In most cases, we don't really know what the consequence is, it's just been found
that they might be overexpressed or underexpressed in a particular type of
cancer. Although, in some cases, we know that
this has a particular prognostic outcome. So, it may be associated with poor or
favourable prognosis. So, just like I mentioned at the
beginning of this, this week, if we don't yet know the molecular mechanism by which
something acts, but we still know it has some sort of consequence for the patient.
In other words, it's associated with a good prognosis or
associated with a poor prognosis, that can still be quite a useful piece of
information, while we can go back and try and find out what's happening at the
mechanistic level. The example I would like to give you
where we know something about the molecular mechanism with the long
noncoding RNA that is overexpressed in some cancers is for HOTAIR.
So HOTAIR, I mentioned when we spoke about the long noncoding RNAs in Week 2.
HOTAIR air is a long noncoding RNA that it acts in trans and so, it's quite
unusual in that sense. It appears to act as a scaffold, because
it binds to both PRC2, which you'll remember lays down H3K27 methylation, and
to LSD1, and LSD1 demethylates H3K4 methylation.
So, what you see here is PRC2 is laying down an inactive mark and LSD1 is
removing an active mark. So HOTAIR can bring these two complexes
together, and therefore, really consolidate silencing at the target
genes. So normally, HOTAIR is expressed from the
HOXC cluster, but then it acts on the HOXD cluster.
This is why it's acting in trans. Okay?
So it seems to act like a molecular scaffold acting in trans, but potentially
has some sequence specificity because it's still acting on a HOX cluster.
What we know about in cancer is that HOTAIR is overexpressed in breast cancer,
and it's particularly associated with metastasis.
In this case, what seems to happen is, if you overexpress HOTAIR in vitro, then it
will re-target PRC2, they didn't test LSD1 unfortunately, but it re-targets PRC2.
What ends up happening is that the gene expression in those cells changes and it
looks more like when HOTAIR is normally expressed when it would, would happen in
embryonic development. So HOTAIR is being expressed in breast,
but it's creating a gene expression signature of more like an embryonic type
of cell, an embryonic fibroblast in this case.
So this suggests that it's not just HOXD which is it a target for HOTAIR, in this
instance of breast cancer, but rather other genes, because what you see is that
metastasis suppressor genes tend to be silenced.
And this might, might be why you have an association with increased metastasis
when you have increased HOTAIR expression.
So potentially, HOTAIR is binding not just to HOXD, but also these other
regions with which we don't know whether or not there is or is not any particular
identity between this transcript sequence and this region in these, in these genes.
It remains to be determined. So, this is a very interesting finding
with relation to breast cancer, but I guess, oftentimes you'll want to know
does this really hold true more broadly or is it just some weird thing that
happens in one type of cancer? And what's being become clear relatively
recently is actually HOTAIR expression is a poor prognostic indicator because of
this assocation with metastasis, not just in breast cancer but also in esophageal
cancer, colon cancer and liver cancer. And so, this mechanism needs to be
established. How is it that HOTAIR is being targeted
to these other regions in the context of cancer?
So, what can we do with this knowledge for HOTAIR, or indeed, for any other long
noncoding RNA? Well, if we just note that it happens to
be associated with a particular prognostic outcome, but we have no real
measure of why that is, or we don't understand yet the molecular mechanism.
We could still use the expression of long noncoding RNAs, whether they be aberrant
overexpression or aberrant decreased expression diagnostically.
So you can detect long noncoding RNAs in in samples if you need to.
And actually, there's already one that's being used diagnostically, and that is
the expression of PCA3. So, this PCA3 is indicative of prostate
cancer, but you can detect PCA3 in the urine.
And this is really good because it's a non-invasive test, it's not even as
invasive as a blood test. It's certainly not nearly as invasive as
having to have a biopsy taken of the prostate itself.
So this is a useful diagnostic test that uses this long noncoding RNA and here's a
picture of this PCA3 long noncoding RNA. Long noncoding RNAs always have this
quite elaborate secondary structure and it's thought that this is how they might
act as some sort of a molecular scaffold because they have so many binding
surfaces a bit like a protein might. But what about therapeutically?
If we know, HOTAIR appears to have an poor prognostic outcome and maybe
mechanistically because it's re-targeting epigenetic modifier complexes.
What could we do about this at the therapeutic level?
Well, we actually know that you can knock down long noncoding RNAs.
In other words, you can reduce their expression with particular other RNAs.
And so, potentially, these could be used in vivo for therapy if it can be worked
out exactly to deliver it to the right cells or to make sure that we can get it
into the cells effectively. And this is something that's being looked
at by pharmaceutical companies at the moment.
So, now, we've spoken about all the different types of epigenetic
modifications we know about and how they go awry in cancer.
So to summarize what we've said so far then, in terms of DNA
methylation, we see genome-wide hypomethylation.
This is predominantly at the repeat, or at the intergenic intervals.
We see tumour suppressor gene hypermethylation.
In terms of histone modifications, there are particular genome-wide alterations,
you see. But you see also gain of silencing marks
and loss of activation marks at the same CPG islands that are methylated.
We know there are actually long range alterations to epigenetic controls.
There are whole regions, megabase regions, that will be silenced or
activated in the context of cancer. And wee know that there's aberrant
nuclear architecture and altered expression of the long noncoding RNAs.
So altogether, this means that essentially, every level of epigenetic
control that we've spoken about over the past five and half weeks now has gone
wrong in cancer. So in the next lecture, I'd
like to start thinking about, although we don't really have an answer to it yet,
how is it that all of these epigenetic aberrations come about?