Tip:
Highlight text to annotate it
X
Welcome to the second part of the iBioSeminar. This
is titled, "Looking for functional rafts in cell
membranes." And I'm, as you can see, Satyajit
Mayor, from the National Centre for Biological
Sciences, talking about some work that we've been
doing in collaboration with a close colleague of mine,
Dr. Madan Rao, the Raman Research Institute, also
in Bangalore. He's a soft condensed matter physicist,
with whom we've been trying to explore the structure
of cell membranes, and in fact Parts 2 and 3 are
work that we've been doing in collaboration in an
interdisciplinary way between a cell biologist (myself)
and a physicist (Madan Rao). And the reason we
have engaged in this collaboration, I think, should
have in some ways been apparent from the lessons
we learned from Part 1, where we examined the
notion of membrane rafts and the operational criteria
that have been used to look for membrane rafts in
cells, and some of the problems associated with
those criteria to look for rafts in cell membranes.
Now, what we have then argued is that, while there
are problem associated with the operational criteria to
look for rafts, the cell membranes and cells
themselves seem to be telling us that membrane
segregation, or segregation of specific components
in the membrane of living cells, must occur. And the
need of the hour then is ways to look for these
segregated regions in cell membranes. So this part is
in fact our exploration of looking for these segregated
regions in cell membranes, potentially imbued with
function. Well, the story in my laboratory, anyway,
starts with trying to understand the endocytic
process, and in cells, one of the best-characterized
means of endocytosis is the one of clathrin- and
dynamin-dependent endocytosis, where these
components collaborate to make membrane pits, and
these pits then pinch off material at the cell surface,
bring them into the cell as endocytic vesicles, and
then these vesicles, which are composed of
components of the cell membrane encased in their
budding coats, these components then move into
the cell, to form endocytic structures, which contain
a fair representation of the components of the plasma
membrane. For example, in this slide, what you see
are transferrin (an iron-binding protein) bound to the
transferrin receptor, shown in the green fluorescent
image here, and a lipoprotein LDL, labeled in a red
dye, coming into endosomes inside the cell, wherein
they are present in the same endocytic organelle.
The questions that we have asked are in fact ones
about what happens to the endocytosis of the
membrane components in general. So if you look at
the underbelly of the plasma membrane, it looks like
something out of a movie, in fact it could be any kind
of a movie, could be a horror movie or a movie of
things on a Martian landscape, where you have
these fantastic organelles. In fact, this is a clathrin
basket that you see here, where the clathrin coats
the membrane and makes the membrane pucker in
to bring in components. And in this context, we want
to ask the question about what happens to
membrane lipids and lipid-anchored proteins when
they get taken up by the cell membrane. GPI-
anchored proteins, as we had discussed in the first
part, are an example of components of the
membrane that are in fact some of the original
reasons why the raft hypothesis was proposed,
because they are sorted to different surfaces of the
epithelial cell, the apical surface of an epithelial cell,
as opposed to being delivered to the basolateral. In
the endocytic process, we find that the GPI-
anchored proteins are also sorted. So at the cell
surface, if we label GPI-anchored protein, here we've
expressed a folate receptor. The GPI-anchored
proteins come in many stripes and flavors. There are
about 300 different types of GPI-anchored proteins.
Many of them are receptors for small molecules, for
example, the folate receptor is a GPI-anchored
protein in all mammalian systems. The folate receptor
brings in folates, and we can tag the folate receptor
with a small molecule ligand, the folate ligand,
attached to a fluorophore, so it becomes easy to
follow the folate receptor as it brings in this tagged,
fluorescent ligand into the cell. So at the cell surface,
if you take cells that express the folate receptor, and
we tag it with a small molecule ligand, at the cell
surface the folate receptor looks fairly uniformly
distributed. And in the same cell, if we incorporate an
exogenous phospholipid, now we've fluorescently
tagged a phospholipid, an NBD (which is a
fluorophore, a green fluorophore) tagged to a
sphingolipid, we find that this fluorophore is
incorporated in the cell membrane, and when it's
endocytosed, the two components of the membrane
(the NBD-labeled sphingomyelin, which is
incorporated at the cell surface; and the folate
receptor), they seem to segregate. The green dots
here indicate endosomes that are enriched for the
NBD lipid, and the red dots here indicate endosomes
that are enriched for the GPI-anchored folate
receptor. So this suggests that these two different
types of lipids are sorted at the cell surface and
brought into endosomes perhaps by distinct
mechanisms. In fact, at the cell surface, if we look by
an electron micrograph, and this is an electron
micrograph from a colleague of mine, Rob Parton,
from the University of Queensland, where he shows
that the GPI-anchored proteins are in fact coming
into the cell where these long, tubular invaginations
that have concentrated GPI-anchored proteins in
them. And work that we have done over the years
has revealed to us that GPI-anchored proteins are
endocytosed via a specialized endocytic pathway
that's called the "GEEC" pathway, for the "GPI-
anchored protein-enriched endosomal
compartments" that they make. And these pathways are in fact a
main means by which the cell seems to bring in a lot
of the fluid that is also endocytosed into cells. This
pathway requires the small molecule GTPase,
Cdc42, but does not seem to require a host of other
components that are necessary for the clathrin- and
dynamin-dependent endocytic machinery. Instead,
this pathway, which is a pinocytic pathway (bringing
in fluid into the cell as well), is rather lipid selective,
it's lipid dependent, it requires the small molecule
GTPase, and is in fact very sensitive to perturbations
of the actin cytoskeleton. But putting these
characteristics of a GPI-anchored protein together,
the fact that it's sorted into a specialized endocytic
pathway at the cell surface, while it lacks any kind of
cytoplasmic extension, is telling us something. It's
telling us that these components must be somehow
segregated in the plane of the membrane to form
regions or domains where these molecules can be
selectively endocytosed. So in other words, I think
it's providing us an opportunity to look for regions of
the cell membrane that could be enriched in these
components, which by all accounts, would
correspond to the functional definition of a raft:
regions of membranes imbued with function. So, this
is exactly what we've been trying to investigate. So
first of all, we know from work in our lab and other
laboratories that membrane rafts, if they exist, must
consist of some kind of a clustering of cholesterol
and sphingolipids; and that they must be platforms
which allow the segregation of specific proteins, so
that they can be sorted or generate signaling
functions in specific lipid contexts. So how do we
look for these rafts, so these components of
membrane which are functional? Well, one way that
people have looked for them are by this notion of this
detergent-resistant extraction method, the DRM
measurement. And as we discussed in the Part 1,
there are a lot of problems associated with that
operational definition, so for the purpose of this
presentation, I think we should ignore that criteria.
Other measurements that have been done to
investigate the segregation of components in cell
membranes are to ask questions about the diffusional
characteristics of molecules in the plane of the
membrane. And if molecules enter regions where the
local environment of the membrane is more liquid-
ordered, for example, one would expect that
components that are moving into these regions
would slow down in their mobility, and this is another
means to investigate whether there are regions in cell
membranes where there are lateral segregations of
components contributing to an ordering of lipids in
the membrane. We, in our laboratory, have not
looked at these diffusion measurements, but I'll come
back to them in Part 3 of this seminar. A method that
we have used is ones based on trying to investigate
the proximity between components in the cell
membrane and ask questions about can we observe
a coming together of these lipid molecules in
membranes of living cells when they are being
endocytosed. So to do that, we need to step back
and have, as I said, a proper probe. So these GPI-
anchored proteins is, by all accounts, a good way to
investigate the characteristics of segregation in cell
membrane, because they are sorted in the plane of
the cell membrane, and they are endocytosed by
these specialized pathways. So do they exhibit any
special segregated distribution in the plane of the
membrane? And if they do, what is the scale of those
segregated domains? So the first question we asked
was simply, by tagging the GPI-anchored proteins
with a fluorescent dye, again the folate receptor, we
tagged it with fluorescent folate and imaged it using a
confocal microscope, and what we found was that
the molecules look rather boring. They look as if
they're distributed uniformly on the surface of the cell,
with some small speckles of these components that
in fact correspond to little microvilli, which you can
see on the sides of these cells. So this doesn't look
like some big raft, or big segregated domain, in the
membrane which could correspond to a sorted site
for endocytosis. So the molecules at the scale of the
fluorescence micrograph, which is at the scale of
optical resolution, could be dispersed in a random
fashion, or they could be dispersed in a fashion
where the molecules are present as very small
aggregates in the cell membrane, way below the
resolution of light. Another analysis that we did was a
crude sort of immunogold electron micrographic
analysis, which also didn't reveal any large-scale
clustering of these proteins at the surface of the cell
membranes. The gold particles that you see are in
these golden arrows in fact look rather uniformly
dispersed on the surface of the cell. But below the
resolution of light, one could in fact have some
aggregation of these components taking place, and
to investigate that, different investigators have used
a variety of tools. So electron micrographic analysis
coupled with some statistical analysis of the
distribution of how the particles in the membrane has
been one very successful approach that Rob Parton
and John Hancock have pioneered. Chemical
crosslinking of components in situ is another
approach that Kurzchalia and Chiara Zurzolo have
utilized very effectively. And we, on the other hand,
have tried to use sort of a nonperturbing approach,
where we try and observe the proximity between
different GPI-linked proteins using Förster's
resonance energy transfer, or in other words, FRET,
which is a technique that allows you to look at
proximity between two fluorophores, and I will explain
some of the tenets of this technique in the next few
slides. So the FRET is a process in which energy is
transferred from a donor fluorophore that is excited by
an excitation light, and the energy is transferred to an
acceptor fluorophore, and that transfer of energy
results in the acceptor fluorophore emitting light of a
different wavelength. And the efficiency of this
transfer can be measured by many means, and the
efficiency of this transfer is exquisitely sensitive to the
distance between the two fluorophores. In fact, the
efficiency of transfer is sensitive to the sixth power of
the distance between between the fluorophores. So
if the fluorophore distances change by a factor of
25%, the efficiency of energy transfer decreases by a
factor of eight, so you have a very steep
dependence of the energy transfer efficiency with
the distance between the fluorophores. And the
scale that is used in this measurement is a
characteristic of the two chromophores that are
engaged in this energy transfer process. The scale is
called the Förster's radius, which is a characteristic of
the overlap of the donor and acceptor fluorophores,
and in fact serves as kind of a spectroscopic ruler to
measure distances in the nanometer scale in any
fluorescence experiment. We have in fact used a
variant of this type of FRET process to monitor the
distance between fluorophores. This variant is called
"***-FRET," where we look at the energy transfer
between two like fluorophores. And the way one can
monitor the energy transfer between these like
fluorophores is by measuring the emission anisotropy,
or measuring the polarization of fluorescence
emission emitted by the donor and acceptor
fluorophores. I'll explain this in a little detail. So if you
have a fluorophore, for example, GFP (green
fluorescent protein), and excite it with polarized light,
the fluorophore is going to emit light in a polarized
fashion, and if you monitor the emission of
fluorescence, you will monitor a fairly polarized
emission of fluorescence, because GFP, being a
large molecule, doesn't allow the fluorophore in its
excited state to rotate very far before it emits. So the
only depolarization that one monitors is the
depolarization due to the arrangement of these GFP
fluorophores. If GFP was a very small molecule and
rotating very rapidly, the emission anisotropy that was
collected from the GFP fluorescence would be highly
depolarized. So rotation can depolarize the
fluorescence emission, and FRET can also
depolarize the fluorescence emission. So there are
two GFP fluorophores very close together, they can
transfer energy to each other, and in that transfer
process, the orientation information of the initial
excitation energy is lost, and the energy transfer
process itself leads to a depolarization of the
fluorescence from fluorophores that are engaged in
close proximity. So, by monitoring the extent of
depolarization due to these two different types of
processes (rotational and energy transfer), if one can
isolate these two different processes of
depolarization, one ends up with a very sensitive measure of the
FRET phenomena. So we've used this technique,
which is monitoring the anisotropy fluorescence from
these fluorescently tagged GPI-anchored proteins, to
ask whether they are present in close proximity to
their neighbors. So for this type of experiment, we
take Chinese hamster ovary cells and transfect them
with a fluorescent GPI-anchored protein (for example,
GFP tagged to a GPI-anchoring signal, or a
monomeric YFP tagged to a GPI-anchoring signal, or
our favorite molecule, the folate receptor) and then
excite these fluorophores with polarized light, and
then monitor the anisotropy of fluorescence emission.
So the first thing that we observe that the anisotropy
of the emission is uniform across a field of cells,
which have very widely varying fluorescence
intensity, indicating widely varying protein levels in
the cell membranes. So across a range of protein
expression in a cell membrane, we monitor a value of
anisotropy from different cell populations, and this
anisotropy is quite depolarized for the different GPI-
anchored proteins that we have transfected into
cells. And in fact that depolarization is sensitive to
the perturbation of cholesterol in the cell membrane,
so if you remove cholesterol from the cell membrane
using a variety of different means, we see a rise in
the fluorescence anisotropy, indicating that we are
somehow losing the FRET component of this
fluorescence anisotropy measurement. And the
fluorescence anisotropy now rises to a value which
corresponds to isolated fluorophores that are present
in cell membranes. So this sort of an experiment
immediately tells us that the GPI-anchored proteins
cannot be organized in a random arrangement in cell
membranes where they do not see their neighbors. In
fact, they must be organized in an arrangement
where there are optically unresolvable clusters of
proteins where this energy transfer process must be
taking place. And what we have been able to show
is that this FRET process arises from an extremely
high-density packing of these fluorescently tagged
GPI-anchored proteins in membranes of cells. This
high-density packing is something that we have been
able to uncover using a time-resolved fluorescence
anisotropy measurement. The details of these
measurements are quite complex, and I think they
are perhaps better referred to papers from our
laboratory. But what we've been able to show using
these time-resolved fluorescence anisotropy
measurements is that the GPI-anchored proteins, at
least the species that exhibit the FRET or close
proximity, are present in incredibly close distances
between each other. They're present in around 4
nanometer separation between each other. The two
GPI-anchored proteins, in this case let's say two GFP
molecules, which are 3 nanometers in size on their
own, are within 4 nanometers of each other. The
centers of these proteins are within 4 nanometers of
each other, so they're almost close packed, they're
almost touching. But that still doesn't tell us what size
these structures of these FRET-competent species
are. What is the size of their structures, are they a
few nanometers, are they tens of molecules, or are
they hundreds of molecules? Because all of that will
still fit within the optical resolution of a microscope,
which is about 300 nanometers. So some clues to
understanding the structure of these FRET-
competent species in cell membranes came in fact
from another type of experiment, where we were
monitoring the fluorescence anisotropy of these
folate receptors labeled with fluorescein tagged to
the folate ligand, and when we monitored the
fluorescence anisotropy of these folate receptors, the
fluorophores all seemed to get bleached, and when
they bleach, their fluorescence anisotropy simply
rises up the scale in very systematic steps, so if you
bleach the fluorophores by 50%, the fluorescence
anisotropy rises by a given value. If you bleach it
again by 50%, it changes by another systematic
value. And in fact our interpretation is that if you
have small clusters of these molecules, when they
bleach, within the clusters you see some change in
the distances between these different components.
So what we've tried to do is try and make theoretical
models for these structures. We've made two types
of models at two different scales. One at the scale
where the domains that these molecules form are
much larger than the Förster's radius, which is the
scale of our fluorescence energy transfer experiment.
And the other model is a model where the
fluorophores are distributed in a much smaller scale,
on the scale of the Förster's radius. But first let's look
at the model where we've looked at the change of
fluorescence anisotropy in a model where the
fluorophores are distributed in domains which are
much larger than the Förster's radius. So what we've
done is we've created a situation where we've
arranged fluorophores in these arrangements, and
then bleached them, and in each of these conditions
we've calculated the extent of anisotropy change
we'd expect from fluorophores that were in FRET
distances from each other. And what we find is in
fact that none of the model that have a size that's
much larger than the Förster's radius are able to
describe the data that we have generated from this
photobleaching perturbation of anisotropy. So for
example, if we look at the change in fluorescence
intensity as a function of the bleaching process, as
you move left across this screen, what we find is that
the fluorescence anisotropy creeps up the scale, and
the different models that we have for domains that
are much larger than the Förster's radius fail to fit the
observed experimental values of anisotropy. In fact,
the only model that fits this observed change is a
model where the density of the fluorophores in the
domains is very, very low. But the measured density
that we have for these fluorophores in these domains
is in fact that of molecular separations on the order of
4 nanometers, and if you put those numbers in, we
fail to fit the data that we get from the experiment.
But the model that best fits the experiment is a model
that looks like this: Where we have very tiny clusters,
three or four or maybe five molecules in size, which in
fact are present in a low abundance in the context of
a large number of monomers. So in this type of an
experiment, we find that the theoretical model
perfectly fits the experimental data, and using some
of the fit parameters, we have been able to come up
with the structure of these FRET-competent species
in the cell membrane, and it looks like this, that these
GPI-anchored proteins are arranged predominantly in
the form of monomers, and a small fraction between
20-40% are present as these little nanoclusters. Not
only that, experiments using this sort of a technique, I
think, have allowed us to ask questions about the
properties of these nanoclusters. One thing that we
have been able to show is that multiple GPI-
anchored species can cohabit one nanocluster. So if
you have many different GPI-anchored proteins, if
you monitor the fluorescence anisotropy for one of
these species, the fluorescence anisotropy in fact is
perturbed by the presence of the other species, and
that indicates that multiple or many different GPI
species can cohabit in the same little nanocluster. So
the GPI-anchored proteins in cell membranes then
are organized as monomers and mixed nanoclusters
in cell membranes. Another feature of this
organization is that it is flexible. If we selectively
perturb one of the GPI-anchored proteins by
crosslinking them with antibodies for example, the
other GPI-anchored proteins which do not get
crosslinked by an antibody (which is, let's say, able to
recognize the green GPI-anchored proteins versus
the red ones), the red ones now reorganize to a
distribution that would have corresponded to a
situation where they do not see the green ones. So
the GPI-anchored proteins seem to be flexibly
organized, and their organization can be perturbed or
changed and altered when you do something to
these proteins that moves them laterally away from
these preexisting structures. So they are, in other
words, flexibly organized as monomers and mixed
nanoclusters. They are concentration independent,
and I think this is a very important property of these
GPI-anchored proteins, that if you look at different
concentrations of GPI-anchored proteins in cell
membranes, the fraction of clusters that are present
is independent of the concentration of these GPI-
anchored proteins in the cell membrane. They're
cholesterol sensitive. And I think, given these
properties, people who are not interested in GPI-
anchored proteins might say, "Well, so what? You're
looking at GPI-anchored proteins and what you're
seeing is a characteristic of these molecules, and
maybe there's something special about GPI-
anchored proteins." But I think what I'd like to show
you now is some experiments that have been done
on other lipid-anchored molecules, for example the
Ras protein. The Ras is a inner leaflet lipid-anchored
molecule, and the Ras molecule is one of the key
players in oncogenic transformation of cells, and the
way it interprets growth factor signals from the
outside determines, in many cases, the ability of a
cell to proliferate or continue to grow or stop growing.
And the organization of that Ras molecule has been
a matter of great interest and contention, because it
seems to dictate the outcome of the signaling
process through growth factor signaling. And from
some experiments that John Hancock and his
colleagues have done, they have been able to show
that the Ras proteins are also clustered in the inner
leaflet of cells, in small,