Session 1: Origins of Life: Protocells and Non-Enzymatic Template-Directed RNA Synthesis
Transcript of Part 3: Non-Enzymatic Copying of Nucleic Acid Templates
00:00:00.00 My name is Jack Szostak, 00:00:01.22 I'm a Professor of Genetics at Harvard Medical School, 00:00:05.01 an Investigator at Massachusetts General Hospital, 00:00:07.17 where my lab is, 00:00:09.02 and I am an Investigator of the Howard Hughes Medical Institute. 00:00:12.17 And in this part of my lecture, 00:00:15.00 I'd like to concentrate on the one aspect of the origin of cellular life, 00:00:21.28 which is the chemistry of copying replicating nucleic acid templates. 00:00:29.29 So let's begin by looking at our schematic version 00:00:34.20 of how we're thinking about a simple protocell, 00:00:39.06 so again we have a two-component system: 00:00:43.12 A primitive cell membrane encapsulating some 00:00:47.01 genetic molecules, could be RNA, could be DNA, 00:00:50.12 could be something related. 00:00:54.08 In the previous lecture, 00:00:56.11 we dealt with the growth and division 00:00:58.21 of the membrane compartment, 00:01:00.15 and this time we want to focus on the copying of 00:01:04.24 templates to make a duplex product: 00:01:07.00 The separation of strands, 00:01:08.26 and the copying of those in a subsequent round 00:01:11.22 so that you can distribute that information to daughter cells. 00:01:16.10 And this is an essential aspect of the 00:01:21.23 emergence of Darwinian evolution. 00:01:26.02 There's some kind of informational polymer 00:01:28.11 to code for heritable functions, 00:01:31.24 could be anything that's useful for the cell, 00:01:34.13 something that helps it to grow, 00:01:36.24 something that helps it to divide, 00:01:38.07 something to helps it to survive better in its environment, 00:01:43.01 almost anything. 00:01:44.29 But we need to have a way for that function 00:01:48.23 to be coded and transmitted from generation to generation. 00:01:54.19 Now, there are two general ways 00:01:59.12 of thinking about the process of nucleic acid replication: 00:02:03.27 The first would be some kind of enzymatic or catalytic process, 00:02:08.15 so the classical example of that would be an RNA replicase, 00:02:13.00 an RNA molecule that's an RNA polymerase 00:02:15.15 that's good enough to copy its own sequence. 00:02:19.11 I worked on that for a number of years. 00:02:24.18 Many people have followed up on that work, 00:02:27.22 and there's been a lot of progress made. 00:02:30.12 I do think that eventually we will have molecules 00:02:33.19 that can do that, but at the present time, 00:02:35.19 we're still far from such a solution. 00:02:38.12 And that has driven us to rethink the process 00:02:41.23 and to step back and look at chemical processes 00:02:46.06 that might lead to the replication of nucleic acid templates, 00:02:50.06 whether RNA or some related molecule. 00:02:54.11 So that's what the main focus will be on: 00:02:56.09 Chemical processes leading to efficient copying 00:03:00.07 and replication of templates. 00:03:03.05 Now, there are two really critical factors that apply 00:03:07.03 in any such system, and these are that both the 00:03:11.18 rate of copying and the accuracy, the fidelity, 00:03:14.29 of copying have to exceed critical thresholds. 00:03:18.10 So the rate of replication has to be faster 00:03:22.23 than the rate of degradation, and with a molecule 00:03:26.00 such as RNA, that's a really important factor, 00:03:29.06 because RNA is such a delicate polymer, 00:03:32.00 hydrolyzes relatively rapidly. 00:03:34.17 That imposes a lower limit on an acceptable 00:03:38.03 rate of replication. In addition, 00:03:41.07 the fidelity of that process has to exceed the 00:03:44.27 Eigen error threshold. If we want to propagate 00:03:47.18 useful information from generation to generation, 00:03:50.27 the accuracy of that copying process 00:03:54.08 has to exceed a threshold, 00:03:57.02 which is basically related to the reciprocal of the 00:04:00.10 number of important nucleotides for whatever 00:04:04.13 function we're talking about. 00:04:06.21 In practical terms, that means we probably 00:04:08.24 need to think of accuracies or, shall we say, 00:04:12.07 error rates below a few percent. 00:04:16.16 Typical chemical processes that we study now, 00:04:19.01 the error rates are in the range of 5% to 10% or 15%, 00:04:22.22 so some improvement is required. 00:04:25.08 And we would also like to see improvement 00:04:27.12 in the rate of replication, so that we can do these 00:04:30.16 experiments on a reasonable laboratory timescale. 00:04:35.04 So, the building blocks that we're going to use 00:04:37.11 for these experiments are quite different 00:04:40.04 from the nucleoside triphosphates that are used in all modern cells. 00:04:46.07 So these are modern substrates. 00:04:48.22 They are kinetically trapped in a high-energy state 00:04:53.07 so that they require very sophisticated catalysts, 00:04:56.20 enzymes that can confer the 10^12 rate acceleration 00:05:00.21 that you need in order to make effective use of this kind of substrate. 00:05:06.12 They are also, as I've mentioned before, 00:05:09.00 highly charged because of the triphosphate group, 00:05:13.23 which keeps them from leaking out of cells, 00:05:15.26 which is a good thing for modern cells 00:05:17.13 but a bad thing for primitive cells, 00:05:19.26 which will require their substrates to come in 00:05:23.28 from the environment, get across the membrane spontaneously. 00:05:27.24 And that means we would like to think about less polar substrates. 00:05:33.12 And so that drives us to think of molecules 00:05:36.11 similar to that that you can see down here. 00:05:39.23 These are nucleoside phosphorimidazolides. 00:05:43.05 These kinds of molecules were first synthesized by Leslie Orgel 00:05:47.22 and his students and colleagues, 00:05:50.16 and studied in quite a lot of depth in the 1970s and 80s and 90s. 00:05:58.07 These molecules are much more chemically reactive, 00:06:01.18 we have a much better leaving group, 00:06:03.14 so they're intrinsically more reactive. 00:06:05.19 They can spontaneously polymerize and copy templates 00:06:10.14 without enzymes, and they're also less polar, 00:06:16.04 which means they can get across membranes 00:06:18.17 much more rapidly, without any transport machinery 00:06:21.28 being invoked. So, the early work by Orgel 00:06:29.11 and his colleagues got us partway to copying systems, 00:06:35.24 but there were a series of problems that they ran into, 00:06:38.18 which we'll come to and try to consider individually. 00:06:42.18 But first, I want to step back a bit and 00:06:45.04 think about how these kinds of molecules 00:06:47.08 would've been generated on the early Earth. 00:06:49.23 And that's really a pretty major problem. 00:06:52.04 It's still a major research area, 00:06:54.25 but I think very exciting progress has been made recently. 00:07:00.15 So in the early days, 00:07:04.22 there were self-assembly processes that were 00:07:07.01 discovered that seemed to suggest that 00:07:11.29 the solution might be really easy. 00:07:14.14 In fact, the classical formose reaction, going back to Butlerov, 00:07:19.27 showed that you could make sugars by polymerizing formaldehyde. 00:07:23.25 And so for example, ribose can be viewed as 00:07:26.25 an oligomer of formaldehyde. 00:07:29.12 Five formaldehyde molecules can self-assemble 00:07:31.27 in a series of steps to give you ribose. 00:07:35.19 The problem is making just ribose. 00:07:38.21 In fact, in this kind of chemistry, 00:07:41.04 typically you'll end up with dozens or 00:07:44.00 even hundreds of products. 00:07:46.09 So, part of the problem that's absorbed people 00:07:51.01 has been how to make just the right sugar. 00:07:54.21 A similar problem comes from thinking about the nucleobases. 00:07:59.19 So early on, Juan Oro did very dramatic experiments 00:08:04.00 showing that he could actually made adenine from cyanide. 00:08:09.05 Simply boiling a solution of cyanide gave you some adenine, 00:08:12.24 along with a lot of other products. 00:08:15.18 But it's very striking that adenine can be viewed 00:08:18.20 as a pentamer of hydrogen cyanide. 00:08:25.06 So again we have the same problem of how do we get just the 00:08:28.08 building blocks we want (the A, G, C, and U), 00:08:31.21 as opposed to all of the other related heterocycles 00:08:35.11 that come out of this kind of chemistry. 00:08:38.19 Now, when people like Orgel and many of his colleagues 00:08:44.26 started to look at the synthesis of pyrimidines, 00:08:47.11 again it looks, superficially, very easy. 00:08:52.03 For example, a cytosine can be viewed as the product 00:08:55.18 of reaction of cyanoacetylene and urea. 00:09:00.21 And in fact there are variations on this chemistry 00:09:02.19 that are extremely efficient. 00:09:05.19 So it was starting to look like you could make sugars, 00:09:11.04 you could make nucleobases, 00:09:13.22 maybe it would actually turn out to be easy 00:09:16.24 to make nucleosides and nucleotides 00:09:18.28 and get us all the way to RNA. 00:09:21.05 But it turned out that there was a major problem, 00:09:26.07 even apart from the problem of making just the molecules we want. 00:09:30.12 If you could have ribose and, say, cytosine, 00:09:35.03 you would need to join them together by making 00:09:37.18 the glycosidic bond that links them, 00:09:40.28 and that chemistry basically just doesn't work, 00:09:43.27 no matter how hard people tried, 00:09:45.20 this was a roadblock for a long time. 00:09:49.16 So one of the most exciting advances in prebiotic chemistry 00:09:52.29 in recent years has come from the laboratory of John Sutherland 00:09:56.01 in the UK, who basically followed up 00:10:00.08 on much earlier work of several other labs, 00:10:03.17 and showed that there's an alternative pathway 00:10:06.13 that can get you to the final product 00:10:08.24 without ever having to make this particular 00:10:12.05 glycosidic linkage by joining together a base and a sugar. 00:10:17.03 And the solution basically comes by making this intermediate, 00:10:22.05 2-aminooxazole, from cyanamide and glycolaldehyde. 00:10:28.10 In a series of very simple and actually remarkably efficient steps, 00:10:32.16 this intermediate can be elaborated into cytosine, 00:10:36.08 and then deamination can give you U. 00:10:40.06 So it looks like there might be a reasonable pathway 00:10:43.00 to at least getting to the pyrimidine nucleosides. 00:10:46.11 The synthesis of the purine nucleosides 00:10:48.20 by an analogous pathway 00:10:51.11 is a topic of active research. 00:10:54.03 And if it turns out that there is a similarly efficient pathway, 00:10:59.29 that will certainly be very satisfying in the sense 00:11:02.07 of providing very efficient and regiospecific chemistry 00:11:08.08 that gives us a restricted set of building blocks 00:11:11.12 leading up to RNA. 00:11:13.29 Now, there's still many gaps in our understanding 00:11:17.14 of how we would make pure, concentrated starting materials. 00:11:22.06 There are some steps leading up to activated nucleotides 00:11:24.23 that are far from clear. 00:11:26.08 So there's a lot of work to be done, 00:11:28.17 but I think this new chemistry has really advanced the field a lot. 00:11:33.09 So, let's skip those missing link steps for the time being, 00:11:39.01 and assume that we can make activated nucleotides. 00:11:42.27 The next problem we have to think about: 00:11:45.01 Is there polymerization into RNA chains, 00:11:48.08 and how could that happen? 00:11:49.21 Well, here, we're in a good situation, 00:11:52.10 because we actually have two solutions to the problem. 00:11:55.19 The first is the finding of Jim Ferris and colleagues, 00:12:02.23 including Leslie Orgel, 00:12:05.02 and this is a finding that a common clay mineral, 00:12:08.05 montmorillonite, illustrated here, 00:12:10.17 is a really effective catalyst for that 00:12:13.03 kind of polymerization. 00:12:15.05 So this clay mineral is a layered hydroxide, 00:12:19.01 aluminosilicate, and in between the layers 00:12:21.27 there are water molecules. 00:12:24.14 And organic molecules also tend to accumulate 00:12:28.03 in the inner layers of the clay mineral, 00:12:31.04 and as they accumulate there and become concentrated 00:12:33.15 and oriented relative to each other, 00:12:35.29 their polymerization is catalyzed. 00:12:40.13 And the nice thing is that this is not the only way of doing it. 00:12:46.09 You can get essentially the same result 00:12:49.29 simply by taking a solution of these activated nucleotides, 00:12:53.25 these phosphorimidazolides. 00:12:56.03 And as a dilute solution at room temperature, 00:12:58.17 almost nothing happens, they don't polymerize. 00:13:01.19 But if you put that solution in the freezer 00:13:04.28 and allow the water to freeze and generate ice crystals, 00:13:09.08 what you see is that the solutes, including these nucleotides, 00:13:13.00 get concentrated in thin layers in between the ice crystals, 00:13:18.02 and when molecules are concentrated that much, 00:13:21.07 even at low temperature, they can start to react with each other, 00:13:24.22 and you will see the spontaneous formation of long RNA 00:13:28.27 chains as a result of freezing. 00:13:31.25 So this is very nice, 00:13:33.21 because now we have two plausible, natural scenarios 00:13:38.25 where either a common mineral or just the process of freezing 00:13:41.25 could generate RNA chains. 00:13:45.21 So the next problem we have to deal with, 00:13:48.00 assuming we can make sets of essentially 00:13:50.26 random RNA polymers, how could they be copied? 00:13:56.05 And so here is where we run into a fresh set of difficulties. 00:14:00.29 So the partial copying of RNA templates 00:14:03.29 has been known for decades, as I said, 00:14:06.10 from the work of Leslie Orgel and his colleagues. 00:14:09.11 This is an example of that kind of chemistry 00:14:11.20 done by David Horning, when he was an undergrad in my lab. 00:14:15.29 We start off with a substrate here, a guanosine nucleoside, 00:14:21.19 activated as a phosphorimidazolide, 00:14:25.08 and we supply that to a primer-template complex 00:14:28.11 in which the template contains a region of Cs 00:14:32.28 where the G nucleoside can bind and result in primer extension. 00:14:38.24 So here would be the starting material, 00:14:40.27 and then over time we want to see 00:14:43.02 the incorporation of Gs to elongate the primer. 00:14:47.04 And that process is shown down here in timecourse, 00:14:51.12 where we start off with just the primer, 00:14:54.17 and then over the first, say, six hours, 00:14:57.21 we observe the incorporation of the first nucleotide, 00:15:02.11 and then over the next day or two, 00:15:04.06 we see the second nucleotide come in, 00:15:07.06 but even after two days, there's very little of the third nucleotide. 00:15:12.04 So that illustrates the first problem: 00:15:15.06 This process is intrinsically rather slow. 00:15:18.07 And because this chemistry requires 00:15:21.26 a very high concentration of magnesium 00:15:25.03 to catalyze the reaction, 00:15:27.15 this rate of synthesis is actually on the same timescale 00:15:32.05 as the rate of degradation of the RNA template. 00:15:36.11 So that's a problem. 00:15:38.24 There are other problems, you can see, 00:15:41.20 in the structure of the ribonucleotide, 00:15:44.16 that there are two hydroxyls on the sugar, 00:15:47.27 and either of them can react to generate either the 00:15:51.16 correct 3'-5' linkage or the incorrect 2'-5' linkage. 00:15:57.17 And so that means you inevitably get a mixture of linkages, 00:16:02.28 some of which are the natural, correct linkage found in RNA, 00:16:06.13 and others which are not. 00:16:09.24 The next slide really summarizes the numerous problems 00:16:14.29 or challenges that must be solved if we're ever 00:16:18.12 to think of a complete chemical process for the replication of RNA. 00:16:24.07 So, we begin with these problems of rate and fidelity. 00:16:29.21 The fidelity is actually closely related to the problem of rate. 00:16:36.27 It turns out that, when you make a mistake 00:16:40.27 in incorporating a nucleotide, so as a chain is growing, 00:16:45.10 you put in the wrong base, make a mismatch, 00:16:47.19 the addition of the next base can be dramatically slowed. 00:16:51.25 We call this the stalling effect, 00:16:53.21 and therefore it slows down the overall rate of synthesis. 00:16:58.00 If we could make the chemistry more accurate, 00:17:01.06 the rate of synthesis would be much better. 00:17:04.11 There's the regiospecificity problem that I mentioned, 00:17:06.26 2' versus 3' linkages. 00:17:10.09 There is a problem that we need 00:17:13.14 very high concentrations of these monomers. 00:17:16.28 They apparently need to be very pure, 00:17:19.18 so if you have other kinds of nucleotides in there, say, 00:17:23.14 with different sugars or different stereochemistry, 00:17:26.03 they will also get incorporated, 00:17:27.27 and that will mess up the product that's made. 00:17:30.23 Also, these monomers, when they're activated as imidazolides, 00:17:36.06 are rather unstable, they're quite susceptible 00:17:40.09 to hydrolysis or cyclization, so those are undesired side reactions. 00:17:47.11 They could be solved if we had the right kind of chemistry 00:17:51.01 to reintroduce the activated state, 00:17:54.14 but that's something that's missing so far. 00:17:57.29 The requirement for a very high concentration of magnesium 00:18:02.03 is extremely problematic both because it's 00:18:04.17 geochemically unrealistic and because it leads to RNA hydrolysis. 00:18:10.20 There's another problem with RNA, which is that, 00:18:14.26 even if you could replicate a strand of any significant length, 00:18:19.14 the melting temperature of that duplex is so high 00:18:23.12 that it's almost impossible to pull the strands apart. 00:18:26.16 Thermally, it becomes impossible to melt them. 00:18:29.27 Even if you could melt them, 00:18:31.21 the strands will come back together again extremely rapidly. 00:18:36.15 This rapid reannealing rate is something that will 00:18:39.15 compete with the much slower template copying chemistry, 00:18:43.11 so this is another problem that has to be solved. 00:18:45.28 And finally, in our experiments, 00:18:48.26 we use primers and watch them grow, 00:18:51.12 simply because that's analytically very easy to do, 00:18:56.14 but of course there weren't primers around on the early Earth, 00:18:59.16 and so we need to think of a primer-independent process 00:19:02.06 for copying templates. 00:19:04.01 So all of these problems together have made it very difficult 00:19:09.12 to think of a plausible pathway 00:19:13.01 for the overall replication of RNA templates. 00:19:18.00 So, what we decided to do was essentially to step back 00:19:22.29 from this and think about other polymers, 00:19:27.08 maybe it would be easier to replicate something else. 00:19:31.04 And in the process of figuring that out, 00:19:33.06 maybe we would get clues that would let us come back 00:19:38.02 to RNA and think about how to solve some of these problems. 00:19:41.25 Now in fact, Leslie Orgel concluded a couple of decades ago 00:19:49.26 that, even though RNA replication looked really hard, 00:19:54.09 he thought that the replication of some kind of 00:19:57.04 informational polymer would be achieved fairly readily, 00:20:01.18 and that in the process of doing that, 00:20:04.27 we would learn something about either RNA replication 00:20:09.21 or how to replicate something that might be 00:20:12.08 relevant to the origin of life. 00:20:14.17 Now, unfortunately, despite that challenge 00:20:17.18 to the chemistry community, few people have addressed 00:20:20.16 the problem, and there is still no example 00:20:24.06 of the chemical replication of any informational polymer. 00:20:27.17 So I think this is a major challenge. 00:20:29.08 It's a really interesting and fun thing to investigate. 00:20:32.09 And this is really the focus of a lot of our attention 00:20:36.02 at the moment. So what can we look at? 00:20:39.19 What would other interesting nucleic acids be? 00:20:44.22 So, what I'm going to do is just show you a set of the 00:20:48.09 kinds of molecules that we've been studying in my lab 00:20:50.26 over the last couple of years. 00:20:53.05 And we're concentrating on 00:20:55.20 phosphoramidate-linked genetic polymers, 00:20:58.22 so these have nitrogen-phosphorus bonds 00:21:01.25 in place of the oxygen-phosphorus bond you see 00:21:05.01 in normal phosphodiesters. 00:21:07.04 The reason for that is that the building blocks, the monomers, 00:21:11.23 for making these polymers are aminosugars, 00:21:14.23 so we now have a much better nucleophile than a hydroxyl group, 00:21:20.02 so that speeds up the chemistry again, 00:21:22.03 giving us another boost in rate. 00:21:24.26 So, here are three phosphoramidate backbones. 00:21:28.07 Here's an acyclic, open-chain backbone with 00:21:32.15 essentially a glycerol nucleic acid backbone. 00:21:37.09 Here is a 2'-5'-linked polymer, 00:21:43.18 so the phosphoramidate version of DNA, 00:21:46.14 with 2' linkages. 00:21:48.16 And here's the molecule closest to DNA. 00:21:51.08 All that's been done is to change to normal oxygen 00:21:53.25 atom here to an NH group, 00:21:57.06 so this is phosphoramidate DNA. 00:22:01.12 And then there are two other molecules that have captured 00:22:03.25 our attention recently, and these are somewhat 00:22:07.10 more conformationally constrained molecules. 00:22:10.19 So this is the phosphoramidate version of TNA, 00:22:13.24 threose nucleic acid; 00:22:15.23 here, the sugar is a four-carbon sugar, threose. 00:22:20.17 And here we have a 2' linkage. 00:22:25.13 These molecules were first made in the Eschenmoser Group 00:22:29.18 and studied. They're perfectly good base-pairing systems. 00:22:34.13 And finally, here you see a morpholino backbone, 00:22:39.08 another conformationally constrained backbone. 00:22:42.19 So in the case of the threose, 00:22:45.01 the conformational constraint comes from the fact 00:22:47.23 that there are only five atoms in the backbone repeat unit, 00:22:51.15 so there's one less rotatable bond, 00:22:53.26 so it's entropically constrained. 00:22:56.09 Here, the constraint is different. 00:22:58.17 It comes from the fact that we have the 00:23:00.09 six-membered morpholino ring, 00:23:03.02 which likes to sit in this chair conformation, 00:23:05.13 so it's conformationally constrained 00:23:07.09 in a very different manner. 00:23:09.11 So, what we've been trying to do is 00:23:11.22 systematically study all of these different kinds of templates 00:23:18.01 and see if we can learn anything from this process that might 00:23:22.07 eventually feed back and teach us about RNA replication. 00:23:26.21 So, at this stage, we're really still heavily involved 00:23:30.17 in studying the copying of these templates. 00:23:33.25 In many cases, 00:23:34.28 these are actually quite challenging to prepare synthetically, 00:23:38.26 so that takes a lot of time and effort. 00:23:41.08 But I'll take you through what we've done so far. 00:23:45.00 So we began by studying the simplest template 00:23:51.26 from a structural point of view, 00:23:53.20 so this is the glycerol nucleic acid backbone. 00:23:56.29 So no cyclic sugar, just an open-chain backbone. 00:24:02.22 And here's the corresponding monomer. 00:24:05.27 So we have the amino nucleophile, 00:24:07.27 we have the activated phosphate, 00:24:10.16 these look very simple from a structural standpoint, 00:24:14.29 but in fact, there's a major problem: 00:24:18.17 That lack of constraint from the cyclic sugar 00:24:24.04 means that the amine nucleophile 00:24:27.25 can directly reach the phosphorus electrophile, 00:24:32.00 and as a result, the activated monomer cyclizes 00:24:36.09 to give this useless product here faster than we can measure. 00:24:42.07 So that tells you right away that this system is not 00:24:45.11 chemically a good system to look at 00:24:47.20 in terms of replication. 00:24:49.15 But, the speed of this reaction actually told us 00:24:53.16 something kind of interesting, which is that 00:24:56.20 the intrinsic chemistry, it can be very fast. 00:25:00.19 So, if you can position the nucleophile 00:25:04.17 in just the right position and orientation 00:25:09.07 relative to the electrophile, 00:25:12.26 polymerization could in principle go very rapidly, 00:25:16.14 even without an external catalyst. 00:25:21.16 The system that we've actually spent the most time 00:25:24.20 studying so far and learned the most about, 00:25:28.14 is the 2'-5'-linked phosphoramidate version of DNA, 00:25:34.21 so here's the polymer. 00:25:36.15 A series a of 2'-5' phosphoramidate linkages, 00:25:40.09 and here's the corresponding monomer, 00:25:42.19 shown as the G nucleoside version, but amine nucleophile. 00:25:48.20 Phosphorimidazolide, so good leaving group, good nucleophile... 00:25:53.22 we should get very fast polymerization chemistry. 00:25:58.09 And in fact, that's exactly what we observed. 00:26:01.05 In our first experiments, after we learned how to make 00:26:03.14 these molecules, we set up the following system, 00:26:07.16 where we have a primer-template complex. 00:26:11.05 The template contains this region of Cs, 00:26:14.01 where the G monomers can bind, 00:26:17.29 and we can then observe the primer being 00:26:20.29 elongated by the sequential incorporation of 00:26:23.23 multiple G residues. 00:26:26.14 The result experimentally is shown here. 00:26:30.14 We start off with the primer, and over the course of hours, 00:26:34.04 we can see the complete copying of the template 00:26:38.12 and the accumulation of the full-length, extended primer. 00:26:44.13 So this reaction is so efficient that, I think, 00:26:47.17 if you didn't know this was just chemistry, 00:26:50.00 you would think this an enzymatically catalyzed reaction, 00:26:53.17 but there's no enzyme, there's no polymerase, 00:26:56.26 this is just the intrinsic chemistry of activated nucleotides 00:27:01.13 binding to a template and extending a primer. 00:27:05.04 So, if we could do this in a more general way, 00:27:09.22 so that we could copy templates of arbitrary sequence, 00:27:14.09 we would basically have the kind of system that we want. 00:27:18.06 We would be able to copy sequences 00:27:20.10 that could carry out functional tasks. 00:27:25.00 One of the nice things about this overall system is that, 00:27:30.11 because of the 2'-5' linkages, 00:27:33.08 the duplex that's formed has a relatively low melting temperature, 00:27:38.28 it's relatively easy to thermally separate the strands, 00:27:42.09 and we could imagine a cycle of complete steps 00:27:47.14 of copying and full replication. 00:27:50.07 So, unfortunately, things are not so simple. 00:27:53.25 This copying chemistry works very well 00:27:56.23 with C templates driving G incorporation, 00:28:00.19 works very well with G templates driving C incorporation, 00:28:04.22 but when we went to try to copy templates 00:28:07.29 that contain As and Us, it basically didn't work at all. 00:28:13.16 So we assumed that the problem was that the AU base pair 00:28:17.16 is much weaker than the GC base pair, which is true. 00:28:22.01 And so the solution was simply to go back 00:28:24.29 to the chemical drawing board and 00:28:27.25 look at different nucleobases that make an AU-like base pair, 00:28:35.14 but that is just as strong as a GC base pair. 00:28:39.14 So it turns out, in fact, 00:28:40.23 this has already been done in other contexts. 00:28:44.04 So, people like Chris Switzer, for example, 00:28:48.10 have looked at the base pair made between D, 00:28:51.26 which is short for diaminopurine, and propynyl-U, 00:28:56.00 which you see over here. 00:28:57.08 So we have a propynyl group at the 5' position of U, 00:29:00.02 this contributes extra stacking energy. 00:29:03.09 The extra amino group in diaminopurine 00:29:07.02 gives us back the third hydrogen bond, 00:29:09.17 and this base pair in the context of DNA 00:29:12.19 is essentially just as strong as a GC base pair. 00:29:16.24 So, we made the corresponding activated monomers, 00:29:22.11 and sure enough, it solves the rate problem. 00:29:25.18 We can now, using this activated propynyl-U nucleotide, 00:29:31.02 we can copy a template consisting of four D residues, 00:29:34.06 and in fact, it's very fast. 00:29:36.02 The reaction's finished within the first ten minutes. 00:29:40.01 We can copy using the activated D monomer, 00:29:45.06 a template consisting of propynyl-Us, 00:29:48.02 it's a little bit slower, but still mostly done within an hour. 00:29:51.12 Not too bad! 00:29:52.27 So we thought, okay, maybe we've really solved the problem. 00:29:56.12 Let's get a little bit more ambitious 00:29:58.18 and try to copy progressively longer template sequences 00:30:03.04 that include progressively more, different nucleotides 00:30:07.21 in the sequence. 00:30:09.16 So, the first step looked pretty good. 00:30:13.02 Here we have a template that consists of three Ds and three Cs, 00:30:17.18 so we're incorporating three propynyl-Us followed by three Gs, 00:30:21.26 and the reaction goes pretty well, within a few hours. 00:30:25.26 If you leave out the G, you stall where you should, 00:30:30.05 if you leave out the U, you basically stall almost at the beginning. 00:30:35.09 So that looks encouraging. 00:30:37.26 You do see a few shorter products here, 00:30:40.16 which made us worry a little bit about the 00:30:42.13 accuracy of the overall process, but it's not too bad. 00:30:46.15 As we go to longer templates, so here a mix of Gs and Cs, 00:30:51.13 we can still copy the whole thing, but it does take longer, 00:30:55.24 and you do see more of these intermediate sequence 00:31:00.10 accumulating. And that actually gets much worse, 00:31:05.02 when we go to an even longer sequence of 15 or 16 nucleotides 00:31:09.21 incorporating all four bases in the template. 00:31:13.23 And now we still get some full-length product, 00:31:17.17 but it takes a long time to accumulate, 00:31:19.27 and we see a lot of these stalled intermediate products 00:31:25.13 accumulating over the course of the reaction. 00:31:28.11 So we don't know for sure, 00:31:29.22 but we suspect that these stalled intermediates 00:31:32.22 are the result of mistakes in the template-copying process, 00:31:39.27 such that a chain is growing, a mismatch is formed 00:31:45.11 by a mistaken incorporation, 00:31:47.19 and that drastically slows down the subsequent polymerization. 00:31:52.03 So in fact, we think that, 00:31:56.06 in order to get more efficient copying, 00:31:59.29 in order to speed up the overall reaction, 00:32:02.24 we need to solve the fidelity problem, 00:32:06.09 and that that might help solve the rate problem. 00:32:08.21 So how could we do that? 00:32:10.10 Well, we could look at different nucleobases, 00:32:13.24 maybe our choice of D and propynyl-U was not so great. 00:32:17.19 In fact, there are chemical reasons to think that that's true. 00:32:21.17 In particular, the propynyl group on U changes the pKa of its N1, 00:32:26.26 which can lead to the formation of other tautomers 00:32:29.18 and mismatched base pairs. 00:32:32.05 We could look at other backbones, which we are doing. 00:32:36.14 So there's the possibility that, if the backbone is 00:32:39.02 conformationally constrained in just the right way, 00:32:43.02 it will favor the incorporation of the right bases 00:32:45.20 and disfavor the incorporation of mistaken bases. 00:32:49.26 We could also consider looking at oligonucleotide substrates, 00:32:53.06 which actually turns out to be a really good idea. 00:32:55.14 There are probably a lot of reasons why this would be helpful. 00:32:58.28 And we could also consider looking at catalysis, 00:33:02.00 either ribozyme-mediated catalysis 00:33:04.26 or perhaps catalysis by small molecules or short peptides 00:33:09.13 that might've been lying around, 00:33:11.23 and that's another approach that we're starting to take. 00:33:16.23 So, let's go back to this idea of looking at different nucleobases. 00:33:22.29 So, it turns out there's actually a very simple substitution 00:33:28.15 that looks extremely promising at this stage. 00:33:31.10 So here's the D-propynyl-U base pair 00:33:34.12 which we think is causing problems with fidelity. 00:33:39.06 An alternative is to just replace U with 2-thio-U, 00:33:44.08 so U with a sulfur in place of the oxygen 00:33:47.01 normally at the 2' position. 00:33:50.03 This is an analogue of U that's actually found in nature, 00:33:55.08 in modern biology, it's a common substituent in tRNAs, 00:33:59.04 where it plays the role of stabilizing an AU base pair 00:34:02.21 and increasing the fidelity of that interaction. 00:34:06.07 And the reason that works is because the much 00:34:08.22 larger and polarizable sulfur 00:34:11.05 contributes to stacking interactions. 00:34:15.18 The larger size is accommodated in a base pair with A, 00:34:19.14 but is not accommodated in a wobble base pair with U. 00:34:24.08 So it seems to both stabilize the AU base pair 00:34:27.14 and disfavor the incorrect wobble base pairing. 00:34:35.03 So, we have preliminary experiments that suggest that 00:34:40.18 this is a promising approach, and we're continuing to look at that. 00:34:44.02 Meanwhile, we're also looking at a number of these other backbones, 00:34:48.21 and so here we come back to the 00:34:52.01 3'-linked phosphoramidate version of DNA. 00:34:56.08 Here is the corresponding monomer. 00:35:00.12 We like this system for different reasons than we like the 2' system. 00:35:06.27 Here we're making the natural 3'-5' linkage, 00:35:12.08 so we're a little bit closer to making an RNA-like product. 00:35:16.28 In fact, duplexes of this 3' phosphoramidate version 00:35:21.26 of DNA have a very RNA-like geometry, 00:35:25.06 so that's kind of nice. 00:35:27.06 These monomers can cyclize, 00:35:29.03 so this amino group can reach the phosphorus, 00:35:31.29 but that reaction is fairly slow, so it's not a fatal problem. 00:35:36.12 What may, in the long run, be more of an issue, 00:35:40.29 is that those duplexes have a very high melting temperature. 00:35:45.00 Nonetheless, it's an interesting system to study 00:35:49.12 because this chemistry seems to go very effectively, and in fact, 00:35:54.14 we see efficient incorporation of all four natural nucleotides 00:36:00.22 using this backbone. So it shows that 00:36:02.19 there's a very important coupling between backbone 00:36:05.18 chemistry and the bases that are involved 00:36:09.12 in forming the Watson-Crick paired structure of the duplex. 00:36:13.28 We don't really fully understand the nature of that coupling. 00:36:20.15 I mentioned before two conformationally constrained 00:36:24.05 phosphoramidate-linked nucleic acids, 00:36:26.21 the threose nucleic acid (TNA) and the morpholino backbone, 00:36:32.25 which we abbreviate as MoNA. 00:36:35.03 These are very interesting systems, 00:36:38.02 and the hope here is that conformational constraint 00:36:44.03 might be a way of increasing both the accuracy 00:36:46.28 and the rate of chemical copying. 00:36:49.26 So why do we really think that? 00:36:52.00 Here's an experiment done by Jason Schrum 00:36:55.02 when he was a graduate student in the lab, 00:36:58.03 and we're here looking at the incorporation of 2' amino 00:37:03.06 nucleosides, extending primers where the template 00:37:09.03 is composed of a series of different polymers. 00:37:11.29 So DNA over here, RNA template, 00:37:15.16 an LNA template (LNA stands for "locked nucleic acid," 00:37:19.14 this is a relative of RNA where the sugar conformation is locked 00:37:24.06 into the RNA-like conformation by a cross-link 00:37:27.21 underneath the sugar), 00:37:29.14 and over here is a 2'-5'-linked DNA template. 00:37:33.01 So what you see in the timecourse of this primer extension reaction, 00:37:37.16 is that the reaction goes fairly slowly on a DNA template, 00:37:41.17 a lot more rapidly on an RNA template, 00:37:44.19 but even more rapidly on the conformationally 00:37:47.25 constrained LNA template. 00:37:52.10 So this was our first real experimental hint 00:37:55.20 that conformational constraint of a template could 00:37:58.22 really have a useful and significant effect 00:38:04.22 on the rate of polymerization. 00:38:08.20 And so that's encouraged us to go ahead and 00:38:12.15 make these conformationally constrained templates, 00:38:16.02 even though the synthetic chemistry is rather challenging. 00:38:21.18 So here is the threose nucleic acid backbone, 00:38:26.12 so again a four-carbon sugar. 00:38:28.11 Here's the corresponding monomer. 00:38:31.09 There has been some structural work done from the Egli Lab. 00:38:36.04 You can see here that a TNA duplex looks, at a gross level, 00:38:40.25 very similar in overall geometry to an RNA duplex. 00:38:45.01 So this, I think, is encouraging for the possibility that this 00:38:49.17 constrained backbone might help to position 00:38:53.19 the incoming nucleoside correctly, so as to speed up 00:38:58.10 polymerization and potentially make it more accurate. 00:39:02.05 So we hope to do those experiments 00:39:04.17 over the next few years. 00:39:07.14 Here is the morpholino phosphoramidate backbone, 00:39:10.16 again conformationally constrained 00:39:12.05 in a very different way 00:39:13.22 because of the six-membered morpholino ring. 00:39:17.25 These molecules are actually much easier 00:39:20.14 to make than the TNA molecules. 00:39:23.29 And so we've been able to start looking at the copying 00:39:27.10 of morpholino templates by morpholino monomers, 00:39:30.07 so we think this is another very promising system 00:39:33.20 in which to investigate experimentally the effects 00:39:36.24 of conformational constraint on the rate 00:39:39.11 and fidelity of copying. 00:39:43.07 So, even though we're still far from having 00:39:48.13 a complete chemical system that could drive the replication 00:39:52.08 of any nucleic acid or any genetic material, 00:39:56.16 we can use what we've found so far to learn about 00:40:01.03 the compatibility of the chemistry of genetic replication 00:40:05.26 with our replicating vesicle systems. 00:40:09.09 So we can do experiments where we encapsulate 00:40:12.00 nucleic acids inside vesicles and look at the copying chemistry. 00:40:17.06 So an example of that is shown here. 00:40:19.14 This was work done by Sheref Mansy and Jason Schrum 00:40:23.04 and other people in my lab, a few years ago. 00:40:26.16 The basic experiment is to take the same kind 00:40:28.23 of primer-template complex that you've seen before, 00:40:32.13 and monitor the extension of the primer by a 00:40:36.04 template-directed synthesis. 00:40:38.08 But this time, the primer-template is inside 00:40:42.06 one of these vesicles, 00:40:43.25 so we're going to add the activated monomer to the outside. 00:40:47.25 It has to get across the membrane spontaneously, 00:40:50.19 without any transport machinery, to get to the inside, 00:40:53.22 where it can do this template-copying chemistry. 00:40:58.01 So, here's the experimental result: 00:41:00.22 On this side, you see the control reaction done in solution, 00:41:04.19 you see the accumulation of full-length material over 12 to 24 hours. 00:41:08.23 Here is the same experiment with an encapsulated 00:41:12.00 primer-template, and you can see that the reaction is slowed 00:41:15.02 down a little bit, but still by 24 hours 00:41:18.10 you can see the accumulation of mostly full-length product. 00:41:22.26 So that was actually extremely encouraging for us. 00:41:26.00 This was a real major advance, because it said that, 00:41:29.08 yes, the chemistry is compatible; 00:41:32.05 these building blocks can get across the membrane; 00:41:34.26 when they get to the inside of the vesicle they can copy templates; 00:41:38.17 and once we developed a more general 00:41:42.08 template-copying chemistry, we should be able to 00:41:45.05 combine it with the replicating vesicles 00:41:48.03 and have the composite system that has been our goal all along. 00:41:54.04 Now, in this experiment, the membrane was made 00:42:00.06 from a convenient laboratory system, 00:42:03.27 these unsaturated C14 fatty acids. 00:42:07.16 We can do the same experiments with a much more 00:42:10.17 prebiotically realistic mixture of fatty acids, fatty alcohols, 00:42:15.24 their glycerol esters, saturated ten-carbon chains, 00:42:20.18 and we see the same thing. 00:42:21.27 Essentially over a period of 12 to 24 hours, 00:42:25.25 the nucleotides can get across these membranes 00:42:29.03 and copy templates. In contrast, 00:42:32.23 when the vesicle is made from more modern molecules, 00:42:36.25 from phospholipids, those are a complete barrier 00:42:40.21 to the penetration of these nucleotides, 00:42:43.12 so nothing happens. You see no primer extension. 00:42:47.21 So what this is telling us is that, 00:42:49.16 for this to work, for this protocell model to work, 00:42:52.27 you need the membrane to be made of the 00:42:55.23 right kinds of primitive molecules, 00:42:57.23 so simple fatty acids and related molecules, 00:43:01.01 and the nucleotides have to be the right kind of 00:43:04.17 primitive molecules, not triphosphates, 00:43:07.21 but something like phosphorimidazolides, 00:43:09.27 something less polar. 00:43:14.10 All right, so with all of this work, 00:43:16.03 are we actually any closer to coming back to RNA 00:43:20.04 with new ideas for complete replication? 00:43:24.05 So, I think that we are, 00:43:26.21 and I'll tell you about one of the way of the conceptual advances 00:43:30.14 that we've made recently. 00:43:32.04 This actually comes from a selection experiment 00:43:35.27 that was done by Simon Trevino when he was a 00:43:38.02 graduate student in the lab, 00:43:40.02 and it addressed this question of monomer homogeneity: 00:43:43.13 How important is it really in a prebiotic setting 00:43:47.01 that the monomers be really pure and concentrated, 00:43:51.20 so that we don't make backbones that have different kinds 00:43:56.01 of linkages, different kinds of sugars, etc.? 00:44:00.17 So the experiment that we could do in the lab 00:44:03.29 is a little bit more limited, 00:44:06.06 but what Simon worked out was a way of taking a 00:44:09.09 library of DNA sequences and transcribing that 00:44:15.01 into molecules that are not just RNA, 00:44:21.08 but a mixture of RNA and DNA. 00:44:23.28 In fact, in every position in these transcripts, 00:44:28.00 there's roughly a 50-50 chance of that linkage 00:44:32.03 being a ribonucleotide or a deoxyribonucleotide. 00:44:36.26 So we have extreme backbone heterogeneity, 00:44:39.16 ribo- and deoxyribonucleotide linkages, 00:44:42.25 and that variation is not heritable. 00:44:47.29 The experiment was to take this library 00:44:50.17 and then select for functional molecules, 00:44:53.01 we select for aptamers just by binding to a target molecule. 00:44:58.01 The targets were used were ATP and GTP, 00:45:00.08 because we'd done this many times years ago. 00:45:03.11 It's easy to evolve RNA molecules or DNA molecules 00:45:07.24 that specifically recognize these nucleotides, 00:45:11.15 but in this experiment, the pool isn't pure RNA or DNA, 00:45:15.23 it's this mixed backbone polymer. 00:45:19.25 So what Simon found is that he could go around 00:45:23.04 cycles of selection and amplification. 00:45:27.00 Every time we do the amplification, 00:45:28.24 we reintroduce this backbone heterogeneity, 00:45:31.25 but of course the molecules get shuffled. 00:45:34.10 The exact order of ribo- and deoxyribo- linkages is randomized 00:45:38.29 at every round of the selection process. 00:45:41.29 Nonetheless, after a few rounds of selection, 00:45:45.04 Simon was able to obtain aptamers 00:45:47.21 that bound to their target with great specificity. 00:45:53.02 They weren't quite as good in terms of affinity 00:45:56.09 as the aptamers we get from pure RNA or pure DNA, 00:46:00.25 but they still work. 00:46:02.21 So this told us that maybe monomer heterogeneity 00:46:06.26 wasn't as important as we'd been thinking. 00:46:09.29 Maybe you could actually evolve functional molecules, 00:46:13.23 ribozymes, in the face of nonheritable 00:46:17.12 backbone heterogeneity. 00:46:19.17 So why is that important in the context of RNA? 00:46:24.23 Well, one of the big problems with RNA is this regiospecificity, 00:46:30.17 the fact that, in a chemical system, 00:46:33.19 it seems almost unavoidable that some fraction of 00:46:37.08 2'-5' linkages will be formed. 00:46:41.00 I think Simon's experiment hints that this 00:46:44.05 may still allow for the evolution of ribozymes, 00:46:48.01 so this is something that needs to be experimentally investigated, 00:46:51.13 something we're doing now. 00:46:53.10 Now, if that turns out to be true, 00:46:55.26 and you can still get functional molecules in the face 00:46:58.11 of this backbone heterogeneity, 00:47:00.18 then the important implication comes from the fact that 00:47:03.23 we already know the 2'-5' linkages in the backbone 00:47:08.00 drastically lower the melting temperature of an RNA duplex. 00:47:12.20 And so, as a result, it would now become possible 00:47:16.21 to thermally separate the strands after the copying 00:47:20.02 of an RNA template. 00:47:21.28 So, it's possible that this 2' versus 3' heterogeneity 00:47:27.07 that we used to think was such a huge problem with RNA, 00:47:31.09 is actually what allowed RNA to work 00:47:34.16 as the primordial genetic material, 00:47:37.16 because it allows for thermal strand separation, 00:47:40.15 and therefore, the repeated cycles of 00:47:44.13 template copying and strand separation that 00:47:46.20 give you overall replication. 00:47:49.17 So this is the kind of thing that we're actively studying. 00:47:53.13 A few more points about more 00:47:56.25 primitive scenarios for template copying... 00:48:03.28 All of the work that we've done, 00:48:06.12 and many other people have done over the decades, 00:48:09.03 has tended to focus on primer extension reaction 00:48:12.13 with monomers, because this is a very simple 00:48:15.28 and analytically tractable approach to the problem 00:48:19.00 of template copying. You get a lot of information, 00:48:21.20 it's easy to analyze the products by simple methods 00:48:25.24 such as gel electrophoresis, 00:48:28.29 but it's probably completely unrealistic 00:48:32.19 as a prebiotic scenario. 00:48:34.22 So what we're being driven to think about is template 00:48:37.16 copying by mixtures of short, random-sequence oligomers, 00:48:42.10 along with monomers, dimers, etc. 00:48:46.02 And so it's a much messier situation, 00:48:48.15 you have a large number of different types of substrates, 00:48:52.29 the number of partial products of template copying 00:48:56.10 becomes enormous, and so the analytical problem 00:48:59.03 gets much worse. But nowadays, 00:49:01.26 we have much more advanced analytical techniques, 00:49:04.23 and with advanced methods of mass spectrometry, 00:49:07.14 we can actually hope to analyze these 00:49:09.23 kinds of reactions and perhaps, we'll see that 00:49:14.08 this kind of system gives unexpected benefits. 00:49:19.23 We have the possibility of nucleating the copying chemistry 00:49:23.24 of multiple sites, the incorporation of oligomers 00:49:27.04 means that fewer catalytic or chemical steps are required, 00:49:32.08 so we're very excited about following up on this kind of 00:49:36.01 more natural, "messier" but more natural scenario, 00:49:40.03 in the hope that this will actually lead us closer 00:49:43.20 to a realistic scenario for full replication. 00:49:47.23 So I just want to end by pointing out that, 00:49:50.07 in this much messier scenario, 00:49:53.15 there are completely new ways in which we can think 00:49:56.29 of ribozymes, catalytic RNAs, contributing 00:50:01.10 to the overall process of replication. 00:50:04.02 Up till now, we have exclusively thought about RNA-catalyzed 00:50:09.03 RNA replication occurring through RNA catalysts 00:50:14.26 that are RNA polymerases. 00:50:18.03 But in these scenarios, 00:50:19.26 I think that it's actually possible 00:50:21.26 that the primordial replicase might've been a nuclease. 00:50:26.13 For example, if an oligomer binds and then gets extended, 00:50:30.25 but a mistake is made, you make a mismatch. 00:50:34.04 As we've discussed, that slows down 00:50:36.17 subsequent primer extension, 00:50:40.05 then that can be a drastic effect. 00:50:42.25 So, if there was a ribozyme nuclease 00:50:46.04 that could chew off that mistake, 00:50:49.18 it would allow chemical copying to go back to normal. 00:50:54.00 So that would be one way of speeding the process up. 00:50:57.20 Another scenario comes from thinking about 00:51:01.16 the use of oligonucleotide substrates. 00:51:04.15 It could be that overlapping oligonucleotides 00:51:07.17 bind to a template, and so this 00:51:09.16 would be a kind of a dead-end situation, 00:51:12.02 unless you have a nuclease that can come along, 00:51:15.10 trim away the overlap, and allow chemical ligation 00:51:18.20 to complete the process of template copying. 00:51:21.22 So, I think these kinds of changes in the way 00:51:25.02 we're thinking about the process have really 00:51:27.15 opened up a lot of new experiments 00:51:30.02 and have made me very optimistic about the possibility 00:51:33.22 of attaining a complete replication system, 00:51:36.14 either purely chemically, or by a combination of 00:51:39.21 chemical and RNA-catalyzed reactions. 00:51:46.07 So, just to sum up then, 00:51:50.00 I think that these considerations tell us that 00:51:52.14 monomer purity may not be as important. 00:51:56.01 It's possible that some backbone heterogeneity 00:51:58.28 may not be fatal. 00:52:02.05 The incomplete regiospecificity may be fine; 00:52:08.15 2' linkages may solve the melting issue for RNA. 00:52:13.18 And we're very excited about studying 2-thio-U 00:52:18.00 as a simple nucleotide substitution prebiotically plausible, 00:52:25.05 something that might enhance both rate and fidelity. 00:52:28.25 So by putting all these things together, 00:52:30.16 we're hopeful that over the coming years, 00:52:33.03 we'll eventually converge on a complete chemical system 00:52:38.02 for the replication of either RNA or maybe some related polymer. 00:52:42.11 And if we can get to that point, 00:52:45.00 combining it with the replicating vesicle system should 00:52:47.21 allow us to observe the spontaneous emergence 00:52:51.21 of Darwinian processes from a purely chemical system. 00:52:55.16 And that's really the major goal of this whole thing, 00:52:57.29 and the part of the project that's most relevant 00:53:01.04 to the emergence of biology from the 00:53:03.21 chemistry of the early Earth. 00:53:05.26 So, again, I've tried to mention people as I've gone along. 00:53:10.21 In terms of the chemistry, 00:53:13.09 many people have contributed to this over the years: 00:53:16.10 Jason Schrum, Alonso Ricardo, Matt Powner, Na Zhang, 00:53:20.22 Ben Heuberger, Craig Blain, Shenglong Zhang. 00:53:24.10 Many people have contributed to this work, 00:53:27.29 and so they've played a very important role in developing 00:53:30.28 all of these new ideas that are leading us, hopefully, 00:53:34.13 towards a solution to this major problem 00:53:37.00 in thinking about the origin of life. 00:53:39.06 Thank you.