So welcome to the second lecture in the series
called Chemical Glycobiology.
My name is Professor Carolyn Bertozzi.
I am in the department of Chemistry at UC Berkeley,
with also an appointment in Molecular and Cell Biology Department
and I am also an investigator of the Howard Hughes Medical Institute.
I.. This is the second part of a two part series.
The first part focused on the background and history
about the field of glycobiology and the structures of sugars.
And this part, which is part two, is going to focus more
on some recent research from my own laboratory at Berkeley
with the goal towards imaging the glycome.
And in fact, what you are looking at here is
a fluorescence image of a zebrafish
that's five days old in which some of the sugars have been tagged
with fluorescent molecules and this allows us to literally visualize those sugars
in a live organism. And so let me tell you about
why we are interested in that.
Okay, now the field of molecular imaging is certainly large
and expanding rapidly, and it's very important
both from the perspective of clinical medicine
as well as in the research laboratory.
So many of you might have found yourself lying on a table like this
at one point in your lives, as have I.
This is a magnetic resonance imaging scanner, or a MRI scanner
and what this scanner can do
is literally take a picture of molecules
inside our bodies without having to cut us open.
Which is very convenient for many of us.
And there are devices very similar to this one which allow
clinicians to look at other molecules in our bodies,
molecules that are radiolabeled for example.
And that is what this image is here. This is called a PET scan.
of a human body that has been injected with a radiolabeled version of glucose.
Now these are important tools for looking at molecules in clinical settings.
but we also like to look at molecules in research settings
particularly to understand what molecules are doing inside
living cells and nowadays we have some wonderfully advanced
fluorescence microscopy tools
that allow us not just to see the collections of molecules in cells,
but individual fluorescent molecules as well.
So it's a very exciting time to be doing molecular imaging in living systems.
Now, I mentioned in my first lecture that one of the most
intriguing aspects of glycobiology is the fact that
the glycome or the complete collection of glycans that a cell makes
is dynamic. It's not fixed.
So, the glycome, or the collection of sugars on the surface of a cell
when it's in one particular state
is different from the collection of sugars on that same cell
when the cell has transformed its state.
This is certainly the case when embryonic stem cells
choose a differentiation fate
and then become a differentiated cell,
like a muscle cell, or a neuron. If you look at the collection of
the glycans on the cell surface glycolipids and glycoproteins,
you'll find that the nature of that collection has changed
as the cell has undergone a differentiation process.
From the perspective of clinical medicine,
it's interesting to note that the glycome changes when healthy cells
transform and become cancer cells.
So, you can understand if the sugars on the surface of the cell
are changing in a manner that is characteristic of a cancerous state,
it would be very convenient if we could actually see
those sugars change in vivo, in living human beings.
Because that would offer the possibility of imaging those changes
and detecting cancers when they are at hopefully very early stages.
So there are many reasons why one would want to image the glycome,
to take a look at how those sugars are changing inside living subjects.
So we can understand how do the glycans relate to differentiation of cells
or to organogenesis. And can we detect changes in the glycome
that correlate with disease states.
So that we can detect those diseases at very early stages.
The challenge, of course, is to figure out how to tag these sugar molecules with
probes that we can actually visualize using imaging technologies.
So even in the research laboratory, this is very challenging.
The most commonly used imaging technique in research laboratories
is fluorescence imaging, or optical imaging.
And we don't even have a way to put fluorescent tags
on these sugars. So that is a challenge
that students and postdoctoral fellows
in my laboratory have taken on.
And it's a project we started working on back in the late 1990's.
So it's been progressing now for over ten years.
Just to give you a little bit of counterpoint,
people who study proteins using molecular imaging techniques,
have a wonderful arsenal of tools available to them,
which is why images like the one I showed on the previous slide
are so common and so familiar.
to most cell biologists. In fact,
most cell biologists probably spend a good portion of their day
looking at images similar to this one.
In this cell two different proteins have been labeled
with two different fluorescent dyes,
one green and one red.
And the protein that appears green has been labeled using a genetically encoded reporter.
So let me just show you what that reporter is.
It is called the green fluorescent protein,
abbreviated GFP, and it's one of the most important tools for fluorescence imaging
of proteins. The green fluorescent protein, because it's a protein that is encoded by a gene,
and it turns out that one can fuse the gene that encodes GFP
to the gene that encodes the protein that you are interested in imaging.
And now, when the GFP protein fusion is expressed,
basically your protein of interest is labeled with a fluorescent molecule.
The GFP has a chromophore inside that is fluorescent.
And this is such an important tool. It's revolutionized our abilities
to study proteins, not just in live cells, but even in transgenic organisms.
which we express these GFP-fusions by manipulating the genome
in the embryonic stem cell stage.
This is so important that it was recognized with the Nobel Prize in Chemistry
in the year 2008, and that prize was granted to these three scientists.
Very well deserved. Very exciting development
in the field of molecular imaging.
Now those of us who study glycobiology, of course,
would love it if we had a tool that was comparable
to the green fluorescent protein.
So that we could label our sugars with a fluorescent tag,
the way that people label proteins with this fluorescent tag.
But unfortunately the biosynthetic machinery,
by which sugars are built, doesn't really lend itself
to these genetically encoded reporters.
That's because sugars, unlike proteins, are not primary gene products.
Each of these complex glycans is not encoded by a single gene.
Rather, they are the products of a complex series of
metabolic steps inside the cell.
They are products of metabolism.
And so for that reason, we cannot use genetically encoded reporters like GFP
to introduce labels onto sugars.
So I drew this cartoon as kind of some comedy, because we would love
to be able to put a tag like GFP on the sugars,
but there is really no mechanism to do that using genetics.
So, in my laboratory, we have been focusing
on using the metabolic pathways of the cell.
to introduce fluorescent probes into the sugars
that are not the GFP but rather small molecule fluorescent reporters.
And let me show you schematically how we go about doing this.
It's a two-step process, and we call this metabolic labeling of glycans
with chemical reporters.
The first step is to feed cells a synthetic
monosaccharide building block
Now this synthetic sugar looks very much like one of the natural
monosaccharide building blocks that you might find
in foods that you eat, and I introduced those structures in my first lecture.
But the difference is that we have altered the structure
of this monosaccharide building block to introduce
a chemical functional group "X"
and "X" is just the generic for now,
and in a minute we will get to what that is.
But "X" is what we call the chemical reporter.
It's a reactive functional group
so that when the cell takes this monosaccharide
building block, and when the enzymes process it and integrate it into the glycans,
when the glycan appears on the cell surface, that reactive functional group
is now on display and can be used in a second step
which is a chemical reaction with a probe molecule.
Now the probe is a fluorescent dye, for example.
And that probe has its own chemical functional group, let's call it "Y".
just as a generic label. But "X" and "Y" have to be designed
so they react with each other to form a bond.
And once that bond is formed
now there is a fluorescent probe molecule
bound to a cell surface glycan.
And that probe now allows us to visualize all of the glycans
that possess this red monosaccharide
building block as one of their constituents.
So, this is a very straightforward schematic
in that what we need to do is feed the cell a modified sugar
and then do a chemical reaction on that sugar
once it's been displayed on the cell surface with a probe.
But it turned out that there was a very significant chemical challenge.
embedded in this problem because we have to select
the chemical reporter "X" and a complementary functional group "Y",
so that these two functional groups react
only with each other and not with anything else inside this cell.
And let me tell you, there is a lot of functionality inside this cell.
All of your proteins and your lipids, your nucleic acids,
metabolites, there's water, you know, there are a lot of chemical entities inside this cell
and "X" and "Y" have to be designed to avoid any side reaction
with those biological groups and still react only with each other.
So that was a pretty significant challenge
and we embodied that concept of that challenge
with this term "bioorthogonal".
So, what this means literally is not interacting with biology.
"X" and "Y" have to react mutually and selectively with each other.
They cannot react with anything
that's found in nature. And of course, they certainly cannot be harmful
to the biological system under study
otherwise that would be a disaster from the perspective of
imaging sugars in live animals.
So we spent many years trying to craft reactions that
basically fit this description where the two
reactive counterparts, X and Y, are bioorthogonal.
And to make a long story short, what we discovered is that
an ideal chemical reporter, the group X that we put inside the sugar
is the azide, which is defined as having
three nitrogen atoms linked to each other
in a manner that is shown here.
Now the azide has a number of properties that make it very well suited
for this particular application as a chemical reporter.
First and foremost, azides are not found in biological systems.
As far as we know chemists invented the azide.
We have not found it in nature.
Maybe nature overlooked this functional group
when she was creating all of the chemical
diversity that's found on our planet.
Furthermore, azides are essentially inert in biological systems.
There's really not much that they can react with
that is normally found in a biological environment.
Azides are very easy to attach to sugar molecules.
The chemistry is very simple.
And importantly, azides are small.
So the van der Wahl's radius of an azide is only about 2.4 Angstroms.
Which means that it is just a little bit larger than a methyl group.
And that's important because we are attaching an azide to a sugar
and then asking all of the enzymes
inside the cell to metabolize that sugar as if
it were a normal metabolic substrate.
And so we can't change the structure too much,
otherwise those enzymes might reject
that sugar as being too foreign, too unnatural.
But the azide is a pretty small modification.
We thought maybe we could sneak it through the door.
of some of those enzymes.
Let me just make a point because
I am asked this question frequently
amongst biologists. Many biologists are familiar with the word azide
in the context of sodium azide
and they know sodium azide to be a metabolic poison.
They often put a little bit of sodium azide into their buffers
to prevent bacterial growth. So they know that sodium azide is cytotoxic.
and therefore they ask me, "Aren't azides cytotoxic?
How can you put an azide on a sugar
and feed it to cells and feed it to organisms? Isn't that toxic?"
The answer is no. Sodium azide is a very different entity
from azides attached to a molecule.
So once the azide is attached to a molecule like a sugar,
it's totally harmless. By contrast, if azide is a salt like sodium azide,
you are right, then it is very toxic.
But we don't actually have to worry about that.
because there is no way in your body to convert a sugar azide
to something like sodium azide.
So, don't worry about that.
Okay. Now what's most important about the azide
is that the azide can undergo chemical
reactions with other functional groups
which remember in the previous slide I called those Y.
The other group is Y. This group is X.
There are several different Y groups that will react with azides.
And the first such group that we explored
is a phosphine. So a phosphine is what you get when you take a phosphorus atom
and you link three substituents to it like I've shown here.
Now back in the early part of the 20th century.
We are talking around 1915 to 1920,
a German chemist by the name of Hermann Staudinger
discovered that phosphines will react with azides to generate
a product in which there is a phosphorus-nitrogen bond
and he called this kind of an intermediate an aza-ylide.
It's a very interesting reaction in that these
two chemical groups join together and make this covalent
adduct. We were intrigued by this Staudinger chemistry
because both phosphines and azides are bioorthogonal
and their reaction is very selective. They are very non-toxic.
And it seemed to fulfill many of these criteria for bioorthogonality.
With one exception. It turns out that aza-ylides
are not particularly stable in water.
They tend to hydrolyze.
So the P-N bond gets cleaved.
And of course, if our goal is to use the chemistry
to attach a probe to a sugar in the living body
where we are surrounded by water
then that hydrolytic sensitivity would be a major problem for an aza-ylide.
So what we did was we modified this reaction in a fairly subtle way
where we attached to the phosphorus atom a benzene ring
that had this methoxycarbonyl group.
So now, once the aza-ylide was formed
faster than it could hydrolyze, this nitrogen atom
cyclized to this carbonyl, cleaved the ester bond,
and formed an amide.
And this cyclization was so fast that it didn't
even matter whether there was water around.
Now once this product was formed,
water could come along at its leisure
hydrolyzed the P-N bond and form this product, but now, because an amide
bond had formed during this intramolecular cyclization step
these two parts, one that came from the azide
the other attached to the phosphine
were still linked together through a very stable linkage.
So this is what we call the Staudinger ligation.
It's a reaction in which we take advantage of the classic Staudinger chemistry
in the first step but then we alter it to form a product
in which the two components are permanently
ligated together in a manner that is
very stable in a biological system.
And the Staudinger ligation was one of the first
chemical reactions that we used to try to image sugars in living systems.
So which sugars would you like to image?
Of course there are many interesting choices.
And as I introduced in my first lecture,
in vertebrates there are nine
fundamental monosaccharide building blocks
that are the constituents of our various glycans.
They are all interesting in their own right,
but when we were thinking about what sugar would we like to image,
we were naturally drawn to this monosaccharide,
called sialic acid. And as I mentioned in my first lecture,
sialic acid tends to appear in some very interesting circumstances
in vertebrate biological systems.
I mentioned for example, that sialic acid
is a component of the ligand for the selectins.
I also mentioned that sialic acid
is the structure that the influenza virus binds to.
Well, it turns out that sialic acid is a pretty busy monosaccharide.
because it also seems to be related to embryonic development
and to cancer. In fact, the glycomes of embryonic cells,
differentiating tissue, as well as cancer cells
have been studied in much detail
and it turns out that many of the glycans
found both embryonically and on cancers
have sialic acid as one of their constituents.
So what I have shown here are just three examples from a much larger set.
These are glycan structures that have been found to be highly elevated
on cancers compared to the normal healthy tissue counterpart.
And many of them are also developmentally regulated.
which is probably not an accident, because we know that often cancers
start to acquire properties that we
normally associate with embryonic tissue.
They've sort of lost their identity
and they are reverting to an embryonic state.
and that's reflected in their glycan structures.
So polysialic acid which is a homo-polymer
made up of repeating units of sialic acid
is highly abundant on a variety of tumors
that are derived from neural crest tissue.
Sialyl Lewis X, which I mentioned before
is part of the ligand for the selectins,
is also found in elevated levels on a variety of different epithelial cancers.
and blood cancers. And Sialyl Tn, a very simple disaccharide,
is actually not normally found on any healthy human tissue
but it appears on a variety of prostate cancers.
and we don't really know why that is,
but the fact that these structures appear in the context of malignancies
has suggested to many people that there might be
elevated levels of sialic acids on these tissues.
So if we could image the sialic acids,
maybe we could detect these tumors in living systems.
That was one of the motivations to study the sialic acids.
Okay. So then the question became, in our minds,
how do we put an azide into the sialic acids?
Well, we have to understand the fundamental metabolism
that produces sialic acid in our body.
And that's what is shown in this slide.
It all begins with this simple monosaccharide
building block called N-acetylmannosamine.
and we abbreviate that ManNAc for short.
So you eat ManNAc and in fact if you eat bread or drink beer,
any yeast product will have some ManNAc.
If you don't eat ManNAc, that's okay
because you are able to biosynthesize it from glucose
through other enzymatic steps that I haven't shown.
But in any event, once ManNAc enters your system,
it has basically one metabolic fate.
It is destined to be converted to sialic acid.
in your cells. And that happen through a series of enzymatic steps
and I am not going to go through all these details.
You can certainly look at the slide in detail if you are interested.
But it suffices to say, that you eat this sugar
and after all of these enzymes have acted on it
at the end of the day a sialic acid is made,
and it appears usually at the end of complex glycan chain
on the surface of your cell, either in a glycoproteins or glycolipids.
So we knew from the work of many labs
that this sugar gets converted to this sugar.
And I've shown the acetyl group in blue.
So that it is clear where that ends up in the biosynthetic product.
And the reason that I highlight that acetyl group
is because we know from the work of several groups
that you can make subtle modifications
to the structure at this position without
significantly harming the efficiency of
any of these enzymatic steps,
which is really quite remarkable if you think about it.
Normally we think of enzymes as being very particular
for their substrates, but in this pathway
there is a little bit of leeway and you
can modify this side chain a little bit.
So knowing that, it was clear to us that this was a site
that we might be able to attach an azide
without confusing the enzymes in the cell.
And that's exactly what we did.
So by chemical synthesis we generate this unnatural version
of ManNAc, which has an azide.
So as an abbreviation we call this compound ManNAz,
where it is the azide version of ManNAc.
And we wanted to get ManNAz into cells, so to do that we
block each of the hydroxyl groups, which are characteristic of sugar molecules
with acetyl esters. Those are protecting groups.
So once the acetyl groups are in place the sugar is now sufficiently lipophilic
that it will penetrate through the cell's membrane, and once it is inside the cell,
we have non-specific esterases that simply cleave
these acetyl groups off and throw them away.
And then the sugar is ready for metabolism.
So the enzymes start processing the sugar
and after several hours what we found is that,
lo and behold, some fraction of the sialic acid residues
on these cells had been replaced with an azido analog.
So in other words, all we do is add this compound to cell culture media,
and the cells do all the hard work
and they produce this azido-sialic acid on their cell surface glycans.
Remarkably the cells don't seem particularly
offended by this transformation.
Some fraction of their sialic acids
have been replaced with this unnatural variant
and that seems fine with them.
But it's very convenient for us
because now we can use the azide to do chemistry.
And so by introducing a phosphine reagent,
that's been linked to a fluorescent probe, we
do a Staudinger ligation between the phosphine and the azide.
now there's a covalent bond between the probe molecule and the sugar.
So wherever there is a sialic acid,
we can now visualize it
using fluorescence microscopy focusing on the probe.
And so in this manner we have been able to image
sialic acids on a variety of interesting cell types.
And sialic acid is not the only sugar
that we can see using the azide as a chemical reporter.
So, just as we can introduce the azide into sialic acid,
by feeding the cells ManNAz.
It turns out we can introduce an azide into fucose
simply by feeding the cells this 6-azido fucose derivative
that has acetyl esters protecting its hydroxy groups.
And likewise, we can introduce azides into N-acetylgalactosamine
or GalNAc, by feeding cells this modified form of GalNAc,
that has an azide on its acetyl group. We call this GalNAz.
analogous to ManNAz, and I actually will be coming back to this sugar
GalNAz towards the end of the lecture.
We also put quite a bit of time into developing probes
that we can use to visualize these sugars.
Now the Staudinger ligation is wonderfully selective
in that it allows us to form a covalent bond between the probe and the sugar.
But more than that we can exploit elements
of that ligation chemistry in order to design what we call smart probes.
So these are probes that are actually invisible
until they find an azido sugar
form that bond through the Staudinger ligation,
and then they become fluorescent.
So they basically swim around in the system,
they look for azides and as soon as they find them,
they make a bond and light up, and you don't see them otherwise.
This allows us to do fluorescence imaging
with very good signal above background.
The way we do this is to exploit the fact that
during the course of the Staudinger ligation,
this ester bond will be cleaved.
In the previous slides I had simply a methyl ester at this position.
where methanol was released during the reaction
which is not particularly remarkable.
But, one could replace that methyl ester
with an ester linkage to a fluorescence quencher.
And if there is a fluorescent molecule bound elsewhere,
that fluorophore will be quenched
by the quencher in this initial molecule.
But as soon as this probe finds an azide,
located within a sialic acid, let's say, or another sugar on the cell,
the phosphine reacts with the azide
the Staudinger ligation unfolds,
the ester is cleaved, the quencher is released,
now that fluorophore is activated.
So only once the fluorophore finds its target
and reacts, only then do you see the fluorescence.
Otherwise it's invisible.
Let me show you an actual chemical structure of a probe
that we prepared that follows precisely this mechanism.
This is a molecule that we call affectionately "QPhos",
which stands for quenched phosphine.
It's got three parts to it:
so over here, this green part is a very well known
fluorescent molecule called fluorescein.
It's a green fluorescent dye commonly used in the research laboratory.
In the middle in black is a phosphine that's all set up for a Staudinger ligation.
You can see that on this benzene ring there is an ester group.
That's positioned for that intramolecular reaction
of the aza-ylide intermediate.
And then finally over here, the blue part
is a well-known fluorescence quencher,
called Disperse Red 1.
And as long as this Disperse Red 1 is linked to this ester
it will quench the fluorescence of the fluorescein molecule.
So let me show you some images that we took using QPhos
to detect sialic acid on cancer cells in culture.
So in this experiment what we did was we grew some cells
called HeLa cells. These are human cervical epithelial cancer cells
that you can grow in a test tube. And while those cells were growing,
we fed them ManNAz in the media.
We just added it to the cell culture media,
they took it up, they digested it, and they put azides into their sialic acids.
Then, after a few days of treatment with ManNAz,
we reacted those cells with QPhos.
And then after a couple of hours we took a fluorescence image of those cells.
The cells were also stained with a blue fluorescent dye
that is specific for the nucleus.
That just helps you to orient exactly where the cells are positioned
since each one has one nucleus.
So, what you can see is that each cell
is surrounded by a nice bright green outline.
Those are the sialic acids on the
membrane associated glycoproteins and glycolipids.
of the cell and they are nicely lit up.
We are actually imaging the sugars.
You might see a little bit of fluorescence inside the cells as well.
That's basically part of the secretory pathway,
the Golgi compartment for example.
And then there's a little bit of staining of
little vesicles throughout the cells as well.
That's because after we react sialic acids on the membrane
some of those sialic acids end up internalized
into vesicles because the membranes are constantly being
engulfed and recycled and turned over.
They are going back into the late part of the Golgi compartment
and then coming back to the membrane.
There's a lot going on in fact during the course of reaction with QPhos.
Now we had some issues once we started
doing these kinds of experiments.
For example, we realized that in order to get this nice bright staining
of the sialic acids, we had to react the cells with QPhos for several hours.
Now that's not a problem for imaging the sugars on cultured cells,
because the cells are in a flask with the reagents
and they can sit in the incubator for several hours.
and just let the reaction unfold.
The reason we had to let the reaction go for several hours
as opposed to several minutes
or several seconds is that it turns out the Staudinger ligation reaction
is intrinsically rather slow. It just takes that long for the reaction to proceed.
And although that was not too problematic
for imaging experiments on cultured cells,
we worried that that could be a problem
for imaging sugars in living animals.
because of course a living animal is not at equilibrium.
It is not cells in a flask sitting in an incubator for many hours.
Animals have an active metabolism.
As soon as you introduce reagents into their bodies,
their bodies are working on clearing those reagents right out again.
And that clearance process can be very rapid for certain molecules
and certain animals. In fact, when we started looking into the Staudinger ligation
as a tool for imaging sugars in laboratory mice
we immediately ran into this problem.
So we injected mice with ManNAz
and we found that in the animal
the cells were perfectly willing to take up this sugar
and convert it to the azido-sialic acid,
that was no problem.
And it seemed harmless to the mice.
So there was no obvious detriment to replacing
some fraction of the sialic acids with the azido version.
The problem came when we then injected those same mice
with probe molecules bound to a
phosphine for the Staudinger ligation.
Because what we discovered is that the
reaction between the phosphine and the azide
was just too slow compared to the rate of metabolic clearance
of the probe out of the animal's body.
So we could only detect a little bit of this product
just on certain organs and tissues
The organs and tissues that were the most accessible
to the reagents after we injected them
or the organs and tissues that had really, really high levels
of the sialic acid we were trying to image.
And that was a little frustrating,
but it basically pointed to a very important element
of this experiment, which we really hadn't considered at the very outset.
You know, ten years ago.
Which is that the kinetics of the reaction
between the azide and the probe molecule
are very, very important for imaging in living animals.
That reaction has got to be fast enough
so that the probe can find the azide and react
more quickly than it is cleared out of the body.
And after many years of experimentation,
we finally concluded that the Staudinger ligation was just too slow.
Its kinetics were too slow compared to its rate of metabolism.
And it probably didn't help the situation that phosphines, in general,
can be oxidized in the mammalian liver
by cytochrome P450 enzymes.
which convert the phosphine to a phosphine oxide
and once that occurs, now this product
is no longer active in a Staudinger ligation.
So, undoubtedly, liver metabolism of the phosphine
contributed to its rapid clearance rate
which just out competed the intrinsically slow kinetics of the reaction.
So the bottom-line is that the Staudinger ligation
was great for in vitro imaging on cells in culture.
Not so great for in vivo imaging in live organisms
that have the ability to clear reagents rapidly out of the system.
So at that point we shifted our attention
to a different kind of chemistry that azides can undergo.
It's another chemistry that has its origins in the last century
in the hands of a German chemist.
In this case, Rolf Huisgen, Professor at University of Munich.
who back in the 1950s discovered that azides
can react with alkynes, undergo a 1,3-dipolar
cycloaddition reaction whose mechanism is shown here,
to form a cyclo adduct product called a triazole.
Now like the Staudinger chemistry,
this Huisgen cycloaddition chemistry
is bioorthogonal in that alkynes and azides
both are not found in biological systems,
at least not in most biological systems
and they react with each other very selectively
without any unwanted side reactions in the living system.
So we thought this would be another very promising avenue to explore.
The problem is that the classic Huisgen reaction
as described with a standard linear alkyne
is also very slow. In fact, even slower
than the Staudinger chemistry with phosphines.
So slow, that when people perform
these reactions in a research laboratory,
they usually have to apply elevated temperatures
or elevated pressures in order to get the reaction
to proceed to completion at a reasonable rate.
And when I say elevated temperatures,
I mean above 100 degrees centigrade for example.
So refluxing toluene. Hotter than the boiling temperature of water.
Obviously we can't do that to cells
and certainly not to living organisms.
But we started thinking about ways that we might be able to
accelerate the rate of this chemistry.
It turns out, back in the 1960s,
there was a publication by Wittig and Krebs
in which they reported that phenylazide reacts with
cyclooctyne which has a triple bond crimped into an eight-member ring
to form the triazole product very rapidly.
Now my graduate students were reading this paper.
It was published in German,
and unfortunately at that time we did not have
any coworkers in the laboratory whose German was good enough
to really understand the paper in its fine detail.
but all of us read this sentence
and recognized three words: phenylazide, cyclooctyne, and explosion.
So as far as we could see, Wittig and Krebs
had combined these two reagents
and they formed a product like an explosion.
which suggested to us that this must be a pretty fast reaction.
Keep in mind however, that what Wittig and Krebs did
is to react neat cyclooctyne,
which is a liquid, with neat phenylazide with no solvent.
So these were very concentrated reagents that reacted like an explosion.
And we figured, if we dissolve these reagents at lower concentrations
and prototypical solvents, maybe the reaction
will be very nicely controlled,
hopefully still very rapid at room temperature.
Now why you might ask does cyclooctyne
react with an azide like an explosion whereas
linear alkynes are so slow that you have to
heat them up above 100 degrees?
That can be understood by looking at what happens
to an alkyne when you force it into an eight membered ring.
Now normally, alkynes are linear.
That is their preferred geometry.
And when an alkyne reacts with an azide, this bond angle
is going from 180 degrees down to 120 degrees in the triazole product.
So that bond deformation costs a lot of energy
in the transition state for the reaction.
That's why we have to heat it up.
To give some of that energy to get over that activation barrier.
By contrast, if the alkyne is embedded in an eight membered ring
now the bond angle is bent to 160 degrees
It's not ideal, in fact, it costs a lot of strain
to bend a triple bond to 160 degrees
But the upshot is that now the difference
between 160 degrees and 120 degrees is quite a bit smaller.
It doesn't cost as much energy to go
from this structure through a transition state
to get to this product.
So our hope was that by straining the starting material
and bending the bond angle so it's closer
to the structure of the transition state
we would lower the activation barrier
so that the reaction can proceed at room temperature.
And that's exactly what we observed when we chemically synthesized
a whole family of cyclooctynes in the laboratory.
So let me just quickly summarize some kinetic comparisons
that we performed using a variety of synthetic reagents.
And so each of these compounds is a cyclooctyne
of some sort in some cases with a heteroatom in the ring
with various different substituents at different sites
and we made these compounds for a variety of reasons
that I won't go into.
but you'll see in some cases there were sp2 hybridized carbons
elsewhere in the ring. Our hope was that we could increase the strain
energy a little bit more by putting
sp2 hybridized carbons in there.
And some of these reagents
have electronegative fluorine substituents.
which we also though might accelerate the reaction
based on some frontier molecular orbital arguments.
Also shown among this collection is a phosphine reagent
which is set up to a Staudinger ligation.
What we did is we took each of these compounds.
We reacted each compound with benzyl azide
in a test tube. And then we measured the
second order rate constant of the reaction.
And in blue are shown basically the relative values that we measured.
So the bottom-line is that we found a compound
that reacts very rapidly with azides in a test tube.
This is a di-fluorinated cyclooctyne
that we have given the abbreviation DIFO.
So that's stands for Di-Fluoro Cyclo Octyne.
DIFO. And the relative rate of reaction of DIFO
compared to some other compounds you'll see,
for example a compound without the fluorine atoms.
is very high. So we can accelerate the rate 75 fold
just by putting some fluorine atoms here
compared to some parent compounds.
Importantly, if you compare the relative rate of reactivity
between DIFO compared to a Staudinger ligation phosphine,
you will see that it is a good 15 times faster.
And so we felt that where the Staudinger ligation
failed in vivo because of its slow kinetics.
we might find success now with DIFO
as a reactive counterpart to the azide.
So we tested that concept in a variety of imaging experiments
starting simply with cultured HeLa cells.
So remember, in a previous slide, I showed you
some images of the sialic acids on HeLa cells
where we reacted the azides with QPhos.
This is now a very similar experiment but
once we put the azides into the sialic acids
this time we react the azides with this compound.
So we call this DIFO-488.
Down here is the DIFO part, the di-fluoro cyclooctyne,
The green part of the molecule is a commercial
fluorescent dye called Alexafluor 488.
It's a bit like fluorescein in its spectral properties.
That is why we call this DIFO-488.
So what you are looking at here are the cultured HeLa cells.
They have been fed ManNAz to put azides into the sialic acids.
Then they were reacted with DIFO-488
and you can see that the sialic acids are shown in these nice
bright green halos around each cell.
The membranes are nicely lit up.
Those are the sialic acids.
This is a control experiment
in which instead of feeding the cells ManNAz,
we now fed them the natural sugar, ManNAc.
There are no azides in these cells.
We also reacted them with DIFO-488.
And now you don't see any green labeling of the membranes.
That is because there are no azides for DIFO to react with.
That's a testament to the selectivity of DIFO.
It only reacts with azides.
If there are no azides around, there is nothing for it to react with.
You don't see the sugars.
But the most important number on this slide
is this one right here.
Now we can see a beautiful image like this
by reacting the cells for less than 1 minute with DIFO-488.
Whereas in the previous experiment with QPhos,
the Staudinger ligation reagent,
we had to react the cells for over two hours.
If we can get this kind of an image within a minute,
I think now we can image the sugars in living animals.
This reaction is fast enough, so we hoped.
So, what animal model would one want to use to image the sugars?
Mice are obviously a very attractive model because one can
study a variety of human diseases in that model organism
including cancer. However, mice are not ideal for optical imaging,
because, as you know they are not transparent.
They are opaque.
One can do fluorescence imaging inside a mouse,
but it requires a specific kind of dye, generally a near infrared dye.
and it's a more complicated experiment.
And for a first attempt at in vivo optical imaging,
we thought we would use a model organism
that is a little friendlier to an optical microscope.
And the obvious choice was the zebrafish.
So zebrafish are translucent.
You can see right through them
and monitor their organs in vivo.
in the live animal. Also, zebrafish
are a very attractive model for vertebrate developmental studies.
They have all the same organs and parts that we have.
They are vertebrates. And their embryonic
developmental program has been very well characterized.
So we know what zebrafish embryos
should look like at virtually all stages of their development.
We also know that there are changes in the glycome.
that correspond with embryonic development.
And so we though this was a very nice organism
to first of all test our chemical tools,
but second of all we might be able to address
some very interesting fundamental questions
of how the glycome changes during development.
And learn something about the field of glycobiology.
So, we started with some proof of concept experiments.
in which we focused not on sialic acid
but rather on a sugar called N-acetylgalactosamine.
abbreviated GalNAc for short.
The reason we were so interested
in GalNAc is because, as I mentioned in my first lecture,
it's the conserved core residue in a family
of O-linked glycans that are associated
with glycoproteins from the mucin family.
Mucins are a general class of glycoproteins
that are characterized by having dense clusters
of these O-linked glycans along the polypeptide backbone.
They're involved in cell adhesion,
regulating cell-cell interactions, and they are known
to be both developmentally regulated
and sometimes upregulated during malignant transformation.
So the mucins are an interesting class of glycoproteins.
They all have a GalNAc residue at their core position.
So we thought if we could introduce an azide
into this GalNAc residue perhaps we could image the mucins
in vivo. And the way we do that is to
simply feed cells, or in this case, zebrafish embryos,
This azido-acetyl derivative that we call GalNAz
once again, the hydroxy groups are protected with acetyl esters.
So that the compound can penetrate cell membranes
and inside the cell the acetyl groups are removed by esterases.
in vivo. Okay. So this was one of our first experiments
we performed to simply ask the question will zebrafish embryos
metabolize GalNAz and introduce it into the mucins,
and if so, can we then tag the azides
with DIFO reagents and image that sugar in vivo.
So in this experiment we fertilized zebrafish embryos in a test tube
and we added GalNAz to the culture media
and just let them develop over the course
of several days and then after about five days
we took the five day old zebrafish larvae
and we reacted it with DIFO linked to a fluorescent dye,
in this case it was DIFO-488,
the same compound I showed on the previous slide
and then we put those labeled zebrafish in a fluorescence microscope
and we took a picture. So what you are
looking at here is the labeled zebrafish
five days old, treated just as shown
in this schematic. And wherever you see any brightness
basically you are looking at the DIFO label
which has been covalently attached to the azido sugar.
There's also a five day old zebrafish in this panel.
You can't see it, I realize.
The reason is that the only difference
is that this zebrafish embryo developed in the presence of
the natural sugar GalNAc, which has no azides.
So when we reacted this five day old zebrafish with DIFO-488,
there was nothing for the dye to react with
so you don't see any fluorescence.
Again it testifies to the exquisite selectivity of DIFO
for the azide. So basically this experiment
demonstrated to us that we can in fact image
those GalNAz residues in a live zebrafish.
And I should point out that these zebrafish
look perfectly normal, perfectly healthy
and happy, despite that fact that some fraction
of their GalNAc residues were replaced with GalNAz,
and that some fraction of those GalNAz residues
were chemically reacted with the DIFO reagent.
These zebrafish were then released back into their tanks,
and they go on to live a normal life as far as we can see.
So it doesn't appear to be harmful to the organism
which is important if the goal is to learn something
about the organism's fundamental biology.
We can also do experiments
where we probe for changes in the glycome,
as a function of time
which again, is one of the most interesting
aspects of the field of glycobiology.
For example, here is an experiment in which we took
a zebrafish embryo fertilized in culture
We add to the culture media GalNAz,
and we let those embryos develop in the presence of the azido sugar
Now, at a given time point,
we will take those embryos out of the media
rinse them off, and react them
with DIFO conjugated to a red fluorescent dye.
In this case it was Alexafluor 647.
Now some glycoproteins, those that were labeled
with azides now appear red,
but we can take those same embryos
that are carrying a fluorescent probe
and just put them back in media
and let them continue their developmental program.
What we do first is a 10 minute reaction
with a reagent called TCEP,
this is tricarboxydiethylphosphine.
What this reagent does is it reduces any
unreacted azides to the corresponding amine.
And incidentally the chemistry by which TCEP
reduces those unreacted azides
is classic Staudinger chemistry.
We are able to take advantage of that.
So we just destroy any unreacted azides.
We take these red labeled embryos,
put them back into the media containing GalNAz,
They continue to develop.
They take up more GalNAz as they are doing so.
And they'll integrate GalNAz into the next
wave of mucin glycoproteins.
And now we can take the embryos out
and label them again, this time with DIFO
conjugated to a green fluorescent dye.
So now, at the end of all of this,
there are two fluorescent dyes on the zebrafish.
The green dye reflect these azides that are
in glycoproteins biosynthesized most recently.
Whereas the red dye reflects the older population
of glycoproteins that were introduced earlier in development.
So this allows us to separate old glycoproteins from new glycoproteins
based on the color that they appear in the fluorescence microscope.
So that kind of an experiment is what led to images like this one
which I showed on the very first slide of this second lecture.
So this is the head of a 5 day old zebrafish
that has been labeled with GalNAz,
followed by three different colored fluorescent dyes
at three different time points.
In this particular image the newest glycoproteins
are appearing red, and you can see there's
a concentration of those glycoproteins
here in the olfactory organs,
which are basically the nostrils of the zebrafish.
Whereas older populations of glycoproteins appear either blue or green.
And you can see that the blue and the green
have quite a different distribution from the red.
And by looking at how these different colors
have moved around as a function of time,
we can learn something about the dynamics of the glycome.
At least from the perspective of these mucin glycoproteins
that we have labeled with GalNAz.
In fact one can use a confocal microscope to walk around
the labeled organism cell by cell
and take a very close look at what has happened
to the mucins as a function of time during the development.
Just for example, here is a panel of epithelial cells
and you can see that each epithelial cell looks as if it has
a red labeled membrane with blue and green puncta inside the cell.
Which is exactly what we would expect,
because in this experiment the red labeling
reflects the newest population of glycoproteins.
Those are the glycoproteins that just
emerged on the plasma membrane
with azides ready for us to label.
By contrast, older populations of glycoproteins
that were biosynthesized many hours before
we took this picture, they've already been recycled.
by endocytic vesicles in the membrane.
And that is why they appear inside the cell.
This is a structure that was rather dramatic
that appeared at around the 72 hour mark post fertilization.
What you are looking at here is a red projection
that is basically coming out of the screen
at you from the junction of three epithelial cells.
That red structure is a mechanosensory hair structure.
These are structures that protrude from the side
of the head of the embryo,
as you can see in this larger scale view.
These are structures that
the zebrafish can use to sense the flow of current in its environment.
And the fact that we captured this structure so dramatically in red
tells us first of all that it must contain these mucin like glycoproteins
otherwise we wouldn't be able to see it
because that is what we are imaging.
Secondly, that the majority of those glycoproteins
were formed within the window between
72 and 73 hours post-fertilization.
Which would be a very difficult window to capture
using other classic imaging technologies.
So these are the kinds of studies that are basically opening the window
through the vision of sugars into developmental biology.
And I think this is a very interesting
future application of the technology.
So what should you take home from this second lecture
in this series. First of all, there is a chemical lesson here,
which really has nothing to do with sugars.
It's the fact that, you know, we've taken advantage
of some very classic chemical tools.
These are chemical reaction that were first reported
as early as 1915. That is almost one hundred years ago.
And these old classic chemistries: the Staudinger chemistry
and the Huisgen cycloaddition chemistry
we have found to be real gems for modern applications in biology.
So by digging into the old chemical literature,
one can find some incredible tools
to bring forth into the modern era of biological research.
And that's very exciting for someone who was trained as a chemist
and at heart considers herself a chemist.
Secondly, you should remember the azide.
The azide has phenomenal properties
that make it so well suited for these applications
as chemical reporters.
And we think that the azide
has a lot of potential not just for imaging sugars,
the way that we do it, but for imaging
all kinds of interesting biomolecules.
including proteins, lipids, nucleic acids,
metabolites, things that are not necessarily encoded in the genome,
for which the GFP and other genetic reporters are really not well suited.
Third we now have shown that metabolic labeling with azido sugars
allows you to image the glycans in living systems.
And we think this will allow scientists
to study glycobiology using all of the hardware of molecular imaging
that so far has focused primarily on the study of proteins.
And finally, where we are going with this technology in my laboratory
is to continue our studies of developmental biologies
until we can understand how the glycome contributes
to some of the early cell fate decision making processes
during the embryogenic program.
For example, how does the glycome change
when pluripotent embryonic stem cells are just starting
to choose their ultimate differentiation fate.
We are hoping that we can capture a snapshot
of that moment from the perspective of the glycome.
And then finally, you know, one of our original motivations
was to develop these tools in a manner that
could be clinically useful. So we are hoping that
we can develop the chemical reagents
the azido-sugars in such a manner that you could
actually introduce these reagents into human subjects.
And use them to detect tumors at early stages
by virtue of their changing glycome.
So stay tuned, and hopefully we will have a lot more to say
about that in the future.
Finally I'd like to acknowledge the students
and postdoctoral fellows in my laboratory.
What you are seeing on this slide
is a snapshot of my laboratory at this particular moment in time
where I am showing you all of the students
and postdocs who are presently in the lab
although I have listed just a small list of alumni
whose work I alluded to during this lecture.
Most of their names were actually mentioned on the slides
along with reference to publications if you are interested in
digging into this in more gory detail.
But it should be said that all of the work that I presented
from my laboratory is in fact the hands-on research
of my students and postdoctoral fellows and I am
largely the mouthpiece that has the privilege of sharing this with the world.
So, thank you very much for tuning in, and I hope you enjoyed learning a little bit
about chemical glycobiology.