How to Create a Body Axis

Transcript of Part 2: How to Polarize the Cytoplasm

00:00:13.09	Hello.
00:00:14.09	My name is Geraldine Seydoux.
00:00:15.09	I am a professor at Johns Hopkins University and an
00:00:18.23	Investigator with the Howard Hughes Medical Institute.
00:00:21.14	And this is the second part of my presentation on how embryos elaborate body axes and how
00:00:29.21	the single-cell zygote becomes polarized.
00:00:33.16	So, in this second part, I'm gonna focus on how the cytoplasm of the zygote becomes polarized.
00:00:40.25	So, in the first part, I described how the sperm in Caenorhabditis elegans polarizes
00:00:49.08	the PAR proteins at the membrane of the one-cell embryo.
00:00:54.13	It turns out that the PAR domains are controlling all of the different aspects of the polarity
00:01:02.24	of the one-cell embryo.
00:01:04.08	And that means that these proteins at the membrane direct what's going on in the cytoplasm.
00:01:09.12	And in the cytoplasm there are several proteins and organelles that become asymmetrically
00:01:16.18	distributed along the anterior-posterior axis.
00:01:19.22	And this happens because you want different molecules to end up in the anterior and  posterior cell
00:01:27.13	that is formed at the first division.
00:01:30.05	So, in this presentation we're gonna focus on how do you generate these cytoplasmic asymmetries and,
00:01:37.08	in particular, we're gonna talk about how this protein here, MEX-5, forms an anterior-posterior
00:01:44.12	gradient in the cytoplasm.
00:01:46.07	So, here's a little movie showing you, again, how the PAR proteins become asymmetrically
00:01:52.16	segregated and, down below, how this MEX-5 protein becomes enriched on the anterior side.
00:02:00.12	Okay?
00:02:01.12	So, MEX-5 is an RNA binding protein, it's present throughout the cytoplasm, and,
00:02:08.28	as you can see, as the PAR proteins become asymmetrically segregated, MEX-5 responds and becomes enriched
00:02:18.25	on the anterior side.
00:02:20.12	Okay.
00:02:21.12	So, how does that work?
00:02:23.14	This was the project of a postdoc in my lab, Erik Griffin, and Erik was interested in figuring out,
00:02:32.27	how do you make this MEX-5 gradient?
00:02:37.02	And at first he considered three possibilities.
00:02:40.10	So, one idea, given that we can see that MEX-5 starts out uniformly distributed and then
00:02:46.16	becomes asymmetric, one possibility is that you just make more MEX-5 in the anterior side.
00:02:53.11	So, you could translate more MEX-5 protein specifically in the anterior cytoplasm and
00:02:59.04	that would give you a gradient.
00:03:00.17	That's one possibility.
00:03:01.26	Another possibility is that, in fact, you're doing the opposite.
00:03:05.18	You're actually degrading MEX-5 in the posterior cytoplasm, okay?
00:03:13.10	Another possibility is that you're doing neither one of those things, but instead you're just
00:03:18.02	taking the protein that was in the posterior and moving it towards the anterior side.
00:03:23.17	Okay?
00:03:24.17	So, you're redistributing... umm... redistributing existing protein.
00:03:28.25	So, to distinguish between these different models, Erik realized that these different models
00:03:35.20	actually make different predictions as to what happens to overall MEX-5 levels.
00:03:42.10	So, in the first case scenario, MEX-5 levels would increase, whereas if MEX-5 is being
00:03:52.24	degraded then the overall level should decrease.
00:03:56.04	But if you're just moving protein around, the MEX-5 levels should stay constant.
00:04:01.26	So, Erik had a feeling that he should really try to be able to study, carefully, what happens
00:04:08.20	to MEX-5 levels during this polarization.
00:04:13.02	He also realized that it would be very useful to be able to distinguish between the protein
00:04:18.16	that's already present at the beginning, before polarization, and the protein that might be
00:04:24.02	made during the polarization process.
00:04:26.25	So, to accomplish both of these goals, Erik created this umm... fusion version of MEX-5
00:04:36.15	where MEX-5 is fused to a fluorescent protein called Dendra that has very interesting capabilities.
00:04:45.24	This is a protein that, when it first folds, forms a fluorescent protein that fluoresces
00:04:53.01	green, okay?
00:04:55.03	But if you expose this protein to UV light, now it fluoresces in a different color, in
00:05:03.08	red, okay?
00:05:05.00	And so this is a way to basically label protein at the time that you expose the embryo to
00:05:11.20	UV light.
00:05:13.09	Whatever protein is there is now going to become red, okay?
00:05:16.27	Any new protein that's made later will still be green.
00:05:21.06	So, if you follow the embryos, just looking at the red fluorescence, you can look at protein
00:05:27.25	that was existing at the time that you started the experiment.
00:05:31.20	Okay?
00:05:32.20	So, this is called a photoactivatable fluorescent protein.
00:05:36.08	And it's very useful to follow proteins over time.
00:05:42.04	So, when he did this experiment, just exposing the whole embryo to fluorescent... to UV light,
00:05:50.23	he saw that this red MEX-5 protein was able to redistribute into a gradient during the
00:05:59.07	polarization process of the... of the zygote.
00:06:04.13	And so, from this, he concluded that you don't need any new synthesis -- whatever protein
00:06:09.04	is already there, it knows where to go.
00:06:11.17	So, that was interesting.
00:06:13.13	Next, he measured very carefully the amount of protein that existed at the beginning,
00:06:20.17	before polarization, and at the end, after polarization.
00:06:25.00	And what he found is that, basically, nothing changes.
00:06:28.05	The total amount of protein doesn't change.
00:06:32.01	What changes is that the amount in the anterior goes up and the amount in the posterior
00:06:36.01	goes down.
00:06:37.09	And that is consistent with some kind of redistribution, okay?
00:06:41.24	And then another experiment that he did that that confirmed that, in fact, the protein
00:06:45.27	was moving around somehow is this one, where he labeled... instead of labeling all of the
00:06:53.07	protein, as was done in this first experiment, over here, he labeled only the protein that
00:06:57.26	was on the posterior side at the beginning of the experiment.
00:07:01.19	And then he watched, what happens to this protein over time?
00:07:04.02	And he found that, yes, it did accumulate on the anterior side.
00:07:09.04	So, somehow, protein is moving around in the embryo and knowing to go to the anterior side.
00:07:16.08	So, how could this happen?
00:07:19.14	So, this really suggested that this middle option is the right one.
00:07:26.26	How is this happening?
00:07:28.07	So, then Erik decided to do another experiment, where, instead of illuminating large parts
00:07:36.00	of the embryo with the UV, he decided to illuminate the embryo in just two areas:
00:07:44.02	one stripe over here and another stripe over here.
00:07:48.13	Okay?
00:07:49.13	So, these two areas were illuminated at the same time, and then you can see what happens
00:07:55.01	to the protein in these two stripes over time.
00:07:58.01	So, over just six seconds, that's what I'm showing you here.
00:08:02.01	And so... it's a little bit hard to... to really understand what's happening when you
00:08:05.14	just look at the embryos like this, so what he did is he created a kymograph from these pictures.
00:08:13.04	And a kymograph is basically just taking one slice, one line across the embryo and...
00:08:22.02	as shown here, and then showing how that line changes over time, here.
00:08:26.13	Okay?
00:08:27.13	So, he could see what happens to the fluorescent protein that's very high in the middle of
00:08:33.04	that line at the beginning of the experiment, and what happens to it over time.
00:08:37.08	So, each of these little lit-up dots represent protein that's now diffusing away from the
00:08:46.00	original spot where that protein was when it was first activated, photoactivated to
00:08:52.03	fluoresce in the... in the red channel.
00:08:54.04	So, what you might notice is that protein is diffusing.
00:08:59.01	And it's diffusing in both directions, in both... towards the anterior, which would
00:09:04.05	be over here, and towards the posterior.
00:09:06.17	So, there doesn't appear to have any directed movement.
00:09:10.23	The protein molecules are just diffusing randomly.
00:09:14.15	Okay... so that was a little puzzling, because we know that, overall, there's more protein
00:09:20.15	ending up in the anterior.
00:09:21.24	So, how can that be if the protein is just moving around randomly?
00:09:26.27	Another thing that you might notice is that these two areas look different from one another.
00:09:32.20	This one stays brighter in this area, in the middle, whereas this one is getting diffuse faster.
00:09:40.01	Okay?
00:09:41.01	And that is not because the laser was different in those two areas -- the laser photoactivated
00:09:45.26	the same amount of protein in those two areas -- but what happened is that, by the very
00:09:51.05	first frame, you can see that the protein in this area is diffusing faster out of the
00:09:58.03	original photoactivated area, whereas the protein in this area is staying around in
00:10:03.17	this area longer.
00:10:04.24	So, that tells us that the rate of diffusion of the protein is different in these different
00:10:11.18	parts of the cytoplasm.
00:10:13.03	So, Erik then decided to redo this experiment by sampling many, many different points along
00:10:22.02	the anterior-posterior axis.
00:10:23.23	So, in this graph, the whole length of the embryo is shown down here, okay, from the
00:10:31.20	anterior over here all the way to the posterior.
00:10:36.03	And at each of these points, Erik measured how fast the protein diffuses.
00:10:41.26	Okay?
00:10:43.01	And you can see, in the blue line here, before polarization, the protein is diffusing very slowly --
00:10:51.02	less than one micron squared per second.
00:10:53.20	Okay.
00:10:54.20	So, it seems to be maybe bound to something.
00:10:57.04	It's not moving very fast.
00:10:59.08	But during the polarization process, something very interesting happened -- the protein speeds up.
00:11:05.06	But it speeds up only in the posterior cytoplasm.
00:11:08.13	So, just to summarize what I've shown you here, we start out at the beginning of the process,
00:11:15.01	before polarization, with a MEX-5 protein that's very sluggish in the cytoplasm.
00:11:21.02	And then, during the polarization process, MEX-5 becomes fast, but only on one side of
00:11:27.25	the cytoplasm.
00:11:29.03	Okay?
00:11:30.03	And this happens at the same time that we see this concentration of protein in the anterior.
00:11:36.12	Okay.
00:11:37.22	So, we knew of another little tidbit.
00:11:41.22	And that is that we knew that PAR-1 is very important to create this MEX-5 gradient.
00:11:47.24	And this was work from Jim Priess' lab, who showed that PAR-1... remember, this is one
00:11:53.18	of the kinases that are part of the PAR group of polarity regulators.
00:11:59.11	And PAR-1 is enriched in the posterior side of the embryo.
00:12:04.09	And what Jim Priess' lab had shown is that PAR-1 phosphorylates MEX-5.
00:12:09.09	And this is important for MEX-5 to become asymmetric.
00:12:12.20	So, we wondered, what exactly is PAR-1 doing to MEX-5 diffusion in the embryo?
00:12:21.00	So, here's an experiment where Erik measured the diffusion rate of MEX-5 in both the anterior
00:12:28.03	and the posterior cytoplasm.
00:12:30.13	So, in wild-type, you get two different values, because it's slower in the anterior
00:12:35.26	and faster in the posterior.
00:12:37.25	And then, what happens if you get rid of PAR-1?
00:12:40.16	You can get rid of PAR-1 by getting rid of it using an RNAi treatment or by using a nice
00:12:47.24	PAR-1 allele that was generated and characterized by Ken Kemphues.
00:12:53.13	In both of these cases, we now see that PAR-1 stays...
00:12:56.22	I mean, sorry, that MEX-5 stays very sluggish.
00:13:01.08	So, if PAR-1 is not around to phosphorylate MEX-5, MEX-5 stays slow in both the anterior
00:13:07.26	and posterior cytoplasm.
00:13:09.06	Okay.
00:13:10.06	So, that suggests that PAR-1 is somehow required to speed up MEX-5.
00:13:15.06	And Erik got a very nice confirmation of this result using a different PAR-1 allele,
00:13:22.00	this b274 allele.
00:13:24.24	This is a PAR-1 allele that actually is a premature stop codon in the par-1 gene, and
00:13:31.20	it creates a truncated PAR-1 protein that, now, is not able to attach to the membrane,
00:13:37.04	and this PAR-1 protein is uniformly distributed throughout.
00:13:41.03	So, now you have the kinase everywhere.
00:13:43.28	And when you do that, amazingly, what you see is that MEX-5 becomes fast everywhere,
00:13:50.19	both in the anterior and in the posterior cytoplasm.
00:13:54.05	So, this kind of experiment said that PAR-1 is both necessary and sufficient to
00:14:01.05	speed up the diffusion of MEX-5 in the cytoplasm.
00:14:05.15	Alright.
00:14:06.17	The next experiment is we wondered whether this phosphorylation of MEX-5 by PAR-1 is...
00:14:13.28	might be reversible.
00:14:15.19	Could it be that MEX-5 is actually cycling between being phosphorylated and unphosphorylated?
00:14:20.08	And so, to test this idea, we took MEX-5 and PAR-1 kinase in a test tube, put them...
00:14:28.16	the two together and looked at MEX-5 phosphorylation using autoradiography, as shown here.
00:14:35.21	So, the dark signal here shows you that MEX-5 has been phosphorylated by PAR-1.
00:14:41.02	So, after Erik did that experiment, he added to the MEX-5/PAR-1 mixture embryonic extract,
00:14:50.08	just cytoplasmic extract from C elegans embryos.
00:14:54.06	And then he saw a really remarkable effect of the embryonic extracts, that...
00:14:59.27	that over time this embryonic extract was able to take away the phosphorylation from MEX-5.
00:15:08.10	And so, it seems that this phosphorylation that PAR-1 does to MEX-5 is actually reversible.
00:15:15.02	It's short-lived, it doesn't last for very long.
00:15:17.23	And, in fact, Erik was able to discover the phosphatase that is responsible for removing
00:15:23.12	this phosphate.
00:15:25.00	And it's present throughout the cytoplasm in the zygote.
00:15:28.07	Okay, so PAR-1 phosphorylates MEX-5 but MEX-5 has a way to quickly get rid of this phosphorylation.
00:15:34.22	So, keep that in mind.
00:15:35.22	The next interesting observation that Erik made is he looked at the size of the complexes
00:15:44.24	that MEX-5 existed in in the cytoplasm.
00:15:47.27	He did this simply by running a whole-worm extract, with MEX-5 labeled here with Dendra,
00:15:57.17	so we can see where it runs in the extract.
00:16:00.20	And he passed this extract over a sucrose gradient so as to separate light complexes
00:16:08.18	from heavy complexes.
00:16:10.28	Okay?
00:16:12.03	And what he saw is that MEX-5 actually exists in both light complexes and heavy complexes.
00:16:20.17	So that suggested to us that, maybe, that's how MEX-5 is slow -- maybe it's slow when
00:16:26.26	it's big and it's fast when it's in those smaller complexes.
00:16:32.04	So, putting these observations together led us to this hypothesis for how a MEX-5 gradient
00:16:41.04	might form under the influence of this PAR-1 kinase.
00:16:45.24	We imagined that, at the beginning of the phos... of the polarization process,
00:16:51.17	MEX-5 exists in these sluggish, large complexes that cannot move very fast into the cytoplasm.
00:16:58.20	They might actually be tethered to something in the cytoplasm,
00:17:02.24	and so that keeps them in place.
00:17:05.22	When PAR-1 becomes asymmetric, and it is enriched on one side of the embryo, it can phosphorylate
00:17:15.06	MEX-5, preferentially, on that side.
00:17:17.21	And we imagine that, when MEX-5 becomes phosphorylated, it now breaks away from these large complexes
00:17:25.00	and exists in smaller complexes, which are faster diffusing.
00:17:29.24	Now these smaller complexes can diffuse everywhere, in all directions, so they do so, all throughout
00:17:36.24	the cytoplasm.
00:17:38.03	But if... remember that the phosphorylation by PAR-1 is short-lived.
00:17:44.18	There is a phosphatase present throughout the cytoplasm that can remove this phosphorylation
00:17:50.14	from MEX-5.
00:17:52.02	And when that happens, MEX-5 is going to return into these larger complexes.
00:17:58.14	And if this dephosphorylation event happens in this part of the cytoplasm, where there's
00:18:04.11	no PAR-1, then the larger complexes will stay there longer.
00:18:11.05	And so, just following this kind of thinking, you can imagine how, by creating this diffusion gradient,
00:18:20.15	you end up with a concentration gradient where you have more MEX-5 in these
00:18:24.22	slower complexes in the anterior side of the cytoplasm.
00:18:29.18	So, this was just a hypothetical model that we came up with based on the amount of data
00:18:36.26	that we had.
00:18:37.26	But a... a model is really only useful in... if it actually predicts something that you
00:18:44.06	can then experimentally test.
00:18:46.16	And this model really predicted that MEX-5 should exist into two species.
00:18:52.20	So, I already showed you that it looked like, on a sucrose gradient, MEX-5 did exist in
00:18:58.19	both light and heavy species.
00:19:00.26	But we wanted to be able to see those directly in the embryo.
00:19:05.06	Could we detect a fast MEX-5 and a slow MEX-5, okay?
00:19:11.11	And so, for this, Erik turned to a different technology called fluorescence correlation spectroscopy,
00:19:19.14	which is a microscopy technology that allows you to monitor the diffusion rate
00:19:27.05	of individual molecules in small volumes of cytoplasm that you can sample with your microscope.
00:19:33.26	Okay?
00:19:34.26	And so, by following the fluctuation in fluorescence in this small volume, you can, using mathematics,
00:19:43.01	deduce what kind of molecules are traversing this small volume, and how many different
00:19:51.20	species are there, and how fast are they diffusing?
00:19:55.11	So, doing these kinds of experiments, Erik found that in the anterior cytoplasm there
00:20:02.10	in fact exist two different types of MEX-5 molecules: a very sluggish MEX-5 molecule
00:20:12.26	that is really not diffusing very much at all, and then a faster one, okay?
00:20:18.24	So, that's the two molecules that exist in the anterior cytoplasm.
00:20:23.13	And, in fact, remember that we had measured the average diffusion behavior of MEX-5 using
00:20:31.20	the Dendra experiment that I presented at the beginning of this presentation.
00:20:37.14	And that average number actually uhh... fits with an average that we can get from computing
00:20:44.17	the average of these two numbers and also taking into account how many of these two
00:20:52.06	types of molecules there are.
00:20:53.22	It turns out that there's a lot more of these very slow molecules in the anterior cytoplasm.
00:20:59.10	And that's why that makes the overall population average quite slow in the anterior cytoplasm.
00:21:06.04	What about in the posterior cytoplasm?
00:21:08.01	Well, there, Erik found that those two species exist as well.
00:21:12.27	There's also a very slow one and a very fast one, just like in the anterior cytoplasm.
00:21:18.05	The difference, however, is in the proportion of those two species.
00:21:23.08	In the posterior cytoplasm, the fast species is more abundant compared to what it was in
00:21:30.17	the anterior.
00:21:31.17	There's actually equal amounts of the fast and the slow in the posterior.
00:21:35.28	Okay?
00:21:37.02	So, now, this experiment is really confirming our model that MEX-5 exists in two species:
00:21:45.11	slow and fast.
00:21:46.17	And what differs between the anterior and the posterior cytoplasm is the proportion
00:21:50.24	of those two species.
00:21:53.23	So, now, we are getting close to a molecular model for how this gradient forms.
00:22:00.26	We know that we have two MEX-5 species, a slow and a fast one, and that these two...
00:22:06.01	the interconversion between these two species depends on this PAR-1 kinase, which creates
00:22:12.22	the fast species, but the fast species can revert back to the slow species through the
00:22:19.03	action of this phosphatase that is present throughout the cytoplasm.
00:22:24.06	So, based on this information, we thought that maybe we would be very close to actually
00:22:31.04	being able to model, using mathematical formulas, the MEX-5 gradient.
00:22:38.20	Okay?
00:22:39.19	So, if this explains everything about how the gradient forms, just by inputting
00:22:44.24	these values into a mathematical model, we should be able to recreate the MEX-5 gradient.
00:22:51.02	So, for this, we had to team up with David Odde, a computational biologist, who took
00:22:57.23	our experimentally determined values, together with the size of the C elegans embryo,
00:23:04.18	and putting PAR-1 kinase in the posterior.
00:23:08.18	And, with all of this information, David was able to recreate, in a computer,
00:23:16.04	the MEX-5 gradient.
00:23:17.23	So, here... umm... the black line represents the total MEX-5, which forms a gradient across
00:23:24.20	the anterior-posterior axis, and then the red and the green lines represent the fast
00:23:31.11	and the... and the slow MEX-5.
00:23:34.23	And you can see that, in the ant... the posterior cytoplasm, there's equal levels of those
00:23:39.01	two species, where... whereas in the anterior cytoplasm, there's more of the slow species.
00:23:45.00	So, this type of analysis is very satisfying, because it suggests that maybe we can explain
00:23:52.14	this MEX-5 concentration gradient just by imagining that the PAR-1 kinase can change
00:24:02.12	the diffusion rate of MEX-5.
00:24:05.10	Okay?
00:24:06.13	So, here we have a gradient that's formed in the cytoplasm without having any directed movement.
00:24:14.26	All we have is a local, reversible phosphorylation that induces a local change in diffusion rate.
00:24:23.03	And that is sufficient to create a gradient.
00:24:25.28	So, if you're interested in learning a little bit more about the MEX-5 gradient, and getting
00:24:32.04	more of a direct feel for how such a gradient might form, you might be interested in this
00:24:38.24	video from David Odde, who collaborated with a dance company to bring the MEX-5 gradient alive.
00:24:49.09	So, thank you again for following this presentation, and I hope to see you another time.
00:24:58.05	Bye.