BioEngineering Seminar Series: New tools to Character Novel Neurochemistry
Sep 5, 2023
It's my great honor to introduce today's speaker, Professor Allison Swidler.
Professor Swidler is currently the James Eisner Chair in Chemistry,
and he is also the Director of the School of Chemical Science in the University of Illinois.
He received his PhD in Chemistry from the University of Arizona in 1988,
and spent three years at Stanford before joining here.
Professor Swidler's research theme involves both nanochemistry as well as neurochemistry.
He developed a set of very versatile tools,
and also used these tools to study the distribution and dynamic release of neuropeptides
and cholesterol transmitters, as well as their metabolism.
Professor Swidler is one of the leading scientists in his own research area,
and he has authored more than 340 manuscripts and has delivered more than 350 in-body lectures.
Acknowledging the impact of his research, Professor Swidler has received numerous awards,
including the Ralph Adams Award from the Pittsburgh Conference,
the Field Award from the Eastern Anatomy Symposium,
and the 2014 ACS Award in Anatomical Chemistry.
Today, Professor Swidler is going to give a lecture about his recent work,
and so please join me in welcoming him.
Thank you.
Thank you all. It's fun to be here.
It's always strange to realize that I go around, like most faculty, and give many talks,
and then very few on our own campus about our research lectures and other things.
I'm an analytical chemist that studies the brain,
and so I develop new tools to understand how the brain works.
This is actually, we were just talking about this a minute ago,
it's become, I guess, a hot topic because this kind of fits in with Obama's brain initiative.
Actually, every president over the last 20 years has picked one area,
whether it's nanotechnology or brain or even the war on cancer, a little older than that,
and Obama's has been understanding how the brain works.
None of these initiatives have been successful if you mean that they've actually answered the questions they set out to
while the president was still president, so I know Obama's will be the same way.
We won't understand the brain anytime soon.
It's actually kind of fun to be here.
My first ever outside peer-reviewed grant that I ever got was a Whitaker grant from bioengineering,
and the chemistry department head said, what is Whitaker and what is bioengineering?
At that time, there was no department of bioengineering.
There wasn't much really going on here 20 years ago.
There were some very important people doing work,
and so I've always felt very warm feelings about bioengineering
because, of course, that was my first, as I say, funding,
which is, as a faculty, one of the important areas that we need to do.
So what I want to do is kind of go over some of the tools.
Now, from an engineering point of view, if you want to interfere,
hopefully alleviate, for example, disease in the brain,
you have to understand how it's functioning in the first place.
I usually end with this slide, and I decided to start with it because that way I have time.
The group has changed a little bit.
This slide shows outside of the Beckman Institute a few years ago.
It was on April Fool's Day, so the group was making fun of me and my hair.
My group really, you know, to join the group, you have to have hair that's bright colors.
So this is the group then.
I became an editor of an ACS journal, administrator,
running the director of the School of Chemical Sciences,
and the group shrunk a little bit.
It's probably in a growth phase, but these are the people that did the work.
I have a lot of collaborators, some of which are listed here
because I'm talking about their research.
So let me give a few slides of background about the brain,
and then I'm going to go into how we measure what's happening in individual cells in the brain.
So, you know, this laser pointer is not too bright.
This is obviously a picture of a human brain.
We don't do anything in the human.
So rats, mice, and other animals, I'll talk about those in a minute.
This is a network of neurons or a piece of artificial addition to the brain,
and there's two things.
One is brain cells are unusual shape.
They have wires connecting them.
The interesting thing is a microliter of mammalian brain has 100,000 neurons.
It probably has a couple hundred thousand red blood cells,
400,000, 400,000, 500,000 helper cells, astrocytes, oligodendrocytes, and others.
So if you take a microliter of brain and you analyze its chemistry,
you actually really have not analyzed what's in neurons.
You've analyzed all the helper cells. There's a lot more of them.
Again, there's more red blood cells in your brain than there are neurons.
So I show this slide.
I have somebody saying, are you trying to say that my student's brain is mostly empty space?
I guess that's another alternative explanation.
The other way to look at this is that even though the wiring in the brain is important,
you can't really change that for an adult too easily.
So it's the chemistry that is the mode of pharmaceutical intervention.
The chemistry occurs in synapses.
There's a picture of one of synapses truly nanoscale.
You see in textbooks it's 30 to 50 nanometers or so.
It depends on the synapse. Some are much larger.
So it's the chemistry that we can do something about.
To be honest, I never worked with drugs of abuse in my research.
The National Institutes on Drug Abuse contacted me
because of the millions of compounds in nature,
the 100 or so that affect this type of chemistry are known as drugs of abuse.
They hit the same types of circuits that the molecules in your brain normally do.
One more basic slide.
Again, most people think of cells as small.
I have cells in my body that are in my spine that go all the way up to my brain and down to my toe.
They're not straight.
So even though I'm not quite that tall, they could be two meters long.
They're not part of the cell cycle.
I'm more than 50 years old.
My neurons in my spine are 50 years old.
When people talk about cells as small and they talk about how they keep on changing,
remember neurons don't follow any of those rules.
You could even ask questions about how neurons take insult,
chemical, physical problems that keep on accumulating during your lifetime.
Whether or not a message goes from the cell soma, the cell body, down to a terminal
depends on all these other connections in and around the cells.
Whether or not this signal continues depends on everything that's happening in and around there.
We want to be able to measure that.
When I started this work 20 years ago, I thought all the neurotransmitters,
all the neuromodulators were known.
I talked to lots of people in bioengineering.
They look in a textbook and they say, here's the growth factors.
What I kind of want to do today is show you the techniques,
but also tell you in every single system we've ever looked at,
many, many neurotransmitters, neuromodulators seem to be not known.
In many cases, they're quite a surprise.
Everything about 20 years of work has really been dealing with complexity.
Chemical complexity, spatial complexity, and temporal complexity.
How do you do that?
One is obvious.
I'm an analytical chemist. I like measurement tools.
If you want to look at small molecules in the cell, you need a big tool.
I think it's an unwritten rule, the smaller your sample, the larger the tool you need.
If you want to look at really small cells, you need to have lots of money.
I'm not really sure.
That's true here for fabrication.
The smaller the line you want to make on a device, the larger the tool you need to make it.
We have some commercial mass spectrometers.
This is a time-of-flight instrument.
This is one that Eric Lanny is just finishing at the Beckman Institute.
It started on a commercial instrument.
It's been home-built or laboratory-built for four or five years.
This is an ICR at the IGB.
We have mass spectrometers.
These work great.
This is a large part of my research group.
I'm not going to go into how mass spectrometers work too much in this talk.
This is a fun one that nobody really thinks about.
This has been hammered home, actually, as I look at science over the last hundred years.
If you want to cure a disease, if you want to understand something,
step back and think what you really need to know.
When people wanted to understand electrical activity in humans,
some people started working at mice and rats.
Other people did other animals.
The group that was smart enough to go to the squid giant axon got a Nobel Prize
because they could fill up something this long with seawater,
change the sodium or potassium levels,
squeeze it out, and do it again.
If somebody wants to do a genetic study on humans and want to do ten generations,
we'll come back in, what, 200 years?
Of course, use a mouse.
Then it's maybe a graduate career.
Work on a fruit fly, which in some ways is a completely, I'll get people mad,
irrelevant, lousy model in every way, except the generations are really quick.
You can manipulate it.
Because of that, it's worth working on.
Biology, over the last hundred years, has had a number of successes
with, in some ways, lousy models that answer really important questions.
This one got the Nobel Prize with Eric Kandel for learning and memory.
I'm going to show you some data.
This is some work with Rainer Gillette.
I'll just give you one example because it's lunchtime and you're not eating candy.
You're listening to me.
This is a pleurobranchia californica.
It's a sea slug.
It's under seawater, and it eats a Spanish shawl.
Now, this is bright purple and orange because it's poisonous.
If you're a fish with eyes, you'd say stay away from something purple and orange.
Unfortunately for this sea slug, this guy, or this hermaphroditic self-fertilizing sea snail,
doesn't have eyes, tries to eat it, gets stung in the mouth, and spits it out.
Its brain, which has maybe 1,000 neurons related to feeding,
remembers never to eat one of these again.
So they can learn.
They don't learn bioengineering.
They don't learn English language.
They learn how to find food, how to avoid being eaten, and how to find a mate.
But this is a great model for learning.
I've done work with Gene Robinson on plastic changes.
You change the expression of a single gene in a bee,
and you can go from a nurse to a forager, and then mice and rats.
And this is just simply to remind people we're at the University of Illinois.
If you ever see a smiling cow, run away.
We have actually done all of these models.
And I will say, here's a good example.
This is why I like working with sea creatures.
This is found off the coast of the state of Washington.
You have to go there over the summer collecting animals.
You can't let anybody else do it because then you don't have an excuse to go to the beach.
And this animal has two behaviors.
It can swim, and it can swim.
Now, as a chemist, I can keep track of those.
One is feeding swimming, and one is escape swimming.
And you can look at how that changes.
And so every one of these animals has good reasons to study it,
and they make the answers much clearer than if you work with something like this or even a human.
And so these are really fun.
The last way you can simplify is to go smaller.
This is a honeybee brain.
It's a whole brain, but it's a honeybee.
This is about a million neurons.
This is a million neurons from a brain slice from a rat.
And I bet some of you work on neural nets when you do programming.
Well, this is a real neuronal net work.
This is the cells that control reproduction in a sea snail,
except we put them on a microscope slide.
And I want to know what changes as this goes into fictive reproductive behavior.
And so you want to know what chemicals are there,
and when they're released,
and so chemical, spatial, and temporal information,
and down to single mammalian cells.
So we oftentimes work here and move back and forth using a zoo.
In the last five years, I think we've had, you know,
well, now we're working on monkey brains and birds and planaria and snails,
and actually we have a paper being reviewed in Nature on ctenophores.
That's comb jellies because it's the animal with the simplest nervous system that exists.
Horses are slightly simpler animals, but they don't have nervous systems.
So we work at a zoo,
and my group oftentimes thinks that they're the zoo when I refer to them that way,
but it's the animals.
So we work on a range of animals.
We try to make and understand what's happening in single cells.
Single cell measurements are hot.
If you go to NIH, they have something called director's cross-cutting initiatives,
or now, you know, there's the common fund is another way to say it.
So the director of NIH picks areas like, to be honest,
last year metabolomics and single cell measurements and all the areas that I work on.
We talked about that a minute ago.
That worries me because once you're working on an area that's really hot,
it means in two years it won't be.
But right now, single cell measurements are considered new.
I think that's kind of funny.
Leon Howick, a few years ago, came up with the first microscope,
and he published some notes in the Royal Society in London,
and his notes basically identified things like red blood cells,
single-celled organisms in pond scum, semen, lots of other cells.
He just discovered an amazing number of cells.
Cajal analyzed neurons, and if you look at this, there are all sorts of different shapes.
To be honest, these people are the ones that got to name the cells in our brain,
and they got to name them using things like pyramidal cells or stellate or others.
He just named them on shape.
And so people think that all the neurons of a similar shape are similar.
They're not.
If we do chemistry, we'd come up with a different naming system.
About 50 years, people have been doing single-cell chemistry.
This is a crazy example I love because it was so ridiculously ahead of its time.
This group did single red blood cells.
They published this paper in Science, and then every one of them quit
and changed to another field because it was so hard.
But they took a silk fiber.
I'm actually not joking.
They took a silk fiber in a humid environment.
They put one red blood cell manually on one spot, put 1,000 volts across,
and then the proteins would move across the silk fiber.
They couldn't even buy capillaries easily at that time.
And then they would detect things using a weird interferometric optical approach.
So what's changed in 50 years?
Mass spectrometry, among others.
So we do a lot with mass spectrometry.
I will show you an example later on of mass spectrometry
analyzing a type of brain chemical called the neuropeptide.
In 1970, Guleman isolated a neuropeptide from a sheep,
and it took him 4 million sheep hypothalami,
the brains of 4 million sheep, to isolate enough material to analyze.
Now you can do that in a single cell.
So that's 15 orders of magnitude improvement in sensitivity.
I always am amazed by that, but I'm also amazed
that I can imagine the graduate student who works with the slaughterhouses,
and they grab 4 million sheep hypothalami,
they have the final thing going from bats down to barrels down to a single flask,
and you show your research director the flask
containing all the biologically active peptide,
and you're ready to do the final purification, and you drop it.
You know, there you go.
Oh, I guess I need another 4 million sheep.
Anyway, things have gotten a lot better,
and it's because of mass spectrometry.
I'm going to not go into details too much,
but this one allows you to do single metabolomics
or the small molecule content, and these supply imaging.
This one, SIM, Secondary Mass Spectrometry, exists at MRL,
so anybody can use it if they get trained on it.
And all of these allow you to look at what's happening in neurons.
And so ignoring this picture, I'll show you how well it works in the next few slides.
So here's our sample.
You can take a creature.
This cell is green. Isn't that convenient?
So we're going to isolate that green cell in a droplet and analyze it.
By the way, some of the experiments we do are heroic.
What I mean by heroic is they're much harder in this creature than in a mammal
because no one would do this.
We've worked where at Columbia University,
one of the groups that's actually got the work on the gill withdrawal reflex,
the part that got this the Nobel Prize,
and they stick an electrode in, and they stimulate it,
and they see if they get the right behavior.
If they don't, they pull the electrode out and stick it in.
Pull it out, stick it in.
Eight to ten hours, they find the cell of interest.
So we're not doing single cell measurement like you take a cell off of your skin.
We're using one cell in the entire creature.
It has to be that cell.
They then take this brain with the green cell, put it on dry ice.
No, actually, I'm sorry, regular ice, so it's still liquid.
It was seawater.
Send it to us.
We isolate the cell.
Now it's not looking too green, and we analyze it.
So these are hard experiments.
The hard part was done somewhere else and here.
So this is low throughput, hard experiments.
I decided to do one example of not mass spectrometry
because I thought most people, a lot of people here use optics.
The type of separation we do is capillary electrophoresis.
Imagine a hair.
It's the same diameter, filled with seawater.
You put 20,000, 30,000 volts across it.
Call it 20,000 volts.
Charged molecules migrate.
Positive go towards the negative side.
Negative go towards the positive.
And if you stick a cell on this side, you get things falling out the end of the capillary.
If the end of the capillary is a UV laser, instead of green, it's, say, 257,
anything aromatic you can detect.
So I had a grad student build this instrument with Whitaker money,
and we're still using aspects of this.
Actually, I don't know how many people saw Warren Chan when he was here yesterday
or earlier this week.
So Warren is the one that actually really built this when he was an undergrad in my lab 20 years ago.
And so Warren built this device.
He thought it was amazing that it still was there.
So he started out as an analytical chemist before, I guess, seeing the light
or whatever you want to call it.
But anyway, this is Warren's and Aaron Timperman's device.
So Warren got an undergraduate degree and Aaron got a Ph.D.
I guess that's the way it works.
And this was done a little bit later, but this is now when you have these sea slugs,
wild-caught off the coast of California.
We pay a surfer dude.
The business office hates this, but those are the only people that will collect them.
And so if the waves are up, they surf.
If the waves aren't up, they catch these animals, ship them to us for money,
and we get the animals.
But these are pleurobranchia.
And we take one cell, the metacerebral cell, out.
We analyze the serotonin levels.
Now, you're going to say, what's serotonin?
If you're in America and you look at the popular press,
you've heard of SSRIs or serotonin reuptake inhibitors in depression
or monoamine oxidase or MAOIs.
Every one of these drugs are worth a billion-plus dollars.
Actually, I think $10 billion a year because they change your mood,
and most people like to feel better.
So serotonin in every animal exists, and it also has something to do with feeding.
And this is the first time anybody ever showed that serotonin levels in a cell
can actually regulate an animal behavior state.
Lots of variability, but these are wild-caught animals.
I know some of you will probably work on rats and mice.
These are even more wild.
I mean, these are wild animals.
We don't know anything about the history.
The student who did this work did it blind,
meaning Nathan Hatcher actually fed some of the animals.
We catch them off the coast of California.
They're shipped to us.
We starve them for a month, and then we randomly feed one squid,
and one stays hungry, and we make these measurements.
So this type of measurement can be made because we have an instrument
that really only sees serotonin, HIA, and a few other compounds.
Just to give you one example in a mammal,
if you take a cortex of a rat and you add serotonin, you see serotonin.
You know it's serotonin because of its fluorescence properties.
It elutes at 7.4 minutes.
But we also, HIA, a catabolite, the conversion of this to this
is what some of the drugs deal with.
But we actually found an unusual, unknown molecule that we now know
is 5-hydroxyindophiazolidene carboxylic acid.
I put this up because there's an example of a molecule
that wasn't known to be in mammals.
Is it important?
Well, every one of the catabolic processes of this molecule
actually are the site of a lot of pharmaceutical intervention.
So this may be important.
The reason this works, though, is there's 10,000 compounds here,
and we only can see those that have this indole ring.
So I made a very selective instrument
that only detects this part of the molecule.
So that works if I'm studying serotonin.
It doesn't work if I want to see everything else.
By the way, those are big cells.
Here's a fruit fly.
This is just kind of showing you
how good some of my students are at manipulations.
I'm drinking coffee.
Some of my students never touch coffee,
and they have hands that are unbelievably fine.
And so you can take a single neuron here.
They're green again. You can pull one out.
They're green because we've labeled them using molecular techniques.
And you can actually now see serotonin
and some catechols in a single fruit fly neuron.
This is unpublished,
but the difference between a 500-micron neuron
and a 5-micron is 100-fold cubed,
or a million-fold less material.
My student takes this cell out
and actually manually sticks it into a capillary
the size of a human hair.
And so if you're good at manipulations,
you can do some neat work.
Now we'll go to mass spectrometry.
So there's the green cell.
What's in the cell?
And so rather than pre-select for serotonin with a laser,
like the laser-induced fluorescent system I just showed you,
we're doing mass spectrometry.
That gives you supposedly everything.
I'm just giving you one example here,
acetylcholine.
Now, acetylcholine is a neurotransmitter
that is a pain to detect
because it's not fluorescent, it's not electroactive,
it doesn't have a UV absorbance.
I don't know how to analyze it.
So I've spent 20 years,
or actually about 18 years of my life,
ignoring acetylcholine because I can't detect it.
And so the best thing you can do
when you write grants and publish things,
if you can't detect a molecule,
you just pretend it doesn't exist.
And that works, actually.
We've gotten funding for molecules that we can detect.
But we finally have mass spectrometry working
in the last few years.
And this is acetylcholine.
I know it's acetylcholine because of its mass.
And even though it's a single cell,
I get a good signal-noise ratio,
I get good resolution,
everything works really well.
Just to show you one other thing.
We're measuring the mass of the molecule,
and it's 146.12.
Here's one single cell.
It's a mass, or electrophyrogram.
So things migrate out based on mass to charge.
This is intensity.
And we actually have four compounds in that cell
that have a mass of 146.12.
And only one of them is acetylcholine.
So this is this weird technique called tandem MS.
You basically take your molecule
and you smash it on the ground
and you look at the pieces
and only acetylcholine breaks into that piece.
So we can do this for any molecule
that's at high enough levels in a single cell.
And every cell is different.
I've had many people ask me this.
Okay, neurons are neurons.
Well, these electrophyrograms
with different colors representing different masses
and acetylcholine, serotonin, glutamate, betaine,
we actually detect 200 to 400 compounds in every cell.
I don't know how to show you that.
So I just picked important molecules here.
And every cell is a little different.
One cell in the entire creature, the R15, has this mass.
I've never been able to figure out what it is
because it's only found in one cell
and there's not enough to identify it.
But it's probably important.
Every one of these molecules, by the way,
are found in your brain.
So I actually, I guess I haven't really emphasized this so much,
but every animal we've looked at,
when you go from a coral and a jellyfish to a human,
the neurotransmitters are the same.
And that makes what I'm doing possible.
And so who knows about that one,
but most of the others are really the same.
And you can get data that looks like this.
And I actually am showing you this slide
because my students usually misinterpret it.
But this is the metacerebral cell.
It's the only cell that has serotonin.
And my students will come to me and say,
look at all the biological variability.
Each animal has a different amount.
And I like this slide because it turns out that's not true.
What I already showed you is this variability
tells you how hungry the animal is.
It's not really random.
And so the important point is when you measure something
and you get a lot of scatter,
maybe it really is variability,
but maybe what it is
is you're not controlling the right variable.
We've actually had other cases
where we see a huge amount of variability,
and then I discover if we always sample at 8 in the morning
when the lights come on at 6,
then the variability goes away.
And so biological variability may be just that,
or it may be something you haven't controlled.
Actually, with serotonin I didn't show it.
Serotonin and serotonin sulfate does follow circadian rhythm.
We actually did a paper on that.
That was actually because I couldn't understand serotonin sulfate.
I had a graduate student making measurements
that was free-running.
He sometimes would make measurements
at 3 in the morning or 5 in the morning
or 7 in the morning or 1 in the afternoon.
And after seeing values that went from non-detectable to micromolar,
after doing this 50 times,
he did something that was quite important.
He looked at his notebook very carefully
and noticed that luckily he recorded when he took every sample.
And he noticed that this followed a circadian rhythm.
We didn't think of that when we started
because it's not melatonin.
It was serotonin sulfate.
So yes, there is a circadian rhythm.
In our body, there is a seasonal rhythm
between melatonin, which is made from serotonin,
and acetyl to melatonin and others.
And there's a seasonal variation.
Let me show you one example
because I know people who work with cell lines
tend not to like to hear this.
There's two neurons in aplesia,
the last example I'm going to give you from the sea slug, I think,
that are called B1 and B2.
There's something to do with buckle and feeding,
so they're 1 and 2.
And we isolated them from an animal
and we analyzed them as fast as possible
or we cultured them for one day.
So all of these cells are freshly isolated,
one cultured for a day and one not.
Here are the B1 cells and the Cs,
meaning they're cultured.
And the chemistry changes.
Now, this is a weird plot.
It's principal component analysis
looking at all 300 to 400 peaks that we detect.
Actually, in this case,
I think it was just the 50 most intense peaks
that between these and these are separating,
meaning there's something different about the chemistry,
the levels between the B1 cells and the cultured B1,
but for the B2 cells and the cultured B2,
they're not different.
And the reason I'm showing you this is this slide.
B1 cells and B2 cells are distinct.
They do something different in the animal.
And here they are, the B1 and the B2,
separate on this principal component plot.
But after one day in culture,
whatever made them chemically distinct is completely gone.
So this isn't a cell line.
These are freshly isolated cells
that were only cultured, primary culture,
for one day,
and whatever made them different is gone.
And let me tell you,
we're good at culturing these cells,
and we can't keep them.
So whatever happened in the intact animal disappears.
So as you're doing experiments in microfluidic devices,
and we do that all the time, and others,
be aware there's nothing less physiological
than culturing a neuron in a device.
And chemistry changes very quickly.
And I don't know what the answer to this is,
but to also sometimes work with intact cells,
if you can get away with it.
Things change quickly.
Here's the hardest experiment we're doing.
And the names, completely ignore the names,
but we're working with two brain regions.
This was with Lee-Cox, and we're still doing it.
The ventral basolateral nucleus
and the thalamic reticular nucleus.
Okay, that's a mouthful and doesn't matter.
We're interested in why we see GABA where we shouldn't.
So this is the enzyme to make GABA.
I'm showing you this, not for this data,
but to say we went to a rat,
and in the two areas we can do this mass spectrometry,
and in this area we see acetylcholine, GABA, and glutamate,
three important neurotransmitters.
And in the VB nucleus we see others.
Okay, nothing new about that,
but I wanted to take this to single cells.
We decided to try and see if we can know
what cell we're sampling.
So we're using something called patch plant method.
Anybody here heard of that?
No, good.
So this is patching onto a cell.
Actually, Nehren-Sackman did this in, I think, 1991,
got the Nobel Prize for it.
Actually, when you do this,
you have access to the cytoplasm,
and we can suck up a little bit.
So here's the physiology of the cell in the brain slice.
Here's another cell.
Some are spontaneous active, some are not.
But we can suck up.
This is the original cell,
and the movie is so hard to see,
and this is hard to see,
but it's 9.6 by 8.5 microns,
and after we suck up about 3 picoliters,
it's 8 by 7.3.
The cell is alive the whole time.
If we really do things right,
we can actually get metabolites from those cells.
So here is the physiology from spontaneously active cells,
from TRN neurons, VB neurons.
This is an astrocyte.
Here is the shape because we injected a dye in,
and these are just partial metabolomic traces
showing you GABA and glycine and serine and other compounds.
So this is the smallest sample we've ever done.
It's the first time I've ever seen anybody do something
where you actually take a living cell,
you can actually do physiology, morphology,
and chemistry at once.
So this gives you a lot of ability.
For example, NIH has all these calls for proposals
because no one knows how different astrocytes are from each other.
People like to say astrocytes are astrocytes.
I don't show you more data
because we've now done about 10 astrocytes,
and the chemistry in different astrocytes is amazingly different.
We see different compounds,
and we haven't done enough to understand what we're seeing.
So this is known because of this part of it,
but the chemistry that's occurring in astrocytes
is pretty very poorly known.
Okay, that's one type of mass spectrometry,
and everything I showed you is what's happening in cells,
but isolating the cell is hard.
So now we're going to go to imaging.
Now, I bet a lot of people here use microscopy of sorts,
fluorescence microscopy.
You can actually do mass spec imaging.
It's horrible resolution.
It's lousy imaging.
What you do is here's a brain slice,
and here are a bunch of red dots.
You can shine the laser there or shine an ion beam there.
Things will explode off,
and you can do mass spectrometry on it.
I move it over 20 microns or 20 microns.
If I do a thousand by a thousand array,
I get a million pixels,
each one at 10 micron or 20 micron resolution.
Nobody in their right mind would do this,
except at each pixel I get an entire mass spectrum.
So it's lousy spatial resolution.
It's hard to do,
but you get an amazing amount of chemical information.
And so we can use a laser.
See, I'm using one up here,
and I get about a micron,
really usually five microns or so.
I can do proteins sometimes.
I can use a gold cluster.
That's certainly for others.
We're using buckyballs up at the Beckman Institute,
and you get a smaller area,
but it's actually smaller molecules.
You can take a single cell.
There's a cell,
and you can actually see lipids.
You can see vitamin E localization,
and this works quite well.
So you can image cells and their lipids, for example.
You can even see what happens
as you go towards a simple neural net,
and you see what changes.
So this is these three cells.
Here's one of them, and here's just a terminal.
And we get some information on phospholipids.
And so this is some work
that we're really just, you know,
ongoing and trying to make work.
You can combine these.
This is actually some...
We're still continuing with work like this.
You take a section of the spine.
You do SIMS for lipids.
You take the next section,
and you do MALDI for peptides.
These are peptides here that control behaviors.
You take the next six sections,
and you do chemical analysis,
and you can work your way down the spine.
And we've done this
and found lots of different things.
This is the first images
from a new instrument we just built.
As I say, it took about five years to build.
And the reason this is good
is because it gives us more information
about some of the lipids.
So it's this buckyball SIMS imaging.
It's on the second floor.
And you can actually get information
on higher-quality mass spectra
than what I was showing you.
So here's two cells.
The imaging resolution is not very good,
but the mass spectral data is quite impressive.
There's lots and lots of lipids in that area.
Just to give you a feeling for this, by the way,
LipidMaps, a consortium of groups
that NIH has funded,
basically, you hear that there's 20,000 genes
in the human genome.
There are 600,000 to 800,000 molecular entities
in your human brain that are called lipids.
So different length chains,
different sites of unsaturation,
different head groups.
And the other thing to think about it,
and everybody gets surprised when I say that,
is if you wanted to go into your refrigerator
and find the item that most resembles your brain,
and you had a chicken breast in there,
for example, most people say,
oh, that's a good model.
No, to be honest,
your brain is like a bar of butter.
A little bit of protein,
a little bit of water,
but it's about 70% lipid.
It's oil. It's fat.
And those lipids turn over,
and a lot of function of the brain depends on them.
And there's no easy way to analyze them.
So the reason these types of tools
you're starting to see a lot
is because no one really even understands
how lipids in the brain turn over
or how they work.
And so I will say,
when I tell some people that,
they get very upset
because they don't like to have their favorite,
most prized possession of their brain
be compared to a bar of butter.
I came up with that
because I'm an editor of a paper,
and a reviewer and an author
are going back and forth,
and somebody actually used chicken breast
as their background approach
for their mass spec imaging.
And the reviewer said,
you shouldn't be using a chicken breast.
You should use a bar of butter.
And the reviewer got really, really mad,
saying, you should throw this reviewer out
because he's comparing my brain to butter.
So I looked it up,
and the reviewer was right.
Anyway, so let me end
with a little bit about neuropeptides.
Neuropeptides are the hardest molecules to analyze
because they're present and active
at really low levels.
I already mentioned they used to take
a lot of work to isolate.
Some classical textbooks say
that four of the seven deadly sins
are mediated by neuropeptides.
You know, gluttony, attraction,
and things like that.
So they're important, or not.
Mass spectrometry has gotten really kind of fun.
You can take a section of your brain,
you can isolate a cell,
or here's a neuron.
If I put TNT, the explosive, around the cell,
and I shine a laser here and it explodes,
I get a spectra right at that spot.
And that spectra,
every one of these peaks
are the peptides that were in that cell.
It really works well.
This is sensitivity.
This is a chemical printer
printing spots of the peptide angiotensin on here,
starting out at 60 femtomoles a spot,
down to three,
keep on going down to 300 zeptomoles per spot,
and we still can detect it.
Our detection limits are about 100 zeptomoles.
So people say, what's a zeptomole?
It's 10 to the minus 21 moles.
That's really good,
except the zeptomole is also 600 molecules.
So this is about 100,000 molecules or so.
So it's good, it could be better,
but it actually allows us to see what's in a cell
in terms of peptides.
This is older data,
but we're the first group that ever did things
like take a single vesicle in a single neuron,
in this case a big one from aplesia,
and we can see some of the peptides are present in there.
Some vesicles that look different don't have peptides.
You can look at a cell in a process,
and everywhere we stick this explosive matrix,
we get peptides.
And so you can see lipids and peptides
going from a cell soma down a process.
And this kind of highlights that type of information.
Here's a bunch of cells that we labeled with an in situ probe.
You can pop a cell out and do mass spectrometry on it.
And that cell is missing because we actually analyzed it,
and we got the neuropeptides that are present.
We've done this probably 40 times
and discovered for publications
like the Journal of Neuroscience,
the neuroscientists don't even know
that single cell mass spectrometry is weird or difficult.
They just come to accept the fact
you can do physiology and mass spectrometry.
If I know that this cell makes this animal go into swimming,
when I do this and I discover a new molecule,
I, even as a chemist, can say
maybe this molecule has something to do with swimming.
And then we test it.
So we've discovered literally thousands of brain peptides,
but we actually have about 100 that we know are neuropeptides.
We have biological activity.
And just to kind of end this type of thing,
you can go even smaller.
This is a dorsal root ganglia, those cells in my spine.
As I say, I like these because they're such strange neurons.
They're long.
They're probably the largest animal cells.
Let's see.
Sometimes I show a picture of a blue whale
because a DRG neuron in a blue whale would be 100 feet long.
That's a single cell.
And so we can look at those in many ways.
Here we've taken a single process from a single cell,
and each one has different chemical profiles.
And so this one has one peptide.
This has a bunch of unknown peptides.
And so we can make this work really small.
I don't know any other group that's doing anything like that.
We can use, in a mouse, molecular approaches.
This is a green cell.
We're good at isolating green cells.
And every one of those,
we see the peptide profile of different green cells.
And it says here, how do those change with pain?
You're probably thinking that about now
towards the end of my talk.
When is pain going to end?
We can take a dorsal root ganglia group of neurons,
separate it in a microscope slide.
There's some cells right there, there, there,
pieces of fibers.
You can separate the fibers out if you want.
But we can image this with mass spec imaging.
And this is 1,000 individual cell spectra.
This is 300 that were undergoing a painful stimulation.
We injected something painful into the foot of a rat.
300 on the other side, water injection and control.
Four neurons.
This is a mass spectra from 340 to 348 Daltons.
Four neurons had a huge change.
Went from, you know, here.
But actually, the other 300 or so didn't change.
So we can find the rare cell using this approach.
And this is the same data shown differently
using 3-axis principal component 1, 2, and 3.
Here's 1,200 or so cells.
And we're interested in the outliers up here
because they have different chemistry.
This is different because we're looking at one mass,
in this case, of one of the peptides at 1,100.8.
And we can actually take that cell,
one of these cells of interest,
and analyze it again because most of it's still on the target.
And so we can take that weird cell
after we've done mass spectrometry.
90% of the material is still there.
And we can then do CEMS on it.
So this is capillary electrophoresis mass spec
of the metabolites after the last data.
Now, let me show you.
I have two students trying to make this better.
One actually is mostly working on software.
These red dots are the...
I'm sorry, these black dots are the cells.
And this is rastering the laser in a normal beam.
And you obviously, wherever there's a red dot,
you've got a little bit of the cell.
But what we're doing now, instead of this approach,
is we actually image the locations of the cell
and only shine the laser at the cells.
So we're not doing imaging anymore.
We're actually just finding the cells.
So this took us a while to make work.
This is a really recent data set.
But we imaged using dark field and autofluorescence.
We found the cells.
And then we did a three-by-three imaging raster around it.
And we hit the cells really well.
And wherever we hit the cells, we get a mass spectra.
And this is the most recent data.
This is 2,000 cells.
We did a principal component analysis on it.
So here's the target with lots of cells all over.
This is principal component analysis.
And most of the cells are here.
But some of them have weird chemistry.
And I'm just showing you random spectra
of a few cells around that spot, that spot,
in different locations
from this principal component analysis.
So if you don't like the brain,
and you've forced yourself to sit through
awake this entire time,
imagine looking through, for example, using cancer.
And you wanted to find the rare cells
that maybe first started to change.
It doesn't matter what it is,
whether it's cancer, whether it's developmental
stem cells, or the brain.
This is a way of probing with chemistry
large data sets,
and then trying to extract information.
The problem really is that it takes us
actually a surprisingly short time to get this information.
And we're just learning how to go through
2,000 cells, each one with a gigabyte of data,
in terms of the mass spectra,
gives you terabytes of data
in the wrong format to do this.
And so we're still learning how to make that work.
I want to conclude,
I've got one more example after this,
but just in terms of single cells,
are we actually done? Have we done enough?
And the real interesting thing is that
from a proteomics point of view,
you can't do single cell proteomics.
RNAs, because of everyone and others,
you can do single cell transcriptomics.
And we're really just starting
with metabolites and lipids.
But if you look at a cell like this,
it's not even the single cell here,
we want to understand the differential distribution
throughout this creature.
How much time do I have?
Do I have three minutes, five minutes?
What?
What? Okay, perfect.
So I'm going to show you, I'm running out of time,
I'll give this example, some work with Martha Gillette,
just to give you more of a functional context.
From every animal around,
time controls everything.
So if you think of it, it's obviously eating, sleeping,
waking, drinking, heart attacks.
To be honest, the National Institute
on Drug Abuse thinks it's important that people
abuse drugs more at night.
I always think that's really kind of weird because
there's more crime at night probably because it's dark.
But they're convinced it's circadian.
And so we're studying that also.
So how does this work?
With Martha Gillette, we're working and studying
the signaling of the suprachiasmic nucleus.
This is a small
10,000 neuron organ
in the brain.
It controls all the body's quarks.
If you take a rat, now I'm showing you a picture of a human,
we didn't do a lesion study in a human,
but if you take a rat, if you lesion,
remove the SCN in a rat,
the animal loses circadian rhythm.
Even 20 years ago, it was known
if you took another SCN and stuck it close to there
and just let it free float, you would restore circadian rhythm.
That implies there's something
being secreted by the SCN
that can entrain all the body's quarks.
In 2005, Science magazine published
the 100 unanswered questions,
top 100 answered questions in science.
Now they can't count because there's 125 answers
or problems.
I would say
100 of them were really good.
What is the origin of the universe?
What is consciousness?
They're great questions.
I just don't even know how to study them.
One of them was, how does the SCN
control all of the body's quarks?
As an analytical chemist, I said,
I can do that one.
We started looking at that.
I'm going to show you progress
in about eight years in this area.
The first thing we did
was we measured the peptides in it.
I hate long lists.
We found 102 peptides
that were present there
including 29 that people
didn't know about. Some were phosphorylated.
This is a smaller list.
It's too high. I don't know how to deal
with 100, so we have to simplify it.
This is one of the approaches to do.
This is a living brain slice.
You take a slice out that has the SCN.
There's little white dots because we put
little collection beads that can
have onto peptides that are being secreted.
We then wash them
off and we measure, pretending like it's
a cell, we measure what was in that bead.
When we do that, we can see
different things released at different places.
This is arginine vasopressin, for example,
near the SCN or the SON,
and not in between. I'm showing you this
because of this type of data.
This is during the day or night.
Arginine vasopressin
is released at this
time of day and
this mass, 1495.75
at night. There's no known peptide
that does it, so that maybe is important.
We're getting some information that's
kind of neat. You can
figure out what that mass is, I told you,
1495, using all this
complicated tandem MS.
We fragmented it, and this is just a way of
saying that with a ridiculously good E value
of about 10 to the minus
7 or 10 to the minus
11, we're able to show that this is cerebellum,
a name of a peptide that was known to be
in humans and rats and mice but has no known function.
We can
start to get some information. We can
modify how we collect. Here's the brain
slice. Here's the optic nerve. We can
stimulate that as if this brain
was seeing light. We can collect the
peptides that are released, and
we get data that looks like this.
Before stimulation, we get
a little bit of garbage and lots of
background. We stimulate the optic
nerve, the retinal hypothalamic tract,
at 10 hertz at that time of day,
and we get all of these peaks. Each one
is a peptide. Then my graduate
student and a post-doc spent a year
I'm not going to tell you about that, but that's really
how long it took, and we identified
every one of those peaks. That's
not very straightforward, but we now
can tell you when this says, for example,
that it's big lantern neurotense
or substance P, we
did a lot of work to identify them. That's not
easy. The real
answer is that the SCN
synchronizes the body's clocks with lots
and lots of different peptides. It's not
very obvious that it's one compound.
You can even look at it this
way. This is at one time
of day, another time of day,
these two are at the same, and as you change
the frequency of stimulation, different levels
of light coming in,
you get different profiles.
It is
a complex system, even though it's only 10,000
neurons. We have that made progress.
There's a peptide
there that starts with serine, alanine,
alanine, serine, and it's a smaller
form. Somebody called it
little space. That peptide
is secreted
by the SCN, so we made the peptide
and dumped it on. This is showing you
the phase shift of the
brain as it sits there. You can measure
this. Martha's group is really good at that.
These are Martha Gillette's group.
These are electrophysiology or single unit
recordings from the control.
You add AVP, there's no change. You add
little space, and there's a phase delay.
You've changed the phase of the brain in the dish.
It turns out it doesn't matter if you add
100 picomolar or 1 micromolar,
it does this phase shift.
By itself, stimulating
the retinal hypothalamic tract does the same thing.
We discovered a peptide that seems
to do at least part of
the phase shifting.
After we did this, somebody
else made a little space knockout
mouse, not us, somebody in New York.
They found that at this
time of day, when you expose the
animal to light, you don't get
a reset of biological clock.
This all does fit.
If you like to see more
physiological data, this is some
Gillette's group with Norman Atkins.
50% of the little space neurons actually
respond to CFOS when you give
light. That means they are light sensitive.
All of this fits. We discovered a
neuropeptide in the circadian
system using mass spectrometry.
Now we know what it does.
The true
last example, VIP, vasointestinal
polypeptide. This one we
couldn't detect. We actually sent our
original paper to science. Two of the reviewers
said if you can't detect VIP, the
most common and well studied
SCN, circadian
peptide in existence, your system
doesn't work. We refused
to let go of that.
Jennifer and Martha's
group has done some functional testing.
We can't detect the VIP, but we
detect a small form.
It turns out after a lot of work
I can say our mass spectrometry was right
and the 500 papers that actually
detected the long form were wrong.
They actually just used immunohistochemistry.
This bar from here to here is
the amino acid sequence of VIP.
We only detect this. It's cleaved
in an RK.
It turns out no one else
has actually done mass spec on the SCN
like this. Here is what happens
when you stimulate that
and you look. You see the small form of
VIP is secreted.
If you actually wanted to know what it's doing,
the same single unit recording.
VIP and the short form of VIP
cause a phase advance.
It's a phase advance or
delay depending on the time of day.
They both
work exactly the same.
They both go through the
VIP receptor 2,
VPAC 2.
If they're identical, why do I say this is
biologically active? Because this is not released.
This was isolated in the gut
of a mammal. This is what's in the SCN.
Now we can at least explain our observations
and have information
that people have been testing the wrong peptide.
Now I am done
and I wanted to leave with
three points.
The mapping of the brain, the brain initiative of Obama
certainly is a
grand challenge. A lot needs to be done.
The reason it's hard
is really the dynamic range of the brain.
If you think of distances, this is
nanoscale, but in my
spine I have things separated by meters.
In the brain itself by centimeters.
The time period is
milliseconds for some neurotransmission
and yet my neurons are
not functioning as well as yours because
they're three decades older and it's the
onslaught and oxidation and protein changes
that occur during decades that actually
changes things also.
Molecules range from nitric oxides
to proteoglycans and so in every case
there's six orders of magnitude
dynamic range in length, time, and
chemistry. The other thing to
end with is I have a lot of people who tend to think
that everything is known in terms of chemistry.
I know lots of people
coming up with a new tool. They said, I looked in a
book, I found that 18 molecules of the cell
that are of interest and I'm going to measure those
and I assure you that when we
use different tools to look, every
cell we look at has surprising neurochemistry.
I didn't go into it but in some
cells the single most intense
by mass peak that we see
in the cells we can't identify yet.
So that tells you there's a lot of
unusual neurochemistry still to be
discovered and that's true in mammals
as well as some of the stranger creatures
we work with. And truly
the last slide, I've had a lot of
fun for 20 years basically
working with a range
of animals and even
the Illinois animals like the
cow, pig, and the honeybee which I guess
are more agricultural in nature but have been also
a lot of fun to work with. And thank
you very much for your attention.
Applause.
I wonder if you can
elaborate on activity and functioning
and you're seeing
activity but it's not dynamic.
It's not dynamic. So
I have, you know, if you think about it
if you, I kind of tend to think
there's dynamic, there's chemical, and there's spatial.
I know most people in this room probably wouldn't have put
chemical as an axis. And mass
spectrometry and the techniques
we use give us a lot of chemical information
at the cost of dynamic.
So dynamic information, time
resolved, is critical. I just have
given that up
because I don't know how to get it at the
same time. Now with fluorescence
for example, you can make a probe
and you can pre-select the molecule of interest.
It could be a neuropeptide, it could be
a small molecule, and then you can get
dynamic and space. Now in one sense
you don't get chemistry except that
you've built it into your probe.
So if your probe only
images nitric oxide, if your probe images
serotonin, then by definition
you're selective. But I would say that
that's because you knew in advance and you built it in.
But you're right. We've given up
a critical axis of information
to keep the chemistry.
And we pay for that
a lot. And so I've picked problems that
are slow. Time of day is great.
It's dynamic. I've had people say
well that's unfair. You're using something that takes
hours. And I kind of think in my mind
that's good. I picked a good problem. If I can do
hours, I'm going to work on time of day
or seasonal changes. I can't do milliseconds.
You know, I've
had a number of people say, well you cheated.
You picked aplesia with
300 micron cells and they actually
act like this is a negative. And I say yes, I cheated.
I picked a big enough cell to work with.
Why would you pick a 3 micron cell that you
can't get the information you want?
And so in one sense, if you really understand
the brain and you tell me that
you have, and this is
actually an important point. I spent 20 years learning this.
If you want dynamic information and you have
a particular molecule, there may be a part
of the brain that that's perfectly well suited.
I didn't go into it, but we work
with D-amino acids.
D-serine or D-aspartate as opposed to the
L-amino acids. And those are actually
like, D-serine is glia specific.
It's only found in astrocytes.
Unless you move to the next region of the brain over
and then it's found in neurons. And in that other
part, it switches back and forth. And so you get to
choose. It's kind of fun.
By knowing enough, you can actually
decide that if you have a set
of tools or criteria, you can actually pick
the problems that are relevant. And in our
case, we've given up
unfortunately, the fast.
You know, it took...
So, we do a...
What do I want to say? We don't do much
for anything really with cancer, with the exception that
we have worked with some cell
lines, like immortalized cell
lines, like ATT20 cell lines, or endocrine
cell lines because they release
and they are a cancer cell line. I don't tend to
think of them as cancer in the sense I think of them as a model
of an endocrine cell line.
It's taken me a decade
to learn enough neuroscience to where I
actually can make my way around. And after 20 years
I actually feel confident about some
neurochemistry. When I think of cancer,
you know, it's going to take me another decade
before I know enough to actually ask the right questions.
And so, no, I
haven't done anything really
with cancer. Oh, it is.
And so I've had a...
Amanda Humman at Notre Dame is a former
student of mine who was using some aspects of this
and she is actually both in chemistry
and in her cancer medical school.
And there's other people doing this with cancer.
And that would be great to look at, you know,
how the environment in and around stem
cells, because presumably that's partially
chemical in nature that helps
determine the fate of a particular cell.
And so I would think
that these approaches are going to be very important
when cell heterogeneity is important. And that's
in certain diseases like cancer,
it's in development, and it's in the brain.
Now there's probably others, but
those are obvious.
Do you find any differences
in this chemical profile between sensory
versus motor neurons? I mean, there's clearly
differences in natural science.
So in aplesia,
every cell is slightly different. The motor neurons
tend to be big and they have
very big differences. So the peptide complement
is different.
So
we don't know
enough in the mammal because we haven't done enough
single cell. We've done about 40 to 50 using the
patch clamp approach. We see
for example in the
TRN, I mean this is a good
example, these one set of homogeneous
neurons. We measure...
Now when you do a patch clamp experiment
sometimes you don't patch on.
We throw those cells out. Sometimes
we think we've done everything right and we don't see
the amino acids that are required for protein
synthesis. We throw those samples
out. But in 25
cases we got really high quality data
that makes us think we patched onto a single
cell. And
these are all glutamate using cells.
And in 21 of the 25 we see
high levels of glutamate. In 4 of the 25
we don't see glutamate. Our physiology
friend says you're absolutely wrong.
That can't be true. Everything else
is right. The physiology looks good.
Why is glutamate missing in 4 out of the 25?
I don't know. We have
seen now 4 spontaneously active
TRN cells where you don't inject current
and they're still active. And all 4
have high levels of histamine. I don't know
why. So we're learning that. But the
heterogeneity, in one sense you should do 10,000
cells. And
when we want to know the physiology
within the context of the brain slice
then we're using patch clamp
and everything else and those are kind of slow measurements.
And so we're
still learning.
So there's huge differences
within a supposedly defined population
and you change populations of cells.
The chemistry is completely different.
And I say that. They still have all the amino acids.
You're still doing protein synthesis. So when I say
the chemistry is different, I may be
exaggerating. Transmitters, some
housekeeping. A good example of a difference
is in one part of your brain
you use nitric oxide.
That's a neurotransmitter.
It turns out
one in a thousand cells use NO.
NO is degraded into
nitrite. It's well known
and established that your brain has about
10 micromolar nitrite.
What we find is cells that use NO have 10 millimolar.
Most of the people
doing modeling
of NO and nitrite
say we're crazy. It can't be 10 millimolar.
But that's what we get. However, if it's
one in a thousand cells, that explains their number.
It's 10 millimolar in the NO producing cell.
Almost zero in the rest.
That averages out to 10 micromolar.
Now, from a modeling of
chemistry and how
the enzyme
works, there's a huge difference between
10 millimolar and
zero. So that's an example where the heterogeneity
is striking. And you really have to
do it single cell at a time because
you get an average otherwise. 10 micromolar
as opposed to really high in zero.
And we're still trying to understand that.
The problem in that case, if it's
one in a thousand and there's no easy way to get to it,
that means you have to do thousands of cells
to get 10. That's kind of
like the cancer. So one of the reasons we haven't done
something like cancer is I can't do
a million cells to find the one.

