The Origin of Life on Earth
Transcript of Part 2: Protocell Membranes
00:00:00.00 My name is Jack Szostak, 00:00:02.10 I'm a Professor Genetics at Harvard Medical School, 00:00:05.22 and an Investigator at Massachusetts General Hospital, 00:00:08.22 where my laboratories are. 00:00:10.21 I'm also an Investigator of the Howard Hughes Medical Institute. 00:00:14.07 And in this lecture, I would like to tell you about some of our 00:00:18.21 recent work relevant to the origin of life on Earth, 00:00:22.12 and we're going to focus this time on the membrane component of simple protocells, 00:00:30.03 how those membranes assemble, 00:00:32.03 what their properties are, 00:00:34.03 how those properties are relevant for early forms of life, 00:00:38.08 and how we imagine them gradually changing 00:00:42.27 as a result of the emergence of Darwinian evolutionary processes. 00:00:48.28 So, let's begin by having a look at our conceptual model 00:00:55.26 of what an early protocell looks like. 00:00:59.09 So, again, we think of protocells having two parts: 00:01:03.24 A membrane boundary that closes up into a spherical vesicle 00:01:11.25 that can encapsulate molecules in the internal, aqueous compartment... 00:01:16.21 and the most important of those would be genetic molecules, 00:01:21.25 possibly RNA or DNA or maybe some related kind of nucleic acid. 00:01:27.04 And what we would like to understand is how, 00:01:33.07 in the absence of highly evolved genetic material, 00:01:37.18 this membrane compartment could first grow 00:01:41.22 and subsequently divide into daughter cells. 00:01:47.10 And because we're thinking about this at a very early time, 00:01:53.14 prior to the emergence of Darwinian evolution 00:01:56.16 and advanced, highly evolved cellular machinery, 00:01:59.17 we have to look to simple chemical and physical processes 00:02:04.15 to drive these transformations. 00:02:08.11 So, to think about that, 00:02:10.22 we need to think about the molecules that are going to be used 00:02:14.11 to build these membranes, and our favorite molecules 00:02:18.28 in this regard are the fatty acids. 00:02:22.01 So, these range from short-chain, saturated fatty acids, 00:02:26.20 like capric acid, which are the kinds of molecules 00:02:29.05 that we think could have been easily made in different scenarios 00:02:33.01 on the early Earth. 00:02:35.05 Many times in our laboratory work, 00:02:37.07 we use more modern fatty acids 00:02:40.03 (myristoleic acid or oleic acid) simply for convenience, 00:02:45.01 but in the end, we always come back to more prebiotically reasonable systems. 00:02:52.12 So what are the biophysical properties of these closed-up membranes, 00:03:00.08 in the form of vesicles or liposomes? 00:03:04.12 So, the most important aspect, 00:03:08.13 and something which is quite different from modern membranes 00:03:12.21 that are based on phospholipids, is that fatty acid vesicles... 00:03:19.20 basically, their formation and existence is controlled 00:03:23.08 by a series of phase transitions. 00:03:26.29 And the most important of those depends on the pH of the environment. 00:03:31.21 So, at fairly alkaline pHs, 00:03:35.17 the carboxylates of fatty acids are basically fully ionized, 00:03:40.24 and then charge repulsion wants to keep those groups 00:03:45.00 as far apart from each other as possible, 00:03:48.07 while the hydrophobic forces want to keep the hydrophobic tail 00:03:52.18 of the molecule as close together as possible. 00:03:55.26 And the result is the formation of micelles, 00:03:59.14 which are very small aggregates, a few nanometers in size, 00:04:05.08 and composed of roughly 10 to 100 fatty acid molecules. 00:04:09.20 Now as the pH in the environment gradually 00:04:12.00 drops towards roughly 8.5, 00:04:16.26 which is the point at which half of these carboxylates 00:04:19.21 will be ionized and half-protonated, 00:04:22.06 there's a phase transition that results in the spontaneous 00:04:25.14 assembly of bilayer membranes, as you can see here. 00:04:29.21 And because the curvature is much reduced in these much larger aggregates, 00:04:36.00 these are actually stabilized by hydrogen bonding 00:04:41.19 interactions between adjacent charged and ionized carboxylates. 00:04:47.07 Now, as the pH drops even further into the acidic range, 00:04:51.07 all of the carboxylates will become protonated eventually, 00:04:55.05 and then these structures collapse into oil droplets. 00:04:59.01 So the existence of these bilayer membranes 00:05:02.27 is only possible at an intermediate pH range, close to the pKa. 00:05:08.23 You'll notice this pKa of 8.5 is considerably higher 00:05:12.11 than the pKa of individual fatty acid molecules in solution, 00:05:17.24 which is closer to 4.5. 00:05:19.20 And the reason for that is that these charged head groups 00:05:23.15 are much closer together in this aggregate structure. 00:05:29.16 Now, when you do form these membranes, 00:05:32.14 they have very interesting properties, 00:05:34.18 which again are very different from the 00:05:37.01 properties of phospholipid-based membranes. 00:05:40.27 And one of the things that is very important to us 00:05:44.08 is the dynamic exchange processes that occur in these membranes. 00:05:49.02 So fatty acid molecules in a fatty acid membrane 00:05:52.25 flip-flop back and forth, from the outer leaflet to the inner leaflet, 00:05:57.28 very rapidly, sub-second timescale. 00:06:01.00 They also exchange in and out of the membrane, 00:06:06.03 so into solution, where they can exchange with molecules in micelles, 00:06:10.25 and again this is a very rapid process, 00:06:14.03 so that the molecules that make up any given vesicle 00:06:17.08 are constantly in flux and exchanging with the molecules in other vesicles, 00:06:22.01 again on the second timescale. 00:06:27.13 Another very interesting aspect of these membranes is that 00:06:34.24 they can be quite heterogeneous at a molecule level. 00:06:40.02 And so these membranes can incorporate not just fatty acids 00:06:46.14 but also their glycerol esters, as you see here, 00:06:50.01 and even lyso-phosphatidic acids, 00:06:53.10 so an ester with phosphoglycerate. 00:06:56.12 They can incorporate alkanes and 00:07:00.02 polycyclic aromatic hydrocarbons as well. 00:07:04.00 Now, what happens when you make membranes 00:07:07.00 with these kinds of mixed compositions? 00:07:10.09 It's very interesting that, in many cases, 00:07:14.15 this results in membranes that are much more stable, 00:07:18.15 they're more physically stable, more thermostable, and also, 00:07:22.25 often more permeable to polar nutrients that we would like to 00:07:27.04 bring in from the outside environment. 00:07:30.27 So this slide simply illustrates an experiment 00:07:35.05 to look at the thermostability of some vesicles 00:07:39.18 made from a pure fatty acid, as opposed to mixtures. 00:07:43.01 So in this experiment, what we've done is to encapsulate 00:07:47.18 a short DNA oligonucleotide within vesicles 00:07:52.08 and watch it leak out over the course of about an hour. 00:07:57.07 And now if the membrane is composed purely of myristoleic acid, 00:08:02.23 this 14-carbon, singly unsaturated fatty acid, 00:08:06.11 you can see that up to about 50 degrees, 00:08:09.04 nothing leaks out over an hour. 00:08:11.11 But as the temperature goes up to about 80, 00:08:14.20 they become more leaky, and then at 90 degrees, 00:08:18.09 everything is leaked out over an hour. 00:08:21.00 Now, if you add some of the corresponding alcohol, 00:08:23.26 you can see significant stabilization, 00:08:27.10 but the most dramatic effect is when you add a fraction of the 00:08:31.27 glycerol ester of that fatty acid. 00:08:34.04 And now, you can essentially boil these vesicles for an hour, 00:08:39.00 and none of the contents leak out, 00:08:41.13 at least these oligonucleotide contents are trapped inside. 00:08:46.02 So that's really exactly what we want. 00:08:49.00 We want the genetic material, long oligonucleotides, 00:08:53.18 to stay inside indefinitely, 00:08:56.24 while small molecules (nutrients) 00:08:59.00 can get across the membrane very rapidly. 00:09:03.06 And again, as the temperature goes up, 00:09:05.07 the permeability to small molecules increases dramatically. 00:09:11.02 So, we got into this area about 10 years ago, 00:09:17.16 and started doing some very simple experiments 00:09:21.20 based on the pioneering work of people like Dave Deamer 00:09:26.14 and Pier Luigi Luisi. 00:09:28.14 Now, the Luisi Lab in particular had shown already 00:09:32.05 that you can make vesicles grow simply by 00:09:35.24 adding more fatty acids in the form of alkaline micelles. 00:09:40.05 When the micelles go into a buffered solution, 00:09:42.21 they become a thermodynamically unstable phase, 00:09:45.21 so there's an energetic driving force for those molecules 00:09:49.04 to assemble into new membranes or 00:09:51.21 to integrate into preexisting membranes. 00:09:56.09 And at the time, 00:09:57.27 Martin Hanczyc was a postdoc in the lab, 00:09:59.29 Shelly Fujikawa was a student... 00:10:03.01 they basically repeated those experiments, 00:10:06.01 but using real-time methods of analysis, 00:10:09.23 so that we could watch the process of vesicle growth in real time, 00:10:14.19 after the addition of food molecules. 00:10:17.15 The assay that we used was a fluorescence assay 00:10:21.21 based on energy transfer from donor to acceptor dyes. 00:10:27.02 It's basically a measure of how far apart the dyes are on average. 00:10:31.15 And as the membrane grows by incorporating new molecules, 00:10:34.28 the dyes are diluted, they get further apart, 00:10:37.14 and energy transfer efficiency decreases. 00:10:40.18 So you can use this as a very effective assay for 00:10:43.05 surface area in real time, so we can watch things grow very easily 00:10:48.27 by this simple fluorescence assay. 00:10:52.17 And the essential result from that is that, if you add the food slowly, 00:10:57.02 almost all of it gets incorporated into preexisting membranes; 00:11:00.13 if you add it very quickly, only a fraction, maybe half, 00:11:04.26 will get incorporated, and the rest will form new vesicles. 00:11:08.00 So that takes care of growth at a basic level. 00:11:11.22 What about division? 00:11:13.24 So in the early days, we didn't really have any good idea 00:11:17.27 of how to drive protocell division through environmental fluctuations. 00:11:23.25 So essentially, in desperation, 00:11:26.04 we resorted to a brute force approach of forcing big vesicles 00:11:31.12 through small pores in a filter, 00:11:33.23 and little vesicles come out the other side. 00:11:37.00 So, on the one hand, 00:11:39.00 it's satisfying because it's a proof of principle that 00:11:41.26 cell division could be driven by a physical, environmental force. 00:11:47.00 On the other hand, it's not very satisfying because 00:11:49.22 we didn't think there was any realistic analogue 00:11:53.26 to that process on the early Earth. 00:11:56.07 And furthermore, when you take a large vesicle 00:11:59.16 and physically divide it into smaller spherical vesicles, 00:12:03.18 you necessarily lose some of the contents, 00:12:06.16 which is also not very satisfying. 00:12:09.15 Nonetheless, that process could be turned into a cycle, 00:12:13.19 where we could do growth and division 00:12:15.28 and growth and division, again and again. 00:12:20.04 So then the problem really became: 00:12:22.15 How could we find other pathways that might have 00:12:27.15 been more physically realistic, more robust, 00:12:30.24 and more plausible for the early Earth? 00:12:34.29 So, the first advance in this area came from 00:12:41.21 thinking about an alternative growth pathway. 00:12:46.03 And so this is a pathway of growth that, 00:12:48.08 instead of relying on the environmental influx of new molecules, 00:12:54.23 occurs in a different way, 00:12:56.04 by competition between protocells in a population. 00:13:01.13 So some of are going to grow at the expense of others. 00:13:04.15 And this was work done by Irene Chen when she was a student in my lab, 00:13:09.09 and initiated by discussions with Rich Roberts, 00:13:12.12 when he was a postdoc in the lab. 00:13:14.14 So the basic idea was that, 00:13:17.08 if you think of our basic protocell model, 00:13:21.17 we have genetic molecules trapped inside a semipermeable membrane. 00:13:27.19 Now, large polymers contribute very little to the 00:13:31.06 overall internal osmotic pressure in a system like this. 00:13:36.13 Most of the osmotic pressure actually results from 00:13:40.11 counter-ions that have to be there to neutralize the charge 00:13:44.11 on the polyanionic genetic molecule. 00:13:47.29 And so this is the classical Donnan effect: 00:13:51.25 These ions contribute to an internal osmotic pressure, 00:13:56.03 and we thought that, as a result of this physical phenomenon, 00:14:01.05 this physical effect, 00:14:02.28 vesicles that had more RNA inside of them 00:14:06.01 should have a higher osmotic pressure, 00:14:09.12 and maybe that could actually drive growth competitively. 00:14:14.21 So, the way that we looked at that experimentally 00:14:19.08 was using the same fluorescence assay that we had been using 00:14:24.17 before to monitor membrane growth following 00:14:28.11 the addition of new membrane molecules, 00:14:31.13 but in this case, the idea is that we're going to take 00:14:37.05 osmotically swollen vesicles that have, 00:14:39.28 for example, a lot of RNA inside, 00:14:42.16 and we'll monitor the surface area of those vesicles following 00:14:47.22 mixing with vesicles that are osmotically relaxed, 00:14:51.25 and we'll ask what happens to the surface area. 00:14:56.14 So here's the experiment, 00:14:58.23 where we're monitoring the surface area of the swollen vesicles, 00:15:03.10 these are vesicles that have a lot of RNA on the inside, 00:15:06.28 and by themselves, they're perfectly stable, nothing happens, 00:15:11.02 same thing for the relaxed vesicles, nothing happens. 00:15:13.24 As soon as you mix the swollen vesicles with relaxed vesicles, 00:15:18.02 you can see that their surface area starts to increase 00:15:22.05 over a period of minutes. 00:15:24.09 And in control experiments, 00:15:25.29 where we mix with buffer or other swollen vesicles, 00:15:29.21 there's very little change. 00:15:31.28 So it seems like the swollen vesicles are growing 00:15:35.19 by absorbing molecules from their relaxed neighbors. 00:15:40.11 We can of course do the converse experiment, 00:15:43.11 and monitor the surface area of the relaxed vesicles, 00:15:47.12 and when they are mixed with swollen vesicles, 00:15:49.25 their surface area declines, so molecules are leaving 00:15:53.05 those vesicles and going into the swollen vesicles. 00:15:56.10 Again, control experiments where relaxed vesicles 00:15:58.17 are mixed with more relaxed vesicles, not much happens. 00:16:03.06 So what's going here is that, 00:16:07.09 as soon as we mix these populations of vesicles, 00:16:09.20 the swollen ones are growing, the relaxed ones are shrinking, 00:16:13.28 and the reason that's happening is that this osmotic pressure 00:16:17.10 creates a tension in the membrane, 00:16:20.20 which is only relaxed by the incorporation of more molecules, 00:16:24.21 which allows the surface area to increase. 00:16:28.04 So the exciting implication of that physical process 00:16:32.17 is that it's a potential link between genome replication 00:16:37.09 and faster membrane growth. 00:16:39.08 So, any process that results in faster replication 00:16:43.15 of the genetic material inside a vesicle will automatically, 00:16:48.28 as a result of the Donnan effect, 00:16:51.05 lead to faster growth of the membrane compartment. 00:16:54.07 So it's a coupling between genome replication 00:16:57.01 and cell growth as a whole. 00:16:59.16 Now, that could be particularly interesting if genome replication 00:17:04.22 is an autocatalytic process driven by RNA-catalyzed RNA replication. 00:17:10.11 So now, any mutation that leads to faster, 00:17:14.01 more efficient replication of the genetic material 00:17:17.18 will automatically go along with faster growth 00:17:20.12 of the cell as a whole, 00:17:22.04 and it's a competitive form of growth 00:17:24.10 that results from these cells essentially eating their neighbors. 00:17:30.06 So that was a very exciting conceptual advance, 00:17:34.17 but there was always this nagging problem. 00:17:38.09 That here, we're looking at growth driven by osmotic forces, 00:17:43.15 which means that these are growing as spherical vesicles. 00:17:48.09 And dividing a spherical vesicle, 00:17:50.13 especially one that's osmotically swollen, 00:17:52.25 is something that's actually quite difficult to do. 00:17:55.08 It takes a lot of energy, 00:17:57.10 and as I said in the case of the extrusion experiments, 00:18:00.18 results in loss of some of the contents, 00:18:02.27 which is not very satisfying. 00:18:05.04 So, we were left thinking for a long time 00:18:08.23 about possible alternative membranes 00:18:11.17 and alternative division processes, 00:18:15.07 and that's something we'll come back to later. 00:18:19.28 So, let's think about more natural systems. 00:18:27.24 Everything that we've looked at up till now 00:18:30.26 has been highly constrained, very artificial laboratory system 00:18:38.16 designed so that we can follow what's happening 00:18:42.29 analytically in a very clear and simple way. 00:18:46.12 But of course, the vesicles that form naturally 00:18:51.21 aren't so nicely constrained. 00:18:55.04 They're extremely heterogeneous, 00:18:58.19 and that generates a system which is much harder to study experimentally. 00:19:03.26 So that's illustrated by this image, 00:19:07.24 which shows the range in the diversity of vesicles that are formed 00:19:11.16 when you just shake up oleic acid with some salt and buffer. 00:19:15.18 You get these huge vesicles, tiny vesicles, 00:19:18.29 compound vesicles, it's a terrible mixture, 00:19:23.27 and if you imagine adding some more oleic acid to 00:19:27.08 this in the form of micelles, 00:19:28.25 so that everything grew a little bit, 00:19:31.22 you couldn't tell that anything had happened. 00:19:34.27 So we need new analytical ways of looking at this, 00:19:38.14 or new ways of preparing vesicles so that we can actually 00:19:42.13 watch what's happening. 00:19:44.08 And our major advance in that area came from the 00:19:48.06 work of a new graduate student, Ting Zhu, 00:19:51.28 who had a background in mechanical engineering 00:19:54.14 and proposed a very simple and elegant way of 00:19:57.25 preparing more or less uniform populations of vesicles 00:20:01.23 that were large enough to just look at in the microscope. 00:20:06.04 So for the first time, we were able to do experiments 00:20:09.08 where we take vesicles, add food, and just watch how they grow. 00:20:14.09 And so what you can see in this image is a field of vesicles, 00:20:18.00 they're all roughly four microns in diameter, 00:20:21.24 so big enough to image optically, 00:20:25.29 and what you're actually seeing here is a fluorescent dye 00:20:29.24 encapsulated in the aqueous internal space of each vesicle. 00:20:35.05 So now, what you'll see next is a time-lapse video 00:20:41.01 of how these things actually grow following the addition 00:20:44.17 of more fatty acids. 00:20:47.04 What we expected was simply that these spherical vesicles 00:20:51.26 would swell up a little bit, we'd just watch them get bigger. 00:20:56.19 We thought that, if the surface area grew faster than the volume, 00:21:01.11 they might become somewhat elongated in shape. 00:21:05.24 But we did not expect what actually happened, 00:21:09.05 which is quite dramatic. 00:21:12.16 So, what you see here is that, initially, 00:21:15.24 a faint filament emerging from the parental spherical vesicle, 00:21:20.06 and over time that filament grows and absorbs 00:21:23.21 all of the material in the starting spherical vesicle, 00:21:27.21 which has now grown into a long, thin, branched, 00:21:32.03 filamentous morphology. 00:21:35.13 So, this was a very surprising observation. 00:21:41.05 It immediately raised a host of mechanistic questions, 00:21:45.16 which we're still investigating. 00:21:48.01 But it, very importantly, immediately solved the problem 00:21:52.27 of how cell division might happen in a manner 00:21:56.29 driven by environmental fluctuations. 00:21:59.28 As I said, spherical vesicles are hard to divide, 00:22:02.23 it takes a lot of energy. 00:22:04.15 But these structures are incredibly fragile, 00:22:07.17 and all you have to do it shake them up. 00:22:11.21 Gentle agitation is sufficient to cause them to divide, 00:22:16.09 and you can see that process in action in the next video. 00:22:21.13 So here you see the pearling instability: 00:22:25.07 The filament forms a set of beads, 00:22:28.16 and then they snap apart and generate fragments, 00:22:32.27 which eventually round up into spherical daughter vesicles. 00:22:36.25 And so this process is driven by very gentle perturbations 00:22:43.01 in the fluid, very mild shear forces. 00:22:46.08 You can easily imagine the shear forces generated by 00:22:50.15 wave action on a shallow pond leading to cell division 00:22:54.06 in this kind of process. 00:22:56.27 Now, this process of growth followed by shear-induced division 00:23:03.10 makes a cycle that can be iterated indefinitely. 00:23:07.09 So you can start off with a spherical vesicle, 00:23:09.20 grow it into a filament, divide it into spherical daughter vesicles, 00:23:14.00 which can grow again. 00:23:16.03 If they sit around, the volume increases, 00:23:18.20 they round up, then you can grow again, divide, grow. 00:23:23.22 This cycle of growth and division can be carried on indefinitely. 00:23:29.19 So, let's think for a minute about some of the 00:23:32.29 mechanistic processes that are involved. 00:23:35.10 We don't understand everything that's going on, 00:23:37.13 but at least we have some experiments that tell us 00:23:41.21 what some of the critical factors are. 00:23:45.11 So, what we think is going on, 00:23:47.21 and what the important factors are, 00:23:49.26 is that surface growth is happening much faster than volume growth, 00:23:57.07 and so that makes a shape change inevitable. 00:24:01.15 Volume is conserved on the timescale of growth 00:24:04.16 because most of the solvents are only very slowly permeable 00:24:08.05 across these membranes, 00:24:10.05 and the solvent can be anything from the rather artificial buffer that we use, 00:24:16.13 to a mixture of amino acids, 00:24:18.12 such as you would get from Miller-Urey-type syntheses. 00:24:22.00 The other and somewhat surprising factor that's critical is the 00:24:26.07 multilamellar structure of these vesicles. 00:24:29.03 So these large vesicles tend to have several bilayers closely apposed, 00:24:35.29 as their sort of outer structure. 00:24:38.25 What we think happens is that when we add food molecules, 00:24:41.23 they first integrate into the outermost bilayer, 00:24:46.28 there's very little space between it and the next bilayer, 00:24:51.00 and so that extra surface area is extruded as a 00:24:55.01 thin, initial filament, which then subsequently elongates 00:25:01.27 into this long, tubular, filamentous structure. 00:25:05.21 We really don't understand the physical forces 00:25:08.04 that are involved in those later stages, 00:25:11.25 but we have some ideas about the initiation phases. 00:25:16.04 So I want to show you first an experiment that shows 00:25:19.25 that volume conservation is very important. 00:25:22.23 If you use a buffer that is very, very rapidly permeable, 00:25:27.24 then you don't see this kind of process at all. 00:25:30.22 Instead, what happens is that the outermost bilayer just balloons up. 00:25:36.25 As surface area increases, volume keeps up, 00:25:40.10 the volume expands, and so you end up with 00:25:43.18 most of the bilayers almost unchanged, 00:25:48.12 and trapped within a much larger unilamellar vesicle. 00:25:53.26 Now, the second important aspect is how 00:25:58.18 multilamellar vesicles grow in contrast to how unilamellar vesicles grow. 00:26:04.10 So when you have these multiple bilayers, as I suggested, 00:26:08.17 it looks like, from these confocal images, 00:26:11.00 that it's the outermost bilayer that grows first, 00:26:14.13 and that extra surface area is extruded as a thin tubule. 00:26:18.21 Eventually, you get a long filament, 00:26:21.04 and when you divide that by mild shear forces, 00:26:25.11 you get back small, multilamellar vesicles, 00:26:29.05 and then you can repeat the cycle. 00:26:31.11 Now, something very different happens if you go to the 00:26:33.22 trouble of making a unilamellar initial vesicle. 00:26:38.16 This grows the way we intuitively thought it would grow: 00:26:41.08 It just expands, becomes somewhat elongated. 00:26:45.25 But now these structures are extremely delicate and unstable, 00:26:50.17 and when they're subjected to shear force fluctuations, 00:26:54.19 they tend to rip apart, the contents spill out, 00:26:57.25 and remarkably, they reassemble into multilamellar structures. 00:27:03.19 So, to summarize this, 00:27:05.23 the two important aspects of the overall process 00:27:09.04 are beginning with multilamellar vesicles 00:27:11.28 and having something in the surrounding solution 00:27:16.08 which can only permeate slowly across the membrane, 00:27:20.02 resulting in slow volume increase 00:27:23.26 compared to rapid surface area increase. 00:27:26.19 And then environmental perturbations lead to cell division 00:27:29.25 in a very simple and natural way. 00:27:34.02 So, to summarize this, in a series of stages, 00:27:38.22 we've gone from a very artificial laboratory simulation, 00:27:43.19 to a process that looks simple and robust enough 00:27:47.06 to have occurred on early Earth environments. 00:27:53.04 Now, more recently, we've come back to the idea of competition 00:27:59.17 between protocells as an alternative mechanism of driving growth. 00:28:05.18 And this is particularly interesting in terms 00:28:09.11 of a way that this could emerge from Darwinian processes. 00:28:16.20 We think that one of the first ribozymes to emerge, 00:28:20.10 perhaps THE first, was actually a ribozyme 00:28:23.21 that could catalyze the synthesis of phospholipids. 00:28:28.15 And so why do we think this? 00:28:30.08 This came out of considering the series of evolutionary transitions 00:28:35.24 that must have occurred to get from 00:28:39.20 primitive membranes based on fatty acids, 00:28:42.05 with the appropriate dynamic and exchange processes 00:28:46.13 for a primitive cell, 00:28:48.01 up to modern membranes, which are based on phospholipids, 00:28:52.18 which are good barriers and which rely on 00:28:55.15 highly evolved biochemical machinery to mediate 00:28:59.07 growth and division and the transport of molecules 00:29:03.20 across the membrane. 00:29:04.27 So you couldn't go directly from the primitive state 00:29:08.20 to the modern state. 00:29:10.10 That would be suicidal, 00:29:11.21 because now a cell that depended on the influx 00:29:14.27 of nutrients from the environment 00:29:16.18 would no longer have access to those nutrients 00:29:18.29 because these membranes are so impermeable. 00:29:22.12 So presumably, the transition had to occur in a series of stages, 00:29:27.02 and what would an intermediate stage look like? 00:29:31.19 Probably it would look like a fatty acid membrane 00:29:35.09 with a small amount of phospholipid doped into the structure. 00:29:41.02 So if you think of a protocell containing genetic molecules, 00:29:46.21 if one of those sequences happens to be a simple 00:29:49.26 acyltransfer ribozyme that could make phospholipids 00:29:54.11 (after all, it's a simple, single, one-step acyltransfer reaction), 00:29:59.05 you could imagine making just a little bit of 00:30:02.09 phospholipid as an initial step in this transformation. 00:30:07.14 The thing that puzzled us for a long time is 00:30:11.11 what possible selective advantage could there be to 00:30:16.11 the presence of a small amount of phospholipid 00:30:19.16 in a fatty acid background? 00:30:22.24 It was very hard to imagine any reason why that would be advantageous. 00:30:27.21 It didn't really seem that a little bit of phospholipid 00:30:31.05 would change the physical properties of the membrane 00:30:33.18 in any particularly useful way, 00:30:37.06 and so the answer to this conundrum 00:30:39.29 came from doing experiments, 00:30:42.10 and this is all work done by Itay Budin, 00:30:45.06 who's a graduate student in the lab. 00:30:47.25 So what Itay did was simply to prepare vesicles 00:30:53.13 containing a little bit of phospholipids 00:30:56.28 doped into a fatty acid membrane structure. 00:31:01.11 And then he asked, well, 00:31:03.00 how would that affect the properties of the membrane? 00:31:05.20 And what turned out to be particularly interesting 00:31:08.12 is what happens when these vesicles are mixed with 00:31:11.18 pure fatty acid vesicles. 00:31:14.23 And it turns out, as you can see on the next slide, 00:31:18.14 that even a little bit of phospholipid confers a really dramatic ability 00:31:24.12 on that vesicle to absorb fatty acids from its 00:31:28.02 neighboring fatty acid vesicles. 00:31:31.13 So here's the experiment that Itay did: 00:31:34.11 He prepared vesicles with 10% of a phospholipid 00:31:41.11 in an oleic acid background, 00:31:43.19 and mixed them with a large excess of pure, 00:31:46.02 fatty acid, oleic acid vesicles. 00:31:48.23 So the vesicle starts out as a spherical structure, 00:31:51.13 you can see the encapsulated dye. 00:31:53.26 Within seconds, you see the protrusion of small filaments, 00:31:59.00 and over time, you can see a transformation 00:32:01.17 into a filamentous structure, 00:32:05.04 much as you saw before following the addition 00:32:08.07 of a large excess of fatty acids. 00:32:11.07 But this time the process of growth being driven purely 00:32:16.20 by the initial vesicle absorbing fatty acids 00:32:20.03 from its neighboring molecules. 00:32:21.27 It's essentially growing by eating its neighbors. 00:32:24.26 The mechanism of this phospholipid-driven growth is extremely interesting, 00:32:30.01 and you can see what we think is going on in the next slide. 00:32:34.16 It turns out there are actually two processes: 00:32:37.07 The first is a simple, entropically favored process 00:32:40.17 in which the absorption of fatty acids molecules into the mixed bilayer 00:32:46.04 dilutes the phospholipids, 00:32:50.10 which are extremely insoluble in the surrounding space, 00:32:53.19 so this is an entropically driven dilution. 00:32:58.04 But it turns out that, in addition to that factor, 00:33:01.02 there's actually a quantitatively more important factor, 00:33:05.00 which is the fact that these phospholipid molecules 00:33:09.01 change the properties of the bilayer such that 00:33:13.14 the disassociation of fatty acids from the bilayer is slowed down. 00:33:18.25 As a result, the vesicles grow because the 00:33:24.06 association rate constant is the same, 00:33:27.27 but the dissociation is slowed down, 00:33:30.27 so they hold onto more of their molecules, they grow, 00:33:34.10 and the pure fatty acid vesicles therefore shrink. 00:33:37.21 So how do we demonstrate this experimentally? 00:33:41.06 Itay took advantage of an assay that was developed by Jim Hamilton. 00:33:48.00 It's a rather indirect but very effective assay. 00:33:51.00 So if you have a fatty acid vesicle, 00:33:52.26 you can monitor the dissociation of 00:33:56.04 fatty acid molecules from that membrane, which is a slow step, 00:34:02.11 by looking at how they dissociate into solution, 00:34:07.06 integrate into the membrane of a reporter vesicle, 00:34:10.26 which is a fatty acid vesicle that contains inside it 00:34:15.20 a fluorescent dye that's pH-sensitive. 00:34:18.20 Now when the fatty acid molecules integrate into the outer leaflet, 00:34:23.05 they flip rapidly to the inner leaflet. 00:34:26.10 It's the protonated form that flips because it's neutral, 00:34:29.16 it can flip very rapidly, and then it re-equilibrates on the inside, 00:34:34.10 releasing protons, which acidify the interior. 00:34:38.05 So we have a fluorescence assay for this pH decrease, 00:34:42.16 and that gives us essentially a real-time assay for this 00:34:46.01 initial slow step of dissociation. 00:34:49.17 You can see that this assay works by comparing 00:34:53.25 the dissociation of fatty acids of different chain lengths, 00:34:58.00 and so here you see that oleic acid, which is an 18-carbon chain, 00:35:03.16 dissociates fairly slowly. 00:35:06.13 Palmitoleic acid, two carbons shorter, dissociates more rapidly. 00:35:11.19 And myristoleic acid, 14 carbons, dissociates extremely rapidly. 00:35:17.06 So it gives you a good readout for the dissociation kinetics, 00:35:21.09 and that allows you to monitor the rate of dissociation 00:35:26.21 as a function of phospholipid content of the membrane. 00:35:31.17 And so if you start off with a pure fatty acid membrane, 00:35:35.28 so a pure oleic acid membrane, dissociation is quite rapid. 00:35:40.21 But the more phospholipid that there is in the membrane, 00:35:44.25 the slower the off-rate of the fatty acid is. 00:35:51.17 So, as a result of these two processes, 00:35:56.24 this entropically driven dilution aspect 00:36:00.19 and this off-rate-driven aspect, 00:36:03.11 the presence of even a little bit of phospholipid will drive growth 00:36:08.29 by essentially extracting or absorbing fatty acid molecules 00:36:13.09 from neighboring vesicles that have either no phospholipid 00:36:17.25 or even just less phospholipid. 00:36:20.24 So what would the implications of that process be 00:36:23.15 in an evolutionary scenario? 00:36:27.15 We begin with this idea that phospholipids drive growth. 00:36:31.12 As a result, of course there would be a very strong selection 00:36:35.08 for a ribozyme that has the catalytic ability 00:36:40.08 to make phospholipids, so a simple acyltransferase ribozyme. 00:36:48.07 That kind of primitive cell would rapidly take over 00:36:51.24 the population because it has such a strong selective advantage, 00:36:55.23 but that would lead to an evolutionary arms race. 00:36:59.11 Because now, in order to grow by eating your neighbors, 00:37:03.15 you have to make more phospholipids than they do. 00:37:07.11 So there'd be an evolutionary race to make 00:37:11.29 more and more and more phospholipids. 00:37:14.27 And so now the phospholipid abundance in the membrane 00:37:19.13 would start to increase, you'd have more and more phospholipids, 00:37:23.05 and eventually that would lead to new selective pressures 00:37:26.25 favoring the emergence of metabolism, 00:37:29.09 because now it's actually useful to have internal metabolic reactions 00:37:33.24 because you get to keep the molecules you make 00:37:36.00 instead of having them just leak out, 00:37:38.20 since the phospholipid makes the membrane less permeable. 00:37:42.09 At the same time, there would be selective pressures 00:37:45.03 for the emergence of transport machinery, 00:37:48.23 which would help you absorb molecules that you still want 00:37:52.05 to have access to from the environment, 00:37:54.00 or release molecules that you want to get rid of, 00:37:56.22 so waste products. 00:37:58.15 So we think that this physical process, 00:38:01.26 this unexpected physical process of phospholipid-driven membrane growth, 00:38:06.05 would set in place a whole cascade of events eventually 00:38:11.20 leading to selective forces for much more complicated cells, 00:38:17.25 where we see the emergence of metabolism and 00:38:20.09 membrane transport machinery, 00:38:22.11 and eventually, we think that could drive the emergence of 00:38:26.10 protein synthesis, initially uncoded but eventually coded protein synthesis, 00:38:32.08 and be a major factor in the emergence of modern biochemistry. 00:38:39.25 So from all of these experiments, just to sum up, 00:38:42.23 what can we say that we've learned that's really relevant 00:38:46.03 to the origin of life on Earth? 00:38:47.22 And I think at at general level, 00:38:49.03 there are two things I'd like to point out: 00:38:52.01 One is that, from doing very, very simple laboratory experiments, 00:38:56.14 again and again we've uncovered a surprising new 00:38:59.08 physical phenomenon, things that you would have a hard time 00:39:03.05 thinking up a priori, but which emerge 00:39:07.16 naturally from doing laboratory experiments 00:39:10.08 where you're trying to build systems and see how they work. 00:39:13.08 The second general lesson, I think, 00:39:16.19 is that in our early experiments, 00:39:20.08 we tend to use very highly constrained artificial systems 00:39:24.23 because they're analytically tractable. 00:39:27.20 But they often don't exhibit the properties 00:39:30.28 that we really want to see that are relevant to the prebiotic system. 00:39:35.24 So it's very important to relax those experimental constraints 00:39:41.20 and gradually develop the ability to look at "messier" 00:39:45.00 and more natural systems that give you insight 00:39:48.01 into physical processes that really could explain some of the 00:39:52.21 apparently hard parts in the process by which life emerged 00:39:57.17 from the chemistry of the early planet. 00:40:01.02 So, again, all of this work was done by many very talented 00:40:04.08 students and postdocs. 00:40:06.26 I've mentioned in particular the work of Itay Budin and Ting Zhu, 00:40:10.28 and the early work of Irene Chen, Shelly Fujikawa, Martin Hanczyc. 00:40:17.09 Many people have played a role in developing these systems 00:40:21.09 and gradually giving us the ability to understand what's going on 00:40:26.13 in these model systems and some insight into the origin of life. 00:40:31.02 Thank you.