Authors: Mary J. Laws*, Athilakshmi Kannan, Sandeep Pawar, Milan K. Bagchi, and Indrani C. Bagchi
We generated a mouse model where aberrant estrogen receptor alpha (Esr1) signaling in the hypothalamo-pituitary-ovarian axis leads to ovarian tumorigenesis. In this model, termed Esr1d/d, the Esr1 gene is selectively deleted in the anterior pituitary. The Esr1d/d mice form palpable ovarian tumors which display expression of ovarian epithelial tumor markers, cytokeratin 8, WT1 and PAX 8. Besides proliferating epithelial cells, tumors also contain an expanded population of stromal cells. The loss of negative-feedback regulation by estrogen at the level of pituitary led to elevated production of luteinizing hormone, which stimulated ovarian cells to enhance steroidogenesis. Interestingly, stromal cells of the tumor display a high level of P450 aromatase expression, suggesting that these cells acquired the ability to synthesize estrogen. In response to estrogen produced by the stromal cells, the Esr1 signaling is accentuated in the ovarian epithelial cells, triggering abnormal cell proliferation and tumorigenesis. Consistent with this hypothesis, treatment with letrozole, an inhibitor of P450 aromatase, led to a marked reduction in tumor volume when compared to controls. We have, therefore, developed an animal model of ovarian epithelial tumorigenesis, which will serve as a powerful tool for exploring the involvement of estrogen-dependent signaling pathways in the etiology of this cancer.
Thank you very much, and I'd like to thank the organizers for providing us with this excellent meeting for the Illinois cancer
Community here. I know it's not an easy task to put together something like this
And I'd also like to thank the organizers for giving the opportunity to present my research with you today
So today I'd like to talk to you about an animal model
We've developed in the laboratory, which we think was going to be an excellent model to understand human epithelial ovarian cancer
So just a very quick introduction ovarian cancer is actually the most lethal gynecological disease here
We're looking at statistics for
estimated cancer deaths related to specific types of cancers on the top and estimated
Oh, excuse me estimated diagnosis of cases of
cancers on the top and
Estimated deaths related to those cancers on the bottom and if we look at those statistics for breast cancer
We can see that there's a high level of number of diagnosis cases of breast cancer per year over
200,000
But the deaths related to those cancers still high number of 40,000 about 19% and this
Picture dramatically changes when you look at ovarian cancer, so we have a much lower number of cases
Diagnosed per year about 22,000, but when you look at the cancer deaths related to ovarian cancer
It's dramatically high accounting for about 96% of the diagnosed cases
So this figure here tells us that we really don't have any good diagnostic tools for determining early
Cancer events for ovarian cancer, and we really don't have very good treatments either
So I've told you that we created a knockout or a transgenic animal that does develop ovarian tumors
So I just quickly wanted to tell you how we created that animal
we took an animal which has loxp sites flanking the exon 3 of the estrogen receptor alpha gene and cross that animal with a
Another animal that has Cree recombinase knocked in underneath the endogenous progesterone receptor promoter
So the animal that we generated will have ER alpha disrupted in all cell types where progesterone receptors expressed and this conditional knockout animal is
Termed ER alpha DD for deleted
So these animals do form very large ovarian tumor masses
These animals actually die around 10 11 12 months because they fail to eat because the tumor mass gets so large it compresses the stomach
and the intestines
Here's looking at gross morphology at the progression of the tumor
So we're looking at three months five months and eight months of age and we have a couple different genotypes here
So let me go over this with you
This is our conditional knockout model so you can see that as the animal ages the ovarian tumor mass does get large
Here's our flocks only controls. So compared to a comparative to a wild-type control
You can see that the ovary obviously doesn't get larger as the animal ages and here's a global knockout
Animal for estrogen receptor alpha. So this animal has estrogen receptor alpha deleted in all cell types
This animal does form hemorrhagic cysts on the fall on the ovaries you can see here and we sometimes do see that in our
Animal model we think our animal model is going to be an excellent tool to study
ovarian cancer in humans
this animal does form tumors relatively slowly over the course of the animals life and the tumors form due to a genetic alteration induced by
Physiological perturbations, which I'm going to be talking to you about in the future slides
So we wanted to first characterize this type of tumor that we're seeing in this animal model
And we just wanted to get an idea of the proliferating cell types in the ovarian tumor mass
So we're looking here at PCNA staining immunohistochemistry
For all the subsequent slides all my controls are going to be on your left and the tumor samples will all be on the right
So in a normal ovary we have proliferative granulosa cells here in the follicles, which is normal
And but in our tumor sample, you can see that we have high proliferating cells here in the surface epithelium
Stained here in red as well as in the tumor sample cells within
Within the tumor mass itself. So we do have a highly proliferative ovarian tumor
Forming in these animals and we next wanted to characterize. What kind of tumor cells are these?
So ovarian tumors can arise from many different cell types within the ovary including granulosa epithelial cell, etc
So we first wanted to look at a granulosa cell type marker
So we're looking at anti-malarian hormone or AMH and in the normal
In the normal ovary you can see that the granulosa cells in the follicle are staining positive for AMH as they should the granulosa cells
In our tumor mass, you can see here. I have a good internal positive control
I found one follicle that is staining positive for AMH, but the tumor mass itself is absent for AMH
So this is clearly not a granulosa cell type tumor in the ovary. We next look at cytokeratin. So we're looking
Here on our controls cytokeratin is a marker of epithelial cells and you can see that the one cell layer thick surrounding the ovary
Is staining positive for cytokeratin here
We're looking at three months, six months and 11 months of age
Beginning at three months of age in our conditional knockout model
You can see that there's several layers of epithelial cells on the surface of epithelium
Which is aberrant for an animal of this age. Beginning at six months of age
You can see that there are a large number of epithelial cells within the ovarian tumor itself
And then by 11 months of age, you can see this is a full-blown epithelial ovarian tumor
So these data do conclude that we do have an epithelial cell type ovarian tumor that we're seeing in this animal model
We next wanted to look at the molecular signature of this ovarian tumor that we're seeing in this animal
So we isolated mRNA from normal ovaries and our
tumor samples and subjected it to microarray analysis and we found that there's over
2600 genes aberrantly regulated both up and down one or 1.2 fold or more in these tumor samples. So
Unfortunately, I couldn't show you all my data
But we believe this is going to be a good model for studying a specific type of epithelial ovarian cancer in humans
The serous type so we wanted to compare our molecular signature to that of different
published studies in the humans
So I exported data sets from the Oncomine database
Which is publicly available from three different studies
here, which compared
mRNA levels of different genes in serous adenocarcinomas versus normal ovarian tissue and I compared that to our data set
We were very surprised to see that there is a high level of homology between the gene signatures in our animal knockout model
compared to human serous adenocarcinoma
So in this study is lowest 25% gene homology aberrant gene homology
That's common between the two systems and upwards of 40% homology and aberrant gene expression between our mouse model and human
Which we feel is very significant considering they are two different species
So why does this animal form ovarian tumors in the first place?
I've shown you that the global knockout for AR alpha does not however conditional knockout does and in order to explain that
I just wanted to quickly go over the HPO axis
So the hypothalamus will release GnRH to act on the pituitary which releases FSH and LH
which acts on the ovary to release steroid hormones estrogen progesterone androgens and
The different level of these steroid hormones in circulating the system will have either a positive or negative effect
At the level of the pituitary or hypothalamus to shut off or promote this system. So if you have high estrogen, for example
Estrogen will bind to an estrogen receptor at the level of the pituitary or hypothalamus and shut off the release of FSH LH
And this is normally what happens in cycling women. So we do have a conditional knockout mouse
So I needed to determine where and the what tissue is estrogen receptor alpha conditionally deleted
So we looked at the hypothalamus. We're looking at the third ventricle
The cells standing in red here are positive for estrogen receptor alpha and you can see that they're not so
Significantly different between our control or our
Conditional knockout. So when we look at the level of the pituitary
We can see that there's a high level of estrogen receptor alpha in the anterior pituitary
In a normal mouse. However in our conditional knockout
This estrogen receptor alpha is almost completely ablated in the anterior pituitary
And I'll come back to that in a minute
surprisingly estrogen receptor alpha is
Highly expressed in the surface epithelium of the ovary and normal mouse and we were surprised to see that this estrogen receptor alpha is intact in
Our mouse model and we think that this estrogen receptor alpha is very critical for driving this tumor genesis
Which I'll I'll follow up in the next couple slides
But we turn next to this loss of estrogen receptor alpha in the anterior pituitary
And we hypothesize that this might be having an effect on the steroid hormone levels in this animal model
So we did a panel of steer of serum hormone levels in these mice. They have very high levels of luteinizing hormone
no statistical difference in
Follicle stimulating hormone, but very high levels of progesterone testosterone and extremely high levels of estradiol circulating in this mouse
And so we know that estradiol can be a driver of tumor genesis in other systems such as breast cancer
So again, we wanted to look at the look and see is estradiol driving the tumor genesis that we see in this animal model
so
Estradiol is formed ultimately from cholesterol just a overview of the steroid synthesis pathway from way back in biochem class
uses a lot of different enzymes, but
Ultimately aromatase p450 is important for the last conversion from testosterone to estradiol
And so we want to look to see if aromatase is aberrantly expressed in this tumor
So we at the mRNA level and whole tumor samples
You can see that there's about a 12 fold or 10 fold increase
Excuse me in mRNA level of aromatase expression and we wanted to look and see well
What cell types are producing this aromatase and ultimately estrogen production?
So we enzymatically digested these ovarian tumors and and cultured them
So we're looking at co-localization here
The top three panels here are looking at cytokeratin and red is a marker of epithelial cells
p450 aromatase as those cells which have the ability to secrete estradiol and then a
Co-localized here so we can see that we have a clump of epithelial cells
They don't express aromatase when you merge them. You can't see any overlap. However, the stromal cells which we marked here in vimentin
And marked also with p450 aromatase do co-localized
So it's actually the stromal cells within this ovarian tumor which are secreting high levels of estradiol
And so again, we have to ask the functional question
Well, if it is excreting such a high level of estradiol, does it have any important effect in the production of this tumor genesis?
So we treated these animals with electro with electrosol
You can see that what the animals treated with electrosol have a decreased level of serum estrogen. However, no decrease in LH
So this treatment
Which is a p450 aromatase inhibitor is specific for modulating estrogen and not LH
Here's gross morphology these animals
This is sham control
So a high tumor mass and when you treat them with electrosol the tumor burden is decreased in this of the statistics here
So in the epidemiological data
a lot of studies that have been published do jive with a hypothesis that estrogen is driving ovarian tumor genesis and
It's had been reported in several different independent studies that that
Women who are menopausal perimenopausal or postmenopausal who are given hormone replacement therapies either estrogen alone or estrogen in combination with progesterone
Do have an increased risk of ovarian cancer?
And so in summary
I've shown you that a conditional deletion of ester receptor alpha and the HPO axis
Does lead to a loss of negative feedback of estrogen at the level of pituitary which increases steroid hormones
LH progesterone testosterone and estrogen
We believe that it's the actually the estrogen that's being formed in the ovarian stroma
Which is acting on AR alpha and the surface epithelium to drive ovarian tumor genesis
And we believe that this model is going to be a good model to study the serious type of ovarian
epithelial cancer seen in human which accounts for about 80 80 percent of what we see in human cases for ovarian cancer and
So with that I'd like to thank my mentors and Ronnie Milan bocce. They've been extremely supportive of me and pursuing this research
I'd like to thank all of the bocce lab members
Who have been giving me great support in the project? I'd like to thank our pathologist collaborators. Dr
Hasher talk and dr. Dr. Tong and then all the other collaborators for reagents and technical help
And I'd like to answer any questions that you may have
Thanks, as you know, there's a lot of people now believe that high-grade
ovarian cancers are
originating in the lobe
So in your model, does your model metastasize?
The period behind
We're thinking it's more likely of a low-grade serious adenocarcinoma
They do not metastasize are currently where our hypothesis is that it needs another oncogenic insult for metastasy to occur
Which is why we think it's going to be a valuable model to study early events in tumor genesis
For the fallopian tubia. I am aware of the studied literature for
brachymutated women being
more susceptible to ovarian or
tube tubule
Carcinoma and then metastasizing the ovary. I did remove the oviducts of these animals and there's no difference in tumor formation
So we think it's an intra ovarian tumor genesis event
In the hypothalamus
Just by immunohistochemistry, we didn't notice any difference in the levels of protein of estrogen receptor alpha
Thank you