|Slide Show 3
This slide show is based on the talks given by Dr. Chieko Nakada, a post doc in our lab, in several occasions.
Diffusion barrier in the neuronal initial segment membrane
My talk today will be on the diffusion barrier in the neuronal cell 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.
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.
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.
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.
First, Ifd like to show you the movement of single molecules of Cy3-tagged DOPE in an 11-day old 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 Ifll 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.
Here is the initial segment region of another 11-day-old neuron. And Ifm going to show you the fluorescence image of the same area in the next video sequence. The scale bar is again 3 microns.
When we used gold-tagged DOPE instead of Cy3-tagged DOPE, they exhibited similar behaviors. Ifd 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.
In the more distal part of the same axon, the lipid probes diffuse rapidly.
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.
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.
This is a distal part of the axon of another 12-day-old neuron.
iwhile trappingjThe optical tweezers are trapping three probes here. The moving structures are not gold probes, but transport vesicles inside the cell.
iwhile scanningjWhen I moved the cell, all of the trapped molecules were dragged along the membrane. When I turned off the laser, they all start rapid diffusion in the membrane.
We also found that a younger neurons do not display a diffusion barrier.
Ifd like to show you an dragging experiment in a 1-day-old neuron in the initial segment membrane.
(video) This is the axon. Ifm going to expand this part. This is the cell body and this area is the initial segment.
Ifm 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.
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 Ifd 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.
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.
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.
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.
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.
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 gapparenth 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.
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.
Letfs 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 ganchored-protein picket modelh 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.
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.
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.
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.
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.)
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.
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.
Finally, I would like to thank my collaborators.
Thank you very much for your kind attention.