Though you may or may not know of
Barack I can imagine someone who

works in the General Theory of
Computation while he was not a very

very has been one of
the leaders in the field.

I mean Barack has worked on many of
these seminal problems in bio physics.

He's extraordinarily well known for
his work.

Looks like a stand study in the Basically
because I guess your ninety's.

I mean what Barry continue on on and

touches you thinks is very important and
uses.

Extreme physical insights not just
merely Let's do a large calculation and

a clear victory.

He actually tries to develop
an understanding of fundamental principles

that he's been interested in other
things like the mutual protein

fold space are the sorts of things you
know how does one classify folds of

one canvas a lot of implications
functional inference.

Very has been intimately involved with
developing was in charge of production and

refinement packages.

He's the author of Almost two hundred
fifty papers he's a fellow of the vi

Physical Society a member of the National
Academy of the American Academy of Arts

and Sciences an extraordinary city but
he's also an extraordinarily nice of

course I always consider always learn
something from Barack so today.

Barry actually is moving
on from his traditional

work which is my heritage computations
I understand that Irene has got

is a born again experiment and he's
going to be talking to us about probably

very interested in today in
particular living so you look for

molecular specificity the recognition
making this you hoping to hear.

Thanks for thank you Jeff.

I'm actually trying out a new talk on you.

So i Pod.

As for that and it's really something
I don't like when people do what I'm

doing anyway which is to give you
two trucks and they are related and

I'll try to make the relationships but
they're related by the title.

Number one and by the subtitle

I'm interested in these days is how
subtle differences between closely

related proteins lead to very large
differences at a cellular level.

How so.

That's the general question we're looking
at families of proteins that are closely

related and very different things and

these are two areas that
I've been working on and

in each case they were started by
a faculty member in my department

who told me that there were some problems
that were interesting to them and

I say I will give a group meeting and see
if the people in my lab get interested and

in both both cases they did and in both
cases they worked out to the point where

as Jeff told you I now have my own
experimental program in these areas.

All in collaboration I know
what to do with my own lab but

I know how to hire people who
work in other people's lab so

that's what is meant by
an experimental program.

The basic logic of what we do in both
cases and this is where computation

comes in as well number one I'm
a structural structural biologist so

we start with structures we do one kind
of computation on houses or another.

Try to understand particular properties

biophysical as primarily looking
at binding energy measurements and

then based on what we've learned
tried to effects our behavior.

So it's from structure computation
by physics to the cells system so

that's the review and I'm going to
start with discussion of hospital.

Basically if you look at developing
embryos and the developing

embryo is partitioned into
different segments and

the outcome of cells in a particular
segment is determined by

combinations of what are called
eight hocks protein.

So which harks protein

is expressed in a particular location will
determine the out come of that location.

You can take the wrong protein
from the say the story of A and

express it here and then you get
how all these months to fry is that

geneticists like to create with legs
growing out of eyes and things like that.

So the basic idea is the outcome.

Is determined by these eight proteins and
the question is how do they work and

how are they different.

Now their transcription
factors they bind to D.N.A. So

my entire story here is going to be
about how specific protein D.N.A.

interactions discriminate
among these curtains and

as you say we think we've arrived at a new
recognition mechanism between proteins and

so we hope at least that what we found
is a quite general applicability.

So let me get right into the problem from
a structural standpoint this is a piece of

D.N.A. and how it's proteins bind to
D.N.A. This is one of the eight X.

and as well as its primary
recognition mechanism is the helix

inserted into major group D.N.A.
Now the way these proteins work

specifically is to work
together with a second team in

this case called the exterior
cofactor which also has a home and

you can see that there's a scratch X.
extends from one here from U.B.S.

and binding to the second honey or
the main and there's a wipe.

Into this property so
this is a number of crystal structures and

what's going to be relevant is that the
connection between the two proteins has

not been seen in any crystal structures as
you'll see in a second the connection is

where specificity lies and here.

This is the paradox of proteins
if you look at the D.N.A.

binding regions all the D.N.A.
residues are the same.

So this is a paradox how can we be
different determine location in

a developing embryo when there are these
other recognition elements are the same

right down to the SO in the end
terminal of the brain in the words

between the structures so
far and I can just tell you

that there is evidence that there
has been evidence that specificity.

So again specificity is where
the crystallographer is don't see anything

that's very strange by the way please
feel free to interrupt me any time

during the talk now it turns out that the
source of the problem has been people who

are crystallizing these molecules these
proteins in the wrong D.N.A. and people.

Find complexes that usually take them
from D.N.A. sequences that bind without.

And there are still acts
experiments that are used and

it turns out that most proteins are not
going to go through the details.

That are well known there in
databases exactly etc but

the actual specificity side by
biologists even though they

sometimes are not
identical to the consensus.

And for those of you that are interested
in transcription factors this is

a fundamental problem or
databases of consensus sites and

those databases hide information
about biological specificity.

So I'm going to show you how we
arrived at address this problem and

this work is done in collaboration
with Richard man who's a geneticist

since we now have a paper
coming out together and

sell and I claim I'm a geneticist when
he tells me I have to be able to tell

the difference between a male and
female fly before I qualify.

And I haven't I haven't
really done that yet.

Anyway.

So this is a dream a dream but

there is a transcription factor that
controls the expression of gene called and

he adds a report or
Gene next to the transcription factor and

transcription factor site binds
one of the proteins called C R.

And you can see that S C R is
expressed only in the segment.

Now and
this is the sequence of the binding site.

Now there is another site which we call
consensus which binds another of a number

of rocks proteins and
you can see that the region

attached to this site
is different segments.

So this you are binds to
a few other proteins and

as you can see they do they differ in only
two nucleotides So the difference between

these two nucleotides somehow affects
this behavior in a developing embryo.

So to make a very long long
short third collaborator and

we'll argue all solve the structure of
SE are bound to the specific sorry.

The consensus which maybe is hidden
from some of you and the specific site.

So you've already seen
the structure is just like this.

The difference here is
that it's particular.

Down to specific and
nonspecific D.N.A. and

what we see here is for the first time.

Their structure in this connecting region
between the two proteins here again

in the consensus there is no structure so
I'll just say right now we think of

the blue pill thing is to pull the strands
across the minute group so that

it's position going a certain way in your
D.N.A. So that's the role of the second.

But we want to understand specificity
in that region and if you look in

that region now we're looking at
a minor group and for those of you.

The don't know most recognition is
takes place in a major group and

that's because there
are more hydrogen bonding.

Because But the reason the more hydrogen
down in patterns available in the major

group in the mine it doesn't
really get her that much.

It's recognition
the difference between things

is in a manner growing in particular and
in origin in that I'll be focusing on.

Now.

If we look at binding a theory is
of thing are one of the proteins

we're dealing with a consensus and
for care about the same

an animal but now if we look at immune.

So we're going to remove these two
amino acids and see how it has

a protein now binds to the side the effect
is very very small from eleven or twelve

to eighteen and that's because there's
no structure seen of the consensus.

So you would expect a mutation to make
any difference to the specificity sorry.

When you.

Remove these two residues I'm not going
to really tell if any girls from nine

to fifty five gets significantly
weaker is that important in vivo.

Well if you look at an embryo C.R.

A couple can express it's everywhere
you see lots of regions light up.

It's not only in the skin region.

But when you make the mutant.

This is lost again so that you only see

expression patterns so
that those two amino acids totally affect

the convivial activity of the binding
site that control element.

Whereas for the consensus.

They make no difference.

Again because they don't
bind anyway to D.N.A.

So the major point here
is that a factor of six.

Is able to affect
the outcome of the control

controlling function of this protein.

Now how does this work and

now I'm going to become a structural
biologist to go back to electrostatics.

This is a surface of D.N.A.
This is the my new groove and

here is an origin in that
proteins are where the males have

here is the same origin in with these
two other other amino acids are three

in his minus twelve we now there are three
is more important than the history and

I'm going to be focusing on so this is the
structure of the D.N.A. in the consensus.

This is the structure of the D.N.A. in the
forehead side something some difference

between less and
this is the source of specificity.

And the difference that we first
saw in the crystal structure is at

a minor groove is more narrow here and
here you can see in the picture.

It's more now and that's a.

We could see that was
different between the two.

So the question is Why
is that important and

I guess fortunately Jeff
said I used to my whole

life in the ninety's focused on
electrostatic properties of molecules.

This is a slide from a paper
we published actually N.D.S.

the way this is an electrostatic
potential country or

diagram around ideal D.N.A. D.N.A.

So what this means the level of D.N.A. is
negatively charged so there's going to be

a negative potential attracting
positively charged you know acids.

And I remind you that we're
dealing with an origin.

So that's going to be the source
of what I'm talking about.

So what this means is that if you
brought in a positive charge D.N.A.

the attraction would be get greater and
greater as you get close to the D.N.A.

and as a consequence of classical
electrostatics which I won't be

talking to you can see that as
you get close to the D.N.A..

The potential is very much.

Follow the shape of the surface of
the molecule was just a consequence of

basically electrostatics So

somehow the shape of the protein what
we know how reflected in the D.N.A.

is reflected in the pattern
of electrostatic potentials.

So the thought we had was that
the attraction of the groove

to positive charges would also be
dependent on its shape and here.

What you see in this is a fork.

The blue curve is a plot.

And you see in the cycles
two minima the red

curve is a plot of
electrostatic potential and

you see that the potential really
follows the with very nicely in

the consensus is only a single minimum and
again this is where.

Corrected in the electrostatic potential.

And in the one in each
minimum in the consensus

there is only a single origin in
there isn't enough negative potential

in the other are so what we're
arguing is that the shape of the minor

group is actually controlling its
ability to recognize positive charges.

Now the shape is controlled
by the sequence and

there is if you look at
the sequences are both eighteen rich

regions are known to be the big difference
is that in one case there's eight

in the middle of the sequence and
one in the right around the right

steps as they're called are known to open
up minor groups the classic example is

some of you will know of
the box eighty a minor group.

So what we thought was happening is
that the sequence is designed to have

two now regions in the sequences designed
to have one a post-doc in my lab.

So I won't go through it has developed a
Monte Carlo technique where you basically

sample lots of D.N.A. confirmations and
is able to predict structure

based on sequence and
the blue again or the X. ray and

hear the green or the results of his
simulations and you can see that

the regions that are seen to be narrow in
the crystal structure are predicted to

be now and not going through this
spec some people will be interested.

But the bottom line for

us is that the source of specificity
that we believe we've identified simply

not hydrogen bonding there's
no specific hydrogen bombs.

It's just the within the group so
be D.N.A. isn't just ideal B.N.A.

there is there's freedom there more
generally what we're saying and where.

Other proteins that I won't have time to
talk about that a common feature is that

the helix sets common to all members of
the family recognize the major groove

the same way and subtle variations in
a minor groove in this link or region.

How they recognize D.N.A. start beginning
start breaking up the family members

in groups and we certainly don't have
a complete answer I've been looking

telling you about S C R which
has our Q Are there and

there are others that have R Q R So
this isn't all this is the whole story.

But what I want to focus on is that there
are other protein is there a consensus

sequence that have so I've told you that

the results means Argentines
fitting into the minor group but

other proteins are the same reason
they don't fit into the market.

So that's what's wrong.

When we don't know we don't yet
have a structure of N R G R proteins but

we have them for many other proteins and

in the case of other proteins where there
is an R G R on the minute the glycine fits

into the groove in the origin
in splay out in the case of C.R.

You can see fit right in this group
of mean spaces out of south so

what we assumed is that if
difference actually in specificity

is not only in the arguments but in how
the origins are presented to the group.

So we would like to do is starting
to do is change the guy not a D.N.A.

contact residue and see if we can affect
behavior we that hasn't been done yet but

we do have to grind you Taishan and
of the boss is a fairly a strong.

As the origin in your.

So you can change you get the same effect.

So the presentation of
the charges is crucial and

that's how we think
recognition is taking place.

I'll just point out that if
we look at other holy domains

look at many other D.N.A.
binding proteins the pattern of

regions of the miners group
is very widely observed.

So we think this is hopeless in
the recognition mechanism and

that's sort of the story that I
wanted to tell you I think I've been

giving you all the messages that I need
to summarize think you are interesting.

I think the choice of a loop is there

to make sure that the protein
doesn't buy into the wrong sequence.

So there's going to be an end tropic
cost of fitting into the group only when

the entropy is used to reduce
affinities only when faced with

the right in a well binding place.

And another message that I alluded to
at the beginning I repeat again is that

specificity requires look at a specific
sequence a lot of what people know

about transcription factor binding sites
today is based on consensus sequence.

So there's a lot to be learned about
D.N.A. control of specificity which

obviously is very important and
we think that working on specific

sequences in sites rather than in vitro
sites is going to be very important.

OK I'm actually through here and so

it's yes we have some
answers some questions here.

They do.

Exactly and that's where we're going
to be playing the obvious claims of.

Measuring infinities different
protein different sites.

But yes you can see that
there is a variation.

That's conserved right up to man of
the site the site is better conserved.

You know the proteins in thirty nine
in vertebrates and the conservation

of the eight to thirty nine go in
groups is remarkable across species.

Yes No it's MT D.N.A.

when we calculate I didn't show
the results the binding affinity.

Of a single origin which is just what
we've done so far to the two sites.

Yes there's a repulsion because of
the salvation of facts for example.

It's about an order of magnitude.

It still is an order of magnitude
greater to the natural part

of the group than the white part of
the group but we put all that and

I just was trying to give you know
ultimately the driving force is

the potential so
it's going to be more attractive.

It's more negative.

It's a very good question and.

There's a long way to go
from a binding affinity or

from a crystal structure to the function
of a pretty what we found so

far is a correlation between affinity and
expression patterns and my guess is that

when the story is complete there going
to be lots of Connecticut facts that.

Yes

yes I can give you examples
nuclear very interesting.

They have their own loop.

I mean they bind.

So they have a tire see that
binds back to themselves.

I got this from Peter right.

So the loop extends from
the protein fits back and

when encountering the right
sequences again these are G R R X R

the same motif is widely used and
almost always associated with

the

same side because the side is
on the back of the major and

look more carefully at what this is
telling your so everybody knows everybody.

I didn't know a year ago but everybody
knows who we are domains by regions.

People didn't know is that the location
of a T.V. in the region is crucial.

So it's that kind of thing that I think
people have to go back and look at.

There's a lot of people have been looking
at chromatin structure on how D.N.A.

binds to his terms or a lot of these
effects they're very specific in fact

there's been a very large
focused field looking at D.N.A.

but the connection hasn't always been
made to biology so I went back and

read that literature and it was just
amazing how much of the stuff we could

understand based on crystal structures
of nucleotides So a lot of this.

Information is known and
I think it's coming out again Steve.

I'll never get it anyway.

So I've been trying so
I don't know the answer.

I've been trying to get people
to crystallise I know I have

some good friends who told me they didn't
get promoted till they went away from

studying the D.N.A. they now have to study
proteins and they don't want to go back.

So that's why I hired a crystallographer
to look at these things.

So the connection is through
simulation remote simulations

do a good job in reproducing the results
of crystal structures of isolated D.N.A.

So right now our only connection
to crystal structures is through

the simulations that seem to work.

So it's a suggestive argument but

until we solve some of
these structures we believe

it could be just an inclination
to look that way but

we know that you know better
than I tell you steps.

There's a connection between OK So

if you've had a break from questions
I'm going to the second part

of my talk as I said connection is

the issue of a single protein
doing very different things.

When you look in detail and there are
other connections as well that I'll get to

so here I was introducing you to proteins
known as here and these are seller

he's in molecules and I don't really
need all the information on the slide.

If you imagine the structure
of the membrane.

Proteins that extend from the membrane
they have five extracellular domains.

And when they really

signal through the membrane to
contain in this case and other

signal molecules that ultimately activate
active but we're going to be interested in

cell cell specificity how that
relates to molecular specificity.

So how does how does the binding of
molecules relate to the binding of cells.

This is a picture from the.

Graph from the textbook the molecular
biology of the cell and

this is the kind of phenomenon we'd like
to be able to explain these are real

cells in a developing ectoderm of
an embryo and they express a cat here and

called here in and I'll tell you there
are six what are called Don't worry

about what that means an E.P.
and I'm going to be talking about

the theory is neural Ronal and

what you can see here is that what is
called the neural tube if I had a really

dark you could see that this is up here
through breaks off from the ectoderm and

the only difference between these cells
and these cells is that the apathy or

cells express reportorial and
the neural cells express here.

So this important logical event
is determined by the here

and you can really
should turn them off for

some of this you can do experiments
which mimic this behavior someone.

So you have our cells express are green.

The first red or green and
they don't normally bind and

then if you express here and you transfer
the cells would entail here in this case

the green and red are both in cat here and
you see the cell stick together.

But if you have red cells expressing
in green cells expressing and

you see that they separate
into two aggregates and

this behavior has been interpreted for
many years to mean.

So that's why you get this pattern but

this is been very controversial and
as you'll see these

pictures were taken in cells that
respond very quickly and it turns out

that the results can be very different but
I'm sort of setting up the problem.

This was what we knew when
we got to the area and

the assumption was that these
properties were reflected.

At the molecular level now if we look
at how Cat here is bind to each other.

This is an electron tomography figure
from David Stokes his lab and well you.

And you can see these come off of the two
membranes and what they did is simply

build the cat here and this is a crystal
structure of one of the care here and.

You see five.

You see they contact only in
this membrane you see one and

you see in the electron tomography they
also make contacts in that region.

There seem to be some contacts
down here on the same cell.

That may actually be be important.

So we know from this there are from
structures but we know from lots

of mutation data that specificity is
only determined in the one region.

If you wipe out one of these
residents in the interface.

You don't.

And here we also know that
the structures of the sort.

You're seeing.

Here are not how cells begin
this is part of a larger picture

of what's called the junction of death
in the soma junction between cells and

ultimately these molecules form these
very tight junctions that ultimately lead

to solid tissue but when they start
recognizing each other they're dilute

their single molecule events when they
start binding they come together.

So let's look at the interface.

Remember the question I'm getting at in
this part of the seminar is how is and.

How do these how do molecular differences
need to sell your differences so

we have structures and a key element of
the interface between the structures is

the exchange of a beta strand
between these two domains.

You see they bind in this funny way with
the with the swapping of the beta strand

in this is a trip to sand that inserts

from one domain into the other
this is the one if you

change this trip to thank Allen cells
don't bind molecules don't bind etc.

Now the problem of understanding
the difference between any here and

begins with a slide and

there is more information than you
need to know except that the right.

Green are residues that
are in the interface between

between any of our structure
it's any in diamonds.

We don't have a structure of any
again we assume that wouldn't bind.

So if we look at either you look
at the interface the crucial

point is everything is
almost the same everything

in the interval very small differences
in interface between any care here.

And so we couldn't really tell from
sequence what might be going on.

So I had a graduate
student a few years ago.

I ask you to go through the following
exercise here is the structure of one of

the domains of this structure the N I

asked him to build a model and
tell me what was wrong with it.

That is right on these binds do
the best job you can building a model.

Tell me what's wrong with it and
then we'll know why cells work and

you get your Ph D. and etc and
after two years of every simulation

known the bottom line was there's
nothing wrong with we couldn't find and

he couldn't I couldn't find anything wrong
with it and even though the fitting is

where the limits of our understanding is
that very sort of there are differences

between these interfaces that were so
small that we just actually wrote a paper

saying that to one another with
each other so that was the problem.

We just couldn't find anything wrong and

then we were faced with a more general
problem of how do these similar molecules

still lead to this very distinct
behavior with a separate

So there was another issue
in all this which is

that I'll just summarize here this
is a stereo view it's an all slide.

I bet nobody under sixty
in the audience can see it.

But maybe some of my colleagues can
anyway this is an interface between

two clear here and.

And the point I want to make is
that it's a very well formed

interface lots of hydrophobic
surface area of hydrogen bonds but

the number had been around that
the interface the binding Ophelia's about

one hundred microbial Robbie giving
you better numbers that's very weak.

So one issue is rising so weak

another there's a biological reason for
it to be weak because a cell.

Exploring moving around.

They don't want to get stuck so

they want to be able to sort of find the
right partner and maybe stick them but.

So that's a kinetic Yes you actually but
that it's always been assuming that these

affinities would be low for that reason
I think there may be another reason.

So it's weak and the reason it's
really going through this quickly but

I think conceptually it's very simple
is if you assume here is a monomer

with a triple fan tucked into its
own structure and here's a dime or

you can see the yellow strand goes
into another structure and vice versa.

And here's a question we don't really
need to look at the basic point is for

this trip to fan to make new contacts here

to a first approximation at least
it has to break the same contacts.

I miss are in the equation and
the start of the whatever equation.

So all contacts here
if it's truly swapping

all contacts here are identical to all
contacts here and without the G. in zero.

Except for the monitor.

So it can't be completely true but
this is the idea.

You can derive a swap interface
still have a large specific but

still be weak because only the right
contacts would break the monomer context.

So we student of mine shall opposing.

Simply did a survey of all structures that
were around structures that are known.

Well that we had both affinity and
structures for.

And you can see the binding affinities
this is plotted against the surface area

of the interface binding a fairly is
range over many many orders of magnitude.

But all the green triangles
are swapped structures and

this is so the basic principle of using

the lower Filreis but still having a large
interface that we ideally need a lot.

Two things that specifically said
you don't buy into the wrong thing.

But if you want to really slow
this is the way to do it.

Entropy is another way.

So this is also relationship to hucks
proteins is find a competitive mechanism

for binding and only the right.

Structures will form.

I think that there's some kind
of semi principle there anyway

I want to go through this slide is
is a cartoon that really displays

I think we're summarizes the logic of
how we're thinking about how this works.

I think there are some interesting
general principles here imagine two cells

with a bunch of in this case here.

And so on the surface and I want to say
they simply asked the question what's

the binding affinity between two cells so
I could write an equation which

says that the binding affinity of free
energy a binding B. of two cells is equal

to the number of diverse times a binding
free energy of individual molecules.

So what you get is an amplification
mechanism that is even if there are so

binding upon these eventually
you get strong effects if they

are multiplied by multiple interactions
on the cell surface because now.

Now nature.

If it's going to work this way
runs into a problem which is if if

you have let's say the cells the molecules
are different than J You don't want them

to stick to each other if too
many of these diamonds form.

Then the wrong cells
will stick to each other.

So what you need is a situation
where you have enough interactions

to amplify individual finding
interactions but not too many or else.

Everything gets stuck and that gets
gives rise to questions of whether

the concentrations of molecules and
cell surfaces.

I don't think because of time I'll
go through that I just want to sort

of appreciate I think the main point here
is that very small difference is that

a molecular level.

The amplified and
that I think is the way specificity is

included on the scale here and because
they're very very similar to one another.

But how similar.

So here's where we go

to recent experimental work
none of it has been published.

We've looked at two domain constructs.

So there's a second the main coming out of
each of these where I'm only looking at

the first one and
now we run into some real surprises.

The first is that the binding
affinity of hearing

is one hundred fifty my Comodo Now
that wasn't a surprise.

That's what that was
intended to support it.

But when we look at here and
the Finlay has twenty microblog So

these proteins that we can tell
the difference between them.

In terms of cocoa is that these are very
similar but the feeling is different.

By a factor of seven eight seven a half.

This came as a major surprise.

Then these were measured with
an analytical interest of a centrifuge Now

we wanted to measure the hetero
feel like a fairly easy to end.

But we couldn't do that in
an analytical and herself centrifuge so

we used a technique known as
surface plasma resonance of course

through the device and again I'm just
going to tell you quickly how this works.

You have a chip when you label the chip
with one kind of care here and.

And you know another care here and
over measure a signal.

So you're basically measuring an affinity
by flowing one molecule in this case

over and over and over and over and

it's tricky because these molecules dimer
eyes on the chip and it's messy stuff.

But if you look at the data.

Here is a chip with an carried here and
on the chair and

you flow over the church and
you see the height of this curve

is the strength of the signal strength
the binding when you flow over the church.

It's much weaker.

Well that's not surprising and environs to

better than EBA surprises down here.

Here's a chip with the red curve
in which we know is weak but

when we buy into any stronger.

So the head of interaction is stronger
than the hunger feel like interaction.

This is the strongest the these arrows
mean affinities free energy and

in a strongest any is the middle is
the weakest how then to sell separately

with these molecular affinities remember
our expectation at the beginning

and end doesn't guarantee this is
totally different than what we expected.

And again I think I'll just
explain this to you in pictures.

These are equations that come from

statistical mechanical
theory of fluid mixtures.

Having said that it's really
easy to understand if you have

the environment and
they don't stick to each other and

you have cells sort of mixing up
then what you get are two aggregates

if you have binding to any stronger

but they also stick to each other then
the ideal situation would be this and

the basic logic is the following and
sticks to N.

so far and forms an aggregate
is left with nothing else.

But the stick the other is
if they buy into each other.

You'll get envelopment depending
on the numbers and so on.

You can also get a situation where N.

binds to be binding is
stronger than the binding.

And then you can get ordered
aggregates like this.

So the basic qualitative bottom bottom
line is by controlling affinities

you can control shape.

You can control how cells aggregate and
what kinds of shapes they form.

So we went back to the cell
aggregation assaye is seen

pictures earlier of the binding these
are different types of mass A So

these are the same molecules and

this is a control which I'll
show you a minute here and

six be the dozen by each of them but
now it's easier to see.

Appear if you look at an end very
carefully under the right conditions.

You see they form aggregates where
the green separates from the red but

they stick to each other.

You see some of that here as well but
when you express them with care here and

six B. that none of them are one stick
to you get separate aggregates so

there are lots of figures like this
that I'm not sure this is the best one.

But the basic bottom line and looking at
a lot of them is you actually do see next

aggregates that have separation which is
actually what we have in our insides.

We have we have touch other organs.

But still prefer to prefer
to stick to each other.

So where are we going with this.

I think I'm the sort of go to the.

We want to turn into an entity and we look

at these interfaces we still don't have
an answer because that was the question I.

You know we have made me swear.

I told you about we can actually
make that make it a stronger

when we try to make actually
make it stronger too so

there's a lot we don't understand yet
but this is where our program is going

to understand what's happening
at the molecular level

and what I haven't told you about
is expressing these proteins in

developing embryos and
creating monsters just like the fly people

work I've spoken about collaborators or
Richard.

These are people in his lab people
in my lab who did the work along

here and work is done in very close
collaboration with Larry Shapiro

a crystallographer in our department
is a statistical mechanician.

Stephen Price and
the work in my lab was done

by Peter Chen thank you for
your attention.

And why you are asking the question and
yes we did the president of the third

main has no effect on binding it so
the two domain

affinities in the three demand affinities
are going to go with an experimental or.

Are you just asking me.

That's my philosophical response and
my my my my my

my my base my scientific
response is the saying which is

to get on with understand this process
I know we have to look at kinetics

we just haven't gotten there yet.

I could have spent a lot more
time talking about the sort.

And that's all kinetics the serving is
all kinetics if you spin and that's why

there's a big debate some people say
the other other guys spin too fast and

so we had two sorts of ways
one are hanging drop acid.

We're not moving at all and
the other are forces and

they do change the total picture but

not the bottom line of what sticks to what
you get very different patterns depending

on on the spinning and
that's clearly kinetics.

So we have an internal seminar series and

I spoke to people about people who work in
various cancers about these results and

they said that

even the need to invade tissue may
depend on and he's of strength and.

So I found out a few weeks or.

Well that is a great funding source.

So I can't answer your question except
I just become aware of how important it

might be but I don't have anything more
to say other than consciousness raising

the entanglement depends on the sheer
forces there is you can separate them.

If you're sure persons
are large enough but yes.

So before they bind to each other
the hearings are diffuse on the surface.

They don't have a direction they're
just you can see from straining

that you just everywhere.

When they make a contact.

Then you want to create tissue with
specific interaction right cells binding

to each other so at least the way I think
about the problem is that recognition

takes place with just a few care here and
once they stick their other cat here and

diffuse into the region and
ultimately create tissues and

create junctions that
activate signaling pathways.

But you need to start with
a pure recognition of that.

That's at least a hypothesis
we're working under

we certainly know that regionally there
diffuse and eventually they concentrate so

that that process occurs and
for sure how it occurs.

So it's a controversial part
of the field the word intern

meaning so there are papers that claim
that they start this way and then

I don't think they ever occurs there's
no evidence for in my opinion but

what happens is they start
forming this way and

then they start making contacts on
the same cell through other domains.

So what we think is that they're weaker
interactions between other domains and

that get a phase transition eventually
driven by weaker interactions but not into

these would be trans on
the same server would be yes

yes yes exactly.

Exactly exactly.