Future, Human, Nature: Reading, Writing, Revolution - An Evening with Dr. George Church


George Church

video: https://vimeo.com/209623759


Thank you for coming. I think we're going to have a very interesting evening tonight. I wanted to start off by telling you a little bit about the origin the idea for this talk and what it's leading to over the next couple of days. We had a very wonderful interaction myself Jennifer Doudna with Bill from Stanford and we realized there was an opportunity to get together and discuss the ethics of genome editing and in particular human genome editing. We wanted to start this off with a very public lecture with someone who represents a real leader and thinker in the field. We were fortunate to have sponsorship from the Templeton Foundation for this. We are grateful to them. It's a great pleasure and honor to introduce our keynote speaker for this evening, George Church.

Let me tell you somewhat about George, although he's someone who doesn't need an introduction but there's really a lot to say about him. He received his PhD at Harvard in 1984 and then he joined the Harvard faculty in 1986. I met him around that time when I was a graduate student in a very different field. You couldn't help but be aware of George and his big thinking and incredible ideas. In those days, he was starting work on DNA sequencing and thinking about how to analyze organisms from the level of the whole genome. His name has become synonymous with the field of genomics. He pioneered the first direct sequencing method. He has made many incredible developments and discoveries especially for genome engineering and genome manipulation. He has advised the government on policies in biosafety, precision medicine, and he has authored over 400 papers, and a book Regenesis which a number of us were fortunate enough to have him sign for us tonight. I think he's someone who has been a real inspiration ... and that's really the topic for us tonight. I am excited about his title, which is "Future, Human, Nature: Reading, Writing, Revolution". I know it's going to be very interesting. George, we are delighted to have you here.

Well, it's great to be here. This is my conflict of interest slide. (laughter). It's also on my website. Transparency is an important part of we technologists who take an interest in the ethics side of these things. I just want to make a special thank you to this organization, pged.org which was initiated by Ting Wu a close colleague and my wife, in 2005. This is when we were at the Office of Science and Technology Policy giving a briefing, not lobbying, giving a briefing on genetics which we do on a regular basis. pged also teaches high school teachers and they visit especially under-represented and underserved communities throughout the United States mainly. They work with screenwriters for TV and films to try to get the facts straight. And this is a picture of map-ed where people will take a test and show that they have some knowledge of genetics. Jennifer was at the congressional briefing and we had the pleasure of being with her when she got her award in Canada, which she then donated to pged.org, so we are grateful to that.

I will start with some audience participation. I'm going to take a guess that most of you think that human genomics is not useful. I will ask you to raise your hand if you have your human genome sequenced. You don't have to have it with you, but if you've had it sequenced, raise your hand. Oh, that's quite a lot. I don't want the whole sequence... I'll settle for 95%. The dirty little secret is that nobody has ever had their genome completely sequenced. So if I am going to make the argument that it might be useful, I'll make the argument later on that you need to know your genome in order to edit it. This might sound shocking to some of you. With preconception genetic counseling, this does not put embryos at risk, you could diagnose serious diseases. These are highly predictable and actionable in that you could marry someone based on that information. That might seem unromantic to some of you-- so there are in vitro fertilization diagnostics and pre-natal non-invasive testing and so on. There is newborn testing, which many people are shocked to find out that their children and sometimes they were genetically analyzed for up to 40 diseases that are actionable, usually it's something like dietary. Then there's genetic odyssey where you have a child with a development delay and you go from physician to physician trying to figure out what went wrong, and very often the parents blame themselves, just finding out what went wrong-- finding an explanation in the DNA is a great relief to many parents even if there's no cure, it means they didn't cause it. So the list goes on. I think there's an opportunity here for preventative medicine. Very often we talk about the cure for this, the cure for that. Well what about prevention?

Let me give just two quick examples. You'll notice that I show pictures of patients. I name their names. You'll think, how can he be talking to us about ethics if on his first slide he's showing John Lauerman as a patient? Well, we have a study where we are sure they are educated. If you participate in medical research, your data might become identified. In fact, whether you are in a medical study or not, your medical records are probably in someone else's hands. Your medical records are worth 20x on the black market than your credit card. If you're interested in that, I can explain why. In this case, John Lauerman... he had a JAX2 mutation. It was not germline, it was in his blood. It was reported that he was healthy, but we found out that he had sciotmas in his eyes and some leg pain, and now he takes aspirin for life. Not too bad.

Here's another one, Angelina Jolie... she did a wonderful thing for the world by being open about her BRCA1 genetics and talking about the decision about whether to trust the genetics and take a preventative medicine. A lot of people think that Angelina Jolie had a lump or some positive result in an x-ray. This was entirely based on her genome. She was said to have an 87% chance of breast cancer before she had surgery, and less than 5% chance afterwards.


  • 1974 Scientist = Enthusiast -> no. ELSI 36 articles. Polarized -> win win
  • 1974 GMO -> 2014 biocontainment
  • 1977 Equality, Wealth-Ed-Health Cycle -> cost reduction
  • 1986 Animal rights -> computers & human organoids
  • 1998 hESCs -> 1981 mESC, 2006 adult stem cells
  • 1999 Unsafe xenotransplants -> 2015 endogenous viruses inactivated
  • 2004 Robot wars -> IGEM human practices
  • 2004 Lawyers & Synthetic DNA -> surveillnace
  • 2005 IRB poor data access & sharing -> PGP & pgEd
  • 2012 Gain-of-function H5N1 -> 2014 pre-discuss gene drives
  • 2015 Patents & Germ line -> addgene, safety+efficacy relative to standard care
  • 2016 Aging reversal = over population -> space
  • 2016 Space radiation, 38% gravity, microbes -> space genetics

This is a list that you will see a few times during this talk. This is my personal ethics list. These are dated in the order they appeared in my life. They are not necessarily significant or logical in any other way. They represent ethical things that I have dealt with. I teach a required course in ethics at Harvard Medical School for the graduate students, and we talk about some of these things. We're not going to cover all of them today.

The first one I want to talk about is this idea that when a scientist stands up here and talks with animation let's say about a subject, that means he or she is advocating that subject. He is an enthusiast. If I was to talk with you about neanderthals, then you would all conclude that I have an active program to clone neanderthals. This should not be considered the case. I should be able to discuss topics that from my point of view are alarming or in possibly in a negative sense... I can say it with animation without being an advocate or enthusiast. And so, to that end, I have been coauthor on about 30 some articles on the topic of ethical legal social issues something that you know I've been very privileged to have close relationship with Jontean Lonsaunft, embedded in our laboratory as a bioethicist. Very often we seek solutions to the polarizations, where we are looking for a win-win. You'll see some examples in this list as we go through them. The win-win for embryonic pluripotent stem cells was induced pluripotency, that's something that satisfied most of the argumentative groups on both sides, although not everything.

Let's talk about equality... it often comes up at the end of a long meeting like this. At the end of the Sunday afternoon.. let's start with it. I'm going to give you a quirky view on it, so take it for what it's worth. What is the revolution we're talking about here? I said reading, writing and revolution. That's my version of math. But, what is the revolution here? And of course, we all know the revolution is CRISPR. I think that, the normal acronym .. none of those letters have anything to do with the CRISPR technology that we're all enamored with (clustered regularly interspersed short palindromic repeats). I am going to give you some other words that I think better capture what the revolution is, and it has many components, not just nucleases (see more), it has the idea of comprehensive data, recombination, informatics, sequencing, pluripotency, and reduced costs. Comprehensive data is a possibility and is often insisted upon. We have recombination: not just cutting, but cutting and splicing and making exactly what you want. Pluripotency is to say you can go from a single engineered DNA molecule to a whole organism. We also have greatly reduced cost.

For some of these things, I know this is a broad audience, some of these pictures will be inappropriate. They will be complex. Just think of them as eye candy. I like the Nova sequencers, it's named after my granddaughter, Nova... No, just kidding. Her name really is Nova.

Cost of DNA sequencing and cost of DNA synthesis

We have this exponentially shrinking cost of reading and interpreting your genome. I have been obsessed with this for a long time, ever since I left Song Ho Kim's lab, when I was 20-something years old. And shortly thereafter, before that we couldn't even conceive of the project. When we finally started to deliver it, it was going to be about $3 billion dollars, the price of the tallest building in the world. And now it has been about $1000 for about a year now. It will be free soon. In the sense that your phone has free software on it. Many people in our project get their genome for free.

Here are the curves. The thing that is most strinking.. this is factors of 10 on the y-axis and it goes back to almost the time where I started molecular biology. It was on a Moore's law curve, it was a steep exponential for computers. It got really fast in 2003-2004. We'll talk about that in a moment. This is for both reading and writing. Reading has improved over 3 million fold, mostly in the past few years. Writing of short oligonucleotides has improved over a billion fold in that period of time. I think both of them can improve by another maybe 1000x to 1 million fold. Don't ask me when and don't remind me when it doesn't happen. But anyway, that's my guess. And so far we have all underpredicted it, rather than overpredicting it. What happened in 2003-2004? I will give a totally self-centered and self-serving synopsis.

These two papers, we figured out how to miniaturize and multiplex and self-assembly could be used. Biology should be really good at this thing. It's atomically-precise, as miniature as you can get. We captured that and the revolution in electronics and microfabrication in both reading and writing. The key thing here to take away is that you can't do one without the other. Our most popular sequencing method is literally called sequencing-by-synthesis. Every time we do synthesis, every time you do editing, you should know the genome you're editing. I mean, you don't have to, in the same sense that you could hypothetically text and drive your car at the same time. If you are doing genome editing, you need to be doing sequencing.

There is-- I've been impertitent enough to redefine CRISPR, I'll be blasphemous enough to say there are more than one editors in the world. These are molecular machines that scan your genome, which for humans is 6 billion base pairs, and looking for one 20mer that the machine likes, and either making a cut or a recombination at that point. How does it find that one in six billion? The scan head, the place where the action occurs for that recognition of 20 bp out of 6 billion occurs, is either DNA (in red) or RNA (in blue) which is what CRISPR-cas9 uses. Or there are some of these that use proteins for recognition-- like Watson-Crick base pairs. And proteins have more complicated but well understood rules. So that brings us to the next question: why is everyone so infatuated with CRISPR? Why do we think it's an improvement? Some people say it's easy to use and low-cost, and that's why it has been exalted to such a high status. I will say that maybe... but maybe not. About the time that we were starting to publish this deluge of CRISPR papers... this lovely paper came out from Jun Sung Kim's lab where they .. with very little fanfare, assembled an alternative to CRISPR nuclease, so called TALENs, for all for almost all of the protein coding genes, over 18,000 of them. It couldn't have bene that hard to do even though it was CRISPR. I realize it's an anecdote, but there you have it. TALENs day in the sun lasted only about a year. CRISPR has lasted now a whopping 4 years in the sun. It may last forever, maybe it's the last editing method. But this is kind of the history-- I could write the same thing for genome sequencing, a complicated plot of a whole series of replacements. It started with homologous recombination in mammalian cells, and some people asked who will give the prize for genome editing, and well it's already been given to Mario Capecchi. There is some progress. There's... this is factors of 10 on the y-axis, there's a lot of constancy in the editing efficiency. I think it's the editing efficiency that we are celebrating with CRISPR.

Genome editing specificity and off-target issues

  • Computational: Jan-2013 Mali.. Church, Cong.. Zhang, Science
  • Paired nickases: Aug-2013 Mali.. Church, Nat Biotech.
  • Truncated guide RNAs Jan-2014 Fu..Joung, Nat Biotech.
  • FokI fusion Apr-2014. Tsai..Joung; Guilinger..Liu, Nat Biotech.
  • Brief binding Apr-2015 Davis...Liu. Nat Chem Biol.
  • Weak Cas9-dsDNA, Jan-2016 Slaymaker ... Zhang, Science
  • Weak Cas9-ssDNA, Jan-2016 Kleinstiver..Joung, Nature
  • First SNP specific method: Jun-2016, Chavez ... Church BioRxiv
  • ... but with multiple changes, the risks multiply.

To be a party pooper again, if you're going to be applying CRISPR, especially in a clinical setting where you're treating a large number of humans, in cells simultaneously... if any one of them gets an off-target event, and most drugs have some kind of off-target event, if you have an off-target event with CRISPR and it lands in a tumor supressor gene, you run into the risk of getting a tumor. So you need the off-target effects to be as low as possible. This list doesn't give it justice, there's a large number of articles that have improved the specificity of CRISPR. One of them, from my group, even claims that you can detect single nucleotide polymorphisms. This jargon means you can make a CRISPR that will recognize, be off by 1 nucleotide, and you can do this fairly routinely. I urge you all to publish on BioRxiv. You will see some other references like that. Even as we bring down the off-target as low as possible, the more changes you make, the higher the risk, and they multiply out. Let's say you have an error rate of 10^-4, one error in 10,000 cells.. if you are treating a billion cells, that's a lot of errors.

What about on-target errors? These are ethical issues, believe it or not. These are safety and efficacy issues. Mostly safety. On-target, CRISPR is not ideal. In fact, none of the nucleases that have been used, meganucleases, TALENs, zinc fingers, they have a race where the double strand break and the repair... I call this genome vandalism instead of genome editing. Very often you only want to make a precise change. There are some solutions. My two favorite are from bacteriophages. In fact, CRISPR itself evolves with bacteriophages. Most of the gifts that on slides I've been mentioning, most of the technologies were not invented, not in the sense of thinking about atoms from first principles, they are gifts from the microbial world. These two are as well. These two work, the specificity we want, on-target, by not making a double strand break, but only once they bring together the donor DNA you want to change and then the target, and then they do their job. These are not as applicable as CRISPR. You don't go out and sell your CRISPR stocks. They are very specialized. The beta recombinase only works in ecoli k12. You might say that's coincidential, that's my favorite organism. I am sure you are all saying that. You all love ecoli k12. But anyway.

Beyond CRISPR = no double strand breaks ((24m 32s))

Codon recoding and multi-virus resistance

One of the world's largest and most radical genome engineering and I think in this case radical is a good thing maybe where we changed a 4 billion base pair genome to eliminate one codon. So all of the organisms in the world, synthetic and natural, use 64 triplet codons. ACGT to the third power. Except for this one, which only uses 63 codons. We expected it to be useful for four things, and in fact it is useful for those four things. It uses non-standard amino acids very efficiently, we are no longer restricted to the amino acids of nature. We can put in completely chemically synthesized amino acids. It's genetically and metabolically isolated, which is an ethics lesson. It's bio-containment. If we're going to release something into the wild, we want to have a reverse button. We want to have some ability to go back. And that's what this is. We've proven it. And finally, multi-virus resistance. I think this is profound both philosophically and practically. It's the idea that you can make an organism resistant to all viruses, including viruses you have never studied and never understood, as long as they follow one rule which is that they use the genetic code of the host. And if you change the genetic code of the host, they say, hey what's up it's not working and they can't-- it's radical enough that the viruses cannot evolve around it. That's a theoretical prediction and it's looking like it turns out to be very likely. We were surprised how multi-virus resistant these were, with just one codon change, and now we're changing seven codons.

Human genome project - write

We're going to jump now from that radical recoding of the codons, radical application to something that goes beyond editing, we call it Human Genome Project Write, or hgp-write. This is writing the whole genome, beyond editing. We're going to change to engineering humans. We're going to go into this gently. We'll start with something probably not controversial, which is engineering human cells for diagnostics and testing therapeutics and testing hypotheses. When we started this exercise as to why you want your human genome, some things are highly predictable, but others are variance of unknown significance. If you get one of those variants of unknown significance, you would like to know, how do I determine whether it's harmful or not? This is a new way of doing that. You can alter a genome by one base pair or however many the difference is, let's say as little as one, and you can make it into a complex tissue or organ, or modestly called organoids, and then test whether that function has changed. You can test cause and effect. You no longer need a cohort of 10,000 or gigantic population samples to get an unconvincing correlation, you could do cause-and-effect. As long as you're changing one base pair and have a good model. It doesn't involve animals, it's human organoids. So at the top we started with PGP1, code for personal genome project. Full disclosure, the individual number one, I said I can release the names of the participants, it's me. We sometimes call me guinnea pig number one (GP1). We take those fiborblasts, turn them into stem cells, put in crispr-cas9 and then use that with a repair oligo to take out that one gene. If you leave out the repair oligo, it makes a mess and vandalizes the DNA. You can change just that one gene... we've been talking about off-target on-target messes, you expect me to change only one change? Well, we sequenced the genome, this is embryonic stem cells, even if you have off-target if you have a clonal where you grow it up from a single cell, you can sample that clone and show that it is extremely unlikely that you have anything off-target. In this case, this baby was hypothesized, in some cases you might have hundreds of hypothetical changes-- the most likely case was missing a G on its x chromosome, males have only one x chromosome in every nucleated cell. We made two cell lines from PGP1 that differed by that one base pair, so tihs was a clean experiment. If you compare my cells to yours, there would be 3 million differences, but changing 1 bp is a single change. Luhan Wang, she of course sequenced the genome, she didn't even ask before doing that.

So then this is... TAZ mutation is responsible for cardiomycote morphology abnormality. When you have a normal stem cell, again from PGP1 on the far left here, and then change one base pair and it's abnormal biochemistry and abnormal morphology. You can determine cause and effect.

If you run this in reverse, then you have gene therapy. This is genetically modified human cells and human organoids. What about genetically modified humans? Some people are surprised to hear that there are genetically modified humans running around. There are 2300 clinical trials on gene therapy. There is only one approved, ironically in Europe where they are not keen on eating genetically modified organisms but they are okay with genetically modified humans. Glybera is the most expensive drug in history, it's $1 million USD per dose. To somebody like me who likes bringing down the costs of things by a million or a billion fold, this is unacceptable. We have various strategies we can talk about for how to bring this cost down. The main thing is increasing the market size. The problem is that for a lot of these they are orphan drugs, with very tiny market sizes. Some infectious diseases will have gigantic markets.

Human germline genetic engineering

So now we get even edgier-- we went from human cell lines, to organoids, to humans, and now to human germline therapy. Gene therapy is typically done in adults, in some cases children. The earlier the better, in some cases, like curing blindness, which happens with gene therapy and is a real thing. You have to do it at a young age because otherwise they will be able to see photons but they won't be able to process faces. You could imagine taking this further back, you would want to engineer in utero or earlier. So what is the status of human germline therapy? Some people take it for granted that it will never happen. Some people think it's not allowed anywhere and it's not happening anywhere. That's very far from true. It's permissible in many countries, including the United States. Human cloning is actually permissible in the United States. The question, though, isn't just whether it's permissible, it's whether it's acceptable. It's already done. Here I am showing a picture of a real patient, the girl in the middle had mitochondrial germline modification. Some people might say, well that's not really germline-- but it is, it's passed on to her children. Most of these therapies are to reverse a problem. You could either think of it as a slippery slope if you think in negative terms, or you could think of this as a way of testing without putting too many people at risk. It will start and move on from mitochondrial therapy to sperm and fertility where you will be changing men with certain kinds of infertility using genetic traits to become fertile. If you could do that to just engineer the soma the somatic cells around the sperm would be considered normal somatic gene therapy not impacting the germline. But since it is a sperm itself that is effected, then you have to at least reverse it; you're changing it to what everyone else has. If it happens in clones, then you could in principle get something that is close to 100% correct, where it has all the low error rate of CRISPR plus all the low error rate of checking the clone by sequencing. This could possibly make it into fertility clinics. No embryoes would be put at risk there, any more so than any other medical procedure, then you could-- the might-- the world might move on with due consideration and conversations like we're having here- to serious diseases like Tahy Sachs which is currently handled either with abortion or in vitro fertilization where... abortion you might sacrifice 25% of the children born with the recessive disease, and with IVF you might sacrifice like 80% of the embryos some of them might be normal but they will not be implanted. And for a large fraction of the United States and the world, that loss of embryos is unacceptable. This might be-- usually the way this conversation is phrased, almost every conversation I've been in, is that we're putting embryos at risk. In this scenario, we might actually be saving embryos, putting fewer embryos at risk than current medical practice. Now there is great practice at making eggs and sperm entirely outside of the body. Mostly done in rodents so far.

Human enhancement

So we've gone from human cells, human organs, adult human therapy, germline therapy which is past-tense, and now let's talk about enhancement. I am not doing this just to be provactive, I think it's important to visualize things in advance even if they never arrive, we need to talk about it. Here are some of the traits that you might want to enhance in human beings:

  • Visible light
  • Hearing
  • Chemosenses
  • Touch
  • Heat sensing
  • Memory span
  • Memory content
  • Heat tolerance
  • Locomotion
  • Ocean depth
  • Altitude
  • Voice

.. or in some cases you might want to prevent the world from enhancing. Right now we can only see a very tiny sliver of visible light with our eyes. We can only hear a limited range. We can only sense certain chemicals. Touch, heat sensing, our memory is very limited. We locomote very slowly, and so forth. But we're already augmented-- we are almost unrecognizable to our ancient ancestors, our ancestral limits can be blown away, we can see the entire electromagnetic spectrum from gamma ray to radio. The list goes on. We can go so fast, we can reach escpae velocity from the earth and we can survive in the vacuum of space and the extremely cold 15 degrees Kelvin out there. Our goals, like being able to go into space and survive in the vacuum and cold, would be incomprehensible to our ancestors. I would argue that most of the augmentation that will exist and already exists will be physics and chemistry, it's not genetics.

If we are to talk about genetic enhancement, then it will probably be intelligence, immunity and longevity or aging reversal. Much of the ethics we will talk about, we might talk about other things, if we get to something safe and effective there's a tendency to drop the conversation. In vitro fertilization was described negatively with like "test tube baby", which doesn't sound as negative as it did back in the 1960s, but when Louise Brown was born and she was beautifully healthy, suddenly the ethics turned around almost 180 degrees where it went from we should never create the monsters of test tube babies, to we should not deprive parents of their right to have babies of their own that is the product of union of two people that love each other. But there are tihngs beyond safety and efficacy, we might drop them if safety and efficacy are proven, these include the possibility that parents might treat their children like commodities which in some senses they already do. They get them the best education and they expect them to perform. Some parents will be making choices where they choose to have children with hearing or not hearing. If you look at in vitro fertilization in the United States, a very common practice is to choose a gender and guess which gender that is 80% of the time-- it's female. I don't know what your expectations were, and I've seen editorials that give reasons, we can talk about that in discussion. The other thing that we need to be cautious about is the loss of neurodiversity. In a classroom, there's a great desire to change the classroom into a Henry Ford mass production exercise. Diversity in skin color but none in their behavior and appearance. If somebody fidgets too much or falls asleep, this is a bad thing and you should medicate them. If you could genetically medicate them, that would be something we should be cautious about, some of our most amazing citizens are those who are on the edge of some spectrum or other. I tihnk we don't want to lose them. We might want to allow them to not be in pain some of the time, maybe to turn it on and off, but not lose them. We talk about diversity, but do we really mean it?

Cis-genics vs transgenics

GMOs. How many people here shop at a market that has the no GMO rule? I do. Wow, you guys aren't the radical Berkeley crowd. Or QB3.. There are some GMOs that almost everyone likes, even the people that are anti GMO foods. It's usually at least adalimumab, erythropoietin, and etanercept. There's a longer list but I picked a few that are recombinant proteins that treat some of them fairly common diseases others are orphan diseases. So there is a difference of opinion when you get to things are extreme health interests and sometimes extreme economic interests. In Hawaii, they banned all GMOs from the islands, except for papayas because that was they-- the papayas would have gone extinct. So there were some room for negotiation on that subject. I would argue the future of GMOs-- this is actually a report from one of the anti-GMO groups, one of the main problems in genetic engineering is random insertion of genes which could create toxins and allergens... well, random mutations are definitely random. We are not doing random genetic engineering, though. I argue that if you want to open a car door, you could shoot it with random shotgun fire, or you could engineer a handle and really use the car door.

So we get to very interesting definition issues here. This is not technology, it's possibly ethics. Transgenics is pretty commonly used word, it's almost the definition that people use for defining GMO plants and natural and organic. You have moved a gene between species, over a great distance, and the distnace is defined in regulations, how far apart the plants have to be. Cis-genics is a less used word-- it's a regulatory, if you think of it generously, is a regulatory opportunity or a loophole if you think less generously. There are 30 GMOs in the last 5 years alone that get through the USDA and to some extent international regulations because they are changing just one base pair. You change one base pair, that can happen in nature. If you change just one base pair, that could have happened by the shotgun approach and you cleaned it up with conventional genetics. The most recent famous example is the white button mushroom where they knocked out a gene to reduce browning at Penn State University. A transgenic that should have gotten a pass from the critics, a matter of life and death.... anti-GMO is usually justified because in most cases it's not a matter of taste or necessity, but actually golden rice-- the thing that it addresses, which is vitamin A deficiency, millions of people go blind and die. They often die within a year of going blind, it's a cause and effect relationship. Golden rice was working by 2002, it started long before that. The decision was made to make it transgenic, it didn't have to be transgenic I think because beta-caretein is already made in rice, it's made in the wrong place in rice, so cis-genically you could move it, but instead they imported a gene from a bacterium. Phlyotene synthase from daffodil, Narcissus pseudonarcissus. And carotene desaturase from soil bacterium, Erwinia uredovora. I don't know where golden rice is going to go. It's an example of a transgenic that didn't escape, even though it's addressing a major health threat.

Here's something you could say is not beyond safety and efficacy, it's a part of safety, but it's not what the USDA, FDA and EPA worry about. They worry about the distant future which might be next quarter, or maybe 10 years from now. But this is 100 years. Here's a juicy example that hopefully many of you. In 1872, Yellowstone was established and the grey wolf was in decline in yellowstone and elsewhere and it was completely gone by 1926. The endangered specifies act was the second such act, and it allowed reintroduction in 1995. The introduction of wolves was dramatic. The elks didn't like it, they started killing a couple dozen elks, and then the beavers came back, the waterfowls came back, and fish and so forth. It was an amazing impact on the environment. We made a mistake in 1872, getting rid of the wolves. What if we're doing similar things today? We can't just say don't change anything, because that can have negative effects as well.


This is even worse than transgenics if you think transgenics is bad. This is moving organs from pigs into humans and human organs into pigs Something like this was in the news recently, and many times before, this was an article by Carl Zimmer on the topic. We have some skin in the game here, with Yang and eGenesis. Humanization of pigs goes back at least 2 decades. There was a $1 billion USD investment into this, there was a good roadmap, and we now think it's about 50 genes or more that needs to be changed. What freaked them out was that pig organs were producing viruses that could impact immuno-compromised recipients of the organs, and that would probably be bad. You don't want to have swine flu or the equivalent evolving in your immune compromised patient. When we got the awesome power of CRISPR, .. she and her team decided to try to get rid of all 62 endogenous retroviral genes from the pig with one CRISPR all at once. And it worked. We were surprised how easy it was. 62 genes at a time seemed completely out to lunch, at the time. In 14 days, in a 37 degree incubator, and then some PCR to screen it, that's all it took. Now there are piglets that we have ultrasounds on these things where many dozens of other genes have been changed as well. I look forward to seeing and holding these piglets when they are born. If they aren't, then we will try again. And what excites me about this is a little more subtle than merely curing the transplantation problem, which is very acute, it's not just that you and I are incompatible for exchanging organs, it's that there isn't enough of us to give organs. It's more than that, though. When we produce organs, we are going to be highly motivated to do preventative medicine-- we should make organs that are pathogen resistant, aging resistant and cancer resistant. It was hard to get FDA approval to get a healthy human and make them resistant to aging, cancer and viruses. Preventative medicine is very hard to do unless it just involves taking walks and eating broccoli. Not something as radical as gene therapy.


Aging reversal is a real thing. We will see it for physical aging disorders as well as cognitive aging disorders.

We have an aging population and this is extremely important. I imagine many of these will be used off-label for reversing aging in people who do not have disabling disorders. There are two classes of example of aging reversal where it has been shown in mice, one of them involves hooking up young mice with old mice, with their circulatory systems. I don't recommend you do this with your children. But it works. Another one recently published is that using the same reprogramming factors to establish pluripotency in stem cells, you can do that on a whole organism basis and it reverses aging.

Gene drives

Gene drives was something that was hypothesized again about 2 decades ago. Actually it was observed in the 70s by Bernard one of my mentors as a graduate student. It was hypothesized by Austin Bert that it could be turned into a technology. Kevin Esvelt and Smidler and Catteruccia published a paper in 2014 where we didn't do an experiment-- .. not talk about it before it was ready to be published like H51N, we didn't do that, we talked about it upfront. We talked about reversing invasive species, revert herbicide and pesticide resistance and also control vector born disease like malaria and lyme disease. Nobody likes malaria. Unlike normal inheritance where half of your offspring inherits the trait, it confers it to all offspring, so it spreads to the entire population very quickly. It uses crispr, which is mechanistically, you know, it's scissors, it's basically scissors. The crystal structure indicates that they are scissors. We put up to 8 scissors in the mosquito experiments to make sure the mosquito genome does not become resistant to the crispr gene drive. Normally we would call this selfish DNA but in this case it's altruistic because it's doing what we want, it's carrying the purple cargo there, purple cargo is resistance to malaria. We're not trying to wipe out the mosquito, we're trying to make them resistant to malaria with antibodies and small peptides. And those scissors are aimed at an essential gene, a ribosomal protein gene for example, so that if it repairs by genetic vandalism then it dies. If it repairs by copying the gene drive, making two copies, it lives. You can now tell anyone in the elevator that you understand gene drives.

Here's the sophisticated stuff. On the top is how fast it will spread, which has been confirmed in some organism experiments, and then we have a new one which was published in Noble et al. Biorxiv 2016. It was called a daisy drive where we have three drives, C, B and A in that they cut themselves the part of the genome they come from but actually the cut the one from the next one. The yellow decays because it's slightly deleterious and it has nothing driving it. The orange lasts longer because it has the yellow driving it. And the blue has yellow driving orange driving blue. This allows you to do geograpihcally and temporally controlled drives.

Space genetics

What if we cure aging and eliminate poverty and diseases of developing nations? We're going to have overpopulation, or at least some people say that. I don't think it's a great solution to say well we're not going to cure diseases of poverty or aging of the industrialized nation. One possibility is going to space. I don't mean this frivolously. It's a good idea for our species because we're at risk from supervolcanoes and asteroids. In space, we have a new set of problems including radiation and low gravity even on Mars. And then there's space genetics issues. We have a consortium on this and space colony challenges.

We have challenges like gravity, osteoporosis, neuro-behavioral issues, microbiome issues.

Rare protective alleles:

  • LRP5 G171V/+, extra strong bones
  • MSTN -/-, lean muscles and low atherosclerosis
  • SCN9A -/-, insensitivity to pain
  • ABCC11 -/-, low odor production
  • CCR5 -/-, HIV resistance
  • FUT2 -/-, norovirus resistance
  • PCSK9 -/-, low coronary disease
  • APP A673T/+, low alzheimers
  • APOE E2/E2, low alzheimers (E2=R112C, R158C)
  • GHR,GH, -/-, low cancer
  • SLC30A8 -/+, low type-2 diabetes
  • IFIH1 E627X/+, low type-1 diabetes
  • TERT overproduction, low aging
  • CDKN2A overproduction, low cancer
  • TP53 overproduction, low cancer
  • GRIN2B overproduction, high learning and memory
  • PDE4B inhibition, low anxiety and high problem solving in mice

Kind of a quirky list here. These are rare protective alleles. They are things that make your bones extra strong which could result in something that could help in space, or on earth. Some that reduce pain sensitivity, which you might want to turn on and off, because if you have it off all the time then maybe you end up hurting yourself like kids chewing on their tongues. ABCC11 will give you low odor which might be helpful in space travel contexts. The good version is common in Asian populations.

There are some things that have been tested in animals, like low cancer and high cognitive ability.

What about radiation resistance? Here's a case in the literature where radiation resistance was improved 100,000-fold. 10-fold using e14-deletion. 50-fold using recA. 20-fold using yfjK. And 10-fold using dnaB. See Ecoli, Byrne et al, eLife 2014 ("Evolution of extreme resistance to ionizing radiation via genetic adaptation of DNA repair"). This only requires 4 mutations. There is a wide variation in natural organisms, but the only difference here is those 4 mutations.

I am going to end on this speculative notion where in the normal surgery ward, we have a choice about what we take into space. Are we going to take the entire Noah's arc and all the whales and malaria and smallpox, or are we going to leave stuff behind, like germs? Well then we don't need to wash our hands for surgery. If we can turn pain on and off, we don't need anesthesia for surgeries in space.

GET conference for personal genome project. Some grants from the BRAIN Initative. Thank you. Questions, or discussion, actually.


Host: George, that was wonderful. We would like to have a discussion and invite questions from the audience. We have some people running around with microphones. We will bring you a microphone and we will open it up to the floor.

Q: Thank you so much. As precision medicine continues to grow and we continue to use genome editing technologies, what are the implications on off-target non-coding regulatory regions, like an off-target deleterious mutation occurs in germline, new pathologies might arise in the population. Off-target trying to fix something. If you're trying to fix something and we don't know the off-target effects... What are the ethical implications of that?

GC: The safest tihng to do would be to insist that you have zero detectable off-target changes. So you get the CRISPR as low as possible mainly from computer analysis, and then you sequence the clone. Most gene therapies are not done on clones. They are done on, you'll treat a billion T cells for example, ex vivo, and put them back into the patient. So every one of those has a chance of being off-target. The most serious are going to be in tumor suppressor genes and tumor suppressor exons. It will go through 3 or 4 phases of clinical trials. When you start finding things that cause cancer, then you will go back and fix it. But it will be very hard to-- it will be very hard to prove it, you can't use animal models because they have a different background genomes. So it's probably going to have to be tested in humans. If you want to do what a base pair does in terms of cancer, then we should use organoids. If it's something that happens in 10^-9 cells, well, organoids are pretty small, they are limited to about half a millimeter, and vascularized organoids will be whatever size you want and we are starting to get those.

Q: Thanks. You mentioned in one slide that one way we could get to germline modification in humans is through synthesizing new sperm from embryonic stem cells. But there are a lot of mutations that are occurring and these attempts in mice so far. How do you think this is doing? Is there a lot more work to be done before we see that working safely in humans?

GC: Right, so. The mutations you're probably referring to are mutations that occur just from growing cells. If you look at your body, you're full of mutations, almost no two of your cells have the same genomes. A lot of your cells are missing a chromosome or have an extra chromosome. This does not mean we should glibly start creating germ cells that have the same kind of mutations, in fact the standards should be higher. A huge fraction of human conceptions never make it to term, and probably a lot of that was deleterious mutations to embryonic development. It's a huge challenge. It's not due to CRISPR. It's due to just growing cells, which have a mutation rate of at least 1 mutation per cell division, and it's a little worse in cell culture than it is in vivo, but it's a problem in both places. It's an open question. That's the fun of being a genetic engineer in 2017.

Host: During your talk you presented a number of different areas where genome editing and engineering is going to have a big impact in the future, what do you think is the first area where we are going to see this kind of impact? Will it be in transplantable organs, in a clinical benefit to patients or maybe something else?

GC: Just focusing on what we already have, we already have genome editing for CCR5 treating patients with HIV infections that already have the infections. That's done with zinc finger nucleases. We also have treatments involving CAR T cells for cancer therapies. It's a pretty safe prediction that these are going to make it through serious clinical trials some of these are already in phase 2 clinical trials. Applications to infectious diseases beyond HIV is going to be a major case where you can get larger populations. That's a fairly safe bet. The orphan drug act has made it affordable to companies by getting reimbursable huge regular orphan drug might be $100k or more per year, and if you can do that with one dose then that could conceivably save money. I think that's where we're going.

Q: Thanks for coming here George. I was super happy to see animal rights on your list there but I was a little confused when you brought up growing organs in pigs. I was surprised not to see a straight jump to like in vitro organ growing. Is there a reason for that? It seems like a conflict of ethics.

GC: That's a very accurate... very observant. People wonder how I can be a vegan but allow my postdocs to start a company on making transplantable organs in pigs. In all fairness, we're trying to get organs to develop in the lab. I showed an example of a cardiac muscle organoid, and I think we're one of the first labs to have vascularized organs growing in the lab, which allows you to pump blood through them and get larger and more realistic ones. Given a choice between, if everyone stopped eating pigs tomorrow for all reasons, or hurting them, that would be great. But this is a drop in the bucket compared to bacon. I'm not wild about it. You probably aren't either. But it's temporary, we don't know if we can deliver organs accurately entirely in the lab. I think we can, but until we're sure about that, millions of people are dying from lack of organs. It's a very tough ethical decision-- is a pig life any where close to a human life? That pig could save 10 human lives because you have heart, liver, lungs, kidneys, intestines, and so on. I apologize in advance.

Host: I see a hand in the back there.

Q: Is it a legitimate analogy to say aging and cancer are the yellowstone wolf and humans are the yellowstone elk. Should any pathogens not be addressed or prevented by genetic manipulation?

GC: Excellent question. There are experiments that show--- so, some people feel that you need, that, we have this sudden infatuation with the microbiome. There are a lot of microbiome products. I am involved in 5 microbiome therapy companies. We don't need our microbiome for life. Every experimental animal, goats, chickens, even humans grown without any noticeable microbiome. You get certain less than letahl diseases that can be cured by carefully chosen species of microbes. For a while people thought it was just so called good bacteria, but now they have shown you can do this with viruses like norovirus. So I think we have deep ignorance that will be fixed hopefully soon from merely observing the human microbiome but also doing experiments and experiments on germ-free mice. It's a great question and well put in terms of the wolves.

Host: Maybe you can edit the human microbiome.

GC: You can. But I think there's an assumption that it's healthy out there. You can confer obesity and diabetes by microbiome to germ-free mice, so you want to be very careful even with naturally occurring microorganisms.

Q: You discussed human enhancement through genetic modification, and I was wondering if you could give us examples of what's in our reach right now, and what kind of modifications would be more challenging.

GC: Right, so. I listed many that some people list on their wish list, their so called post-human or transhumanist wishlist. But most of those can be handled by chemistry and physics quite adequately. I wouldn't say I want to change my genome so that I can fly into space without a spacesuit. But, what I listed, those are possible not addressed by chemistry and physics alone. Longevity, immunity and cognitive and behavioral traits. Some of these will be addressed in adults through conventional therapies, you can already see some conventional therapies that treat some of these caffeine, metformin, these sorts of things. And there's, it's hard to raise an ethical issue when you're talking about small molecule drugs. If you have safe and effective gene therapy applied to adult human beings. A lot of the red lines that we draw, I'm not a big fan of drawing red lines in ethics, I'm a fan of having strong discussions and algorithms, but it's much more complex than red lines. The red lines about worrying about embryo manipulation are distracting us from real issues. If you enhance an embryo with cognitive improvements, it's going to be 20 years before you see any impact on society from that. Whereas if you change an adult cognitively, that could spread through weeks on the internet where everyone is doing do-it-yourself gene therapy on their brain. And that sounds ludicrous, but we live in a time of exponential change and I personally know several people already doing do-it-yourself gene therapy on themselves too and these are not wealthy individuals either. I'm "not" encouraging you to do this, but I'm just reporting that there are people doing this. Look at the latest MIT Tech Review for example.

Host: Okay let's take one or two more questions over here.

Q: I'm sure you're aware that the vast majority of diseases are epigenetic in nature, and not genetic. Epigenetics can depend on stress, diet, and lifestyle. Should we be using genetic engineering as a crux instead of cleaning up the environment and altering lifestyle?

GC: So the question is how many of these things can be fixed by lifestyle and diet, and Tahy Sachs to take an extreme example, is probably hard to fix with lifestyle and diet. It's a genetic disease and it's well understood and there are hundreds like it. Diabetes on the other hand is epidemic and it's probably environment and lifestyle and there might be a microbiome fix or something. So yeah I think we should take a deep breath and make sure we're not using a bazooka when we could use a fly swatter. But not everything is going to be that easy. There is one way in the back then.

Host: Please, in the back.

Q: I know you said not to ask this, I'm really curious, what are the advances to make the next leap for read and write for the thousand or million-fold improvement for DNA synthesis?

GC: I am going to ask you to.. restate that so I can get it right.

Q: At one point in your talk you were mentioning that you thought we could make a million-fold improvement in read or write. What are the enabling technologies for that?

GC: If you had asked me in 2003 or say at the beginning of that plot, how are we going to get a billion-fold improvement? I would have said, very glibly, miniaturization, self-assembly and multiplexing. But that's not really an answer. If I had a recipe in the 1980s, things would be different. Roughly speaking, I'll give you an example that we are actively pursuing is that right now DNA synthesis is done by organic chemistry with phosphoramidites, it takes 3 minutes to add a base, you can do this in parallel... while biochemistry, polymerases can go up to 100,000x faster than that. That's not a recipe, but it's an indication of where one could go. And that speed turns into cost because the equipment, whatever the parallelism is, however you're directing the pixels in an array, the cost of that machine has to be amortized before it's-- over a period of years, and hte more cycles per year then the better. So that's a factor of 105 for example.

Host: A question in the front please.

Q: Thank you. Machine learning has really impacted medical research such as in imaging, like identifying cancers more effectively than doctors. In your opinion, what sort of applications do you see for deep learning, machine learning and others in genetic engineering.

GC: So the question is about machine learning in biology in general. I think the simple, hopefully not too glib answer, is almost everything we currently do could be augmented by machines. I've gone on record saying human intelligence has a lot of advantages over machines. We're a 20 watt computer, while IBM Watson is 85,000 watts. Putting aside small human chauvinist comments like that, I think that the deep learning method has huge promise in biology. There will be a loop--- we have a grant from the BRAIN Initiative from IARPA where we are analyzing a single millimeter of visual cortex on live behaving animals and getting the connections wiring diagram for that activity map to synapse level resolution... and in order to do this goal of the IARPA project to help the machine learning community to handle complex visual tasks like self-driving cars, viewing faces and security footage and so forth, these are not solved problems in their communities. So there will be a virtuous cycle where they help us read out the brain, and then what we read out from the brain might help us with our connectome algorithms. It will be amazing if we can iterate this a few times.