Discovery of the Structure of the Nucleosome

Top

00:00:07.14 Hello, my name is Karolin Luger.
00:00:09.20 I'm a structural biologist at the University of Colorado Boulder
00:00:13.11 and an investigator for the Howard Hughes Medical Institute.
00:00:17.21 And today, I want to tell you a coup...
00:00:20.02 about a couple of key experiments
00:00:22.04 that enabled a breakthrough in our decade-long efforts
00:00:26.01 to determine a three-dimensional structure of the nucleosome.
00:00:30.07 But before I do this, I want to highlight
00:00:32.09 the importance of this problem at that time.
00:00:34.27 The human genome is over two meters in length
00:00:37.11 and has to fit into a nucleus
00:00:40.09 that's about five microns in size.
00:00:42.09 A good analogy for this would be if you had
00:00:45.09 to fit a human hair that is ten miles long
00:00:47.25 into a golf ball.
00:00:51.00 And you... now imagine that you have to take that hair
00:00:55.08 and split it into two during a process called DNA replication,
00:00:58.26 or imagine you have to find certain bits of information
00:01:01.23 on that piece of hair,
00:01:04.00 all the while leaving the hair in the golf ball
00:01:06.14 and avoiding tangles or knots.
00:01:11.00 A good analogy for using DNA
00:01:13.20 as an informational molecule is a cassette tape.
00:01:16.20 That was used in prehistoric times by humans
00:01:19.16 to record music and play music.
00:01:22.10 Just like the cassette tape,
00:01:25.02 human genome packaging poses many challenges.
00:01:28.19 We have to package the information
00:01:30.29 in a very confined space.
00:01:32.16 We have to protect it from physical damages and tangles.
00:01:35.26 We have to regulate access to the stored information --
00:01:38.19 so, you need to find your favorite ABBA song on your tape.
00:01:41.13 You need to be able to tag that somehow and find it rapidly.
00:01:44.21 And we also have to accurately and rapidly
00:01:48.03 replicate the information that is encoded on the tape --
00:01:51.11 let's say you want to share that tape with your friends.
00:01:54.24 Early views of human genome packaging
00:01:57.07 came from taking chromosomes out of cells
00:02:00.03 and spreading them very, very thinly
00:02:02.24 on electron microsco... microscopy grids.
00:02:05.20 This revealed that the fundamental organizing units of chromatin
00:02:11.18 are spherical, about 10-nanometer-sized particles.
00:02:16.10 These can then be separated out
00:02:21.22 from the chromosome
00:02:23.26 by using an enzyme called micrococcal nuclease.
00:02:25.19 This enzyme cuts unprotected DNA,
00:02:27.24 but it will not cut the particles themselves.
00:02:30.07 And then further analyses of these particles
00:02:32.22 revealed that they're composed of DNA,
00:02:35.06 roughly 150 base pairs of DNA,
00:02:37.27 and four to five small, very basic histone proteins.
00:02:42.15 Efforts were undertaken to determine the structure of the nucleosome
00:02:46.20 by X-ray crystallography,
00:02:48.04 which was the method of choice in the 20th century.
00:02:51.09 This method entails taking your molecule
00:02:53.28 that you're interested in
00:02:56.03 and convincing it to form ordered crystals.
00:02:58.17 You do this by subtly changing the solvent conditions
00:03:02.10 around the molecule to...
00:03:04.23 molecules to convince them to aggregate with each other
00:03:08.01 in an ordered manner
00:03:09.24 to form these beautiful structures here,
00:03:11.22 as shown in this green droplet.
00:03:13.15 These crystals are about 100-500 micrometers in size,
00:03:17.09 so they're very, very small.
00:03:19.03 So, you use a microscope to fish them out
00:03:21.19 with a tiny little loop -- it's almost like a spoon --
00:03:24.22 to mount them in this crystal loop.
00:03:27.13 This loop is then immersed in liquid nitrogen
00:03:30.21 to rapidly vitrify the solvent around it
00:03:33.22 and is then mounted on a goniometer head to be...
00:03:37.13 to be illuminated by X-rays.
00:03:41.27 The electrons in our molecule
00:03:43.22 will diffract the X-rays,
00:03:45.12 and those diffracted X-rays
00:03:47.09 are collected on either a film
00:03:49.26 or, in later times, by electronic media.
00:03:53.00 And it is this diffraction pattern that then,
00:03:55.15 through a series of quite complicated
00:03:57.23 mathematical operations that I will not go into,
00:04:01.07 will then give rise to what we call
00:04:04.25 the electron density map.
00:04:06.07 And this electron density map
00:04:08.11 is then used to build your macromolecule,
00:04:11.00 which then allows you to identify
00:04:14.15 the location of every atom
00:04:16.07 with respect to every other atom.
00:04:18.21 Now, you hear crystallographers talking a lot about
00:04:22.08 high resolution.
00:04:23.26 High resolution is really needed
00:04:25.27 to understand function.
00:04:27.10 What does that mean?
00:04:28.28 Imagine you're an alien and you're trying to figure out
00:04:32.00 the function of a human hand.
00:04:34.04 Imagine now all you... all the...
00:04:37.07 the only information that you have is a... is a mitten.
00:04:39.04 You find a mitten on the road,
00:04:40.25 and from that you conclude that humans
00:04:43.10 have a large paddle-like protrusion
00:04:45.10 and a smaller protrusion,
00:04:47.02 and this constitutes the human hand.
00:04:49.02 Now, imagine that you're finding a ski mitten.
00:04:50.28 You're getting quite a lot of...
00:04:53.12 more resolution.
00:04:55.05 You now know that humans have five fingers.
00:04:56.26 But it's still quite limited
00:04:59.00 in the details that you can observe.
00:05:02.28 The highest resolution structure you would obtain
00:05:05.13 from a latex glove,
00:05:07.06 and there you can figure out details in the finger joints,
00:05:09.27 and you could begin to understand how the human hand functions.
00:05:14.24 This also illustrates a major problem
00:05:17.07 that we have in crystallography,
00:05:19.06 because the images that we obtain
00:05:21.29 -- the structures that we obtain --
00:05:23.25 are static images.
00:05:25.08 And from static images,
00:05:26.23 it's sometimes hard to understand functions.
00:05:28.22 And that's why crystallographers
00:05:30.17 and other structural biologists
00:05:32.12 always use orthogonal approaches
00:05:34.03 to obtain different states of the molecule in question
00:05:38.28 so that we can interpolate between those different states
00:05:42.00 to see how they get from here to there.
00:05:43.29 And we also use additional methods
00:05:45.26 to observe these machines and molecules
00:05:49.24 in the act of doing their job.
00:05:54.14 Now, back to resolution.
00:05:56.00 What determines the resolution in a crystal?
00:06:00.12 It is determined by order.
00:06:02.05 A single crystal is contained...
00:06:04.06 is composed of billions and billions of macromolecules.
00:06:08.11 In the ideal case, all macromolecules
00:06:10.21 would look the same,
00:06:12.04 as seen in the example to the left side.
00:06:14.02 We have all of our macromolecules
00:06:16.11 very happy and smiling,
00:06:17.28 and so the average structure that we get
00:06:20.12 from that crystal lattice
00:06:22.09 would be a smiley.
00:06:24.19 And this would be a very high resolution structure.
00:06:27.17 In contrast, if some of our macromolecules
00:06:29.22 are not happy -- they're frownies --
00:06:32.00 they are incorporated into the crystals,
00:06:33.29 and what we get from these crystals
00:06:35.23 is a composite of smileys and frownies.
00:06:39.04 And so, we really have a hard time
00:06:41.26 to determine what their facial expression is.
00:06:44.00 And this is what would constitute a low resolution structure.
00:06:49.09 Now, back to nucleosomes.
00:06:51.09 The way nucleosomes were isolated in previous days
00:06:55.11 was by isolating chromatin from natural sources
00:07:01.03 -- from chicken blood or calf thymus --
00:07:04.10 and by cutting out nucleosomes
00:07:06.18 with micrococcal nuclease
00:07:08.05 and using those for crystallization.
00:07:10.27 Now, this posed a problem,
00:07:12.17 because by definition the DNA sequence of every single nucleosome
00:07:15.03 would be different,
00:07:16.21 because we sampled the entire genome.
00:07:18.15 The DNA length would also be subtly different
00:07:21.18 because the enzyme isn't that accurate
00:07:24.02 in chewing back all the way.
00:07:26.01 And the histone components, the protein components,
00:07:27.27 are also highly variable
00:07:30.05 because they exist in many flavors.
00:07:32.27 These many flavors have very, very important biological implications,
00:07:36.07 but as far as crystallography goes,
00:07:39.08 they really represent a big problem
00:07:42.15 because they preclude the formation of highly diffracting crystals
00:07:46.14 because the material is just simply too heterogeneous.
00:07:50.08 And so, this could be...
00:07:52.12 this should come as no surprise,
00:07:54.09 that the first efforts to determine the 7-Ang...
00:07:56.23 to determine the structure of the nucleosome
00:07:58.16 were limited to 7 Angstroms.
00:08:00.02 So, this would be the equivalent of a mitten.
00:08:02.29 And we really couldn't see a lot of detail from these structures,
00:08:06.16 but we could discern that the DNA
00:08:09.09 was wrapped around the outside of the protein.
00:08:11.12 The protein constituted a dense core.
00:08:15.01 And we could discern the super helical path of the DNA
00:08:19.11 around the outside of the cylinder.
00:08:21.21 Details were not distinguishable.
00:08:23.18 We... this structure did not allow us
00:08:27.01 to determine how the histones hold to the DNA
00:08:29.18 and how they afford this quite profound distortion.
00:08:34.00 I should also point out that
00:08:37.05 these low resolution structure,
00:08:38.25 more than anything else,
00:08:40.11 rely on orthogonal methods to verify the conclusions
00:08:42.09 that we draw from them.
00:08:43.27 And in this particular case,
00:08:45.08 very beautiful cross-linking data
00:08:47.10 from the Mirzabekov lab
00:08:49.10 really helped us to verify that this structure was true.
00:08:55.09 Now, how do we get to higher resolution?
00:08:58.03 How do we get from these diffraction pattern
00:09:01.00 that have very...
00:09:03.10 only very few diffraction spots
00:09:05.14 to these beautiful Starry Night-like-looking diffraction patterns,
00:09:08.26 where we have lots of information,
00:09:11.06 and as we...
00:09:13.06 as we say, they diffract to very high resolution?
00:09:15.12 There’s three fundamental improvements that had to be made.
00:09:19.06 First and most importantly, sample homogeneity.
00:09:22.09 So, we had to convince our nucleosomes
00:09:24.10 all to be happy and smiley,
00:09:25.25 and this proved to be the most time-consuming
00:09:28.07 and most difficult aspect of this project.
00:09:31.01 The second was that we had to find
00:09:33.11 a way to determine the phases,
00:09:35.07 and that was not a trivial matter,
00:09:37.19 and this is not something I will go into in great detail.
00:09:40.10 And third, we needed access to stronger X-ray beams,
00:09:42.26 and it was very fortuitous that at that time,
00:09:46.15 third generation synchrotrons were developed
00:09:49.14 that allowed us...
00:09:51.02 that produced very hard X-rays
00:09:55.00 that were required for this project.
00:09:58.01 So, how do we get to greater sample homogeneity?
00:10:01.22 Obviously, cutting nucleosomes out
00:10:04.26 from native tissues did not work,
00:10:06.17 so we had to individually express
00:10:09.14 every single histone in bacteria.
00:10:11.06 We then had to refold those histones
00:10:12.28 to form what we call a histone octamer,
00:10:15.11 which is the protein component of the nucleosome.
00:10:17.28 We then had to find a DNA sequence
00:10:20.02 that was uniquely suited to position nucleosomes
00:10:24.13 and to form nucleosomes.
00:10:25.26 And we had to prepare vast amounts of that DNA,
00:10:29.18 and that was, in and of itself, a major undertaking,
00:10:32.11 and involved a lot of phenol extraction...
00:10:35.14 not a lot of fun, I should say.
00:10:38.15 You then combine those two components,
00:10:41.26 and I had to develop methods
00:10:44.18 to form the nucleosomes in vitro.
00:10:47.23 Once I've done that,
00:10:50.03 I had to convince myself and others
00:10:52.03 that these nucleosomes that I made from scratch,
00:10:54.12 from bottom up, from their basic components,
00:10:56.08 are exactly the same --
00:11:00.18 just better than the native nucleosomes
00:11:02.11 that we would isolate from tissues.
00:11:04.06 And that required quite a long...
00:11:07.14 a long effort in...
00:11:09.19 efforts in biochemistry and other approaches.
00:11:12.02 But finally, we had nucleosomes that had...
00:11:15.02 were unique with DNA sequence and length.
00:11:17.10 They had highly uniform histones.
00:11:19.22 And furthermore, this approach allowed us to introduce mutants into the histones,
00:11:23.22 and this proved to be essential for our efforts
00:11:26.09 to determine the phase problem.
00:11:29.01 So, all is well. Everybody's happy.
00:11:31.08 We should be well on our way to determining
00:11:33.15 a high resolution structure,
00:11:35.05 but of course this was not so.
00:11:37.01 As science goes, two steps forward, three steps back.
00:11:39.15 The crystals that we obtained
00:11:41.21 from this allegedly much better material
00:11:43.24 were actually much worse.
00:11:45.13 They diffracted badly.
00:11:47.07 Sometimes they didn't crystallize at all.
00:11:48.22 We got thin window panes
00:11:51.04 that were really not easy to mount at all.
00:11:52.21 And so, it seemed like all our efforts were for naught.
00:11:56.03 And so, this was really a low point in this whole effort.
00:11:59.27 And I really...
00:12:02.06 we really had to dig ourselves out from this hole.
00:12:03.22 I should point out that the material that we put into these crystals,
00:12:06.28 when we analyzed it on native gel electrophoresis,
00:12:10.18 looks kind of fuzzy.
00:12:12.13 And that's just how they looked like.
00:12:14.15 But in order to find out
00:12:17.22 why these crystals all of a sudden
00:12:19.21 were so ill-behaved and didn't produce...
00:12:23.10 didn't... didn't diffract as well,
00:12:25.12 I had to figure out what was incorporated into the crystal.
00:12:28.19 Maybe something had to happen in that long time
00:12:31.06 when they refused to crystallize for them...
00:12:33.17 so that they could be converted into a form
00:12:37.20 that allowed them to crystallize.
00:12:39.11 So, one way to do this is to just take your hard-earned crystals,
00:12:41.19 and you dissolve them,
00:12:42.26 and you analyze them on a gel.
00:12:44.14 And much to my surprise,
00:12:45.26 I found that rather than being fuzzy,
00:12:47.20 like the input material,
00:12:49.29 these crystals were actually...
00:12:51.19 the contents of the crystals was actually a very nice crisp
00:12:54.24 and tight band.
00:12:56.09 So, this was really weird.
00:12:58.12 What happens?
00:12:59.23 There's two possibilities...
00:13:01.25 formal possibilities.
00:13:03.13 In one possibility,
00:13:05.19 we have a mixture of smileys and frownies,
00:13:06.28 and only the smileys make it into the crystal.
00:13:09.13 So, when I dissolve my crystal,
00:13:11.17 I can only see the nice tight band composed of smileys.
00:13:14.13 And the frownies would still be
00:13:17.14 in what we call a mother liquor.
00:13:18.25 This is the liquor surrounding our crystal.
00:13:20.18 In the second possibility,
00:13:22.07 it could be that these two forms
00:13:25.01 are interchangeable or in equilibrium.
00:13:26.20 And it could be that only our smileys
00:13:28.16 are incorporated into the crystals
00:13:30.02 and thereby shifting the equilibrium,
00:13:32.00 pulling the frownies over to the smileys.
00:13:35.06 In this case,
00:13:36.20 I would have smileys in my crystals
00:13:38.28 and nothing in the mother liquor.
00:13:40.14 So, this is a very easy hypothesis to test.
00:13:42.12 And to our delight,
00:13:44.29 we saw that there was really nothing left in the mother liquor.
00:13:47.10 And so, this really means that this hypothesis
00:13:50.03 was correct,
00:13:51.29 and now we had to find a way to convince
00:13:54.10 all of our frownies to be converted into smileys,
00:13:56.22 and that would really solve all of our problems
00:13:59.20 and allow us to move ahead in our efforts
00:14:02.03 to determine the structure.
00:14:03.27 So, what are the possible reasons
00:14:06.07 for this sample heterogeneity?
00:14:10.24 We know that our histones are uniform.
00:14:13.11 We know our DNA is uniform.
00:14:15.01 So, the only variable that was left was the position of the DNA
00:14:17.18 on the histone octamer.
00:14:19.27 And this is illustrated by this little yarn spool, here,
00:14:22.16 with the white DNA wrapped around it.
00:14:24.15 In the top example, the DNA is positioned centrally
00:14:28.15 and nice and tightly tucked in.
00:14:30.10 In the bottom example,
00:14:32.19 I tucked on the left-hand side of my yar... of my...
00:14:35.22 of my DNA and pulled it off a little bit.
00:14:38.15 This leaves a little bit more of the yarn spool exposed,
00:14:41.27 and leaves some of the DNA sticking out.
00:14:45.04 And indeed, this is what we...
00:14:47.05 what we believe has happened.
00:14:49.00 This is shown in this cartoon
00:14:51.23 underneath the gel here.
00:14:53.04 The three bands, we believe, are...
00:14:59.03 constitute the central...
00:15:01.09 centrally positioned nucleosome
00:15:03.29 that would be the most fast-moving,
00:15:05.16 because it's the most compact.
00:15:07.06 And then we'd have two nucleosome species
00:15:09.02 that have DNA hanging out either on the left side
00:15:11.11 or on the right side at different angles,
00:15:14.09 and that would change their electrophoretic behavior,
00:15:16.13 and that thus would result in those three bands.
00:15:19.19 So, how could we convince these two top species
00:15:23.12 to move to the centrally positioned species?
00:15:27.04 Now, we surmise that
00:15:29.22 the centrally positioned species
00:15:31.14 would be the thermodynamically most stable,
00:15:33.21 because most of the DNA is bound tightly,
00:15:35.17 and so this would be a more stable situation
00:15:37.20 than having DNA kind of flopping off the side.
00:15:41.04 And so, we figured that
00:15:44.22 a simple heating step could actually do the job
00:15:47.00 of converting these species into each other.
00:15:48.23 And much to our surprise,
00:15:50.24 just introducing a little bit of thermal energy
00:15:52.28 -- heating our samples at 37 degrees for half an hour --
00:15:57.24 actually did the job,
00:15:59.21 and we now got beautifully homogeneous particles
00:16:02.03 that actually gave rise to beautiful crystals.
00:16:06.03 And what was even better,
00:16:08.06 these crystals diffracted to very high resolution,
00:16:10.11 giving rise to these diffraction patterns
00:16:12.28 that still...
00:16:15.16 I get goosebumps when I see them,
00:16:16.29 just because they're so nice to look at.
00:16:20.17 Now, you think you're done, right?
00:16:22.23 But you're really not done.
00:16:24.05 Structural biology is an endurance sport,
00:16:25.24 maybe like a triathlon.
00:16:28.07 So, we've solved the first...
00:16:30.05 we solved the first problem,
00:16:32.02 the structural homogeneity.
00:16:33.18 We've obtained well-diffracting crystals.
00:16:35.17 Now, we have to solve the phase problem.
00:16:37.20 Without going into detail, the phase problem...
00:16:41.02 to solve this problem involved
00:16:44.13 me making about 25 or 30 mutants in histones,
00:16:46.16 isolating each of these mutants independently,
00:16:50.27 making 25 different flavors of nucleosomes,
00:16:53.09 and collecting data on these nucleosomes
00:16:57.18 so that we could obtain
00:17:00.06 heavy atom derivatives.
00:17:02.03 And the third hurdle that we had to overcome
00:17:04.21 was using a high-energy X-ray source.
00:17:07.19 These are very weakly diffracting crystals,
00:17:09.19 and they require what we call very hard X-rays
00:17:12.03 so that we could get enough intensity
00:17:14.10 from our diffracted crystals.
00:17:16.11 And luckily, the third generation synchrotrons
00:17:18.13 were under construction
00:17:20.20 and were becoming operational just at that time.
00:17:22.12 And synchrotrons are large particle accelerators.
00:17:26.08 They're about two kilometers in diameter.
00:17:28.12 This is the one in Grenoble, the ESRF.
00:17:31.29 We were among the first users.
00:17:34.04 And rumors have it that the first users
00:17:36.24 used their waiting time
00:17:39.09 while the samples were exposed
00:17:41.19 to play squash in empty hutches
00:17:43.24 and use the smooth concrete floors for rollerblading.
00:17:46.29 I would not know anything about these rumors,
00:17:49.19 so I will deny all...
00:17:53.04 all accounts of this.
00:17:54.16 That's all I'll say about it.
00:17:56.04 So, now remember what the...
00:17:58.08 the result that we get from diffraction experiments
00:18:00.23 is not the structure itself.
00:18:02.29 It is actually the electron density.
00:18:05.14 The electron density without a model built into it
00:18:09.14 tells us nothing.
00:18:10.26 The electron dens...
00:18:12.17 so, then I had to sit for about a year in a dark room,
00:18:14.03 painstakingly and manually
00:18:16.16 building a three-dimensional structure into this model.
00:18:19.26 Now, this was really, really difficult,
00:18:21.14 because we had no idea where to start.
00:18:24.22 We had no idea
00:18:28.20 what the secondary structure of these proteins was.
00:18:30.14 We had no idea where the DNA was.
00:18:32.28 This is the raw data that we had.
00:18:34.15 And so, this is like a big three-dimensional puzzle,
00:18:36.22 and only when all the parts
00:18:39.13 are fit in correctly,
00:18:41.10 you can account for all the electron density.
00:18:43.10 And so, this is what the final product, then, would look like.
00:18:46.09 And this required, as I said,
00:18:48.21 about one year of work
00:18:52.09 in a dark room in front of a computer monitor --
00:18:54.00 not a very social experience.
00:18:56.11 Here's a detailed...
00:18:59.25 a detailed view of the same density.
00:19:01.05 And you can see that now we can actually obtain
00:19:04.17 the orientation of the main chain and of the sidechains of the protein.
00:19:07.23 And this allowed us to really...
00:19:09.21 to describe how the DNA
00:19:14.11 is held so tightly by the histones,
00:19:16.01 how the histones interact with each other.
00:19:18.13 And this gave the...
00:19:20.18 gave rise to this iconic picture, here,
00:19:22.19 of the final nucleosome structure in two different orientations
00:19:25.21 that now graces mouse pads and textbooks and T-shirts.
00:19:31.19 And it's just generally very, very pretty to look at.
00:19:36.09 Like all scientific endeavors,
00:19:38.23 this particular project took very many years,
00:19:45.15 and it was a huge group effort
00:19:47.27 and couldn't have been done without my colleagues
00:19:50.02 at the ETH Zurich,
00:19:52.11 with the guidance of Tim Richmond.

This Talk

Speaker: Karolin Luger

Audience:

Student
Educators of Adv. Undergrad / Grad
Researcher
Educators

Recorded: January 2020

Talk Overview

The nucleosome is the central organizing structure of the eukaryotic genome. It consists of DNA wrapped around histone proteins. Dr. Karolin Luger shares her discovery of the three-dimensional structure of the nucleosome using X-ray crystallography, which provided a deeper understanding of chromatin organization.

Speaker Bio

Karolin Luger

Karolin Luger earned her bachelor’s degree in microbiology and her master’s degree in biochemistry from the University of Innsbruck. She completed her PhD in biochemistry and biophysics at the University of Basel. In 1997, she published the first high-resolution 3D structure of the nucleosome, determined using X-ray crystallography. The Luger Lab uses structural biology to… Continue Reading