Dr. Rafael Yuste
🧠💡 They’re Mapping Your Mind: How Neuroscience Is Building the Blueprint for Remote Brain Control
🔬 Guest Speaker: Prof. Rafael Yuste
🎓 Professor of Biological Sciences & Neuroscience at Columbia University
🏛️ Investigator at Howard Hughes Medical Institute
🏆 Pioneer in brain circuit mapping and creator of radical imaging techniques
📚 Table of Contents
- 🧠 The Cortex: Humanity’s Final Frontier
- 🔄 Modular Brain Theory & the “Double Helix” of the Mind
- 🔍 Mapping the Invisible: Reverse Engineering the Brain
- 🎥 Imaging Neural Activity with Calcium Tracers
- 💡 Turning Neurons ON with Laser Light
- ⛔ Silencing Brain Cells with Photonic GABA
- 🎹 “Playing the Brain Like a Piano” with Holographic Light
- 🧪 What This Means for Targeted Individuals
- 🛡️ Detection & Countermeasures
- 📎 References & Further Reading
🧠 The Cortex: Humanity’s Final Frontier
Professor Yuste opens with a bold statement:
“Understanding the cerebral cortex is the greatest scientific challenge of our time.”
The cortex — a bark-like outer layer of the brain — is where cognition, emotion, and perception originate. It’s only about 2 mm thick, but it’s what makes you you.
💡 Psychiatry still struggles to “fix” it. Why? Because it’s the only organ we don’t truly understand.
🔄 Modular Brain Theory & the “Double Helix” of the Mind
🧬 Just as DNA’s double helix unlocked genetics, scientists now believe the brain might run on a repeating modular structure — a single algorithm duplicated many times.
This means:
- 🧱 Each “module” could be a universal processor.
- 🐭 Mice and humans may share the same blueprint.
- 🤯 All human cognition might be built on one simple circuit.
And they’re racing to find it.
🔍 Mapping the Invisible: Reverse Engineering the Brain
Yuste proposes a straightforward method — the same engineers use:
“Take something complex. Watch it. Tweak it. Watch again. Build the diagram.”
But doing this with the brain requires:
- Imaging all brain cells firing in real time
- Controlling individual neurons
- Simultaneously stimulating patterns across the brain to discover its “software”
They call this:
🎹 Playing the brain like a piano
🎥 Imaging Neural Activity with Calcium Tracers
The team uses a method where neuron activity is visualized via calcium signals. When neurons fire, calcium floods in, and a fluorescent marker lights up.
📸 Result:
They can watch 4,000 neurons firing at once — in a live brain slice.
👀 These high-speed optical recordings give the first visual “readout” of thoughts at the cell level.
💡 Turning Neurons ON with Laser Light
⚡ They developed a ruthenium-based compound called Rubiglutamate, which is inert until hit with a focused laser. Then:
- 🔓 It breaks apart instantly.
- 🧠 Releases neurotransmitters like glutamate.
- 🔥 Triggers exact neurons to fire with 1-cell precision.
They can activate only the dendritic spine of a single neuron — under 1 micron wide.
This is optogenetics 2.0 — no genetic modification needed, just photo-chemical hacking of brain cells.
⛔ Silencing Brain Cells with Photonic GABA
They also created Rubigaba, a compound that blocks neurons.
☠️ It mimics GABA — the brain’s natural inhibitory signal.
🔦 Shine the laser → Rubigaba activates → neuron silenced.
They’ve tested it to:
- 🚫 Stop epilepsy seizures
- 🧠 Control neural feedback loops
- 🔇 Selectively disable brain areas in live tissue
Even better? It can be applied topically or injected into cerebrospinal fluid (CSF) — making it viable for remote or full-brain silencing.
🎹 “Playing the Brain Like a Piano” with Holographic Light
Using a Spatial Light Modulator (SLM), the team can:
- 🪞 Split a single laser into dozens of simultaneous beams
- 🎯 Target neurons in 3D — across multiple layers of brain tissue
- 🧠 Trigger complex patterns (e.g., memory recall, sensory activation)
They demonstrated:
🎼 Stimulate 5 neurons → the whole circuit bursts into spontaneous activity.
🎮 Stimulate just the “right” combo → it completes the rest of the pattern.
🧠 This proves the brain has pattern completion logic — like AI filling in a missing puzzle piece.
🧪 What This Means for Targeted Individuals
🎯 While Prof. Yuste’s intentions are academic, these technologies are dual-use — meaning they can be weaponized:
| Technology | Research Purpose | Weaponization Potential |
|---|---|---|
| Calcium Imaging | Track neural activity | Monitor thoughts in real time |
| Rubiglutamate | Trigger precise circuits | Remote stimulation of emotions/memories |
| Rubigaba | Silence circuits | Induce neurological shutdowns |
| SLM Holography | Test brain computation | Inject artificial thoughts or induce confusion |
| Epilepsy Photo-Cures | Therapeutic inhibition | Targeted neural paralysis |
This is the future of neuroweapons — not science fiction. It’s already happening in labs.
🛡️ Detection & Countermeasures
How do we detect and defend against this?
🛑 Many of these signals use near-infrared (800 nm – 1064 nm) light. It is invisible to the naked eye, but can:
- Penetrate up to 1 mm in biological tissue
- Cause neural modulation without awareness
- Be scanned with specialized TSCM equipment at low RBW
🛡️ Suggested Countermeasures:
| Threat Vector | Detection Tool | Countermeasure |
|---|---|---|
| Two-photon lasers | Optical backscatter cameras | Optical shielding (e.g. graphene film) |
| Ruthenium compound activity | EEG voltage artifacts | Detox / chelation + monitor CSF implants |
| Pattern activation | RF backchannel signature | Scan for multi-node RF hopping patterns |
| GABA-based silencing | Sudden mental blackouts | Real-time BCI feedback monitoring |
📎 References & Further Reading
- 🧪 Yuste Lab: https://www.yuste-lab.org
- 🧠 Connectomics Review in Science (2023)
- 🔬 Two-photon Microscopy Overview (NIH)
- 📄 DARPA BRAIN Initiative Grant Archive
- 🧨 “Neuroweapons” – UN Report on Emerging Threats
- 📘 “The Brain’s Last Secret” by J. Giordano (Georgetown Lecture)
🎯 Final Thoughts
While the mainstream sees these as medical miracles, Targeted Individuals already know:
🔹 Neurons can be hijacked
🔹 Perception can be rewritten
🔹 Thought surveillance is not science fiction — it’s state science
This lecture proves that remote neuromodulation exists — and the tools to do it are already here.
Full Transcript:
we Professor RAF as our guest speaker
today is a professor of biological science sciences and neurosciences at
colum Columbia University he was born and raised in Madrid Spain where he attained his MD at the University at the
unity aut easier uh she was after that she spent
some time at Cambridge in UK with a group of Disney brers and then he did
his PhD with L cats at the Rockefeller University in New York he then moved to
BS and work with David tank and winred B
and in 1996 he joined the Department of biological Sciences at columia
University in 2005 he became an how Medical Institute investigator and Co
director of the C Institute for brain circuits at Columbia
and since working there and obtain many
awards for his Works including the New York City moror majors and Society of
neurosciences young so welcome to thank thanks for [Applause]
coming you’re so wonderful that uh applauded me before I speak I hope I’m
up to the to your applause so it’s I’m delighted to be here finally um and I
spent the whole day talking a great science with uh some of you uh and um I
wanted to uh talk about uh how the types
of techn ology that some of you this type of group as a whole is developing
could matter crucially to answer fundamental questions in Neuroscience okay so I’m going to try to present to
you what are the big questions in Neuroscience the way I look at it and how and then show you the types of of
technical challenges that we have and how we are trying to solve them in New York in my lab but that hopefully will
make resonate very well with many of you that are working in at the interface
between biology chemistry and physics and Engineering so um I would say that
understanding the cerebral cortex is the biggest challenge in science of our
times and some people may not agree with with this but let me tell you why I say this so um the cortex is the largest
part of our brain okay and it cortex means bark the cerebral cortex uh and
it’s a little bit of the this bark that covers our enfin um and this uh bark is responsible
for all our cognitive abilities and we know this for sure both from uh human
case lesions and animal work so in fact you could argue that What Makes Us
humans human what the humanity in our in our in our psychology is so due to the
cortex so um so it’s fundamentally important for
humans to understand the cortex because we will understand ourselves for the first time okay and it’s the only part of the
the of the body that is not understood I can tell you that in general other parts of the body medicine knows them well
enough to be able to cure diseases and when you go up from the nose I mean these psychiatric diseases it’s just
hopeless and we don’t have it’s like trying to fix a car without understanding how it works it’s as
simple as that so why is it so difficult why people have been looking at at the cortex now for 100 years I’ve been
studying for 100 years why haven’t we figured this out so um so everyone says
it’s because it’s very complicated there’s buil out of many cells and these are example of some of the layers in the
cortex from the top to the bottom it’s about 2 mm but this is very similar whether it’s
in humans or in mice it appears very late in evolution in the Maman lineage
and once it appears years it actually increases in size very quickly so um because of that people
have thought have are hypothe hypothesizing that the way the cortex is
evolved is by repetition of the same um circuit that
is a single modular unit that’s getting duplicated in evolution very quickly and
this is in in uh this this thought this hypothesis is
inspired by the way the rest of the body is put together in evolution that we’re metameric animals and Gene duplication
is essentially the trick that Evolution uses over and over again to grow bodies so the it is possible that the brain of
a mouse the cortex of a mouse is built with the same basic module as the cortex
of a human the reason we’re sort of smarter than mice is because we have more modules but that the module is the
same and this is based on evolutionary arguments is also based on the development the way the cortex develops
in onogen is stereotypical and it’s the same whether it’s in a mouse in a cat
and a monkey and a human it’s exactly the same It’s Quickly through the same stages of of
steps but there’s something uh also quite deeply fundamental about the idea that the cortex could be modular and
that is that if you look at different parts of the cortex we know that they’re involved in completely different
computations for instance our visual cortex in the occipital L is involved with visual recognition he’s doing a
very sophisticated visual analysis of the world meanwhile the front of the cortex is involved with motor planning
or thinking ahead about the future calculating probabilities of things that will happen in the future and then you
look at the auditory cortex is involved in analyzing sounds Spectra so if you if
you buy the idea that we’re dealing with a modular structure and if you think that all
these different computation how so different that means that if they’re done by the same Hardware the common
denominator has to be very simple okay so you have all these different computation that
mathematically have apparently nothing in common if they’re performed by the same Hardware this Hardware by
definition has to be simple okay so this is uh why I’m so excited about trying to
understand the CeX now in in in science because I would argue that we’re not dealing with a complicated structure
it’s the opposite has to be very simple because otherwise it wouldn’t be able to do all these things so I would venture
the idea to you that we’re dealing with something like the double helix they were looking for the brain’s double
helix just like in genetics had all this data people said genetic was very complicated and
difficult and then Watson and cre came out with the double helix model and bingo it was as simple as pie and it
explained essentially everything it’s possible that this cical module could be
like a brain stubble Helix it could be a simple algorithm that you could write down in a piece of paper and would be a
fundamental computational power it would be analogous a biological touring machine an algorithm that can be used to
compute any optimization problem or any problem so this is what we’re looking
for okay no one has found this yet uh but I hope that within our lifetime
we’ll see this happen someone will show up with the brains double helix uh and
uh why haven’t people found this yet so the cortex is complicated in the sense
that is not that is buil with many different cell types and this is an example of different types of neurons
that are present in the Maman cortex and again it doesn’t matter whether they’re Neons from the mouse or the human they
look exactly the same these are the layers that I was showing you earlier this is the top of the corex lay one the
bottom layer six again this is about a millimeter and a half or so and these are some of the cel types found in cats
or in mice and neurons come in different shapes uh in fact we still don’t know
how many classes of neurons are there in the cortex it’s possible that it’s about 100 different subtypes of neurons and
people have assumed that each type of neuron may have a different function and
that’s why they’re they look so different they have different dendrites which are in uh in blue here or in Black
here uh and axons which are in in Gray or in in red and these difference in morphologies are
indicative that maybe they could be a different in function so unfortunately in this drawings these neurons are
separated side by side but in reality they’re all mixed in next to each other
so um this is a picture of the cortex of a mouse this is a living brain slide uh
and every one of these white dots is the cell body of one of these neurons that I was showing you earlier okay so the
little cell body here could be one of these guys but every one of these neurons has a
tree as big and as complicated as these trees so they’re all mixed
in so that’s why people have been afraid and have always assumed that we’re never
going to understand this in fact the uh the first investigator who really made a Headway in the cortex was someone called
Ramon kahal who was Spanish 100 years ago he wrote a book in which he argued to Young
investigators that they should never work in the cortex because he he said these are the impenetrable jungles where
many investigators have lost lost themselves he was talking about himself because he worked in the CeX his whole
life and could never figure it out so in a picture like this we could be looking at these impenetrable jungles but maybe
of a universal computer okay this could be one of these modules maybe there’s several modules up here in front of our
eyes we just don’t know how to decipher that that information but it’s present
there okay so this is the challenge and
uh what we need to do to the cortex is the same research program that Engineers or elect Engineers do in a in a
day-to-day basis which is to reverse engineer it and turn pictures like this
into circuit diagrams like that okay um and our group and many groups around the
world are trying to do this reverse engineering job and um what we need to do is to be able to first image the
activity or record the activity of all the cells okay this is the first technical challenge we have to be able
to image that that activity the second thing we need to do is to turn cells
on or to turn them off okay if we could do this then we could actually reverse
engineer because we’ be able to essentially poke the circuit uh and then the last thing is
what we call to play the piano which is to simultaneously manipulate the activity of all the neurons in arbitrary
spatial temporal pattern because if we could do this we may be able to figure out the transfer function if you were an
electric engineer you have a circuit like this and you can actually play the piano with the circuit and start putting
any arbitrary functions you can figure out the transfer function okay so this is the uh this is the
dream so obviously we need new methods and this is a nice quote from criek who
like kaha uh After figuring out DNA he spent the rest of his life trying to
figure out the cortex and he failed like AA but he was arguing in this uh quote
that we need better techniques that we’re dealing with a technique called bottleneck and he imagine a method in
which you could inject neurons with a substance and label all this that are connected to it so this would be a
method that would give you the circuit diagram of the cortex um people are trying to do this
now in Neuroscience is a big revolution of methods getting invented left and right and developed some of them are
using electron microscopy and this is what you may have heard as connectomics it was actually a nice review in science
last week on connectomics from liman another approach is using electrophysiological methods pair
recordings record at the same time from pairs of cells to find out connected this would be equivalent to the
electrical engineer going in with a voltmeter and poking systematically every two positions in the circuit to
see whether they’re connected or not there are genetics methods that try to label the circuits using genetic tricks
using retrovirus Ral methods to uh label the circuits and um we think that uh
we’re pushing for the optical methods that using light you could actually carry out this research program and make
uh Creek happy okay so um but because all these cell
touch are all mixed if you want to go in with Optics
to reverse engineer the circuit you need to have the same Precision at least that the circuit has okay and that’s why um I
think we need single cell Precision that you’re not going to be able to reverse engineer the circuit with Optics if you’re
stimulating or many groups of cells you need to go one cell one at the time the
solution for that in our hands is to use two Photon excitation because it’s a
technique that enables you to uh use uh laser light uh with Optical sectional
properties to go deep into scattering media and has a special resolution effectively of much smaller than the
size of a neuron so with two Photon we can image and we can optically manipulate and play the piano with this
single cell Precision that we to do this reverse engineering job so uh I’m going
to describe a series of techniques that we’ve developed using two photon lasers to try to go through this research
program I will not tell you what is the answer of the brain double helix but I will tell you how people like me and
other groups would trying to get there and how some of the work that’s getting done in this campus could matter so let’s talk first about Imaging
so the idea again with a two Photon image is to be able to capture the special structure of the circuit and
read out the temporal Dynamics the activity
patterns um and this is something that started through a chance discovery that
I made when I was with cats that you could use calcium as a proxy for the electrical
activity of a cell okay so why is that happen so this is an experiment in which we’re Imaging calcium in a neuron that
we’ve patch with an electrode and we fill the cell with the calcium indicator and with the same electrode is a we’re
depolarizing the cell and making it fire so this is the pattern the membrane potential as a function of time and
these are the famous Action potentials and they’re regular because we’re injecting current we’re making the cell
fire and simultaneously we’re measuring calcium in the Su of the cell and every
time there an action potential there’s a little increase in the calcium concentration and when you fail to
generate an action potential you fail to generate that increase so that means you can use this calcium
trace and back calculate the pattern of spiking okay and the reason this happens
is because neurons like every cell have calcium channels on their
surface and when the neuron and the neuron has essentially zero calcium inside a very low concentration of
calcium and when the neuron fires in action potential calcium’s Channel open
and this lets in enough ions calcium ions to increase the calcium
concentration significantly inside the cell and you can read this out optically if you have a good calcium
indicator okay so through this um I would say uh consequence of the neuronal
biophysics that first of all that we’re lucky that every neon has calcium channels in the surface and second of
all that these calcium channels open with calcium with action potential you can actually use calcium to read out the
the uh the action potentials and this is say you could say well what’s a big deal
in this cell you can measure the action potential with an electrode but how about measuring that in all these 4,000
Neons you cannot put 4,000 electrons but you can do this image is exactly a two
Photon calcium image of a living brain slice and in this type of experiments
we’re monitoring at the same time the activity of 4,000 Neons so this is
equivalent having 4,000 electrodes so with this Imaging
two Photon calcium Imaging of circuit is not perfect yet but at least
we’ve solved the first problem being able to monitor the activity of if not the whole circuit a big chunk of the
circuit okay so then we moved on we’re still working on developing better
techniques to image not calcium but voltage but this is still very uh it’s not working really um but uh with
calcium at least we can move on and then uh this is an example of the the data that we’ve acquired with to
Photon calcium Imaging so this is the the analyze movie here you’re looking at about 500 neurons the the the neurons in
Gray are the cell bodies of the cells that are not firing and the ones in red are the ones that are firing and this is
the spontaneous activity present in a brain slice is an example of uh what I
mean by Imaging the activity of a circuit so you should be able to make movies like this of big chunks of of
brain tissue if possible possible of the entire brain of an animal and being able to capture the activity of every neuron
and detect every action potential okay yeah this movie was uh 50 milliseconds
per frame which is it’s too slow if you want to catch a reaction potential but
we’re improving the method and some of the things I’ll discuss could actually help us get to the regime where we could
actually monitor this with 1 KZ uh um frame
rate so let me move on from the Imaging to the how do we turn things on
okay so uh there again and we have to do this with two foron excitation people
have been using optogenetics using genetic constructs that are turned on by light that make depolarize the
neuron um but as of today they work very poorly with two Photon excitation so
that means that they have been used only with traditional light sources like LEDs
or lasers like this one but this as they penetrate the tissue they
scatter um and then the spal resolution essentially becomes uh enormous so they
don’t have the Precision that we require for the analysis of the circuit so what we have done instead is using uh
chemistry uh in fact we use uh the solution came from ruthenium which sits
in the metal transition metal that sits in the middle of the periodic table that
uh people uh that are building solar cells have been working with ruthenium
because it has ideal Optical properties you can build uh antennas with ruthenium
at the core uh that are extremely efficient at Gathering light um and one
of these chemist is an Argentinian a chen and he was visiting my lab and he was saying well we have this great Rian
compounds the problem is that when the sunlight hits them they break apart so
they’re terrible for solar cells um but uh that’s what that would
make them ideal to do photochemistry and photo release so there are actually technical
reasons why ruthenium is better than all the other meths metals at this no um I will not go into into this technical
reasons but let me point out that uh with collaboration with Jenik in buenosaires we buil a series of
ruthenium cage compounds in which you have ruthenium coupled with two bipin
and phosphine and then with the remaining Bond we can um derivatize it
with a neurotransmitter with excitatory transmitters like glutamate inhibitory
transmitter like Gava or with neurom modulators nicotine serotonine dopamine
all kinds of things and then when uh when we shine light so this these cage compounds are
inactive so if you put glutamate here next to ruthenium this glutamide will not bind to The receptors will not
activate the cell but if you shine light um through this these properties
that linium complex have uh the metal Bond will break releasing the
transmitter in this photo release in fact uh it’s extremely fast this is a measurement of the photo release of
renum it happens in about 4 NS it’s two ERS of mag itude faster than traditional
cage compounds that use an organic moti here normally based on Nitro benil
chemistry so because we’re using a metal Bond one of the properties it has is that extremely clean and very fast D
caging so we’ built this series of tools that are photo activatable that break
apart very quickly and you can also excite them in a wide variety of wavelength this is the excitation
profile of some of these renum compounds and they work in the in the visible
light in fact we’ve done encaging of uh R ruinan compounds with lasers like this
one with just uh laser pointers and they’re strong enough to fire neurons when we engage um Ruby glucamine for
example but they also work in the two Photon regime and this is exactly what
we need them for okay so using this chemistry that again came from the
people that were building solar cells okay inorganic uh
chemist we uh we helped bring it into Neuroscience to solve this technical issue have been able to turn cells on
with light and the way we did it is by building Ruby glutamate renum by pring
glutamate in which you have a glutamate molecule you can photo relase with light
and when you uncage uh rubig glutamate on top of a neon you make it
fire okay this currents that are generated are blocked by glutamate
receptor antagonist so they are they are due to activation of glutamate receptor they’re not due to photo damage or
something else and they also reverse that the reversal potential for glutamate so this is a veritable clean
glutamergic activation of a neuron um so using this Ruby glutamate
we can actually uh do this experiment for in in this case we have two neurons literally next to each other we’re
monitoring them with we have two electrod to monitor their action potentials and then we’re encaging on
top of one and nothing happens to the other one you cing in the second one I
think happens to the first one so this is I mean look at the special resolution that we have for this this is a squis it even here that are next to each other
you be able you w be able to turn one on and then the other without mixing so this is the type of tool that we need to
do this circuit uh um taking apart the circuit and then of course we can go
inside a cell and we can turn on not just a single neuron little piece of a neuron these dritic spines they’re
Illustrated here and this the special resolution here is even smaller this is about half a micron or so so you have
literally submicron resolution in terms of the encaging of ruthenium of Ruby
glutamate in this case so this is an example of um two Photon and caging to
far this neuron this is the special resolution I was talking about we’re patching this cell and when we enage uh
ruthenium rubig glutamate on top of the cell we make it fire if we move the laser a little bit off then there’s
essentially no response of the neuron we have singles of precision so uh an
example of what you can do with this type of technique um so then you have you can have a brain
slice and you patch one particular neuron can you record the membrane
potential in this cell and then you incubate this slice with rubig glutamate and you go go in with your two Photon
laser and you go and turn on every neuron in this slice one by one and if
this neuron that you’re activating is connected to the neon that you’re recording from you will detect a
response okay and doing this simple experiment serially in about 10 minutes we can
actually test 500 neurons and find out if they’re connected to the new we’re recording from so going back to cric’s
dream this is actually one vers of cric’s dream is not exactly what he had in mind but we we can solve the problem
of mapping circuits we can go in an image like this patch a cell and stimulate all the other cells to see
who’s connected to that neuron uh and to do this efficiently we first uh we do this automatically okay
we cannot do this by hand it takes too too much time so we first detect The Contours of the neurons using an
algorithm that analyzes the image and looks for these Contours and this happens in a couple of seconds and then
uh we actually compute the traveling cman problem to move the laser beam from
nearon to neuron to figure out the fastest route to stimulate them all one by one and then we use this trajectory
and essentially stimulate every cell and build Maps like this one so this is now a map of the connectivity of a piece of
the brain slice and we’re mapping all the connection to this particular neuron this is the neuron that we patch with
the electrod and the neurons in Gray are the ones that are not connected the
neurons in color are the ones that are connected and the color code corresponds to the strength of the connection so
these neurons up here in red are very strongly connected to the neuron in Black so this is an example of using
these Optical methods to decipher the circuit using two Photon caging of this
uh glutamate in this case to be able to quickly this map was obtained a about 10
minutes to map all the connections that are present in that uh from this piece
of of the brain to this particular neuron okay of course if you want to map the
connection from all the neurons to all the neurons it’s another story but we’re working on that too not using electrod
but trying to do it in an all Optical method as I’ll tell you later is that a single
plane this actually is only a 2d this is a single plane but we’ve done this in 3
to so um actually that so um the this is
a brain slice right so this is physically removed from the one plane is removed from the volume right this
actually it’s sort of in the middle of the brain slide so it’s it’s sandwich between this cage on the top and in the
bottom were there a lot of connections between neurons that were out of that plane that were removed we don’t know
but we presume that they are in fact the connectivity increases as we go deeper into this slice these slices
incidentally are about 3 to 400 Micron thick they have approximately 10 layers
of neurons on top of each other uh and they extend the slices extend for
several millimeters okay they have maybe 10,000 neurons maybe 50,000 neurons
depending on how rough estim you can see the connection within that slice exactly
so for that slice we can principle with two foreign Optics we can stimulate every neur we haven’t done it yet in Vivo in
the mouse that’s another challenge but it’s not impossible to think that you could
do this also in Vivo patch an in Vivo and stimulate all the cells not all the cells in the brain but all the cells in
a larger and larger territory and map out the connectivity one of the things we’re finding as you can see here is
that the connectivity tends to be very local and this is something people know and suspected from the way the axons are
shaped that most of the connection this actually something that could be really important in terms of the siering the
double helix of the brain that it is possible but dealing with a very very local module where most of the
connections are intrinsic internal okay yeah so um so I discuss Imaging how
we turn things on how do we turn things off well we use the same ruum compounds but now we CLE it to Gava which is the
main in inhibitor neurotransmitter in the brain in the meman brain and uh it works the same way except in Reverse so
now if you unage rubika on top of a neuron you generate these currents that reverse at the Gava reversal potential
they’re blocked by Gava receptor antagonist and we can use this to uh map
gabic receptors along a cell so this is an example in which we’re moving our two Photon laser up and down the cell and
we’re mapping the garic response so this gives you a little bit of a map of the
heterogeneity of the function of these G gabic receptors in the neuron so this is
something that you couldn’t do before because you needed two phot on Precision to be able to map this uh to this level
but of course we’re interested in using this to turn cells off and this is an example of of how we use Ruby Gaba to
turn neurons off so let me explain this experiment um so uh let’s look at this
panel first we’re patching a neuron in a brain slice and then we’re enaging rubby
Gaba right next to it the neuron is bathed in the slice is bath in rubby Gaba and we’re releasing it with our two
Photon laser right next to it and uh in this neuron we’re making the neuron Fire by injecting with the electro the
current that’s depolarizing the cell and this is an example of the action potentials that occurring of the neuron
every time we depolarize the cell and in blue is what happens when we depolarize
the neuron and anage Raba at the same time time so we’ve silence you see silence the uh the neuron by encaging
Gaba on top of it and then if we move the laser beam a little bit to the left or to the right then uh we lose this uh
silencing effect so in this case 10 microns away we still can silence the
neuron but 20 microns away we cannot any longer so this tells us that Precision
we have at silencing is on the order of 10 15 micr okay so this is just to illustrate that
it’s a two Photon effect uh this is the same type of panel in C in C the point spread function of
the photon is much worse so the spatial resolution is a little bit worse than it is in
XY so um so that means that you can actually optically silence neurons using with two photon
lasers and um we’ve used this for different experiments one of them I
wanted to illustrate here is in collabor with Steve Rosman the University of Minnesota uh and this is to use rubby
Gaba to stop epilepsy so you know there’s uh um about a third of the patients with
epilepsy do not respond to drugs uh and most of them have to go uh to the
surgeon and as of today it may seem brutal but surgeons go in and scoop out
pieces of the brain that’s the only solution for this patient and maybe there’s a method to use Optics to help
by uncaging G on the pieces of the cortex that are generating the seizures and stopping the seizures optically and
this is an experiment from an animal in which we this is a mouse in which we’ve induced these seizures this is the uh
electrographic evidence for seizures in the mouse and this have been induced
using a pharmacological treatment that makes the mouse epileptic this 4 AP model and in a control experiment
without Ruby Gaba we turn on uh blue light and nothing happens to the
seizures as it’s it nothing should happen but if we have rubby Gaba on top topically applied to the brain and we
turn on the blue light we can actually stop the seur so in this case we’re actually this is not a two Photon job
this is a one Photon job but we’re silencing seizures over a significant period of time in the the poor poor mou
So This Could Be Imagined as the potential therapy Rothman is a
neurologist interested in developing this as a therapy for for epilep for chronic epilepsy that doesn’t respond to
pharmacological treatments so uh let me talk to you about the piano and this is actually
those I think are the most fun experiments so again the idea is to be able to image all the cells and at the
same time going with Optics and turn cells on and off in any arbitrary spatial temporal fashion and I think
these experiments could actually lead us into the transfer function of the circuit um so so we’ve tried to do
this um with two Photon encaging of glutamate um using a single two Photon
laser beam and this is Illustrated here so this is an experiment in which we have 50 neurons in a brain slice and
we’re Imaging calcium in the 50 cells by raster scanning the laser beam and
monitoring the fluoresence okay and this is Illustrated here this is the cell
number as a function of time and every time you see these little black uh lines
is when the neuron fires an action potential so this is you’re watching all the firing a movie like the one that I
showed you earlier displayed now in a rasa plot in which you see every neon and every Spike now in the middle of
this experiment with the same laser we’re going and uncaging
glutamate in five neurons are Illustrated here by the Five Arrows and we do this by quickly moving the laser
from neuron to neuron and then back to Imaging so we’re Imaging all the cells
and in the middle we’re enaging five Imaging all the cells and then we move and engage another five and we’re
actually playing these musical scores as a speak we’re walking our way through the circuit turning on five Neons at a
time so this is Illustrated here so these green lines are the Neons that
we’re uncaging we’re firing simultaneously we’re Imaging all the circuit okay but look what happens here
in whatever 200 seconds into the experiment we uncage these five neurons and
boom the circuit goes bistic the all fire okay and then we continue our March
boom something interesting happen and a little bit later something interesting happens there no so we don’t understand
what’s going on yeah but something’s going on no and
you would agree that this is not like a random assment of Spike and that we’ve done something to the Circuit so this is
like Electro Engineers so you don’t know what the circuit is doing but you you know you’re do with some sort of uh
business here not some interesting uh transfer function so this experiment was
very exciting we published in 2007 but we did this by moving the laser quickly
so it’s a little bit like playing the piano with one finger no and uh
it wasn’t satisfactory so we thought about well what we really need to do is
to play the piano with all the fingers there with 20 fingers or 10 fingers at least and be able to change the pattern
in any any way we want stimulate any neuron in any order so um this is the type of
experiment that we have in mind we record a pattern of activity first these three neon fire this is like the movie
that I showed you earlier this could have been frames of that movie and and this could be spontaneous activity or
activity and response to S stimulus and then uh we could uh generate patterns
that resemble these patterns and see if we do this to the Circuit does the circuit complete the the
song okay if we could do this experiment we could actually test whether the C is BU for pattern completion this one of
the ideas that is floating around as what could this double helix of the brain be built for maybe it’s some sort
of par completion type of circuit okay so uh again so we have a
fundamental problem with two Photon microscopy because all the laser scan in microscope is serial you point a single
laser to one point and then with galvanometers or aods you move it very quickly to scan all the sample and that
just not not going to cut it because the you can move the laser as fast as you can but the faster you move the laser
the smaller the Dell time per pixel which means that if you want to get enough photons out of those pixels
you’re going to have to increase the intensity and this is going to lead to photo damage and you’re going to be saturating the
chromophor so we thought that the better solution was to split the laser beam
into many beamlets using either refractive Optical elements or special light modulators that I don’t have to
introduce to this audience okay so we essentially uh have explored this uh
multiplexing of the laser light as a solution for this problem of playing the
piano with uh all with 10 fingers so um I’m not going to discuss
uh the do doe sorry I’m going to focus on slms because I think they’re much more
powerful and uh as you know they operate essentially in the fourer plane of light
and you could argue that mathematically they could mimic any Optical transfer function at the at the
end of the day a microscope with all its component is nothing more than an OP iCal transfer function so you could
mimic this with this slm uh in the forrier plane okay so um that enables us
to shape the light in space and in time and this is an example of the type of
setup that we use this is a fento second laser going through a bunch of Optics a pockle cell and then uh we run it into a
Olympus modified homemade um to photo microscope and then we detect the
signals with with uh pmts but in the middle of this we have the little module
which where we have a reflective slm where we can actually uh split this
pin into any uh arbitrary set of pinets and with two Photon light you can then
take images uh Desir images like this one calculate the face mask and then
essentially write Columbia with light in your sample with two phot light this is the ronal I was talking to you earlier
so people in my lap uh it wasn’t my idea but they decided to write the picture of Ramon kahal in the in the in the
brain um and the way we do this as you probably do here is through these algorithms in which we we have a
calculated uh um um um ideal Target and then we uh
essentially um threw a a a series of Loops we improve the face mask to match
it until we obtain a really good match to the Target image that we want to to uh to mimic no and um
so the idea is Um this can help us tremendously with our research program
because um first of all you can do you can get around this serial problem of
laser microscopy um because you can take an image that you can obtain using
traditional laser scanning slow frame rates and then you can
calculate where the cells are detect where the cells are okay then calculate the center of mass of each cell and this
is done in a couple of seconds and then upload these positions onto the slm to
generate the mask through this algorithm I I quickly describe and then you split the the the beam into a series of
beamlets that each of them Target one of the neurons and then you can use a camera
and simultaneously acquire the fluorescence from all the Neons at once and if you use a fast camera that runs
at 100 HZ or a Kilz then you can essentially read out with Kilz frame rate the activity of all these neurons
that are getting illuminated continuously so you don’t have to pray pay a price uh in the number of photons
that you um collect by Imaging fast okay so you get around this fundamental
problem of of laser scanning microscopy so let me show you this at works so in
this case we’re Imaging I think it was 50 neurons uh this case it was uh less
whatever 20 or so and here let me just point out this this uh we’re running the camera at 16 milliseconds per frame this
is two Photon images taken using this approach so we split the laser beam and
we’ve Target each of the beamlets to each of the neurons and we’re monitoring fluorescent versus time and in this
particular neuron we we have an electrode so we know what type of action potential is generating and these are
the calcium signals that correspond to the action potentials and in this case we detecting single action potentials so
we’re Imaging single action potentials in this group of neurons at 60 Mill frames per second so this can be done
we’ve done it with 100 neurons now and we could in principle do it with even with a thousand neurons without any
significant uh problem so you could image a th000 neurons at let’s say 10
100 Herz and detect single action potential so I think this this this actually can really help this uh this
type of research so uh but with the slm not only
you can use it to speed the imageing you can use it to play the piano and this is
an example of some of these experiments in this case we’re encaging glutamate on
several dendritic spines at once so in this case the encaging is done truly simultaneously we’re not moving the
laser we’re Computing the positions of the vendu spines generating the mask and
using that Mass to flash the laser beamlets on top of them and releasing glutamate at once and these are the
electrical responses of the neuron when we do this so in this case we’re essentially uh playing the dendri spines
with six fingers in this case is we’re playing a chord it’s synchronous in this case we’re actually
using the slm to generate a Mas that looks just like the cell okay so we’re
enaging glutamide on top of the cell but only on top of the cell so we’re essentially to PL the cell on top of
each other and that should make this type of optical experiments very uh
efficient because you you can compute the structure of what you want to stimulate and use that shape to
optically stimulate that structure um let me show you a little
bit more what you can do with slm so the slms are essentially holographic techniques you can do this in 3D and
this is an example of a two foron stimulation in 3D so so these are now
two neurons that are sitting at different focal planes this case we’ve actually projected the image in the same
uh um in the same plane but they’re sitting at different focal planes and we have one electrode in each of the
neurons okay and uh and then we have essentially uh again um a similar setup
with an slm in the middle before uh before the uh the the objective and
we’re with the slm we have a mask that stimulates the top cell or the bottom
cell okay and the results are actually shown here so in this case we’re stimulating
both cells at the same time and they’re both find together but now we’re
stimulating only the bottom cell or only the top cell and look how
there’s essentially no cross talk in the optical stimulation so this is I think the one of the first cases maybe the
first case where people are doing optical stimulation 3D in this case it’s only two cells it’s a proof of principle
you could do this with many cells uh we don’t know how many but probably uh with
our setup we can probably play the piano with 20 fingers far 20 cells in a volume
uh and generate uh Action potentials on them so
um so this is a little bit where we are um because of
the idea that with an slm you can mimic practically any Optical transfer
function we went ahead and built a little microscope that has a a
transmissive slm I’m sorry transmissive slm um that’s
um uh essentially with a light source which is a laser pointer okay um and then with a little
diic and objective we can put it on top of a sample and then have a little camera uh to image the fluoresent coming
from this uh sample and with this type of pocket scope which is not smaller
than my laptop you can actually do this multi-side calcium Imaging and
stimulation of groups of neurons um with single C resolution so a lot of what we
need this big monster microscope to do we can do them with a little pocket scope that cost maybe $110,000 to
build so maybe slms could be used to advance to bring into the typical
Laboratories or into the field um this Advanced microscopic uh
techniques um so just to summarize I discuss with you um our improvements in
terms of using ruthenium compounds to uncage glutamate Gava and then we’ve
also done a bunch of other things um and the advantage of this renum uh cage
compounds is that they’re they’re very fast uh they’re clean chemically they’re
of course water soluble they don’t damage the cells uh you can use visible
light or two Photon light to turn them on to uncage them and they’re very chemically
versatile and then I discuss our use of scanless laser
microscopy for fast Imaging and for this type of piano experiments multide
Optical stimulation in principle you could do the same thing with Optical in activation uh and uh I would make the
pitch that these slms are ultimate Optics because they can mimic most Optical transfer functions I haven’t
show it to you but we can use slms actually when we get lazy instead of focusing by focusing the mechanically
you can actually focus with the slm in the software so you have essentially a microscope that has in principle you
could build with no moving parts and you could do in software a lot of the things that are traditionally done in Hardware
uh so just like in the electronics Industry there’s a translation to software of things that used to be done
in Hardware maybe the microscopy something like this could be happening in the future we haven’t shown this to
you but you can do aberation compensation Adaptive Optics with the same exact slm and you actually probably
very uh doing this year much better than we are um and uh yeah and maybe we could
be uh looking at the microscopy of the future and finally just the the person
that we responding for for the work was nikolenko was a phenomenal grat student who did most of what I presented eloo
was responsible for the work with Ruby glutamate and emilano real with Ruby
Gava Dary petera joined the lab and he’s now taking the lead in the slm work and
I think you’ve met Darcy when he was here last summer and our collaborators the chenik in Buenos Ares is inorganic
chemist and Steve rman in Minnesota is the neurologist and this work was funded by the generosity of the National
Institute of the NIH and the hhmi thank [Applause]
you they’re actually the the people in the lab and we just moved to a new building at Colombia for
interdisciplinary science anyway yeah um I was wondering
considering the diversity of neurons in the in the core text uh using your approaches one can identify the synoptic
partners for individual neural is there also a way to couple it to
so we yeah I didn’t I didn’t go into this but we do all these experiments on mice on purpose because with mouse
genetics you can precisely do what you said there are thousands of mouse gfp lines in which different subpopulation
of neurons are label with gfp because it’s under the control of different promoters so in our lab we have about 10
lines that we use for our experiments and each line identifies a subtype of of neuron so we focus a lot of our work on
the inhibitor neurons the So-Cal interneurons G interneurons um and for those we we have
this gfp m so we’re not mapping circuits at random we’re essentially going
systematically through the main cell types uh and uh actually the good news
is that the circuit diagrams that we are obtaining for these interneurons they look all identical so that means that at
least for the inhibitory cells in the in the cortex the reverse engineering of the circuit diagram is
pretty much almost finished in terms of understanding their circuit uh connectivity the exitor Neons are the
more the bigger challenge because they represent about 80% of all the neurons
yeah you you
show resp by strength of response do you mean
is it a multistep connectivity two neurons or yeah this is the connectivity
between uh yeah this case is the connectivity let’s say this this yellow Cell between
this yellow cell and this black cell so this one is about whatever 2 molts that
connection is about 2 MTS in amplitude and the connection from the red cells is
maybe five times bigger yes so so that that’s the strength of connectivity but what does it mean in terms of physical
network does it the connected via let’s say four other neurons to no no no this
is direct connectivity so we know for sure that these cells have axons that contact the
dendr of the black neuron so this is the strength of the direct monoptic
connection so that means why is the Red Cell five times stronger there’s two
possibilities because maybe it’s making five time five times more
contacts okay or maybe it’s making the same number of contacts but they’re just five times
stronger okay so all of these are direct connectivities that’s right exactly so
I’m discussing mapping direct connections yeah yeah you com the
idealic US yeah okay so uh so um to to do the
brain slices um with slms um it would be good to have a
little bit more power than what we normally have we have about four Watts with this uh coherent um Ultra whatever
they call so if the we have pass after L that have were more powerful would be
ideal if you want to go in Vivo and do this type of mapping job and reverse engineering in the cortex in an on an
animal you go through the entire six layer
or something like that uh and then
well yeah so there there’s enough for the calcium indicators now come in all different colors thanks of the work of
Roger Chan and other Pioneers that on on Whose shoulders we’re
standing and so that problem is almost effectively solved there not that many dies in the infrared but I think people
are coming with them now I think I saw the a group in Russia that’s coming out in Moscow with infrared
gfps um so that part uh for the Imaging um a more fundamental problem for the
Imaging is to be able to image voltage rather than calcium I would rather image
voltage so that I don’t have to I can put images like this in which instead of
measuring the voltage with an electro we measure the voltage optically because if we could do this then we wouldn’t be we
would be able to map all the connectivity all the cells to all the other cells because then we wouldn’t
need any electrode okay that’s a more fundamental problem but um yeah in terms of the uh
uh this um encaging of renum compounds we’re encaging with 800 nanometers um I think the encaging
becomes less efficient if you move deeper into the red so yeah maybe to play the piano it would be good to have
maybe a laser uh two lasers one to image maybe a 1064 and another laser to un
cage at 800 or something like that yeah okay so two questions theum
compounds you when you were talking about brain slices and also the mouse work it sounded like those are applied
on top of a slice on top of the brain topically yeah is there any chance of being able to get these sort of inen so
when you’re you’re looking Ino Mouse exactly yeah that’s right so so in the with our collaborators they essentially
put them in the CSF and I actually I at the beginning I thought it was a crazy idea I was discussing with Rothman you
want to put this in a patient you want to inject Ruby a patient said listen the epileptics that are not responsive to
pharmacological treatments they walk around with canas in their csfs already so they have already they’re injecting
into their CSF some of these antipeptic drugs so it’s trivial to put in Ruby Gala in the
entire brain on the inside okay so so that part uh in in this case it’s a
mouse with an open skull and he’s putting it on the top and we’re putting it um in the CSF okay and the second
question so the you know I’m sure you’re with this workout of Jia Farms we’re using uh slms to improve point spread
functions deep in tissue is there a hope that you can combine these two things on a reasonable time scale yes absolutely
we we sort of uh we were discussing with uh bet six group using
slm for that purpose before they even started working on that so so I take
part of the credit for that work even though we didn’t do it yeah so it’s absolutely I mean you in principle could
do it all together in one software suit you you’re able to just press the button said here I want to optimize and I saw
some very nice work today of pulse shaping that could be done with slms on the fly to essentially optimize the
delivery of light to different depths and say well not only are we going to be playing the piano we’re actually compensate while we’re at it we’re going
to compensate for the for the uh scattering of the tissue in 3D and that
may depend from animal to animal or from piece of the brain to piece of the brain
so the beauty of it you could do it all calculated in software and solve all these problems at
once if you want to play s
y means you need toate a sequence that’s right how much computation work is that
well so we’re we’re doing a little bit of that we have slow slms that are modulated at 60 HZ or so uh and the way
we’re doing it we’re actually building a series of movies or sequences offline and then we essentially upload that uh
and in those uh sequences we have a series of Imaging targets and then in
the middle of that we have some of encaging targets okay so that part it’s
it’s not that complicated essentially you know slms the the ones that we use are are offshoot of the consumer in
electronics Industry know so they use movie formats they can bring take in a u
t file sequen of t files so that’s essentially what we’re doing I to to
make the movie though to form the movie that you actually want to upload yeah so
to decide which NS to play yeah so that that we haven’t really gotten very deep
onto this but some of the ideas that we discussed is using uh compressive
sensing that I’ve seen in U discussed today um so because you may be able to
get away if if you assume sparseness if you assume that the circuit is sparsely
connected you may be able to get away by using uh one of these compressive sensing strategies and without having to
sample every possibility but sampling playing the piano with five news at the same time and using a random assortment
of those you may be able to solve The Matrix of connectivity so if if I were to do it tomorrow I mean we’re still not there
we’re still fighting these other battles but if you were to do tomorrow I would suggest maybe that could be one one
interesting experiment to use a compressive sensing strategy to design these patterns of movies to play the
piano and see how far we get in terms of predicting the
[Applause] connectivity