TEDxCaltech – Stephen Quake – The Integrated Circuit of Biology

Reviewer: Elisabeth Buffard So in Feyman’s famous lecture
that we’re celebrating today, one of the things he foresaw was that the computers of his time could become incredibly more
powerful by shrinking them and increasing their paralellism
with the components. And that certainly came to pass, and the invention
that was key in doing that is the integrated circuit, or
– as we call it today – the computer chip. And through the incredible
paralellisation available through the integrated circuit, we now realise that
these devices are useful. They automate computation
and they’re useful for much more than solving maths problems. We use automated computation in
virtually every area of our lives today, from giving presentations to writing, to communicating. It’s really amazing. And what I’ve been interested in
for the past decade or so is trying to find the biological
equivalent of the integrated circuit. We’re here… What we’ve been
looking for are ways to automate biology. And since the root of nearly
every aspect of biology today boils down to your ability
to maneuver fluids around, this means that instead of
looking for devices with transistors and wires on them, we’ve tried to create devices
that have pipes and valves and pumps and so forth. And back in the early 2000s, my collaborators and I
managed to do that, and we were able
to create fluidic circuits which have, nowadays,
up to tens of thousands of valves on them. And I’ll do decree
any plumbers nightmare. And it really is small plumbing. In the spirit of getting
the baloney out of the presentation: you know: plumbing is one
of the least glamorous of all the trades. And, you know, if you’ve ever had
the plumber in your house versus your electrician, you can tell that it’s not
nearly the same level of regard. And… So trying to invent
better plumbing is not one of those mysteries
of the universe that generally keeps people up at night. However, that being said, plumbing – once you have
this sort of automated plumbing in hand – it allows you to explore
a wide variety of problems in biology, and to do so in a way
that lets you ask questions that aren’t practical to do by just
pipetting things by hand. And in this movie
you’ve been able to see, one of these chips in action, with the dimensions of the channels
are about the size of a human hair, and different colour food dyes
are being manipulated in the device, and this is something where
the dimensions are much larger than you see in a computer chip, and
are correspondingly easier to make and one of the fantastic things
about this is you can make devices of nearly arbitrary complexity
with really a very small number of steps. And I’m going to take you
on a little tour of my laboratory, and some of the things we’ve been
doing over the past few years to demonstrate what one can do
with this plumbing in hand. One of the first things
we began to look at is using these devices
in structural biology. So you’ve seen these
beautiful pictures of proteins that have been imaged
with atomic scale resolution, and the vast majority of those
are obtained through crystallography, where one has to grow the protein into
a perfectly atomically ordered crystal, and then diffract X-rays off of them. And it turns out that one of the least
scientific parts of this whole process is getting the crystals to grow. And we’ve spent quite a bit of time,
trying to bring physics to that problem, with some degree of success. And it turns out, that
these small volumes are useful, not only for achieving engineering
economies of scale, but also the laws of fluid physics
get into interesting new regimes as you shrink the volumes
down to nanolitres and below. And so one can…
one’s intuition about how fluids behave from the swimming pool
and the ocean don’t always apply, and one can take advantage
of novel behaviour that’s not available at the macroscale. And in particular one can manipulate the crystallisation of proteins in ways
that aren’t possible at the benchtop. And so we took advantage of that
to design a series of devices that allow us to grow
crystals and study them, and parts of that became comercialised and are used in labs around the world, and there are now many structures that
have been determined using these chips, including ones of proteins that are
involved in Ebola and Avian flu, and therefore have
potential health benefits and the people who have done so, – this has been reserved generally
for the hardest proteins to solve – and these people say they couldn’t
have done it without the chips. Either because they didn’t have enough
protein to do it conventionally, or they couldn’t manipulate
the crystallization kinetics. Another area where
we’ve used these devices, has been to try to measure
the interactions between molecules. And to do so in
a very highly parallel way. And to do so at a scale where you can
make measurements that weren’t possible to do before. One area which we’ve done that
has been looking at transcription factors, which are the proteins that turn genes
on and off in your cells, and we’ve in fact have been able
to measure the binding energies of transcription factors
to lots of different DNA sequences, all possible combinations in fact. And from those purely
physical measurements we’ve been able to show that you can combine that with
knowledge of the genome to predict which genes are regulated
by any given transcription factor. That’s been a very nice kind
of basic science application of this, in trying to understand
how transcription factors work, and what evolution has allowed
them to do and not to do, and how much of that capability
has been explored. But it’s also lead
to practical applications, for example in drug discovery. And we were able to discover
an interaction that was – with a collaborator, J
eff Glenn at Stanford, – that an interaction that is important
for the replication of Hepatitis C virus, we used chips for the first time, to biochemically prove that
that interaction happens, and then screen small
molecules on a chip to try to inhibit the interaction. And in doing so, discovered
a new drug for Hepatitis C, which is now in clinical trials. And so, one practical application
of this plumbing, and its use to measure interactions
has been in drug discovery. Let me take you to another application, and this is going to require
a small digression. We got interested in using
the plumbing to count molecules. It turns out that it’s fairly challenging
to do precise quantitation of DNA. What’s generally done is to use
a biochemical amplifier called PCR, but because the amplifier is
non-linear it’s difficult to do it precisely, and there’s this beautiful idea
that’s been floating around for a number of years,
called digital PCR, and it’s reached sort of its fullest
fruition in the microfluidic context. And the idea is very simple: if you have a test tube
full of molecules, and then you dilute them into
many different test tubes, so that on average each test tube
has one molecule or zero molecules in it, and then you do this amplification
process in every test tube, and don’t worry about trying to do it
quantitatively, let it run non-linearly. What you’ll see at the end of the day is test tubes that either have amplified
product in or don’t. And if you count up all the ones that
have amplified product in, that tells you how many molecules
you started with. And this so-called digital PCR provides
a beautifully precise way to do DNA quantitation. As you can imagine, this is not something
you’d want to do by hand, and it lends itself very well
to automated plumbing. And in fact, if you look
at this image here, each bright spot you see is the amplified
product of a single molecule, and these panels are loaded with
increasing concentrations of molecules, and you can see that one can just
count up what was there. What’s the use of this? Well, one practical application has
to do with making babies. For those of you who are parents,
you know it’s one of the most remarkable things you can go through, but it’s also somewhat terrifying
when doctors want to stick needles in. And I had this done
to my wife and unborn kid, and it really made me think
there must be a better way to do diagnostics of the baby. And it turns out that gross
chromosomal disorders like Down’s Syndrome affecting
about 1% of live births, and these invasive tests have
their own risks associated with them. And it happened one day, when I was
sitting in my office, reading an article, and discovered that a certain amount
of the baby’s DNA floats around in the mum’s blood, and I realised that by using this
molecular counting we’ve been working on, you can actually measure, whether or not
the baby has Down’s syndrome non-invasivaly, simply by having
a little bit of mom’s blood, which is drawn from her arm
where there’s no risk, and you’re not putting the needle
anywhere near the baby. And so we published a proof-of-principle
study of this some years ago, and now others have gone on
and used this method on real clinical samples to show that
you can in fact diagnose Down’s syndrome and other genetic disorders
non-invasively, and it’s something that’s moving
very quickly towards the clinic. So a diagnostic application of
small plumbing and counting molecules. (Applause) So in the last minute here I’ll give you
one more biological example, which is using the plumbing
to analyse single cells. We built this device, shown here,
with which you take a single human cell, digest it, and distribute its chromosomes
randomly across those different chambers. And then we use very similar
amplification processes to amplify those single molecules, and then
recover the components independently. And what that allows you to do, is analyse an individual’s genome
with sort of exquisite sensitivity, and in fact, as you know, you get
one chromosome from each… one copy of each of your chromosomes
from each of your parents, and virtually all of the genomes
published today have sort of not been able to
deconvolve that in any meaningful way. And being able to separate
those molecules and amplify, and analyse them independently, lets you look at a synthetic
karyotype like this, and say for a given person,
which chromosome came from your mom, which came from your dad. And in fact you can do
even better than that. If you analyse a few cells from
the mom and the dad, you can work out exactly
how their chromosomes were mixed together in the particular
sperm and egg that created that child, so a chance to look back in time, and look at the specific genetics of
those single cells that created that individual. And this sort of haplotype measurement is something that is gonna to be an important part of genome
analysis going forward. The last example is leaving biology
and moving to chemistry, another area where the ability
to manipulate fluids really matters. The ability to use microfluidics in
this field has just had it’s beginning, and you may ask, why would
you want to make nanolitres of stuff? Turns out for areas like
gene and genome synthesis and medical imaging for cancer, PET imaging, you want small amounts of stuff, and we’ve been able to demonstrate
in principle you can do that, and that’s something we look
forward to seeing developed. And so, I’ll just summarize here, by saying we’ve had a lot of fun,
exploring different areas, and you can see there’s applications
of this plumbing in both basic science and
in diagnostics and drug discovery, and the hope is that in the near future
these devices will be in labs everywhere, and liberate graduate students from
the tyranny of pipetting. Thank you! (Applause)


Add a Comment

Your email address will not be published. Required fields are marked *