 Okay, well, it's late already. It's half past five, so I think we continue with the next bit of discussion. Like this morning, I think it's a good idea to start with the more technical questions that are addressed at the speakers specifically. And briefly discuss some of the broader issues that arise when we discuss non-new genetic systems. I would like to point out that actually tomorrow morning there will be also, there will be more talks on related issues. So we don't have to finish this discussion tonight. So does anyone have a more technical question for one of the speakers this afternoon? Of course, we would like to know more about the transcription, the unnatural transcription. So if that was a teaser, can you tell us more about what kind of molecules are you using? So we've transcribed GFP, Superbold GFP, three different codons to look at early, middle, late. And we are only now beginning to look at sequence biases, so we haven't heavily done that yet. We just got our own LC-MSMS, so we're setting up to do rigorous fidelity assays. I mean, NAB has been great, but we just can't use them in the developmental fashion, because it turns out they have other things to do. But so right now everything is based on the buy it and shift, and it's not a very quantitative measure, so we don't really care about that yet, except looking for differences. What I can tell you, and I alluded to this, I think, is that TRNAs transcribe better. And we attribute that. Usually T7-RNAP is resistant to row determination, because it's fast, but we suspect that maybe running across transcription of R-Guy slows it down a little bit, and so structure prevents row determination. So I alluded to this to you as well. I hope none of you are reviewers on my space because you'll be bored. We've worked hard to get E. coli-RNAP work, and it doesn't, but there are anti-termination systems, which we can deploy, which turn it on. So that's... So you had a transporter to bring in here. Oh, that's what my boss puts. Yeah, what do you want to call it? And you said that you mutated it to reduce the toxicity. Do you know how that mutation worked? I'm a transporter person, so I feel like they didn't know. Yes. Can you tell me? We're publishing that. So I apologize, I can't, because some of this work is funded by a small biotech company, and they get hyper-sensitive to things like that. The camera. So two questions. The first one is you didn't say whether the cells turned green, and the second one is what happens when you put consecutive unnatural bases in the template? So they didn't turn green because in no case have we transcribed both the tRNA and mRNA, but what I will tell you in getting to that, we've isolated mRNA and the majority of its full length, but of course unlike replication, you get some that are abortive early, and so we get some that do abort at the unnatural base pair, and what's really interesting is they don't abort there, or they abort 11 nucleotides past reproducibly, and that's a footprint of a ribosome, and other people have already shown that, and we've gone in with specific nucleases to digest them in a certain pattern, and it is certainly that. And so all I can tell you is that the RNA is spewing out of T7, and the ribosomes are recognizing it and they're stalling. Yeah, and so the hope is that once we do it together. The second question. So we have PC or amplified contiguous unnatural base pairs. It's certainly more difficult, and I like to be clear about this. We are not creating an orthogonal system. Our base pairs require the context of a natural system. We've never even tried, and I don't think they would even form duplexes if they were all our hydrophobic analogs. Having said that, we will create more codons than could ever be used if we never put them anywhere near each other, and that's the goal. It's not in any way to imitate nature and to make our analogs as good as GC and AT. It's simply to get to the spot where you can stably store and retrieve and evolve increased information. There's the question up there. It's again for Floyd, it's anyone actually, for a technical comment. So one of the beauties about natural systems is that they've evolved a lot of very sophisticated DNA repair systems, including basic decision repair, which deals with spontaneous damnation of cytosine, urosil, spontaneous loss of base where you end up with apyrinic apyrinic sites, etc. Plus all the redox perturbations on the DNA itself and all sorts of other chemical damage that occurs. So more natural systems have highly evolved to deal with that and obviously keep the code completely faithful and replicable between subsequent generations. So how is the synthetic base going to cope with normal chemical perturbations and how will the natural system deal with that? For example, if you have a synthetic base next to say a cytosine that becomes deaminated and the urosil is removed, you end up with an ap site, do you have any information on how that might work or have you already done that? So from our perspective, the unnatural base for itself appears not to be generally recognized by damage repair pathways, MMR, BER, NER. What we think, and so none of this is published yet, but we've now looked at a whole bunch of different sequence contexts and some of them are better replicated than others and it turns out that for the worst replicated sequences, SOS plays a role and if you make a lexate, uninducible strain, fidelities go up. So we think, and that makes sense because all of the other pathways involve recognition of the neclavis and presumably they just can't, so they're probably undergoing fetal cycles, they're probably maybe recognizing a distorted duplex or something but they just can't do anything about it. These are different because it's damaged, specific, independent. Damage type independent. The more subtle question about damage proximal where the same thing then happens and you prevent repair in a proximal manner, sure, but you've got to remember all of those strains are viable at least for reasonably long periods of time because that sort of mutation, they just don't accrue enough. Certainly on the time scale of a protein expression or even frankly a long-term evolution experiment as we run them, we just wouldn't expect that to get done. Our goal with XNA is going to an XNA to a plasmid system where this XNA is producing enzymes, is producing XNA enzymes also, which is in fact a completely isolated system where I don't think repairs should be a question. You have a very short question also about the orthogonality thing. Can you show that DNA and RNA may have these spiral structures but if you change the backbone you can get these ribbon-like things. Do you think this really can be worked into a feasible information carrying system? There's presumably reasons for DNA to have this particular spiral form. The burden is indeed by the polymerase, it should be replicated at all. Of course the helicastrictor is shorter, and it's also more folded than a straight-fold. So it's certainly again not a system which you will be able to use in a chromosome or in a completely replacement in the organism. Again it should be restricted to something where it is applicable. As far as I think you can go much further as a plasmid, but that's also the reason. I would perhaps comment on this. It might be that double AX is so emblematic of biology and so on, but when you look at the choreography of DNA in the cells you see that there is an awful lot of proteins, energy expenditure and so on to unfold it. So it might be that backbones that are prone not to make helices, but that would in an induced fit like Floyd told us, be able to replicate nevertheless, might lead to replicas that would be much, much more easy to manage and evolve at least ex vivo. So it might not be an inconvenience after all, not to have proper folding in regular helices. With changing the backbone you have a whole generation of possible structures. It might be that, yes, it might be that template. You have a whole choice of possible structures. This is just one example. You have to see what has to be processed, what has to be done, what is the task of this change, the number change, topologically change to a teometry change. Again it's a question, if you use it for certain functions, I think it should be used in a cell for certain functions. Can you be more specific? I would like to know, it's a different functionality, but what do you think of? What is your dream? The dream is to encode in this artificial nucleic acids four reactions from which the whole cell becomes dependent. To put a headquarters in the cell to control the cell functions. To catalyze some cell functions in that system. And that the whole cell is dependent on catalytic functions. You put in this artificial nucleic acids. But why is there a structure better than the... It's not better, it's about autogonality. If you have a helical structure, if you have something with a helical structure, like DNA, you produce it in a cell. It will recognize DNA, RNA and block the natural functions. That's the problem. There are several reasons for that. You could think of non-Euclidean biology. So something like that, try to implement systems that have not occurred spontaneously and that are not in the axioms of molecular biologies. Just for the sake of it, it is scientifically interesting to do. There is another answer to your question, sir, which is that when you say what are the natural specifications respected by this unnatural stuff and so on, frankly, the notions that we have of natural design and natural specification is very short. So making artificial systems and interrogating natural systems with such systems as Floyd, etc. is a way of understanding. So synthetic biology is not only a way of constructing a tower of table and stone, it's also a matter, a way of understanding life as we know. I would like to have it more precise because something has to be optimized. And if it's a better shape or better structure, then I understand maybe autogonality, chemical organality might be a profit. It's an optimization organization, of course, but there's this among my fellow evolutionary biologists. There's a lot of discussion right now about whether evolution is optimizing something and if it does, what is it optimizing? So it's not always so obvious. That's not what you're talking about. So it's still a question about, I think it's very funny that you can create this ribbon like DNA structure like this. The question to me is, well, why has nature not already experimented with this before? I might at least see this in case. But then you already claimed that it somehow has tried to do that, but it didn't take it because it wasn't successful enough. No. Behind your argument. A question. Is there any plan to try to put each year old and your basically the same organism? I think it should be attempted. I think not. I think the way I think of it, and Chiro can comment, we've actually talked about this. I guess, I don't think that ours and Chiro's would work together because I think that, I mean, we've looked at lots of mis-pairing and we occasionally, in the process, we actually, in evaluating candidate or natural basepairs, we've of course evaluated what in the end would amounted to mis-pairs between our analog and other hydrophobics. Water excludes water really well. It doesn't, oil excludes water really well. Oil doesn't exclude oil all that well. The question is, do they recognize each other to your basis or recognize by his base? So do we have four basepair systems together? I don't think they would be orthogonal. I think that, so I think that they would probably go on at some intermediate level against each other. But it goes to twice. Maybe. They're each commercially available, so we could have, each of us could have bought the other guys, so obviously there's... I had a question. I mean, we've heard from George, that back as much information as possible in DNA. We use it as a long-term storage of computerized information. And then the unnatural basis that you're incorporating with, I guess, enhanced stability, certainly against biological degradation, would they also enhance stability against physical-chemical degradation? In other words, would these be better formats to store information, like for eternity, if you wish? I think they would be better, because as well the systems which Floyd is using with his modified base are splitting an artificial backbone where you don't have a glycosidic bone anymore. They are eternally much more stable on DNA. So DNA is one of the most unstable all the nucleotides we can have once you change the basis. Because we cannot have any protonation anymore no matter in the base, because they have no nitrogen analog anymore. So they are intrinsically chemical more stable. HNA backbone is highly nucleic acid which is much more stable. So your answer is yes. If you want to store information, better do it with XNA or with modified bases. And the worst you could take is DNA. Isn't there a question that DNA needs to be mutated in the evolution context? Doesn't DNA need to be mutated and changed and altered? So isn't the fact that the instability of DNA is inherently part of the evolutionary process? I think it's a metabolic issue. I mean, I'll just lay it out there. That's why you have why you have uracil bases instead of timing bases in RNA because people claim that RNA becomes DNA and uracil makes more mistakes as timely as DNA. So there is much more evolutionary power in DNA which is base dependent. Of course, this plays a game. I think that people have argued about this for a long time and I think the one argument is that it just was it exceeded sort of a cost ratio benefit to optimize to the point that it optimized because they probably could have optimized further. But I think the ability to evolve, the evolution of the ability can't anchor a subject and the evolutionary biologists can all wound up about. But I think that as a result it certainly drives evolution and it's hard to imagine how it could happen without. That's the topic which I sort of emerged in my life when I was listening to all this talk about changing your genetic code. My colleagues who are doing research on evolution by these machines, you can put animals in them and you get the code. So you get these machines and you put a bit of the animal in it and it reads the code for you and it produces the genetic code. But actually it turns out it's become very clear that in many cases this code is not sufficient. Not all of the information that individuals use to create their phenotypes is encoded in DNA. So the field of epigenetics is now a big thing. So I was wondering if to what extent it has been known whether these alternative bases are also prone to things like methylation and that sort of thing, to what extent they can be modified this way. Some of them may not be able to do so. But in natural systems there is a whole sort of enzyme system that is doing this. Enzyme systems are quite critical base if you see if you do some small modifications. You put a simple halogen or pyrimidine or you remove a nitrogen and put a carbon atom. They are not anymore recognized by restriction enzymes at all. And they are very small modifications. So I think it's going to be very methylation, very similar. It's going to be difficult. So I would give you from our perspective sort of two reasons that I don't suspect that is coming around the corner. And one is they are inherently less functional because they have less reactive functionality in them. You can look at the natural pyrimidines and you see reactive centers. Secondly, evolution works by tinkering or acceptation or whatever you want to call it. You really have to have something that's close to start from. And there really shouldn't be much in a cell that recognizes the analogs. They are pretty far off. So the idea that a small number of mutation or the lateral gene transfer in some new activity, it's probably... I would guess it's unlikely. I think there's plenty for us to do before we worry about that. Also, we shouldn't forget that there is a lot of epigenetic information also in the relations between proteins, how they make for switch or whatever. There is a lot of information that this is still available. What would not be available is the type of information by analogy that is contained in the fact that there is a lot of information about the gene technology that is contained in the fact that this science is methylated. I think it's actually more likely that our analogs themselves would act as sort of epigenetic switches by altering histone packing or unwinding during transcription or lots of things like that. If you go to artificial system, I think you should always go to the more simple system. And it's, for a part, not unfair to completely evolve the system and expecting that you would need the same functionalities for life and survival. If the goal is to have artificial amino acids in your protein sequence or an autogonal system, could you both comment on the alternative approach with the pocodon base pairs from Jason Schinquins and how that kind of is complementary to that? Yeah, so I know Jason well. To me, it's an interesting approach because there's no, ribosome production is tightly coupled to cell growth. It's really hard to imagine a robust, healthy cell expressing a completely orthogonal set of ribosomes that have altered ribosome binding sites to recognize the pocodon base pair. It's a really creative idea. People have been working with pocodons for a while. And of course you get suppression effects as a 3-codon, as you'd expect. In this case, they're not RF mediated. They're just the natural anti-codon mediated. So people often talk about combining this with sort of a Schultz and then maybe a re-synthesized genome that's had more basic re-appropriation done. But we look at that as really elegant and really hard approaches to getting at most two unnatural base pairs in. And then amino acids. And that all might work and it's really interesting, but we think that a perfectly interesting sort of different approach is just to rewire it up from the bottom and have to not worry about suppression as a 3 base pair codon trying to read it in a 4-code. Or to have more than amber. So there's all sorts of issues about trying... I mean people, George Church tried to synthesize essentials and I mean there's really interesting growth effects there because it turns out that essentials are enriched, they're GC biased, right? And they're enriched in the beginning of genes. And they affect structure and you can't delete them. So our perspective is the code and codon usage as part of the code is extremely, is pleotropically entangled with lots of other aspects of the cell. And it's just I would argue an interesting different approach to try to rewire it up from the bottom where you avoid those pleotropic conflicts. Regarding the quadruplet coding and so on, there is also it has to be recognized that what has been proven so far is translation of quadruplet between triplets. I don't think that there is a demonstration that several quadruplets can be correctly translated and that the frame has been, and as Floyd said, re-engineering the ribosome so as to make a multi-plat coding and so on would require an amount of work that is really daunting. I was wondering if you could have like ten different amino acids in a protein, ten new ones I was just wondering about could you give a hint of the kind of applications we might want to do that. What could we do with ten new amino acids that we couldn't do like two or three right now, like real application? So ten new unnatural amino acids as opposed to two or three unnatural new. So first of all, the options we have right now is one and it's not efficient. There's an immense amount of optimization required and remember the whole notion of trying to recode and maybe delete RF1 that will never work outside of bacteria you karyotes have one RF, that's it you don't have the trick to play there so the notion and from a pharmaceutical perspective, the reason all the unnatural base pairs are put at the beginning is because if you get truncation it's not as much of a separation problem if you can't do that game over because you can't separate, I mean, pharma is not going to run columns. So I would argue there's justification for a single. Now for two there's all sorts of other issues, for example in sort of simplest thing, if you go to pharma and you ask well what are the two things they're interested in they're interested in ADCs and they're interested in antibody drug conjugates and they're interested in well, peglation or preglation or PK issues so right from the start the hope is that we could do both that's a pretty easy argument to make everything else when you start to get but right now people can't do two getting to three you start to get to some more esoteric things like well that plus a metal binding site or something that facilitates purification or but ten maybe I put in those three additional colors by different GFP just because it would be fun to see color that too might be okay but I think there should be, there might be ten different but not ten at the same time we could have a cassette of two and always two different because the problem with a lot of enzymes is that the number of the reactions I can catalyze is limited. So you cannot catalyze these reactions, you cannot catalyze a lot of reactions which is due to chemistry which is not which you cannot use in biotechnology and there's going to be a reserve demand for new proteins that can catalyze different reactions for making chemicals, for making drugs so I think there is certainly a need for more than two but for generating new catalytic functions you might only need two but another cassette of two, another cassette of two dependent on the reactions you want to catalyze so to continue in this and actually I think after we should actually tomorrow morning we are actually going to continue on this because actually I am a bit skeptical to what extent we actually can design the next step I think if you it would be a big thing if you could make an alternative genetic system but then you would need to have it evolve to weed out all the mistakes that it will make right? but this is something which I suspect we will come up tomorrow also in one of the talks. There was one question in the back I think. I just wanted to comment on that question that there's two companies PECTI DREAM and RAF pharmaceuticals which use a whole spectrum of unnatural amino acids in PECTI discovery but then in in vitro transcription translation system so if you can chemically charge or enzymatically charge your tRNAs with whatever you want you can reassign the codons again whichever way you like and because it's in vitro you don't have to worry about poisoning the rest of the cell so in that case certainly for peptides circular peptides specifically the amino acids are obviously avatages and so on and so on so I think there's no shortage of cool things one could do in vitro once the code has been expanded Ok, I think I propose that we will continue this discussion tomorrow morning and I'd like to thank both speakers this afternoon applause