Thank you so much for this very nice introduction and also for inviting me. I'm actually very happy to give a talk here because my work has really tried to move the glia field, and I'm a glia biologist, but trying to move the field into looking at the brain as an organ and really looking at the physiology of the brain as a whole organ instead of looking at microscopic signaling from synapse to glia cell or glia cell to synapse. So, as a way of introducing it, I would say that the field of neuroscience for many years has really been inspired by computational research. And if you look at most neuroscience meetings, you'd see figures like this that came out of the Obama and the European Brain Initiative, that really present the brain as a highly wired structure. And the idea is behind that, that if you understand all the connection and all the signaling between the billion of knowns we have, we would understand how the brain works.
If you're MD, you realize that the brain doesn't look like that. The brain is an organ similar to the kidney, for example, or the liver. And it is composed, of course, of cell and tissue that need to work.
And it also has to take into consideration the very high metabolic rate. demands and the very long lifespan of humans. So how do we keep the, and this is really what our work is about, how do we keep the brain healthy for the many decades of human life? We actually don't know a lot about that.
We know, for example, from other organs like the heart, that cardiovascular exercise is a key. We know we have to drink a lot of water if we have a tendency to kidney stone. And Probably all of you also know, although it's not really a developing field, we all know that sleep is completely essential for how we feel when we wake up in the morning. If we have been up most of the night, we know we are not very good to learn.
We might be in a bad mood. At a little bit longer sleep deprivation, people might have seizures or hallucinations. So, Why is sleep, and I'm not a sleep researcher, and actually when I was trained in medicine in Copenhagen in the late 80s, we didn't have a single lecture on sleep. So I went into the field because I really wanted to understand why is sleep so restorative. We all know that, and we all have a relationship to sleep.
And it's actually, it's a field that has developed quite rapidly, at least since I trained, and it's been shown that sleep is important during development. and probably REM sleep from the deep brainstem drive the neural circuitry formation, and essential for that. It's known that sleep is a key for memory formation, and there's this beautiful hypothesis about synaptic homeostasis, where we keep increasing the number of synapses and the size of synapses during wakefulness, and then we downscale the synapses that are storing information that's not useful by keeping... but keeping synapses large that are storing newly learned material.
We also know that immune function is way better when we sleep. We get tired when we have an infection, and opposite sleep really helps fighting the infection, because the anti-inflammatory process is way more efficient during sleep. One thing we know is that we do not sleep to preserve energy.
We only save a little bit of energy while we sleep, and that is, I think, probably one of the most surprising features, because why would the brain use almost the same amount of energy? And this is primarily glucose when we are sleeping, because we are decoupled from our surrounding, we are not processing information, we are not making decisions, and we are not walking around. So this suggests that the brain might actually be pretty active during sleep.
None of this explains, and this is really where it came from, why all species, and these are fly, mice and humans, why they die after days or weeks of sleep deprivation. So this observation suggests that sleep must observe a very fundamental biological role that's not explained by any of these features here. A less dramatic example is Sigrid Visay's beautiful studies from Penn Sleep Clinic.
And she basically sleep-deprived mice for three days, very mildly. They were kept awake for three hours when they're supposed to go to sleep in the morning, and then they were laid bedding in the cage. So they could basically recover their sleep loss.
She did that for three days, and then she looked at the dendritic complexity in Leucocevilius, which we know. is very sensitive to sleep deprivation. But what she saw from the three days of very mild sleep deprivation was you actually had a loss of dendrite and the complexity was decreased.
So this again suggests that sleep actually, even on a structural level, might be important. So searching for why sleep is so important, it's fairly obvious that the brain, and this is both the CNS, it's both the brain and spinal cord, are completely devoid of lymphatic vessels. And maybe that could explain some of the need for sleep.
So what are lymphatic vessels? Lymphatic vessels, as K.A.L.A. Tarlow described this morning, they are, of course, very important immune function, they're important in cancer spread, but they also serve an essential role in return of waste product from peripheral organs. So any protein that's present in the extracellular space would, by convective flow, basically be transported into the lymph vessels here, a blind lymph capillary, and that would be dumped into the venous system by the collecting large lymphatic vessels. It's not trivial, because we return more than 20 grams of protein on a daily basis from peripheral tissue, which I think is quite substantial.
And maybe one of the surprising things about the protein you return is that these are not limited to extracellular proteins or albumin. You can actually find not only the extracellular matrix proteins in the lymph, but you can also find a very large proportion of cellular proteins within the lymph. So this suggests that most tissue are actually utilizing the lymphatic system as a garbage disposal system that's similar to your weekly pickup of garbage would pick it up without even sorting it.
The beauty of this system is it doesn't require any specific transporter. You dump it and you're not sorting it. Of course, all the endogenous protein degradation requires much more energy on a single cell level. So what about the brain?
The brain is, of course, similar to all other tissues. It's always rebuilding its own protein. So there is a protein in the brain that has a very fast turnaround, about one to one and a half hours.
So this is... receptors, for example, and then there are proteins that have a week to a month's half-life. These are cytoskeleton proteins and many others. But a very simple calculation of protein turnaround suggests that the brain actually, on a daily basis, is producing and degrading 7 grams of protein. And then you can say, is it really such that the most precious organs of all is the only organ that's forced to recycle all its own protein?
This concept is based on two things. First of all, we don't have lymphatic vessels. Second of all, the blood-brain barrier has essentially no transporters for protein waste. So therefore, the field of neurodegeneration has basically stated that all protein in the brain has to be recycled on a cellular level by autophagy or ubiquitination, for example. This might be a very key question in neuroscience, because if you look at the neurodegenerative diseases and you actually step a little bit backward, you can state that all these diseases of aging, that are Alzheimer's, Huntington's, Parkinson's, ALS, and many others, they're actually characterized by accumulation of proteins.
Each of them have their own specific protein. It's amyloid, it's tau, it's nuclein, or Huntington. But each of these proteins, they are actually, they differ from other proteins by having a tendency for aggregation. So if these proteins are present in larger concentration or larger amount for a long time in the brain, if it's stagnated, you'd actually start to see aggregation. And a characteristic of these protein aggregation is once you start the aggregation, you basically suck.
other proteins into these aggregates. So if you take amyloid plaques and look at what is in amyloid plaque, you have a whole bunch of proteins, although it started as an amyloid plaque. So we are not the first to propose that the serpospinid fluid might act as a semi-or quasi-lymphatic system. So the serpospinid fluid is produced within the ventricles, and we have four ventricles. And in the ventricles, all the ventricles, we have the choroid plexus, that basically produces half a liter of cerebrospinal fluid on a daily basis in humans.
This fluid would pass through the ventricle, leave the inner part of the brain by forvarium magentae, and then it would basically bait the brain. And this is a very clever system, because baiting the brain actually reduces the weight of the brain from 3 pounds to just 50 grams. And this is important if your organ is floating in fluid, that you are not that heavy, because otherwise you would collapse.
And second of all, the fluid would protect the brain from mild traumatic injury. And that happens, of course, once in a while. The cerebrospinal fluid will, after the textbook, leave by the aragnoid villi, the cranial nerve, especially the olfactory nerve, and spinal nerves. These are the major efflux pathways from the brain, but I'll come back to that later. So how can this system function as a quasi-lymphatic system?
It was discussed in the 50s and early 60s fairly extensively, and the argument against this was that the brain environment is very complex. Basically, the brain consists of trillions of processes interweaved. And if you have all these processes interweaved, the process of diffusion become very, very slow, because basically you have to take a tortuous path getting out to the cerebrospinal fluid.
There exists, however, another pathway, which is convection. And convection is basically like a river. It's fluid flow driven by a pressure gradient or a gravity, for example.
And in convection, it does not matter whether you're a small peptide, like beta amyloid, that's about 4 kilodalton. versus a larger waste product as tau that is 10 times larger or 40 kilodalton, because these proteins would flow with the same velocity as a fluid flow. So that's a clear advantage compared to diffusion.
So could we have convection brain? We probably have, and this is actually just a very simple movie downloaded from Wikipedia, showing that the cerebrospinal fluid is always in motion. And why is it in motion? Because the brain... receive a tremendous large arterial supply that form the circular villus at the base of the brain.
And that pulsatility these large vessels set up always move cerebrospinal fluid not only within the ventricle but also outside. Furthermore, in brain the vasculature is actually organized very different from in peripheral organ. In other organs, artery and veins, often a lymph vessel and a nerve would follow each other into the organ.
surrounded by a little bit of fibrous tissue. One of the characteristics of the brain vasculature is that artery and veins are always separated. They're never one in parallel.
This is not really discussed in textbook, but it's really a characteristic of the brain. The second is that these large vessels that enter the brain, they actually enter by diving directly down into the tissue. And if you look at that in a little bit higher magnification, you see here we have a...
large pia arteriole. It would send penetrating artery directly into cortex. And opposite you would have in striatum penetrating vessel diving directly into striatum.
The difference of this vasculature from all other organ is that the arteries are surrounded by glia end feet. So glia cell and these are astrocytes would actually send vascular processes out that plaster on the vessel wall. These glia end feet cover 98.4% of the entire vasculature, and that is ranging from arteries, capillaries to veins, so the entire vasculature. And what does that create?
It creates a very specific space that only exists in the central nervous system, so brain and spinal cord, and this compartment here is called the perivascular space. The perivascular space differs fundamentally from the rest of the brain. The rest of the brain, you have trillions of processes crossing each other. In the perivascular space, it's nothing, except occasional microglia cells, macrophages, or p-acids, and otherwise just loose fibrous tissue. So this creates a pathway that is basically like a donut-shaped tunnel.
It's donut-shaped, so you have the inner wall is a vascular wall, and the outer wall are the glia end feet. So this perivascular space presents a highway. for fast transport of fluid, potentially, because there's almost no resistance if you try to move fluid fast. In the rest of the brain, it would need the very high resistance from the very complex brain environment. So our model was that you actually used arterial pulsatility to drive cerebrospinal fluid down into the perivascular space because these PL arteries are actually laying, they are resting.
directly in the subarachnoid space that contains the cerebrospinal fluid. So they're resting there, and you have them connecting directly by the perivascular spaces, also called the vacant robin space, directly linked. So our model was basically that you use the artery, the pumping function of the artery, to drive cerebrospinal fluid in along the perivascular spaces with very low resistance. So that was the model, and we set out to test that initially. using two-photon emitting.
So in this experiment, we would basically anesthetize the mice, we would prepare a cranial window, and we would inject a tracer, in this case a green dextran tracer, probably. We either use dextran of various sizes or albumin tagged with flucine. So we would inject an cyston of magma. The only reason for injecting an cyston of magma is it's a nice big fluid-filled compartment so we can inject without causing injury.
All the other spaces are pretty tight, so it's hard to inject. So we inject into the stent of magna and look in mice, and this is over the parietal cortex, so it's about two centimeters away from where we inject, and in order to be able to actually observe something, we would have filled the plasma with a red tracer. So these arrows are actually placed specifically to point out the penetrating artery. And remember in our model, the penetrating artery played a central role, because that's where you have big perivascular spaces, and we would expect, if our model is right, that the cerebrospinal fluid first enters along this penetrating artery. So now we inject the tracer, and the first thing you would see is that you have basically, you see the outline here of a PL artery on top, but you very quickly...
would see that you have actually influx, not within the vessel, but outside the vessel, of the green tracer. And then it distributes equally across the cortex. So we are looking here only about 180 micrometers below the surface, because this is where you have optimal imaging. If you go deeper, it's actually harder to image these tracers well.
But I think the most surprising thing about this movie is how fast this process occurs. Because this entire movie is collected over 15 minutes. So in 15 minutes you have basically the same amount of tracer in the extracellular space as you have outside in the cerebrospinal fluid. So being a glia biologist, we were really interested in this because of one question, and that had been a mystery in the glia field.
So here we have stained with two markers that are specific for astrocyte. One is GFAP, and GFAP is an intermediary filament, so it would stain the cell body and the major 3-5 processes of astrocyte. And astrocyte, I actually named astrocyte because this looks like stars. So this is a very classical staining of astrocyte. Then we have also stained nuclei, and you see the nuclei of neurons are really nice, large and beautiful compared with astrocyte, which is normally a little bit smaller.
But then we have also stained with another marker for astrocyte that is totally specific in brain for astrocyte, and that is for the water channel, aqua point 4. So aqua point 4 here are expressed, and this is shown here, it is expressed primarily in the vascular end feed of astrocyte. And not only is it polarized to the end feed, it's actually polarized, such that it's only expressed at high densities. at the face of the glia infeed that faces the vessel wall. And that had been a mystery for many years in the glia field.
Why the brain would need so many water channels right there? Because one of the characteristics of brain endothelial cells is that they have no water channel. They're devoid of water channel. Whereas endothelial cells in peripheral organs express aquaporin-1, there's no such expression of brain endothelial cells.
So they're basically a closed faucet. There's as little as possible water flowing to the brain endothelial cell. So why would you need in the glia infeed, facing the capillaries, another vessel, why would you need such a high expression?
It's actually calculated that this expression is 10 times higher than kidney cells in the discending tubuli. So the expression is about 4, I forgot, but it's about 10 times higher. And you can see the expression is so polarized, it actually looks like a vessel stain, but it's not.
You're standing in the infeed plastered around the capillaries in this case here. So we were actually postulating that maybe this makes sense. Maybe this is because you have fluid pumped down along the perivascular spaces and you would place this water channel in a strategic favorable position to basically facilitate fluid inflow or water inflow into the brain. And to test that, we worked with a group of Ole Peter Otteson that was at that point at the University of Oslo. And we basically compared tracer influx.
Here it's injected, this is tracers injected, about 30 minutes before we sacrificed the mice and prepared coronal sections. And we compared it to a mouse here generated, where there's a global knockout of ACRO.4. And we saw that tracer influx was very significantly reduced in these mice. There's no known inhibitors of ACRO.4, at least there's not at that time point. So this was the best way of defining it.
So based on that, we basically defined a new system. We also did a lot of other experiments that have been published where we basically could separate arteries from veins. Because the idea was to describe that there are actually a systemic fluid transport in the brain that's highly polarized. And how does it work? So it works in this way that we have this hypospinal fluid that is baiting the PLR tree.
Pumping of the arteries would drive the fluid, the cerebrospinal fluid, down into the perivascular spaces. Here it would leave the perivascular space and enter the brain tissue, facilitated by the very dense expression of aqua point 4. That would set up a fluid flow within the brain tissue that would, without any discrimination for sizes or composition, it would drive any... compound in the extracellular space towards the venous space, the perivinous space. I'm not documenting that here, because we have done that, and we basically use reporter mice to document the polarity. This fluid would then leave the brain by the perivascular spaces and meet the lymphatic vessels.
And we call this a glymphatic system, because we saw the shared features with the lymphatic system in peripheral organ, and the G stands for glia cell, because... 0.4 expression glia cell is a key for this function. Since then, there's been published beautiful studies showing there is actually traditional lymphatic vessels, not within the brain, but in the meninges that surround the brain.
I hope you heard the talk by K. Alatalo today. This was actually very much welcome, these studies, because this provided a link that was missing in our study, and that is how do you transport half a liter of of CSF out on a daily basis from the brain. And what this study showed was that you actually have lymphatic capillaries that are positioned around the perivine system, primarily around the sinuses, the superior sagittal sinus and the transverse sinus here.
And that would form traditional lymphatic vessels that express all the lymphatic markers and then basically merge into lateral lymphatic vessels and dump. Again, the lymph out into the systemic circulation, where then all the waste product from the brain and from other tissues can be degraded by the liver that really is the professional recycling plant of our body. So in this model, it's actually very simple.
It might look complex, but it is actually simple. We have peri-arterial influx. Then we would have supported by ACO.4, convective flow within the brain tissue. It would collect peri-venously and would be drained out by meningeal and also cervical lymphatic vessels. So once you have this model, you can start to use...
literature to test, this is a white model, and one thing that's very easy to do is to ask, what if you express beta amyloid in the brain? Can we detect it further down? Can we detect beta amyloid expressed within the brain tissue here?
can be actually detected in lymphatic nodes further down. So here in this study here, a group I actually don't know, they used the APPPS1 model of Alzheimer's disease. They make a lot of beta amyloid, a human beta amyloid.
And what they did was they measured beta amyloid in brain. Not surprisingly, it's very high. But I think the surprising thing is you actually find almost as high level inter-lymph node that receives CSF from... the brain, but not if you look further down. So this was supporting it.
So we started thinking about maybe we should develop a more clinical relevant platform to study glymphatic fluxes, because obviously optical signaling would not be ideal. So I worked here with Helene Benveniste, who is now a professor at Yale University, and she replicated our studies, but she used MRI to detect in real time the whole brain pattern. of fluid flow and she also moved to rats because it's a bigger species and better for the MRI imaging. So she would basically inject a contrast agent here in Cystunum magna and follow its movement.
And as you can see the movement of the tracer follow the perivascular spaces and then it basically leaks out, or is cleared out. So this movie is two hours long but influx is similar to mice pretty much over in about half an hour. And it's mostly clearance by this point.
So looking at that, we started to ask this very basic question. How can the brain function almost as a kidney and pump this very large amount of fluid around and at the same time be awake? Because you should imagine that neural activity requires a lot of energy. So this was basically our introduction to that.
Because how can the brain actually pump all this amount of fluid around? So movement of fluid is very energy demanding because it requires all the movement of ions across the plasma membranes. So what we did was to ask, is it always active? And one of the things we realized at that point was that all the studies we had done so far were done in anesthetized mice or rats.
They all received anesthesia. So these two young students, Lu Lu Xi and Hong Ni Kang, undertook the next study, and that was to study, is it also active when they are in a... are weak.
And the model was fairly simplistic. It was basically to implant a cannula into Cystuna magna and then have it by two ports connected to the same size tracer. So we used five kilodalton dextrins, but attacked with either texasite or flucine and injected in the same animal at different stages of its sleep-wake cycle. A mouse, you cannot tell when you look. look at a mouse whether to sleep or wake if it's sitting still.
So the standard in the field is that you can basically decide what state the animal are in if you measure the prevalence of the slow delta wave in combination with EMG. So if the prevalence of the slow delta wave that are very prominent during non-wamp sleep, if it's low, the animal is awake. So this was an awake mouse.
We gave it ketamine silicin anesthesia. No surprise, it went into... it got a higher prevalence of the slow delta wave.
On the other hand, you could start out with an animal that was already asleep because the prevalence of delta wave was high, and you could wake it up by gently touching its tail. This is all done in darkness, and the mice actually tend to fall asleep because experiments are done during daytime when they are supposed to sleep, and they have been trained for a minimum of three days before we do the experiment. And it's always the same experimenter. So the idea was... didn't want to have inter-animal variability, so we looked now at the vasculature here in the same animal, in the same field of view, and we would inject a red tracer when it's under natural sleep, and a green when it's natural awake.
And this is determined by EG electrode in the opposite hemisphere. So you can see the sleep movie here basically replicate what I showed you in anesthetized mice. You see an influx where you first have influx along the... Maybe show it again.
You first have influx along the PL artery, and then you have the tracer dipping down into the vacant robin space surrounding the penetrating artery. There's almost no influx in the awake mice. So, I'm sure you all have it, like we have it. If you see something that's so dramatic difference, you wonder, so what did I do wrong here? So we replicate it many times, and we use different techniques because you imagine that the awake mice might be struggling, or the some reason, mechanical reason, for it doesn't reach it, but we found that consistently with many different type of approaches, we found that in a wake state, the mice is basically closing off its brain for influx of traces.
So I will show more evidence for that later. I just want to show, since it's a physiology meeting, what is the mechanistic background for that. And we asked, so why is the ECSF influx so different in the sleep-awake?
And for that purpose, I worked together with... with my old collaborator, Charles Nicholson, from NYU. And what Charles has done is that he has basically developed the only technique that is actually available to quantify very accurately the size of the extracellular space. So the extracellular space volume, if you look here, is about 20% of total volume in the brain. And the way his method is called, the TMA method, this method is fairly simple.
You would inject, you would use... use current injection to avoid movement of the electrode, you would use current or ion to forasis to inject TMA into the extracellular space. And then you detect TMA by a microelectrode that's selective for TMA in a distance that's normally about 150 micrometer.
Then you can, based on the curve and the arrival of it, you can calculate back what is the accurate size. So you have developed a software that's freely available to do. do that.
So if the extracellular space is small, you would see a faster rise and a larger amplitude, basically less diluted. If you have a larger extracellular space, you would see a slower slope and also a reduced amplitude because you're diluting the tracer. And this is how the measurement look like. The animals are awake here, so you would see movement in the recording.
So what we found, and this was maybe some of the most surprising observation, what we found is that consistent increase in the extracellular space volume when you transfer from the awake to sleep state or to ketamine, psilocybin, anesthesia. And it's fairly dramatic because we have, in awake state, we have what's called the extracellular space volume is about 14% of total volume and increases to about 20%, which is what is the standard that has been measured in all other studies using brain slices and in vivo observation in anesthetized mice. so the obvious question is what triggered these changes these very dramatic changes and I would say that other studies have supported this they have not directly replicated it but there is one set of studies where you use MRR to look at water diffusion and water diffusion has been shown to be more and more restricted the longer we are awake and as soon as we fall asleep it becomes wider the other line of evidence is a beautiful study that was published in in science last year by Tononi and Cervalis group, where they had fixed animals at different time points in the sleep-wake cycle and basically quantified the sizes of different compartments in the brain.
And what they showed was that all the elements, and these are astrocytes and especially synapses, they keep expanding during wakefulness and then the decreases in size during sleep. And especially sleep deprivation is very harmful because it can cause sleep deprivation. keeps increasing during that. So anyway, it has not been directly confirmed, but there's many indirect evidence from other groups that this very dramatic change we reported is correct.
We have also replicated this observation by just placing TMA on top of the brain and just recorded what happened when we give the animal anesthesia. And what we can see is that this change actually happened over just a few minutes. So it's fairly fast. So it must be shift in fluid fluid.
between the extra and the intercellular compartment. But what triggered these changes? Here we can actually go back to the beautiful slip literature that have shown in very, very convincing details that it's the deep brain nuclei that drive arousal. So especially the concerted release of many of these snow modulators, and especially...
in norepinephrine and orexin that drive wakefulness. So if you use optogenetic studies, it's been shown that leucocervilius that produces norepinephrine is probably the most important mediator. of wakefulness because you can transfer a sleeping mouse into a wake mouse in second if you stimulate the look with civilians neurons so we focused on that and we also went back to a study that was been has been published by investigator Bruno and this is old study it's from 2011 so he published this beautiful study where he showed that if he took a mouse that was a word that was anesthetized no he took in a wake mouse you start with an wake mouse.
He makes a little cranial window, and then he basically drips norepinephrine receptor antagonists, so this is for alpha-1, alpha-2, and beta receptor antagonists, on top of just that piece of tissue. He doesn't do anything else, no cutting to block the thalamic input, but he just drops it all, and he can actually transfer the EEG recordings in this little piece of cortex to a sleep-like pattern because the prevalence of slow delta... wave increases. Whereas the mouse is still awake, it will still be behaving on the stage, but this little piece of the brain is still asleep. So we saw that's really neat.
Let us replicate that. So we replicated that and measured interstitial or extracellular volume, and we found that just adding this cocktail of norepinephrine receptor antagonists could immediately increase the volume of the extracellular space in our awake mouse. And then we looked at glymphatic influx. and we found that we could transfer the very low inflow, this is plotted in orange, of their wake mouse to a very substantial influx if, in this case here, injected the cocktail systemically. So we changed it for the whole brain.
So this suggested that the arousal that's already been described very nicely here is actually the arousal system that drives no one to wake up. And we wake up and start to sense our surrounding are also the ones that shut off. the glymphatic system. So looking at that, and this is just, it's actually studies that have not been published, looking at a system we believe is organ-wide and looking at something as global, as different brain state, as sleep and awake, looking at that using two-photon imaging and a cranial window is obviously not ideal. You're looking in a pinhole at something you believe is brain-wide.
So we went on to... In later studies, we have really preferred to use macroscopic imaging. Macroscopic imaging is basically just a photographer lens placed at the microscope, and it's connected to LED light that's pretty strong.
And since the skull of the mice are actually fairly, what's called, it's not so dense in blocking photons, you can actually flip the skin away and look directly down at tracer distance. within the brain without opening it and without preparing cranial window. This is obviously a major advantage if you want to look at dynamic within a closed cavity that you're not doing any surgical preparation.
So how does that look like? So here we are looking down at sleep. This is actually an anesthetized mouse.
This would be the olfactory bulb. This would be cerebellum down here. And here we have the middle cerebellar artery. So this movie is basically just imitating fluorescence molecules.
that are mixed up into cerebrospinal fluid and moving along the perivascular space. It looks like blood flow, but it's not. But it's still a very fast movie.
It's actually a half-an-hour long movie. And what you can observe here in a wake mouse, and this is head fixed, so we have basically a frame here and here so it doesn't move out. You can see there's very much reduced glymphatic influx in the awake state of a wake mouse. So looking at some...
something so global, we start to ask very simplistic questions, and this was based on choosing. So we asked, does the body position matter? And we asked that for two reasons.
One is the lymphatic system in peripheral bodies. Obviously, it's very important how your position is, but also all known species, and these are man to fly, they actually sleep with their head down. And you know, if you travel transatlantic, how much you would actually pay for getting a business car seat to be able to put your head down. because it's hard to obtain stable sleep if you're forced to sit up. So we did a very simple study.
I will not show you the data, but we found that the lateral right side seems to be favorable to glymphatic flow, and we actually had 30% higher inflow in this study if we placed the rat, in this case here, on the lateral right side. It probably is because when you place the animal on the lateral right side, you have the heart in a higher position and systemic... noradrenaline goes down because the heart has an easier time pumping and venous returns easier. So it's very beautiful studies that have measured the level of systemic noradrenaline in standing, sitting, to laying down and laying down on the lateral side, and it's all declining in that order. So it's probably because we have a lower sympathetic drive when we lay on the lateral right side.
So going into more systemic, and this is very recent data that's not published yet, we have got into more quantitative description of it because we really want to know the driving factors and how those driving factors might be affected in diseases that lead to neurodegeneration. And for that purpose, I work with a team of fluid dynamicists, and this is needed because these are studies I cannot perform, and these people are incredibly smart. They're actually studying. When I came to them and asked them to study glymphatics, they said, why should we leave studying fluid dynamics around the sun? So they don't even know what a biological specimen is.
And what we have done is that we have moved from looking at dextran or albumin traces to look at particles. And why do we do that? We do that because I've developed software that can do particle tracking and with very high accuracy basically define the velocity. So if you inject now, instead of injecting tracers, you inject microspheres, and these are one micrometer.
And image, again using two-photon imaging, but using very, very fast imaging. This is 30 hertz. You can actually start to correlate the velocity of the microsphere that's tracked here. So this is velocity of the microsphere.
These are the heartbeat measured by EKG, and this is respiration. And what you can see by just looking at the particle as a move along the perivascular space, and this is a pial artery that's filled with a red tracer just to follow the blood. So all this is red blood cell and white blood cell moving that appear black in this movie here.
But if you look at those particles, some of them are aggregated, so we actually have many different sizes of the particles. But what you can see is they don't move evenly. They have these increases in velocity.
And those increases in velocity seem to really correlate with the R waves of the EKG. They occur about 40 milliseconds later, which suggests that this is actually the time it takes a pulse wave generated from the heart constriction to reach that artery up there. And you can correlate that, and the beauty of this is you get an enormous amount of information.
So the mean number of particles that are tracing each mice is 20,000. So when you do the correlation, you get very strong correlation. What we find here is blue, this is velocity. I hope you can see it, it's plotted against the EKG. And you can see there's a very nice correlation here.
Opposite, if you correlate it to respiration, we see a fairly weak correlation, but there is a correlation suggesting that respiration might also, at least under some circumstances, drive the movement of the perivascular flow. One thing that's obvious to ask is, how would hypertension? affected. Such known hypertension is a major risk factor for small vessel disease and for Alzheimer's. These are the primary reasons for dementia in old age.
So we ask, what does hypertension do to this flow? Because if the fluid flow around the arteries are reduced, we would also expect that beta amyloid clearance is reduced. So this was basically our question, does acute hypertension affect this activity?
And here we have, the nice thing about this is you can study the same mouse. So here we have basically plotted in the same mouse the microsphere movement during normal blood pressure and after we induce hypertension by endutension. So it's the same mouse, the same field we are looking at. What you can see is that there is, especially if you look down here at the tractories that's plotted out here, you can see the tractory is somewhat disturbed in the hypertension. So there's more stop, it stops more, and it actually moves slightly backwards during these speeds.
So it's like in the normal case, in the normal tension, they would move forward, slow down, move forward, and slow down. In the hypertension, you would actually in some cases see even a little back movement, not just slowing down, but a little back movement. So if you plot the overall movement of this microsphere, you would see that the mean flow speed, is reduced when we induce acute hypertension.
So we think that could contribute. This is still very preliminary, but we believe it could contribute to the reduced lymphatic clearance and the association of hypertension with Alzheimer's disease. Furthermore, if you have chronic hypertension, you of course have a tremendous reorganization of both the smooth muscle cells, but probably also of the fibrous tissue surrounding the smooth muscle cell that could explain some of this event.
So what it can show, this is just basically going back to that we believe that this method gives very convincing evidence for we are not looking at diffusion. Because there actually is a diffusion we would predict that the particle would move according to this little radius here, but it actually moves much, much further. Another characteristic is that the microsphere always moves in one direction, and that's in the same direction as the blood flow is running. And one thing we noted, if we took a movie, so this is a movie of, normally we image for about 20 minutes, and we would basically integrate or just superimpose all the tractories of the microsphere we have tracked.
And one thing we noted when we did that was it looked to be huge. So normally when you look in the literature, you would see that the perivascular spaces are described as a space that's about 50 micrometers or less, fairly small. actually much smaller, about 10 micrometers.
But what we saw here was more something that looked like much larger spaces. And we could actually replicate that if you use two-photon imaging and just looked at the perivascular space here with a Dexan trace. it looked much larger than what's been described in the literature. So we did C-scanning to scan the tissue too so we could actually visualize, at least in the pial arteries, the perivascular space don't seem to be...
donor-shaped tunnel, it actually seems to be two triangles laying on each side of the arteries. We don't know why it is so, but you can actually here also see it's very few times that we observe the particle crossing the vessel wall. What we observed was, and this is actually very interesting for the Alzheimer field, what we observed was that the velocity is fairly fast and even along the artery. But if we get particles that move into this branching point here, basically the particle particle would stay there and never move. That's why they're not plotted here, because it did not move.
And this is actually where you would see amyloid plaques start to develop as in this branching point here. So anyway, we also looked at post-fixation, because you're interested in trying to explain why would the tissue samples, the histology, why would the perivascular space look so different and be so small after fixation. And you can see here, if you look at a C-stack to the fixed...
tissue, you would see there's basically no perivascular space. It's collapsed. Not only is the artery collapsed, as we would expect, but we also have a collapse of the perivascular space.
So these two differences are really dramatic, and we also seem to see an uptake of the perivascular traces, especially if you use dextran, into some cells or some fibrous tissue after fixation. So in order to approach that, this is just a quantification that we actually see a tenfold decrease in the perivascular space area looked from the top after fixation. And this is why we're interested in it, because many modeling studies have used the data that's derived from fixation and come up with different models for perivascular flow. And there's actually several people working at perivascular flow going the opposite direction of what we see, which always, when we do in vivo imaging, that it flows through the blood, always in the same direction of the blood.
So these... parameters are actually important for modeling studies. So we set out to actually image the process of death. So image what happened to these perivascular tracers when we fixed the animal.
And this is a study here. So here, just before I show the movie, I'll just say that at this point here we are five minutes into the imaging. So we have already filled the plasma with a red tracer and we have a perivascular tracer here in green. We have again injected an interstitial staining mesh. and we wait about five minutes to see it nicely around the trunk of the middle cerebral artery, which is here.
So what you would see is that we first image, and it looks fine. Then we start to use basically clean blood out by infusing PBS, as we all do in perfusion fixation, and then we would start to perfuse with performaldehyde. So it's indicated in the movie when I show it.
You would first see it's flowing nicely as before, and you see the aggregate of the dextrose. are slightly, what's called, aggregated. You could actually see when the animal was alive.
Now it's starting to die because we are washing blood out. You saw the red traces leaving. So it's starting to die, and you now very soon see that the smooth muscle cells constrict during the process of death. But you also see that the dextran aggregate starts to move somewhat in the opposite direction.
And during the constriction, it becomes quite prominent that the move is... you see fairly high velocity moving the opposite direction. And we think that, you know, the artery constrict.
And at this point here, we think now we perfuse paraphernalia hide, and you'd see that the perivascular tracer that was very nicely located in the perivascular space now actually transfer into the basal laminae in between the smooth muscle cell. You start to see lines there. So this suggests, and I think that's a very important point in our studies, is that...
Looking at histology and fixed tissue cannot tell anything about tracer movement in live animals. We really have to use live preparation when we can get away with it and try to understand how perfusion fixation works. I was very surprised about this redistribution of the tracer that occurred during perfusion fixation. So this is just, I don't think I have to show it again, this is just a small cartoon here showing how to do it. showing what we saw.
So what we see in the alive mice is that we have the perivascular green tracer moving in the perivascular spaces along with blood. When we perfuse with PPS and the animals start to be exposed to hypoxia, We start to see the treasure moving in the opposite direction, possibly explaining the controversy in the field where Other groups have suggested we actually have movement in two directions within the perivascular space. We have argued that it's not possible to have that, because you cannot have fluid moving in two directions within one compartment. But we think that might explain some of the studies, and then we have the redistribution of the traces going into the perivascular space.
Then I'll just end up with just a few remarks of why has this not been described before, because we didn't come out of the blue. So it was actually a young investigator, Patricia Grady, that that in the late 80s and early 90s described basically what we are describing. But she didn't have any access to a microscope to actually visualize what's going on. So she was interested in axonal tracking, and she was injecting horse rat peroxidase into cats or dogs. And she noted that if she by accident injected into the cerebrospinal fluid and then killed the animals, just four minutes later, she would see the entire perivascular space label.
and this would be the whole brain and actually including the spinal cord. So she published these studies and people got really excited and there was a pretty strong field in studying the blood-brain barrier at that time. And these two very big groups at Brown University at NYU, they basically replicated her studies in rats and they concluded, no, this is wrong. If perivascular influx exists, it's inconsistent and minor. And...
basically the field died out, and Patricia Grady left science and became an NIH officer. What those two fairly larger groups did, and they are all big groups that become a bit arrogant, what they did was they actually opened the skull to observe tracer movement, and if you open the skull and don't close it with a cover slip, you are eliminating a low-pressure system. So they were basically wrong.
The good thing is that Patricia Grady has acted as a director for Institut of Science. of nursing, so she became a big shot at NIH, and she's actually implemented all the treatment now in the U.S. for treatment of chronic pain and cancer. So she actually went on to do something really important. The sad thing was I gave a talk a couple of years ago down at NIH, and she was in the audience, and she was happy, and that was fine.
It was no big deal for her. She went on in her own life. But the sad thing was that her colleagues came up to me and asked, did Patty really do this? She had not spoken about her research.
for 30 years. She kept it quiet. And I think that's really sad that she didn't do that.
The other thing is, does it exist in human brain? There's very good evidence, I'll start to round up very quickly because we are out of time, but there's a group from Oslo headed by Pierre Eide that have shown if you replicate the studies Helena Benveniste did in rats in humans, so you inject contrast agent into cerebrospinal fluid, you'd actually observe that the contrast agent enters the entire brain within 24 hours. hours. And the biggest increase actually happened when he sent the patient back home to sleep.
He hasn't started sleep deprivation, but he's doing it right now. The second and very, very beautiful approach is from Ulu University in Finland. And this is an investigator, Vesa Kivinivi. What he's doing is he's scanning using T1 imaging the brain at 10 hertz.
It's incredibly fast. So he really undersamples when he does it. So the brain is big.
big pixels, but he collects a large data set and then he takes this large data set and sorts the images after heart rate of respiration. So in that way he can integrate hundreds or thousands of images at one time point after the heart rate. And when he does that he makes these incredible movies of fluid flow in the brain because with MRI you're emitting photons and photons is in the brain that is water.
So here is driven by the cardiac cycle, but you also have very different type of movies if you image after respiration. And there's very good indication that respiration might be important in clearance or outflow of fluid, similar to venous outflow occur when we inspire and we make a vacuum, that that might be important. So I think that this has a lot of hope because this is non-invasive.
It's basically just putting a patient in the MRI for 20 minutes. And then... analyzing the data for a week after that. Then there's an investigator at NYU, De Leon, that did a beautiful study published last year.
And what he did was he used PET of Alzheimer patients. So this is normal control. And what he did was he injected a ligand, or a tracer for other amyloid or tau.
And then he didn't image, or he actually didn't use it, but he first started to look at pixels intensity fairly long time after he injected the tracer. So the tracer would bind to amyloid. And you would see, you would basically use it in the clinic as a diagnostic for Alzheimer's. So if it bind a lot and bind plaques, the patient has Alzheimer's. But what he did was, he said, I don't care about pixels within the brain.
I look at only extra cranial pixels. that show amyloid staining. And what he found was that if he looked at the control, he could see the efflux of cerebrospinal fluid with amyloid here in the olfactory region.
And then when he compared it to the Alzheimer patient, he saw that their efflux was reduced, and he suggested you could use this method to maybe detect early Alzheimer. It's interesting because... In an Alzheimer's patient, one of the first signs of Alzheimer's is actually that you lose your sense of smelling. So this is another, so many different imaging approaches are on the way.
So what I was trying to tell you was that I think these studies are most exciting from the standpoint of looking at non-neuronal cell in brain. So it's been established for hundreds of years that the nerve cell... changes activity during the sleep-wake cycle. So any first-year resident in Neurology can take an EEG and basically say, this is an awake patient, this is a sleeping patient, or this is REM or non-REM sleep.
You can subdivide it. So it's totally like global that all the nerve cells in the brain change its activity during the sleep-wake cycle. What this study suggests is that maybe the non-neuronal cells do the same. Maybe astrocytes during wakefulness are really large cells that support, so they're basically expanding to insulate synapses better and support the synaptic transmission is precise not only in space but also in time.
During sleep, astrocytes convert more to like a kidney-like cell that pump large amount of fluid out. And this is needed because you cannot have large amount of fluid pumping around in the brain if you want to be precise in your neural transmission. connectivity would be lost if glutamate diffused from one synapse to the next.
So I think it's a coordination of multiple cell types that explain why the brain is so unique with regard to sleep. And finally, I don't think I want to show that. I just want to say I mention all the people who have done the work, and I've also received funding from NIH and also in Denmark for these studies. So thank you so much for listening. I know I'm keeping you away from lunch.
Thank you.