The Membrane-skeleton fence and anchored transmembrane-protein picket models shown in the Slide Shows 1 and 2 have now a strong support from the results of electron tomography of the membrane skeleton, shown in this Slide Show 4. Here, we show that the compartment size observed for phospholipids and proteins agree well with the mesh size of the membrane skeleton right on the cytoplasmic surface of the plasma membrane. To determine the mesh size just on the cytoplasmic surface, conventional electron microscopy was insufficient, and we have applied recently developed electron tomography technology to the plasma membrane specimen, prepared from the ripped-off membrane samples, which were then rapidly frozen, deeply-etched, and platinum replicated.

This slide show is based on the presentation given by Nobuhiro Morone, who is now Group Leader at the National Center for Neurology and Psychiatry (Kodaira city near Tokyo; his mail address is available upon request to us) in various occasions. The main part of this presentation has been published in The Journal of Cell Biology (174, 851-862, 2006). One of his results appeared on the cover page of the journal (shown below), and the Journal was kind enough to attach the 3D viewing glasses to the distributed book form of the journal. This work is also introduced in the "In This Issue" section of the Journal, by an editorial staff, Nikki LeBrasseur.

Nobuhiro Morone

In this Slide Show, I would like to present "Three-dimensional reconstruction of the membrane skeleton at the plasma membrane interface by electron tomography.".

The part of the cytoskeleton at the interface with the plasma membrane plays a unique role for the function of the plasma membrane. The structure and the molecular compositions of the cytoskeleton at the interface are different from those in the bulk cytoskeleton. Therefore, the portion of the cytoskeleton that are closely associated with the plasma membrane is often called the membrane skeleton for short, and I would use this term to indicate the interface structure of the cytoskeleton near the plasma membrane. In fact, the membrane skeleton is considered as a part for both the plasma membrane and the cell's interior skeleton. This is an electron microscope image showing the cytoplasmic surface of the cell membrane.

Throughout my talk, I'm showing the inner surface of the cell membrane, viewed from inside the cell.

This electron micrograph shows a bird's-eye view, a very large area of the cytoplasmic surface of the cell membrane of NRK cells in culture. It is the cytoplasmic surface of the upper cell membrane. Previously, it was difficult to observe such a large area in a single specimen. As I magnify this image,

we can tell that the membrane skeleton consists of many filaments, with a diameter around 7-9 nm.

As I further magnify this image, I hope you can see a striped pattern of a 5.5-nm periodicity, which indicates that these are actin filaments. This shows that the membrane skeleton is primarily composed of actin filaments. The actin filaments with striped pattern are everywhere. (It is difficult to prepare electron microscope samples in which all of the actin filaments that are present in the image show the striped pattern of the 5.5-nm periodicity, as shown here.)

Now we are back to the bird's eye view of the cytoplasmic surface of the plasma membrane. These results already lead to one of the most important conclusions of this presentation. Namely, the membrane skeleton is likely to cover almost the entire cytoplasmic surface of the plasma membrane. The exception appears to be clathrin-coated pits, caveolae, and other internalization apparatuses, and the cell adhesion structures.

Here, let me give you a glimpse of 3D views of the membrane undercoat using so-called anaglyph. Our friend, Dr. John Heuser at Washington University is a world leader of electron microscopy, and has helped us a lot in this project. We also learned the power of EM anaglyphs from John, and he taught us how to do it.

Here, I would like to show you several EM anaglyphs of inner surface of the plasma membrane. Please have the viewing glasses, with the red glass on your left. These glasses are widely available online ( for example, in Japan ) or via novelty sources such as comic book shops and toy stores. Please make sure that you have the red glass on you left, otherwise, you would not get the correct 3D effect. If you cannot find the viewing glasses, we could send you upon request at singlemoleculesfrontier.kyoto-u.ac.jp)

This image shows the cytoplasmic surface of the upper plasma membrane. As you can see, the inner cell surface is uneven and covered with undercoat structures. Let me magnify this image.

Here, you can see the presence of caveolae (protruding structure with striated patterns) and clathrin-coated pits (another protruding structure located near the left bottom corner of the image with hexagonal patterns) on the cytoplasmic surface of the plasma membrane.

Here is the magnified image of these the clathrin-coated pit. These structures are coated with proteins called clathrin heavy chain and light chain, and thus called clathrin-coated pits. They are generally around 140 nm in diameter, and this particular one is somewhat greater than average, around 200-300 nm in diameter, Perhaps, two clathrin-coated pits are formed at the same place. Its surface has a polygonal lattice pattern like a soccer ball with hexagonal and pentagonal patches combined to form such curved surfaces.

Caveolae have striated coats on invaginations of 50-75 nm in diameter.

Both the clathrin-coated pits and caveolae are surrounded by and linked to the membrane skeleton.

Here is a stereo image of the membrane skeleton in an area where clathrin-coated pits and caveolae are scarce. The skeletal filaments frequently cross each other.

When the filaments are magnified, the stripe pattern of a 5.5-nm periodicity could be seen very clearly, showing that these filaments are actin filaments.

We often find filaments coming vertically, toward the cytoplasm, from another actin filament lying horizontally on the plasma membrane. Probably, this way, the membrane skeleton is connected to the bulk cytoskeleton.

Here, probably I should go back, and explain to you how these EM specimens were prepared.

I'd like to shift gears now and would like to show you the method for 3D-reconstruction of the membrane skeleton, using electron-microscope-based computed tomography or electron tomography.

We collected many images of the platinum-replicated samples by tilting the stage between plus minus 70 degrees, and obtain the image every 1 degree. Therefore, we collected 141 images for a single view field.

This movie shows a typical series of the tilted images of the membrane undercoat structures.

Let's enlarge the images. This series was obtained at a different view field. Many actin filaments can be seen.

From the collection of tilted images, a series of sliced images parallel to the membrane were calculated. I would like to walk you through a typical stack of sliced images, starting from the cytoplasmic side, toward the cytoplasmic surface of the plasma membrane. The thickness of each slice here is 0.85 nm.

Here, we move from the cytoplasmic side toward the cytoplasmic surface of the membrane.

Please stop the movie, whenever you want to inspect any particular slice. For example, at the middle of the sequence, a bundle of actin filaments or a stress fiber can be seen, which runs diagonally from the left bottom of the image toward the right top of the image. Later, one would start seeing the membrane skeleton. Just prior to reaching the membrane, one would see the last layer of the skeleton on the membrane, which can interact directly with membrane proteins and lipids. Therefore, this represents the part of the membrane skeleton that directly interacts with the lipid bilayer membrane.

Previously, by looking at the movement of single membrane proteins and lipids, Fujiwara and others in our group showed that the cell membrane is compartmentalized with regard to the lateral diffusion of membrane molecules, and these membrane molecules undergo short-term confined diffusion within a compartment, and long-term hop diffusion from one compartment to the next in the cell membrane. These colored areas mean the compartments in the cell membrane. In the next slide, I will show you the hop diffusion of a transmembrane protein transferrin receptor on the video.

This video clip shows in real time the movement of a colloidal-gold-labeled transmembrane protein, called transferrin receptor, in the plasma membrane of NRK cells in culture. In the second round of this video, the same video sequence is shown but with the particle's trajectory superimposed on the image sequence. Based on many experiments like this, leading to many publications, it was concluded that the plasma membrane is totally parceled up into apposed domains of 30-230 nm (the compartment size varies from cell type to cell type, but independent of molecules. Different molecules sense the same compartment size although their hop rates may vary greatly). Within a compartment, the molecule moves very rapidly, and repeatedly bounces off the compartment walls.

We hypothesized that the mechanism behind this motion is hidden on the intracellular surface of the cell membrane. Namely, we assumed that the membrane skeleton located right on the cytoplasmic surface of the plasma membrane acts like fences between the compartments. Receptor molecules collide with the membrane skeleton fence, which induces their temporary confinement in the membrane skeleton mesh. The receptor molecules hop across the fence, to travel to an adjacent compartment, when sufficient apace between the membrane and the membrane skeleton is formed due to the thermal motion of these structures, or when the actin filament temporarily dissociates. This is called the "Membrane Skeleton Fence Model".

In addition, as shown in the Slide Shows 1-3, we also found that phospholipid molecules in the outer-leaflet of the plasma membrane also undergo such hop diffusion. Such hop diffusion and short-term confined diffusion were independent of the extracellular matrix, the extracellular domains of membrane proteins, and the partial cholesterol depletion, but the presence of actin-based membrane skeleton. Again based on many results, we proposed "anchored-transmembrane protein picket model", in which transmembrane proteins anchored to and lined up along the membrane skeleton practically form rows of pickets in the plasma membrane along the membrane skeleton, and by their steric hindrance effects as well as hydrodynamic-friction effects (immobile proteins slow the movement of lipids and proteins around them due to hydrodynamic dragging effects), inducing corralling of all the membrane-incorporated molecules within the membrane skeleton mesh. Note that the membrane skeleton is not totally occupied by the picket proteins, because the hydrodynamic friction effect propagates quite far from the surface of the transmembrane protein, of the order of several nanometers.

We wanted to compare the compartment size determined by observing the diffusion of membrane molecules with the mesh size of the membrane skeleton right on the cytoplasmic surface of the plasma membrane. This is a conventional electron micrograph, and we can also construct the normal stereo view based on the difference in the view angle between left and right eyes, as I showed in the previous slides.

However, these methods cannot quantitatively speak to the distance between each individual actin filament and the inner surface of the plasma membrane. For such analysis, quantitative 3D reconstruction of the membrane skeleton on the cytoplasmic surface of the plasma membrane would be necessary. In the present research, we obtained the distribution of the membrane skeleton mesh size right on the inner surface of the plasma membrane from the 3D-reconstructed images. Then, we were able to examine if the tomography data are consistent with the single-molecule tracking data.

By using 18 consecutive images taken every 0.85 nm from the cytoplasmic surface of the plasma membrane, the mesh size of the membrane skeleton on the cytoplasmic surface of the plasma membrane was determined. Each green area is the compartment surrounded by the membrane skeleton (see the published paper for details).

In this histogram, the open bars represent the distribution of the membrane-skeleton mesh size determined by electron tomography for NRK cells. And the closed bars show the compartment size determined from the diffusion of a phospholipid. As you can see, these two histograms agree reasonably well.

As we had known that FRSK cells had much smaller compartment sizes by observing phospholipid diffusion, we then determined the membrane-skeleton mesh size in this cell type. The image on the left is the normal deep-etched image, and that on the right shows the compartments determined by electron tomography (white area are the places where we were unable to determine the compartment size due to the presence of actin bundles, condensation of actin molecules, the presence of other plasma membrane structures, and the presence of actin filaments whose end is located within a compartment without ending on another actin filaments).

See the blue open bars. As expected from the diffusion measurements, the membrane-skeleton mesh sizes of the FRSK cells were found to be very small, as can be seen from the blue histogram on the left. The size distribution agrees well with that of the diffusion compartment size determined for a phospholipid (blue closed bars). Red bars represent the distributions for NRK cells, which is the same as that shown in the second previous slide.

Namely, we were able to show the good agreement between the membrane-skeleton mesh size and the compartment size sensed by membrane molecules in two different cell types with very different mesh/compartment sizes.

Taken together, these results strongly support the membrane-skeleton fence and anchored-transmembrane-protein picket models, previously proposed by us.

Here are the conclusions of the Slide Show 4.

Finally, I would like to end by thanking my collaborators. I extensively collaborated with Dr. Jiro Usukura at Nagoya University and Shigeki Yuasa at the National Center for Neurology and Psychiatry. Yukiko Hirata at FEI company started us with electron tomography, We obtained important technical advices from Drs. T. Baba and S. Ohno at Yamanashi University School of Medicine, M. Setou from Mitsubishi Kagaku Institute of Life Sciences, and K. Kozuka at the National Center for Neurology and Psychiatry, In particular, Dr. John Heuser at Washington University gave us a lot of helpful advice and encouragement.

Here are John and his wife, Tanya. As explained in the presentation, John gave us a lot of help for this work. This photo was shot at Kiyomizudera Temple in Kyoto, in February 2003.

And here are the authors of this presentation. Nobu Morone is on the left, and Aki Kusumi on the right. This photo was shot by John, also at Kiyomizudera Temple.

I would also like to thank the sponsors for the present research. And above all, I would like to thank you very much for your interest in this Slide Show 4.

We can send you a complete copy of the presentation in Microsoft PowerPoint format (version 2000) with full resolution video (in AVI format) upon request. Please contact us at:
singlemolecules111kusumi.frontier.kyoto-u.ac.jp