 You mentioned that you improved the efficiency of the synthetase with mage. How close did you get to the efficiency of the wild-type enzyme? So it's a good question. We're still a ways off. So I didn't have this slide that showed the biochemistry data, but going into the experiment we were about 800 fold lower in activity. And now we're about 75 to 80 fold lower. So we improved it by about an order of magnitude. We'll still have more than the order of magnitude to go. So we're getting there, but we're not there yet. And just to give some context, we're working basically with synthetase, synthetase is the derived from archaea. I'm putting it into bacteria. So I think there could be a number of explanations as to why we're still not only to be, but it's still a significant improvement that allowed us to drive many incorporations into proteins. Maybe it's getting too detailed in your question, but did you focus on active site mutations only or the full protein? Right now it was focused just on the massive binding targets and the specific domains of the synthetase that we'll recognize and bind to the anticoagulants of the tRNA. So what we're looking at now is expanding beyond those, as well as going deeper into those regions as well, because I think both of those two strategies need to be pursued. So being able to go beyond, so in those experiments we talked about 12 residues in the amino acid binding pocket, but if you do the math there's no way we can sample that genetic diversity. So we're coupling that also with confidential protein design as a way to better inform the new genesis as well and then going to other domains across the protein. Yes. May I ask you a little bit, a more general question, not technical, about the safety and a little bit of a strike of what you're saying. Maybe I didn't understand. You said that you increased the resistance to virus, T7 virus. That is one factor because there was an infection and there was a company who lost a lot of money. And then you were speaking about, over seven days you checked, seven days there, and so you have long term stability. What do you mean by long term stability? You have not the hindsight of the nature of billions of years, and millions and trillions, somebody spoke about trillions of micro-articles, so for me that's not scientific, absolutely not. Okay, so you're touching on, I would say, excuse me just to be direct. That's fine. You're focusing on two points really. One is the first is on genetic isolation, in terms of the virus experiments. So what I try to do is motivate those experiments both biologically as well as thinking about how they can solve real world problems. So the Genzyme example was a real world problem, and it's a problem that is actually common to all biopharmatization processes, where about 15% of biopharmatization processes are compromised by phase infection, okay. So that's a real world motivation to create organisms that are going to be resistant to viruses. And so what we showed is data that, where T7 infectivity was produced, and some data that didn't show where that actually expands to other viruses. So that, what it does is it shows two things. It addresses that problem I described, but it also shows the notion, which is scientific, that if you have organisms that have alternate codes, they are now incompatible with genomes of that exist in nature, and those genomes can come from other species, or they can come from viruses, which have been evolved over billions of years. And what we're showing is that by creating organisms with alternate codes, you can create barriers to the facile transfer of genetic material, okay. So that is scientific. The second is on biocontainment. Again, that goes after a different, motivated by, I would say a complementary problem, which is now thinking about engineering GMOs that go beyond biopharmatization processes. So there's a lot of people, and I think we'll probably hear a little bit about this later this afternoon, about thinking about engineering microorganisms as new types of probiotics to treat disease, for example, or for bioremediation. So when you think about applying these organisms in open system applications, what you want to do is endow safeguards into these organisms. One of those actually is genetic isolation. Another is restricting these organisms such that they can't grow in the wild, or in other words, you limit their growth to defined synthetic environments. And what we've shown is that you can basically, by virtue of recoding, link the viability of these recoded organisms to synthetic amino acids, which don't exist in nature. Of course, what you said, they are compatible, so they can be exchanged. And then, secondly, the containment, I think if I understood correctly, it's 10 to the minus 12. But if you have trillions of trillions of trillions, that's it. Now, it's a general question. It's valid. It's not personal. I'm not... You're a sister, aren't you? Yeah, so I would like to hear your answer. Sure. So with regard to those experiments, we pushed the limit of what we could really achieve in the lab. So if you're growing up trillions of cells, that's quite a bit of cells. And then being able to test those on plates and liquid culture is a big experiment. So that's the limit of what we could do in our lab. To go beyond that, and address the point that you just raised, what happens if you go to trillions and trillions and 10 to the 15th cells? We haven't tested that yet. And so I can't really speak to that. But it's something that I agree needs to be done. The quantity and the time. There's reasons for that. Thank you. Right. But we did test it over time. Over time. Yeah. We basically tested these over the course of three weeks. That's a long time for a graduate student to be a tracking experiment. It's not nature, you know. No. But... Come back when you have to, Billy, and you're later. We do. We do. We do that with the experiments. It has resources. One can be very astonished. You're absolutely right. And I would say this is important. It will be important in two centuries, you see. I would say this is an important first step, okay? And there's more that can be done. That's it. Okay. That's it. Okay. Danielle. So I guess, building on that, can you do a back of envelope calculation on how long it would take to revert or overcome changes, or even adapt by introducing new changes into the genome. I mean, certainly, even while you're doing your experiment, you're introducing other mutations that you're not intending, right, that have nothing to do with your experiment that are just arising. And what rate are those appearing? Maybe we don't know what rate they need to appear at, but at least we can compare what rate those are incorporated. Yeah. It's a good question. I would actually say in the course of those experiments, as we were sort of arriving at our solutions that we presented, we did observe, I didn't have time to discuss that, but we did observe secondary mutations that led to escape, and we characterized those. And actually, that was informative in then creating next-generation solutions to actually overcome those problems. And so we observed what you would typically observe. We observed the formation of amber suppressors. We observed mutations in proteins that are important in fidelity at the ribosome, for example. And those were in earlier strains that didn't have the mutations that were targeted at conserved functional residues of those essential proteins. And that's what sort of led to that solution is really going after an approach that distributed the multiple TGs across conserved functional residues so that you wouldn't allow one or two or three, who knows how many, mutations to overcome that sort of degree of containment. Something beyond that I think will require larger populations, possibly a much longer period of time than what we were able to do in the lab. I'm not sure, Jeff, if you have a comment on that as well, because I know you did some of the similar experiments, but. Right, so the other point to make with regard to the trillions and trillions is that there are many different ways to contain microorganisms. So, Ference got a factor of 10 to the 12th, which is probably the world's record at this point, I would say. But we showed in a paper last year that if you use two completely orthogonal containment systems, each one contains to a level of one in a million, that when you combine them, you get about 10 to the minus 12, as you would expect for two independent control systems. So, there's no reason in principle why two, three, four or five orthogonal control systems couldn't be combined. It's a good point. Sure. We actually observed the same results in an analogous system in that period. I can also make a comment on that. A spin-off company at our institute called Actogenics, I think they were the first ones to enter clinical trials really by a contained lactococcus for probiotics. And so, the only thing they were required to do from regulatory therapies to be able to enter clinical trials was to have an oxotrophy, classical oxotrophy for uracil built into the organism, such that the organism had to take a uracil from the gut components in those patients. And as soon as the bacteria were released out of the patient, concentration of uracil dropped so highly to such a low level that they considered this sufficient. This was EMA approved in this clinical trial. I mean, approved for clinical trial is not the same as approval for real use in the end. So, I'm sure they would have to build in our terminal systems to bring that up further. But it's not impossible to get these kind of products approved to enter development in this case. Yeah, I think that's really important. Was that a naturally occurring probiotic or an engineered probiotic? It's an engineered probiotic producing a human side effect. What do you think are the chances that you could actually manage to build in this artificial amino acid and some vital function for the bacteria itself? So, then you would make it really dependent on? Yeah. So, we are looking at that in two different areas. One is to do more molecular evolution of these active sites in these proteins and as well as see if we can further expand and enhance the activity of enzymes by the presence of a synthetic amino acid. Touch them out a little bit at the end and sort of seeing if we can actually enhance their, in some way, function or maybe thickness by encoding these amino acids. Because if they get dependent on it and they don't produce it themselves. Yeah. Maybe I should add, I mean, currently we do know almost nothing about how these synthetic amino acids are taken up. They provide always a huge excess of these compounds in the hopes that somehow they take it up. But currently there's not much known about uptake and I mean availability, I mean the costs are immense. When I talk to our chemists sometimes it's really difficult to produce these compounds and then at the end we get more nuclear. So, there are limitations still. I wonder if there's been any consideration to refactoring, recoding the organelles. We have a much smaller constrained system, basically a translational transcription machinery, very few genes, like 20-odd genes. Like mitochondria? Yeah. We have considered it, we haven't done it yet, but we have considered that. Because you wouldn't have to worry about changing piecemeal because of the whole thing. Possibly. Yeah. It's 120,000. Just like that. Karen, I was wondering if you could comment on the growth rate of these GROs. I know you showed sort of the relative growth rates between RF1 plus versus RF1 minus. Can you comment about absolute growth rates? Sure. So, in the context of doing the experiment we inherited some secondary mutations that reduced the growth rate to about 60 to 70 percent. And the original one that we published that we can attribute to these secondary mutations. And as you mentioned, we didn't see any impairment due to the RF1 knockout. And so what we've actually been doing since that observation, and when that paper came out, is going back and actually using mage to revert some of those secondary mutations to improve growth rate. And we have observed improvement. And is it still like a M-U-S background? Which I presume would also affect growth rate, but I don't know for sure. So, M-U-S knockout does not affect growth rate. What it does is enhance the background mutation rate. Right. And so, we and others, in fact, as opposed to out there, I would point out number one and two, that has shown that you could basically create these dominant negative protein mutants where you can basically transiently silence M-U-S as a way to relax the genome, drive high-efficiency mage mutation, but then remove those dominant negative proteins and then stabilize the genome to reduce that background mutation rate. So, us and others have actually improved that aspect of the mage process. Yes, Victor. Mr. Farron, you have an idea of what happens to your recorded organisms where they are released into an actual little environment like soil or some kind of mesocosm or something like that, because, you know, you may have surprises. You think that this is completely to have a certainty of containment and maybe at the end it would not be like, I don't know. Is there any data on that? So, it's a good question. So, we haven't done that exact experiment. The closest we've done, and we did this in our paper where we showed the containment is rather than taking them and dumping them into the soil outside, what we did is we bought plates that were made from soil extracts and blood extracts and we showed that they weren't able to grow on those so that they weren't basically able to scavenge from the environment and, you know, escape through a conventional type of metabolic cross-feeding. But we haven't yet tried to compete them in nature and I think they would perform miserably, to be honest with you. Same lines. What I was thinking about was, you know, I should have already mentioned UNESCO mutations, you might not need it for your technology, but along those lines for, if you were thinking about industrial streams or non-domesticated streams, is there any limitations to your technology or it's able to be used on industrial streams? It's a good question. So, as you can see, as showed early on, the technology is really based on re-engineering the replication fork and because that's basically conserved across biology in principle, you can do that. And so, we have efforts of basically expanding this into some other bacteria as well as in yeast and those are shown to be promising. And I also know of another paper coming out in PNAS and you should go look at the poster outside that actually shows a system that a group from Hungary developed to basically port the mage, certain components of the mage process into some of these undomesticated species. So, you'll actually see some papers come out on this in the coming months. You mentioned that by removing two copies of the tyrosine tRNA out of three, you avoided the emergence of one species of suppressor tRNA. What's the impact on fitness of reducing the copy number of the tyrosine tRNA by a factor of three and is there some way to compensate for that with a stronger motor? Right. So, we were fearful of that experiment before we did it and we only saw about a 10% drop in fitness in doing those experiments. And I do think that if it was more dramatic than that, we would have basically put it under a stronger promoter to compensate for the loss in copy number. But for those experiments, it wasn't striking enough that warranted those experiments. Any questions? One more hypothetical question. In principle, using your method, you should be able to swap columns for normal amino acids, too. So, have you any ideas about such an organism to do with it? Not sure I understand your question. You could use your methods. You have introduced an artificial amino acid. If you repeat this two times, you could actually swap to an amino acid. Yeah, that's a really good idea. It's something that we have thought about and at least mapped out from a design perspective. And I actually think that would create an even greater degree of isolation, if you will, than what we've done. I think it's possible. It hasn't been something that we've initiated yet, but it's certainly an intriguing idea to pursue. And I definitely think that that would work. Because many species depend on things like sex, and today we have a block also. Yeah, it's a way to fundamentally change the meaning of a codon within that species such that not only would you be preventing the introduction of foreign DNA, but you would also genetically isolate the DNA from this organism going out as well. Yes. I have a very naive question. You replace the UAG with a UAA. What happens if you change the UAA to UAG? So, you're right, we convert UAG to UAA. And that was driven by two main reasons. One is the frequency of those codons throughout the genome, as well as how those codons are decoded at translation. And so I showed this during my talk, but I wouldn't try to convert UAA to UAG for two reasons. One, the number of those codons is on the order of two and a half to 3,000. Two, more importantly, is that UAA is decoded by both RF1 and RF2. So then the route to not just recoding, but then eliminating that function from the cell would be very challenging because you would have to probably knock out a release factor and then engineer the other one to alter its specificity away from UAA. So that wouldn't be a natural choice that I would pursue. Could that implementation change the system because you have to be careful? You have to check, but I mean that's for me the most reasonable thing. Clearly, I also would do that, but you have to take into account that you may... Yes, with our connection to the phosphate network. I mean, we just now opened a completely different door. We didn't expect this, but of course when you think about it it makes a lot of sense. Farron, you want to comment on the question? I mean in a lot of ways I sort of anticipated that question with certain things I covered in my talk where if we do want to redesign biological systems I'll just quote what I have up there. We have to do it in the context of evolution and leverage what biology does and does well and is distinct from other, sort of more... like distinct from other, say, conventional engineering systems. And in many ways I think the exercise in recoding was one way to sort of test a sort of permutation on the evolved natural genetic code. So I think in many cases a lot of the work that we're doing addresses your overall question, but I would say that... Yeah, no, I agree. I'm just curious. Okay, so given what you've learned from these experiments what does that mean? Can we draw more general lessons out of it? I would say that really I think the most elegant solutions are going to, let's say, might come through the lens of engineer but are going to be inspired by how biology solves the problem. And I'm a big believer in that. So I know that there was some discussion yesterday afternoon on the approaches of modeling versus more evolution. I think they both have a role at the table because we are talking about... they all have to solubate the laws of physics and chemistry so there are opportunities to model as a component of design. But biology always has some exceptions to the rules. I think we need to appreciate. And that's where their ability to adapt and evolve and find new solutions or landscapes needs to be anticipated. Well, can they even be anticipated? To Kirsten's point about not realizing the links between... We might not be able to predict what the solutions will be but anticipate that they are there. Fair enough. Great. To cut in there, I think to design for evolution we have to consider something that hasn't been mentioned here but there's very few labs working on that which is engineering more than one microbe or more than one species and let them collaborate. That's basically what we see in real evolution is competition with other species and then you have consortia, farming, biofilms and everything. And I think it's not my area of expertise but I think there's a general observation that if you have mixed populations or some fermentists this tends to stabilize the fermentation processes in those clear samples that is done already. So you would have to engineer more than one organism to actually come up with a more evolutionary, stable solution that would be my intuition. And then there is a whole body of theoretical biology of competition and fitness in those replicator systems that have more than one participant. So there would probably be a lot of interesting theory to try to bring on the table as well. Okay. Well, thank you very much for an interesting discussion and with that I think we can go to lunch.