What Genomes Can Tell Us About the Past

Part 1: Genomes Tell Us About the Past

00:00:12.05 Well, I'm going to talk to you today about the most fascinating
00:00:18.17 problems in biology that anyone can imagine.
00:00:23.25 And the interesting thing is that you don't need a big lab to do this work.
00:00:31.28 You can do it at home. You can do it where ever you are.
00:00:38.05 But before I explain this, let me tell you about a very important development
00:00:47.29 that took place in the last century (that's the 20th century)
00:00:52.18 in regards to our knowledge of the history of the universe.
00:00:58.07 See, what I'm going to talk about is the reconstruction of a past.
00:01:04.13 And if we can find scientific evidence for that
00:01:09.27 so that we could recreate the past at least in theory.
00:01:14.26 That seems to me to be an enormously challenging problem for us to solve.
00:01:22.24 Now what happened in physics was the following
00:01:28.02 it soon became realized after Einstein's principles of relativity were established
00:01:37.28 that light, which has a constant speed in the universe,
00:01:44.17 that light coming from afar actually takes time to reach us.
00:01:51.03 So, the further out in the universe we can look,
00:01:57.18 the earlier the events we are seeing.
00:02:00.27 The light left there many, many millions of years ago.
00:02:07.16 And so the question is: if you could measure the distance of these remote stars, galaxies,
00:02:17.21 you could begin to measure exactly what they tell you about the early stages of the universe.
00:02:27.09 And of course we know have a means of measuring them.
00:02:31.03 It's called the red shift because this is a function of distance and
00:02:37.28 it was noted that the lines of the elements in the spectrum were shifted to the red.
00:02:46.06 The further the light came to us, the more the red shift took place.
00:02:53.29 And so, we can therefore, knowing that distance,
00:02:58.23 tie those two things together and
00:03:02.22 we can now begin to reconstruct events that took place in the remote past.
00:03:09.26 And so the question we can ask for our science of biology:
00:03:17.02 Is there anything like that?
00:03:19.13 Do we have within ourselves, so to speak, evidence of the remote past?
00:03:27.28 Now, as all of you know, over the last few decades and particularly over the last few years,
00:03:37.12 there has enormous development of DNA sequencing techniques.
00:03:46.04 And as a result of this, the genomes of a large number of organisms
00:03:52.19 have been sequenced, including the genome of ourselves,
00:03:58.04 of our nearby primate cousins (chimpanzee, the orangutan),
00:04:05.10 and going all the way back through many invertebrates
00:04:11.04 and going all the way back into the unicellular organisms and many bacteria as well.
00:04:19.16 And what we'd like to know by the study of these genomes we'd like to know two things of course.
00:04:29.00 The first is, could we actually read and understand these genomes?
00:04:36.00 Could you pick up a genome, look at it and say, "Oh, yes. This is the genome of a zebra."?
00:04:43.16 That would need an understanding of almost all biology.
00:04:48.17 It is very remote for us to do this at the present moment.
00:04:54.01 We can do a little in microorganisms. We can recognize some proteins that they may produce.
00:05:03.00 But, for the main, that still remains an enormous challenge;
00:05:08.29 the interpretation of the meaning, that is the function, of these genomes.
00:05:15.23 This is important because as you all know biological systems contain
00:05:23.11 a complete description of what they are and will be in the DNA code of their genomes.
00:05:34.23 And of course, since that is what is propagated, that is the structure that
00:05:41.17 is changing during evolution with the consequent changes in the phenotype.
00:05:47.28 But the problem I'll deal with today is not so much that problem,
00:05:55.26 which in itself is enormously challenging, and will occupy biologists for a long time to come,
00:06:03.29 but with the other one,
00:06:06.01 which is to ask: Can we see anything in these sequences that tell us of the remote past?
00:06:16.08 And what has happened now, very frequently,
00:06:20.07 is that a large number of people have begun to compare these sequences,
00:06:27.06 notice and calculate the differences between them.
00:06:32.29 So, for example, the chimpanzee genome can be compared with the human genome.
00:06:43.05 They are extremely similar, about 99% the same.
00:06:48.20 And we can also compare this with animals that are more distantly related to us
00:06:56.07 such as the mouse and the cow and even down to frogs, lizards, fish and so on.
00:07:03.24 What people then do is pick a region of the genome, pick a gene as they say.
00:07:11.04 Then, make an alignment, that is, put the things into line,
00:07:16.11 because amongst the many mutations that occur are insertions and deletions.
00:07:23.24 And then one can compare the two and count the changes.
00:07:30.10 And of course, it stands to reason that the greater the number of differences,
00:07:37.15 the greater the number...the more distant these two organisms are.
00:07:45.00 Now, you will see that if you do this comparison, you get out of this
00:07:53.15 a number which is essentially compounded of two other numbers.
00:08:00.26 The two other numbers are the time in the past at which the two organisms diverged
00:08:12.10 and the rate at which they diverged.
00:08:17.07 So you can see you get exactly the same if
00:08:21.15 you go twice as far into the past and evolve at half the speed.
00:08:28.06 The numbers will come out exactly the same.
00:08:31.20 So how is this dilemma resolved?
00:08:36.19 In most cases, it is by accepting a number of assumptions
00:08:43.18 and one of the difficulties of this field is the number of hidden assumptions that
00:08:51.19 lie behind all the programs and equations that people use to calculate these differences.
00:09:02.04 Now, I think that unveiling these assumptions is one of the best ways of approaching this.
00:09:13.21 And in order to do this, I think we need to focus on a particular case.
00:09:22.03 What I'm going to do for the rest of this talk is to ask a lot of questions
00:09:27.27 about mice and men. How distant are we?
00:09:31.26 And what can we learn, really, about the comparison?
00:09:37.21 Now, of course, I'd like to make some other, more general comments first.
00:09:45.26 And those deal with the difference between genes that change under selection,
00:09:56.07 that is a favorable mutation that enhances the reproductive success of the animals it is found in,
00:10:06.03 will of course, come to then be, as we say, fixed in the population by selection.
00:10:16.00 This can be extremely rapid and that is one form of change.
00:10:26.07 There's another form of change where the actual change has no effect on
00:10:31.21 the phenotype so far as it effects reproductive success neither one way or another.
00:10:42.22 These are called neutral mutations and it has been assumed that
00:10:51.05 if there are genes under selection, then, of course, the genomes will change more quickly
00:10:59.05 and if there are genes that are not under selection,
00:11:05.02 then they will become established by a process that is called genetic drift.
00:11:14.11 I will explain these in a moment but before I go on, I want to say one other thing.
00:11:23.06 A perplexing problem in the study of the genomes of higher organisms like ourselves,
00:11:32.14 is that most of the genome appears to be unnecessary. It is, if you like to call it, rubbish.
00:11:41.11 Now, as you well know, there are two kinds of rubbish; the rubbish you keep,
00:11:46.13 which we call junk and the rubbish you throw away which we call garbage.
00:11:51.27 And the question has been debated fast and furious over many decades
00:11:59.28 what was these meaningless sequences that we find in our genome and
00:12:09.20 in fact in our genome it could be as much as 95% of the sequence.
00:12:15.22 Now, in this study, you're not allowed to have the issue that
00:12:23.19 genomes can plan their future because they obviously can't.
00:12:29.06 You can imagine this if you think of the primordial bacterium in the primordial soup saying,
00:12:38.03 "I can't change this gene because I'm going to need it in 2 billion years to make actin."
00:12:45.01 So that is absolutely not possible to do.
00:12:49.16 So, you have to ask whether what does this junk do?
00:13:01.08 The best answer I know is that it does nothing.
00:13:05.17 It doesn't do you any harm, so you have it.
00:13:08.23 It doesn't do you any good so it might as well...you could lose bits of it.
00:13:15.10 It doesn't have any effect at all.
00:13:18.23 It is certainly the balance between two processes that go on in all genomes
00:13:25.27 which is the increase of DNA and the loss of DNA.
00:13:32.09 And these are events which take place in our genomes
00:13:39.27 in which genomes increase in size with one process and genomes reduce in size by another
00:13:49.03 and usually the balance between these will depend on whether the genome expands or contracts.
00:14:00.17 Now, in at least mammalian genomes, there are events
00:14:06.23 that lead to massive expansion of the genome and these are, effectively, transposons;
00:14:17.10 that is, pieces of DNA that can copy themselves more than once in each cycle.
00:14:24.00 And there are events in our germ line that allow these expansions
00:14:30.24 of a large number of elements, composing at least half the DNA.
00:14:37.08 One element called the MU sequence composes about 27% of your sequence,
00:14:47.02 which is a lot. There's millions of copies of this strewn all over the genome.
00:14:54.00 The fact that sometimes these will be used doesn't prove...
00:14:59.20 in fact, they can be used but purely incidentally. If they are used, they will be used.
00:15:08.06 And the problem then we have is of how we're going to analyze
00:15:13.21 the parts of the genome that we want to get information about.
00:15:22.15 So what I have been interested in is analyzing coding sequences.
00:15:31.04 Now, coding sequences, of course, specify your proteins.
00:15:36.16 They are the ones which are most conserved because
00:15:40.23 you have to conserve them in order that you have a protein that is functional.
00:15:46.21 But they change partly because some amino acids can easily be substituted for others.
00:15:54.25 And indeed, sometimes they acquire new function as well.
00:16:00.29 The interesting thing about the genetic code, is that parts of that sequence do not matter.
00:16:10.18 That is, the codon are, as we say, degenerate.
00:16:16.06 So you can make the same protein by using different genetic sequences
00:16:24.18 and will make the identical protein.
00:16:27.15 And, without getting into detail,
00:16:32.06 you will know that some amino acids have two codons coding for them.
00:16:43.19 They differ only in whether the third base is an A or a G or in other cases a C or a T.
00:16:54.17 And there are some other codons which are coded for by all four bases.
00:17:00.20 For example, a glycine is coded for by GGX where X can be A, G, C, or T.
00:17:11.10 So that in the code itself, in the protein sequences that we find in the code,
00:17:21.26 that we find in the genome, we are able to look through these
00:17:30.05 at the frequency of neutral changes
00:17:34.16 without having to have any postulated selection event.
00:17:41.28 So let's ask ourselves whether we can see in rigorously conserved amino acids,
00:17:51.26 that's lysine, whether there are changes in a given lysine, in a given protein
00:18:00.24 which have happened throughout the lineage of organisms
00:18:05.13 whether in some it's coded by a G and in others it's coded by an A.
00:18:13.12 Because that change is neutral.
00:18:17.25 And in the same way, we can look at the four codon amino acids
00:18:26.02 and ask whether we can have all four possibilities at the same place,
00:18:33.06 keeping the protein completely the same, and whether there are these switches.
00:18:40.04 Now, I want to draw your attention to one of these.
00:18:45.00 If I look in the third base of a codon,
00:18:54.22 like leucine or lysine, I will see that sometimes I have in one organism
00:19:09.28 I have CTG and if I go down, I can have in another organism CTC.
00:19:21.04 Of course, I can have A and T as well.
00:19:24.28 But I only want to consider these two.
00:19:27.22 The reason for this is that if you see: that switches to that.
00:19:35.19 That mutation from G to C is technically a mutation in the DNA of a G-C pair to a C-G pair.
00:19:49.25 And simply, the base pair orientation, which does not matter in this case,
00:19:58.22 that is, it just depends whether G is on the Watson strand and C is on the Crick strand
00:20:06.20 or C is on the Watson strand and G is on the Crick strand.
00:20:11.07 So then the forward and reverse mutation rate is exactly the same.
00:20:20.02 So that if we take every gene we want to compare in this case,
00:20:28.28 and we now look at the switches from G to C and vice versa,
00:20:40.15 that is we decompose every protein we compare into a string of Gs and Cs
00:20:52.00 just looking at the third bases of comparative codons.
00:20:56.02 So one protein might go something like this.
00:21:02.10 We don't worry about the distances between them.
00:21:05.18 And the other one might look like this.
00:21:23.08 OK. So, in this one there's been an exchange of a C for a G. We'll call it a flip
00:21:31.06 because that's essentially the base pairs flipped over.
00:21:34.28 Here's another flip and there's another flip.
00:21:39.10 So, if you like, these are separated by three flips.
00:21:48.06 So we can count the separation. This turns out to be, fortunately, a slight process.
00:21:55.21 And when these are counted for, we can compare genes from different organism
00:22:03.11 We will find that, as expected, the further the organism is back in the evolutionary tree,
00:22:13.07 the less frequent are these flips.
00:22:20.06 So, for example, if I compare human with a cow (bos), then this number is 0.94.
00:22:40.14 If I compare human with a mouse or the rat, you get the same number,
00:22:51.18 which is fine because rats and mice...I mean after all, mouse is just a little rat.
00:22:58.15 I get a number 0.90.
00:23:03.24 If I take the human and I compared it with the chicken, that number is 0.70.
00:23:10.24 And if I compare this with the fish, I get numbers that are 0.53.
00:23:24.17 Now, you'll notice that 0.5 means that these are uncorrelated.
00:23:33.10 You can't get lower than 0.5. It means then there's no history left in the genome.
00:23:41.04 So every time you can see you can find, by comparing two organisms,
00:23:48.15 you can find a value that is above 0.5 for this measure,
00:23:57.08 only then can you say that we have information and this is what the information means.
00:24:06.18 If it's less than 0.5, there's no information.
00:24:10.05 It could be any time in the past. In fact, the history is lost in the mists of time.
00:24:19.17 So, now what do these numbers mean?
00:24:24.01 And this is the sort of measure...
00:24:26.15 this just happens to be a very particular one that I choose because
00:24:31.20 I didn't want to get confused between differing rates of forward and reverse mutation.
00:24:39.05 This is symmetrical, it doesn't matter anything about the organism
00:24:44.22 and if the chemistry changes or there's a repair process
00:24:50.22 that works in one direction, it has to work in the other direction at the same time.
00:24:56.08 So, this is the actual measure that is free from all kinds of other interpretations
00:25:07.00 and gives you this sort of measure.
00:25:09.13 Very well. So, now, what does this mean?
00:25:12.23 Now, in general, we want to find out how far back the separation time is.
00:25:21.10 We want to do phylogeny. What do people normally do?
00:25:28.14 What people normally do is just say, "Well, there are 3 differences.
00:25:35.19 How many of them occurred in the human? And how many of them occurred in the mouse?"
00:25:42.08 Because, this could have been a C in the precursor organism,
00:25:50.00 and changed to a G in the mouse, and this could have been a G and changed to a C in the human.
00:25:58.08 Well, what people normally do is split the difference.
00:26:03.18 In other words, they'd say, "Well, of the changes here, there's 10% flip,
00:26:10.27 we'll put 5% in the human (0.05) and we'll put 0.05 in the mouse."
00:26:22.26 Now, you can't do that because, as I'll show you in a moment,
00:26:29.16 mice and men are evolving at different rates.
00:26:34.04 And this is the first time that this paradox can actually be resolved.
00:26:40.08 And I'll go through the argument as it goes.
00:26:45.00 I also point out that before we go to anything else,
00:26:55.11 you can't tell from the absence of any information, which direction evolution is preceding.
00:27:06.22 For example, if I compare a fish with a human genome,
00:27:13.16 I see there are lots of differences.
00:27:16.11 I see the human appears to have more genes and so on
00:27:21.15 and I can have two equal hypotheses in the absence of any other information.
00:27:30.19 One is, half a billion years ago there were a lot of people running around
00:27:36.23 and they degenerated into fish. Some of them, of course, they left some humans behind.
00:27:44.14 Or other other hypothesis is, half a billion years ago, there were lots of fish around,
00:27:52.29 and they actually evolved the human genome.
00:28:00.25 So you can't tell where you gained, so to speak, the sign is plus or minus
00:28:05.26 in the absence of other information.
00:28:08.20 And the information we have is the fossil record.
00:28:12.15 And that's very important information
00:28:16.01 because it tells us what was around there which is the origin of this
00:28:26.02 and it also can give us a very good date.
00:28:29.22 So, remember when you are doing this,
00:28:33.29 you must have something that gives you a clock in a sense,
00:28:40.00 but it can't be a clock in the organism as I will show you now.
00:28:45.17 Many people thought that the rate of mutation remained constant over time.
00:28:51.29 They thought there was a molecular clock but I think the molecular clock hypothesis is broken
00:28:59.20 because the clock rates, as you will see in a moment, differ markedly.
00:29:05.10 Very well. So, what I want to know is, of all those changes,
00:29:13.00 and you can go through lots and lots of genes and collect these changes.
00:29:17.09 It's a lot of work.
00:29:18.15 But the question now arises: I can't just split them between the mouse and the human.
00:29:26.08 I must have a way of telling how many of these changes occurred in the mouse.
00:29:32.29 And so what I can do is take an organism that's further back in evolution
00:29:40.10 for which I have good evidence lies further back,
00:29:44.01 and take that organism and see what its gene is.
00:29:49.19 So that if we have one that branched off earlier...
00:29:55.10 So let's just say that what we're going to have here is the human (and I'll draw it here).
00:30:03.16 Then we've got, at some point, the mouse branched off. I'll say that's the branch.
00:30:13.00 And if I have another branch that's further back, giving me another organism over here,
00:30:21.23 and I look in this organism's genes, then I can actually say that,
00:30:29.11 if the human has got the background and it's also present in this organism,
00:30:36.01 then the change happened in the mouse.
00:30:39.25 Of course, I'm going to make a few mistakes
00:30:43.10 because this one happened to change in one other direction but we can correct for that.
00:30:50.20 Now, if you do this on this, going through now three sets of genes,
00:30:56.14 one from the cow, one form the human, one from the mouse,
00:31:02.00 you find that, in this case, 2.7 or so changes occurred in the mouse
00:31:17.27 to every one change in the human.
00:31:24.14 In other words, mice are evolving...
00:31:27.20 and this can be said... we don't want to give, we don't have to give any reasons for it,
00:31:36.14 we don't have to say this is the way they behave
00:31:40.02 because everything we know about modern mice
00:31:42.14 would be very hard to project backwards in time to this branch for all we know.
00:31:48.20 What branched off here, what branched off here were very different animals all together.
00:31:55.01 But, we can certainly say that the length of this for the mouse,
00:32:03.02 is 2.7 times the length for the human.
00:32:08.10 And you can do the same for the cow. You can do the same for a lot of organisms.
00:32:15.16 In fact, what we can discover from this is now a very interesting paradox
00:32:24.09 because we have strong evidence, because it doesn't matter...if I'll just do this now for you.
00:32:32.21 We can then draw this line here.
00:32:48.21 Which must be 2.7 times the length of this line. This is the human.
00:32:58.11 So basically let us say that the fossil evidence suggests that
00:33:05.05 this is -80 million years, that point of division is 80 million years.
00:33:13.27 Then I have to conclude from this that mice are
00:33:19.12 on the human scale plus 125 million years in the future.
00:33:35.03 So here's today and what mice have done is evolved into the future
00:33:43.02 to the extent that if we were to run the human line we would only catch up in 125 million years.
00:33:52.28 And I think the mouse genome is a pretty good genome and so we can quite safely say,
00:34:01.00 so far as the genome is concerned, we could guarantee this for 100 million years.
00:34:07.05 And that emerges strictly by knowing the rate of evolution.
00:34:12.12 And we can do the same for the cow
00:34:15.20 and you'll find that the cow is about 30 million years into the future, it's evolving quite slowly.
00:34:28.00 And we can do the same for all the animals.
00:34:31.18 Now, of course this gives you a high resolution because if I'd have taken my numbers,
00:34:39.29 (you'll remember I think they said the cow was this and the mouse was this).
00:34:47.03 It actually turns out that for this flip rate calculation,
00:34:53.23 I would have deduced that the mouse is a more ancient point of division than the cow,
00:35:02.19 if I had assumed everything was going at the same rate.
00:35:06.15 In fact it's not true because,
00:35:10.12 as we've seen now, the cow division point is older than the separation from the mouse.
00:35:20.07 Now, with this approach you can do absolute phylogeny
00:35:27.20 with quite high resolution certainly for mammals.
00:35:31.28 You need a lot of data but the data is now available.
00:35:37.14 We can do the comparison with let us say marsupials which are very far back.
00:35:45.28 And we can work out their rate of evolution in much the same way
00:35:52.08 by doing the comparison with further and further back.
00:35:56.06 The more distant you go back in time, the more difficult it is to actually make the measurements
00:36:07.15 because you are getting closer and closer to the point of equilibrium.
00:36:13.03 And where you have a very high resolution is in this region of 0.8 to 1.0.
00:36:23.24 One is identity.
00:36:25.14 Oh, by the way, on this scale, the chimpanzee,
00:36:29.09 which actually is evolving at the same speed as ourselves,
00:36:33.02 so is the macaque monkey,
00:36:35.08 but the chimpanzee on this scale, this division is very close up here because the genomes are so close.
00:36:45.24 So we can actually, by the same token, measure in a set of primates
00:36:53.05 we can get the absolute point of departure
00:36:56.27 because we can measure the rate of evolution in these organisms.
00:37:02.16 And I've done it for the macaque, the chimpanzee, and of course ourselves.
00:37:12.01 Well, this absolute phylogeny says that you can actually reconstruct the past
00:37:21.01 simply from the study of the genome and simply by looking at these changes.
00:37:29.14 Now, where I'd like to go on is to tell you about another problem
00:37:37.13 which seems completely unable to be resolved unless you take a
00:37:49.11 very clear cut attitude towards unbarring all the assumptions that lie behind what people say.
00:38:00.24 We'll see here that basically by saying we can't just simply assign all the changes to this,
00:38:08.25 we must measure the rate.
00:38:10.17 Once we've got the rate, then we can actually make a measure of distance
00:38:17.24 and once we've fixed one time, let us say mouse-man separation,
00:38:25.27 we can then put in all the other animals that's we've compared on the absolute time of separation.
00:38:34.00 And so you can construct a tree which is one of absolute phylogeny.
00:38:41.04 which I think is a terribly important thing to do.
00:38:45.26 So, by doing that, we can scale at least all the mammals
00:38:52.12 together with the marsupials onto one scale and we can study them.

Part 2: Genomes Tell Us About the Past cont’d

Videos in this Talk

Part 1: Genomes Tell Us About the Past
Part 2: Genomes Tell Us About the Past cont’d

Speaker: Sydney Brenner

Total Duration: 1:21:40

Recorded: April 2008

All Talks in Archive

Talk Overview

By looking at the light from distant galaxies and having well-established calibration methods, astrophysics can make hypotheses about the history of our universe. Do we have similar “rulers” in biology that could allow us to reconstruct the remote past and the evolution of species on this planet? The answer is likely “yes” and the clues are undoubtedly contained in the many whole genome sequences that are now available for inspection. However, it is critical to evaluate the assumptions that one makes in analyzing such sequence data. The first part of the talk discusses how one might reconstruct an accurate phylogeny between species based upon examining neutral mutations (particularly the degenerate third base in the triplet codon of certain amino acids). Critical to the approach is the realization that different species (e.g. mice and men) are evolving at significantly different rates.

Speaker Bio

Sydney Brenner

Sydney Brenner is currently a Distinguished Professor of the Salk Institute for Biological Studies in La Jolla, USA. Dr. Brenner received degrees in Medicine and Science from the University of Witwatersrand, South Africa and a D. Phil. in Chemistry from Oxford University, England. Dr. Brenner was a member of the Medical Research Council Laboratory of… Continue Reading