Original subtitles

Hello. My name is Randy Schekman.

I'm from the Department of Molecular and Cell Biology

at the University of California, Berkeley.

In parts one and two of my presentation

I described two approaches

for the study of protein secretion and vesicle traffic in eukaryotic cells.

In the first approach, I described the use of genetics

to develop an understanding of the contour

of the secretory pathway.

The genes and protein molecules that are necessary

to move membranes, membrane proteins, secretory proteins

among the stations of the secretory pathway:

the endoplasmic reticulum, the Golgi apparatus,

and mature secretory vesicles.

In the second part of my presentation, I described

in a rather more focused effort

an understanding at a biochemical level of how

membrane proteins are packaged into transport vesicles

through the intervention of a cytoplasmic coat protein complex.

A kind of a skeletal structure that sculpts

membranes and sorts molecules

into a bud and then cleaves that bud

by a process of fission from the ER membrane.

And I indicated that that process is likely recapitulated

at several stations in the secretory pathway elsewhere in the cell.

Now in my last part,

I am going to talk about more recent experiments

that describe two fortunately fairly rare human diseases

that affect the very machinery that we uncovered

in our genetics and biochemical work.

Of course, as I illustrated in the first part of my presentation,

when you introduce a mutation into an essential amino acid residue

of a protein required for cell growth and division, the cell dies.

Some of these mutations produce a protein

that is thermally unstable

and thus if it's a simple microorganism like yeast,

the cell may survive at room temperature,

but die when the cell is warmed to body temperature, or 37 degrees.

Now such mutations in essential amino acids of essential proteins in humans

almost certainly would not survive.

They would arrest development at a very early stage.

However, of course one can have more subtle mutations

in otherwise essential proteins that produce

a partially functional molecule that survives even at body temperature.

In the case of the machinery,

the coat machinery that I have described in part two of my presentation,

another solution to this problem comes from the fact

that humans have multiple copies of the genes that

encode the Sec proteins that comprise the COPII coat.

I mentioned the set of actors at the end of part 2 of my presentation that comprise

the coat. They are a small GTP binding protein called Sar1.

Yeast cells have a single essential Sar1, but mammals have 2.

And you'll see in a moment that this gives rise to a

very interesting and fortunately rare disease

when one of the two genes is mutated.

The coat itself, consisting of two layers, Sec23/24 and Sec13/31,

these subunits are also duplicated.

Humans have two copies of Sec23 whereas yeast have only one.

And humans have four copies of Sec24,

so there is an opportunity for some genetic mischief

in these genes by introducing mutations into one of the copies,

some cells may be affected more than others,

and that is precisely what I will illustrate with two diseases in the next few minutes.

The first disease that directly affects the COPII machinery

was described by Carol Shoulders and her colleagues in London

in a collaborative effort that engaged clinicians

from around the world, and it was published several years ago

in Nature Genetics. And I'll just illustrate what she discovered

with the abstract from her very interesting paper.

What she discovered was the genetic basis of a disease that affects

the production of a giant lipoprotein particle called a chylomicron

Now when you absorb fat in your diet, it is taken in the digestive system and

in the lower intestine it is absorbed in cells

that line the intestine that are called enterocytes.

These cells swallow up the fat, internalize it into the cell

and then assemble these fat molecules with a protein molecule called ApoB.

And the form of this lipoprotein

is very different than the lipoproteins that you are more familiar with, LDL and HDL.

It is instead a giant lipoprotein called a chylomicron.

These chylomicrons assemble in the lumen of the ER,

and they must pass through the bottleneck of the secretory pathway.

by being packaged into a transport vesicle

and delivered downstream...ultimately secreted eventually into the bloodstream.

Chylomicrons once in the blood stream are absorbed by other tissues

for example the liver, and the fat is then taken away

from the chylomicron and reformulated

into the lipoproteins that you have heard about, LDL and HDL.

Now until the advent of this paper, the mechanism of packaging of these

very large lipoproteins into transport vesicles was rather obscure.

Indeed, the structures of these chylomicrons are much larger

than a COPII vesicle, so it was not easy to understand

how they may pass along in the secretory pathway

in these enterocytes until Carol Shoulders discovered

the genetic and molecular basis of a disease called Anderson's disease

or chylomicron retention disease.

And in a small number of families with this genetic lesion,

chylomicrons are produced when the patient mistakenly takes fat into his or her diet,

but the chylomicrons that are produced remain sequestered,

hidden away in the lumen of the ER,

failing to be packaged and transported elsewhere down the secretory pathway.

Shoulders and her colleagues identified the locus defined by Anderson's disease

and discovered that it is one of two copies of the human Sar1 gene.

We'll call it the second copy. This second copy has a number

of different mutations in it, depending on the family of origin.

And these mutations, many of which are likely, very seriously affect the activity

of that Sar1 molecule, producing a molecule that may have very little if any function.

And yet these patients obviously can secrete molecules

They grow to adulthood, and as along as they control their fat intake

in their diet, they can control the productions of chylomicrons and survive quite normally.

Which means that the other Sar1, the one that is not affected

by this disease, must be doing the lion's share

of the work to engage the traffic of all other secretory molecules.

So a very interesting investigation, one that has yet to really be understood why

two Sar1s would have such different function.

I am going to spend the rest of this part talking about a disease

that we have had a direct hand in helping to understand

and that is another fortunately very rare disease

that was uncovered by Simeon Boyd and his colleagues at

John Hopkins University and a collaborator in Saudi Arabia

who uncovered the genetic basis of a rare form of a craniofacial disorder

that, as you'll see, affects the secretion of a lot of different molecules,

fortunately only in a subset of cells.

So this disease, called CLSD, cranio-lenticulo-sutural dysplasia,

produces a apparently subtle defect, at least as you see

in these two homozygous patients. These patients grow

to adulthood, but they have a facial malformation.

Their eyes are farther apart. They develop cataracts.

There are bone deformities, and most interestingly as you see in this slide,

there are issues with the closure of the fontanelles at the

top of the head. The soft spot on the head,

which normally closes very early after birth,

and in these patients it takes up to two years to close.

So there is some lesion in bone morphogenesis

that leads to a failure in the normal fusion of the bone

to make in intact scalp.

So these patient then were studied. Samples prepared.

The locus responsible for this craniofacial disorder was cloned

by Boyd of Johns Hopkins, and one day he called me all excitedly,

telling me that the lesion was in an absolutely invariant amino acid residue

found in Sec23, the molecule that is the integral subunit

of the coat that I described in the last part of this presentation.

An absolutely invariant residue found in yeast and in humans,

and yet in these...in patients with this disease,

this amino acid was changed from a

phenylalanine to a leucine, which is a fairly subtle mutation.

Nonetheless, it is in an absolutely conserved residue.

which is highlighted in two forms, both from the sequence of the protein

molecule in the region of this mutation.

in yeast, and in the two human copies of the sec23 gene.

This is the phenylalanine residue, shown here

as absolutely conserved and in all sec23s that are sequenced and available

in the genome database, there is always a phenylalanine here.

And imposed in a structure

of the atomic resolution structure of a complex

of the core of the COPII coat.

A structure determined by a wonderful young crystallographer

by the name of Jonathan Goldberg and his students

at Sloane Kettering in New York.

So to orient you, we are looking at the top of the coat

as though one was looking down onto the surface of a membrane

onto which the Sar1 molecule and the Sec23/24

subunits of the coat had just assembled.

So we are looking from the cytoplasm down to the membrane.

Here is the Sar1 molecule, the GTPase that begins this process,

and it's seen here touching the Sec23 molecule shown here in mustard color

directly adjacent to a green copy of the Sec24 molecule.

Now you'll see in later images that this top side surface of the coat

is the surface that comes into contact with the scaffold complex

that you saw in the last part of my presentation

that forms the outer layer of the coat and allows the membrane to bend

to form a vesicle bud.

Note that the phenylalanine substitution

with a leucine at amino acid 382 of the sec23a gene

in humans likely changes a surface feature on the top side of the Sec23 molecule.

This phenylalanine is not at the very surface of the protein molecule.

It is somewhat hydrophobic, so it is buried

under a loop just beneath the surface

and the placement of a leucine residue

may make this region slightly misshapen, perhaps more flexible.

And clearly with some pathologic consequence

that affects these patients in craniofacial development.

Now our collaborator in Saudi Arabia prepared fibroblasts,

skin cells, from patients, homozygous patients, where this mutation F382L

is present in both copies of the sec23A gene.

And Simeon Boyd, again in collaboration with our favorite morphologist, Lelio Orci,

had a look by thin section electron microscopy

at cells either normal, or mutant, to see if one could see

a Sec mutant phenotype such as I described

in the first part of my lecture that affects the traffic in yeast.

Indeed, as you'll see in these images, a surprisingly profound effect

on traffic is evident in primary fibroblasts

cultured from homozygous patients with this disease.

To orient you to this image, the top panel shows a section of a control,

normal fibroblast. Here we see endoplasmic reticulum tubules,

as you've seen before, studded with these dark

staining ribosomes that are engaged in the production

of secretory molecules, one of which, in fibroblasts,

is the secreted molecule collagen.

In a heterozygote, a parent of one of the children

that we saw in an earlier slide,

there is a slight distortion of the ER tubule, some distension of this, but

this is otherwise in the normal frame of what one sees,

within the range of phenotypes that one sees.

And the heterozygote, the parent of the patient, has no obvious pathology.

In contrast, fibroblasts taken from a homozygous patient

show a dramatic distention of the ER tubule.

A huge increase in the volume of the lumen of the ER,

in some ways more dramatic than we saw

in the original Sec23 mutations in yeast.

Very clearly defective, and yet surprising because

these fibroblasts grow and divide in the laboratory, so clearly

they are capable of enough secretion to permit cells to grow,

and it's surprising that such a defect would be consistent

with a rather subtle pathology characteristic of CLSD.

Now to look at some of the protein molecules that are handled

by fibroblasts, either wild type or CLSD homozygous fibroblasts,

Orci and Boyd collaborated to evaluate the production

and intracellular distribution of two different proteins

that mark different stations in the pathway.

On the very left, highlighted in red, is a molecule I'll just refer to as PDI.

It's a protein that is involved in protein folding in the lumen of the ER.

It's a resident of the ER, it just works there; it never leaves the ER.

In a wild type fibroblast, a single spread out cell,

one sees PDI defining a reticular network, the endoplasmic reticulum.

In the homozygous mutant, this same reticulum

is now blown up to create vacuolated structures

that are much larger, rounder, such as you saw

in the electron micrograph in the previous slide.

In parallel samples, a different staining technique using an antibody directed

against collagen with a green fluorescent dye was applied,

again in a wild type cell a reticular network is apparent,

representing collagen molecules that have just been made

and are being assembled in the endoplasmic reticulum

and are on their way out to the cell surface.

And likewise, in contrast, the same collagen molecules accumulating in the

homozygous CLSD Sec23 mutant,

this vacuolated appearance. So these proteins accumulate

apparently in the very same structures that harbor the ER resident protein, PDI,

because one can superimpose these two images of the same cells

to show that the two fluorescent dyes overlap; thus

collagen is accumulated in this cell, and it is accumulating in the ER

just as you can tell from the control red fluorescent dye.

Now let's have a look once again at the structure of the core of the coat,

the Sec23/24 molecule, because I am going to borrow yet another

very important observation from Jonathan Goldberg,

the fellow who crystallized this coat structure,

to show you in some detail what we now know about why this mutation

in the CLSD patient has its effect. So you have seen this image

before of the core of the coat looking down from the cytoplasm

onto the surface of the membrane

with the region that's affected by the leucine substitution at residue 382,

but I superimposed onto this diagram a backbone structure

of a portion of the scaffold in this case, the Sec31 molecule

as shown by Goldberg to register right over the affected region of Sec23.

So quite surprisingly, and very dramatically, from his purely structural analysis

we have a very clear impression that the F382L mutation

may affect directly the interaction between the core of the coat and the scaffold

that must assemble to cluster coat molecules together.

This is apparent also from an end on view,

so now we are looking at the membrane along the side

looking at the edge of the coat, and you can see

the top side of the Sec23 with the F382L mutation

and the backbone of the scaffold subunit, the Sec31,

resting directly on top, perhaps in contact with the affected region.

Chris Fromme, a postdoctoral fellow in my lab,

who has done the biochemical investigation of this

has shown very directly that this mutation affects the function of the COPII coat

using a vesicle budding reaction such as I described

in part two of my presentation.

And also showed very directly that this mutation blocks,

or at least retards the interaction of the Sec31 scaffold

with the Sec23/24 coat subunits.

Now on additional exploration Lelio Orci observed something

that we hadn't anticipated. A rather surprising additional feature

of the defect that is produced by this

mutation in fibroblasts cultured from homozygous patients.

I showed you a moment ago an image of the ER membrane

blown up with a very large lumenal content,

but if you look more closely at this ER membrane,

you see an additional feature that we hadn't anticipated.

And that is, at the region of the ER membrane

that would normally give rise to COPII vesicles

the so-called smooth transitional face, note the rest of the membrane

has ribosomes, these are engaged in producing secretory molecules,

but one portion of the membrane is smooth, ribosome free.

This portion of the membrane normally is the platform for the assembly of COPII vesicles.

And in the homozygous CLSD patients

instead of COPII vesicles, we see long smooth tubules,

as though the budding event is arrested, and in its place

long membrane tubules form that for some reason

can't consummate the process of fission and then continue

to project tubules, which in other images Orci has shown

project many microns away into the cytoplasm.

So what's going on here?

Why would a defect in the completion of the binding of the scaffold

to the inner core of the coat

produce this aberrant tubulation? Now in order to explain this aberrant behavior,

I want to take a brief digression into another subject,

a subject that we have investigated with the yeast proteins,

to explore how the Sar1 molecule initiates the formation

of a bud on the surface of a chemically defined phospholipid membrane.

To illustrate these experiments, conducted by Marcus Lee,

I want to very briefly tell you about another pioneering

study conducted by Brian Peters and Harvey McMahon

in Cambridge, England.

Peters and McMahon were studying the effect of a set of proteins,

including one called amphiphysin,

responsible for the first step in the formation of a bud

at the surface of the plasma membrane,

a bud that is responsible for endocytosis

particularly for the endocytic internalization of membrane proteins,

including membrane receptor molecules.

Peters and McMahon discovered that amphiphysin and other molecules

like it, wedge shaped molecules, have on their surface

a structure that they refer to as a BAR domain

and this BAR domain interacts very tightly

with the charged surface of the membrane

to change its shape, to likely change the structure

of the bilayer to allow it to initiate the formation of a sharply curved bud.

One way to show this is to mix protein molecules like amphiphysin with

chemically synthetic phospholipid vesicles

and to explore what happens to these membranes when amphiphysin binds.

Here's an example, two examples, taken from some of these experiments

in which the molecules called endophilin

and amphiphysin, each of which has an apolar structure

associated with one helix of the molecule exposed

on the concave face. When these two separate molecules

are mixed with pure phospholipid vesicles or with

liposomes, the starting liposome gives rise to

very long, rapidly, very long tubules that project many microns from the starting membrane

It's as though the simple binding of endophilin or amphiphysin

to the membrane destabilizes the bilayer locally

and allows this sharply curved tubule to project.

Now we looked on the concave surface of the Sec23/24 molecule.

Knowing that this part of the surface is responsible in part for sculpting the bud,

we looked for a BAR domain and were unable to find such as structure.

Indeed, if you mix Sec23/24 with liposome membranes in an experiment like this,

it doesn't even bind by itself.

However, another molecule, the one that initiates COPII vesicle budding, Sar1,

has an amphipathic N-terminal helix that may serve a similar role

to these amphipathic helices in BAR domains

implicated in membrane curvature in endocytosis.

And to illustrate this point, a simple cartoon should suffice.

Sar1, when it touches the nucleotide exchange catalyst, Sec12,

as I indicated in the last part of this presentation,

becomes activated by binding GTP

and then it extends an amphipathic N-terminal helix that likely embeds directly

into the exposed monolayer of a bilayer.

And one idea is that this embedment

may cause the relative surface area of the two monolayers to change.

This is a variation of an idea that was published some 35 years ago

by John Singer and Michael Sheetz,

called the bilayer couple hypothesis.

And we would posit that once Sar1 activates,

it may bind to the surface of the membrane,

cause this change in the surface area of the exposed monolayer,

and thus cause the membrane to buckle, to change its shape,

and possibly to initiate the formation of a budded vesicle.

Indeed, when pure Sar1 molecules are activated with GTP in the test tube

and simply mixed with synthetic phospholipid membranes,

these membranes develop long tubules just as was shown by Peters and McMahon.

These long tubules project from a donor liposome membrane

and can go for many microns in length.

So this may then well be the initial stage in vesicle formation

Indeed, I would like to leave you with an interesting suggestion

that it is this feature of Sar1 in the patient with CLSD that

may result in the production of the thin tubules budding in a futile effort,

budding from the transitional smooth face of the ER.

Thus explaining why this may be seen in fibroblasts from these patients.

Now let me summarize this part of my presentation

and then all three of my presentations,

in the form of two model slides.

The first model slide is simply a recapitulation

of the pathway that I've described for the assembly of the COPII coat,

as it applies to all eukaryotic organisms.

Once again, Sar1 in the cytoplasm is in its inactive state

bound to GDP. It becomes activated by touching

the active nucleotide exchange catalyst, Sec12.

It then extends an amphipathic helix causing Sar1

to embed part way into the bilayer.

This act alone may be sufficient to cause the membrane to begin to bend,

to at least initiate the formation of a bud,

possibly a tubule, but this process would be capped ordinarily

by binding the core of the COPII coat,

to Sar1, and then through interaction and sampling of different proteins

to capture a non-covalent complex cargo molecules that are destined for transport.

And eventually the process would finally be consummated by the

acquisition of the outer layer of the coat,

the scaffold complex that clusters the cargo molecules

and the inner layer of the coat into a bud,

which is then separated by pinching from the donor membrane.

Now this model applies to the endoplasmic reticulum,

but does it apply to other stations in the secretory pathway?

As I have illustrated, there are many such events in the eukaryotic cell.

How general is what we have learned as it applies to other situations?

Let me conclude then by showing you that we believe that it likely does apply

because there are several other coats, and likely more coats to be found,

that explain many different stations in the secretory pathway.

The process that I have spent more time discussing involves the assembly

of the COPII coat and its role in traffic from the ER to the Golgi apparatus.

But an entirely complimentary biochemical investigation

conducted by Jim Rothman's lab has shown that COPI and its cognate GTP binding protein,

called Arf, is responsible for pinching a vesicle and delivering it from

this station back to the ER, and likely at other stations of traffic within the Golgi apparatus.

Likewise, at the other end of the pathway, another kind of coat complex that I

haven't had time to describe, called clathrin,

uses a GTP binding protein, likely Arf,

to form a coat on the surface of the trans-Golgi and deliver cargo molecules

to and from another membrane station called the endosome.

One area of fruitful investigation that a number of students

in my lab are now conducting

concerns other...the activity of other possible coats

involved in traffic of the majority of plasma membrane proteins.

from the trans-Golgi to the cell surface.

We recently described a new complex called exomer

responsible for the recruitment of a set, a small set of membrane proteins,

and it is my conviction, or at least suspicion,

that there will be many other such coats that remain to be discovered.

And so for those of you who find this topic interesting,

I would welcome your engagement in further explorations

on the molecular basis of selective protein traffic

in the secretory pathway. Thank you.