We are interested in how signaling molecules in the plasma membrane are organized and also how they are reorganized to form (transient) molecular signaling complexes and signaling domains in/on the membrane when they are engaged in signaling after extracellular stimulation. We are paying special attention to the roles played by the membrane skeleton, scaffolding proteins, and raft domains.
In this slide show, we deal with compartmentalization of the plasma membrane that even acts on diffusion of phospholipids in the plasma membrane (therefore almost any molecules in the membrane). Such knowledge forms the basis for understanding molecular interactions and recruitment occurring in the cell membrane.
Let me start by summarizing the diffusion data of transmembrane proteins in the plasma membrane and explaining the membrane-skeleton fence (corral) model.
If you wish to know more about these studies, please take a look at Slide Show 1 and publications cited there.
Our group has been looking at movements of various membrane proteins in the cell membrane at the level of single or small groups of molecules.
This is a typical trajectory of a transmembrane protein called CD44. This particular molecule moved about rapidly, but was confined in a region of about 600 nm for a few seconds, and then hopped to an adjacent compartment, and then again started confined but fast diffusion, and hopped to the next, and so on. We have examined various membrane proteins and carried out quantitative and statistical analysis, which I don't have time to explain to you, but we found that many membrane proteins undergo macroscopic diffusion over many compartments by repeating such confinement plus hop movements, which we call hop diffusion.
These lines here are to help the eye, but we showed that, in fact, these compartment boundaries are made from the membrane skeleton.
Therefore, we refer to this model as a "membrane-skeleton fence model". Transmembrane proteins are sticking out into the cytoplasm, and, in this model, their cytoplasmic domains collide with the membrane skeleton, which induces temporal confinement or corralling of transmembrane proteins into the membrane skeleton mesh.
The transmembrane protein can cross the barrier when the membrane skeleton or the membrane itself fluctuates to give space between the membrane and the membrane skeleton, or when the membrane skeleton temporarily dissociates or when the transmembrane protein has kinetic energy sufficiently high to get over the membrane skeleton fence barrier when it is in the barrier region.
Membrane biophysicists have been puzzled for these 25 years by the observations that diffusion rates of membrane proteins in the plasma membrane are reduced by a factor of 10 to 100 from those in artificial reconstituted membranes, and we thought we largely solved this puzzle with this membrane skeleton fence model. Namely, diffusion in the plasma membrane is slowed not because the diffusion rate per se is small, but because transmembrane proteins are corralled by the membrane skeleton mesh.
I said "we thought" using the past tense, and this will be clarified toward the end of this Slide Show 2.
I would like to show you the structure of the membrane skeleton observed by electron microscopy. Here, the upper (dorsal) membrane of NRK cells was ripped off to expose the cytoplasmic surface of the cell membrane, and was labeled with anti-actin-colloidal gold probe. This electron micrograph was obtained by Dr. Nobuhiro Morone in our group using a method called rapid-freeze, deep-etch, immunoreplica electron microscopy.
If you have already looked at the Slide Show 1, please skip the Suplementary slides. They are the same as those shown there, but are reproduced here for those who have not seen Slide Show 1.
One of the critical issues with the membrane skeleton fence model is whether or not membrane proteins actually collide with the membrane skeleton. To examine this issue, we carried out the following experiment using red blood cells.
We attached a large bead to the membrane skeleton. At the same time, we labeled a transmembrane protein band 3 with small colloidal gold particles. And we only observed the band 3 that are undergoing hop diffusion.
Then the membrane skeleton was deformed like this by dragging the beads using laser tweezers. If band 3 molecules collide with the membrane skeleton, they will be moved along with the membrane skeleton as shown. If band 3 molecules are not colliding with the membrane skeleton, their movements will not be affected by such displacement of the membrane skeleton.
So, let's look at the result in the next video sequence.
Supplementary slide Video 1
This particle with a very high contrast is the 1 μm bead attached to the membrane skeleton, and these particles are band 3 molecules that are undergoing hop diffusion. When we start dragging the large bead at a dragging rate of 2 μm/s, these band 3 molecules were at the places indicated by + (there are two + signs in the image). As we drag the membrane skeleton toward the west, all band 3 molecules follow. When the membrane skeleton was moved back to the original place, the band 3 molecules are also moved back, but they did not go back to the positions indicated by + signs. This is because the band 3 molecules are undergoing hop diffusion even as they are being dragged by the membrane skeleton. When we labeled spectrin, spectrin moved back to the original place.
It is possible that this result was induced by the hydrodynamic effect due to dragging the spectrin network, and the membrane proteins bound to the spectrin network. To test this possibility, the membrane skeleton was dragged slowly at 0.1 μm/s, which is 1/20 of the previous rate. Even at this slow rate of dragging, band 3 followed the bead.
As another control, phospholipids on the outer surface of the cell were observed. This is the gold particle attached to phosphatidylethanolamine. It was not moved by dragging the cytoskeleton. This shows a slow play back. (When the membrane skeleton was moved very very quickly, the lipid was also moved along with the movement of the membrane skeleton. This can be explained by the "anchored membrane protein picket model" in which transmembrane proteins anchored to the membrane skeleton fence effectively act as rows of pickets against the free diffusion of phospholipids, due to the effects of steric hindrance and hydrodynamic friction (near anchored membrane proteins).
These results indicate that the cytoplasmic domain of band 3 in fact collides with the spectrin membrane skeleton.
Based on these observations, we think that the membrane skeleton fence structure is a basic feature of the cell membrane.
Next, I would like to show you that not only proteins but also lipids are corralled by the MSK. This occurs due to the effects of transmembrane proteins that are anchored to the membrane skeleton, and we call these anchored proteins pickets, because they act as posts for the membrane skeleton fence.
As a lipid, we used an unsaturated phospholipid called DOPE, or dioleoylphosphatidylethanolamine. We tagged DOPE with a fluorescent dye called Cy3, and carried out single fluorophore observations. By the reasons I will explain later, we also tagged DOPE with colloidal gold, and carried out single particle tracking.
I am going to show you the movement of individual Cy3-DOPE molecules in the cell membrane of NRK cells. This is in real time and recorded at the video rate, and the scale bar shows 1 micron. Since these fluorescent spots are photobleached suddenly in a single step, these spots are likely to represent single Cy3-DOPE molecules.
Next is a gold tagged DOPE. This is in real time and recorded at the video rate, and the scale bar shows 1 micron. As you can see, gold-tagged DOPE appears to diffuse as fast as Cy3-DOPE.
This table summarizes the diffusion rates of Cy3-DOPE and gold-tagged DOPE. As long as we stay within time windows less than 100 ms, the diffusion rates are about the same, which are about 0.5 mm2/s on average. However, this value is smaller by a factor of 10 to 20 compared with that in artificial membranes, such as giant liposomes. This result therefore suggests that there must be some kind of slowing mechanisms in the cell membrane which could not be resolved at the time resolution of the video rate, which is 33 ms.
However, achieving higher time resolutions in single fluorophore observations is difficult due to the problem of poor S/N ratios. Therefore, we had to depend on gold-tagged DOPE and single particle tracking. Using a high-speed video system and a modified microscope, we were able to achieve a temporal resolution of 25 μs.
Here I am going to show you the movement of DOPE tagged with a gold particle. This sequence was recorded at a time resolution of 25 microseconds. (Start the video.) For showing, it is slowed by a factor of 270 from real time. Shown is a gold particle. The scale bar here is 1 micron. It is a little hard to see what's going on, and so in the next sequence, which is the same as this one, the particle's trajectory is superimposed. Different colors indicate different plausible compartments. We carried out a statistical analysis, and indeed we found that more than 85% of DOPE molecules were undergoing hop diffusion.
These are representative trajectories of DOPE recorded at a time resolution of 25 microseconds. The scale here is 1 micron.
A quantitative analysis showed that the compartment size was 230 nm on average. The residency time within each 230-nm compartment was 11 ms. No wonder we did not see hop movement at the video rate, whose time resolution is only 33 ms. In fact, these trajectories are 62-millisecond long, and if we had used video rate observation, there would only be 2 or 3 points in the whole trajectory.
The diffusion rate within the 230-nm compartment is interesting, which is 5.4 mm2/s on average. It is as fast as DOPE molecules observed in artificial membranes, such as giant liposomes. Therefore, lipid diffusion in the cell membrane is slow not because the diffusion per se is slow, but because the plasma membrane is compartmentalized with regard to the translational diffusion of phospholipid.
Then the next question may be what makes the boundaries between these compartments. Here, in this table, I'm comparing the compartment sizes for DOPE, which is a phospholipid, and transferrin receptor, which is a transmembrane protein, and the compartment size turned out to be about the same. This raises a possibility that the membrane skeleton mesh is involved in the hop diffusion of phospholipid because it's the membrane skeleton that determines the compartment size for transmembrane proteins such as TfR.
We also observed DOPE diffusion on blebs on the cell surface. The bleb is a balloon-like structure of the cell membrane where the membrane skeleton is largely lost, and we found that DOPE was undergoing very rapid simple Brownian diffusion rather than hop diffusion on the blebs, indicating again that the membrane skeleton may be responsible for the hop diffusion of DOPE.
Therefore, we examined the involvement of the membrane skeleton, as well as that of extracellular matrices, the extracellular domains of the membrane protein, and cholesterol-rich raft domains in the hop diffusion of phospholipid. I do not have time to explain all these, and so you kind of have to take my word for it, but we found that only when we modulated the membrane skeleton, the movement of phospholipid was affected. Therefore, we think that the membrane skeleton is responsible for such corralling of phospholipids.
However, this is very strange because the DOPE we are looking at is located in the outer leaf of the membrane, whereas the membrane skeleton is located on the cytoplasmic surface of the membrane.
Therefore, we are proposing an "anchored membrane-protein picket model" in which various transmembrane proteins anchored to the membrane skeleton effectively act as pickets or rows of pickets against free diffusion of phospholipids, due to both steric hindrance and hydrodynamic-like friction effects by immobile anchored protein pickets. It is important to realize that transmembrane proteins that are not immobilized on the MSK are not effective diffusion obstacles.
In this model, transmembrane proteins that are anchored to the membrane skeleton are coupling the membrane skeleton, which is located on the cytoplasmic surface of the membrane, to the phospholipids that are located in the outer leaflet of the membrane.
We carried out a series of Monte-Carlo simulations to test this model, and found that when 20 - 30% of the MSK surface is covered by the anchored proteins, we were able to reproduce the experimental data, such as the hop rate of every 11 ms on average (the anchored protein pickets do not have to physically close the entire compartment boundaries. Due to the hydrodynamic-like friction effect, which propagates over distances of several protein diameters, the coverage of only 20-30% of the membrane skeleton boundaries is sufficient to cause temporal confinement).
These anchored-protein pickets would act on any molecule incorporated in the membrane. Therefore, in the case of transmembrane proteins, both membrane-skeleton corrals and anchored-protein pickets are responsible for slowing of their diffusion in the plasma membrane. This is the reason why I used the past tense in the previous part.
In the case of NRK cells, we found a second compartmentalization: the cell membrane has double [nested] compartmentalization (cf. T. Fujiwara, K. Ritchie, H. Murakoshi, K. Jacobson, and A. Kusumi. Phospholipids undergo hop diffusion in compartmentalized cell membrane. J. Cell Biol. 157, 1071-1081 (2002)).
Compartmentalization in the plasma membrane in other cells has been investigated in our laboratory. A paper on this subject is under preparation.
By now, you might have started wondering what are the physiological roles or functions of the pickets and fences in the membrane?
A possibility I would like to suggest is that these diffusion barriers are important for maintaining the spatial information when chemoattractant or chemorepellant molecules are received by respective receptor molecules. For example, when an extracellular signal is bound to a receptor monomer in the center, the receptor molecules tend to form oligomers and/or cytoplasmic signaling molecules are often recruited to the liganded receptor molecules. This would induce a dramatic decrease in the diffusion rates of such signaling complexes, due to the enhanced effects of corralling or binding by to the pickets and fences, and these signaling complexes would immediately be arrested within the compartment where the signal was received. Namely, monomer molecules tend to undergo hop diffusion with relative ease, whereas oligomerized receptor molecules and those bound by cytoplasmic signaling molecules cannot cross the pickets and fences very easily (right), and also tend to be tethered to the membrane skeleton (left).
You might intuitively assume that the diffusion rates of molecules in the membrane would decrease if they form oligomers or larger molecular complexes. However, this is not true. Since two-dimensional diffusion is very very insensitive to the diffusant size, even if the receptor molecules form 10-mers, the diffusion rate will decrease only by less than 20%. Without the pickets and fences, oligomerization-induced or signaling-molecule-binding-induced decrease of motion or immobilization would not occur. Such an arrest could be particularly important for the receptors involved in chemotaxis, the cytoskeletal reorganization, or protrusion of processes, where the receptors have to "memorize" where the signal was received.
We have previously shown that oligomerization of E-cadherin-GFP greatly decreased its diffusion rate, and explained this by an "oligomerization-induced trapping model", which forms the basis of the above argument
This is the end of Slide Show 2. Thank you very much for your interest in this work.