We Solve for X: Shawn Douglas on targeted therapeutics

innovation that we’ve seen over the last century, you might
wonder, how is it that we haven’t already solved
cancer as well? And it turns out that there is
a few really big challenges, actually, several very big challenges, in treating cancer. One of which is that the drugs
that we’ve been able to find really lack specificity. So this means that, if we’re
trying to target cancer cells, these drugs also go to
non-cancer cells and cause all sorts of toxic side effects. So that’s one thing that
we need to overcome. Another is that the cancer
cells, themselves, evolve resistance to these drugs. So even if you find a specific
drug, it doesn’t necessarily keep working. And after a matter of weeks or
months, the cells find ways to get around that. So thinking about these two
issues of specificity and overcoming evolved resistance,
it occurs to me that maybe we don’t just need new drugs
to treat cancer. Maybe we actually need entirely
new classes of drugs. And so, in my lab, at UCSF, we
work on building nanoscale devices out of DNA, such as the
one I’m showing here, that I’ll explain in a moment. And I think this is novel,
because we usually don’t think of drugs as devices. We think of them as small
molecules or maybe proteins. And we also don’t
think of them as being on the nanoscale. We actually think of them as
being much smaller than what I’m showing here. So, for example, here’s a small
molecule drug that you probably all have heard of. It’s called penicillin. And this really revolutionized
medicine when it was introduced back during
World War II. And it’s only 41 atoms. And about a 1 and 1/2
nanometers across. And it’s been estimated that,
over the last 70 years, it’s saved anywhere between 80
and 200 million lives. And so, clearly, small molecule
drugs, like this one, can be extremely effective
and very powerful. And I think it can be
instructive to look at how do small molecule drugs work. Why do they work really
well in certain contexts and not in others? And then maybe that can help
us understand how we should approach developing
new cancer drugs. So what happens when you get
a bacterial infection? Well, if you zoomed in with a
microscope and looked, you might see that some unwelcome
bacteria have moved in and started to divide. And this doesn’t really tell
us what’s going on with the drug interaction. So we actually have to zoom in
even further and look at the nanometer scale. And if you could look at this
surface of the cell, you could see that the bacteria actually
manufacture their cell wall out of a polymer called
peptidoglycan, actually. And they also manufacture an
enzyme, that they secrete, called transpeptidase. And this enzyme, actually,
spot welds that polymer together. And this is what gives the
wall it’s integrity. So it turns out that penicillin
is able to bind to and interact with that transpeptidase really perfectly. So when we take a penicillin
drug, it’s actually specifically targeting
that transpeptidase. And if we looked at the
molecular structure, we’d see that the penicillin actually
binds perfectly in this little active site, this little pocket
in the enzyme and inhibits that function. And you can kind of think of
the small molecule as a key that fits, perfectly,
into this lock, this pocket of the enzyme. So with this transpeptidase out
of the picture, the cell wall no longer gets
made properly. And then, as the cells try to
divide and grow, they actually burst under the pressure
of all the stuff that’s building up inside. So the main take home here, that
I’m trying to convey, is that small molecules are really
great when you have orthogonal targets. So what I mean by this is that
the biochemistry of the bacteria is really different
from our own cells. So to extend this analogy, you
could think of the locks that the bacteria use are totally
different from our own locks. So we can eat tons of what is,
effectively, poison to the bacteria, and it doesn’t really
bother our own cells. But in the case of cancer, it’s
important to keep in mind that, actually, all of our
cancer cells, they had a healthy ancestor in the past. So the cancer cells are using
the same locks that the healthy cells use. So if you try to target a cancer
cell, like, say, you find a drug that targets cell
division, because cancer cells are dividing a lot. Well, you’re going to hit
that cancer cell. But you’re also going to hit
healthy cells that are dividing as well. And so this is where
this specificity issue comes into play. And you get all these,
so-called, off-target effects. So that’s one challenge
to overcome. I also mentioned the
resistance issue. So going with the lock and key
analogy, you can think of cancer cells as changing their
locks, constantly. So there’s so much redundancy
built into cells that they can actually turn off certain
pathways and just stop making stuff that we use
as drug targets. So even if you have this really
specific drug that kills like 99% all these cancer
cells, that 1% that survives is going
to come back. And, of course, the drug
is going to not do anything in that case. So this is another
big challenge that we have to overcome. And I have some animations that
I’ve showed you here. But I think that it’s useful
to actually look at the practical outcome of
these challenges. So here, I’m showing a
melanoma patient who participated in a phase
one clinical trial for this PLX4032. And so you can see that the
patient has several tumors all over his body. And so this is right when
he’s starting this drug. And after 15 weeks, this looks
like a miracle, right? All these tumors have
disappeared. He’s gained a ton of weight. It looks really great,
actually. But unfortunately, this is the
photo that was taken right when he’s, basically, eliminated
all of the nonresistant cancer
cells in his body. And only in a matter of weeks
later, basically, all of these tumors come back. And unfortunately, this patient
died several weeks later, after this last
photo was taken. And so I think this really
drives home the point that you can have something that you
think works really well, but the cancer cells somehow manage
to outsmart that. So thinking about these issues
of specificity and resistance, how do we actually
solve these? And keep in mind, we have
to solve both of them, simultaneously. So for specificity, I think
it makes sense to use targeted delivery. So we need molecules or drugs
that can actually interrogate the cancer cells and figure out
whether or not they are cancer cells or healthy cells,
before they deliver the drug. So that’s one thing that
we need to solve. Another, for the resistance,
one approach we can take is combination therapy. And so what this means is that
we deliver multiple drugs to those cell surfaces or to the
inside of these cancer cells. And the idea, here, is that
maybe a cell can evolve resistance or find resistance to
one drug or maybe even two, but it’s probably not going to
be able to outrun three or four or five drugs, all
simultaneously. And, of course, we can’t just
give patients five drugs, because they’re so toxic
and nonspecific. So, basically, we need
to solve both of these problems at once. And to me, if you think about
this, it seems like we’re asking too much of these
small molecules. How could we program so
much sophistication into those 41 atoms? So I think that maybe we
need to build more sophisticated drugs. And, in fact, we need to build
larger molecules, on the nanoscale, that are actually
devices that can, actually, figure out whether or not a
cell is the right place to deliver a drug and then deliver
multiple drugs, only to those types of cells. So how do we build
these devices? Well, we found that, actually,
DNA is a great building material for constructing
nanoscale shapes. So we can program DNA to
self-assemble into these nanostructures. And, actually, the field was
pioneered by Ned Seeman over the last 30 years. And back in 2006, Paul
Rothemund, actually, introduced a really
revolutionary advance. He was working at Caltech. And he figured out
how to make, what he called, DNA origami. And this is using templated
assembly. And I’m going to show you a
quick animation to illustrate how this method works. So, basically, what we use is a
long, single strand of DNA, which we call a scaffold
strand. And we can make this
in the lab. And we know that sequence. And so what we do is we use
computer software to program or design short, so-called,
staple strands of DNA that will bind to the scaffold strand
and force it, in a one-pot reaction,
to, basically, adopt a target shape. So when I was a post-doc at
Harvard, I met another post-doc in the Church lab,
named Ido Bachelet. And we asked the question, could
we actually use this technology to build a device
that could target specific cells and actually deliver
multiple payloads to those cell surfaces? And so this is a prototype
of what we build. It’s a 35 nanometer
diameter barrel. And it’s able to sequester
an antibody payload on the inside. So we can dock drugs. That could serve as a drug. And, basically, it’s a DNA
origami structure, with two domains, so the top
domain, shown in blue, bottom, in orange. And the domains are connected
in four places. There are two hinges in the back
and these two, like twist tie locks, in the front. And I’ll explain how those
work in a moment. But first, here’s what this
looks like under an electron microscope. We can load gold nanoparticles
on the inside, or we can load antibody fragments. So these are 20 nanometer scale
bars to give you an idea of the size. And the way the targeting
mechanism works is it’s actually built into what
we call a DNA lock. And so the DNA lock is basically
a special sequence that people have figured out
how to make, which can actually bind to ligand or a
key molecule and, actually, unzip or separate from its
complimentary strand. So we thought, could we use this
as a building element in making our devices? And so here’s what this looks
like in the unlocked confirmation. The ligand is bound. And it unzips those locks. And then the entire thing swings
open like a clam-shell and exposes its payload to the
surrounding environment, which could be a cell surface. So that’s what that
looks like. So under the electron
microscope, again, we can image this and see the
open versions of what we call a nanorobot. And we want it to perform a
simple selectivity assay, to see if this would
actually work. So what we did was we mixed
together our DNA nanorobots and two cell types. And we loaded, as cargo, these
fluorescent antibodies that could actually bind to
both cell surfaces. And we used this
human leukocyte antigen as our target. But then we locked the
nanorobots with these locks that could only respond to the
keys that are expressed by then NKL cells. And, specifically, this
a PDGF molecule that’s made by the NKL cells. So what we expect is that our
nanorobots will be able to recognize the NKL cells,
open up, and bind to the cell surface. But they’ll leave these Jurkat
cells alone, since they’re not expressing those keys. So we did a simple experiment
to test this. We made a locked version
of the nanorobot as a negative control. And what you’re looking at
here is a flowcytometry experiment, where we can tell
the two cell types apart on the y-axis. So we have those two peaks. And then, on the x-axis, we’re
looking at the fluorescence of the antibodies loaded inside
the nanorobot. So here’s the negative
control. It’s locked. It doesn’t open. So both of those peaks
are on the left. If we add an unlocked version,
so we just leave it open, it’s able to bind to both
cell types. So both of those peaks shift. And so what would you expect if
we had a selective version? Well, what we want to see is
that only the bottom peak is shifting And we were really
excited when we actually got this result. So we mixed these together,
and you can see only the bottom peak is shifting. So we’re able to actually target
those cells, in the presence of all these bystander
cells, which, maybe, look similar, but they’re not
expressing those locks. So we could swap out
that payload and do a killing assay. So we basically use antibodies
that are known to, basically, push the kill switch on the
surface of these cells. And so when they bind, they
actually induce signaling events in the cells that
induce cell death. And so we tested
this, as well. So we have this control. So basically, what we’re looking
at is a dose response curve, where we have increasing
concentrations of the nanorobots, and then we’re
measuring the percentage of cells that we’re able to kill. So the control, basically,
stays flat. The unlocked version starts
killing the cells at a 1 nanomolar and higher. And then the gated version also
does the same thing, once you get to the right
concentration. So we published this paper
about a year ago. So if you’re interested
to learn more, you can check that out. And I just want to spend one
more minute talking about future goals. So one, we obviously want to
move in vivo with this stuff. So we want to test these
in animals and then, later, in humans. In order to do that, we
need mass production. So this is very expensive
right now. We need to figure out how to
synthesize these things on a mass scale. We are also interested
in coming up with new applications for this
technology, this unique positional control that we have
at the nanometer scale. And for that, we actually need
new talent, new students to join the field and
work on this. And to that end, when I was at
the Wyss Institute, as a post-doc, they generously gave
me some seed money to start a student design competition,
called BioMod. And so this has been going
now for two years. We’re starting the third year. So these students all got
together and worked over the summer to design nanoscale
devices. And then they came to
Harvard to present their work in the fall. This was 2011. We had over 150 students
last year. And I think this is actually the
most important thing that I’m working on. Because young students are
really going to be the ones that solve all these problems
and, actually, implement this stuff in the lab. So I’d love to get more people
working on this stuff. And with that, I’ll just
thank all the people who made this possible. And thank you for
your attention. [APPLAUSE] [MUSIC PLAYING]