I just wanted to mention Leslie Orgel, who is a giant in this field, and it's actually

very, very hard to think of anything in this area that Leslie hadn't thought about or thought of in some way or other, and I think also as the theme of this morning session I think I would like to quote Orgel's second law, which is evolution is smarter

than you are. This is really nice, okay. Well, there's one question. I have a question for Philippe Marriere, an intellectually demanding question for me. How do you know which mutations are adapted or relevant when you have 6,000 mutations

in your R1, 8,000 in R2? So one heuristics you actually propose in some way or suggested is to look at the overlap, so you do more experiments, look at the overlap. If you have in the end 20 mutations that are overlapping, it doesn't mean that they are sufficient, so you could cleanly put them in a new strain and see how it goes,

but you probably require some of the other mutations that are found within the 8,000 or 6,000. Do you have other heuristics to find which are adapted or relevant? I completely agree with you. We are so happy to see that many mutations.

You know why? Because we thought of referees, if you find only 15 adaptive mutations, they are going to ask you to take each of them and do that in a separate stream, 6,000. No journal ever nasty and pharmacophobic and whatever will ever pass.

You must have found ones that made some sense, though. DNA binding proteins, dah, dah, dah, dah. Definitely. The replication apparatus is under fire. RNA polymerase is under fire. The repair system is mostly noble enzymes of nucleic acid, but not so much nucleotide metabolism.

There is thymidine kinase also is a target of mutations. Is it HNL and other proteins that bind DNA? Well, some of them are. Yes, yes. But polymerase, in each lineage, we find that polymerase. Now, I don't think that we are going to really study, because in a trajectory like that,

you find interesting things all along the way, but our main target is to really leave the solar system. All five polymerases? I'm sorry? All five or just three? Mutations?

All three, yeah. No, no, no. More than that, you have now up to ten mutations in polymerase. I know. In all five, in how many different polymerases? Just pull three or one in three? No, no. Several. I think it's three out of the five, yes. The question, I think that maybe one has to go from the chemostat or better, from

the tortostat to the gravostat, because I think this is clear that gradient is better. You can distinguish spatial gradients. Some parts of it might be ice and some parts might be not ice. Then we get coexistence via niche effects, and this creates a lot of interesting patterns.

Especially, you showed that picture there. It's very familiar. There, if you have really environment involved, then you get crazy patterns. Yes, there are also unexpected ways of helping evolution.

For instance, having predation. You know, if you have predate, you go faster, even in a completely orthogonal way. Yes, you pointed out an interesting situation where the direction of mutation went differently

after you finished the initial adaptation phase. So you went A to G in one culture and B to A in the other. Yes. Could that be explained by the acquisition of a specific mutator mutation in one of the two lineages? Actually, we tried to see whether there was such a pattern, and we did not find

it. There is no clear ... Actually, because as of sudden, as soon as it was only accelerating and the division generation time was no longer imposed, it has to ... We saw this trend at the turn of the evolutionary regime.

Clearly, we already don't understand that. At the transcription side, you have mispairing, but it's also possible. I was curious if you saw more mutations in the transcripts, but also how you would

tell that it's from the RT and so on. We had mutations in the RPOS and the transcription machine in RNA polymerase. Now you no longer have TATA box, but CLA-CLA box and stuff like that. So RNA polymerase has to adapt to that, but we did not do a thorough transcription

analysis of the diversion process, so we don't know. I was wondering, almost the cell remains able to execute a browser program, for instance, a phage or something like that.

Yeah, yeah. Since they are always growing, they no longer have a stationary phase, so you can freeze and thaw them okay, but you cannot just ... E. coli is probably the model organism because you can just let it on the bench for the weekend, and then you come back on Monday morning,

and you can resume your experiment. Now they die very, very quickly, so everything that is not to be maintained during the evolutionary process is lost. Phages, they are less sensitive to, I think, T7.

T7 was tested, and T7 seems not to infect it as well as previously, as a white type. A kind of overlapping question, what is the genetic stability of these mutations? How stable they are? Well, well, well. They are as far as we can study because it's difficult to just have

an isolated strength from that and study it because they die out and so on. So we have problems of reproducibility, but we don't know whether it's purely phenotypic because they die or whether it's genetic. We don't have indications that they are very, very mutated, for instance.

They are steadily mutated. They accumulate the order of one mutation per cell, per generation, which is high enough because we will reach salmonella in the foreseeable future and so on. But that's what we can say. Please. So I don't ... maybe this is ...

As I understood yesterday's talk, you're looking for new pairs that are stable. That's part of the criteria you're looking for. What if you look for ones that were stable? You'd have to do a co-selection on A and B or X and Y, whatever you want to call it. But where the cross-hybridization with the naturals are not stable,

you could enforce speciation. Absolutely, definitely. And that's actually one of the nice goals to consider, is to try to effect this speciation through exaculation. Yeah, I agree. Yeah, yeah. I agree. Good, thank you. I have also a question. Sorry.

Now you're turning screen. All right. Please. In evolutionary biology, it's very often the case that if you have these alternating environments, that you get special phenotypes that allow themselves to adapt this by switching and pathaching. Do you see that? I cannot say that we have other media in which we could test them and so on.

We saw that they were very, very unfit. I don't think we can answer it. You mentioned the thymine became toxic.

Yes. Have you considered running your experiments? Backwards, yes. Yeah, yeah, yeah. You see that the cockwheels are turning in the right direction in your audience when people ask this question. I don't think that we are far enough.

I mean, at the beginning, you know, we would put it on thymine and see a reversal mutant pop up. They no longer pop up. So we are not far enough. But time will come and we will definitely do that. I think you just pick up too many compensatory mutations and it just becomes a one-way street.

Yeah, definitely. Question for Phil Holliger. You rightfully pointed out the importance of local concentration effects in these stories and scenarios. I was wondering whether it has been explored that small oligonucleotides that have the possibility to pave a surface or a volume would allow for the concentration of other or same oligonucleotides

that would have some catalysis properties and would at the same time protect them from dilution.

Yeah, I think surface effects have been explored by some people, but simply by tethering oligonucleotides. And yes, obviously you can then go through concentration and water cycles and you can overcome product inhibition to some extent. One of the problems of coupling things to surfaces is that at some point things need to come off the surface again.

So you need to have some sort of regime which will then strip the surface so the cycle can start again. We haven't explored that, but some people like von Kiedrowski for example has looked at things like that.

And certainly there's also simulations which show that you need compartmentalization as a prerequisite for Darwinian evolution, but under certain circumstances just local surface patchiness might be enough to isolate, for example, a replicator from parasites.

So yeah, I mean these are interesting ideas, but we have not explored it yet. Phil, I have a related question. First of all, very nice talk. And did you ever check whether pH changes also play an effect in these protective phases?

Because it's known that ice crystallization or pure crystallization, certain for example in the lab buffer components, would crystallize later or earlier and then this might affect then the pH in these new phases. Yeah, the pH changes by about 0.5 depending on the buffer.

It changes by about 0.5 units, but when we try to compensate for that literally nothing changes. So I think the pH change is not the dominant effect we observe. I think the dominant effects are the temperature change, the concentration, and possibly, although this is very hard to prove,

there's surface effects of the ice crystals because there's a very, very large surface that is generated as the ice crystals grow. So we were just discussing what historically was the temperature situation back in the day.

And also related to that, I assume most of the ice is salt water ice. So are your conditions closer to what you would have by freezing salt water ice or fresh water ice? Okay, so the first question I can't answer and neither can anybody else because I think there's no rock surviving from the very early days on Earth.

I think there are estimates kind of based on so-called CERCOM inclusions which give you only very average temperatures. But I believe the consensus, although I'm on thin ice there because I'm not a geophysicist,

I believe the consensus is that the planet acquired a hydrosphere very quickly and cooled to about a global temperature of approximately 40 degrees within the first 100 million years. So assuming that and actually the fact that the solar output was only about 70% of today,

I think it's certainly not far-fetched to propose ice maybe at the poles or seasonal kind of at higher elevations. I mean, I would say, but it's impossible to be totally certain about that.

I think the conditions of our assays are mostly compatible with fresh water ice. And what you'd find is that the high salt concentrations that you'd find in sea ice are mostly incompatible with not just nucleotide chemistry but also assembly of membranes or generation of amino acids.

So I think all of the sort of most credible scenarios for generating these things need to take place I think outside the ocean. I think there is some chemistry which has been proposed to occur at these various types of hot vents.

But it's not necessarily, you don't necessarily have to have the origin of life where you make all the stuff. I think some of these chemistries kind of are okay as foundries sort of where you assemble all the bits that you need to build the car. But I think the car needs to be built in a more benign environment.

In the beginning of your talk you showed this nice picture of Campbell's primordial soup. Do you have any idea how diverse this soup must be before the auto-assembly thing of your...

Yeah, I'm not a prebiotic chemist. So I think there's various people have advanced various scenarios but there is now a number of I think very credible studies which yield a number of amino acids and lipids.

And certainly some of the nucleotides from things like HCN and cyanogens and acetylenes. So basically highly reduced components which you find for example in comets and in outer space highly abundant.

So presumably given the probably almost incessant bombardment from outer space on the earlier, certainly these molecules would have been present. But this is sort of outside my expertise a little bit.

I mean what I would say is that there's credible scenarios to take you to nucleotides. Some other people for example Chuck Shostak and David Diemer and others have shown some credible pathways to go from nucleotides to oligomers. And really kind of this is where our work would start.

But presumably this is pretty thorough cycle. You would create different combinations of oligomers every time. So you would need some different amounts of diversity I think to start with. Yeah, I don't know actually. I mean I think that's a very good question.

What would be the minimum amount of diversity that you would need to start with through iterative recombination to build up the sufficient... ...diversity from which phenotypes could emerge. I think these are open questions. I really don't know the answer to that. There's a question in your abstract. You mentioned the thing that it might be an accidental freezing of the actual set of...

Yeah, I think the frozen accident refers to the chemistry, not so much to the process. I think, well, again, like everything to do with the origin of life, this is controversial.

But I believe that the chemistry of life is built, is basically opportunistic. So life got started with building blocks that were abundantly there given by Earth's pre-biotic chemistry. And once you've made that choice, it's very hard to change. As Felipe has shown, it takes a lot.

Could you use this idea to get something going in synthetic gravity? That's right. Well, I think, as I've shown, in principle there is probably a number of backbones, of polymers that are capable of heredity, genetic information storage, propagation and evolution.

And certainly, probably not just backbones, as we've seen yesterday, there are probably different base-pairing schemes to encode information as well. So I would say there is a large but finite number of ways to store and replicate information.

And I think in time we might be able to explore it. Just to be clear about that, to my knowledge there's still only one biopolymer that's capable of amplification and heredity, and that's DNA. No, thiod DNA, four prime, thiod DNA is also able to... Okay, but we talked about still using... Well, RNA, this RNA virus is clearly capable of heredity and evolution.

Okay, so DNA is predominantly RNA in some cases, but just for the language to be clear, I think you should be careful about calling those genetic systems in heritability. Because you still are using the DNA in the amplification. And amplification is the huge component of...

No, no, I agree. I don't call them genetic systems, but I think heredity as a process of information storage and propagation is a nice sort of short form of kind of saying. Yeah, but at some point things become short enough that people misunderstand them. Fair enough. Yeah, but XNA polymerase, XNA, it's only a matter of time.

Sure, but it does require... But it's not yet here. What's that? They are not yet available, but there is... I will not ask the question when they are. Perhaps I wanted to make a comment about spontaneous evolvability and so on.

You know, you find, you read it very often, you know, people like to see star evolution, then chemical evolution, then biological evolution, as if it was a single process, and presumably it's not true. When heredity kicks in, you have a completely different regime, including generating

people who think like us, and can transform and morph completely things. So in terms of implementation in synthetic biology, I would advocate for going as artificial and improbable and non-spontaneous as we can, and not the other way around.

For setting up life, you know, start again, because actually the specification that you have to respect is enormous. If you really want to close the gap between some chemicals or ribos or whatever, and real life or evolving system, you have to...

Not only it's a matter of chemical design, but you have to follow all the continuity of these logical steps, which is daunting. Now, with organic chemistry you can simplify a lot, because you can arrange design systems in non-aqueous solvents and so on.

And implementation might not have all these constraints of continuity and logical design, stepwise. You can make big jumps. So you engineered E.coli with a new material metabolism, and it has a single growth rate after 40 days.

I was just wondering, if you mix that population with the Y-type, what would happen? Would it auto-concrete? Would there be two populations at the same time? Actually, one of these lineages was evolved long enough, you know, in turbulence, as was done in Berlin by Otto Hundsen.

And actually, at the end of the evolutionary process, he made competitions with Y-type, and it grew faster. But of course, it was not such a demonstrative experiment, because he did not adapt Y-type E.coli under the same conditions,

so it should be the outcome of the same competition independently to measure that. I tried to allude to that earlier with this Turing test, you know. It is confronting adaptive challenges that we can already measure, improvement in availability and so on.

Such experiments, this is the future, you are right. But at this stage, these machines are not too costly, but we have a limited number of them. The first objective is to reach the outskirts of Mother Nature ASAP,

and then we play competition games and stuff, but we are not yet there. And it is a matter of resource allocation and so on. About this Turing test, is there something, is there a Turing test as a kind of theory?

Isn't that going to be innate and natural? Well, as I said, then you can use competitions, you know. You can adapt Freud's strain and put it in competition with Evolvit,

so that the bases feel good, invade the genome for a short while, do the same with Ishiro's strain, then put them in competition. At some point, you know, this kind of experiment, there will be different, among the different bases that you mentioned, you could put the bases in competition for invasion of E. coli, you know.

The races are not only between organisms, but you could make, you could use organisms to sort out the benefits, the functional benefits of different bases as well. Actually, I'd like to invite some questions from the audience of this idea of the Turing test. Has anybody thought about when we decide that synthetic biology has become real biology or something?

It is real biology. We don't need the test. No, it just depends on what you're interested in. If you're interested in solving a specific question, then one can do that.

I think what we're referring to here is how do you tell whether something's gotten as good as nature produced? Yes. The situation reminds me very much to the problems we have to do with designing new algorithms. If you want to sell it to industry, they say, we know yours is better,

but we know ours to implement that. It costs a lot of money and we know what it makes wrong, but we don't know what your scheme makes wrong. This maybe is an obstacle. Life is algorithmic chemistry, definitely. When you talk with computer scientists about this kind,

they see organisms as competing programs. I think it's very pertinent, this way of thinking. Now, you can do that in silicon, the kind of computing that you have. The evolutionary process itself has a kind of statistical programming and so on.

We don't master for money. You're obviously pushing very hard on this idea of completely synthetic life and building new chemistries and algorithms and new solutions to what we have now. How do you think society and the public would think about that?

Because, as I say, in this room, we're all kind of scientists. Don't tell anyone, yes. My son, it's a bit of a problem for him. He seems to be permissioned, saying, we're going to make life that you've never seen before. It's not even following the rules of life. How are you going to do that? You're so right, I don't know. People look at domestic animals.

They like them. You see these chamole, cattle. They look so happy. They are completely genetic variants. They are morphic. They no longer look like rocks or stuff. We would like to make them like xenobiology creatures, like their pets. Pets are horrible monsters.

I totally agree with you that it would be super. We are thinking of having this big European brain, the graphene, larger than life. We are going to use everything, all competencies, all kinds of different skills and so on,

so as to make a second nature that will be very respectful, of course, of everything, gender, things and stuff, but completely artificial, so as to help the industry. I think it's easy to convince technocrats of that, but the public is just…

I think maybe, to me, the most interesting thing in varying the chemistry is to understand what makes the chemistry of life special, potentially, or not.

Turing test analogy. You said something about, once we get to a certain point, we have to contain things ironclad way, but it wasn't clear to me. I mean, if you select for something that has a gain of function, that outperforms wild type,

and you've been doing that for a long time, but almost every time when you do that and then you test them under some other fitness condition, it never can compete with the wild type, so is this really a problem? Yes, I think it is. You remember Jurassic Park, you know, life finds a way, right?

And you know, for chloro uracil, I would not bet on the fact that E. coli will never find a way to make chloro uracil and become prototrophic again. You know, if you treat with hypochlorite cytosine or whatever, you get chloro uracil in the juice

or with some chloroperoxidase and stuff, so I think chloro uracil is accessible, it's within reach. So if we think along this line and we want to contain organisms, we will have to go a long way. It's not going to be easy, and scientifically it's going to be very demanding, actually,

to get to a stage where we can make sure that there are something like 10 to the power 30 genomes at work, cellular genomes, at work on Earth. Well, that's big. That's not that big. So we might be clever enough to find chemistries that the whole biosphere could not tame or master itself.

I think it's a big issue to go in the chemical reprogramming of life beyond the search that the biosphere alone is able to do. But of course, I know that I'm here in a mathematical institution.

It's a question for mathematicians. How can we measure the search of what is searchable by the biosphere and with our little knowledge, skills and whatever,

how can we say at this point there will be no way of having spontaneous evolution to make these Jurassic Park monsters of ours? Isn't that also not only a question of the conditions that life evolves into?

It's also about what you're doing is creating artificial conditions that life may adapt to and it doesn't have the need before to go there. That's right. But we should not be too optimistic on the fact that they are de-evolved

so that they were just in some plumbery and that they will never re-adapt to the wild world. We don't know. We must be pessimistic about this. I think that there is another strategy to reach a bit of a cycle which is not to change the chemistry but to change something else.

That's a concept called life. Definitely could. Can you comment on that? Do you think you can really reach the same level of…? You know, proliferation is serious business. When you have cattle fed with beef, you get triumphs in the cycle. So proliferation, when you have proliferative latency,

having a cycle, a nutritional cycle and so on, will lead to proliferation and improvement by natural selection. We saw it in many infectious diseases, prions and so on. I think it's a matter of nutrition for containing

because if it is only a matter of any entity that is able to take things, to assemble them and proliferate, if it can find them in the wild or make them through some chemical, metabolic reactions,

it will invade its environmental. This is Darwinian selection. So to prevent that, even if we have different genetic codes and so on, it will not be enough. We have to prevent their material proliferation. So we have to make them absolutely dependent on nutrients that they cannot find or make.

I think it's a question chemically that is very well stated. They must not find their nutrients. Otherwise we will have prion life, whatever. And we have to be super pessimistic about it, not say, life will find a way.

For myself, I tend to come down more on this side. I think that related to the question that came up yesterday, sort of what new activities, looking at the case of proteins, could we evolve. I think those would be very specific activities that you chose specifically for a certain goal, like a therapeutic activity or something. And I think the question as to whether or not

something could grow to dominate in the wild, I'm not sure that anything, at least as I look at it, there's anything out there that is going to make things grow faster or be more adaptable in general. Nitrogen fixation is a shame. You know, it's one of my favorite stances. I show you metabolic pathways.

Some of them are nice, some of them are just... So I think there are some pathways that you could engineer where the argument changes then. Nitrogen fixation, if you make it easy, you know, and if it is made in the common genetic language, it will spread out. Well, so you'd have to have specifically the ability to fix nitrogen contained in one selectable thing.

Yes, but there is such an incentive. You know, nitrogen and so on, there are probably other ways to do it. Nature has found this one. There are many chemical reactions very easy in the industry that life has not found yet.

So introducing such capabilities like metathesis or stuff like that, you might provide incredible incentives in terms of selection in metabolism. Okay, the last question. It's more a question, more as a remark. We have been discussing yesterday the modularity. I think we have to take into account not just the chemistry.

We have a lot of other modular level. Biomechanics and the signaling is getting more and more involved. We have to understand it better. Therefore, I'm not against biochemistry, but only in the opposite.

But we should not forget there's more behind, especially biomechanics. And motility of species is very important, but which you cannot catch with the common chemistry. Well, when you look at a ribosome, it does very fine-tuned, superb chemistry, but you can also see this as a mechanical, robotic thing.

Yes, but it's a mixture. Yes, the fields are not so well separated. It's not separated exactly, but I'm coming from a modeling side. It's overlooked. We look too much on chemical reaction systems, not taking into account biomechanics,

which is an important feature of structure formation. Empirically, making covalent bonds is what we master best, as seen from very far away. We are very good at that. You heard yesterday these talks about elaborating completely base pairs.

It is very miraculous. Engineering biomechanics at the same level and doing physics in the same way would be much more difficult, I think, with the current knowledge that there is. Okay, with this, I'd like to thank both speakers again.