Finding Genes that Control Development
Transcript of Part 1: Finding Genes that Control Development
00:00:16.10 You know, a lot of times, when you--as a scientist--think back to experiments that you've done 00:00:22.29 even really successful experiments, you realize that at the time that you were doing them, 00:00:31.00 you didn't really fully understand the science involved or the science behind them 00:00:35.21 or the science which was going to affect whether you were going to get results or not, 00:00:39.06 and you probably also didn't really appreciate how lucky you were. 00:00:44.24 Today, we're going to talk a little bit about one experiment--a wonderful experiment-- 00:00:48.22 that I did many years ago with Christianne Nüsslein-Volhard 00:00:59.04 at a time when we were both starting out our scientific careers. 00:01:02.25 We, like many people at the time (this was the late 1970s) 00:01:06.16 were interested in the problem of embryonic development, 00:01:10.18 but at a very basic level. 00:01:11.26 We knew that genes controlled things that you could see happening in the embryos. 00:01:16.00 We wanted to identify what those genes were and what they did. 00:01:20.18 And there were a lot of different molecular strategies--conceptual strategies-- 00:01:24.17 that you could use to do this. 00:01:25.18 We decided that we would try a genetic approach and what that meant for us 00:01:30.28 was that to build on our lack of knowledge and to just randomly mutagenize all genes-- 00:01:38.24 knock them all out and see what happens, 00:01:41.13 and get some estimate at least or some global picture of what it was the genes were doing during development, 00:01:51.18 how gene activities were organized in patterns and in sequences 00:01:56.19 to produce these extraordinary phenomena that you see in development. 00:02:02.00 We knew that there were lots of genes, 00:02:03.16 and we knew that understanding the process would probably require that we identify most of them, 00:02:09.23 so we had set up in our minds at least this goal that we wanted to identify everything. 00:02:17.16 We wanted to identify all the genes, and this meant a really big experiment. 00:02:21.19 And for those of you who have any sense or experience working with Drosophila, 00:02:26.00 what a big experiment looking for lots of genes means is lots of tubes, 00:02:31.24 growing lots of flies, and setting up crosses. 00:02:35.20 Basically, our plan was to feed flies mutagens, establish inbred lines from single males 00:02:44.08 in the first generation and then carry it through an appropriate number of generations 00:02:48.12 to produce inbred heterozygous lines that we could collect homozygous embryos from. 00:02:56.06 And we thought about it, and our estimates of what we needed to do was something like 00:03:01.04 in the range of maybe 40,000 lines established through many generations. 00:03:07.14 And for us to do that, we realized 00:03:10.13 (and this is actually the first real secret and the unappreciated aspect of this experiment 00:03:16.06 is that we were starting our jobs in Heidelberg) 00:03:19.14 and we knew that we had to set up lots of tubes, and we would identify all these genes 00:03:26.03 and if it was going to be effective, we would have to have ways of almost a mechanized inbreeding of flies. 00:03:34.23 We couldn't sit and set up crosses manually by hand. 00:03:37.25 We would have to have selective procedures that allowed us 00:03:40.19 to avoid collecting individual flies and set up crosses. 00:03:48.06 The first two years of our stay in Heidelberg was involved in this. 00:03:54.22 Setting up and designing genetic crosses that allowed us simply to put a fly in a tube. 00:04:02.07 Wait two weeks, there would be lots of flies... Put two flies in a tube, actually. 00:04:06.28 Wait a couple of weeks for the next generation, 00:04:11.20 and by raising it at a different temperature, or with some other selective technique, 00:04:16.00 we could just shake the surviving flies over and go through and gradually inbreed the lines 00:04:21.20 such that, at the end of the cross down here at the bottom, after three generations, 00:04:25.03 we had balanced, inbred lines so that we could collect homozygous embryos. 00:04:28.12 Now, it took us two years to set that up and to get crossing schemes that would actually work. 00:04:35.07 We set out to do the experiment, and the experiment actually, itself only took us... 00:04:38.15 well, two rounds of experiments, each of which were less than two months in advance. 00:04:43.29 So, I guess the first lesson is the idea that if you have a vision and the vision's clear enough 00:04:49.12 you may still have to spend a lot of work just getting the assays and getting the procedures right. 00:04:59.04 But then the next thing, and this is the thing that I really wanted to talk about 00:05:01.25 We did actually start out with trying to make 40,000 lines. 00:05:10.18 As you'll see, we only really carried about 27,000 of them through to the third generation 00:05:15.29 But, by the time we got to that generation and began to get our first results... 00:05:21.18 We'd gotten our first important result. 00:05:22.01 We'd used enough mutagenesis that, in the lines we were looking at, 00:05:29.25 we had about, in these 27,000 lines, we had about 18,000 different mutations. 00:05:35.29 So that meant that our mutagenesis procedures worked--everything was working fine 00:05:41.18 Our interest, though...these mutations kill the fly at any time during its life cycle, 00:05:46.20 and what we were really interested in were genes that controlled the way embryos developed. 00:05:52.09 And, this is the point, actually, where I think we were extraordinarily lucky. 00:05:56.28 Because, what we had to do was now, from all these 18,000 different lines 00:06:02.21 we had to collect mutant embryos, look at the embryos that die, 00:06:08.10 and decide whether the gene that was mutated in those flies 00:06:11.03 actually killed the embryos and killed them in an interesting way. 00:06:14.11 And the great result--the thing we couldn't have anticipated 00:06:18.01 was that, of the 18,000 different lethal mutations, only a tiny, tiny fraction 00:06:24.26 actually changed homozygous embryos in a way that was really meaningful. 00:06:29.00 Most of the time, the homozygous mutant embryos hatched and crawled away. 00:06:35.26 It didn't have to be that way, and what would have been really, really horrible for us, in thinking back to it 00:06:42.00 is if every one of those 18,000 lines, we picked up and looked at embryos and there was some phenotype... 00:06:47.15 something had gone wrong. 00:06:49.07 We would never have actually been able to, I think, since there was really only the two of us doing these experiments 00:06:54.20 to sort things out and come up with a coherent picture. 00:06:58.01 So, we were really lucky that what had actually happened was that, in a way that we didn't fully anticipate, 00:07:04.15 the screen was very selective... 00:07:07.15 selective for mutations that caused phenotypes in homozygous embryos. 00:07:12.04 That is to say, we were selecting for genes that had to, themselves, be transcriptionally active in embryos, 00:07:18.13 and that, in flies, that number is very small. 00:07:23.09 And so, the number of mutations was about 586, 00:07:27.13 and actually the number of genes, if you do complementation tests among those mutations, 00:07:34.04 there are only about... we only found 139 different genes. 00:07:38.02 So that's a really small number that reduced the problem 00:07:42.13 of thinking about development down to 139 components--139 genes, 00:07:48.07 each of which were producing different phenotypes 00:07:50.21 and could be arranged in pathways and sequences that made, for me at least, 00:07:56.01 the whole process of development almost understandable or structural for the first time. 00:08:02.15 That understanding came from the fact that the numbers are small. 00:08:07.24 And the numbers are small, not because we planned them that way. 00:08:11.15 The numbers are small because of the peculiar biological features... the way flies develop. 00:08:17.03 It turns out that flies' embryonic development is very rapid, the mothers puts everything the embryo needs... 00:08:22.01 [everything] that she can possibly put into the egg is put in by the mother, 00:08:26.27 and the transcriptional requirements early during these patterning stages are very minimal. 00:08:33.13 So, the fly has stripped down its transcriptional requirements to the rare genes that can't be supplied 00:08:40.28 by the mother--that must be supplied by transcription in this cell and not in this cell. 00:08:46.26 There's genes...the embryo is using transcription to control patterns, to control transitions in time. 00:08:53.19 So, we did a genetic experiment that was oriented towards identifying phenotypes in homozygous embryos 00:09:01.03 and because of the biology of flies--the way they work-- 00:09:05.00 we ended up identifying the central set of core regulatory genes 00:09:12.18 whose activity temporally and spatially control the development of pattern. 00:09:16.25 And if I think back to what both Janni [Nüsslein-Volhard] and I, at the time... 00:09:22.11 We thought a lot about this experiment. 00:09:25.10 As I said, we planned it for more than two years, we actually spent two years trying to get the thing set up, 00:09:31.03 and yet we didn't really fully understand how the system worked-- 00:09:36.22 how the fly embryos developed until we actually had the numbers 00:09:40.15 and had the results. 00:09:41.09 And I think that's actually a great lesson in biology... 00:09:44.05 On the one hand, you don't understand the experiment really that you're doing 00:09:50.19 until you get the result. 00:09:51.29 And it's really only when you get the result that you potentially understand 00:09:58.07 what biology and the whole system that you work with is actually giving you. 00:10:02.25 That's true, and I guess it's pretty much the nature of science. 00:10:13.07 It's really great to do experiments where you don't know the outcome.