- Thank you very much.
It's a pleasure being here.
I was told by one of the fellow fellows
yesterday, or two days ago, that I don't even
understand your title.
So have no fear.
It was a bit of a challenge as a chemist
because I'm so used to talking with structures.
Structures, and the beauty of structures,
is really a part of what gets us excited about chemistry.
But I've cut out a lot of structures
in the hopes of giving you a sense for what
are some of the key problems that chemists
are trying to solve.
How does my group go about trying
to solve a specific aspect in this larger
picture of chemistry?
And then, I also want to give you a little bit of a sense
for how does an organic chemist think about problems?
How do we approach the world?
And so first, I really, really want
to thank the Daniels family.
So when Judy called me up last year
and told me this news of my fellowship,
I had the challenge of explaining this to my parents.
So my dad worked in textile plants.
My mom in home construction sites.
And so the fact that their daughter
was going to get paid just to think for nine months
was a little bit mind blowing.
Sort of magical like a fairy tale.
So I really want to thank the Daniels family here
for endowing this fellowship to enable this fairytale.
And hope their generosity and vision really
inspires a lot of people beyond me too.
So with that, I'd like to start, and give you
a sense of first of all, how is it
that we've gotten therapeutics?
Things to treat disease?
Historically, humans have gotten this from plants and things
in their environment.
So we'll for example, chew on birch bark or willow bark
to extract an aspirin-like component from that.
As you can imagine, getting something
from a natural source, it's always
dicey as to how much of that compound is actually
in the source.
And so, in the late 1800s, chemists
started figuring out ways of generating
these compounds in pure form, and to make something
like aspirin.
And many of the therapeutics that you
might be familiar with beyond aspirin
that you might buy at a pharmacy,
are what we call small molecule therapeutics.
These are relatively small compounds.
Many of them were inspired by natural products
from various sources.
And they've been really changing the way
that we can treat disease.
So in the course of these studies,
and really going back to the early 1900s,
with the isolation of insulin--
that protein-- the small 51 amino acid peptide or protein
from the pancreas, we realized that there was actually
some larger molecules that could also be therapeutically
interesting and vital.
So many of the newer therapeutics,
including vaccines as well as all these monoclonal antibody
therapies you may hear about in the news now,
are based on what's called recombinant DNA technology.
And I'll describe that in a little bit of detail
in the next two slides.
And these are known as biotherapeutics.
These are quite a bit larger than the small molecule
therapies that chemists have been synthesizing or extracting
from natural sources.
And these are known as biotherapeutics
even when they are chemically modified in some way.
And this is actually the fastest growing sector
of the pharmaceutical industry now is biotherapeutics
for various reasons.
And interestingly, in 2014, the World Health Organization
identified increased access to the biotherapeutic products
as a global health priority.
So not just small molecule therapies, but also
these much larger molecules.
Now, this is a challenge from a chemistry standpoint.
So we started out by just isolating insulin
from cow pancreases and dog pancreases.
And then, as you can imagine, this is quite tedious.
Some people would have an immune response against a cow
form of insulin versus human.
And so the first version of insulin
was synthesized in the 1960s.
Then with the advent of recombinant DNA technology,
this ability for not just your own selves
to take genetic information and make a protein out of it,
but also harness other cells--
whether they're human or bacterial cells.
We can take those genes, and then
have that new cell make a protein of interest.
And now, there's not nearly as much of a background.
Purification is still a challenge.
But basing recombinant DNA technology starting in the 70s
really transformed our ability to get to biotherapeutics--
these larger biomolecules.
And this was really a result of lots,
of lots of interesting chemistry that was developed.
So why is it that it is so easy to sequence a gene now?
There are a lot of chemists who worked on ways of enzymatically
synthesizing genes, being able to find ways to analyze
those genes, as well as being able to synthesize
the little fragments that are used in what's
called a polymerase chain reaction to make
many copies of a gene.
So these are all different chemical
and biochemical techniques that had
to be developed in order for structure function
relationships to be easy to do with genes, genetic material,
nucleic acids.
While I was a post doc at Stanford in the late 1990s,
there's a lot of technology that was
starting to be developed on this end too.
So Bruce Merrifield won the Nobel Prize
in chemistry in the 1980s for coming up
with an automated way of stringing together
amino acids to make peptides.
So that now, if you're interested in a specific gene
or peptide, you don't have to contact a chemist.
You can just go online, put in the sequence,
and order your favorite gene or protein.
There's a lot of different methods for analysis
of proteins, purification of proteins,
as well as enzymatic methods that
now make it relatively straightforward although still
challenging depending on the sequence,
to access genetic material and to access proteins
from synthetic methods so that you can get very well
defined compounds.
So it's one compound.
Not a mixture the way you might get from a natural source.
Now, we discovered after insulin it doesn't end there.
What is genetically encoded, you make the protein.
But after that, there's all sorts
of modifications that can happen to that protein that
affect its ability to for example,
transfer across membranes.
How long it will circulate in your blood
will depend on these other modifications.
And one of those very important modifications
that happens to proteins is the addition of lots of sugars.
Here's just a schematic of a large sugar.
These sugars can be large enough that they're
as large as the protein itself.
And historically, biochemists would just
cleave the sugars off and study the protein calling it good.
Because oftentimes, the same protein
could have a slightly different bunch of sugars on it.
So it made the problem a lot more complicated.
Now, especially with all the monoclonal antibody therapies,
we're discovering the sugars actually matter.
They matter not just in the pharmacokinetic properties.
In other words, how well does this biotherapeutic
circulate in your bloodstream for example?
How quickly does it get cleared?
But it can also make a difference
on the actual biology of how that monoclonal antibody
interacts.
And so this has really led to an increased understanding
that we need to figure out how to study sugars too.
Not just ignore them.
And so now, you look at this scheme here.
Why is it that the sugars are so difficult to study?
And I'll give you a few slides specifically
as to why these are so challenging.
But basically, when I set up my independent research career
in 2000, I thought well, it would
be nice to be able to get these sorts of tools--
to be able to have commercial chemical
synthesis of a carbohydrate that doesn't take years
and costs tens of thousands of dollars
for a simple trisaccharide stringing
three sugars together.
We need better analytical methods.
We also ideally, could perhaps harness biology to help us make
some of these compounds.
And fortunately, for the field the United States National
Academy of Sciences put out a report in 2012
that identified these exact areas
as increasingly important areas that we need
to really fund and develop.
The inherent chemistry.
Essentially, it said, chemists we need you.
We need to figure out the tools and techniques
to study this class of biomolecules
that mimics the way we've studied peptides
and nucleic acids.
So why is it that the study of carbohydrates
has lagged so far behind the study of peptides
and nucleic acids?
Well, here is your first chemical structure.
So here is a very, very basic carbohydrate.
So it's basically a backbone of carbon with water on it.
That is carbohydrate.
Now, if I just draw it this way, glucose
is the most common carbohydrate.
And that is what's powering your brain right now so you
can pay attention.
But I can't tell what this actually is just
from looking at its structure.
Because it turns out, in three dimensional space,
there's many possibilities.
And in fact, this is the many possibilities
that a structure drawn flat like this actually could be.
Now, not only for many of you, but for many chemists,
they run screaming when they see this
because it's just confusing.
This is like looking out at the Harvard Marching Band,
and seeing a whole bunch of crimson blazers and black pants
and going, where is my friend?
So I have color coded some of these sugars for you.
So basically, the difference between each of these sugars
is just how this hydroxyl group as it's called, this alcohol,
is oriented in space around this three dimensional structure.
And so to me, this is fascinating
because you really need to have an understanding of how
this looks in three dimensions, not just two.
But it makes it a killer problem from an analytical standpoint
and a synthetic standpoint if you
look at all these different possibilities.
And so if you look at the possibilities
for making different kinds of sugar chains,
even something very simple like just hooking two things
together, there are four different nucleotides
for deoxynucleic acid.
You get four different possibilities.
You can get 16 possibilities when you make a dimer.
There are about 20 amino acids for proteins.
That leaves you with 400 diners.
If you only take these 16 sugars,
you can find 2,560 different possibilities
of connecting those two sugars.
And believe me, there's many, many other
naturally occurring sugars.
There's literally hundreds.
So you can start seeing this becomes an incredible problem
in terms of if somebody hands me a sample from a bacteria
that I've never seen, what is the sugar?
It is extremely challenging to tell you what that sugar is.
And so then, of course, it makes it very challenging
to develop a vaccine based on that sugar
if you can't even tell what it is.
And so this is part of the fun part of this project.
But it's also part of the frustrating part
of looking at sugars.
So this is part of a basic problem
that chemists are facing right now.
If you look back to the 1800s, chemists were empirically
looking at a lot of different reactions,
and trying to figure out what are
all these different elements.
And this is my redrawing of Mendeleev's Periodic Table.
This is how he drew it.
But I basically just turned it on its side.
So it looks a little bit more like the modern periodic table.
But he basically took empirical information
from different reactions with different elements,
and then started seeing patterns.
And the beauty of this periodic table
is the fact that it gives you predictive power.
So these question marks were elements
that he predicted or later discovered.
And I can look at this and say, ah, sulfur,
which is one way that you can link a sugar to a protein,
should have similar but slightly different
properties than the oxygen linkages that link
some other sugars to proteins.
So if you're looking at two different elements,
the periodic table is a fantastic way
of generating hypotheses.
Now, one thing I should make you aware of
is the fact that the vast majority of these elements
we really don't know that much about yet.
And the reactivity and how they can react with other compounds
is largely still unknown surprisingly,
even though we've had this basic heuristic for studying these.
Now, my focus has been primarily on the carbon and the oxygen
up here.
And this is in the study of organic chemistry.
It's all the different cool structures
that carbon can make especially in natural settings
like in human biology.
And so the next sort of leap was this idea of how does
a group of these atoms react?
And so many of you are familiar with these two compounds
in their aqueous dilute form--
ethanol, or vodka in it's aqueous
dilute form, or acetic acid, or vinegar.
And ask anyone who's tried vinegar versus vodka,
they're quite different.
And you see it from a chemical standpoint here,
this is an alcohol.
And this is an acid.
These are what we call two different functional groups.
And interestingly, the entire undergrad organic chemistry
classes that I and many people have taught,
are based on this idea of functional groups.
It allows me to compare and have some predictive power about how
an alcohol might react and interact with other compounds
versus how an acid might react.
What I found amusing was this quote from Saul Patai
and this whole first volume in the series
on functional groups in 1964.
It made me argue that the treatment
of a single functional group is not in accordance with the most
modern principles.
Still, organic chemistry and in most
places taught according to functional groups.
And the mnemotechnic advantages of this division
are so great that it will probably not
be displaced for many decades to come.
And in fact, it still hasn't been displaced.
It's exactly how I still teach this.
Now, you may recall from my structures of sugars,
that it's all alcohols.
Now, what do I do?
So here, I have those same 16 sugars.
And now, all I've done is remove one of these hydroxyls
to give you the entire 16 deoxy series.
I could add acids.
I can make literally dozens and dozens
of different carbohydrates.
So the problem now is there is no heuristic model.
And now, that all of these functional groups
are identical, they're just altering their placement
in space.
We don't have great ways of predicting
how one compound will react versus another compound
will react.
And this is one of the frontiers in chemistry
now is, how do you deal with a molecule that
has all these identical functional groups
in the same molecule?
How do I predict how this molecule will react?
And where for example?
Now, when I look at something like this, and go,
OK, the differences are not going
to be nearly as great as the difference between ethanol
and vinegar for example.
How am I going to deal with these more subtle differences?
And so one thing that my group has approached
taking this problem is looking at this entire set as opposed
to just focusing on something like just glucose
or just galactose, which are found in humans.
So three of these sugars have been studied a lot.
The rest have not.
In part, because they're difficult to access.
So we're looking at the entire isomer set.
And we also realize that we're going
to have to do some very carefully
controlled experiments to be able to tease out
subtle differences.
Now, one of the other fun things about being an organic chemist
is actually doing chemistry.
It's really pretty much the whole mind body experience.
I get to use my senses of sense.
A lot of chemists used to taste, but we
tend not to do that anymore.
Sights.
Sounds.
When you're running an experiment,
you're fully all in using your brain as well as your body.
The problem then is there can be a large lead time in learning
the techniques of chemistry.
And so, it's also I realize we're
going to have to start thinking about how to take some
of the human factor out of this in order to run experiments
reproducibly to be able to get access
to some of the underlying phenomena of nature
of comparing these compounds.
So today, I'm going to tell you primarily
about our chemical synthesis.
But my group really approaches these problems as interrelated.
And that also distinguishes my group from many others.
I realized on the analytical end for example,
we don't have authentic standards of compounds
to develop analytical methods because they're
so difficult to make.
The scale at which I need to make compounds
is in large part dictated by the analytical methods
because they require a fair amount of the compound.
And so I realized these were all interrelated problems.
And so my group is trying to study these basically
as a whole system.
And trying to not just optimize one reaction.
But it pushed the entire system forward.
And I'd be happy in question and answer or later to talk more
about this work.
But today, I'm to tell you a little bit
about how we're approaching the chemical synthesis problem.
How do we start looking at something
that's so complicated?
So I mentioned the fact that we really
wanted to be able to take a little bit of the human factor
out of making these molecules, and why is it so difficult?
You saw the basic structure.
Stringing these things together can be quite lengthy syntheses.
And the reproducibility is not because we're somehow
uniquely unable to reproduce so much as if you have a 20 step
process or a 40 step process, it's unlikely
that many people are going to be motivated
and rich enough to reproduce the process.
And so that's a big problem.
And this is a problem with nucleic acids and peptides
too until commercial automated synthesis
platforms became available.
Now, the ways that these became available
was attaching everything onto a solid phase.
Literally, you can attach your first building block
onto a solid phase, grow your biopolymer chain,
rinse away all your reactions, reagents that you don't need,
and then you can grow your chain.
In fact, that's exactly what we tried initially.
But there are various problems with that approach
when you look at sugars.
There's still a lot of difficult linkages
that I won't go into details.
You still need stuff to put on your machine.
The actual building blocks that you put on the machine.
Analyzing the products can be challenging.
Purifying the products can be challenging.
And still is for peptides from automated peptide synthesis.
And then ideally, we'd also be able to interface anything
that we use to be able to make carbohydrates reliably.
We should be able to also interface
that with the already available methods
to make peptides in order to look
at that bio therapeutic space in a very systematic way
to do systematic structure functional relationships
as we're calling it.
So I had the opportunity 3 1/2 weeks ago to go to the German
apothecary museum in Heidelberg.
And frankly, I felt right at home.
In fact, going to the museum I passed by Bunsen's house.
And many of you might be familiar with the Bunsen
burner.
And I was so excited.
I told my husband you've got to get a picture of me.
And the people walking by probably thought I was crazy.
But I didn't care.
Look, how many times have I used the Bunsen burner?
It's great because now I don't have to use bellows.
That was the only thing in that museum
that I realized electricity was a great thing
to be able to control heat.
But otherwise, we still have mortar and pestles,
we still have sand baths, we've got all sorts
of really cool glassware.
The only problem with this approach
is that many times frankly, in my graduate career,
I would reproduce the prep from the 1800s.
And it worked beautifully.
Crystals would crash out.
You're like, you felt this visceral connection
to this chemist who's been long dead.
The problem is there was a long lag time
in learning all the techniques.
When you read a procedure, they just
assume that you know how to put the glassware together,
and how you might heat or do anything.
And so there's quite a bit of training period
in learning how to do this.
So I realized that we're going to have
to start-- if we're looking at these very subtle differences
in reactivity that we're looking at with carbohydrates,
this sort of manual approach is introducing enough errors
that they might obscure or obliterate
the subtle differences that we're looking for.
And so, basically, that was one of the major driving
motivations in automating senses.
Of course, it can be important in terms
of trying to develop the biotherapeutics,
but I also see it as central to being
able to understand the underlying fundamental
chemistry and how these molecules react.
So why do we want to automate this?
Some of you may recognize these sugar building blocks.
We have individual building blocks
that we'd like to put into a machine,
and then just string them together.
Three to nine of these sugars would
be enough to be able to start doing
a lot of fundamental biology work, as well
as the biotherapeutics [INAUDIBLE]..
And of course, the reason we want to automate it in part
is to eliminate a lot of tedious labor.
There are only a handful of carbohydrate chemists
on the planet.
Not nearly enough to feed the need
for being able to make these kinds of structures.
So ideally, we'd be able to easily generate
diverse sugars in very well-controlled sequences
so that we can look at the function based on the sequence.
And the thing that I didn't fully
appreciate until we started doing automation
15 years ago in my lab is that it also
forces you to develop readily reproducible chemistry.
So where a needle goes into a solution
can make a difference in the reactivity,
and the outcome of the reaction.
And so all of these things turn out
to matter when it comes to actually running your reaction.
And that's what we need to discern the underlying
logic of these relatively subtle differences between compounds.
So how do we go about doing that?
So here's Kevon, who is a senior graduate
student in the lab with one of our three synthesizers.
Our goal, like I said. is just to string three to nine
of these sugars together reliably.
And this, I have to say just the actual robotics of this
is quite a bit more complicated than a solid phase peptide
synthesizer because now, I'm not just rinsing resins.
But I'm asking everything to be done in solution.
Not enough solid phase where you need
lots of excess reagents and building blocks
but in solution.
And that's really what's unique about our platform.
And ideally, the reactions must be amenable to liquid handling
because this entire thing we have to make a solution,
have a robot deliver that to the glass flask,
and I'll show you here.
Basically, we have our component with a little basically tag.
It's like a little affinity tag that we
attach under our sugar--
maintains a solution.
And then, we can couple each of our building blocks
and grow the chain like that.
So basically, to get this to work,
we need to develop a purification system
to be able to get this fish out from the mess
of other compounds that are in there.
We need sufficient amounts of these building
blocks, which is something we're still very actively working on.
And then, we of course need methods to put the building
blocks together in a way that a stupid machine or a robot
can deliver them.
And so I don't expect you to look at all--
here are some of the details if you're interested.
But basically, this machine has a whole bunch
of glass reactor blocks.
A little bit step up from the old apothecary lab.
Certainly costs more.
And then we have a robotic arm that delivers the reagents
as we need them.
And it's all basically controlled
from the central computer.
And then it does all the purification steps
to be able to do this sort of iterative process,
and string together sugars one by one.
Now, the development of the process
has allowed us to make oligosaccharide synthesis
a much more reproducible process than it
was before when we had to rely on each individual person
doing these things.
We use the computer to tell the various machine
parts what to do.
And all the variables, including the timing, are specified.
So how long does it take from where
one reagent goes into that flask versus another one?
That is programmed now as opposed to a person going
in there and manually adding and not necessarily writing
down how long it took.
And you see a huge difference between a first year graduate
student and a fifth year graduate student--
how long it takes to get the second reagent in the flask.
And the other great thing is these programs
can be easily shared between researchers
in a much more exact way than a written protocol can be.
Though the other thing that's nice about the automation
platform, and I won't go into details,
it's forced us to develop new chemistry.
And that's always exciting from a development point of view--
is that OK, what is now that we have this process what are
the next set of limitations?
And that's led us into developing
some chemistry of something an element that Mendeleev
didn't get the molecular weight correct but did do existed.
Bismuth, which is once again, on of those elements
that there's surprisingly little known
about in terms of its reactivity and chemistry.
It's also allowed us to adapt.
Look at how do you adapt a manual process into something
that you can do on a machine.
And sometimes the chemistry doesn't just
translate automatically.
You have to develop new kinds of chemistries.
And so that's also been a large part of my lab
is thinking about how do we go about developing
new chemistries, and what kind of new chemistries
can we discover now that we have this essentially
creative limitation in having to use an automation platform.
Now, so far, very few reactions have been shown
to work on automated chemistry.
So the vast majority of different kinds of chemistry
that are run in the world are still done the old fashioned
way-- manually.
In this particular case, we use a computer program
to tell the various machine parts what to do,
which is not a very common skill among chemists.
Very few chemists have computer programming skills
or background.
The interesting thing though is if you took that code
off this computer, you'd have no idea
what it was actually running.
Nowhere in the code does it tell you
what this entire complex actually is.
And the other criticism is this machine
is only accessible to those of us who have
extremely well-funded labs.
And so then, how do you also make this more accessible?
If you want to take all the manual processes,
the hundreds of different chemistry
processes that are being run around the world,
and make them into something that
could be machine readable so we can ultimately
mine that data to look for these more subtle effects that
are not as obvious as a difference between one
functional group or another or one element or another.
But more subtle differences like, why is this hydrogen
different than that one?
Or why is this hydroxyl different than the other?
Ideally, we'd have a method that all reactions will be produced
in a way that's reproducible, and produces
data that's machine readable that we can ultimately
mine that data.
So that's the bigger picture long term goal
in the idea of automating.
So part of what I also did in coming to Radcliffe
is thinking about OK, we have this fancy machine.
It's interesting training students
to use the machine because I mean
I was trained as a traditional chemist
too in that you have to give up some
of that visceral fun of running your reactions
when you have the machine do it.
And as soon you make a mistake in programming it,
and the needle doesn't go where you say it does,
I just want to go back to the bench
and run it the old fashioned way.
So there's definitely issues of just adapting
to having a machine run your reactions for you.
And so in thinking about this, part of it
is this is a huge leap from those glass
retorts in the Heidelberg Apothecary Museum.
It's very, very different.
And so when I think about how most chemists set up reactions,
we have a round bottom glass flask,
we have some way to heat our reaction for example,
and some way to add our reagents.
There are all different parts.
And we put this together.
And I remember being trained as a first year
by a senior graduate student in another group,
you look in there and this set up was a beauty to behold.
It was an aesthetically pleasing experience just
to look at this reaction set up.
And learning how to do that, and do it reproducibly
takes a little bit of time.
So one of the things I'm doing with Benjamin Lee, who's here
with one of my Radcliffe's research partners,
and Alex [? Majelous ?] who's a senior graduate
student in the Pentelute Lab at MIT where I'm also working,
is think about how do we make setting up
an automated chemical process as easy as setting up LEGOs.
I played with a lot of LEGOs and building blocks as a kid.
And the beauty is you have a limited set of building blocks,
and you can build just about anything.
And we'd like to be able to do that too.
But in a way that you can reproduce it by making it
machine readable and codeable.
So how do we make setting up reactions using automation
as easy as building the usual glassware setups?
How do we code the setup itself?
Not just the program that tells the setup what to do.
But you should be able to look at the program
and say, ah, this is what I need to go set up
this particular setup.
And then, we also have to think about what should the output
files be in order to actually make the whole process machine
readable so that all the data ideally
from every lab across the planet can be produced in such a way
that you can start looking at this data using machine
algorithms.
And so this is something that Benjamin, and Alex,
and I have been working on in the last semester.
And it's been a lot of fun.
So basically, we have components in the lab
like this valve that can switch from one thing to another.
And you look at the back of it it's got a little serial port.
And basically, now, the little components--
you'll see you'll notice a difference here--
is basically another computer chip that has a Wi-Fi signal.
So this is basically a relatively straightforward
simple method where we can take some of the components
that we have in the lab and Wi-Fi
enable them to put them onto what
we call the internet of things.
How many of you have heard of the internet of things?
OK.
So yeah.
So we don't have a lot of internet of things things
yet in the chemistry labs.
And so how do we go about doing that?
So the nice thing about having this as opposed to cabling
is now, you can literally mix and match
all these pieces in your fume hood or on your bench.
And then, have them all talk to one other
with a computer program.
And so what Benjamin, Alex, and I've been working on,
is actually the program.
How do we think about this in a very modular approach
so that the program spits out what were the parts?
What did we actually put together?
Because we're going to have to connect
all these different parts usually with tubing.
And then, how do we get also a program
that actually runs the process?
So the idea is at the end of this project,
hopefully by May or earlier, we'll
have computer program code that if we gave you the code,
you could not only have a list of all the components you would
need that you could then physically assemble and put
together--
I promise not to do it in Byerly Hall--
in your fume hood, and then basically
also have the program to run those components so that you
could run the reaction in Bangladesh, and Ghana,
in China, and India, in Canada, anywhere
around the world exactly the same way
that we ran it here in Cambridge, Massachusetts.
And so to that end basically, we have
the components, the program that codes
not just the protocol which has been done with automation,
but also the modules set up.
So Benjamin programmed this so that he hears the components
that we need for this particular reaction
in a continuous process.
Here's the manifold, the pumps, prints out the components.
Tells you exactly the type of tubing to hook up
all these different components.
And it also spits out a nice little diagram
so you get a sense of OK, if my set up doesn't look like this,
I've done something wrong.
I'm missing a component here somewhere or the other.
And so now, we're going about also developing ultimately will
be a graphical user interface.
But we notice graphical user interfaces date themselves
quite quickly compared to the underlying programming.
So that's why we're waiting for that
to be last so that it can also be easily updated.
And so now, we have the first sort of modular way
to be able to put components together.
And we're also hoping that all this data will be in a machine
readable format so that we can deposit it,
and anyone across the planet could actually
start mining that data from the outputs.
So part of what we've been using this setup for initially,
is to be able to link a sugar to an amino acid.
The very basic component of a peptide.
So this goes back to the title of the talk,
how do I marry the automated oligosaccharides synthesis
to the automated peptide synthesis?
Well, the peptide synthesizers use amino acids
in a protected form.
So the reactive group is blocked with something
to make it less reactive.
And ideally, we'd like to be able to put a sugar on there.
Right now, you can certainly do that.
And they're extremely costly.
In the case of this, an amino acid with an oxygen
linked here you can buy that sugar,
but it's about $10,000 a gram.
And you start thinking, ah, no wonder lots of people
aren't working on this compound.
Labs aren't that well-funded.
And the starting with a process my post-doc at IU developed,
we have now, developed this into a more automated process
whereby we can flow in, make a solution of our sugar,
a solution of our protected amino acid,
and then in one step be able to link these two things together
so that after purification, this is
ready to go into a solid phase peptide
synthesizer in the Pentelute Lab at MIT in this particular case.
So it's the first continuous process
to make sure modified amino acids is ready for solid phase
peptide synthesis.
And it's in the process now of being automated.
And it requires only one step from these two
commercial starting materials, which
also makes it more feasible.
So we basically figured we brought the price down
for about $10,000 a gram to about $10 a gram,
so that we can start putting these pieces together
to be able to start synthesizing glycosylated or sugar modified
peptides, and looking at that entire space.
So one of the other appeals of this fellowship
is that there is a large glycobiology
center that moved to Harvard Med School two years ago.
And I'm also working with Rick Cummings and his lab
there to start looking at generating antibodies
for example specifically against sulfur linked
sugars versus the more commonly oxygen and nitrogen linked
sugars.
Because frankly, we didn't even know until two years ago
that humans had sulfur linked sugars naturally occurring
in their proteins.
So we're quite excited about this possibility that
opens up based on the synthesis itself
developing the synthesis.
And so, how far along are we to making
precision biotherapeutics?
Well, I hope to give you a sense today
that the chemistry for stringing sugars
together is becoming much more well-developed than it was even
in the last 10 years.
We do have machines now that can string together the sugars.
But access to the sugar building blocks
themselves has turned into the major hurdle.
And so that's the other reason why
I'm excited about lab scale automation
of continuous processes to be able to bring the steps
to make the sugar building blocks themselves
into something feasible so we could easily
get kilos of material to drive that chemistry.
We have an easy modular approach process
to automate the chemical synthesis
that we hope should allow the much more widespread adoption
of automation to build databases of reproducible chemical
reaction data.
The goal that Alex, and Benjamin,
I set, and Miles Ingram, who's the new Radcliffe research
partner who just started this week on this project,
is to make this cheap enough that a primarily
undergraduate institution can easily adopt these automation
processes and incorporate them into their labs
in training their students and in generating data sets.
Integrating automated sugar and peptide synthesis
is ongoing with now, these much more affordable building
blocks.
So we're quite excited about looking at and exploring
this glycopeptide space now.
And then, ultimately of course, I'd
like to be able to do enough chemistry to start developing
rules to predict the reactivities of all
those different sugar structures I've shown you.
And that's where I think automation
and artificial intelligence algorithms are really
going to be needed.
But for that, we need the data.
And so the automation, we're one step closer
to being able to get that reproducible data that we
can start to machine read.
And so with that, only thing left for me to do
is thank the many researchers that has helped me.
This is my research group back at Indiana who's helped me.
Here's Miles and Benjamin, who are Radcliffe research partners
along with Alex in the Pentelute Lab at MIT.
The various funding sources.
And that basically is part of the joy
of being a professor is working with a lot of talented students
from around the world.
So with that, I'd be very happy to answer questions.
Thank you.
For more infomation >> Chinese New Year Message 2018 - Professor Paul Boyle University of Leicester - Duration: 1:24. 
For more infomation >> I am Leaving from Cinema..? - Kamal Haasan Clears..! | Harvard University | kalakkal cinema - Duration: 3:56. 
For more infomation >> University of Nebraska supporters pack hearing to oppose budget cuts - Duration: 2:27. 
For more infomation >> Chapman University | Anything Imaginable - Duration: 0:31. 
For more infomation >> Kimberley Seed, University of California, Berkeley - Duration: 31:45. 
For more infomation >> Andrew Mehle, University of Wisconsin - Madison - Duration: 31:43. 
For more infomation >> Barbara Valent, Kansas State University - Duration: 35:23. 
Không có nhận xét nào:
Đăng nhận xét