RNA Processing, Alternative Splicing, and the Spliceosome

Part 1: Split Genes and RNA Splicing

00:00:00.28 Hi. My name is Melissa Moore
00:00:02.22 and I'm from the Howard Hughes Medical Institute
00:00:05.04 and the University of Massachusetts Medical School
00:00:07.14 and I'm here today to tell you about split genes
00:00:09.24 and RNA splicing
00:00:12.24 Most of you are familiar with the central dogma of biology
00:00:15.21 which was elaborated by Sir Francis Crick in 1956
00:00:21.03 and what the central dogma says is that DNA is copied into
00:00:26.29 or transcribed into RNA and then that RNA is translated
00:00:33.18 into a different language and that is the language of proteins.
00:00:36.15 So one of the central tenets of the central dogma of biology
00:00:40.16 is that one gene encodes one protein.
00:00:44.23 and this is very much true in bacteria where if one
00:00:49.03 breaks open a bacterial cell and spreads out the DNA and RNA
00:00:54.11 you can see very clearly that here is the DNA
00:00:58.00 and as the DNA is being transcribed into RNA, it's
00:01:02.09 being joined by ribosomes. So these big black blobs here
00:01:07.03 are polyribosomes and those are making protein.
00:01:10.03 And so the RNA is copied or translated directly into proteins.
00:01:15.29 So here I'm showing a eukaryotic cell next to a prokaryotic cell.
00:01:21.20 So these are bacteria next to a white blood cell.
00:01:24.28 And you can see that the white blood cell is much
00:01:27.22 much larger than the bacteria. And so not only
00:01:31.14 are eukaryotic cells larger than bacteria but they also
00:01:36.01 are more complicated internally.
00:01:37.27 If this is the bacterial structure down here and this would be
00:01:43.19 the eukaryotic internal structure and you can see
00:01:46.19 that in the eukaryote, the DNA is in the nucleus
00:01:51.24 whereas the ribosomes are out here in the cytoplasm
00:01:54.21 as you can particularly see these on the rough ER here.
00:01:57.28 And so the big difference for a eukaryotic cell is that
00:02:03.04 the DNA is not directly accessed by the ribosomes.
00:02:08.09 Also in the eukaryotic cell, this is a gene that is being transcribed
00:02:14.01 from this direction to this direction. And so you can see
00:02:17.28 that the RNA coming off of the DNA is getting longer and longer
00:02:21.26 as it's being transcribed in this direction. But at some point
00:02:25.08 there are these loops that form. And in fact, these loops
00:02:29.26 are introns that get spliced out. So unlike prokaryotic genes
00:02:37.26 eukaryotic genes are split in nature. They have segments
00:02:42.20 of them that need to be spliced out before they can
00:02:45.14 be used to make proteins. So this slide shows the
00:02:49.00 current-day view of eukaryotic gene expression.
00:02:52.08 And that is that eukaryotic genes are split, meaning that
00:02:57.09 they contain sequences that are not contained within
00:03:00.05 the final mRNA, and which are not translated into
00:03:03.08 protein sequence. Those sequences are called introns
00:03:07.06 and they are represented by the white line here, and
00:03:11.09 the exons are the little colored boxes. Those are the
00:03:14.18 expressed regions. When a gene is turned on and
00:03:18.19 transcribed, the entire gene is transcribed first into
00:03:23.26 a pre-messenger RNA or pre-mRNA and that pre-mRNA
00:03:27.19 undergoes several steps of processing.
00:03:31.20 First it is capped with a seven methylguanosine cap.
00:03:37.01 At the 3' end it is cleaved and then a poly-A tail is added.
00:03:43.07 And then in the middle, these intron sequences are literally
00:03:46.28 spliced out. So there is a molecular scissors and tape
00:03:51.05 that does the job of pre-mRNA splicing. After all of
00:03:56.15 processing is done, the mRNA migrates to the nuclear
00:04:00.19 envelope where it is exported and used to be translated
00:04:05.19 into proteins. This shows the structure of a typical
00:04:10.15 human gene. So the typical human gene has 23,000
00:04:16.01 base pairs and 7 introns. Now I'm using a median here
00:04:21.04 because if I use the average, it would be very much
00:04:24.00 thrown off by some of the very large genes that
00:04:27.23 we're going to talk about in a little bit. The other thing
00:04:31.20 that you'll notice is that the typical human gene
00:04:36.23 has a median intron length of over 10 times that of
00:04:41.07 the exon lengths. So that means that whenever a
00:04:45.16 human gene is transcribed, 90 to 95% of the RNA
00:04:50.21 is immediately spliced out and thrown away. And that
00:04:54.28 seems rather wasteful. So we'll be talking about
00:04:58.05 why it is that we have these intronic regions.
00:05:04.12 What good are they if we're wasting that much RNA.
00:05:07.28 But before we talk about that, I want to tell you about
00:05:12.09 a particular gargantuan gene and that is the dystrophin gene.
00:05:17.11 So dystrophin encodes a protein that is necessary
00:05:22.14 for your muscles and mutations in this gene are one
00:05:26.26 of the causes of muscular dystrophy. The DMD gene
00:05:30.19 is the second largest gene in the human genome. It's
00:05:34.17 2.2 million base pairs long. It has 79 exons, 78 introns.
00:05:42.15 and one of those introns is 400,000 nucleotides long.
00:05:48.12 Now it's really hard for you with just me saying this
00:05:53.01 to really get a sense of the scale of this thing.
00:05:56.16 So next I'm going to show you a little movie so that
00:05:59.08 you can really see just how big this gene is.
00:06:05.26 Here we are with our scaled model of the dystrophin gene
00:06:08.22 that is represented by this rope. And I want you to
00:06:12.11 look at the end of this rope. So the little colored tape
00:06:17.19 marks here are the exons. And the white rope is the
00:06:21.26 intron. And you can see that here is the next exon.
00:06:25.06 So the introns are much, much longer than the exons.
00:06:28.22 Now, just to give you a sense of how big this gene is
00:06:32.27 I'm going to pretend I am RNA polymerase. And I
00:06:37.03 start transcribing this gene when you get up in the morning.
00:06:40.14 So this is your time point of breakfast here. So here I go.
00:06:46.22 Here's RNA polymerase. I'm transcribing this gene.
00:06:50.03 It's not evening. It's still morning. Here's our mid-morning snack.
00:06:55.21 So we're still going. Polymerase is incredibly processive on this gene.
00:07:00.25 We haven't even gone a million base pairs yet.
00:07:04.10 Now we're about halfway through the gene so this is
00:07:08.28 lunchtime. The polymerase is still transcribing this gene.
00:07:13.07 We're still going, still going. Here we are at mid-afternoon.
00:07:20.05 We're getting into dinnertime now. And we've been
00:07:24.08 going for about thirteen or fourteen hours now. Finally
00:07:28.28 when you're about ready to go to bed after 16 hours
00:07:32.18 we get to the very end of the gene, the last exon.
00:07:36.14 It took 16 hours for polymerase to transcribe this
00:07:41.20 entire gene of dystrophin.
00:07:59.11 Alright, so you've just seen how long dystrophin
00:08:03.26 RNA is. And I can tell you that was a 99.4-foot rope.
00:08:09.08 Now once all of those introns are removed, to scale
00:08:14.08 this is the size of the messenger RNA. This messenger
00:08:18.22 RNA is very long. It is a 17,000 base messenger RNA.
00:08:23.07 But it is less than 1% of the original RNA that was transcribed.
00:08:29.29 You know, really pretty amazing.
00:08:31.29 So now we want to talk about the question of why
00:08:36.24 do we have all of these introns. It's really quite
00:08:40.22 amazing that we have so much apparent junk DNA
00:08:45.00 that is copied into RNA and then immediately thrown away.
00:08:49.05 What good are these introns? One thing that introns
00:08:54.16 do for us, is to allow eukaryotic cells to evolve new
00:09:00.06 genes readily. And they can evolve new genes by
00:09:03.20 exon duplication. So let's take for example,
00:09:06.25 the fibronectin protein. You can see here, the fibronectin
00:09:09.14 protein is made up of many repeats of different domains.
00:09:13.25 These domains are also found in other proteins.
00:09:16.21 So for example, the yellow domain is found in cell
00:09:21.18 surface receptors and other extracellular matrix proteins.
00:09:25.03 The blue domain is found in blood coagulation proteins
00:09:30.03 and the red domain is also found in TPA, or tissue plasminogen activator.
00:09:35.01 So the way the fibronectin was created was to take
00:09:40.01 all these different domains and hook them up and
00:09:43.08 bring them in from other places. And the way that
00:09:46.18 that can be done is that you see if you look now at
00:09:49.14 this diagram of the fibronectin gene, you can see that
00:09:53.02 each of these domains consists of either one or two exons.
00:09:58.07 So the advantage of having introns in this case is that
00:10:02.21 by non-homologous recombination big chunks of the
00:10:06.29 genome can be taken from one place to another
00:10:10.04 and they don't necessarily have to be hooked up perfectly
00:10:13.03 because the splicing machinery hooks them up.
00:10:15.17 whereas in bacteria, where there are no introns,
00:10:18.24 if there is a non-homologous recombination event to
00:10:22.01 make a new protein, then it has to be perfectly in frame
00:10:25.07 and it is much harder to do because there's not the
00:10:29.01 genetic space in which to do the recombination.
00:10:33.10 So that is a one real advantage of having introns.
00:10:38.13 Now the second major advantage we will discuss by
00:10:44.12 looking at the gene number versus complexity problem.
00:10:48.01 So let's consider how many genes different kinds of
00:10:51.29 organisms have. So here for example, is E. coli, a bacterium,
00:10:57.17 and S. cerevisiae--budding yeast, or the yeast that is
00:11:00.24 used to make bread or brew beer. E. coli is a bacterium
00:11:07.01 in your gut. Now we've already shown, that bacteria
00:11:10.17 are much simpler cells than eukaryotes, so you might
00:11:13.22 expect that E. coli not to have as many genes as S. cerevisiae
00:11:18.22 and that's true. So E. coli has about 3,200 genes and
00:11:24.00 S. cerevisiae has about 6,000 genes. And by genes
00:11:27.24 here, I'm talking about protein-coding genes.
00:11:30.17 Now if we go up the evolutionary ladder, let's consider
00:11:34.05 two other model organisms. C. elegans or the roundworm
00:11:39.15 and Drosophila melanogaster, the fruit fly. Now on the
00:11:44.20 surface, it might seem that Drosophila is the more
00:11:47.18 complicated organism, so it would have more genes
00:11:50.13 than the roundworm, but in fact, it's the opposite way.
00:11:54.21 So C. elegans has about 19,000 genes and
00:11:58.17 Drosophila melanogaster only has about 13,000 genes.
00:12:02.14 So now you have to ask yourself, "How much more
00:12:05.23 complicated am I than a fruit fly or a roundworm?"
00:12:10.00 So how many genes do we expect humans to have?
00:12:13.16 Or humans, mice, and even a mustard plant, so
00:12:19.20 something that might be a little more complicated
00:12:21.23 than a roundworm. So, if a roundworm has 19,000 genes
00:12:27.15 before the human genome was sequenced, people
00:12:30.26 were thinking that humans would have around
00:12:33.03 100,000 genes. The big surprise has come that in fact
00:12:37.10 all of these organisms have about the same number
00:12:39.27 of genes and they're all around 25,000. So it's really
00:12:44.21 not a very big number. And just by this count,
00:12:49.03 we're no more complicated than twice a fruit fly or
00:12:52.25 about 1.3 times a roundworm. That's sort of troubling.
00:12:57.26 So how can we get away with that? Well the way we
00:13:00.08 get away with this apparent lack of complexity is
00:13:05.08 alternative splicing. So eukaryotes in addition to
00:13:10.24 having split genes, they defy the central dogma of
00:13:18.24 biology, the original central dogma of biology. And
00:13:21.19 that is that we may have one gene, but one gene can
00:13:26.13 make many, many proteins. And so for example, here
00:13:30.08 is one gene, where if all of the exons are spliced in,
00:13:35.00 then it makes one protein, but if the red exon is left
00:13:40.01 out, it makes a different protein. And then here, the blue
00:13:43.21 exon, exon 4, is left out and that makes yet a third protein.
00:13:48.26 So on this particular gene, there are three proteins
00:13:51.28 coming from one gene. And we now know from recent
00:13:55.26 deep-sequencing efforts that about 95% of all human
00:14:00.15 genes exhibit alternative splicing. So that means that
00:14:03.27 our proteome, the number of proteins that we have,
00:14:07.14 is much, much larger than our genome.
00:14:11.03 There are many different kinds of alternative splicing.
00:14:15.26 So, there can be alternative promoters, which is the
00:14:21.02 beginning of the gene. There can be alternative poly-A sites.
00:14:25.00 So that's the 3 prime end of the gene. Alternative 5 prime
00:14:28.26 splice sites. We'll talk about what a 5 prime splice site
00:14:30.29 is in a minute, but it's the beginning of an intron.
00:14:33.12 Alternative 3 prime splice sites. And then there are exons
00:14:37.14 called cassette exons that are either spliced in or
00:14:41.01 spliced out. That's the example I was showing you a
00:14:43.18 minute ago. And there are mutually exclusive exons,
00:14:47.07 where either one exon is put in or the other but not both.
00:14:50.15 And then of course the simplest form of alternative
00:14:53.00 splicing is to not splice at all. So you can splice or not
00:14:55.21 splice and that would be a retained intron. So there
00:14:58.27 are many different types of alternative splicing. The
00:15:01.06 most common type of alternative splicing is the
00:15:03.17 cassette exon, where an exon is either spliced in or spliced out.
00:15:09.02 Now how many different proteins can be made from
00:15:12.18 one gene? So this is the alpha tropomyosin gene from rat
00:15:18.03 and these are some of the different splice forms of
00:15:22.00 the alpha tropomyosin gene. And so you can see
00:15:25.05 that there are many different splice forms in fibroblasts.
00:15:31.10 Those are essentially undifferentiated cells.
00:15:35.07 Then here's other isoforms in the brain and the smooth muscle.
00:15:38.22 And so one of the important things about splicing
00:15:42.16 is that it can be developmentally and tissue-specifically controlled.
00:15:45.21 And so one gene in one tissue might make one protein
00:15:50.20 but in another tissue, it makes a very different protein.
00:15:54.10 And so, again that's how we add complexity.
00:15:58.05 So just how complex can it get?
00:16:01.13 Well, here's the current record-holder. So this is the
00:16:04.09 Drosophila DSCAM gene, which is involved in axonal
00:16:08.04 guidance in the brain. And Drosophila DSCAM has
00:16:12.15 three regions of mutually exclusive exons. So there's one here,
00:16:18.03 one here, one here, and one there. This region has 48,
00:16:22.23 33. There's two over there, and there's 12 back here.
00:16:27.07 So if you do the math, there are over 38,000
00:16:32.28 different possible spliced isoforms of the DSCAM gene.
00:16:36.25 And to the best of our knowledge, all of these isoforms
00:16:39.12 can be made. So that means that this one gene in Drosophila
00:16:44.14 can make three times as many different proteins
00:16:47.24 as there are genes in the Drosophila genome.
00:16:51.07 So it is very likely that in higher eukaryotes, such as you and me,
00:16:55.19 our proteome is well over hundreds of thousands to
00:17:02.10 millions of different proteins. So just with that thought in mind
00:17:06.14 Now let's look back at our complexity problem.
00:17:09.03 But this time, instead of looking at genes, let's look
00:17:12.16 at how many introns each organism has and then how that
00:17:16.14 scales with complexity. So in E. coli, it has no introns.
00:17:22.12 Prokaryotes do not have these types of introns.
00:17:25.19 S. cerevisiae, or budding yeast, has a few introns.
00:17:28.29 It only has about 250 introns and as we go up the
00:17:32.15 evolutionary ladder, the roundworm has about 99,000 introns.
00:17:38.15 Again, more than the fruit fly simply because it has more genes.
00:17:41.24 But as we go up now to humans and to mice, you can see
00:17:47.05 that the number of introns is going up dramatically.
00:17:51.14 And because the number of proteins you can make
00:17:55.03 scales let's say exponentially with the number of introns
00:17:59.28 you can imagine that our proteomes can be
00:18:03.17 much more complicated than those of the other organisms.
00:18:06.15 Now we want to talk about the problem of how does
00:18:12.28 the cell know where the introns are. In other words,
00:18:18.09 how does it know where to splice? And in this sense,
00:18:21.23 a cell faces very much the same problem as a film director,
00:18:27.10 who has shot thousands of hours of film and yet is
00:18:32.16 now overwhelmed with film and needs to then make
00:18:37.16 a final movie. And so traditionally with film before the
00:18:41.22 digital age, the film director would go through the film
00:18:46.13 frame by frame, find exactly the place where he or she
00:18:51.11 wanted to make a cut and then there would be a
00:18:54.15 splicing machine that would literally cut and splice the
00:18:59.09 film back together to make the final version of the movie.
00:19:03.08 And in cells, that's exactly what happens. Because of
00:19:07.22 course introns are made up of individual nucleotides
00:19:10.23 and the nucleotides are very much like the individual frames
00:19:15.00 in a film. So introns have the following structure.
00:19:21.21 The 5 prime end, or the beginning of the intron, is
00:19:25.02 many times called a spliced donor but splicers tend to
00:19:28.17 call it the 5 prime splice site. And it, in 99% of genes,
00:19:33.26 is signified by a GT as the first two nucleotides of the intron.
00:19:40.05 And then much further downstream, actually close to
00:19:44.23 the other end of the intron is an adenosine called the
00:19:49.01 branch site, and we'll be talking about what that
00:19:52.04 branch site does in part II of my lecture. And then
00:19:55.28 at the very end of the intron is the AG, which is the
00:20:00.15 last two nucleotides of the intron at the 3 prime splice site
00:20:03.24 or the "splice acceptor." Now I've told you that some
00:20:07.23 introns are up to 400,000 nucleotides long and certainly
00:20:12.18 this amount of information content is not enough
00:20:15.18 to even signify a small intron. So there must be
00:20:20.08 additional sequences. And so in addition to these
00:20:24.24 universally conserved nucleotides, there are consensus
00:20:28.09 sequences in all of the introns. So here are the
00:20:31.23 consensus sequences for budding yeast and these
00:20:34.12 are called logos, motif logos, and the height of the letter
00:20:43.03 tells you how conserved that site is. So for example,
00:20:46.15 here you can see the GT at the 5 prime splice site
00:20:50.06 and so in addition to the GT, there are several other
00:20:53.07 nucleotides that are highly conserved and a few that
00:20:56.21 are less conserved. At the branch site in yeast, there
00:21:00.07 is a very highly conserved TACTAAC sequence
00:21:04.12 and then upstream of the 3 prime splice site there's
00:21:07.07 not too much conservation until you get right to the
00:21:09.14 to the 3 prime splice site. So there's additional information
00:21:12.03 in these so-called consensus sequences. Surprisingly
00:21:16.28 in humans--so if you remember yeast only have about
00:21:19.10 250 introns and they generally are short--they're all
00:21:23.15 less than a couple hundred of nucleotides whereas
00:21:26.17 human introns are much, much longer. And the
00:21:29.18 surprising thing is that the human consensus sequences
00:21:33.00 are less conserved. So you can already see that
00:21:35.22 because you can just see that there are lots more letters
00:21:38.18 down here than there are up here particularly for the
00:21:41.06 branch site much less conservation. And then for the
00:21:44.10 3 prime splice site, there's a little more information
00:21:46.08 because there's this so-called pyrimidine track here
00:21:49.06 that is a bunch of C's and T's upstream of the AG.
00:21:54.11 Now we can quantify the amount of information that's
00:21:58.29 in these sequences by using the terminology that's
00:22:03.07 used in computer science and that is bits of information.
00:22:06.00 So how much information content do these things have.
00:22:08.24 And so Chris Burge and coworkers did a study on
00:22:14.24 a number of very short introns, the shortest introns
00:22:18.09 in humans and also the introns in yeast. And they
00:22:23.03 derived these consensus sequences, and these consensus sequences
00:22:26.27 have this much information in them. So you can see
00:22:29.24 already just by looking at the numbers of bits that
00:22:33.10 the human introns have less information in their consensus sequences
00:22:39.01 than the yeast introns, even though the human introns--
00:22:41.00 there are many more of them and they're much longer.
00:22:43.20 Now if we add up all of this information, the yeast intron
00:22:48.24 sequences clearly have more information in their consensus sequences
00:22:53.03 than the humans but I can tell you what Chris Burge
00:22:57.16 and his colleagues estimated in this paper was that
00:23:01.02 for humans, in order to uniquely specify an intron,
00:23:05.21 even of this short size that they were looking at
00:23:08.26 so below a thousand nucleotides, you need about
00:23:13.05 37 bits of information and yet only 23 bits of information
00:23:17.13 are contained in the splice site consensus sequences.
00:23:20.16 So where is that other information coming from?
00:23:23.07 So from the same paper, here is a graph showing
00:23:27.23 the relative contributions of intron features to
00:23:31.02 intron recognition in different organisms and these are
00:23:34.02 the same organisms that we looked at before in terms of
00:23:37.19 gene count or intron count and complexity.
00:23:40.09 And what you can see is that in S. cerevisiae all of the
00:23:43.23 information is coming from the 5 prime splice site,
00:23:47.02 the 3 prime splice site and the branch site consensus sequences
00:23:50.07 but once we get up to humans, only about half of
00:23:54.14 the information needed to specify an intron in humans
00:24:02.00 is coming from those three consensus sequences.
00:24:04.24 Some of the other information is the length. The composition
00:24:09.08 of introns. They're somewhat different in their base composition
00:24:14.04 than exons. But then there's this big grey quadrant
00:24:20.28 and one has to ask, "What is this other information that
00:24:25.22 human introns require?" And a clue to that came very early
00:24:32.06 soon after introns were discovered. So this was a paper
00:24:36.06 in 1983. Introns were discovered in 1977.
00:24:40.29 And the Maniatis and Orkin labs were mapping out
00:24:46.13 mutations in the beta-globin gene. So mutations
00:24:50.27 in beta-globin cause a disease called beta-thallasemia.
00:24:54.03 It's a very debilitating blood disease because you
00:24:57.12 do not make enough hemoglobin. And what they found was
00:25:01.25 that many of the mutations that cause beta-thallasemia
00:25:06.19 were mutations in the 5 prime splice site consensus sequence
00:25:10.26 of the globin gene. But instead of completely killing
00:25:17.17 the splicing of the beta-globin gene, what these mutations did
00:25:22.23 was they activated cryptic 5 prime splice sites in the
00:25:26.24 nearby area. So that means something was telling the
00:25:31.21 splicing machinery that there's supposed to be a
00:25:35.01 5 prime splice site in this region and even though this consensus
00:25:39.15 sequence was mutated, there were other sequences
00:25:43.00 nearby that could serve as a viable 5 prime splice site.
00:25:46.17 So what is that other information?
00:25:49.18 So to think about what that information might be
00:25:53.07 let's take a look back at that model of the dystrophin gene.
00:25:56.16 that we looked at. So if you'll remember, here's my long
00:25:59.13 99-foot rope and when your eyes are looking at this rope,
00:26:03.14 probably what they're drawn to are the exons, the
00:26:09.07 little pieces of color, and not the rope itself.
00:26:12.18 And so you can actually think about a human gene
00:26:15.16 as islands of exons in the middle of oceans of introns.
00:26:20.26 And so very much like a map of the Pacific Ocean
00:26:25.11 if you're going to identify the islands, you're going to
00:26:29.14 look at the islands, not at the ends of the oceans.
00:26:32.19 You're going to look at where the islands begin and end
00:26:35.09 not where the ocean begins and ends. So let's think
00:26:39.01 about how that might be applied to a human gene.
00:26:42.16 What we now understand is that the way that
00:26:47.06 introns and exons are recognized in human genes
00:26:50.11 is through what we call exon definition. And that is
00:26:55.12 that even though the introns are of highly variable length--
00:26:58.24 going from about a few hundred nucleotides going up to
00:27:02.03 tens of thousands, hundreds of thousands nucleotides--
00:27:05.01 the exons tend to be very much uniform in their length.
00:27:09.21 The average exon as we said was 123 nucleotides
00:27:13.09 but they vary in length from about 25 (there are some
00:27:17.05 few shorter. There are exceptions to every rule.) but
00:27:20.15 in general 25-300 nucleotides is the size of an internal
00:27:23.26 exon in a human gene. And so what the human splicing machinery
00:27:29.10 looks for is a good match to the consensus branch site
00:27:34.07 and 3 prime splice site followed with 25-300 nucleotides
00:27:41.06 by a match to the 5 prime splice site. And so this
00:27:44.11 then defines an exon. But even with that distance information
00:27:48.24 between a 3 prime splice site and a downstream 5 prime splice site
00:27:53.02 that's still not enough information to uniquely identify
00:27:57.09 an exon and it's also not enough information to allow
00:28:00.18 for alternative splicing of that exon. So in addition
00:28:04.01 to the consensus sequences, there are also other sequences
00:28:08.13 called exonic splicing enhancers or splicing silencers.
00:28:12.21 So these enhancers or silencers can be either in the
00:28:19.14 exon or they can be in introns. And if they're in introns
00:28:24.25 they're called intronic splicing enhancers or intronic splicing silencers.
00:28:29.00 So that would be the ESE, ISE, ESS, and ISS sequences.
00:28:35.08 Now these sequences tend to be clustered around exons.
00:28:39.17 and they are recognized by two different kinds of proteins.
00:28:47.14 in general. And these proteins are called the SR and the hnRNP proteins.
00:28:53.25 Now SR proteins are RNA-binding proteins that have
00:28:59.09 a C-terminal domain that is rich in arginine/serine dipeptides.
00:29:04.03 And hnRNP proteins are another set of RNA-binding proteins
00:29:09.17 They do not have this RS region. But in general, the SR
00:29:13.28 proteins tend to recognize splicing enhancers. They tend
00:29:17.26 to increase splicing and hnRNP proteins tend to
00:29:23.00 recognize splicing silencers and so they tend to inhibit splicing.
00:29:27.12 And it's the balance of the SR and hnRNP proteins
00:29:32.02 that are in a cell that determines the splice pattern
00:29:36.18 for any particular RNA. So you can think very much about
00:29:40.20 splicing decisions being made as a committee decision
00:29:45.25 so in addition to the SR proteins and the hnRNP proteins
00:29:49.22 there's the core splicing machinery and we're going to talk
00:29:52.05 about that in part II of my talk. But the core splicing
00:29:55.15 machinery is what's responsible for recognizing the
00:29:59.03 consensus sequences at the splice sites. And then
00:30:02.07 in addition, at the same time there would be these
00:30:05.18 SR proteins and hnRNP proteins and then it is the
00:30:09.22 conjunction of all these different factors that then
00:30:15.14 makes the decision as to whether or not an exon is going to be
00:30:19.21 utilized or not and made into mRNA. So one thing
00:30:26.18 I'd like to point out about these splicing enhancer
00:30:29.14 and silencer sequences is that mutations in those sequences
00:30:33.03 can lead to human genetic disease. So about 20%
00:30:38.05 of human genetic disease is caused by mistakes in splicing.
00:30:43.22 And these mistakes can occur if a mutation occurs
00:30:47.11 at one of the consensus sequences--we've already
00:30:50.04 seen that for the beta-thallasemia genes--
00:30:52.24 or when a mutation occurs in one of the silencers or
00:30:57.12 enhancer sequences. So a common mistake that ismade
00:31:01.14 when scientists are hunting for the mutations that
00:31:06.29 cause a particular disease, if a mutation is found
00:31:09.27 in a coding exon and that mutation changes the amino acid that's
00:31:16.04 coded by that exon, it's often assumed that that mutation
00:31:20.11 is having its deleterious effects by changing an amino acid
00:31:25.26 in the protein. But in fact many times the mutation can actually
00:31:31.00 change the splicing pattern. So for example, leave out
00:31:34.18 a whole exon or change a splice site and that itself would be
00:31:38.26 much more deleterious. So it's important to think about that if
00:31:43.19 you're trying to hunt for mutations that cause a particular disorder
00:31:50.02 in either humans or some other model organism.
00:31:52.15 And it's also important to note that we can't ignore
00:31:55.19 mutations in introns. Not only are mutations at
00:31:59.07 the consensus sequences going to be deleterious
00:32:01.20 for splicing but mutations in these intronic splicing
00:32:07.13 silencers and enhancers can also change pattern. So
00:32:10.21 in the future, we're going to have to be sequencing
00:32:14.26 introns as well as exons when we're looking for mutations
00:32:17.26 in human disease. So here are some examples of
00:32:21.10 those kinds of mutations. The top here, we have
00:32:24.04 the SMN gene. Now all of us have two copies of this gene
00:32:28.06 there's SMN1 and SMN2. If you have a mutation
00:32:32.26 in SMN1, such that it is disrupted, this is a problem
00:32:40.28 because the SMN2 gene is identical to the SMN1 gene
00:32:45.23 except that it has a T mutation in an exonic splicing enhancer
00:32:51.05 such that in the SMN1 gene all of the exon 7 is included
00:32:57.19 but in SMN2 only 20% of exon 7 is included. So if you're
00:33:03.02 lacking the SMN1 gene, then you do not make enough of the
00:33:07.06 SMN protein and children who do not make enough of this protein
00:33:12.10 have problems with development of their motor neurons
00:33:15.20 and have spinal muscular atrophy. Here's another
00:33:18.24 example. Dystrophin is a mutation, which causes a
00:33:23.29 stop codon. So one might assume that the problem is
00:33:31.20 that it makes a truncated protein. But in fact
00:33:35.08 this stop mutation, this nonsense mutation, is in an exonic splicing silencer,
00:33:42.01 or it becomes an exonic splicing silencer such that
00:33:46.00 that exon is left out. And it still makes the protein
00:33:50.15 but it's lacking that exon. And then this is a third example.
00:33:56.10 Patients that have frontotemporal dementia often have
00:34:01.02 a mutation in the tau gene. And that mutation
00:34:04.28 causes an exonic splicing enhancer to be stronger such that
00:34:13.01 they're making more of the version of the protein
00:34:17.05 that has exon 10 included and less that has exon 10 excluded.
00:34:22.22 And this 50/50 ratio is very important to have
00:34:26.09 normal amounts of the tau protein. So in this case,
00:34:29.28 the patients are overexpressing this truncated form of the protein
00:34:34.24 and that leads to frontotemporal dementia.
00:34:39.28 So finally I'd like to leave you with my take home messages.
00:34:43.16 The first is that eukaryotic genes contain introns.
00:34:48.13 Not every eukaryotic gene contains an intron
00:34:50.24 but certainly in humans the vast majority contain
00:34:54.10 introns. These introns facilitate the evolution of new genes
00:34:59.11 and by allowing for alternative splicing, they allow us to
00:35:04.17 greatly increase our proteome complexity. And
00:35:08.29 higher eukaryotes utilize exon definition to define
00:35:12.26 splice sites because the exons are like islands in
00:35:17.16 oceans of intron. And then finally, many human
00:35:22.05 hereditary disorders are cause by mutations
00:35:24.24 that affect exon inclusion or splice site choice.
00:35:28.19 So this ends this part of the lecture. If you're interested
00:35:33.11 in learning something about the machinery that
00:35:36.16 removes the introns, load up my next lecture which is
00:35:40.26 spliceosome structure and dynamics. Thank you.
00:35:45.04

Part 2: Spliceosome Structure and Dynamics

00:00:00.02 Hi, I'm Melissa Moore from the University of MassachusettsMedical School
00:00:03.24 and the Howard Hughes Medical Institute and in this lecture
00:00:06.28 I'm going to tell you about some work from our own lab
00:00:09.27 on spliceosome structure and dynamics. So as we saw in
00:00:14.15 the first lecture, eukaryotic genes are split, in that they
00:00:20.00 have expressed regions or exons, up here, and introns.
00:00:26.13 And the introns have to be removed and in my previous
00:00:30.21 cartoon I just used the scissors and tape. So in this lecture
00:00:34.15 we're going to be talking about "What is the nature of those
00:00:36.27 scissors and tape, and how do they actually do the reaction?"
00:00:39.25 So first let's talk about how the splicing actually happens.
00:00:45.17 And it happens in two chemical steps. In the first step
00:00:50.18 of splicing, the branch site--so we talked about that, that's a
00:00:54.11 conserved adenosine in the intron near the 3 prime splice site.
00:00:58.19 The 2 prime hydroxyl of the branch site attacks the phosphate
00:01:05.07 at the 5 prime splice site, and that generates a lariat intermediate,
00:01:10.11 which has a 2 prime, and 5 prime and a 3 prime phosphate
00:01:13.23 all coming off that adenosine. And it releases the 5 prime
00:01:17.23 exon. Now that 5 prime exon doesn't float away, it's goingto be
00:01:24.10 held onto by the splicing machinery, which we'll be talking about
00:01:27.09 in a moment. The second step of splicing--so here are intermediates
00:01:33.22 in the reaction. This 3 prime hydroxyl that was generated
00:01:38.15 in the first step on the 5 prime exon--it now attacks the
00:01:42.16 phosphate at the 3 prime splice site, and now it kicks out
00:01:47.28 the intron and ligates the two exons together. Now we
00:01:53.09 know this is the chemistry and that these occur as single
00:01:57.27 transesterification reactions. The two steps of splicing
00:02:01.28 are catalyzed by a large complex in the cell called the spliceosome.
00:02:09.06 The spliceosome is arguably the most complicated macromolecular machine
00:02:13.26 in the cell, as we'll see in a moment. And the spliceosome consists
00:02:17.24 of four major chunks, large pieces, subunits that need to
00:02:23.23 come together with each round of splicing in this complicated
00:02:27.04 dance called the spliceosome cycle that we'll talk about
00:02:30.07 in a few slides. But what I want to show you here is that
00:02:34.12 the pieces of the spliceosome, the major pieces are these
00:02:38.13 components called U1, U2, the triple snRNP U4/5/6, and the NineTeen complex,
00:02:46.00 or the NTC. Now what are these things?
00:02:49.17 Now what are these U1, U2 things that I showed you in the last slide?
00:02:57.06 Well they are so-called snRNPs for small nuclear RNA protein complexes.
00:03:03.21 And so each of these snRNPs consists of a small nuclear RNA
00:03:07.27 somewhere between a hundred and a couple hundred of nucleotides.
00:03:12.19 It's a stable RNA complexed with a set of proteins. So for
00:03:17.00 example, U1 snRNP has U1 snRNA in it and a set of core proteins
00:03:23.05 called the Sm proteins. There are seven proteins. They make a
00:03:26.08 ring-like structure, and you can see they are common to
00:03:29.18 a lot of the snRNPS. And then some specific proteins
00:03:32.26 that are common to U1 snRNP and in U1's case--70K, A and C.
00:03:38.03 U2 is more complicated, and these are all of the proteins that are
00:03:42.10 associated with U2 snRNP. And then with the triple snRNP,
00:03:46.09 a so-called triple snRNP because it has three small nuclear
00:03:50.14 RNAs in it, and it has even a larger set of proteins.
00:03:55.17 In addition to the snRNPs, there's the NineTeen complex, so
00:04:00.12 named because it contains a protein called Prp19 and
00:04:05.15 its associated factors. So it is just like a snRNP except it
00:04:10.02 does not contain an RNA component. And then in addition
00:04:13.25 to these main stable components, there are also things called
00:04:18.11 splicing factors and these are proteins that come and go,
00:04:22.10 but are not stably associated with any single snRNP.
00:04:26.05 And included in this class are RNA helicases that change the
00:04:32.29 structure of RNAs or can change the structure of RNA protein complexes.
00:04:39.02 Certainly there are RNA binding proteins, we talked about
00:04:42.12 two of the classes of those in the last lecture--the SR proteins
00:04:46.06 and the hnRNP proteins. And there are unexpected proteins
00:04:51.05 like cis/trans prolyl isomerases and ubiquitin ligases.
00:04:54.25 Altogether the complete spliceosome parts list as we now understand it
00:05:01.23 consists of 5 snRNAs (U1, U2, U4, U5 and U6) and a hundred
00:05:09.26 proteins in yeast, a hundred different proteins, and about
00:05:12.25 three hundred different proteins in humans. And the reason that the
00:05:17.13 human splicing machinery is so much more complicated
00:05:19.17 than the yeast splicing machinery, you can imagine
00:05:22.11 why because we have so many different introns, and we do all
00:05:25.07 this alternative splicing that the yeast don't. And so most
00:05:28.08 of these proteins, the extra ones, are involved in alternative splicing.
00:05:32.00 Now let me tell you a little bit about the snRNAs. They're called
00:05:36.03 "U" snRNAs because they're uridine-rich RNAs, and their
00:05:40.23 numbering came from--if one just purifies all of the stable RNAs
00:05:47.27 out of the nucleus and runs them on a gel, the most abundant
00:05:53.01 one is U1, the second most abundant is U2, and so forth.
00:05:57.06 And it turns out that U3 does exist, but it is involved in ribosome biogenesis
00:06:03.28 as is U7. But the other five of the six most abundant
00:06:09.29 are all involved in pre-mRNA splicing, or the spliceosome.
00:06:14.13 Here we are at the spliceosome, back at the spliceosome cycle,
00:06:17.15 and we're going to look at this a little more closely because
00:06:20.18 I'm going to be telling you about some experiments we've recently
00:06:22.29 done to test this cycle. So in the first part of the cycle,
00:06:27.24 at the beginning of spliceosome assembly, U1 snRNP interacts with
00:06:33.23 the 5 prime splice site and U1 snRNA actually base pairs
00:06:38.22 and recognizes the 5 prime splice site. Similarly, U2 snRNA
00:06:44.29 base pairs with and recognizes the branch site consensus sequence,
00:06:49.22 and U2 snRNA, when it joins the U1 snRNA, forms A complex.
00:06:56.03 E complex stands for "early" complex and then A, B, and C is--
00:07:02.21 we'll see in a minute--were just named by where they ran
00:07:06.07 on a gel. So A complex comes first, and then the next big chunk
00:07:11.12 of the spliceosome that comes in is the triple snRNP--
00:07:15.17 U4/5/6. Once the triple snRNP is there, we have B complex.
00:07:21.10 Then there's a structural rearrangement, where U1 snRNP is actually
00:07:26.24 ejected. It's kicked out, and somehow things are majorly rearranged such
00:07:31.20 that U6 snRNA is interacting with the 5 prime splice site. In another rearrangement
00:07:38.27 U4 snRNP is kicked out, and then the NineTeen complex comes in,
00:07:47.17 and now we have the catalytically active spliceosome, where the first step
00:07:53.18 and then the second step of splicing occur. Once the second step of splicing
00:07:58.01 is complete, the splice product is released, and the spliced out intron
00:08:05.21 leaves with the rest of the spliceosome. That splicing machinery has to disassemble,
00:08:10.27 and then it's reassembled anew on each new round of splicing, so ergo
00:08:15.29 the spliceosome cycle. So how do we know a lot of these details of the mechanism
00:08:22.04 that I've been telling you about? Well one of the ways that we know is by
00:08:25.19 performing in vitro splicing reactions. So in an in vitro splicing reaction,
00:08:31.15 we take a piece of RNA, usually a piece of RNA that would be a couple
00:08:37.19 hundred of nucleotides that would consist of an exon with one intron
00:08:42.16 followed by a downstream exon. And the little asterisk here, the red asterisks,
00:08:49.01 are to indicate that this RNA would be radioactively labeled, so we would
00:08:53.04 transcribe it in vitro and put radioactive nucleotides all throughout.
00:08:59.23 We then mix this RNA with either whole cell extract if we're working on the yeast spliceosome
00:09:05.08 or nuclear extract if we're working on the human spliceosome. And for example, we
00:09:09.20 might get that from HeLa cells, which is a very common tissue culture cell for humans.
00:09:15.01 And then also ATP because ATP is essential for splicing because many of the
00:09:21.21 spliceosome transitions that I was showing you here all of the ones
00:09:26.12 after E complex formation, each one of those steps requires ATP and
00:09:32.07 also going around the backside here. Now why do we use whole cell extract
00:09:38.01 or nuclear extract? Well I've just told you that the splicing machinery is
00:09:43.08 incredibly complicated. It has in yeast a hundred different polypeptides;
00:09:48.15 in humans, three hundred different polypeptides. There's really just no
00:09:51.26 way at present that we can purify each one of those proteins, have that
00:09:56.23 in a test tube and fully reconstitute the machinery. So right now, the
00:10:01.10 best way to study the spliceosome is simply almost in its native
00:10:04.12 environment, and that is in unfractionated cell extract. So if we then take
00:10:11.16 those splicing reactions and take out time points and then purify
00:10:17.23 the radioactive RNA and run it on a denaturing gel--and in this case a
00:10:22.07 pretty high percentage denaturing gel--what we can see is that over
00:10:26.16 time (and this time course goes from about zero to sixty minutes
00:10:31.23 in vitro) we can see the substrate gradually disappear. And then at early
00:10:40.11 time points, the two intermediates of splicing appear, the lariat intermediate
00:10:44.11 and the 5 prime exon. And then you can see at later time points the
00:10:49.18 lariat product, the intron product, and the spliced exon product
00:10:56.22 appear. And the reason that the lariats run high in the gel, even though
00:11:01.26 they're smaller than the pre-messenger RNA, is that they have this
00:11:06.21 circular structure, and because of that unusual structure, they're retarded
00:11:11.17 in the gel more than a linear RNA and so they actually run higher than
00:11:16.16 you would expect. But now if instead we take that same splicing reaction
00:11:23.27 but don't purify the RNA and just run it on a native gel, so now we are
00:11:30.05 looking at the complexes that contain the RNA. Here we can see those
00:11:35.12 complexes I told you about before. So here's E complex, the early complex,
00:11:39.18 A, B and C and they build up and go away over time, as you might expect.
00:11:45.29 And so that's where the names of these different complexes came from,
00:11:49.10 simply by their migration on the gel. Alright, now these different
00:11:57.05 complexes can be purified. They're stable. They're stable enough to
00:12:01.11 survive a native gel, and there have been many different ways devised now
00:12:07.13 to purify the different complexes. Here's one example from my laboratory.
00:12:11.17 What we did was we took a splicing substrate, where we mutated
00:12:15.20 the 3 prime splice site, so the splicing machinery could build up on
00:12:19.28 the RNA. It could do the first step, but it couldn't do the second step
00:12:24.11 because the 3 prime splice site was mutated. And into that intron, we built
00:12:30.04 stem loops that were recognized by the MS2 protein, which is a viral coat
00:12:35.06 protein that binds very tightly to its recognition sequence. That viral coat
00:12:40.09 protein called MS2 we linked to maltose binding protein. Maltose binding
00:12:45.13 protein likes to bind amylose resin, so we could use that as an affinity
00:12:50.12 tag to pull down these spliceosomes and purify them. And you can see here
00:12:56.07 an EM image of those spliceosomes and the spliceosomes are all
00:13:02.24 around 20-30 Angstroms in size. Now from those EM images, you can do
00:13:11.07 single particle reconstruction to start to get structures of the splicing machinery.
00:13:17.27 And at this point, in comparison to the ribosome, for example, the
00:13:25.08 structural information we have for the spliceosome is rather limited.
00:13:30.13 We do have crystal structures now--two different crystal structures of U1 snRNP,
00:13:36.20 which is the most common of the snRNPs. And those are at 5.5 and 4.4
00:13:43.08 Angstrom resolution, so enough to see the RNA and the proteins.
00:13:49.13 But for the bigger complexes, we are still at the electron microscopy
00:13:54.09 stage, and so for example, here are some images from Reinhardt Luhrmann's lab
00:13:57.28 of the yeast splicing complexes--the B complex, the activated B complex
00:14:04.04 and then the C complex, which is the spliceosome that has intermediates
00:14:08.12 in it. And so in the coming years, we're really--the splicing community--is
00:14:14.08 really struggling hard but looking forward to having high-resolution structures
00:14:19.09 because we would really like to see where all of these parts bind
00:14:22.19 and how they all fit together to form these just really remarkable machines.
00:14:27.09 Now in the meantime, that's where we are on the structural front. In
00:14:34.11 addition to structure for any biochemical complex or machine, you
00:14:38.29 really need to note something about their dynamics, so I've gone over this
00:14:42.29 whole splicing cycle with you but as I explained before, the splicing cycle is
00:14:47.17 based on the complexes that are stable enough to be resolvable
00:14:54.14 on a gel or you can affinity purify them. But it doesn't tell you about
00:15:00.19 the kinetics of things coming and going. So for example, is it going to be true
00:15:07.00 that on every intron U1 has to come before U2, and do these two have to come before
00:15:14.01 the triple snRNP? And all of these arrows that we're showing here are one-way
00:15:19.12 arrows, but most biochemical reactions and chemical reactions really are
00:15:23.11 two-way. So are these arrows really one-way: is it a one-way street?
00:15:28.07 Or is this process in any way reversible? So to get at that information
00:15:34.07 my laboratory has recently been collaborating with Jeff Gelles's laboratory
00:15:39.02 at Brandeis University and Virginia Cornish at Columbia University
00:15:43.29 as well as some co-workers at New England Biolabs to develop new
00:15:48.12 tools in order to look at the dynamics of the spliceosome. The main
00:15:53.27 method that we have been utilizing is called total internal reflectance.
00:15:59.08 Now this is an experiment you can try at home. So this is simply a laser
00:16:03.21 pointer that's going into an aquarium of water. And if you correctly position
00:16:11.12 the laser pointer at the critical angle, when there's a change in
00:16:15.23 refractive index, in this case between water and air, then all of the laser
00:16:20.23 light will be completely reflected, that's called total internal reflectance.
00:16:25.05 Except right at the point where the laser contacts there is a little bit of
00:16:34.15 energy that goes through the other side, called the evanescent wave.
00:16:38.13 So the evanescent wave--in this case now we're going to be having the lasers
00:16:44.26 come through the air to a microscope slide. So here is the change
00:16:49.02 in refractive index is going from the microscope slide to the aqueous
00:16:56.14 layer above the microscope slide. The evanescent wave will go about
00:17:00.05 a hundred nanometers into the solution above the microscope slide.
00:17:05.25 So imagine having not just one laser but let's say three different colors
00:17:11.17 of lasers. It turns out we can now do five lasers. I'm only going to show you three,
00:17:16.05 three today. But imagine having three different colors of lasers all going
00:17:23.23 in at their critical angles, and having something tethered to the surface
00:17:27.25 within this 100 nanometers and having molecules that are fluorescent
00:17:34.07 in the colors that are excited by your three different lasers. So molecules
00:17:41.14 that are in solution above the evanescent wave are not fluorescent
00:17:45.25 because they're outside the area of where the light energy is. And so
00:17:51.10 only the molecules that are tethered to the surface are going to be fluorescent.
00:17:55.07 So we can use this to then ask, for anything that's tethered to the surface
00:18:01.02 what different colored molecules at any one time are associated with the
00:18:05.13 molecule that's tethered to the surface. So let's just see how this looks.
00:18:09.04 So imagine you're looking down at this surface and we're going to be
00:18:13.27 looking at the molecules on the surface. So we call this technique colocalization
00:18:20.21 of single molecules spectroscopy, or CoSMos. And this technique was
00:18:25.13 pioneered by Jeff Gelles and his co-worker Larry Friedman at Brandeis
00:18:29.00 University. So here we are looking at the fluorescence, and each one of
00:18:35.29 these spots is a single molecule on a glass coverslip that has different colored
00:18:42.21 things on it. In this case, the molecules are a strand of DNA, and the
00:18:49.16 colored things are different oligos that are complementary to that strand of DNA,
00:18:53.22 but they have different fluorophores on them. And so you can see for
00:18:57.12 example that this molecule of DNA had all three of the oligos bound to it
00:19:04.18 but this molecule of DNA only had this one--only had the green and the
00:19:10.04 blue bound to it. And you can see, here's one that only had the blue
00:19:13.19 molecule bound to it. So it's very simple because all we're doing is
00:19:18.25 we're looking at this, say, constellations, different constellations of spots
00:19:23.17 and we're going to learn something about our biological system.
00:19:27.09 And in particular if we can look at how these spots change over time,
00:19:33.13 we can learn about the dynamics of the system. Now in order to
00:19:37.19 use this to study the spliceosome, we had to develop a number of
00:19:42.01 different or new technologies to enable us to label parts of the spliceosome
00:19:46.22 so that we could see them. And so one of the things that we had to do
00:19:51.18 was to create fluorescently tagged pre-mRNAs because we need to know
00:19:57.21 where the pre-mRNAs are on the surface. Also our pre-mRNAs have to
00:20:02.26 have some way to be tethered to the surface. The way we do that is to
00:20:06.13 put a biotin molecule on one end, and then we also have biotin on the
00:20:12.05 glass surface. We have biotinylated PEG--polyethylene glycol.
00:20:17.19 And then we make a sandwich, where we have streptavidin. Streptavidin
00:20:21.24 can bind four molecules of biotin, so you can use that to make a
00:20:25.11 sandwich and bind your RNA there. Now the other thing that we had to
00:20:30.07 develop were other ways of tagging the snRNPs because what we really
00:20:33.27 wanted to do was look at the snRNPs coming and going in real time.
00:20:37.22 So the way that we're tagging the snRNPs is using two protein tags.
00:20:44.22 One is the SNAP tag that was developed by Kai Johnsson and is now
00:20:49.17 available through New England Biolabs. SNAP is based on a protein that
00:20:55.12 is a suicide enzyme that removes alkyl groups from guanine nucleotides
00:21:01.23 of DNA. And so it transfers those alkyl groups to itself. So in this case
00:21:08.03 if you have benzyl guanine--so here's guanine and then there's a benzyl group
00:21:13.19 on it and if you attach to that a fluorescent dye, the SNAP tag protein
00:21:20.05 will transfer that dye to itself, and if you've made a fusion protein
00:21:25.21 between the SNAP tag and your protein of interest--in this case a
00:21:30.08 snRNP protein--then you can specifically label your snRNP protein.
00:21:34.29 Here the other tag that we've been using is the E. coli DHFR,
00:21:42.03 dihydrofolate reductase tag, and bacterial dihydrofolate reductase binds
00:21:50.08 very tightly to trimethoprim--this molecule down here. It's a non-covalent
00:21:54.15 interaction. Trimethoprim is an inhibitor of E. coli DHFR but this molecule
00:21:59.29 does not bind to eukaryotic DHFR. But this is a very tight interaction and
00:22:07.29 again if we tether a dye with that, this dye will interact with our DHFR
00:22:13.14 tag and allow us to label that protein. And this technology was
00:22:19.18 developed by Virginia Cornish and her coworkers at Columbia University.
00:22:23.22 So how do we get these tags on our snRNPs? The way that we do this is
00:22:29.17 we're using the yeast system and we're using homologous recombination.
00:22:33.16 So we make versions of different protein genes that we want to
00:22:38.09 tag--in this case two U1 proteins and a U2 protein. We then place the tag
00:22:45.22 of interest at the C-terminus of that protein, or the gene for that protein
00:22:50.00 and then we have a selectable maker. And we use homologous
00:22:54.25 recombination to put these modified genes into haploid yeast. And
00:23:01.11 that means the only gene that is encoding that protein in the yeast is
00:23:07.17 our protein of interest--our tagged protein. So then from those yeast
00:23:13.15 strains, we can make a whole cell extract. And in this case we have U1
00:23:18.15 having two DHFR tags on two different proteins, or U1 having two tags
00:23:26.03 and U2 having a SNAP tag on it, so a triple-tagged strain. We then
00:23:33.06 take those extracts and we either can simply add the TMP to label
00:23:40.17 the DHFR or to label the SNAP tag, we take our Benzylguanine that has a
00:23:46.15 fluorescent tag on it. We react that with the whole cell extract. We remove
00:23:52.07 the excess dye by gel filtration and then now we can add our TMP. And so
00:23:58.04 the really great thing about this system is first of all we know that the
00:24:04.00 proteins we're tagging are active because 1) they're the only copy of
00:24:08.09 the protein in the cell and we're only tagging essential proteins.
00:24:14.29 Many of the proteins in the spliceosome are essential, and so we know
00:24:20.07 that if the cells grow, because splicing is essential, then that protein
00:24:24.06 must be active. Secondly, there's absolutely no protein reconstitution
00:24:29.04 required, so we're not making any recombinant proteins, purifying them
00:24:32.25 and putting them back in. We're using the endogenous proteins. We just
00:24:36.08 added a small protein tag to the thing. So let's look now at how these
00:24:43.03 experiments are going to go. So we're going to have our pre-mRNA
00:24:50.18 that's attached to the surface via this biotin-streptavidin sandwich.
00:24:54.12 It has a fluorophore in it so we can keep track of where the pre-mRNA
00:24:58.24 molecules are. And this is actually a view through the microscope of
00:25:04.09 what a field of these pre-mRNAs look like, where each one of these spots
00:25:08.21 is a single molecule of pre-mRNA. And in the movies I'm going to show
00:25:15.11 you we're going to be looking at U1 snRNP binding to those pre-mRNAs
00:25:21.20 over time in splicing reactions. One of the things about single molecule
00:25:27.09 reactions is that you're really seeing everything that's going on--
00:25:36.11 anything that's fluorescent, any kind of dust or anything you can see,
00:25:41.01 so you really need to do a lot of controls to make sure that you know what
00:25:44.21 you're looking at. So the first thing I'm going to show you is a movie
00:25:48.19 where we're doing some controls, where either we have left off the
00:25:54.13 fluorescent RNA, or we don't have the tags on U1 snRNP (and so we
00:26:00.28 wouldn't expect signal) or we have the complete reaction where we
00:26:04.19 have the fluorescently tagged RNA and the fluorescently tagged snRNP.
00:26:09.07 So let's watch that movie. This movie shows two control fields of view
00:26:19.21 and then one experimental field of view on the right. The field of view
00:26:24.28 all the way to the left, we have the wild type pre-mRNA present.
00:26:31.21 We can't see that in this field of view because we're not looking in that
00:26:35.13 channel. We're looking at the Cy3 channel, which is the TMP channel.
00:26:39.18 And we also have the Cy3-TMP in the extract, but there is no tagged
00:26:44.05 protein. So you can see that we do have a little bit of background with
00:26:48.07 material binding non-specifically to the slide and so this is why it's important
00:26:55.07 to do those controls, to make sure your background is not too high.
00:26:59.06 In the middle panel, we now have the tagged U1 and the Cy3-TMP
00:27:07.24 but we don't have any pre-mRNA on the slide, so again we only see
00:27:11.26 background binding. And then in the rightmost panel, which is the one
00:27:16.14 with all the blinking lights, we have all three components. So we have
00:27:21.05 tagged U1. We have the pre-mRNA on the surface, and we have Cy3-TMP.
00:27:28.11 Now one thing that you can see immediately from this is that U1 interaction
00:27:35.03 with the RNAs is highly dynamic. So even in the absence of ATP, U1
00:27:40.27 is binding and releasing multiple times from each pre-mRNA. Now that we
00:27:48.28 know our system's working, let's really do some experiments. And the
00:27:53.09 really cool thing about these experiments is that you can just see
00:27:58.09 the answer with your eyes. So I'm going to show you some movies
00:28:04.15 next, where we've put two fluorescent tags on each of the major subcomplexes
00:28:10.00 So in one extract, in one quadrant, you're going to see extract that
00:28:16.08 has labels on U1 snRNP, as you've already seen on two different proteins.
00:28:20.25 Then we have another extract that has labels on U2 snRNP, on the U5 component
00:28:26.21 of the triple snRNP, and also on the NineTeen complex. And in this first set of
00:28:34.12 movies, we're going to not have ATP present, so in the absence of ATP
00:28:40.27 we've known from the studies on gels that the only complex that should
00:28:49.15 form is this E complex, so only U1 should be able to stably interact with
00:28:55.02 the RNA. So now let's look and see if that's the case.
00:29:02.22 Here's a movie, showing four different extracts with a different snRNP
00:29:11.02 labeled in each extract--either U1 in the top left-hand corner, the triple snRNP
00:29:17.27 in the bottom left-hand corner, U2 in the top right-hand corner or the
00:29:22.06 NTC in the bottom right-hand corner. And in the absence of ATP, what
00:29:28.05 you can see as we saw in the previous movies, that U1 is coming and binding
00:29:34.18 reversibly, but for all the other snRNPs, we do not see any significant binding
00:29:41.04 over background. If we take the data from each of those fields of view
00:29:46.26 and simply count the number of spots over time--the total number of spots
00:29:53.20 over time--what you can see is that in the absence of ATP only U1
00:29:58.27 builds up. And none of the other snRNPs or the NTC really have much
00:30:05.07 occupancy at any one time in the absence of ATP. So now let's run the
00:30:12.16 whole spliceosome cycle, so now we're going to add ATP and see what happens.
00:30:19.17 This movie is now in the same order as before but now we've added ATP
00:30:25.19 to the reaction. And if you watch very carefully, you can see the
00:30:32.06 apparent order of addition of the snRNPs. So early in the movie,
00:30:37.03 and the movie is continually looping, you can see that U1 is binding
00:30:42.10 and coming and going. The next snRNP to build up is U2 and then
00:30:52.02 after U2, we start to see U4, 5 and 6 come up. And then the NTC, we
00:30:59.10 see less of it, but it accumulates much later in the reaction.
00:31:03.24 So what you can see from this movie is that we can see in real time
00:31:12.04 all four of these snRNPs binding to the surface that's covered with
00:31:16.22 pre-mRNA molecules. And as I will show you in the next slide, we can see
00:31:23.10 that all of the snRNPs are binding dynamically--that is, they're coming
00:31:27.08 and going, that none of them are coming and staying permanently.
00:31:32.07 One of the things that you can see from those movies is not only can we
00:31:37.18 see that all of the snRNPs are binding in the presence of ATP, but unlike
00:31:42.11 the spliceosome cycle that I showed you before with all the one-way arrows,
00:31:47.12 all of the snRNPs are binding reversibly. And we can see this by--here's looking at
00:31:52.24 one individual RNA molecule, and we're just looking at the intensity over time for that
00:32:00.07 one RNA molecule and you can see for U1 it bound twice. Here's an RNA
00:32:05.16 molecule where two molecules of U2 bound, U5, and the NTC. The reason
00:32:11.08 we don't just see two binding events often times especially for U1, we'll
00:32:16.00 see three, sometimes up to ten binding events. The reason we're showing
00:32:21.19 these particular traces is it shows you that this binding is due to reversible
00:32:27.09 binding and not due to photobleaching. So photobleaching is always a problem
00:32:31.08 in these single molecule reactions because under the intense laser light
00:32:36.09 the dyes can often photobleach and then they go blank and then when
00:32:41.04 the signal disappears you don't know if it's because your complex has
00:32:44.13 gone out of the evanescent wave or your dye has photobleached. So this is
00:32:49.02 why we attached two different fluorophores to each snRNP because
00:32:53.13 when we see the stepping behavior, this is either due to dye release
00:32:58.27 because we're using the DHFR tag or it's due to photobleaching. But that
00:33:03.26 means that this molecule which went away in one step really had to be a
00:33:08.26 molecule that went away. The whole complex went away. Because it's very
00:33:14.28 unlikely that you'd have two simultaneous photobleaching steps or two
00:33:20.02 simultaneous dye steps. And also you can see it went away and then
00:33:27.28 another one came back. So again this is another one of the controls that
00:33:32.13 you need to do when you're doing single molecule experiments. Alright
00:33:36.14 so thinking back to those movies, if we count up the total number of
00:33:41.16 spots in each frame and just plot that number here, and this would be the
00:33:46.19 number of dyes per pre-mRNA molecule. You can see now in the
00:33:51.02 presence of ATP U1 builds up first, then U2, then U5, and then the NTC
00:33:58.11 after that. So this gives us an apparent ordered process for spliceosome
00:34:04.05 assembly, and it's consistent with that. But it doesn't actually tell us for any
00:34:09.13 one molecule that the spliceosome assembly was ordered. But we can
00:34:15.14 test that directly with our single molecule methods simply by following
00:34:20.07 two snRNPs at once. So now we're going to do three-color experiments.
00:34:24.04 So one color on the pre-mRNA, one color on--in this case--U2 snRNP,
00:34:30.25 and another color on U1. And in the same experiment, by watching these snRNPs
00:34:36.13 simultaneously, we can see does U1 come first or does U2 come first?
00:34:43.25 And so here's where these data look like. So here again is one of these
00:34:48.22 individual single molecule traces, but this is one pre-mRNA molecule where
00:34:53.06 U1 and U2 came to the same pre-mRNA in the same extract. And you can see
00:34:59.22 very clearly that U1 came first and then U2 came. But we can quantify this by
00:35:07.26 measuring the on time for both U2 and U1 and taking this difference,
00:35:14.00 the time of arrival of U2 minus the time of arrival of U1, and that's the
00:35:20.10 delay time between the two. So if that number's positive, that means
00:35:24.24 U2 came after U1. If that number is negative, that means U2 came before U1.
00:35:32.08 And then we can look at this number over many different pre-mRNA molecules.
00:35:38.08 So this is a so-called probability density plot. It's a bar graph where the probability
00:35:45.22 density is the bin height divided by the bin width. The important thing to see
00:35:51.16 is that here we're looking at 82 different molecules and that almost all of them
00:35:55.20 had a positive number for this tU2 minus tU1. So that means that on almost
00:36:02.20 all of them, U2 came after U1. Now there were a few here where U2
00:36:08.18 apparently came first, but that would be consistent with the amount of
00:36:15.16 labeling of our extracts because we can't get completely a 100% labeling.
00:36:20.11 It is impossible without our extracts dying. So we have about 90%
00:36:27.05 labeling efficiency in our extracts, and this level would be consistent with
00:36:31.15 a dark U1 coming before U2. So it's not inconsistent with an ordered model.
00:36:39.22 So we've done this for all of the pairs of the complexes. Here's U2 versus
00:36:46.28 U1. This is another set of data now with 111 molecules. Here's U5
00:36:52.05 versus U2, and here's the NTC versus U5. And you can see all of these,
00:36:57.27 most of the events, gives you a positive number so therefore the
00:37:04.21 second complex came after the first complex. And so that leads us to
00:37:11.24 conclude that for this particular pre-mRNA that we've been working with
00:37:17.07 (and this is RP51A--it's a model splicing substrate in yeast), it is a highly
00:37:22.29 ordered assembly pathway with U1 almost always preceding U2 and
00:37:31.19 then the triple snRNP, and then the NTC comes after that. But what we now know
00:37:37.07 that's new that we didn't know before is that every step in this pathway
00:37:42.11 is reversible. So that means that the pre-mRNA is not necessarily
00:37:52.21 committed to splicing at the very first step with U1 addition but that
00:37:58.17 it increasingly gets committed as you go through the spliceosome
00:38:03.00 assembly pathway. Also, in terms of alternative splicing in the human system
00:38:08.20 if the spliceosome can be disassembled, for example here at later points,
00:38:15.06 then you can imagine you could inhibit splicing at particular splice sites
00:38:19.02 anywhere along this pathway because it could go backward along the
00:38:23.14 pathway if it's inhibited. So this has important implications for our understanding
00:38:27.24 of alternative splicing. Finally, I need to thank the people who actually did
00:38:32.13 the work. And obviously anything this complicated took the input of many
00:38:39.07 different people. And so from my laboratory I showed you data today from Melissa,
00:38:46.28 Aaron, Danny, Eric, Jing, and Nick contributed. Also all of this work was done in
00:38:53.28 collaboration with Jeff Gelles's lab, which developed the CoSMos
00:38:57.10 technology, in particular Larry and Alex. And then also the Cornish
00:39:03.24 Laboratory, who developed the DHFR labeling technique, and finally the
00:39:08.23 New England Biolabs for their help with the SNAP tag. And this work was
00:39:14.09 funded by HHMI and the National Institutes of Health. Thank you very much.
00:39:22.07

Videos in this Talk

Part 1: Split Genes and RNA Splicing
Audience:
- Student
- Researcher
- Educators of H. School / Intro Undergrad
- Educators of Adv. Undergrad / Grad
Part 2: Spliceosome Structure and Dynamics
Audience:
- Researcher
- Educators of Adv. Undergrad / Grad

Speaker: Melissa Moore

Total Duration: 1:15:30

Recorded: July 2011

All Talks in Genetics and Gene Regulation

Talk Overview

In the first part of her talk, Dr. Moore explains that eukaryotic pre-mRNA contains long stretches of non-protein coding sequences interspersed with protein coding regions. By recognizing specific sequences, cellular machinery splices out the non-coding introns leaving just the protein-coding exons in mRNA. Although at first glance this may seem like a wasteful process, it is splicing that facilitates the evolution of new genes, and alternative splicing that allows a limited number of genes to produce a large number of proteins.

Moore goes on to describe the cellular splicing machine, the spliceosome, in greater detail in Part 2. She lists the components of the spliceosome and where each works in the spliceosome cycle. Moore also explains how the innovative use of fluorescent protein tags and total internal reflection microscopy has allowed her and her colleagues to better understand the ordered assembly and function of the complex splicing machine.

Speaker Bio

Melissa Moore

Melissa Moore received her PhD in biological chemistry from the Massachusetts Institute of Technology. She became interested in understanding RNA splicing during a post-doc with Phillip Sharp, also at MIT. Since then, Moore has spent much of her career working on the spliceosome and other mysteries of RNA processing. Moore was a Howard Hughes Medical… Continue Reading