Tip:
Highlight text to annotate it
X
Hello. My name is Tony Hyman. I'm the director of Max Planck Institute in Dresden, Germany.
On the background, you see a movie of a C. elegans embryo
going into mitosis.
And, one of the things that strikes one, of course, when one looks at a movie like this,
is that the spindle itself has two poles.
You have a pole here, in red, and you have another pole over here.
And that's, of course, at the heart of all cell division,
because the chromosomes, when they decide, go to these different poles.
You have to have two poles in order for the chromosomes to segregate into 2 masses.
So, you can ask the question then, why are there two poles?
This is a question that's interested biologists for more than 100 years.
This is a picture from the work of Bovary, the great 19th century German cytologist.
And, this is from a book by E.B. Wilson, called the Cell in Development and Heredity,
where he summarized a lot of this knowledge
that was discovered around the turn of the 19th and 20th century.
And Bovary was also fascinated by this problem,
where you could see that the spindle always has 2 poles,
and how is this bipolarity set up?
What I'd like to talk to you about in this segment of this talk,
is the construction of a very complex protein complex
called a centriole.
Over here, you can see that centrioles are quite large on our scale.
Here we have our tubulin molecules, when we were making microtubules.
And centrioles are again another order of magnitude bigger, and therefore, complex,
in terms of thinking of their organization.
And, at least in some systems,
it's the way the centriole duplicates which defines the fact
that there are two poles of a mitotic spindle.
Bovary was also interested in this problem,
when he stained mitotic spindles with different dyes,
he didn't have access to fluorescence in those days,
but he was still able to see lots of substructure,
and you can see in this particular picture, which was actually taken by Joe Gall,
from an original microscope slide of Bovary,
you can actually see that he could see this little structure in the center of the centrosome,
which we now know is likely to be the centriole.
The centriole is in the middle of the centrosome.
The centrosome, known as the pericentriolar material,
surrounds this centriole, which tends to exist as a pair of linked centrioles,
which tend to be orthogonal to each other.
One of the questions that's always been interesting in that field is
how do centrioles grow?
It's fascinating that once per cell cycle,
each centriole makes a duplicated daughter centriole.
Just like DNA, you also make one copy of each DNA strand.
The same is true for centrioles, and that's therefore interested people very much
over the years.
And the work of, really, three people sorted this problem out
in C. elegans embryos, at a morphological level:
Thomas Mueller-Reichert and Eileen O'Toole, two electron microscopists,
and Laurence Pelletier, he was a cell biologist.
And they decided to attack this problem in C. elegans embryos.
Now, the problem with studying centrioles are
they're extremely low abundance -- there's only about 2 per cell.
If you take ribosomes, there are many 1000s of ribosomes per cell.
The biochemistry is extremely difficult.
And, they also change through the cell cycle.
So, most large complexes so far studied,
by structure, have normally been isolated biochemically - proteasomes or ribosomes.
So, how are we going to study this problem of centriole growth?
Well, let's ask a simple question first.
What I've done here, and you're going to see this through this talk,
is I've outlined the timeline of the cell biology of C. elegans on this axis.
And you can see the different events here, and you can see the time on this bottom axis here.
So, during this process, we can ask the question, when do centrioles duplicate?
Now, centrioles are small, and we can't see them by light microscopy.
We certainly can't see... well, we can see individual centrioles,
but we can't see very easily if there are 1 or 2.
And so, in order to do this, you really need to use electron microscopy.
So, then you want to ask the question,
during this time, when is it that centrioles actually duplicate?
And, in order to do that, you need to say, alright, here I am
at this time point X. What I want to do is look at the centrioles by electron microscopy,
and that's what we know as correlative light microscopy and electron microscopy.
We use light microscopy in a living organism
to get the timing of the system,
and then we have to go by electron microscopy
and look at that system and ask, what did the actual centrioles look like?
And a way you can do that is by using fixation.
But, we need to be able to fix at certain timepoints.
And we can do that, due to a nice little trick in C. elegans,
which is the embryos have an egg shell.
Now, this egg shell has been beautifully evolved over many millions of years
to keep everything out. These embryos exist in soil,
as far as we know, and they have to be able to resist any outside insults.
But, you can actually penetrate the egg shell with a laser.
You can take a laser beam, in a very space age experiment,
you can shine it on the eggshell and pop a little hole.
So, you can do this nice experiment
where you can surround the embryo in glutaraldehyde,
which is a fixative, and it doesn't go across the egg shell,
because it's such an amazing structure.
Then you pop a hole in the eggshell, the glutaraldehyde goes in,
and fixes the embryos. Let's look at that in this movie.
What you're going to see is the embryo move a little bit --
that's where you pop it with a laser,
and you'll see it fix. So, here it goes.
Pop! You see it's fixed.
And, pop! The other one's fixed.
And did you notice how, when we fixed it,
all the movement stopped. It's a very quick process, fixation.
Glutaraldehyde is a very small molecule, goes in, fixes it, then you can process the embryos
for electron microscopy
by serial sectioning.
And, what that experiment showed is that
centrioles are unduplicated about here.
And, if you look a few minutes later, they've now duplicated.
So, by metaphase, they've actually duplicated.
So, that's a very fast process, the centrioles have gone from unduplicated to duplicated.
And so we can conclude, then, that there's a duplication process that happens
early on in the cell cycle of C. elegans.
But then you can ask the question,
well, how do they duplicate?
And the technique we were using then--glutaraldehyde fixation--
was not good enough to tell us how the centriole is itself being made.
We were able to see unduplicated centrioles, and we could see these nicely formed centrioles,
but we weren't able to pick up the different stages of centriole duplication.
How do they form?
They're very complex structures, and that's the question we wanted to ask.
Now, in order to do that, we had to move to a different kind of technique,
which is known as electron tomography.
In order to do tomography, we needed to be able to come back and stop the embryos,
but we needed to be able to stop them by freezing.
So, one way that we preserve structures in biology
without disturbing their ultrastructure is by freezing... very fast freezing
preserves biological components without disturbing them as much in the ultrastructure
as does the fixation.
And what you do, is you freeze at very high pressure and then,
high pressure prevents the formation of ice crystals,
and then you can infiltrate the fixation at very low temperatures,
and that's known as high pressure freezing
and that's a way to preserve the ultrastructure of the system.
So, what we needed was a way where we could freeze the system,
in a time-resolved manner.
And so, we came up with a particular way of doing that,
using little tubes that you use for kidney dialysis,
you can suck embryos into them, you can follow the development
of the embryos under the microscope,
and then you freeze them in the high pressure freezing machine,
and then you process them for tomography.
Now, the problem was, when we started this experiment,
it wasn't easy to actually freeze them
at the time that one's interested in.
So, we used a new machine, developed by Leica
which allowed us to do time-resolved tomography,
and I'm going to show you this machine in action here.
What we've done is we've taken the embryos, we've put them into a tube.
We found out they're just at the right size, and at the right stage.
We put them into the high pressure freezer, and now we're going to freeze them.
So here it goes.
We're going to bring our hand in, and we're going to push it in to the freezer,
and then POOF, it's now frozen.
So, we rapidly freeze the embryos.
Now we can take them, and we can process them for tomography.
Now the key thing about tomography is that you look at very thick sections.
So, normally, in a standard electron microscopy, you look at 50 nanometers,
but you can look at 300nm sections, and you get a 3D picture of the way it looked.
I won't go into that in detail in this talk. You can find it elsewhere if you're interested.
But it's a way of looking at a 3D picture by electron microscopy.
So, we can look at centrioles at metaphase, and we can see how beautifully they look.
You can see over here, for instance, a very nice picture of a centriole
with microtubules around the outside.
And then we can see one over here.
So, what we want to do is look at these in these tomographic sections.
And, as I mentioned, one of the main tools for a cell biologist to link phenotype to structure
is electron tomography.
It's a way that we can actually go in and get at high structural resolution the way things look.
An in situ phenotype.
So, the problem is that we do genetics and we get a phenotype,
we want to know how that's changed the ultrastructural level,
we generally can't isolate them from the cell and look at them.
Rather, we have to look at them in situ.
So, what I'm going to show you now is an electron tomogram
of 2 centrosomes and centrioles early on, after duplication.
So, what we're going to... We're stepping through the section.
You're going to see, we're looking at one centriole
with its microtubules, then we're going to step through further.
Then we're going to come to the other centriole pair. They're about a micron apart.
And you can see what we've done there, in that tomogram,
we've reconstructed both the distribution of microtubules
and the centrioles. And you can see there's a duplicated centriole pair at each spindle pole.
So, then, we said, now we're going to go back in high resolution,
and we're going to try to understand what are the intermediates
in making centrioles using our techniques.
And what we learned there was quite fascinating.
We learned that the initial step in centriole formation was formation of a central tube.
And you can see that here, by tomography,
and a cartoon next to it. This little tube is forming next to this centriole.
It doesn't have any microtubules around the outside yet. It's just a naked tube.
What happened then next was the tube elongated.
So, it's the growth of this tube from what's known as the mother centriole,
And so that's what we've learned so far, is that the centrioles duplicate by...
They separate into two individual components,
and then the daughter centrioles grow from the mother centrioles by elongation of this tube,
and then the next stage was very fascinating,
because we found that the microtubules then associate around the tube.
But, what we found was that the microtubules...
eventually there are going to be 9 microtubules all the way around the tube.
But, in intermediates in centriole formation, there are fewer microtubules.
So, in this case, there are 7 microtubules.
And also, they have intermediate lengths.
And you can see over here these hooks that we found around the inner tube
that seem to define in some way the nine-ness of the tube.
If you look at a number of different centrioles, you can see intermediate products,
so that somehow the microtubules are binding to the tube and forming this 9-fold symmetry.
Here's a cartoon of the process.
You can see the tube elongating, and the microtubules binding from the outside
growing, and forming the microtubules around the outside of the tube.
Now, we've made this cartoon with reconstructions, of course, of fixed material.
We haven't seen the microtubules growing,
but we've inferred it by looking at many different particular specimens.
So, what we learn from that is that centriole assembly
proceeds through structural intermediates.
You have a tube. The tube grows over about 8 minutes,
microtubules associate with the tube over about 2 minutes,
and the mother centriole then matures during this process. I didn't discuss that
in the tomograms, but the mother centriole changes slightly during this process.
So, what was exciting about that discovery was we'd shown that centrioles
have almost a virus-like assembly, where they have structural intermediates
you can define by looking at the growth of the centriole itself.
But what we wanted to do next then, was to say, now what we want to do
is find the genes required for that process.
That's the morphology... what is the genetics?
What are the genes required for that particular process?
And, that turned out to be something which was relatively straightforward
using our RNAi screen, because of the work of my PhD supervisor, John White,
and a post-doc in his lab, Kevin O'Connell.
And to do that, you have to understand a little bit about the biology of a C. elegans embryo.
Have a look at a wild-type. You see the blue centriole pair.
They come in with the ***. Now, they then separate, and each pole gets one centriole.
But, the centriole is duplicated, so you have a centriole pair at each pole.
Do you see that? The blue centriole... the *** has brought in its centriole pair.
It's separated, so you look at each pole,
and you'll see that it has one blue centriole from the ***,
and the orange one represents the duplicated centriole that duplicated
during the process of preparing a spindle, as I showed you in the early part of the talk.
And then you look at the two-cell stage... the same thing happens again.
Let's see what happens if you prevent duplication of the centriole.
What happens when you do that is a very interesting phenotype
because the *** brings in a centriole pair, the RNA interference
for reasons we don't really understand doesn't work very well in the ***
So, that's unaffected. And then, the centrioles separate, and one goes to each pole.
It turns out, you don't need duplication to form the pole.
So, if you don't duplicate your centriole, it doesn't affect the mitosis.
But, the problem comes in the next cell division,
because then each cell only has one centriole, not two,
and now it only makes a monopolar spindle.
So, normally instead of bipolar, it just makes a monopolar spindle.
And I'm going to show you some movies of that.
So, here is a centrosome duplicating at the end of
cell division, and you can see it moving into two different centrosomes.
It's duplicated, and at the two-cell stage, you have two centrosomes,
and you've got the chromosomes, which I've shown you there as well.
So that movie has both labeled centrosomes and labeled chromosomes.
Now, let's see what happens when centriole duplication is failed.
Well, everything is looking fine at this point.
We've made a spindle, it's all divided.
But what happens to the spindles at the two-cell stage
is these beautiful little half spindles will form without a second pole.
A really gorgeous phenotype.
I can't stop looking a those... they're just so beautiful.
And, in fact, what we then did was to go back to our RNAi screen and say
how many genes are required for centriole duplication?
You can take the 800 genes required for cell division.
You can rescreen them by fluorescent microscopy,
and you can look for ones that have that phenotype.
And from that screen, it turned out that we now know there are 5 genes
required for daughter centriole duplication.
So, in the end, also quite simple, there's not many proteins required.
You would think, wow, that's a complex process.
Doesn't that require a lot of genes? But, no!
It seems like these 5, as far as we can tell, seem to be sufficient.
Now, the analysis of these genes and a detailed characterization
was published in a number of different labs,
and I've illustrated some of the papers over here.
And what all of those studies showed was the same thing.
If you remove the function of any of these genes, you get a monopolar spindle,
as I've shown over there in the fluorescence.
And if you then do electron microscopy, you then prevent centriole duplication.
So, that's quite interesting.
We've identified a set of genes that we know are required for centriole duplication.
But, always when you do a study like this, you have the same problem,
which is, how are the proteins themselves related to the structure of the process?
We've done two different experiments now.
I've showed you the structural experiments where we've shown how centrioles duplicate.
I've shown you the genetics which shows you how
we identify the proteins involved in that process.
But how are we going to link the proteins to the structure?
What aspect and which proteins are required for which aspect of building this structure?
So then we linked the two of them together,
and that's what's so nice about doing this time-resolved tomography inside the embryo
is we can now go back and look at the mutant phenotypes by tomography
and ask how does that affect the duplication?
And when we do that, what we find is the following.
Here I've laid out again a timeline of duplication,
and also I've put at the top the proteins.
And as a little hierarchy of organization where there are two proteins Spd-2 and Zyg-1
which are required for all the other proteins to go onto the centriole.
Now, if you remove Sas-6 or Sas-5 from the cell, and then you do electron tomography,
you find no duplication either.
So, that suggests that Sas-5 and Sas-6 are probably required for forming the central tube.
But, Sas-4 was more interesting in its electron microscopy phenotype
because when we removed Sas-4, you still formed a tube, but
you don't form any microtubules around the outside of the tube.
So that tells us then that Sas-4, in some way,
is required for form the microtubules around the tube.
Let me just show you one tomogram.
of formation in Sas-4 RNAi embryos.
You can see that the mother is fine,
but the daughter only has a tube with no microtubules around it.
Can you see that here? That little purple...
The green is the mother, and the purple is the daughter.
So, that's what we conclude then from this study, which is that
the set of proteins forms onto the forming centriole,
and then we can show that Sas-5 and Sas-6 are apparently required
for forming the central tube, and Sas-4 is required
for forming the microtubules around the outside of the tube.
So, in that study, what I've tried to show you
is another very complex, intricate protein complex
forming from the arrangement of different molecules.
It forms an interesting structure, which is a different one from microtubules.
Microtubules are polymers.
This one seems to be a more virus-like assembly,
with steps of assembly process that we can isolate
And we can also find the genes required for it.
And we can show, in outline, how they're required for different aspects of centriole formation.
And the next stage, of course, will be to do more detailed structural work
to try and understand how the individual proteins affect, for instance,
the formation of the tube itself.
So, centrioles then, we believe now, form by a sort of virus-like mechanism
with steps in the assembly process.
And coming back to our scale, you can see that we've gone out
quite a few orders of magnitude now,
from our initial tubulin molecule, so we're actually looking at fairly complex structures,
which are a couple of orders of magnitude bigger than the molecules that make them up.
And so, you can see that slowly, we're putting the cell into subcompartments of organization.
We're not working on individual proteins, but they make these very complex structures.
Some of them more machine-like, say ribosomes, which make protein,
but other ones are more complex-like... polymers or like centrioles.
And by thinking about how these things are put together,
it helps us to understand the organization of the cell.
I'd like to thank... finish, by just... of course, the genomics itself is a very, very
time-consuming process involving many different people.
But, some of the key players are mentioned here.
As well as those involved in the centriole assembly.