 All right, questions. Yvonne, yeah. In the beginning, one of the interesting questions you can prove is whether or not splicing is really essential. And I was wondering if it's not possible for alternative splicing to still occur, even though you don't have introns. Is that some of these silencers that I said are actually exons, right? So in yeast, we have only 5% of genes have introns. And the vast, vast, vast majority of those have one intron. So there essentially is no alternative splicing in cerevisiae. There are a couple of genes that have more than one introns. And one of those shows alternative polyadenylation. But that's something we could easily refactor. So I would never recommend trying this with your favorite mammalian or plant gene. So Jeff, I'll go ahead while we're waiting. So if I understand you right, for your redesign, you were able to do like, there wasn't really any part, except for the one exception that you said that you weren't able to redesign. And even when you were assembling all these, you never had a problem. I mean, or my overstating. I guess I'm curious, what was the biggest thing that you couldn't redesign that you were surprised by? Maybe besides the example you gave? Actually, so yeah, I usually spend a lot of time on this at the beginning when I talk, and I didn't do it today. But we spent almost nine months thinking about what would be the smart things to change and what was too risky to go for. So in fact, so far, nothing has crashed due to the fundamentals of the design. One change of plans that we made after chromosome three, chromosomes six, and chromosome eight were already designed and underway was this paper came out from a group that had first said, you can delete introns with impunity from many genes, which is a true statement. And there are some exceptions to it, so we leave them in for now. There came a second paper, and we didn't know yet whether syn three was gonna work. The second paper comes out and says, actually if you delete ribosomal protein introns, and there's quite a few of those, there's often a fitness defect associated with it. So we actually pivoted and said, okay, from this point on, we're gonna leave the ribosomal proteins in, the introns in, and then come back and take them out later very carefully. Because it turns out you can get around these fitness defects by fiddling with gene expression levels. So I'd say that one is probably the one that, it's the one place where we actually made a change on the fly, but everything else, we haven't changed the original design. So it's very consistent from chromosome to chromosome. Yeah, so almost a social engineering question. You were able to get all these different groups and countries to all agree on the same design. So is it a uniform design across all the chromosomes? Yeah, that was a mandatory aspect of the project. So you had to get your own funding, that was one. Second one was we had to do the design, so that enforced design uniformity. And the third one was you had to agree to give away all the strains and all the DNA sequences without restriction, basically to any, academic or industrial partner. So those are the three major things. And we got some pushback on some of those, but we basically didn't budge on those three principles. Other questions, yeah? So you can think quite a lot with this individual, basically, but I'm wondering what happens if you put this strain into competition with the natural strains? Right, so very good point. So there's fitness and there's fitness, right? So we have a very nice way to do co-cultures and measure the fluorescence with the RFP in one and GFP in the other. And so we have that assay up and running and we've so far applied it to chromosome syn-10 and seen no defect after many cycles of co-culture. Now that's under one condition, of course. And I'm sure we will find conditions where it's not going to be as fitness-wild type. But we have also seen some instances of apparent gain of function after scrambling. So if we, one way perhaps to get around fitness defects will be to do some light scrambling and select for things that recover fitness. That was one of the things that we worried about and prompted us to include the scrambling What about sex with your synthetic sero-inherit? No problem, as sexy as hell. They, so far, and they've never been selected for meiosis, right? But when we put them through meiosis so far, all of them have been proficient at it. Whether it's in the neutral or the wild type? Both, both to themselves and to the wild type. Yeah. I'm not sure we've done that with all the synthetic chromosomes we've completed yet, but definitely for chromosome 9R, 3, and 6, we've done that experiment. If I were to look for an explanation where you only find one type in your cultivation, it would be somewhere there, which you need to conform to the natural configuration. So you can, for functional point of view, you can scramble as you wish, but if you want to mate with other types, you need... Oh, I'm not talking about the scrambled strings. I'm just talking about the base strings. Yeah. Scrambling is something you do after you finish the synthesis. So after scrambling, there will definitely be strings that have mating and meiosis defects. I have no doubt about that. So when you use the scrambling, probably you get a lot of cells that die, right? Yes. And you would expect that this effect would be even more pronounced when you start combining all the different synthetic chromosomes because you have even more possibilities to mess up with the cell. Is that something you're worried about? Is that something you'll even have a shortened notion of this killer we're seeing? Well, we have two knobs we can turn. One is the short, the time, and the other is the concentration of the ligand, you know, the compound that activates the creep. So in fact, what we do now for our evolution experiments for evolving these strains is we actually run them without the estradiol and there's a tiny bit of background scrambling and a little bit of leakiness. And then there's another knob we can turn is the promoter that's driving the creep. So we're pretty confident that we'll be able to handle that one. But it's definitely a valid concern. That's fine. It's a question for Jeff. So as I understand it, Biostudios is a software to design genomes both in an automated and in a manual way. And my question is whether it has been, I mean the design software has been designed in order to be used beyond the SCP project. Is it flexible enough? It is. In fact, Tom has been very frustrated because he hasn't been able to get it to run on his computer. But I'm happy to tell you, Tom, that we now have an AWS instance that I think can be working anywhere. So we're happy to send it to anyone. It is completely generic. It's based on Gbrowse, which is a generic browser. Or you can do Jbrowse, but we mostly use Gbrowse. So it should be quite generic. And we've even done circular bacterial genomes on there just as a demonstration. That's the question here, maybe? Oh, actually, sorry. I forgot. There's been a request to have the question repeated. So I apologize. So from now on, we can repeat the question before you answer it. Thanks. So Jeff, you had mentioned 21 fitness conditions. But I was wondering for maybe genome or chromosome 3, if you look at kind of transcriptional profiles, like essentially profiling, any kind of genetic profiling, things like that, do you see it maybe doesn't matter. But I'm just curious whether you see signatures that the places you change, there are changes. Right. So the thing is asking, did we do transcriptional profiling or other types of profiling? And of course, we did. The vast majority of cases, there's essentially no change, no statistically significant change to the expression other than sub-telomeric genes, quote unquote, that we deleted, which are no longer there. So they come way up on the volcano plot. I was telling you earlier, we have done more limited proteomics on some of these strains. And again, occasionally, like the one I showed pre-4, we turned something up. But it's probably on the less than 1% of the time level. Yes. I had a question about the bacterial CRISPR. So you deliver these systems using bacteriophages. So have you observed any resistance coming up against, not against the CRISPR, but against the bacteriophage delivery itself? Yes. So you would definitely expect to see different kind of resistance to these strategies. So you could definitely resist to the CRISPR by introducing mutations in the target and also block the entry of the CRISPR by just mutating the receptors of the FH capsids. We haven't observed that so far. And the reason is quite simple, is to these events you expect them to happen, but at a relatively low frequency. And in our case, we see survivors at a much higher frequency than the frequency of mutation. And the survivors we observe are just cells that did not receive the CRISPR at all. And that's just because even if you deliver 100 times more phage mid particles than cells, you will always have bacteria that just don't receive it. And we also observe that there is actually some variability. We would expect it to follow some sort of Poisson distribution, but it actually does not. And it indicates that probably some cells tend to have more receptors than other cells in the population. So there might be some variability that also explains why some don't receive it. Actually, I have three questions. So first, the neochromosomes, are the real chromosomes like linear chromosomes with telomeres and ars and centromeres and things? Why don't you give them one at a time, if you could question them? Do neochromosomes have arses, centromeres, and telomeres? And the answer is they always have arses and centromeres. And we can make them either in a circular format, in which case they don't need telomeres, or we can make them linear. We described a little tool we called a telomerator, which is a little device you can insert into any circular chromosome. And then when you cut, it reveals two seed sequences and it very efficiently establishes a linear chromosome. OK, and then the second question was on the ribosomal DNA regions. You said that they insulate the chromosome regions. And I wonder if you could tell us why that could be, or how that could be. Oh, right. Yeah, so the question is, how does the ribosomal DNA kind of insulate two domains of the chromosome? And I think the nucleolus is a gigantic structure, and so at one end, if you will, the DNA is coming in from the centromere. And at the other end, it's going out to the telomere. And those two sites are not close to each other at all. They're very far apart, in fact. So as a result, those chains are really separated from each other by, I think the nucleolus is estimated to be about close to a third of the nuclear volume. And so I think that's what makes it essentially behave like two separate chromosomes from that perspective. Then related to that, you showed that on other chromosomes, you would have potential insertion sites for the ribosomal DNA up front. And what is a potential insertion site? Any phosphodiester bond you want that doesn't disrupt the essential gene? So it could be. In theory. Okay, it wasn't nothing special about the two-sided pocket. No, nothing special. I had a question as well about the perembled genomes. So you see these copy number alterations. And so am I right seeing that most of those are like direct repeats in the genome? And then my question would obviously be, how about the stability of these things once you've selected them, just propagate them. Okay, so the first question is, are the repeats, tandem repeats? And the answer is they don't have to be, in fact, many of them are inverted or partially inverted, which is a signature of something called rolling circle amplification, double rolling circle amplification, which has been described for the two micron circle. And we also see what we call transpositions, where a copy just jumps in between, into another box piece site on the same chromosome. I see. And your second question was? Stability of such strains. So we haven't really studied that stability of the scrambled strains very extensively, but what I can say is that it's stable enough that we can determine an unambiguous genome sequence from short read sequence. But that's about all I can say. So they've had like 30 generations? Yeah. 30 doubling? Yeah, but of course that doesn't mean that somewhere in there you didn't have small clone that underwent something. We do know that if you pick the little pollinings, they are often very phenotypically unstable. So there may be some ongoing instability there, perhaps from dicentrics or something like that. So we have decided to leave those alone. Yes. I did not understand what you said about murdering human disease and pathway transplant. This is fine. And the attribution of patient disease is all that was erroneous. What I didn't get. Right, so the question is just to clarify what I said about patient mutations. So there are many, many case reports in medical literature correlating a neurological or other disorder with the existence of a patient mutation in some gene, right? Because the person found this mutation and they think it might be associated with the disease, but they're very descriptive. And so we've actually were two for two on this. There were two reports of mutations in this pathway. What disease? They were rare, very rare complex neurological syndromes in one case and I can't remember the detail of the other one, but it's not, it's something that's arisen twice or once and the data were limited. And what we did was we moved that exact mutation into the yeast and it's still, we cannot distinguish a change in the phenotype. Okay, but that is one explanation. Don't you think, because this happened several times, especially I don't remember the details, but in mouse. And it turned out that physiology, the pathway of the mouse is different. So you cannot prove the phenomenon, the mutation, or the defect because your choice, the choice of your organism is wrong. And then I think the way to cure this is to humanize the whole pathway. And then you find what you're like. Right, so I totally agree with you, first of all, but we did humanize this entire pathway. We humanized and did every step from starting materials to ANP, every step was catalyzed by a human enzyme. That being said, it doesn't mean that, for example, the protein doesn't fold exactly the same way or something like that. So it's not by any means definitive. But it was surprising to me that zero out of two cases did we recapitulate any defect. Question here? So I have a question for Jeff. And I apologize if I mis-seed or misunderstanding something in general, but I was reading about aging and it says maybe it's connecting to a stability. So did you look at maybe your history and could live longer or are there going to impact it and quantitatively or it's going to? The question is whether we've looked at the aging phenotypes of any of these strains and the answer is we haven't. But it's a great idea. Yes. Could you please make a comment on the choice of the recipient strain you used to make these synthetic chromosomes? You do look to answer that. Okay, so question is what is the strain we chose? And the answer is BY4741, which is a special favorite of mine because we constructed it in our lab. But it's also isogenic with the strain that was sequenced. And so our whole, you know, we're building on the shoulders of giants, okay? All the sequencing that was done was done on strains very closely related to this one. And we didn't want to take a chance on, you know, drifting to another sequence and running into problems related to genetic background. So that's really the main reason. It's a totally horrible strain for doing anything practical in the industry or anything like that. And so you described the lack of chromosomes rearrangements in natural isolates of yeast. And how does that get together with the observation that tRNA genes and transposons are a high recognition? So why did you take them out if you don't have recognition? Okay, so, right, I can see why you're asking the question. So why did I say on the one hand that tRNA genes are hotspots for genome instability, but there are hardly any natural translocations? And the answer is that there are many studies where people have done evolution in vitro, in chemostats and turbidostats and things like that. And when you do that, you often have translocations or duplications and a vast percentage of those are due to change, they occur at tRNA genes. So that's where the data comes from. The natural ones also, some percentage of them also occur at tRNA genes. Yes. Oh, sorry, Tom. I've got a question for you. Thank you for you, actually. So you constructed this flip-flop, this genetic flip-flop system in New York and you remarked that they rise to kind of an epigenetic mechanism for... Right. May you look at some of the interaction networks that actually occur so far? Do you see similar things? Does it have a similar flip-flop type of... So the question is, the flip-flop toggle switch style network which we used in Houston has been used by others in bacterial systems. Do you see those in natural networks? Yes. Yes, they are seen. I don't think they're seen in particularly large numbers exactly in that orientation where it's two things repressing one another, other than maybe in the land-of-fabe switch is a common well-known example. If I remember reading... There's one case in East, actually. There's one case in East? Yeah, pseudo-hyphal switch. Okay. Not the major gene, but the two ones before. Is it a bi-stable refresh? I think lack of the whole thing. Lack mono, but he never spoke. Okay. Yeah. Okay, my question. Please. So both of you talked about gene essentiality and deleting genes. So you could do that with CRISPR and do a screen and scramble can do it. You can get down to minimal genomes. And previously with yeast, people have done knocking every single gene out and now knocking pairs of genes out to look at essentiality as well. But I think I've read recently that it's not a black-and-white system that someone should... I think it was in yeast that you could knock genes out and then depending on how long afterwards you ask whether the cells are getting on, they can kind of recover. So a gene that should have led to an essentiality if you let the cells have some time and then the cell can survive without that gene as long as it's got a period to adapt. So the only thing I can say regarding the... I think CRISPR screens for sure are not going to be very well adapted to see this kind of phenotype because you do things in a pool. You only have a certain dynamic range of depletion you can observe and if something is depleted too strongly at some point it just disappears from the library and you cannot see if it recovers later. So yeah, I think probably for this kind of thing you will still have to go manually or maybe with some eyes to put like Jeff is doing, but... Yeah, I heard a talk on this at the east end this summer. It's quite interesting. I think the scrambling system could be a good way to accelerate the discovery of genes that allow bypass of that special class of essential genes that's only essential in the absence of some other generally changed. So I'm actually very interested in that topic. I think a related question that's much harder to answer is how many different ways do cells die? Jim? Can I follow that up? Because it seems to me that the dynamic aspect of networks which is kind of what Tom's talking about with the metabolic networks which can dynamically adjust to different conditions after perturbation and find different stable states is characteristic and we heard talks about that earlier in the Mayan cells. And there's also the issue of chromatin-based auto-assembly or auto-regulation for epigenetic control as well, which is clearly, you know, these scrambling experiments suggest this is what something like that is going on in the yeast genome. Have you got any plans to implement probes for that kind of behavior? I'm thinking, for example, a D-Cas9 type of approach is to recruit chromatin modifications or something like that where we can create specific modifications and then potentially follow changes that they might propagate. It's going to have a hard time repeating that question, but I think part of your question, which was also asked before that I didn't really address was do we need to worry about, are we recapitulating the epigenetic states that would be found in the native genome, right? Well, that's more the mechanisms involved and how you get at those in a more direct way. Right. So I think the general question is really important, especially as we look towards modifying more complex genomes where we have DNA methylation and histone methylation complexity that we don't have in the yeast. In the yeast, we're very fortunate in a way that it's a very stripped down system and telomere silencing and HM locus silencing are very well known to recapitulate the naked DNA within the time it takes a column to grow. But obviously there's a lot more to be learned on that, so we are very interested in introducing other forms of epigenetic regulation from mammals, for example, into yeast. But yeah, the CRISPR would be an interesting way to go. We haven't thought about that. Do you have any thoughts on that? Tom? Yeah, it'd be, you know, we thought about that when we started first working with talofectus. Maybe, you know, and other people have now gone on and shown both with talofectus and decas9, the recruitment of silencing and these kind of things. And that could be part of creating, you know, synthetic sub-telomeric regions with metabolic pathways that we can control the silencing of. We'd be to have the more coordinate regulation in that region of silencing and activation. I have a question about that. The tal, the vector stuff you talk about, it looked like it operated over a range of tenfold like there was a high background. Yeah. So is that a real background or how are you going to deal with that? So that's a real background, but part of it is, so we have improved on it in the latest version. I think it's more like 50-fold. But it generally comes from the fact that with intentionally this promoter system we're using is a weak promoter because when we overexpress talofectus, we want to use this system to chain talofectus in logic systems together. When we overexpress talofectus, we get a lot of stress on the cells. We're trying to work with kind of a weaker system and this modified PFY1 motor is a lot less than something like GAL, which is a great on-off. Can you take, sorry. Sorry, there's been a couple of people waiting, general apologies. In the yeast genome, there are many genes that are convergent overlapping. Did you maintain the overlap or did you separate the open reading frames? So the question is, how do we handle overlapping genes? And it turns out when you look hard, the vast majority of overlapping genes involve a so-called verified open reading frame with a so-called dubious open reading frame. And so for example, we had many instances where there was a TAG code on that we had to recode. So we applied the rule, always change the dubious work in those instances. Once you go beyond those instances, there are very, very few bona fide overlapping genes. There was one that was really clear cut with two verified works overlapping and we had a TAG and you couldn't change it without changing your amino acid. So what we did was we went to nature and we looked at whether there was anything in nature that had a TAG, the TAA mutation. And sure enough, there was one, so we'll just move back to it. So Jeff, I had a question getting back to the humanized metabolic pathway. I was thinking back to Sven Ponke's talk where he was talking about how to reconstitute tenants and kind of having to minimize it to the point where it was all in vitro. So I was thinking in your case for the humanized system, was it just coding regions? Or in the future, if you were trying to reconstitute it, do you think you'd have to tweak promoters and things to get levels and so on? Right, so the question is, when we did the pathway transplant, how do we do that? And that's a slide I took out. So we took the human or encode unoptimized it for yeast and then we assigned to the human or the corresponding yeast promoter and terminator. So the regulatory sequences are yeast, the coding sequences are human. And of course, I mean, that's a limitation and you might wanna play with that in the future, but unfortunately, human promoters generally don't just work in the yeast. So that would take a lot of it, and it would take a lot of it. Jim, did you still have a question? Yeah, it's a question of whether can you maintain two chromosomes, identical chromosomes that might be in different states. So clearly you've got loxites taking each gene to the idea that you might override local control by a D-S9 with a repressive domain targeted to each gene on a whole chromosome. So you could have active chromosome, silent chromosome and flip back and forth between each one by induction, which gets back to this idea, making substantial perturbations which you then follow the consequences of that perturbation. That's a really cool idea. So you're saying target the loxp sites with the D-C9 and then you make essentially a disome or you have one native chromosome and one science. Yeah, you've got roxites as well, so you could have chromosome three A and B for example. Have to check and see if there's any PAM sites in the loxp site. Yeah. Yeah, have you made diploids of your strains and do you see any, like, comments? They do recombine, myotically. And actually that's the problem with trying to combine multiple synthetic chromosomes to get all these patchworks out. But we come up with a way to homozygose the synthetic chromosome. So that basically becomes a non-problem. But they definitely recombine. We're working on building recombination maps to see whether the frequencies of recombination along the chromosome are similar to or different from the native. Another question. I remember I heard sort of scuttle, but on the Blatner stream where he removed all the transposons. As soon as you introduced any foreign DNA, you basically sort of start contaminating the genome back again with transposons. A, does the same thing happen in yeast? And B, have you seen it? Okay, I don't know about the scuttle, but I hadn't heard of that. But in our, so the question is, when you bring the transposon back in, will it be invaded? Yeah, or are transposons hopping from your native yeast chromosomes into your synthetic? Okay, so to that point, do the transposons jump from the native to the synthetic? That was one of the other reasons we made the party chromosome, because as we build the individual chromosomes, we take out the tRNA genes so all the targets are gone, because those transposons love to go near tRNA genes. So that keeps them clean. They don't go elsewhere? Well, they will go into the other native chromosomes. But they won't go elsewhere on the synthetic. They always go to the tRNA side. Well, never say always. Yeah, yeah. Yeah, so the worrisome part is what's gonna happen when we start reducing the number of native chromosomes, because these transposons have copy number control systems, and they're gonna start getting really antsy and jumping like crazy. So, but we have a plan to outwit them, which is we know all about the transcription factors they love, and we're going to destroy one of those to incapacitate them when we get to that phase. That's the plan. Okay. Any last questions for almost out of time? Okay, perfect. Then thanks very much to all of our speakers.