George Church has been a towering figure in genomics for more than three decades, dating back to the early days of DNA sequencing and the inception of the Human Genome Project. Since then, his interests and influence have spread to systems and synthetic biology, next-generation sequencing technology, and personal and consumer genomics.
Church has also had an important role in the launch of CRISPR technology, epitomized by a landmark paper in 2013 that marked one of the first demonstrations of CRISPR genome editing in human cells. His group remains among the most creative of research labs, pushing CRISPR into one scientific frontier after another, including multiplexing technology, de-extinction, xenotransplantation, and gene drive.
Church recently sat for an interview in his Harvard Medical School office, where he entertained questions posed by Kevin Davies, GEN’s editor-at-large. The interview—lightly edited for length and clarity—is presented below:
Davies: What are some of the key things that are going on in the Church lab at the moment?
Church: I think the theme is technology development—radical, enabling, transformative technologies, if we can. Most of them are paired with some really cool applications, a little out of reach with current technologies. We pair radical technology and radical application.
For example, Genome Project-Write (GP-Write) is something we’ve been working on for a long time, before it was even named that. The idea is to recode genomes, and the application is to make any organism resistant to all viruses, even viruses we’ve never seen before. That is recoding, which we’re doing from bacteria to mammals, humans, and plants.
Another is organoids. We make organs in pigs by making dozens of changes, maybe 80 different changes, in the pig genome, to make the organs more human-compatible and eliminate viruses. But we could also make organs from humans. And these are surprising me, how quickly all this is going, and in particular, how you can get definite insights, even diagnostically important insights, into very late-onset diseases, such as Alzheimer’s, schizophrenia, and bipolar disorder, in about two to four weeks, even though the onset should take 20 to 70 years. We’re making mostly brain organoids, but we have vascularized ones, so we are keen on getting good blood flow so we can make bigger organoids.
We’re doing aging reversal, and we’re doing gene drives, which allow us to cure diseases by dealing with carriers, the animal vectors. We’re also looking at other approaches to Lyme disease, including vaccines and people who are exceptionally resistant, analogous to what we’ve done with HIV.
Encoding data into DNA: The most recent breakthrough there is we encoded a terabyte of developmental data into mice. Every mouse in the colony encodes a fresh terabyte of data, and it takes only a billionth of the mouse’s body.
Davies: You first got into genome editing before you got into CRISPR, correct?
Church: Oh, way, way before. I got into recombinant DNA in the ’70s—our editing is an honorary member of the recombinant DNA family. I was one of the first employees at Biogen, which used recombinant DNA. As soon as I established my lab in 1986, we had two things—one was reading the genomes, the other was writing genomes. We did homologous recombination, which was precise editing.
Since then, we’ve gotten grade inflation and distortion of the word “editing.” But back in the ’80s, when we started working on editing, it meant doing exactly what you wanted—making exact changes, whether they’re big or small. Now it’s like, if you can mangle a gene—I call it genome vandalism—that counts as editing! Most people who are actually editing would be appalled by that definition.
Davies: Did the system described by Jennifer Doudna and Emmanuelle Charpentier in 2012, the system with the single-guide RNA, influence your thinking?
Church: That was extremely fundamental. I’ve said this in the past. CRISPR had been on our list, but Jennifer’s work bumped Sp-Cas9 way up on our list at that point. We ended up using a different guide than she used. . . . Feng showed in his [Science 2013] paper, side-by-side with ours, that her guide did not work in his hands.
We didn’t make a claim that it didn’t work. We just said, “Here is one that does work,” and that was the one that we published. It ended up sweeping through. Everybody ended up using that, at least for the S. pyogenes work.
Davies: You’ve said that CRISPR-Cas9 will not be the last word in gene editing. There’s a lot of excitement around base editing and prime editing. Do you feel fundamentally that is a better way of going about it?
Church: Definitely. We’ve been involved in the deaminases, the base editors, since before CRISPR. . . . David Liu’s group definitely took it places where we had not gone, and it was fantastic, especially the A-to-G base editor. We pioneer multiplex editing and [must consider the] consequent toxicity of CRISPR, whether it involves a base editor or not—almost every editor that everybody is using is highly toxic for multiplex editing, which is my obsession, multiplex everything—my middle initial “M”!
So, we did the study where we showed that we could get 26,000 edits. Our previous record was 62 edits in the pig, and we did 26,000 in human cells. And that required scrupulously getting rid of every source of nicks—not just double-strand breaks, but even nicks that you didn’t really think were nicks. . . .
All these damages are extremely toxic in a multiplex environment, which I think is where we are going for a lot of applications. It is just that people talk themselves out of it because it just seems like science fiction. But when you’re on an exponential, science fiction becomes science fact while you are blinking. We try to stay a little bit ahead, and so we did this toxicity study, which is now published in bioRxiv.
Davies: Clinical results for Victoria Gray, the first American patient dosed for sickle-cell gene editing ex vivo, look promising, and Editas is enrolling patients for its LCA10 trial. What are your hopes for the clinical application of CRISPR over the next few years?
Church: It’s great to see gene therapies making it to clinical trials—and that some therapies are emerging from Phase III trials victorious. Gene therapies hit a major speed bump in 1999, 2000. But now we’re seeing the field blossom. And the therapies are not just about subtracting a gene—CRISPR is mainly subtractive. Some of the therapies are about adding a gene, which could be missing or too low. The latter is something we are doing in our aging reversal studies—adding back genes that drop during aging.
My hope is to greatly lower the expense. I don’t want my legacy to be the most expensive drugs in history. We’ve brought down the price of reading human genomes from $3 billion now to $600, or even $0 [with Veritas Genetics and Nebula Genomics]. That is something I am proud of. I’m much more excited about that than I am about my contribution to expensive therapies.
A key, but underutilized, alternative to gene therapy is genetic counseling. If it’s really $0 for a whole genome sequence now, then as long as we address the privacy and utility issues, everybody could now get their genome sequenced and avoid a huge fraction of these expensive orphan drugs and gene therapies by instead using genetic counseling.
Davies: As CRISPR expands into the clinic, there seem to be three technical concerns: the off-target question, the antibody question, and the delivery question. Do they still pose major challenges?
Church: I think the off-target problem has been overstated slightly. I’m partially responsible, because academically, it is a really cool problem that we know how to solve. We and others have published a lot of papers on this. But even in the very first papers, we were getting off-targets that were close to the spontaneous mutation rate. If you’re going to worry about CRISPR, you should also worry about the spontaneous mutation rate—which is very, very low.
The immune question is hard to evaluate, since I think it is more applicable to CRISPR interference and regulatory mechanisms than it is to the hit-and-run of editing. It is “one and done”—that’s the beauty of it. It’s not like many drugs that you have to take for the rest of your life. From that standpoint, the immune question is exaggerated. We’re working on ways of making completely human-derived editors, for example, human integrases.
I think delivery is definitely the wave of the future. I have two startups just launched, one called Dyno Therapeutics and one called Ally Therapeutics, focused on reducing the immune response. Adeno-associated virus has already very low immune response, but you can take it a bit lower. The side effect of that is, you can get away with lower doses, which brings down the cost and the unintended negative consequences. We’re also working on ways that you can target it better. One of the nice things about CRISPR therapy for Leber congenital amaurosis 10 is, you’re not dosing the whole body, you’re just doing it in the retina. It is the first editing approved for use in vivo.
I would add to your list a fourth issue: CRISPR still does not do its job on target. I think its on-target sins are greater than its off-target sins. The base editors are not perfectly specific for a particularly base, and they’re limited to transitions, A’s to G’s and C’s to T’s. It’s very limiting. They’re not the Holy Grail, which is precise editing. I think we should hold that concept as our actual goal.
Davies: Do you oppose germline editing? When the technology is ready, and if society approves, what genes would be top candidates, beyond devastating disease genes?
Church: I’m not even sure about devastating disease genes. I’m not wildly enthusiastic about it. I’m not opposed to it. We should be focusing on outcomes, which is what you’re asking about, rather than methods. The outcome is, you don’t want there to be sickle-cell disease, cystic fibrosis, and thalassemia.
But as I mentioned earlier, most of these can be cost-effectively dealt with by genetic counseling. In vitro fertilization (IVF) is not pleasant. Whether you’re eliminating them by IVF selection or by IVF editing, you still have to do IVF, which is not a pleasant procedure. In fact, many religions would categorize nonimplanted embryos as murder. It is neither ethically nor medically pleasant. . . .
Most genetic diseases you could solve in an IVF clinic. If one of the parents is a double-homozygote for a dominant disease that maybe is partially penetrant, and they make it to reproductive age, but they’re not sure that their kids will, that would be one example. Infertility is another, where there is no other treatment. Germline does have three intrinsic advantages. First, it is better than other delivery systems at reaching all cells in the body. Second, after the first generation, it could be free rather than $2 million for somatic gene therapy. (So, like smallpox, an initial investment eventually results in equitable distribution if we work at it.) Third, germline goes through a single cell while most somatic therapies impact millions of cells, any one of which could become cancerous. In a single cell, such an event is, a priori, a million times less likely, and if it is derived from a clonal set of cells, it could be checked by whole-genome sequencing.
Davies: You told the Telegraph, “I just do not think that blue eyes and an extra 15 IQ points is really a public health threat or a threat to our morality.” But intelligence is such a polygenic trait—is this even worth discussing?
Church: This is a very interesting point. Many of my colleagues are dismissive, that such-and-such is not worth discussing because it is so far out of range. My experience has been things that we did not have to worry about arrive six decades early, right? The affordable genome was supposed to take six decades. It arrived in six years!
This is one of those things where I think it’s better to worry too much than too little. . . . Just because it is polygenic doesn’t mean it does not have a monogenic solution. For example, there are seven different clinical indications which are treated with somatotropin, human growth hormone—a single-gene product, in the midst of the most complicated genetics of almost any polygenic trait, which is stature. . . .
Furthermore, we’re getting better at polygenetic solutions. I mentioned earlier that we can do 26,000 [simultaneous] edits. That’s very different from solving a polygenetic trait, but it is telling us it is going faster than you might think.
Finally, with intelligence in particular, there are a number of mouse experiments that show that two or three genes have enormous impact on either specific tasks or general tasks that would be categorized as cognitive enhancement. So, does it help putting blinders on or sticking our head in the sand and saying, “There’s no way …”
But it is much more likely this is going to be debugged in adults than in the germline, because germline takes 20 years of debugging to figure out whether you’ve got a genius or not. The market is bigger for adults to become more intelligent and/or reduce cognitive decline, and the development cycle is faster. In principle, you could see an effect in weeks. Adult gene therapies are much scarier to me than germline, because you can debug them faster, and that will spread like crazy viral memes.
Davies: In 2017, a cover of Science showed the CRISPR pigs that had 25 endogenous retroviral knockouts. What is the next step?
Church: We’re preparing an article describing the rest of the wish list. [The article will identify the additional gene changes that will be needed to make pig tissues compatible with humans.] The full list [reflects the work of] everybody who has been working on [pig xenotransplantation] over the past 20+ years.
We were latecomers, invited by the pioneers, and we’re very indebted to them. When we published our first CRISPR article in 2013, they said, “Oh, that’s what we need,” because they’d originally hoped [that incompatibility was a matter of just] one or two genes, but then realized it was much more—on the order of 43 genes. We’ve now “engineered in” that full wish list.
We’ve already started primate preclinical trials with a nine-month survival so far. We employ an immunosuppression protocol that is completely compatible with current practice for human-to-human transplants.
Davies: You’ve joked about being pestered by journalists who ask about the woolly mammoth project. You went to Siberia last year. What was that trip like?
Church: It was overdue. We’ve had a very collaborative relationship with Sergey and Nikita Zimov, who are running the most likely initial location for our cold-resistant elephants, if/when we get them—Pleistocene Park. There are two sites now, one near Moscow, which is more convenient for wealthy and powerful Russians, and one in Chersky, which is where I went. It takes about 50 hours to get there! It’s a beautiful experiment and park; they really have established nearly all the megafauna they need (bison, elk, musk ox, etc.). They’re just missing a huge herbivore that likes knocking down trees.
In addition to evaluating the park’s progress, I also collected mammoth specimens to develop a new technology for analyzing genomic structure. All the ancient DNA people say you cannot get long-range structure, and I think you can. . . . Whenever I get to do experiments with my own hands, I’m a very happy guy. We dissected six mammoths … and [got] the samples back to the United States. We’ve done some testing. I’m not going to spoil that story but can say that we’re very excited about every aspect of the Siberia trip.
Published first in and adapted from The CRISPR Journal.