deaminase that does the base switch in these systems. Many of these are metabolic errors that can be addressed in the liver (where these two new papers are working). Eventually, despite the high barrier to trying it in the clinic, the prenatal route may well be the way to go: there are far fewer immune complications in treating the developing fetus, the diseases can be caught early by amniocentesis screening, and then can be corrected as early as possible in development.
This is going to be a fast-moving field, and in some cases it may push aside some earlier CRISPR human-therapy attempts. One way or another, this idea is heading into people, and it’s going to be doing so as quickly as scientifically and ethically possible. I hope that there aren’t any unexpected complications (because those could well be bad ones), but for now, the way looks as wide-open as anything like this will ever be. Here we go. It’s time to stop playing the genetic hands we’ve been dealt, and heal the sick.