No you live by.

It live.

A lot.

Right.

Now.

But.

Well thank you very much shopping for
inviting me.

Thanks for all of you for coming here and

I can say that I've already
enjoyed the warmth of Atlanta.

Both at a personal level and for

those of you who know anything about
what's happened in the northeast.

This year.

At a physical level.

So it's a great pleasure to be here and
I'm going to tell you about my labs

efforts to perform functional
molecular imaging in the brain and.

Let me begin with this.

So one of the greatest scientific and

engineering challenges of the twenty
first century is the challenge of reverse

engineering the brain and
what's reverse engineering the brain.

Well it means building
an understanding of brain function

to the point where one could build
machines that emulate brain function.

Possibly figure out interesting ways to
repair brains and obviously understand

how brains work and probably hardly need
to tell you that the applications of

being able to reverse engineer the brain
are multifold and range from artificial

intelligence design of neural pretty she's
neurological treatments and obviously

my personal favorite basic science just
understanding how the system works.

So in some sense reverse engineering
the brain is the goal that

all of neuroscience strives towards So
why is it particularly

an engineering challenge as well as
well as a basic science challenge and

I think this picture shows
some sort of you know

egg headed types working on what
presumably is a dysfunctional brain.

You see it seems to see as the Bryza
chair illustrates part of the problem.

Notice what kinds of tools they're using
to do this work and I think you might you

might imagine that using a set of wrenches
on our brain something that really

performs unbelievably complex tasks is
is really not the right thing to do now.

First this is an analogy but the point
I want to get across here is that

one of the key reasons why this why
reverse engineering the brain is really

an engineering challenge is because we are
we need to we need to find better tools to

facilitate analysis of neural
mechanisms just to illustrate

you know the magnitude of the problem
that we're up against here is the brain.

Brains have an extraordinary degree of
complexity so many of you know that

human brains have on the order of
one hundred billion neurons for

even a rodent it might be many
millions each neuron has on

the order of a thousand synopses
connections to other neurons and

there are all sorts of different
cell types that are defined by.

They're patterns of gene expression
neurotransmitters the chemical signaling

molecules that they use and
of course highly complex anatomy and

connectivity in a structure like this.

So it's an unbelievably complex task.

Yet.

The standard the gold standard technique
for trying to understand brain function is

based on tools that were developed
decades ago electrophysiology an approach

where Microelectronics are buried into
the brain and record the electrical

impulses which are folks of the lingua
franca of neural activity but generally

only from a relative handful of cells
across the entire brain so if you look

at the fact that we have you know millions
are in the indoor billions of cells.

Usually we can only get these traces
from at most a few hundred cells.

So we have very specific information
about electrical signaling but very poor

sampling and very little contacts about
where in the brain and how cells.

We record are connected to
the larger neural networks.

So how can we get a more
complete view of brain function.

Well we're very fortunate that
there are techniques that allow us

to measure brain activity at a whole
brain level and Nell the most

popular sets of techniques are based
on something called functional magnetic

resonance imaging which probably many of
you know about in detail are F. M.R.I..

F. M.R.I. allows us to to obtain
maps like the one shown here that

illustrate the distribution of
neural activity across brains and

this is data from rats that was getting
an electrical stimulus to it's for

Paul and you can see that there are parts
of the brain in the region called some at

a sensory cortex and in a secondary
region over here that light up

when the four Paul gets stimulated and we
can detect this using F. or I over here.

This panel compares and electrical
recording of neural activity in this

region ever here shown in gray
so-called Alef P. with the F.

M.R.I. recording this is from
a contrast technique called bold.

And you can see that the trajectory
that F. M.R.I. measures is

reasonably closely correlated with would
be electret with the electrical trace so

it is a measure of neural activity and
it's really a unique tool.

Because it's noninvasive
it's based on M.R.I.

this medical imaging technique
that obviously owes its success to

the fact that we don't have to slice
up patients to get internal images and

we can see parts of the whole brain
using this using this technique but

it's also a very limited tool for
analysis of neural systems while limited.

Because the contrast that gives us
the ability to recognise brain activity in

these images is based not
directly on neural activity but

on the coupling between neural sneer aunts
and what they're doing and blood flow and

this panel over here illustrates
a comparison between a set of neurons

that corresponds to probably about two
voxels two pixels in this image here.

And a corresponding network of blood
vessels in a roughly comparable size and

what you can see is that there is this
intricate network of neurons that are that

are connected like a falling web and
we know that many of those connections

are functionally importance
by this we can see if we look

at the blood basket at the vasculature of
the blood vessels in a comparable region

that the spacing in the blood vessels
obviously much more separated.

This is this is a structure that
responds with a point spread

function on the order of a millimeter
to changes that take place here.

Why is it coupled at all because of the
metabolic demands of of the neurons and

the chemical messengers that they
release that in turn trigger response

in the basket.

So we obviously lose spatial information
as we go from neurons to vessels.

We also lose temporal information I think
you can see that over here the F.M.R.I.

trace is to lead and

broadened with respect to
the underlying electrical activity and

what may be most important is that we
lose information about specificity.

When we look at neural activity
through the window of F. M.R.I.

and why is that well here's
an example that illustrates this so

imagine a hypothetical neural network
that's made up of excited Torrie cells

that use a particular nerve transmitter
and inhibitory cells that use a different

neurotransmitter color coded in red for
an addition and green for excitation here.

Well you can imagine that there.

This neural network might might be
active with different balances of

excitation versus inhibition
as indicated by these.

Bar graphs over here.

Yet the metabolic demands
of those two situations

would be different would
be quite comparable.

So we might see a similar F.M.R.I.
response for really radically different.

Neuronal patterns of neuronal activity and
that's an illustration of how specificity

the lack of spare the loss of
specificity in moving from neurons to

to F M R I keeps us from
interpret at a functional level.

What's going on.

So how can we do better.

Actually only if the side.

How can we do better.

So.

I've shown you that we have on two
ends of the neural recording spectrum

electrophysiology a very specific
technique that provides very poor coverage

and on the other end of
the spectrum we have this F.M.R.I.

often based on contrast called bold
that gives us whole brain coverage but

limited specificity in resolution.

How can we merge the advantages
of these two techniques.

So my lab spends most of its time by
trying to create molecular tools that

are detectable by M.R.I. but that responds
to neural activity at a molecular or

cellular level so the idea there
is that we we have a molecular

level readout that that includes
the specificity of neural signaling and

that's combined with the M.R.I. the
ability to image these things by M.R.I. So

this is part of an overall kind of
emerging field that's often called

Molecular imaging we try to make
molecular imaging agents that detect.

Neural activity and
specifically their M.R.I. contrast agents.

What are M.R.I.
contrast agents Well most M.R.I.

contrast agents or paramedic chemicals.

Here's a quintessential one ever here this
is a molecule called gadolinium D.T.P.

A which was one of the first.

F.D.A. approved M.R.I. contrast agents and
most of them at least the power magnetic

ones exert their affects by
interacting with water molecules many

of you know that when you look at M.R.I.
image you're actually looking at water.

The end of March signal of water
molecules and it's because

contrast agent interact with water
molecules that you get these are facts and

in particular the small
magnetic perturbation due to

the paramedic neta chor of this molecule
in this case an atom of got a little.

Creates two types of affects on water
molecules that promote something called T.

one relaxation.

Which can be detected as a brightening in
M.R.I. scans that are T. one weighted so

more contrast agents if it's a T. one
agent will brighten an M.R.I. signal and

some contrast agents also create so-called
T two relaxation of facts those are often

referred to as T. to contrast agents and
they will tend to do.

Arkan M.R.I. signal as a function
of their concentration so

if you measure concentration going
up here at C one agent will brighten

it he two Agent will darken and you're
going to see examples of both of these

things in the rest of my talk so that's
why I'm taking the time to explain this.

Now the beautiful thing about M.R.I.
contrast agents is that they don't

just do this in test tube so these
are images of micro titer plates here.

That's pretty easy but
if you introduce these things into

living organisms they produce
similar contrast of fact.

So this is.

An example that follows the work of
Alan Korecki at the end I age where he one

contest agent manganese has been injected
into the I have an unfortunate rats.

And you can see that
the I became very bright.

It's a T.
one agents there were things same effect.

That's illustrated here.

And part of what's neat about this
particular example is that the contra

stage is actually transported along
fibers that connect to this eyes so

here are two regions of the brain.

This is called without limits and

this is called the superior click
us that are connected to the eye.

The contrast agent moved to the structures
and so we can see them as hyper intense or

bright spots.

Now my lab is not just interested
in tracing anatomy and

in lighting up certain areas
we're actually interested in in

detecting neuronal function
ultimately in real time.

So we want things like this that
brighten or darken dynamically

as a function of what's going on in the
brain in response to our own elective and.

So the challenges involved in this of
course are designing validating and

applying effect of contrast agents for
M.R.I. studies of neural systems and

you're going to see as I talk us forward
that a lot of this has to do with

designing appropriate chemicals that
have the properties that we need so

how could we do this.

Well you can imagine that
we could design M.R.I. can.

Trust Agents that target a bunch
of different aspects of neuronal

function neurons do account
all kinds of stuff they signal

extracellular early they release
neurotransmitters first and foremost

they're also ions in the extracellular
space that change their concentrations and

we can detect all those things interest or
signaling is a key

key aspect of neuronal function many
of you know about neural plasticity and

signaling pathways that involve calcium
and kindness activity inside cells.

Those could be handles on or
on the activity that we could sense with

a contrast agent and gene expression of
course is also a potential target and

we've done workin in all these areas but
fortunately for you.

I'm not going to talk
about all of those things.

I'm just going to talk
about a couple of areas.

So let me just summarize
briefly the main topics

that I want to cover in
the rest of my talk.

And I'm going to speak primarily about

two things The first is neurotransmitter
sensing with engineered proteins This is

the work that shopping referred
to in his kind introduction and

here I'm going to tell you about
a new strategy for producing M.R.I.

contrast agents that's partly interesting
for its chemical content in a sense.

I'm going to show you the first.

Rudimentary M.R.I. that we've been able to
for perform with a molecular sensor and

then I'm going to show you work that's
now ongoing in my lab towards mapping

behaviorally relevant patterns of
neural activity with these sensors and

then I'll transition to
the second topic where I'm going

to tell you just a little bit about
some new genetically encoded M.R.I.

probes protein based M.R.I. probes for
detecting interest cellular signaling.

Here I'll tell you about in particular
protein contrast agents sensitive

to calcium and
kindness activity which are.

Key interest signals and.

I'll show you some initial evidence
that we have about calcium detection

inside cells and this is some of
you may not care about calcium

it is actually very important and in fact
optical imaging of neuronal activity

is typically based on calcium signals
inside cells so it's actually a really

important potentially and
generalizable handle on neuronal activity.

OK let me start out with telling
about neurotransmitter sensing and.

So we have the idea that it might be
it might be good to be able to detect

specific signatures of neural activity
outside cells in the brain out

why outside cells because we don't have to
think about delivering the contrast agent

why a neurotransmitter is because cells
in the brain often play functionally

distinct roles that are dependent on
the neurotransmitters that they use so

inert transmitter dope I mean
is a famous example it is

particularly important in some some
things you may know about reinforcement

learning is an example some of you know
about the so-called Skinner box where you

put a rat in a box and get it to
do whatever you want in return for

some kind of reinforcement some kind of
reward and the nerve transmitter dopamine

is key to the establishment and
maintenance of those kinds of behaviors.

Many of you also know about the role
of dopamine in movement control and

in particular these things are famous
because of their dysfunctions so.

Dysfunction of movement control is
Parkinson's disease which involves

the death of dopamine producing neurons in
a part of the brain called the substantia

nigra addiction is often thought of as a
dysfunction of the reinforcement learning

circuitry because of the chemicals
the addict and the addictive substances

that interfere with these neuronal
pathways so dopamine is important.

We wondered whether we
could try to understand

the functioning of dopamine in neural
networks by producing a contrast agent

that could detect opening and
a student in my lab McCall Shapiro

had the idea of trying to make
a document sensor based on a protein.

B M three P four fifty.

So why why work with this protein Well
first of all for fifty's like globe and.

Our team proteins and their paramedic so
they seem which is diagrams in this

brownish color here makes this protein
an M.R.I. contrast agent because it's

a paramedic remember the slide
I showed you a few about.

There are other reasons to work with
proteins and one of the key ones is that

we can engineer proteins with a speed and
facility

that at least in my lab can spring to
the design of synthetic contrast agents.

And that was important in part because
when you take an arbitrary protein like

this even a paramedic protein that's
already in M.R.I. contrast agent.

Well it doesn't necessarily do
what we want in particular for

instance this protein binds a league and
called arachidonic acid it's a fatty acid.

That is very interesting in biology but
it's not a neurotransmitter and

it's not what we wanted to achieve.

So we tried to engineer
this protein to bind and

to sense it by M.R.I. How did we do that.

Well we can we partnered with a lab the
lab of Francis Arnold at Caltech who has.

Designed high throughput ways to engineer
these P four fifty proteins to do all

kinds of stuff here we were engineering
it to sensor transmitters and

it uses an approach cold's directed
evolution how does this work

where we take the gene for the protein a
piece of D.N.A. and we randomly mutate it.

So we get a bunch of different copies of
the D.N.A. each with a different mutation

color coded here then we take each
one of these genes put them in E.

coli and express them in multi well.

We lice the cells break them apart.

And then assaye them in the micro titer
plates we add successive amounts of

dopamine or of the thing that we don't
want these proteins to bind arachidonic

acid and what we try to find are those
mutants that bind Open mean better and

bind arachidonic worse
arachidonic acid worse.

This is a screen that's based on
an optical measurement of affinity.

But of course what we really wanted to
detect were M.R.I. changes and so we then

do a detailed characterization of the hits
from this high throughput assaye the hits

the ones that show the best properties
by M.R.I. and optical measurements.

And what makes this not just a screen but

actually evolution directed evolution is
that we then take the best variants that

we isolate at the end of this procedure
and we feed them back into the cycle.

So we go through successive cycles
of mutagenesis selection and

repeated Genesis kind of the way
at least most people think.

Animals of all the nature or
even plants right.

Even bacteria.

OK So this is directed evolution.

And this is interesting in part because
it's a new approach to generating M.R.I.

contraception so
people abuse this kind of technique to

modify protein functions over
some period of time but.

Not for M.R.I. contrast agents.

How did it go.

Well the first thing we needed to
tell was is this protein in M.R.I.

contrast agent and does ligand binding
to its alter its properties as an M.R.I.

contrast agent and we detect We started
out by measuring something called

the relax evictee which is kind of
the strength of the contrasts age and

it's sort of like quantum yield for
a floor of four.

So we measured relaxed city of the protein
the wild type starting protein

in the absence of ligand in
the presence of the natural leg and

directed on a cow sued or
in the presence of very high concert.

Of dopamine far above anything
physiological and what you can see

in both of these cases where we added
Liggins is that the relax it is to press.

So we did get a change in the relativity
and this is sort of abstract but

I think you can see the meaning over here.

The higher the real activity
the brighter the signal in M.R.I.

again based on the T.V. One relaxation
effect that I told you about and.

In particular these Liggins
both dim the signal here.

So we can see that the protein at least
is capable of functioning as an M.R.I.

sensor initially for
either arachidonic acid or

huge amounts of dopamine
could we make it better.

Well it was key actually to our approach
that the M.R.I. signal changes accompanied

by an optical signal change so
we got shifts either in the red direct or

the other in the direction of the red
direction depending on which leg and

we added and that was really useful
because it enables us to assaye

the proteins at much lower concentration
than we could do by M.R.I. and

we can measure these titration curves so
what you can see here.

Is that the binding affinities
measured optically now for

Iraq atomic acid is around
a few Michael Miller.

So ten to the minus six and
concentration for dope I mean it's

really Miller intensive on this three and
why is this happening while it's happening

because we think these Liggins
whether they be or economic asset or

don't mean are coming in near the team
the paramedic part of the protein and

kicking out water molecules so
that's probably how this thing works.

We now actually I don't
have this slide of it but

we have crystallography data
now that backs this up.

We've solved the structure
of one of these proteins and

complex will begin that shows
of the water is being displaced.

OK So this is how it works
now how did the evolution go.

This slide shows the basic results and
what these are are titration curves again.

Now Showing binding of dopamine so
dopamine is on the horizontal.

Access to different variants of
the variance that we produce after

different numbers of cycles
of that evolution process and

what you can see here is that over four or
five cycles reaching this.

Color coded yellow and purple mutants
over here we're able to achieve

affinity down to about single
digits of micro Moeller.

So we started out with affinity on
the order of one million dollars

we found up with several hundred fold

improvement in affinity using
this direct evolution approach so

it's a three hundred fold improvement
which I was pretty impressed by.

When I saw it when I saw it and
some of you may be wondering well OK so

it sticks to dopamine better but
what if it sticks to everything.

What this graph shows is that as
we knock down the affinity lower

Katie improve the affinity I'm
sorry lower Katie for dopamine.

We lost affinity to record on a graph of
this is over rounds of evolution here and

you can see that after about
three rounds of evolution.

We have negligible affinity for
the while for the natural link and

directed on a cow.

So we do have some specificity for do.

I mean does it still function as an M.R.I.
contrast station all those data

are about affinity and what this graph
shows here is again this relaxed city

parameter the strength of the contrast
agent in the absence on the left or

the presence of dopamine and
this is for our best to mutants here.

And again you can see there's
a contrast difference and

of roughly eighty percent change and relax
of it which is almost all or nothing.

You can use these relics of any changes to

measure affinities in the numbers
come out pretty much right.

You can ask well does it bind
other neurotransmitters so

it doesn't bind arachidonic acid.

But it may bind other neurotransmitters
and so we looked at a whole collection of

neurotransmitters that might
produce relaxation changes.

And we found that most of them
don't produce any effect.

So for instance some of the most.

Famous.

Neurotransmitters glutamate glycine Gabaa

no effect at all with
these contrast agents.

Nora Ephron which some of you know is
chemically very very similar to dopamine

did produce a response but
it was a weaker response and

the inset shows that the difference
in affinity is about a factor of four

between dope I mean in or of an effort
in for one of her mutants Sarah tone and

another famous neurotransmitter that we'll
hear more about briefly later produced in

fact so that mutant the one color coded in
yellow that showed a response to certain

an abbreviated five H T That one is less
specific and as a result we use it for

we use the more specific variants for
more of our IN VIVO tests.

Nevertheless these data show us
that we really have here the first

neurotransmitter sensors for M.R.I..

And we thought about taking them and
using them in for

initial experiments in vivo even though
of course we can make them even better.

So what was the first experiment that
we did well wasn't really in vivo

it was in in vitro experiment with cells.

And what was involved here.

What we have cells type
of cell called P.C.

twelve cells that professionally
secrete dopamine.

And we could elicit dopamine release from
these cells by hitting them with potassium

and ionic stimulus.

If we give sodium you know
dopamine release a curse but

it's a control because the onyx the onic
nipple ation is quite comparable.

We put our sensor.

B M three H. or one of the variants of
B M three in the super named above these

cells and that's what's going to enable us
to detect opening release in this test.

OK How did this work.

Well we're here we're looking at
relative signal relative M.R.I.

signal in a microchip in micro tighter
wells that have these conditions in them

and you can see here that the signal
in the case of the potassium

stimulus is lower than in the case of.

Asserting stimulus and

actually I should have pointed
this out on the slide over here.

So note here that the relax a bit
in the presence of dopamine

the strength of the contrast agent
goes down the brightness goes down.

So this result over here is exactly what
we would predict based on binding of

dopamine to the contrast agent the signal
should go down here is a modest decrease.

We can collect data in
a way that enables us to

measure the underlying
parameter called our one.

And that's changing also.

In this case because we knew the exact
amount of contrast agent that we had.

We could actually use these data.

To form a quantitative estimate of how
much dopamine was present in this mixture

and what this graph over
here is showing is the don't

mean concentration in the super Natan
detected by the contrast agent either

with the potassium stimulation or
with the sodium control and

you can see that the concentration in
the presence of potassium is somewhere in

the fifty's of Michael Moore whereas it's
around twenty micro in the case of sodium.

So it's a reasonably big change.

Now is this right and
this is what we calculate from our M.R.I.

data is that right.

Well we did it totally independent
asked say something that doesn't

involve M.R.I. at all.

It's based on an allies a technique
you know plate reader and

the body that actually came up with
independent measurements of dope and

concentration and I think you can
see that they're not spot on.

But they're pretty close.

So in particular in the task in case
where with an error in the case of sodium

will we see a pretty comparable difference
between these two conditions so

I think we're able to do what
we can say from these data

is that semi quantitative detection of
dopamine with these agents is possible

at least in vitro now can
we do anything in vivo.

So this is a raft and
we OK so we did a very crude.

And in vivo we use the rats brain as
a test tube but the right was alive so

it's not really a test.

We made to craniotomies we drilled
to kohls in the rap skull and

we injected Kenya Lee to
infuse in our contrast agent.

And in one case we infuse
the contrast agent only

on the right color coded in blue.

And in the other case we infuse the
contrast agent with our own dope I mean

why did we bother doing this.

Well one of our greatest fears with these
things was that as soon as we injected

them into a living environment they would
get chewed up and stop functioning at all.

So we wanted to have a case where we knew
that we had over mean there are sufficient

Open mean to produce a significant change
in the binding of the contrast agent and

that's that's why we did this test and
let's see is this going to work.

You know.

OK I'm going to touch my computer
here to play a little movie.

What I'm going to show is
a movie of the injection.

Into the rat's brain over here and

over here on the left is the side that
got the contrast agent plus dope I mean

on the right is where the contraception
is going on without opening.

And I think all of you can see
that there's this brightening

here in the M.R.I. contre in the M.R.I.

scans as we infuse the contrast agent and
playing it again.

Without opening and
the brightening is gone.

Basically on the left.

Why is that it's because
remember that dopamine

turns off this contrast agent
that's a contrast agent that works

in the absence of dopamine in
the presence of gets turned off.

That's why it's dark on the left and right
on the right where there's no dopamine.

So this was nice to see we can
quantify this basically by looking

at the average signal over these time
courses you can see that there's a big

signal change about twenty five percent
where we put the contrast agent in but

not when we put in dopamine with it.

You can look at a time course
here this is the contrast agent.

Plus an honest open.

In the gray area here is where
where we were doing the infusion

then we stopped the infusion and
it's just sitting there.

What about a control what we
actually have a great control for

these experiments specifically we have
the wild type protein that wasn't

engineered to bind up me one question you
might ask is could the protein be doing

something to the system that interacts
with dopamine in some nonspecific way.

Well if we inject the wild high
protein the one that doesn't bind open

specifically then we could test for
that and I think what you can see here and

this in these data is that we're seeing
very comparable results in the presence or

absence of dopamine when we
infuse the wild high protein so

what this shows is that we really
are specific to dope I mean.

And these data though
a bit crude are the first

evidence of dopamine detection in vivo.

You know how can we detect a real process.

So we cheated in that experiment
because the dopamine came from us.

We're not looking at a biological
process all we're looking at is

you know what we're infusing the good
thing was that what we infused didn't go

away immediately it didn't seem to be
disrupted by the living environment but

it's still not as good as
detecting dopamine release in vivo

indulgence dopamine release and
even so we designed a paradigm.

So it's sort of somewhat similar to
what I just showed you that test for

that and what we're doing here is we're
infusing the dope I mean sensor here

called eight CA on one side or
infusing the wild type protein

to control protein on the other side and
with each of these proteins.

We're giving pulses of high
potassium stimulation.

So remember that potassium with the
stimulation that evoked opening release

from our cells in our cell experiment.

Well dumping potassium into the brain

evokes you know essentially
broad brain activity and

if you do this in an area of the brain
where dopamine is naturally released

as a neurotransmitter then you'll get
a lot of dopamine release in this as well.

Well worked out in the literature.

So this was a stimulate.

Paradigm and then.

So we did this.

And then what we did was more or less
a sort of a conventional style functional

M.R.I. analysis where we correlate
the signal to the signal trace as

a function of time with the stimulus and
we found where in the brain.

We found where in the brain was the M.R.I.
signal correlated to the stimulus

and here the stimulus again being
these three pulses of potassium and

what you can see over here is that
near the infusion point of our sensor

we found pixels that were correlated.

Whereas near the control infusion
point we didn't find that this is from

an individual animal and
we could take a group of animals and

combine the analysis from
all those animals and

actually map out the signal changes
in response to these stimulate.

These maps don't really
mean anything because

you know they're influenced by where
the contrast agent spread from the from

the from the to tip each injection
is those of you know who've done.

You know in infusions of things each
injection is a little bit different.

So this isn't really
a biologically relevant result.

But what it shows is that at least
a biological release of dopamine

can be detected consistently
by the sensor and

of course the control
doesn't show this response.

We could see little
difference in the M.R.I.

signal over time and if you look at
the signal trajectory the map of signal

over time during an injection cycle you
can see the signal change building up and

then subsiding over the course of
one injection cycle over here.

So this is again a relatively crude
experiment obviously we're not learning

new things about biology here.

We've used this crude
injection protocol but

it is the first time that we or
really anyone else.

I think has been able to demonstrate
neural activity detection with an M.R.I.

based sensor so it's a key step
in our overall plan to try to

create a tool kit of these
molecules that deter.

Act aspect of your on neuronal activity.

What next.

Well these M.R.I.
sensors are not are not very powerful.

So one of the things that.

We have one of the reasons why we have to
do these injection procedures is because

we need to get enough contrast
agent into the brain to see it so

less invasive methods
may not be good enough.

Could we boost their
efficacy as contrast agents.

Well that's equivalent to boosting their
relex civically this or one quantity.

That's a measurement of the strength
of the contrast agent and so

we have the idea of trying to do that by
increasing the spin the electron spin

really increasing the essentially magnetic
power so to speak of the molecule.

That's acting as a sensor and
my post-doc Victor let me build student

now post office will evolve developed
a pretty cool system for doing this.

So we're producing all
these proteins in bacteria.

And it turns out that if you modify
bacteria with something called with

a porphyrin transporter called
Shoei you can get them to take up

whatever whatever poor for
an analog you want to the poor.

For in this the metal complex that lives
in the heart of all of these proteins and

makes them contrast agents.

And so our idea was to fool these
cells basically to take up a manganese

analog of what would normally be an iron
porphyrin in the heart of the proteins.

So we use the system to do this and

we were able to make fully folded
functional proteins using the system.

These are spectra maybe a few
of you out there chemists and

you feel more at home with this than
anything else that I've shown and.

What they show are that we can take
essentially arbitrary metallic poor

friends and dump them into these
cells get them to be incorporated.

These are in the inset here.

So called CD spectra that show
the folding of the protein

of the perfect overlay of the blue and

black traces here some of you probably
can even tell that there are two traces.

That indicates that
the proteins are well folded.

So they are just like our sensors.

But now they have manganese
instead of iron and

this is the key point here which
is that if we now look at their

key one Relax city the same strength
of the contrast agent business.

In the presence or absence of Lincoln's A
minus leg and in the solid colors and and

plus like in the in the pale colors
you can see that the manganese variant

has over two full better
response properties so

it's a stronger contrast agent and
it has a stronger response.

And so this this worked as an approach for

you know reasonably seriously boosting
the strength of this contraception and

I think it's again that that
translates into our ability for

instance to use a factor of two or two to
three fold less contrast Asian in vivo or

potentially to get a factor of two to
three fold more signal in vivo for

it as a specific experiment relax
it improved by a factor of two and

a half right now we're impatient.

So we want to actually start trying to
learn things about biology using these

tools.

And what this figure shows here is a raft.

That is obsessive Lee pressing this lever
in return for rewards in return for

reinforcement in this case the
reinforcement it's getting our electrical

impulses through an electrode implanted
in part of its brain called the M.F.B.

It's a pleasure center.

So the rats actually having
a pretty good time even though

it's embedded in an experiment
here you can see right.

Doesn't look happy.

Anyway so we know the don't mean it's
really important for this process and.

It's actually this emma the reward
paradigm is a very commonly

used paradigm in studies of addiction.

Could we mapped over.

Unreleased patterns as a function
of the stimulation paradigms and

the behavioral contacts
in this kind of system.

So this is very preliminary
results results and

they're from my postdoc tech lonely and
he started infusing the sensor.

Into rats that are being stimulated
in the way that this movie shows and

what you can see at least in this this
example here is that he's starting to

see photos side of what looks like
activity associated with the infusion

of the contrast station and the
stimulation here in a region of the brain

called the nucleus accumbens which is one
of the main dopamine targets in the brain.

If you look in there.

This is this is sort of choppy but
I think you can see each one of these

great bars here is when
we turn on the stimulus.

What makes the rat press.

And I think you can see that there are
little dips in the signal that tend to be

correlated with out if you average
over multiple cycles you see a pretty

decent trace of signal as
a function of stimulation here.

And so I think this is you know
this is quite preliminary data but

it's a step towards taking
this new technology and

trying to use it in a behaviorally
significant context where we want to get.

Of course two and
ultimately three dimensional maps of

dopamine release patterns in the animal
ultimately while it's performing behavior.

So this is towards dopamine
release mapping in reward.

We're also combining this with sort of
more conventional F. M.R.I. studies and

on a few IOS who just graduated
from I love some decent

work sort of learning about the overall
network that these things function and

so I think it's it's going to
be interesting to rection.

All right.

So I've spent much of my time so

far talking about dopamine
neurons are very important.

I illustrated that before and
we designed these M.R.I.

don't mean sensors to detect
what these neurons are doing.

Well that would mean neurons project
to large parts of the brain but

they're really only
a small part of the story.

And you know again I flashed
these figures up here.

How can we see more how can we see more
about what's going on in the brain.

How can we target other aspects of
neuronal function that are represented by

these different color codes here.

Well one obvious idea that some
of you may have thought about

is that we could use exactly the same
platform to try to create sensors for

different neurotransmitters and
we're doing that.

So in particular here.

We've developed and
this is again a preliminary to

republish data developing a sensor for
search home and

that's another neurotransmitter another
month to mean that many of you know is for

instance the target of S.S.R.I. is Prozac
and so on related to preside and what

this figure shows here are real acts of
any changes from a Sarah tone in sensor.

So we can see change.

That's quite comparable to what we observe
with the dopamine sensor these numbers

here in the red box show the relative
specificity relative Bonnie for

me I'm sorry for these different.

Transmitters and you can see that
it's very specific for Sara tone and

the Katie is much lower.

This is and Michael Miller So
it's Michael Moore and

we're able to do these kinds of crude
experiments in vivo with this so far.

Obviously we're moving towards more
interesting in vivo experiments again.

But a more general approach.

To looking at your own elective eighty
would involve targeting signals that all

neurons share.

And that therefore could where we
could apply the sensor to any part of

the brain and
detect signaling patterns and so

we've had we have a reasonably
longstanding collaboration with a chemist.

Steve Lippard at MIT to develop small
molecules that are contrast agents and

that detect ions there
based on Porfiry ins.

And the idea behind that is that poor
friends at least often are so permeable so

we can modify poor friends for
their self permeable I should point out.

They're also power magnetic if
you put in the right metal.

This is something that you've seen
already because they're in our proteins.

So we modified them to
be analyzed specific and

in data that we published pretty recently
we injected them into rat brains we could

show nice staining patterns in a region
of the brain called the hippocampus and

then we did sub sailor fractionation of

these brains to show where within cells
our contrast agent is sitting and

we found that it had most of it's on
the side it's also really a cell permeable

there on sensors and
right now what we're working on.

Is trying to make a calcium
sensor based on this platform and

this is really important I
think because again calcium

signaling is really the sort of gold
standard for cellular neuro imaging so

most of the optical studies
of neuronal activity use.

Calcium base agents and
this would be the M.R.I.

equivalent of that we're far along in the
synthesis but we don't have it working yet

instead I'd like to spend the last
really five minutes of my talk or

ten minutes talking about a different
idea for trying to sense intracellular

signaling one that doesn't
involve the small molecule or or

protein based agents with one metal on but
instead involves

nano particles and the work started
in two thousand and six with

a paper that we published this is work
of touch on the Tennessee vision my lab.

She made and then a part of
a magnetic nanoparticle based M.R.I.

calcium sensor Why did we work with
that of particles while some of you

also work with magnetic now to particles
and one of the reasons why you do it

is because they are known as
the kind of atomic bombs of M.R.I.

contrasts agents there are particularly
strong particularly on a per particle

level as contrast inducers So
initially it was.

The work of Ralph Weiss later and
colleagues that show that if you cluster

magnetic nanoparticles there relax if
he changes their potency as M.R.I.

contrast agents changes they are two T.
two agents.

So it's their so-called T.
to relax of eighty.

That's changing in these examples.

What we did was we functionalize two
populations of magnetic nanoparticles

with two different proteins a cow seen
binding protein called Cow module and and

a peptide cold ORUs twenty that
binds only in the presence but

not in the absence of calcium and what
the way I usually describe these things is

you can think of them sort of like velcro
balls that get turned on by calcium but

are off when there's no calcium so they
tend to clump when calcium is there and

what these two panels here show.

These are so-called atomic
force micrographs that show

individual nano particles the scale
of our individual nano particles

in the absence of calcium
the presence of E.D.T.A.

and in the presence of calcium you can
see that these large aggregates form.

OK And you get a different mix depending
on exactly what conditions use and.

Obviously this would be interesting if
there weren't an M.R.I. response and

what this is showing is the teeth to
the underlying M.R.I. parameter associated

with these contrast agents as a function
of calcium concentration what you can see

is that as we add calcium we move
from dissociated to aggregated and

we increase the T two in this system and
that translates

into a pretty robust change in M.R.I.
contrast and you can sort of titrate up or

down the magnitude of the contrast effect
a penny on how you run the scan itself.

OK so it was a pretty strong
way to modulating M.R.I.

contrast based on these nanoparticles and
dependent on calcium.

But here's the catch these are big
nano particles they're about fifteen

animators in diameter.

How are you going to get them inside
a cell where the calcium changes

are taking place.

Good sensitivity but how to deliver
to sell so this is where we were

inspired by the work of a couple of
other labs on the protein ferrets and

ferrets in is an iron storage protein
of about twelve nanometers in diameter

that accumulates iron oxide in
sort of a biology biological

analog to a magnetic then a particle
that you might synthesize in your lab

the result is comparable in size to
synthetic in a particle and it's got T.

to relax sooty and
it also shows this clustering dependent

effect that I just illustrated in
particular if we clustered Farrington's

you know using a struct
evident by if an interaction.

We can get changes in T. to your one over
T two this plot it versus relative radius

quite systematically based on the amount
of clustering Now this is interesting for

yet another reason and this is part
of why we work on ferret and for

this purpose which is
that a couple of labs.

Eric Aarons at Carnegie Mellon and

we call naman at the vitamin
Institute had already shown that.

Over expression of ferrets and
in vivo actually can induce M.R.I.

contrast agent contrast changes.

So in other words there are contrast
agents contrast changes.

That we could potentially modulator.

If we could do something like this
if we could modulator the clustering

of these ferrets and particles within
cells dependent on the signal.

We were interested in and so

we started out with kind of a toy system
where we modified ferrets and genetically.

With protein domains that allow
them to cluster based on a signal

in the first signal that we
looked at was kindness activity.

We were interested in that because it's
closely related to calcium signaling in

living systems.

And it's an enzymatic process so there
are certain stoichiometric advantages so

we made a whole series of modified ferrets
ends with different domains on them and

in particular the the blue an aura.

Means of the ones that really matter the
blue domain is something called the kid

domain from code protein called crab and
that gets passed formulated by kinases

when it does it sticks to the orange
domain but not without phosphorylation And

so this is the picture of what we
thought would be going on in the system.

And when we tested it in vitro now.

We were indeed able to see relaxed of any
changes of this is are to the measure

of the strength of these ferrets and based
contrast agents and you can see that in

these two yellowed conditions
with phosphorylation or

the addition of enzyme prefix for
a lesion of the addition of ends and

the real activity goes up by
about a factor of two and

there's a corresponding image change that
that we could see in micro tire well OK.

We could do that we could use this to
monitor enzyme activity over time and

you can see that the more enzyme you
add obviously the more change you see

in a given amount of time.

So this showed that this kind of
affects clustering of ferrets in for

genetically modified ferrets and
could actually induce M.R.I. contrast.

But obviously this is a long way from
actually doing anything useful in vivo in

particular we had a lot of constructs
that went into make these proteins and

getting all these things to express
simultaneously in a given population

of cells is not easy.

We tried it.

So we needed to move to
a simpler system and

one of the key things that we tried to
do was to split up the two interacting

species between two protein platforms so
that I think one thing that you

could imagine is if you try to express all
those domains and one population of cells.

You might not get orange or
blue ferret ins.

You might get brown ferrets and
the have them all mixed up.

OK so we tried to address that problem
directly by fusing cow module and

this calcium binding protein
that I looted to before

that one type of one protein keratin.

And we fuse the interaction partner
R Us twenty two a different protein so

that we couldn't get the sort of mixed
formation of the lake immerse so

we wind up with two species the red
Eriksson's that have come on.

Julian and these pink green proteins
that have the binding partner and

here the idea was that in
analogy to our synthetic

nanoparticles the addition of calcium
would promote the clustering of calcium

is here the trigger not kindness activity
sorry for baiting and switching.

And we were expecting to
see these clusters again.

And so.

What this panel is showing is that
actually initially our results

were terrible.

So using the wild high variance
of these proteins comparing minus

to plus calcium we saw nothing much but
under certain conditions when we mutate

did this construct here
the pink construct.

We saw a pretty robust contrast change
between minus and plus calcium it

required getting the right balance of
the two interacting species and for

those of you who like to think about these
things that won't be surprising to you and

in fact it is a liability with
the approach on the other hand we

were able to see some evidence
of responses inside cells so

we took we took these proteins we stuck
them in Saw a particular cell line and

we could see clustering because
one of these interacting partners

was fluorescent kind of a neat trick
that Victor who did this work.

Built into the system and

I think what you can see here is that with
the addition of calcium on of mice and

which is the an eye on a for
that promotes calcium flux inside cells.

You can see these little clusters
of fluorescents building up here.

So there's some evidence of
fluorescence response inside cells.

Evidence of calcium responses.

But if we look at the progression
here well OK we went from synthetic

nano particles that are hard
to deliver to these concerts.

It's that that are just too
complicated to implement inside cells.

Now we've got this calcium
responding ferret's and.

That is still quite dependent on
the ratio of the two interacting species.

So we still want something that
was going to be more robust

to signaling changes inside cells and
really the question we asked was Can we

perform this kind of calcium sensing
that I showed in the previous slide but

with a single protein and this is
current work in our lab and we have

some preliminary results that are that
are pretty encouraging in this respect.

So we were able to make
a single ferret and.

A single ferrets and that's also
fluorescent stick it inside cells here

we're using P.C. twelve cells this is
the patient for the for the construct and

when we add this calcium on the mice and
again you can see this nice punctate

formation that probably reflects changes
in the clustering in the organization

of these constructs inside the cells of
course we've done appropriate controls.

Here you can see more blown up example
over here and possibly most encouraging.

There seem to be M.R.I. changes at least
that we can start to see when we take

these cells pellet them and scan them.

And right over here what we're looking
at is minus versus plus on of mice and

so again adding calcium in the right.

Absence of calcium here
we're looking at are to this

underlying parameter you can see a little
M.R.I. change here and an arc to change.

We've tried it with two different
variants of our constructs and

we see qualitatively similar changes
of this is showing the degree of

this decrease and here we have
a control protein that doesn't

show very much change although
there's something going on there.

All right so
this is really the first example of

intercellular calcium detection with
a with an M.R.I. contrast agent and

I should emphasize that it is
really particularly important.

This is being done with genetically
encoded contrast agents and

that's both because of course we can
use the gene constructs to deliver

the proteins by just expressing the D.N.A.
inside the cells but also because we can

potentially target different cell types
with these constructs so if you know.

Supposing we want to
sense dopamine activity.

Well we have a different way to do it here
if we can just target the the genetically

encoded sensor two cells that
used opening instead of also and

so we in context of some other work where
we're working with viral constructs that

can selectively in fact different
types of cells so this shows.

BLOOM a terrific premeal cells and
important cell type in the in the cortex

being stay and we can also use viruses and
we can potentially ultimately

perform cell specific targeted
functional imaging obviously we still of

a fair amount of work for that and
and I don't want to go off over that.

We're also working on ways to boost
the relax a bit if these ferrets and

sort of an analogy to how we
had to deal with a problem

with our protein based on things so.

Let me sum up.

I told you about my labs efforts

to develop molecular imaging agents
to sense neuronal activity so M.R.I.

contrast agents that responds to neuronal
activity with changes in their efficacy

and in turn with changes in
the brightness or darkness of an M.R.I.

signal in the brain and I feel this is
a sort of an approach that is pretty new.

You can see that many of our
results are rudimentary.

Nevertheless you can see that there's
a lot to do and we're trying to do it.

Approach that we call molecular F.
M.R.I. and

we're trying to get this off
the ground and we I focused my talk.

On a few different areas and
in particular I focused on our F.M.R.I.

first F. M.R.I. experiments with.

These metallic.

Metallic protein based opening sensors
that being three neurotransmitter sensors.

I also told you just in these last
few minutes about genetically

encoded calcium sensors and
I touched on a couple of other topics.

There are more projects that we have in
the lab that I didn't have time to talk

about obviously the next steps involve
doing a better job with delivery.

You can see that all of our experiments so
far have involved injections.

We want to cover larger parts of the brain
with these things and see more and

obviously we have a lot of a lot more
work to do to continue to improve and

validate these M.R.I.
contrast agents both in vitro and

in vivo ultimately leading to whole
brain molecular level investigation

of neural systems which is really
the long range goal of all this work.

I like to think of
people who did the work.

Let me call up some specific people
touching on it is the calcium nanoparticle

work.

Going on here type tech on lead to the
work with this in public porphyrin sensors

in vivo and he's also involved and

in the in vivo in the reward
related dopamine release work.

Victor low evolved worked
on the metallic substituted

protein based sensors and
he also made these ferrets and

based sensors for calcium Mikhail Shapiro

did the in vitro work with our
protein based oakum in sensors and

also our initial work with ferrets and
based company sensors and

Gill vest fire did the in vivo
work with the dopamine sensors.

We have some great collaborators Francis
Arnold has been our close collaborator.

She's a protein engineer
at Caltech she was working

closely with Recently Eric restocked and
earlier.

CHRIS of the infielder marrow.

I also alluded to our collaboration
with Steve Lippert on synthetic.

Just agents you have some other people who
work with I'd like to acknowledge funding

from all these agents and
smart molecules and all of you for

your attention so thanks very much.

Yes So there's quite a lot of work with
Magnes as a contrast Asian in vivo and

as you know Alan Korecki
was the one who first.

Explore that in detail and I think.

The lesson from that work is that

it does seem to be taken up in
an activity dependent way but

the dynamics are extremely slow and it
doesn't come out on a useful time scale.

So that's that's the answer.

I mean basically if one is
interested in performing F.M.R.I.

on anything like the sort
of behaviorally relevant.

Ronal time scale manganese I mean
it's better than nothing but

it's definitely not
the desirable end point.

OK So people done good
work with it up thank God.

Very well.

I thought.

Yeah.

So you're absolutely right that
that a contrast agent like you

know essentially any chemical intervention
is going to disrupt the system and

in particular the contraception
will do something called buffering.

It will soak up some of
the neurotransmitter that we're trying to

detect in much the same way that
a PH buffer soaks up protons and.

So currently.

A.

You know obviously the.

Hope in the long term of
course is simply to use not

did not to use enough to need
very much of a contrast agent.

Now we're not at that point yet we need
a fair amount of the contraception eyes as

I alluded to and so
we will get buffering and so

the there are two ways to deal with the
one of them is to design experiments and

look at phenomena where that's
simply not going to be a big deal.

So in particular if we're interested in
looking at what is the spatial temporal

profile of dopamine
release due to a stimulus.

Well that's largely
a feed forward question.

It doesn't even it doesn't really depend
on what dopamine is doing after it's been

released.

OK On the other hand it's possibly less
interesting for a behavioral standpoint.

So method one is designed to look at for

another look at phenomena that are reports
us fall less a number of approaches or

to is of course to do all the behavioral
tests in the presence of the agent so

that it least we have a good
understanding of what it might be doing

to to the actual function of
the neural network so we do expect.

Of course some interaction and
it will be important to do those tests and

we're set up to do them too.

And they're.

You know.

So for the work that I showed
here we used error prone P.C.R.

So I mean this is basically the broader
question for those of you who don't know

about these methods is you know
essentially did you just mutate every part

of the protein and assume that each or
at least allow each

one to have an effect on the properties
of the protein the answer is yes.

In this case we did.

In some other work that we've done.

We did mutate specific residues that we
thought might be particularly important.

Yeah sure your mouse.

Thanks again the other.