So the cell.

This is a.

Oya of my building at Scripps and
it's a painting by David Goodsell and

it represents a small part of a macrophage

engulfing bacteria and if I were to show
you the entire macrophage you would feel

most of this building to the point
here is that circles are these

incredibly complex factories
of working moving parts.

Nothing happens in a cell by osmosis and
by a fusion.

Everything has to be moved has to be
transported has to be traveled around and

made by machines and it's this molecular
machinery of the cell the things that go

on and turn the idea of D.N.A.
into a working engine and

working factory that we're interested
in with electron microscopy So

it's not really a matter of size
an electron microscopy it's a matter

of resolution and the electron microscope
can study those cellular like.

Structures and can tell you where these
various machines are in the cell and

to do that we use the techniques
called electron cryo tomography and

I'm not going to talk about that
today although we do do there.

And I'm sure in there will
be doing that as well.

Ican talk about all of it.

So I'm going to instead focus on part of
electron microscopy that studies these

these molecular machines the machines
that move things around that copy things

that transport things they do all the work
in this enormous effect of the cell.

So what is this more of this disastrous
transmission electron microscopy

Macroscope that can go away
from cellular to to atomic

there's a very old run the sort of sits
in the front foyer of our building.

It's an entire week from the one nine
hundred forty S. one of the very first

commercially electron microscopes that was
ever made and the one that sits in one of

the Nazi controlled rooms is
something along these lines and

very similar to what India has and
in basic design it hasn't changed

much from the one nine hundred forty S.
to now but it has done what Fineman.

All sport has gotten many many
many more times powerful.

So the design you really like from Mike so
it's very similar to that of a lot

microscope in a lot match script you have
a source of photons electron microscope

you have a sort of electrons that
mask up a series of glass lenses

electron microscope a series of electron
optical lenses and you have a specimen

somewhere in a lot more so if you drive
the photons through the specimen to some

magnification and look at the magnified
image same thing electron microscope

in a transmission electron microscope
we transmit the electrons through

the specimen magnified with some
lenses and look at the result.

So what distinguishes an electron from
a lot microscope is simply resolution

the wavelength of light limits you to
about point two microns the wavelength of

an electron of course is very very
small much smaller then angstroms

the limits of an electron microscope
is about two angstroms for

reasons I'll explain shortly.

So it a lot microscope level X.
in bundles hold bangles and bundles of

actin look like just read files in the
Latin are said at the electron microscope.

We can look at every molecule along at
Acton and see what's contacting it and

how it works as a machine and
the same here.

This is an example for

microtubules in the lot microscope level
they just verify the bundles of things

in an electron microscope we can
look at every molecule along it.

So here's the transmission it come out of
the average run that you can get that in

English led that's in my lab is capable
of a resolution of two angstroms.

Well to angstroms pretty darn good.

I mean why be in contrary just
look at the thing like Phantom and

see it and do everything in the world
the two extremes resolution I'd be

very happy with that you
could sequence D.N.A.

with a two Angstrom electron microscope so
what's wrong with a picture.

Why we can't do everything
is that biological

specimens came to be hydrated and
exquisitely sensitive to

radiation damage if we want to
look at them with an electron.

Microscope.

We have to first make them fit in.

So we can get the electron
beam through them they have

to withstand a very high vacuum in
the middle of the electron microscope and

an incredibly high level of radiation and
when I was learning electron microscopy

My professor said you know your spacetime
is about here compared to nuclear bombs.

When you stick it into
their electron microscope.

So we have to find some way of preserving
our very delicate biological tissue.

So it can withstand these circumstances
and there's lots of ways of doing that

most of you will be familiar with
things in your biological textbooks.

Well maybe you're all too young for that.

But we use sliced and diced things up and
you put them in plastic and

Ukiah chop them up and on the race bed and
I'm not going to talk about that because

that doesn't go up to
very high resolution.

So the technique we use is called cryo
electron microscopy and I'll explain why.

So the substrate the SUV the.

Support thing is a grid we use for
our spacemen is shown here it's about two

millimeter confirmation and on top of it
we sprayed a carbon film with holes in it.

So if we look at this is a hole in so on
that copper carbon form on the copper mesh

we put a drop of liquid and then we blot
it within it down to a very thin film and

then we smash that into a really good
cryogenic very fast so fast that the heat

is taken out of the space when faster in
Africa will the will Congress to last.

So the water molecules are just
locked down in place and so

we're looking at a fully hydrated specimen
when we have it in this now solid state

it's not it is it doesn't crystallize
crystallization would be a disaster.

It would smash the proteins apart.

So the trick is to get it to a sin film
and then smash it down fast enough to

freeze it rapidly enough to form
art we call vitreous or glossy ice.

So when we now have our little
protein particle a molecule machine

imbedded in its hydrated
solution we can drive a beam of.

Electrons through it and we can look at
it as if it was in its native state and

that's the real uniqueness of
cryo electron microscopy So

what does a protein look
like an electron microscope.

This is a tiny little copper group about
three minute meter across with your

specimen on you for smash it into
liquid a thing solidify it and

now you stick it in your electron
microscope and this is what it looks like.

So we zoom in on this.

So we pick up particular square there
this is a new copper mesh and we zoom up.

Now we see that this.

This is the cause of the carbon
film with the holes in and

if we go a little bit higher.

We pick on one particular hole.

Now we're looking just at our
protein in it's a quiz buffer that's

in a frozen state and we can keep zooming
and now we see little speckles here.

And if we look at one of them and
take a picture at higher magnification.

This is what we get and this is in fact
bacteria phages P. twenty two the D.N.A.

packaging engine that in the referred to
and I'll talk about a little bit more.

Well that's all very well but
this is terribly disappointing and

this is a two Angstrom instrument has
about one hundred nanometer that doesn't

look terribly impressive for
two Angstrom resolution and

the problem is radiation
damage if we take a picture

at higher magnification of an area of
the grid we get something like that.

These are Hittite is B. virus particles we

take a picture with very few electrons
we take just a quick snapshot of it and

we get something like that but
if I want better signal and

I expose it for longer use more electrons
all I do is boil my protein away.

I totally destroy it because of
Kong withstand the radiation so

cry is this constant
battle between signal and

noise and we can't use enough of
the electrons to get a good enough signal.

So we have very noisy images if we tried
to use more we would cook the space and

we'd lose anyway.

So the big problem in cryo is really
battling that signal to noise problem.

The second.

I'm is a mathematical one.

It's a transmission electron microscope so

when we collect images we're collecting
projections of a three D. structure.

So all we have are the projection
images and what we want is the three

dimensional structure and I always show
my students this picture to persuade them

never interpret a projection image as
a three dimensional object because you

can get it wrong very easily if you
looked at any one of these images.

You'd be pretty sure you knew what
the object looked like and you'd be wrong.

Every town.
So

if you want to know what a three
dimensional object looks like from

projections you need more than one.

And of course we all intrinsically
know this we have stereo vision.

The reason we can sense dates is because
we have two different views of the same

thing and if you want to know distance.

What you in training you know
instinctively do is move your head from

side to side and gives you an idea of
distance of course this is all exclusively

well worked out in mathematics.

In Tamara Toma graphic techniques.

So the Masons we will use the in to

get a book away from a terrible
signal to noise problem and

to get to the three dimensional images we
need from our two dimensional picture.

Projections are made as a known typically
as electron crystallography angers

an expert on net where we used
to the sheets of crystal and

sheets of proteins or a method which is
known as single particle reconstruction

methods which is a terrible name because
in fact you need millions of particles.

It shouldn't be called single
particle probably that you could

possibly give it but nevertheless
that's what that's what we use and

these are all structures that have been
solved in my lab over the last few years.

So the challenge in single night particle
microscopy is how do we go from this.

This is an image of grow the L..

Single image taken at the doses there
are enough not to destroy the protein and

this is what we can produce
from images like that.

OK this is a three dimensional
reconstruction of growing able to

raise Aleutian of about five or
six angstroms how do we know it's five or

six angstroms because on that
also he knows he's we can stay.

To see the handedness
of the elephant here.

So we know you know we're
at about that resolution.

So the trick is going from this to that.

So of course what we use
is the power of averaging.

So this is a picture of a very well known
person everybody will recognize them

at a signal to noise of about the same
rate as we see an electron microscope.

OK We have one copy of it.

But if we were ed to add ten
copies together or fifty copies or

one hundred copies two hundred four
hundred six hundred and a thousand.

That's what we get out of the noise OK and
this makes sense that

we add signal together the signals
are always the same the noise or

sometimes positive sometimes negative and

enough of them together the noise will
cancel out but the signal will always add.

So that's what we have to do but remember
that we also have to get a three D.

structure out of these two D. projections
and for that we use mathematics that's

very similar to what you use in a CAT scan
where you want a picture of the inside

of your head you get put into an X. ray
machine you take lots of PIC projection

pictures from there is Directions can
mathematically put those together

to see a cross-section of your head in
electron microscopy it's pretty similar

except instead of having run particle and
rotating it and looking at it.

We have many particles that are identical
but happen to fall in different

orientations on the grid and we can use
them mathematically combine them so

in this case we'll find early ones and
this is our intention.

Well average them together until we get
good enough signal in their few or early

ones in a different orientation average
those up until we get a good enough signal

and they're few and then all we have to
figure out is what their relative are.

And that we have to figure out is one of
the big challenges of the field when you

don't know anything about a particle and
it might be floppy and it might be small.

That's a real hard problem but plenty of
people are working on it including us.

So how many do you need in this average

is dependent on the resolution will
be dependent on how many you use.

So we take Here's a typical P.T.B. map.

If if with an electron
microscope we image that and

we averaged a thousand of them together.

What we would see is maybe some domain
structure we know where the major domains

were and we'd be at a resolution of
about thirty angstroms three nanometers

if we took ten thousand of them and

averaged them together at now we'd start
to see some details of the domain.

We take a hundred thousand of them.

We'll start to see some of the alpha
Hyllus is coming out in resolving it and

if we took a million of them would be able
to actually trace the carbon back burn.

So this is sort of theoretically
what people have figured out and

in practice this is what we experience all
of the structures in that have been solved

through these kinds of resolution
have required at least a million

subunits average together and in some
cases ten million subunits everything

together because we can't actually go
quite way physics will allow these other

problems that was trying to
solve at the same time so

you need a million of these things and
you.

Starting with pictures like this
with a few of them scattered around.

OK this is a problem right for automation
and when I first learned electron

microscopy as a grad student at University
of Chicago and I learned how to do this

the first thing I flew to of is a computer
would be better at doing this than I am

because it's less impatient and it's
going to do the same thing every time.

So what we do in single particle
cross we take our protein in solution

we put it in the ether and
we put the grid in the scope and

then it's a matter of hunting down areas
of the grid with nice thin ice with well

distributed particles and taking pictures
like this all of them hundreds and

hundreds perhaps thousands of pictures and
then once you have the pictures you have

to pick out these individual particles one
by one to a whole bunch of alignment and

average into them and
get this three D. reconstruction.

It's incredibly repetitive process and
very tedious for people to do

not at the ten thousand particle level
perhaps but if we want a million.

Nickols it's impossible to do
it except with automation and

so that is what our
national resource that run

scripts is the mission of their
resource is to urge to mate.

To develop a supply of technology aimed
towards completely automating this process

from the moment of making samples but
a way to getting a three D.

map and all goes well not just to make the
process easier although we would like to

do that but we want to increase three for
throughput because we want those million

particles to get to high resolution
we need this level of automation.

So we want to optimize resolution
one to get the best resolution for

that this specimen is capable of allowing

We want to expand the possibilities out to
all kinds of people who want to do this.

This shouldn't be a graduate student
laughter more post-doc laughter I'm

involved in getting a three D. structure.

If you have a good protein to
start with and you start the X.

ray crystallography as you know as soon
as they have a three D. Crystal and

after the synchrotron and twenty four
hours later they have this structure and

that's what it should be like and then we
run into the open technology up to a wider

audience and in fact to mention
we worked a lot on the human

papilloma virus that's the basis of the
Gardasil vaccine and doing these kinds of

very highly controlled automated processes
are very important to pharmaceutical

companies they don't want you know a whole
graduate student grinding away for

five years in order just to get
a simple answer to a simple question.

So that's what we're about and the way
we've gone about this is very much.

What the X.

ray crystallography is did about a few
decades ago is to develop a power plant.

Basically you know you
start with a specimen.

And you end with your three D.

map and there's a whole bunch of steps
along that process and there's back all

makes in the process in these leaks in
the process you lose data all the time so

many ways where plumbers you know we
would use in the bottlenecks we stopping

the leaks and we're trying to make
this process as smooth as possible and

to streamline it and
increase the throughput at all steps.

So what they're.

So that's what we doing for five years and
to some extent we have been successful so

what the pipeline at the National Resource
looks like sample comes in

it gets taken off to be vitrified or
we have the vitrification robots and

some other equipment to make that
straightforward takes about half an hour

or so to make a specimen once it's made
it gets put into the electron microscope

scripts aligned and then automated
software starts to acquire images so

you're sitting in front of a computer
maybe tweaking a few knobs but

most of the time so I'm watching paint dry
you're watching your images come in one

by one maybe a thousand of them and
we have a very nice control room couch so

people can sleep there overnight and
keep an eye on things.

And then while their data is collected
being collected we already starting to

process that we're picking those
particles using automated routines.

We're doing the alignments we're starting
grinding through that whole process

of taking a million sub units and
averaging them and

align them using of course
a lot of high in computing and

we're at the stage where we have
a well behaved specimen block and

I cross it across the road virus
in this case hip that is be virus.

We can take it from this point to this
point in under twenty four hours.

So the system is is streamlined it.

It does have a reasonable throughput so

we can go to something
that's seven then a meter.

So we can start to see
over here is it cetera.

In about a twenty four hour period given

an African Putin power along the way
because this is also a bottleneck.

So.

So read my letters to a large extent is
develop the computer vision algorithms

the computational algorithms some of
the devices some of the technologies and

mechanisms to do this I would rather than
tell you about that because I'm sure you

would rather hear about some of the
biological problems that we solved using

this pipeline rather than
the technology that went into it.

So I'm going to very briefly cover full
case studies I'll do these two together.

Of how we use the power plant
to solve structure in the.

I'm going to talk a little bit about
lambda sage about the P. twenty two D.N.A.

packaging machine about a device machine
that transports proteins around in

the cell and then finally something that
I'm particularly very interested in at

the moment is looking at transients states
of proteins and in this case we used.

We looked at the assembly of the thirty S.
Robison and.

Very briefly cover all of this.

So the first two case
studies are these D.N.A.

packaging and delivery machines and
this was done by a grad student in my lab

in collaboration with Alex a village
showed castings Peter privilege and

Jack Johnson A lot of who are very
interested not only in bacteria phage but

also in the evolutionary principles behind
how phage is a changing and how they work

as D.N.A. packaging engines so viruses
the RIAA bacteria phages interesting.

They are interesting because they infect
bacteria and this is an arms race.

They are tamed something like tame
to the thirty one bacteria phage on

this planet more biomass than
anything else on the planet.

And without them we would be swarming
with bacteria and probably wouldn't have

survived so there's an exquisite sort of
balance between the bacteria phages and

the bacteria that they invade and they
constantly evolving different mechanisms

the one to protect themselves and
the other to infect and here's a bunch of

bacteria phages all attacking a bacteria
all trying to get their D.N.A.

from its packet its head into
the bacteria the bacteria

have evolved all kinds of mechanisms to
protect themselves against this one and

in this case as a gram negative bacteria.

It has this huge polysaccharide layer
around it as a protective layer.

But this twenty to fade
can eat through that

part of second ride lair using specialized
machinery it then injects its D.N.A.

into the bacteria an hour later this
has become an entire factory producing.

P. twenty two which when the pressure gets
big enough to burst open the bacteria and

each one of these can then go off and
do the same thing.

So it's an entire engine for
simply making new fade

which are basically a head full of D.N.A.
So it's an interesting system.

So what these bacteria phages do inside
a bacteria is they have some D.N.A.

and they use their D.N.A. to manufacture
some proteins which simply self assemble

first into an immaturity
haid with as a porter or

here whose job it is to take the D.N.A.
that's been copied and

pump it into this head at the same time
as it pumps to the head the head and

it goes some giant conformational changes
when the head is full and the bacteria

phages knows how knows when it is full
somehow a bunch of other changes happen

in birds an enormous machine which
is the infection machinery and

it's done and it's ready to go off and
in fact in fact a new bacteria.

This is an amazing machine.

OK this is a self assembling
machine that can replicate itself.

So make an ism of action of all
of this is quite fascinating.

So gay people study various
aspects of this and

one of the aspects he's studied is how how
do you change from this immaturity head

this Prokop said into a mature capsule
which is then stuffed with D.N.A.

and has to withstand enormous pressure
the is to mate and pressure inside of this

head when it's full of D.N.A.
something like six atmospheres.

There's way more pressure than
what's inside a champagne bottle.

So this shell of protein has to
withstand that kind of pressure and

not let the D.N.A. come bursting out so
we start with the sickest cell.

This is but this is a quite a thick
cell here about fifty angstroms

by the Thomas mature it's only twenty
angstroms So during this process this

thing gets larger and finesse and
somehow it has to stabilize itself.

So gave has done cry and
reconstructions of both the prohibited.

And the mature capsule and
this is what they look like.

And because we have the resolution that we
can start to dock in crystal structures

you could dock in a similar crystal
structure not exactly the same one and

then study the way this.

Conformational change happens and if you.

This is sort of invented you know we don't
know that this is what's happening but

from going from the pro head to
the mature head this sort of small and

folding happens so by unfolding that
those couple of little loops this

fictional goes into a very thin
shell manages to expand itself and

then in order to make it safe against
pressure capping proteins come on and

stabilize some of the joints these
weaknesses at the threefold cemetaries

on this prohibits are these
capping proteins that come in and

interact with the this device to
strengthen it against the higher pressures

and that's a whole long story and
it has evolutionary implications.

It's a trick and
I don't have time to go into that.

So that's one interesting thing about the
D.N.A. how it withstands of pressure but

of course the most interesting thing
about it is the sort of machinery

that does the D.N.A. packaging and
then does the infection because they have

the working part of the machine
the rest is just holding it together.

It's just the.

The shell.

So we're really interested
in that machinery.

Now in this case we need
a lot more particles

This is where high throughput and
automation really comes into play because

when we can look at the outer shell that
outer shell is sixty faults symmetry.

So every time we pick up Brian of these
viruses we can average of sixty fold.

So we don't need too many before we have
ten thousand one hundred thousand five

units when we can multiply by sixty.

But if you want to look at the tail
machinery that has no symmetry this thing

now becomes an asymmetric object.

So if we went ten thousand of them.

We have to take ten thousand pictures and

this is about how many you gate in one
picture so gave using a little animation

we have had our hands collected about
five thousand images just like that.

And pulled twenty five thousand individual
particles off them Everetts them together

not assuming any symmetry and was now able
to reconstruct the structure of this tale.

Because it's a true three three
dimensional reconstruction we can

go in and examine the machinery
inside the bacteria phages and

what they were particularly
interested in was this device here.

The political which is where the D.N.A.
comes through on its way into the head and

one of the questions about bacteriophages
How does it know when it's for

D.N.A. These packaging machines don't
stop when they see a particular sequence

they'll package any D.N.A. given piece of
D.N.A. It all just pump it into their head

until they sense that
the head has a NAFTA D.N.A.

in reach Hypatia and then they stop.

So one of the questions is how do
they know when to stop if it's not

sequenced related.

And based on these results.

What they have prophesies is that when the
D.N.A. is packaged Taki into the cell it

wraps a tight loop around this
mechanical device and that acts as

a prisoner saying to him switch that bin
tells the rest of the packaging machine.

OK there's too much pressure in here
you have to stop pumping D.N.A.

and then the whole cascade of
interactions happens after that.

So based on these kinds of studies they
hypothesized this precious angel sensing

shit switch and so they were all kinds
of other features about this tail so

this is the political that is responsible
for the pressure sensing and for

pumping D.N.A. in and then the rest
of this machinery is responsible for

the infections stage of the D.N.A.
So we're interested in there too.

So this is what it looks like in
that asymmetric reconstruction

of the entire bacteria phage and
it has all kinds of gene products

something like tramps and
five types of proteins

fifty one sub units all together come
together to make up this large machine.

So what they did is take indeed
took the tails out of the phage.

Sorry.

Some of.
These

structures are have been solved with X.
ray crystallography and the X.

ray crystal structures doc into these cry
instructions with very high fidelity which

gives us a lot of confidence that
the cry and structures are right.

But we wanted higher resolution So what
gave took was individual tail so the tails

can be blown right out of the phage and
be intact and here in our individual.

Tails just on the phone and
then he could take what

three hundred fifty thousand particles
now because they're much smaller so

we can collect many more of them and
we can get to a resolution

of about nine angstroms at nine
Angstrom resolution we can start to see

alpha heal a cs we can start to
understand the packing arrangements.

Even when we don't have high resolution X.
ray crystal structures.

But still it's still five different
protein sub units and we don't know how to

separate them from each other so
then he got individual protocols.

He took this structure alone.

We took pictures of that took fifty
thousand particles in this case

every just those all together and
then got a structure of that on its own so

by sort of teasing this machine apart
into its component pieces and doing high

resolution reconstructions often we can
then balls up quite a high resolution

structure of the machinery and then start
to understand exactly how it works and

what the mechanism of action of each of
these parts are and what the the role

is that they play during inspection and of
course there's lots and lots of other work

to do the reason you want high throughput
in automation is this is not the end

of the story now they want to mutate every
one of these components and do it again.

They want to find out.

Yes So.

You're

saying that this guy may not be
quite the same as that sibling.

If you make.

A thousand of these they
may not all be identical.

Exactly.

Lots of pictures of P. twenty two.

So a bacterium in effect is identical
identical copies of P. twenty two and we

have lots and lots of them on the grid and
we take lots and lots of pictures of them.

So it's simply P.
twenty two which is that they.

Well that's an excellent question.

So we have to treat them as identical in
order to average them and in a minute I'll

talk about a project where we can start to
think about what if they're not identical.

What if there's conformational
variability and then.

We'll know if they from
the same gene products and

they sell for same bowl well they
probably are again Tickle And

I mean to the extent that you know
these crystals structures docking

extremely well you know
they are identical.

Enough that the confirmation
of variability and

name is probably not relevant
in this particular case.

OK so there is this claim given
you more to talk about D.N.A.

packaging machines in the labs been
studying them quite a while but

then we go on to a more cell biological
problem and this is work done

by a post doc in my lab who's now an
assistant professor at Florida State and

he was looking at the structure
of cup two coats in cages.

What is cup two cup two is
the device the machine that

takes proteins manufactured in
the India players we particularly and

has to deliver them elsewhere in the cell
a third of your proteins are manufactured

in India plasma particular but they have
to be delivered to the goal to deliver to

the cell membrane delivered
all over the cell.

So they have to be gotten
from there somewhere else and

this device copped to is
what does that delivery.

It's really packages up.

Proteins and

transports them off is made up of
a bunch of different proteins saw one.

Twenty three twenty full and six thirteen
thirty one and a bunch of helper proteins.

But one of the interesting things
about this caging mechanism

you can see the cages if you
just look at images like this

from across me is they have to be all
kinds of different shapes and sizes.

There's all kinds of different
proteins made in the E.R.

some of them are very long.

And some of them are very small and this
one devise has to be able to package all

of them up and truck them out of there and
deliver them to the right place.

That's one of the interesting things
about it is how the biology has evolved

to make a cage that can expand
to meet the requirements.

So what Scott did is he looked
at one of the components of that

package of machinery six thirty and
thirty one on its own and

found that itself assembles into cages.

And then he looked at six thirty and
thirty one with its other components

a twenty three twenty four and found
that that formed cage like structures of

the largest size Now I know these are hard
to see remember that signal to noise but

we average enough of them together.

These are the structures we get.

So that's what the K. cage alone
looks like six hundred thirty one and

then if we add the other component in
it forms a cage on the outside with

a lair on the inside and I think of
some movies him if they'll start

with sorry and see if we can get stoned.

And so.

These are rather beautiful of
the devices rather elegant geometrical

design of these things and so
what it what we think is happening based

on the results of these reconstructions
is that this inner layer the sick twenty

three twenty four is sequestering
the proteins that it needs to package and

the science of those
proteins that sequesters is

causing it to interact with this cage and
make their cage larger and smaller.

And to understand that a little bit more.

Let's strip away that inner
layer of the larger cage and

this is the structure we get and

these are the smaller cages that self
assemble just from the single protein and

these are all they are both with
the same geometrical principle.

Well they all are four struts
coming together at a vertex so

we can use a very simple geometry fellow
struts come together at one vertex and

they firm the structure and

we can predict what kinds of structures
would be formed by geometry from.

Using that kind of some codes building
mechanism but we don't find all of them.

We only find certain types and
what we find is there

the angle between to between
two of these stress is

always Hans tent in the one direction and
then varies in the other direction.

And that's why we think that the.

The second protein the six twenty three
twenty four which senses the size of

the probably can change that angle and
therefore can build a bigger cage.

So there's a very simple adaptation of
the cell to use the same simple geometry.

But allow one flexible point in
order to build a larger cage and

recently the very recently
actually after we did this work.

The X. ray structure was solved for
the six thirty and thirty one and

by docking that into our cry E.M.'s
even though this map is a very

low resolution relatively it's only about
twenty three angstroms by docking it in we

can understand something about where
the interesting components are and

where they're flexible hinges solar cell
is doing then is leaving these components

alone and changing that angle and
there flattens this out and

makes a much bigger cage and
in fact if you could flatten it out more.

If you do if that angle can become
bigger you can build enormous long

troops using the same geometry and
that may be one way these very.

Very very long proteins get
transported out of the E.R.

and into the goal the operators although
we have not seen those tubes but

it's certainly one of the things
that Scott's will be working on.

OK One last quick example.

And this is something I'm
particularly interested in.

In our lead is we're interested in
looking at transients States and

looking at sort of functional
components so cry is great for

doing structure but
what we really want is function

structure on its own is kind of
boring it doesn't tell you anything.

So what we're really interested in is
function so that's why I was a cop to

Robin once you know how the cage changes
how it shapes because how does this thing.

How does this machine work
to do its work in the cell.

So one of the problems we started on
because it's a straightforward and

accessible problem is
the self-assembly of the ride

is a small subunit the thirty
is sub unit so this is a rabbit.

And it's made up of a small sub unit and a
large subunit these things surface sample

independently and they basically come
together when they're ready to transcribe

R.N.A. into a protein but
on its own the small sub unit

consists of a launch piece of R.N.A. and
twenty proteins twenty one proteins and

if you simply take them as
individual components and

you shake them up in a Chub
they will assemble into working

machinery into rabbits arms OK So that is
saying the process is really fascinating.

I mean this thing is completely
independent It needs no other machinery in

the cell in order to assemble itself.

All you have to do is make
the proteins they will assemble and

there are things that
can make more proteins.

So it's a sort of a fundamentally
profound way that biology

preserves itself OK taking D.N.A.
into proteins goes through this

device which is a self assembling a very
robust device so we're kind of interested

to know how the thing self assembled how
does it come together and there's been.

Tons and tons of biochemistry done on it.

In the lab of J.B. Jamie Williamson
at Scripps apart from anybody else.

So we decided to do is
just look at the thing.

So we took rabbits twenty one proteins in
the R.N.A. shook them up on a tube and

then we took snapshots at time
points along the same point so

every as fast as we could do it.

We took the step shots we
had eleven time points

starting you know at about ten seconds and
going on.

So in the only time points if we
pick out whatever these structures

are some kind of protein structure and
we pick them out and we do some averaging.

We get some idea that there's some kind of
structure forming but we don't see much.

But it by the time we
get to three minutes.

If we strip picks structures out and
averaging them them together.

Now this thing looks very much
like an assembled drivers or

with some changes in the graduate
student who worked on this call these

the sonograms of the baby rabbits.

You know the embryo.

Roberson It's a little bit what they look
like they're not you start to see is that

we've every two lots of these
together we've classified them and

said Tell me what the differences
are not just the every ten but but

tease apart the slight differences and
what you see is that this big body.

Here is Mehran is constant
through all of the averages.

But this region up here is
changing quite a lot and

this region here is changing
quite a lot and if we could.

Here's a here's another picture of
the same thing and I think I have a movie.

So here's four examples.

Here's a fully assembled.

Here's one with this hate
group hasn't assembled yet.

Here's one this little
bit is not quite there.

And here's one way this
little bit isn't quite there.

But there it is OK so we sort of seeing
snapshots of various states of protein

folding in this assembling device and
if we take the.

See if a movie will play really me.

These movies are a little slow.

If we take Earl of these various sort
of snapshots and we just play them.

You can see that this part of
the process is not very much.

Where is this is varying quite a lot these
two these two pieces we can win this

is highly speculative and would have
to be backed up by enormous amounts of

biochemistry of course which it is but now
you can start to assign what these various

varying regions are and assign him to
various proteins and now you can start to

say something about who folds first which
put faults and then which but folds next.

And what's coming and

what is coded Pailin versus
independent in this folding pathway.

So this is terribly speculative because
we don't know the time sequence here.

They're all on the grid at the same time
then we have to tease us apart still but

wrote you might imagine is happening
is this kind of assembly pathway.

So the rabbit.

They are in is folding and

they in helping sequester some
of these other folding bits in.

So this is an ongoing project another
graduate student in the lab has now picked

up a movie working closely
with a biochemist to do it but

what it illustrates is the thing
that I particularly am interested in

pursuing in our There are those this is
low resolution we're not out to angstroms

here we're not doing any of the fishing
but we're asking interesting questions

about an interesting process and what we
want to extend this to now Rob isms a nice

easy project what we'd really like to
start looking at is trans in protein

protein interactions particularly in
membrane proteins so membrane proteins for

the most part are a big you know
cascade of individual individual

protein protein action interactions that
are transient they don't last very long.

So we'd like to be able to use these
techniques to tease apart those individual

events and find out what is
coming together even if we can do

it at low resolution at least we'll be
talking about mechanism and action.

So.

I think that's all I have to say this
is my group scripts at the moment.

So we're supported by some highly
skilled professionals these

German andone are professional software
engineers who also can design devices and

build little mechanisms.

Joel is an incredibly
well qualified senior

research associate who is marvelous with
specimens he can do just about anything

with a specimen that needs to be done and
she's a senior research scientist and

she's the one who basically takes
the software that bees guys bolt and

make sure it's actually useful to human
beings you know not just a very elegant

piece of software and Moxon and a new
technician it helps the collaborative and

services preserves the project that coming
this is an incredibly talented group

of grad students did all the work on
the twenty two in faith caves about to

graduate soon pick ways
a new graduate student and

he's working on the risk
loading complex the machinery.

That's responsible for our in AI
delivery in the cell there takes the.

Micro R.N.A.'s and
turns that into an R E R N A.

I silencing engine anchors picked up the
Radisson project when we continue on there

and Dimitri is a brand new grad student
he's going to be working on components of

the nuclear pour complex another
fascinating membrane embedded machine and

Craig did the initial work on the Robson's
and really become an algorithm person he

lost to biology once again work at Google
and he runs to develop algorithms.

Neil is a post stuck in the lab and
then we have a systems person and

administrators and the group is
run by myself and Clint Potter.

I was going to in just
showing these movies but

let me just add Ron's here mentioned
we run a large training course

in Cry electron microscopy every other
year at the Scripps Research Institute and

it is a large course I think last
time we had a hundred people in.

All but it's a lot of fun
sometimes run for nine days and

it's a more basic course in this case we
ran it for seven days it was a more so

we did our advanced topics and
I think we'll probably alternate So

next year we'll probably go back to having
a basic course and do basic training.

So if anybody is interested in that
you can keep an eye on the website and

apply for it.

I'll just start this again and
let the movies play

any questions if there are any Thank
you very much for your attention.

Thank.

I.

Exactly.

That's what we think that these two
bits can be independent of each other.

First the bottom has to fold nobody seems
to do anything on so let's hold it and

then sometimes look at bottom for all of
this but is folded and sometimes that is

but it's folded and sometimes both but
you can get any combination of those so

what we think is the one has folded
successfully and the other is still

misfolded and hasn't quite I'm Dr but
is very speculative I mean we do have

to back this up was biochemistry but
the bike industry that has been done.

Does seem to support that.

So it's exciting it's fun.

So it's a really fun project one
that I'm particularly keen on

at the moment because it's new.

Right.

Well that's where a lot
more work could be done.

WE USED TO SCHOOL classifications you
know just like anybody else does and

we could certainly stand to have
better algorithms in that regard but

the ones that are they so it works with.

It's like this.

But what we think we're going to have to

spend a lot more time working on is better
castigation when we go smaller when

we're looking at something like a tiny
membrane protein in a transition state.

One would hope you know we could maybe
improve those classification techniques

the other thing that would probably
help a lot is labeling if you can

respect gold label on took you know
a couple of different points of this thing

that's going to be give you all
the signal you need to do the alignment.

So there's various ways.

One might get around that.

But it's a struggle signal
to noise is the better wall.

Of what we do and the other question.

OK

if you just think you know if you have
a copy of my hand and it's noisy and

now take another copy the hand will
be identical but here the noise in

here might be negative at the spot but
in this copy might be positive.

So if I add a positive or
negative together get zero or.

So we basically relying on the fact
that the noise is sometimes up and

sometimes down sometimes up
here sometimes on the end every

day between the mean of that
the average of there is zero.

But the signal is always the same
neighbor varies from image to image so

if we add enough of those they're
just going to add an add an ADD but

the noise is just going to
eventually average off his era.

Exactly.

So we have found means of them
spread out on the grid and

we just go pick them out
one by one individual.

And to some extent they have to be
identical really we have to be able to

classify them as different and deaths you
know really one of the challenges for

doing these kinds of transience States.

I personally will viruses
you know we can do those

I mean we can do them to seventy for
a resolution.

You know without too much weight.

So that's sort of a solved
problem getting them to too and

some resolution is a lot harder and
I think we have to control all kinds of.

Factors with the microscope and
the imaging to get that and

to get that to work.

Very.

Exactly.

So that's the other
thing they want to do so

I leave one out so that's one of
the way to test the sap offices.

If you think that's what's happening.

It's pull one out and
see what happens then.

But it's a lot of work.

We haven't.

Exactly.

So that's where the project is going now.

And what we really need to do is to get
it down cold you know said you can really

make a grid every ten seconds and
that's not as easy as it should be so

that's one of the things we're
working on to try to make that easier.

There's only just one.

I mean there are in a nice contrast and

I mean that's mostly what you're
looking at you know it's mostly R.N.A.

and the proteins are not not
as big as that social and

that seems to fold up in the head
region first be the driver but

I'm no R.N.A. the same body I mean Rob is
an assembly expert I think it's a great

project an interesting driver but
don't take my word for any of unseemly.

And we never have our thanks.

But that's a great question.

So all of thanks.

I mean sure every technique is out of X.

but we're pretty darn sure what we're
getting is real because you know things

have been sold with cryo they've been
sold an X. ray you know they agree.

There.

I'm.

Some structures are gotten
wrong just like they are an X.

ray but for the most part I think
those artifacts are not serious.

So you can make mistakes but but when you
do it right you do get the right answer.

Taking things out of the cell that is
an issue and I mean what you'd like to do

is do everything in the cell but
that's that's not always possible.

So the the cup to cage for instance
that is a very interesting question

is what we're seeing you know this
different sized cages It's a procedure

is that just an artifact of being in
the test tube as opposed to in the cell.

I don't believe so but you'd really
have to go back into the cell and

do something like tomography and find a
cage in situ to do that but that's because

really really hard to see how crowded that
cell environment is you're not going to

find anything very easily to look
at a few of them in the cell and

you know you're looking through
this enormous mass of protein and

activity says really a tough tough problem
except for maybe you can label things

that we can label things and
maybe you can go off to them.

Yeah.

Exactly.

I wish I don't remember
them making a mess of that.

Yeah so.

So two hundred is about you know where
things get really tough like the risk

loading complex the saw in a science and
thing that's two hundred and

it's going well in the lab I mean you
know he's getting a structure from it but

it's not easy and you know probably what
we want to do is make it easy to do those

two I mean it's all very well doing and
I cost a little virus but

they're much more interesting to
do something of that size easily.

So there's a few things coming along.

I mean it would be nice to have some
kind of contrast enhanced alack

cryo negative stain there's some
hints that maybe that would work but

I don't think it's been
worked out exactly.

Well yet labeling you know would
be great if that would work.

And then maybe on the instrumentation side
there is a new device come along called

a phased plate which will boost
the contrast of the images without

destroying the high resolution information
and Gates That's about a five years away.

Pessimistically ten years away
on commercial instruments but

that could make an enormous difference
to be able to look at little things as

well as big things but
at the moment there are two hundred is

maybe you can squeeze down
to one fifty one hundred.

Although you know we were looking at
G.P.C. our with our K. or something.

The other day which I think is one hundred
twenty and we could see it and we couldn't

tell much about it and we can look at
individual antibodies you can look at

a single antibody and you can see it has
a Y. shape again you know it's floppy so

you can't do much averaging so
it's hard to tell much more about it then.

Here's the and to join and
here's the antibody but

maybe that's enough sometimes or maybe
that's the only question you're asking.

OK Richard Quest thank you
very much very exciting.

Thank you thanks.