Tip:
Highlight text to annotate it
X
Hello, my name is Angelika Amon. I'm a professor in the
Biology Department at MIT, and I'm also an
Investigator of the Howard Hughes Medical Institute.
In Part 1, I introduced aneuploidy and its impact on
human health to you. In Part 2 of this series, I told you
a little bit about the work we're doing in the lab. And in
this part, the third part, I would put aneuploidy in the
context of human diseases, and again, I would like to,
with work from my lab, illustrate to you how we think
about the role of aneuploidy in human diseases,
especially cancer. And I want to raise some possibilities
by which we can use or exploit aneuploidy, perhaps,
even in developing new therapies for this disease. Like
all my other talks, I have basically subdivided this part
of the talk into a few subsections. I will first discuss
with you the role of aneuploidy in cancer, how we think
about it, and how we're going to elucidate its role in
these terrible diseases. I will then sort of raise the
possibility that perhaps aneuploidy could be developed
as a new therapeutic target in the disease. So, in the
second part of my talk, I tried to convince you that
aneuploidy is bad news for cells. It causes them stress,
it causes imbalance in the genome, and that is
associated with a lot of detrimental features, and it
basically leads to a stress response. So I hope by now
you are actually sitting there and asking yourself, well,
if aneuploidy is so bad for cells, if it antagonizes cell
division, how come that most solid human tumors are
aneuploid? And that's really a question that the field
has grappled with a long period of time. And there's
many answers out there, and many people have
different ideas about this. Here are my two favorite
hypotheses, okay? The first one is best summarized by
a German proverb that goes: "In the land of the blind,
the one-eyed is king." Let me explain to you what that
means. Even if you're not very good at something --
for example, seeing, because you only have one eye --
if no one around you can see at all, i.e., is blind, you
will still be top dog. And that's how you can consider
effects of aneuploidy. For example, if in an organ of
terminally differentiated cells, if no cell can divide at all,
then the one cell that acquired a copy of Myc, because
it had an extra copy of the chromosome that the Myc
gene is located on, now acquired the ability to
proliferate, even though it proliferates badly because
this extra copy of Myc comes with all the baggage
that's associated with the entire extra chromosome. It
will still prevail. So one idea that a lot of researchers,
and including us, are very excited about is this idea
that, under some specific conditions, aneuploidy can be
advantageous even though it interferes with
proliferation, and there's a number of wonderful
examples for this from experimental evolution studies,
and I've given you an example before: the fluconazole
resistance in Candida albicans that I discussed in Part
2. And so you could envision that aneuploidy might be
a quick and effective way of providing new traits to
quickly adapt to a new environment, for example,
adapting to a new niche during metastasis, and so
forth. So aneuploidy could provide ways to allow you to
adapt to a new environment. The other idea that we
were intrigued by, and that we actually set out to
experimentally test, and I'm going to show you a few
data about this, is the idea that because aneuploidy
causes protein imbalances, that because proteins that
are important for, you know, copying and segregating
the genome may be imbalanced, and that could cause
defects. And it could lead to genomic instability. And so
to test this hypothesis, we asked a very simple
question. The disomic yeast strains, this series of yeast
strains that we developed that carry one particular
extra chromosome, do they show increased genome
instability? And I'm showing you here a large table, and
I don't want you to look at every single one. You can
do that later. What I want to do now is I want to just
show you that this is the series of disomic yeast strains
that we looked at. And then each column represents
one assay for genome instability. We looked for
increased chromosome missegregation, mutation rates,
and various signs of DNA damage, and I'll come back to
those in a minute. And what I would like you to take in
is that there is a ton of pluses on this table. And what
that tells is that every single disomic yeast strain at
least shows one form of genomic instability. Okay? So
this idea that whole chromosomal aneuploidy basically
induces genome instability raises the very exciting
possibility to us that, basically, aneuploidy could be a
driver of evolution. It basically allows the cell to roll the
dice more often. One particular form of genome
instability we were particularly excited about, and I
want to highlight here, and that is this realization that
virtually every single disomic yeast strain that we
looked at had an increased number of Rad52 foci.
What is Rad52? Rad52 is a protein that is important for
DNA repair, specifically it's important for homologous
recombination. And when you start seeing foci in the
cell of that particular protein, that tells you that there's
ongoing DNA repair. And so, these increasing Rad52
foci was not again specific to the disomic yeast strains
that we originally observed. We also saw that again
these more complex aneuploid cells that we generate
through inducing meiosis of triploid yeast cells, we also
saw that there was an increase of cells with Rad52
foci. So a critical question that sort of comes from
these particular studies is sort of: Are there more
breaks, or is the repair defective? We're working on
this, our data right now suggest that there's actually
more breaks being made. We're trying to understand
why they're being made. And so while we are still going
into the details of trying to understand the molecular
mechanisms underlying this genome instability, I want
to highlight another very exciting finding to us. This
was work done by my colleague, Osami Niwa in Japan,
and he for a long time has been studying the effects of
aneuploidy in fission yeast. And his data too suggest
that aneuploid fission yeast cells shown here on the
right, versus euploid cells shown here on the left, they
too have a dramatic increase in number of Rad52 foci,
raising the exciting possibility that these genome
instability-inducing events of aneuploidy could be
conserved at least across yeast species. So obviously a
very important question that we haven't addressed yet
but that are critically important: Are these genome
instability-inducing events due to single gene
imbalances that either interfere with DNA metabolism or
chromosome segregation? There are examples for this
in the literature. We know that when you change the
stoichiometries of centrosome components that that
can lead to chromosome missegregation. Or, is this yet
another facet of the general effects of aneuploidy,
causing cellular stress? There are examples for that in
the literature, also. We know that heat shock causes
sensitivity to DNA-damaging agents, so the general
proteotoxic stress that we in the aneuploid cells could
also lead to increased genomic instability. So clearly,
one way of thinking about how aneuploidy could impact
tumorigenesis, as I said, by inducing genome instability,
it may allow the cells to roll the dice more often, and so
we are now basically interested in trying to understand
if aneuploid cells basically are more quick to adapt to
various stresses that they're imposed on. So to
address this question of the importance of rolling the
dice more frequently. So, irrespective of whether and
how aneuploidy contributes to tumorigenesis (the
gene-specific effects, the genome instability-inducing
effects, and many other possibilities), the fact remains
the vast majority of human cancers are aneuploid. And
so, with that statement, there are two questions that
immediately arise from this. The first one is, these
aneuploid cancer cells proliferate really well. So they
must have learned ways in which they can tolerate the
adverse effects of aneuploidy in order to take
advantage of potential beneficial effects. So the
question is, they must have some mutations that allow
them to tolerate aneuploidy. And if we understand
what kind of mutations allow you to tolerate
aneuploidy, that might actually might help us to
understand how tumors arise, and how tumors evolve.
So one question we're particularly interested in is, is it
actually possible to identify suppressors of the adverse
effects of aneuploidy? And I'm going to show you a
brief example in budding yeast that suggests that that
is actually the case. And a flip side to this question is,
can we identify enhancers of the aneuploid state? If
we find drugs or genetic manipulations that specifically
target either gene-specific effects in aneuploidy or the
general effects of aneuploidy, well maybe then that
would be a way for us to develop new cancer
therapeutics. And after we'll talk about the
suppressors, I'm going to tell you about study in
mammalian cells that we've taken as a proof of principle
that suggests that indeed such compounds do exist. So
basically, we were interested in understanding the
question, can we identify suppressors that allow
aneuploid cells to grow better? The way we did this is,
very simply, we took some aneuploid yeast strains.
They grow really badly, and we keep growing them
until they grow better, and then we simply ask, what
has changed, okay? So this is growth doubling times,
that's a graph that shows how quickly various strains
divide. And in red, you see the parental aneuploid
strain, and in blue right next to it, for example, those
are basically evolved isolates that clearly now grow a
lot better than the parental strain. So what we then do
we take these strains, we submit them to whole
genome sequencing, and then identify the mutations.
And then we've generated a large list of mutations that
allow various aneuploid strains to grow better. What
was interesting is that among all these evolved
isolates, we actually found three mutations multiple
times in various different aneuploids. For example, we
found a truncation mutation in UBP6 in both strains that
had an extra copy of chromosome 5, or an extra copy
of chromosome 9. We found mutations in VPS64,
excuse me, in two strains. And we also found a
segmental duplication in multiple ones of them. So we
decided to pick one of these and look at it in more
detail. And so we started with UBP6 mainly because it's
the first one we found. And so wonderful work by Dan
Finley at Harvard Medical School had previously
characterized this protein. They found that it is a
deubiquitinating enzyme that's associated with the
proteasome. The way you think about UBP6 is that is
competes with the proteasome. So it antagonizes
proteasome function. And so the first question we
wanted to ask is, does deleting UBP6 actually indeed
help these disomic strains, right? Is mutation of UBP6
actually responsible for the phenotype you observe?
As I'm showing you these experiments here, what
we're doing is we're doing what's called competition
experiments here. So we take the, excuse me,
parental strain and we mark it in green here, and then
we take the same disomic strain that's deleted now for
UBP6. We co-inoculate them here at zero, and then we
grow them over time, and then ask, what fraction of
the co-culture is due to one strain versus the other
strain. And you see very clearly that disome 5 and
disome 9, deleting UBP6 really clearly helps them, which
was good news, because that's how we actually
identified mutations. You further might notice that
there were two other strains that were also helped by
deleting UBP6, and actually only one of the disomic
strains actually grew more poorly. So indeed, deleting
UBP6 helps four disomic yeast strains to grow better.
And so Eduardo Torres, who did this work in the lab,
developed the hypothesis that deletion UBP6 increases
the degradation of an unknown number of proteins,
and reverts the protein composition more back to that
of wild type, okay? Sort of because deleting UBP6
unantagonizes the proteasome, so now the
proteasome is more active, it degrades excess
proteins. So using SILAC mass spectrometry
experiments, we can actually test that hypothesis. And
again this was work in collaboration with Steve Gygi,
and what you see here is if you take two wild-type
strains and you compare them, you get a certain
distribution with a certain standard deviation. If you
now compare a disomic 5 strain with a wild-type strain,
obviously there's some extra proteins here, so the
distribution gets wider and the standard deviation
increases, okay? And what's really obvious just from
this bird's-eye view, ten miles up in the air, you can
see, if we now take a disome 5 strain that's deleted for
UBP6, okay, and ask what's the standard deviation of
that is, you see that compared to, excuse me, disome
5 alone, it becomes more narrow. It doesn't reach wild-
type, but clearly things happen. So Eduardo was very
curious to figure out, well, what exactly does happen
to these cells? So he did a simple experiment. He
basically rank-ordered the proteins according to
whether they're overrepresented in the disome 5
strains versus that are underrepresented in disome 5
strains. And then he created three bins. He created a
bin where nothing changes, like within one standard
deviation of the mean (obviously the vast majority of
proteins don't change). Then there's a small group that
gets upregulated in disome 5 versus wild type. And
then there's a small group that gets downregulated.
And he specifically asked, when I now look at these
three groups, what happens, when I delete UBP6, to
these three groups? And that's shown here. That
middle graph, where nothing happens, that basically
shows you that deleting UBP6 doesn't really do
anything to the proteins that didn't change in disome 5
strains in the first place, okay? It makes sense. But
what was very striking is, so these are the proteins
that are overrepresented in disome 5 strains. When we
now delete UBP6, you actually see that they get
dramatically attenuated, okay? So they're getting
downregulated. And he could that this attenuation is
both due to transcriptional and post-transcriptional
mechanisms. And this kind of makes sense, right? So
the proteasome works better, so all these excess
proteins get degraded. But what was unanticipated
and exciting to us, that there, actually the
downregulated proteins got attenuated, too. And he
could show that this was due to indirect transcriptional
effects, okay? So this basically led to a very simple
hypothesis, two hypotheses. One possibility is that
UBP6 attenuates the proteome of the cells that get
improved by deleting UBP6, whose growth properties
increased. The other possibility is that deleting UBP6
attenuates the proteome of all the disomes, but only
the small number that benefits are the ones where
proteins that interfere with proliferation are sort of
cleared, and the guys that don't benefit, well, there,
the reason why they proliferate poorly, deleting UBP6
can't really help them. So there's a very simple
experiment that we can do to distinguish between
these hypotheses. We can take a disome where
deleting UBP6 doesn't make them grow better, okay?
And then prediction is, if hypothesis A is right, then
nothing should happen to the proteome in these guys.
If hypothesis B is right, they should still get
attenuated, but they are not being helped. So we take
a disome 13 strain, we do exactly what we did with
disome 5, and what we found was, like disome 5, the
proteins that were overrepresented in disome 13 were
also attenuated by deleting UBP6. Interestingly, and I
don't have time to get into this, the attenuation of the
downregulated proteins does not occur, and we
actually suspect that that's the reason why these cells
are not helped by deleting UBP6. But what this
experiment tells you is the deletion of UBP6 attenuates
the proteomes of all strains, okay? But disomes in which
UBP6 clients interfere with proliferation are rescued by
deleting UBP6, but disomes by which other proteins
interfere with proliferation that are not touched by
UBP6, obviously deleting UBP6 won't help them. Okay?
So, there's two general conclusions that I want to make
here. Again, our data suggest that the ubiquitin-
proteasome pathway is really important for the life of
aneuploid strains. And sort of a more general
conclusion: Yes, it is possible to obtain mutations that
allow aneuploid strains to grow better. And sort of
obviously, identifying such mutations in the context of
cancer will be instrumental in trying to figure out how
tumor cells evolve from benign to more and more
aggressive disease. So, in the last part of my talk, I
want to address of question of whether the aneuploid
states of cancer can be used in targeted therapy. And
Yun-Chi Tang, a postdoctoral fellow in the lab, asked a
very simple question: Is it possible to identify drugs or
compounds that preferentially impair the proliferation
of aneuploid cells versus euploid cells. And she initially
did a very small, targeted screen that again showed
that indeed such compounds exist. She identified an
energy stress-inducing compound called AICAR and a
proteotoxic stress-inducing compound, 17-AAG, as
having these properties. So initially, when we did this
screen, we sort of rather than screening sixty, a
hundred thousand compounds at the same time, Yun-
Chi sat down and asked a very simple question, well,
what have we learned about aneuploidy over the
years? We've learned that these cells are under
proteotoxic stress, we've learned that they're under
energy stress. And so Yun-Chi reasoned that if she
now took drugs and compounds that also induced
energy stress or that also induced proteotoxic stress,
that that could synergize with the aneuploid state, and
could create what yeast geneticists call "synthetic
lethality." Either condition on their own won't kill the
cell, but if you combine them, you create what we call a
synthetic lethal situation. So used a small number of
compounds, around 30, and started screening them.
And I'm only showing one set of data here, where she
actually looked at this compound AICAR and its effects
on cells. So the cells that she's using here are trisomy 1
cells and she uses trisomy 13 cells, and then she
basically, excuse me, cultures these cells under various
different concentrations of AICAR. And I want to just
you focus on basically the open squares and the filled
squares. So the filled squares here are wild-type cells
treated with high concentrations of this energy stress-
inducing compound, AICAR. They still can grow, albeit
very poorly. But look what happens to the trisomy 1
cells. Not only do they stop growing in the presence of
this concentration of the drug, they actually start
dying, their numbers go down, okay? And the same is
true for trisomy 13, and then to a less extent, for
trisomy 16 and 19. These trisomies are trisomies of
smaller chromosomes, so obviously the energy stress is
predicted to be less. This data here basically shows you
the same data as this graph, except now we normalize
them to the untreated control. I told you in the second
part of my talk these trisomic cells have proliferation
defects, they proliferate poorly, so to specifically look
at the effects of the drug, what she did is she
normalized the growth to "one" for both untreated, and
then asked what is the percentage of the various
growth, respect to zero. And you see very clearly
dramatically different effects for both trisomy 1 and
trisomy 13. I'm not going to tell you all the details. We
characterized these effects in more detail, we know
that it's due to a p53-depedent stress response that
gets induced in these cells, and we don't exactly know
yet the mechanism by which trisomy induces this p53
response, but we're currently working on it. What was
more important to figure out initially is to ask, well,
these drugs that we found to be selective for aneuploid
cells, do they also work for cancer cells? Okay? And so
to address this question, what she decided to do is she
decided to take two colon cancer cell lines (groups).
Colon cancers come in two flavors. They come in "MIN"
flavor, and they come in a "CIN" flavor. The MIN flavor
has low-grade, low-complexity aneuploidies, and the
CIN flavor has high-grade, high-complexity
aneuploidies. So if our drugs indeed has some
selectivity for aneuploidy, the clear prediction is that
they should be more effective inhibiting proliferation of
CIN cells, compared to MIN cells. So that was a very
simple prediction. And so she tested this prediction, in
the data shown here. AICAR and 17-AAG alone have
not that great effects in the cancer cell lines, so she
combined the treatment, and in black you see euploid
controls; then in blue and in green, you see the MIN
cell lines; and then orange, red, and purple, you see
the CIN lines. And you can very clearly see a
differential effect between the MIN lines and the CIN
lines, okay? So down here is the chromosome number
and also whether or not p53 is functional or not. It was
very interesting to see that actually p53 status did not
appear to influence the sensitivity. So she
subsequently asked whether she also sees effects in
xenografts. How these experiments work is you take
an immunocompromised nude mouse, you inject these
cancer cells into both flanks, the MIN line in one flank,
the CIN line in the other flank. And then you treat the
mouse with either control, PBS, or with the drugs
combined. And you see both in the untreated, that
both sites, the MIN and the CIN cancers, grow equally.
But what was very clear is that in the presence of the
drug, the proliferation of the CIN cell line was
dramatically reduced compared to the MIN cell line. I
want to be very clear here, I do not mean to imply that
the only reason for this differential response is the
degree of aneuploidy. I'm sure there's lots of other
reasons why MIN versus CIN lines could respond
differently, but our data are consistent with the idea
that the degree of aneuploidy contributes to this
differential effect. And so with that, I would like to
conclude that aneuploidy could be a potential target in
tumor therapy. So with this I want to sort of end this
part on cancer, and I want to summarize and sort of
also discuss a little bit of what kind of implications our
studies have on how we think about the impact of
aneuploidy in cancer. Our data very clearly indicate
that aneuploidy causes a proliferative disadvantage.
It's bad news to have the wrong number of
chromosomes. And we propose that the proliferative
disadvantage caused by aneuploidy needs to be
overcome during transformation, during tumorigenesis,
for the cancer to take advantage of potential beneficial
effects of the condition. Our data also raise the
interesting possibility that cancer cells may be more
heavily dependent on the mechanisms that help cells
deal with the stresses associated with aneuploidy, for
example, proteotoxicity is something that's very
prevalent in these cells. We would also like to propose
that the phenotypes that are shared by aneuploid cells
may provide a new therapeutic target. I've shown you
one example, AICAR and 17-AAG. Now we're actually
in the process of screening large-scale, chemical
libraries for new compounds, unanticipated compounds,
that perhaps show this differential effect. And then
finally, I think one way to think about how aneuploidy
could promote tumorigenesis is by increasing, basically,
adaptive potential by allowing cells basically to roll the
dice more often. So on that note, I would like to again
thank you for your attention. And I want to end by
acknowledging the people who did this work. The
aneuploidy project was started by Eduardo Torres
many years ago. The transcriptional response work I
told you about was by a graduate student, Jason
Sheltzer. Ana Oromendia did the proteotoxicity and
polyQ work that I showed you at the very end. The
protein aggregation work was also by Stefano
Santaguida. And the hunt for drugs that specifically
inhibit proliferation of aneuploid mouse cells was the
work by postdoctoral fellow Yun-Chi Tang. And last but
not least, funding. Thank you, NIGMS and the Howard
Hughes Medical Institute. And thank you very much for
listening.