Slide Show 3 The Diffusion Barrier in the Neuronal Cell Membrane.
Slide Show 3 The Diffusion Barrier in the Neuronal Cell Membrane.
My talk today will be on the diffusion barrier in the neuronal cell membrane.
2. Introduction: polarized distribution across the plasma membrane.
My interest in this topic arose from the fact that neurons are highly polarized into distinct domains, which are functionally different.
As you may know, the cell body and the dendrites, which are together called the somato-dendritic domain, receives and integrates the electric signals from other neurons, and the axon, which is at the opposite end of the cell, emits the integrated information as the output. These domains are divided by the region called the initial segment, the part of the axon nearest to the cell body.
Previous studies have shown that each functional domain has a different set of molecules such as these membrane molecules listed here, and that selective transport is necessary to deliver such molecules into each domain.
3-1. Fluid mosaic model
These are schematic drawings of the cell membrane. This is a static view. These are membrane proteins spanning the lipid bilayer and these are the actin filaments comprising of the membrane skeleton mesh on the cytoplasmic surface of the cell. These are the static views, but in real life, these membrane constituent molecules are undergoing rapid thermal diffusion or Brownian diffusion within the two dimensional plane of the membrane, making the membrane a two dimensional fluid entity.
Therefore, to keep polarized distributions of membrane molecules, the selective transport of newly synthesized proteins and lipids may not be sufficient. Since the plasma membrane is one continuous membrane, delocalization of membrane molecules by thermal diffusion has to be blocked in some way, i.e., some mechanism is required to prevent diffusional mixing, like the presence of a diffusion barrier at the initial segment.
5. Summary of the previous work
These are previous key papers on this subject, discussing whether the diffusion barrier does exist or not in the initial segment membrane, and their conclusions are different. One of the major difficulties of believing in the presence of the diffusion barrier is that it is difficult to conceive of a barrier mechanism by which diffusion of molecules that span only half of the membrane could be blocked.
Futerman AH, Khanin R. Segel LA, Lipid duffusion in neurons. Nature362: 119. (1193) PubMed
Wincker B and Poo MM, No diffusion barrier at axon hillock. Nature 379: 213. (1996)
Kobayashi T, Storrie B, Simons K, and Dotti CG. A functional barrier to movement of lipids in polarized neurons. Nature 359:647-650.(1992)PubMed
Winckler B, Forscher P, and Mellman I. A diffusion barrier maintains distribution of membarane proteins in polarized neurons. Nature 397:698-701.(1999)PubMed
6. Questions addressed in this research
Therefore, we first asked if there is a diffusion barrier in the initial segment membrane.
And if so, when is that barrier formed during development in cultured neurons, and what is a plausible mechanism that can block the diffusion of even phospholipids.
We studied the dynamics of an unsaturated phospholipid, DOPE (dioleoylphosphatidylethanolamine) at a single-molecule level, incorporated in the cell membrane of cultured hippocampal neurons. We selected such an unsaturated phospholipid as a probe, because this type of molecule appeared most difficult to block diffusion. We used both fluorescent and colloidal gold tags for studying the movement of DOPE. Fluorescent probes are good in the sense that single fluorescent spots truly represent single molecules and in addition, they are smaller than gold probes. Gold probes give better time resolutions and better spatial precisions for localizing the probe. For the analysis of slow diffusion, good spatial precisions are important. Also, it has the added benefit enabling dragging experiments employing optical tweezers.Neurons were taken from the hippocampi of new-born rats (within 12 hours after birth) so that days in vitro (DIV, days in culture) match the developmental age in vivo.
First, I'd like to show you the movement of single molecules of Cy3-tagged DOPE in an 11-day old neuron.
8. Video: Movements of Cy3-DOPE in a dendrite of 11-DIV neuron
Let me show you lipid diffusion in a Dendrite, which is this part of an 11-day-old neuron.
Here lies a dendrite. Let me start the video. All videos will be shown in real time. Now I'll start the video. (play) DOPE molecules are rapidly diffusing here, through a wide range of the membrane, as you can see through their trajectories. The scale bar represents 3 microns.
When we used gold-tagged DOPE instead of Cy3-tagged DOPE, they exhibited similar behaviors. I'd like to show you the movement of gold-tagged DOPE in the IS membrane in a couple of video sequences.
This is a 10-day-old neuron. This is the cell body and this part is the initial segment. These are the DOPE probes and they are not moving. Now, please watch around here. A new particle just arrived on the membrane, and was immediately immobilized. This is a slow motion replay. So, DOPE in the initial segment is not diffusing in a 10-day-old neuron. By the way, these moving structures are transported vesicles.
This summarizes the lipid diffusion in various parts of a 10-day-old neuron.
Lipid diffusion was suppressed severely in the initial segment, whereas they were mobile in other regions. The mobility in the IS was as slow as the immobilized molecules on the cover slip, suggesting that there is a diffusion barrier in the IS.
Next, we used optical tweezers and tried to drag lipid molecules through the initial segment. I would like to show you these experiments on video.
15. Video: dragging experiment of G-DOPE at the IS membrane of a 12-DIV neuron
This is the initial segment of a 12-day-old neuron. I trapped two gold particles. One is attached to DOPE and the other one is trapped in the solution. When I move the cell, the one in solution reports the location of the optical trap. And when I turn the laser off, it goes away. But the particle attached to DOPE stays stationary, with respect to the cell, escaping from the trap, indicating that this molecule is fixed to, or hits, some solid structure in the cell.
We also found that a younger neurons do not display a diffusion barrier.
I'd like to show you an dragging experiment in a 1-day-old neuron in the initial segment membrane.
(video) This is the axon. I'm going to expand this part. This is the cell body and this area is the initial segment.
I'm trying to drag a DOPE molecule along the membrane, through the initial segment. Now the DOPE probe is being dragged through the initial segment, and now it has reached the cell body. When I turn off the laser trap, that particle starts diffusing on the membrane. This indicates that no barrier exists in this one-day old neuron.
This is a 1-day-old neuron. Phospholipid diffusion was high everywhere including the initial segment. Barrier formation is therefore thought to occur during maturation of neurons.
19. Dmicro at each part of the cell (6&10-DIV).
In this slide, I am comparing the diffusion rates of DOPE in various parts of the neuron. At day 6, which is shown in open black circle, the diffusion rate is high everywhere. Solid black circles shown on the right represent molecules immobilized on the coverslip. So here we displayed the number of the molecules which showed the movements below the noise level in these numbers because their values are meaningless. At day 10, which is shown in red, the diffusion rate dropped by a factor of more than 800 at the IS, whereas in other regions, the diffusion rates only slightly decreased.
We then asked what causes lipids to stop their diffusion in the initial segment of mature neurons, and how the barrier is formed during developtment.
To explain our idea, first I'd like to introduce you to the models we have proposed to explain why the diffusion rate of membrane molecules in the cell membrane is reduced by a factor of 5 to 50 compared with that in reconstituted membranes or in liposomes. I would like to start from the MSK fence model.
20. EM picture showing MSK (FRSK cell)
We have previously shown that the actin-based membrane skeleton (MSK) as shown here play a key role in suppressing the diffusion in other cultured cells.
21. Typical trajectories of Transferrin Receptor in the cell membrane of NRK cell
These are typical trajectories of a transmembrane protein called transferrin receptor in normal rat kidney cells in culture. This particular molecule moved about rapidly, but was confined in a region of about 200 nm for several tens of milliseconds, 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, and found that many membrane proteins undergo macroscopic diffusion by repeating such confinement plus hop movements over many compartments, which we call hop diffusion.
These lines here are to help the eye, but we showed that, in fact, these compartment boundaries are made of 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.
22. Video: Gold-DOPE in NRK, Time resolution of 25 ms/frame
Here I am going to show you the movement of our phospholipid probe DOPE tagged with a gold particle, again in an NRK cell. This sequence was recorded at a time resolution of 25 microseconds, which is faster than video rate by a factor of 1350. (Start the video.) The video is slowed by a factor of 270 from real time. This is a gold particle. The scale bar here is 1 micron. It is a little hard to see what's going on, so in the next sequence, which is the same as this one, the particle's trajectory is superimposed on the image. Different colors indicate different plausible compartments detected by computer software we developed. We carried out a statistical analysis, and indeed we found that more than 85% of DOPE molecules were undergoing hop diffusion.
23. Typical Trajectoriess of G-DOPE in NRK, Time resolution of 25 ms/frame
A quantitative analysis showed that the compartment size was 230 nm on average, the same as that sensed by a transmembrane protein transferrin receptor. The residency time within each 230-nm compartment was 11 ms. Therefore, without employing high time resolutions, such fast hop diffusion could not be observed.
These results indicate that 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.
24. Summary of the experiments showing effects of several drugs on G-DOPE diffusion
What makes the boundaries between these compartments?
After a variety of experiments, as shown here, we found that the membrane skeleton is responsible for such corralled diffusion of phospholipids.
(This slide gives further details regarding the involvement of the actin-based membrane skeleton for compartmentalization of the plasma membrane even for phospholipids. You could skip this slide.) The membrane compartmentalization could be due to (1) the extracellular matrix (ECM), (2) extracellular domains of membrane proteins, (3) exclusion from rafts, (4) partitioning into rafts, and (5) indirect interaction with the actin-based membrane skeleton. First, cleaving off the majority of the cell surface proteins (including extracellular domains of membrane proteins and extracellular matrix proteins) did not affect DOPE diffusion (the second row in the Table below. In addition, both gold (40 nm) and Cy3 (small) tags for DOPE gave the same diffusion rate in video-rate observations, showing that the interaction of the gold probes with other protein components on the cell surface has negligible effects on DOPE diffusion. Therefore, interactions on the extracellular surface are not major constraints of DOPE diffusion. Further we found that partial cholesterol depletion with MbCD or saponin had little effect on the compartment size or the hop rate (data not shown), excluding the third and fourth possibilities involving rafts.
When very mild latrunculin treatment was carried out so that very small fractions of actin filaments may be depolymerized (50 nM - 1.7 mM for 2 min at 37 Cº on the microscope stage, and the observation was finished within 15 min), the compartment size was increased by a factor of about 1.4 in diameter, or by a factor of 2 in area (Table 1 below. Note that the purpose of this mild latrunculin treatment is not to wipe out the actin filaments on the cytoplasmic surface, which may often be done in biochemical or cell-biological studies, but to increase the mesh size only slightly by very low level of actin depolymerization). Treatment with jasplakinolide, which stabilizes F-actin, reduced the compartment size only slightly (Bubb et al., 2000), but increased the residency time in a given compartment by a factor of about 1.5-2 (Table below). Taken together, these results indicate the involvement of the actin-based membrane skeleton in compartmentalization of the plasma membrane and hop diffusion of even lipid molecules. DOPE molecules primarily undergo simple Brownian diffusion in membrane blebs where the membrane skeleton is largely depleted or in large unilamellar vesicles.
25. Comparison of G-DOPE trajectories in the cell membrane with that in LUV (liposome)
Here are the trajectories of gold-tagged DOPE recorded at a time resolution of 25 microseconds. As shown in the left and middle trajectories, DOPE molecules undergo hop diffusion over the membrane in the plasma membrane of NRK cells. On the contrary, in the membrane of large unilamellar vesicles (LUVS), DOPE molecules undergo simple Brownian diffusion [the trajectory on the right; different colors indicate fake domains, which may somewhat look like membrane compartments; one might want to say that one could see domains in this trajectory like those shown in purple, red, and yellow near the bottom. However, according to the statistical analysis and the computer program that detects the hop movement (both developed in this lab), they do not represent compartments. Since diffusing molecules tend to stay at the same place (thereby the average position is always the same), they often exhibit such "apparent" domains. Although compartments can only be determined by statistical analysis, even by eye, compartments can often be distinguished because compartments are tightly apposed to each other, and the determined points (coordinates) are densely and rather uniformly populated within a compartment]. The results shown in this slide further support that temporary corralling of phospholipids in these compartments are due to interactions with a cellular structure, namely, the membrane skeleton.
However, this is very strange and we were in fact very surprised to see such hop diffusion of phospholipids. Because the DOPE molecules we are looking at are located only in the outer leaflet of the membrane, whereas the membrane skeleton is located on the cytoplasmic surface of the membrane, they do not see each other. There must be some mechanism to couple the membrane skeleton with phospholipids located in the outer leaflet of the membrane.
26. Anchored-Protein Picket Model
Therefore, we are proposing an "anchored membrane-protein picket model" in which various transmembrane proteins anchored to and lined up along the membrane skeleton effectively act as rows of pickets against free diffusion of phospholipids, due to both direct steric interactions as well as hydrodynamic friction-like interactions with the anchored transmembrane protein pickets. This model was confirmed by a series of Monte Carlo simulations. Theoretically, it is important to realize that transmembrane proteins that are not immobilized on the MSK are not effective diffusion obstacles.
27. Picket model applied to the barrier mechanism at the IS membrane?
Let's go back to the original topic, i.e., on the mechanisms for formation of a diffusion barrier in the initial segment membrane. We thought that the "anchored-protein picket model" might explain the barrier mechanism. Namely, if the picket membrane proteins are highly concentrated and anchored to the concentrated actin-based membrane skeleton underneath the IS membrane, lipid diffusion may be greatly suppressed. In fact, previous studies have shown that the membrane skeleton is more concentrated in the IS compared to other regions.
28. Trajectories on the axon 10-DIV, with ankyrin signal
This is the axon of a 10-day-old neuron. The red signal shows the expression of a membrane skeletal protein called ankyrin-G. Ankyrin is highly concentrated in the IS region and then decreases toward the distal axon. These are typical trajectories of an unsaturated phospholipid DOPE. It is clear that lipid diffusion is suppressed where ankyrin is expressed, and that with an increase of the ankyrin level, DOPE movement was decreased.
This figure shows that actin is also densely distributed in the IS region, which is this thin white structure.
29. Developmental accumulation of ankyrin-G and sodium channel
This slide shows the diffusion rate of DOPE in the initial segment membrane plotted as a function of the developmental stage of the neuron in vitro.
The diffusion rate of DOPE falls off greatly between day 7 and day 10, with a great variation among individual cells. The numbers below represent the number of molecules that exhibited the movements below the noise level. And by day 10, the diffusion rate decreased by a factor of about 800, and became as small as the noise level.
In the top graph, the concentrations of ankyrin-G and the sodium channels in the initial segment membrane are plotted as a function of the days in culture. Sodium channels are known to be bound to ankyrin-G and therefore it is expected to be anchored to the membrne skeleton.
This top graph shows that the concentrations of ankyrinG and sodium channels increased, as a function of days in culture, indicating that membrane skeletal proteins and the picket proteins that are anchored to the membrane skeleton in the IS became accumulated.
Therefore, this is consistent with our model that accumulation of transmembrane proteins anchored to the dense membrane skeleton netwerk forms the diffusion barrier in the IS.
31. Developmental accumulation of ankyrin-G and sodium channel (same as 28)
However, strange thing is that in the period when the diffusion coefficient decreased greatly, the accumulation of the membrane skeletal proteins and picket proteins takes place rather gradually, presenting an apparent contradiction with the model.
32. Monte Carlo simulation
So we ran a series of Monte Carlo Simulations to examine the relationships between the diffusion coefficient and the % coverage of the compartment boundary by anchored picket proteins. The diffusion rate of lipids dropped by a factor of more than hundred as the % coverage increased from 15% to 25%. Thus, a small increase in the amount of anchored membrane proteins can cause a dramatic decrease in the diffusion rate, (because the free area in the compartement boundary region for lipid diffusion decreases rather suddenly as picket fences become packed.)
33. Summary of Single Fluorophore Video Imaging (SFVI) data: Phospholipid and Na+ channels are immobilized in the IS in actin-dependent manners
Finally, we performed some experiments to investigate if the membrane skeleton and the picket membrane proteins are actually the constituents of the IS diffusion barrier.
These are the distributions of the mean square displacement (MSD, for 200ms) of fluorescent probes, observed at the single fluorophore level.
First, we observed changes in phospholipid (DOPE) mobility when the cells were treated with Latrunculin-A, which depolymerizes actin and therefore partially destroys the membrane skeleton. DOPE was originally immobilized in the mature IS membrane, but after the treatment with Lat-A, most of the molecules became mobile, as shown here with black-hatched bars (the second graph from the top).
On the contrary, treatment with Jasplakinolide, which stabilizes actin increased the immobile fraction of DOPE (red hatched bars in the graph second from the bottom) in the IS of an immature neuron (5-DIV), where most of DOPE molecules were originally mobile. These results together indicate that phospholipid diffusion is suppressed by the presence of the MSK, which mainly consists of actin.
Also, most of the voltage dependent channel, which is thought to be bound to the MSK, were immobile in the IS of mature neurons (the top graph, gray bars), but they became completely mobile after partial destruction of MSK by Lat-A treatment (SFVI was carried out using Alexa-labeled antagonist against the voltage-dependent sodium channel.). Therefore, this result suggests that the sodium channel is immobilized in the IS in an actin-dependent manner, thus being one of the constituents of the picket fence barrier.
Our conclusion is summarized in this slide:
A diffusion barrier in the IS membrane is formed by a local accumulation of various transmembrane proteins that become anchored to concurrently concentrated membane skeleton. Sodium channels are major candidate for the pickets, but other transmembrane proteins are probably involved.
Day 10 is about the time that these neurons start emitting electric signals through the sodium channels assembled at the IS membrane. Therefore, it is interesting to note that the accumulation of sodium channels at the IS may have the dual role forming the diffusion barrier that maintains the required polarity for activity in neurons as well as being an integral part for producing the action potential, thus synchronizing the two events.
35. Diffusion barriers found in various type of cells
Such a mechanism for suppression of diffusion as explained by the concentrations of anchored protein pickets, might be generally used to maintain polarity in polarized cells, such as budding yeast, epithelial cells, and mammalian sperm.