To investigate breast cancer characteristics as realistic as possible, we use a mouse model able to faithfully reproduce the human breast cancer progression and regression in all its steps.

6. Discussion

6.1 Mouse model

Breast cancer, as any type of cancer, requires a closer investigation regarding the different underlying molecular characteristics, due to the intrinsic heterogeneity of the disease. The huge differences in the tumorigenic cells profile within the tumor and among the different patients with the same type of tumor, make understanding the cause of tumorigenesis much more difficult.

In general tumor growth is usually driven by oncogenes or loss of tumorsuppressor. The oncogenes are genes that in normal cells promote cells growth and proliferation, and are often mutated and hyperactive in tumor cells. The tumorsuppressor are genes that normally block cell proliferation if the cells are not considered “healthy” after different checks. These cellular brakes, are often inactive in tumor cells.

To investigate breast cancer characteristics as realistic as possible, we use a mouse model able to faithfully reproduce the human breast cancer progression and regression in all its steps.

The mouse model I used in my research work, is a doxycycline inducible system, already used for different investigations in cancer field, that allows us to regulate the expression of two oncogenes.

The tumor growth in many human breast cancer cases is induced by the overexpression of one or more oncogenes. Furthermore tumor regression is the result of targeted therapy that blocks the overexpression of a specific oncogene, interrupting the “driver force” for tumor growth.

In our specific mouse model (TetOMyc/TetONeu/MMTVrtTA), that is still under investigation, the same process is reproduced by the simple administration of doxycycline for around 60 days to induce the overexpression of these oncogenes. Within this time range, the human endpoint in terms of all grown tumors combined (2 cm) has been usually reached and the administration of dox through the food is stopped. Following a period of around 20 days of oncogene inactivation (no dox food), we regularly observe a complete regression of the tumor/s to a nonpalpable state.

The dox withdrawal and inactivation of oncogenes mimic optimal targeted therapy in the clinic.

Nevertheless, even though as radical as possible a treatment, dox withdrawal is not enough to block the further appearances of relapses. In fact, after around 200 days off dox, our TetOMyc/TetONeu/MMTVrtTA model gets relapses in over 50% of the mice that went through one cycle of doxycycline induction.

This means that even in an optimal setting there is not a complete capacity of targeted therapy to eradicate transformed malignant cells. Similar to the patient situation cells remain in a latent state for months before re-growing in a stronger and often fatal tumor recurrence.

To be sure that recurrences in our mouse model are not due to incomplete shut down of oncogene expression after dox withdrawal, I performed Immunohistochemistry on normal, tumor and regressed samples using an antibody against human c-Myc antibody. The expression of c-Myc in tumors was strongly localized in numerous nuclei and lightly present in the cytoplasm, due to the overexpression and the transgene nature of c-Myc. The expression indeed was completely absent in normal and regressed samples.

6.2 In vitro system, oncogene dependence, survival of residual cells and relapse

One way to investigate tumor relapses in detail is to try to study them in vitro. As described in the results, for the in vitro cultures we used a 3D system in which primary mammary epithelial cells can grow in a three dimensional space, forming organotypic structures.

Using primary mouse mammary epithelial cells instead of cancer cell lines, we circumvented

problematic issues of immortalized cell lines. As published by a team of Memorial Sloan-

Kettering investigators, the tumor cell lines that are most often used for cancer research are not

ideal models for investigations of this disease. The reason for it is the genetic makeup of cells

lines that often differs significantly from the one of the patient’s tumors [96].

Using primary cell lines we are able to control the genetic characteristics of our transgenic cells.

Furthermore, the importance of having an in vitro system is that it requires a shorter time to recapitulate tumor progression and regression and give us the possibility to follow single cells.

After reproducing in vitro all the steps of tumorigenesis, as described in the Results 5.3, we focused on the important cellular substrate of recurrence, cells that survive treatment. As mentioned in the Results, also other members of the lab are investigating these aspects with promising results. I am going to mention some first interesting results on those residual cells.

First experiments start from regressed tumor samples, extracted from tri-transgenic mice. When the tumors reached the size of the “human end point”, the mice have been transferred to cages with normal food. After 4 weeks “off dox”, the tumors were completely regressed and the cells of the mammary glands that were hosting the tumors collected, processed and seeded in 3D gels.

These regressed cells, taken after 3 weeks since the mice were off dox and grown in 3D gels, were not able to give any tumor relapse, even if left growing for weeks. However, if reseeded several times (Materials and methods 4.5) they started growing without the administration of dox.

This led to the conclusion that the proliferation stimulus to reform acini structures caused, in the regressed cells, an accumulation of mutations, maybe linked to a different metabolic status of these surviving cells. Right now different analysis are taking place in order to shed light to the transformations occurring in these polarized, regressed but still anomalous cells.

The second experiment provided the collection of cells from regressed mammary glands of mice fed with food without dox for 12 weeks instead of 4 weeks. These cells, maybe due to longer time spent in the fat pad proliferating and accumulating mutations, were able to grow as tumorigenic cells and give relapses after few re-seeding. In this case could be also taken in account the potential role of signals coming from the surrounding stromal tissue helping the accumulation of tumorigenic mutations.

The third step was to test the potential of cells growing in vitro since the beginning, meaning cells taken from 6/8 weeks old mice, seeded and afterwards induced with dox and then left in normal media to regress. The aim, in fact, was to check if mammary gland epithelial cells growing in 3D gels, were able not only to recapitulate acini structures, tumor growth and regression, but also to form in vitro relapses. As described in the Results, tumorigenic cells growing in vitro, if left in normal media for 7 days, are able to completely regress to a re- polarized structure. The question is if these cells are able, after a certain period to regrow and form a relapse without the administration of dox. If this would happen, we would be able to completely reproduce in vitro tumor growth, regression and recurrence. To test this capacity, the cells have been re-seeded in the absence of doxycycline for around 12 times. Cells that had undergone one cycle of doxycycline induction and “treatment” (withdrawal of doxycycline and silencing of the oncogenes) were able to start to grow in solid and filled tumorigenic structures – in contrast to the “never induced” controls. This encouraging result is giving us the possibility to have an in vitro system that can reproduce human breast tumor progression, and is permitting investigations on the molecular characteristics of meaningful in vitro relapses.

As described above, some cells became resistant to the switching off of the oncogene expression upon several rounds of re-seeding and acquired the possibility to regrow and proliferate in an uncontrolled way. One could argue that recurrence could be due to alterations acquired in a few cells during the oncogene overexpression or tumor initiation phase. We do not believe this to be a realistic scenario, since in vivo mice need a timeframe of about 6 months to display relapses and

We can also exclude leakiness of the system, since neither cells nor mice, have been observed to

give rise to tumorigenic growth in the absence of doxycycline, even after one year. To be sure

that also in vitro the doxycycline dependent system is correctly shutting down the oncogenes

expression in the absence of dox, I performed qPCR on in vitro growing cells. I used RNA

extracted form normal, tumorigenic early and late off dox samples, to follow the expression

levels of the oncogene taken as reference, c-Myc. The result was clear and showing a strong

overexpression of the oncogene exclusively in “on dox” samples and a rapid expression level reduction when off dox, even decreasing more with the increase of the number of days off dox.

As I described in the Introduction, most tumor cells are dependent on their initiating oncogene, meaning that they need oncogene overexpression for their survival and proliferation. We observe this principle also in our cells, because after dox withdrawal and consequent block of the oncogenes overexpression, they lose the capacity to proliferate, undergo apoptosis and regress to polarized mammary epithelia structures. My thesis focused of the possible inhibition of the survival of these surviving cells. The rationale would be that a reduction in the cellular substrate for relapse would help reduce or prevent recurrences.

6.3 Investigations on potential candidates for inhibition of cell survival during targeted treatment

I tried to set up a chemotherapeutic approach that, associated with targeted therapy could decrease the onset of relapses. In order to do so, I started by identifying important molecules for mammary gland epithelial cells survival –in developmental, pregnancy and tumorigenic situations.

The rationale for this approach rooted in the observation that tumorigenic cells almost completely lose a structural and polar organization, but when the oncogenes are switched off they survive as a somewhat nicely re-polarized rim. While most tumor cells apoptose due to the dependence of these cells on the oncogenic signals, the cells of the outer layer survive after the dox withdrawal (Results Fig.5.10). This could be due to the presence of strong survival signals coming from the basal membrane surrounding the outer layer of epithelial cells in tumorigenic structures. This strong signal could promote the survival of the outer layer cells, which probably hosts the future surviving and later relapsing cells.

Therefore we decided to try to reduce the probability of tumor recurrence, by reducing the number of survival cells after dox withdrawal. To increase the cell death among the resistant cells to the shutdown of the oncogenes expression, we decide to inhibit key molecules for cell survival and polarity pathways.

As already described, two important molecules, ILK and FAK, involved in proliferation, survival, migration and polarity pathways, often found overexpressed in different cancer types, have been chosen as targets to potentially inhibit. The molecules chosen have been already shown to be involved in tumor regression, decrease of invasiveness and proliferation rates upon inhibition.

Their inhibition, in fact, should target the cells of the out layer that are supposed to be the resistant one to the doxycycline withdrawal and that give rise to relapses after a certain period.

This is due to the importance of receiving survival signals from the surrounding basement membrane. During the regression of tumorigenic structures, the cells surviving dox withdrawal seem to be the outer layer of epithelial cells, the ones juxtaposed to the basement membrane. For this reason we decided to target molecules involved in intracellular signals coming from the basement membrane. The adaptors between the basement membrane signals and the cytoplasmatic pathways, the integrins, are the starting point for several pathways. We decided to target 2 of the key molecules involved in these pathways.

By respectively inhibiting ILK and FAK, we decided to block the Akt and Src pathways,

important for cells growth, proliferation and inhibition of apoptosis.

Fig. 6.1 AKT and Src pathways.

The artificial molecule chosen is inhibiting FAK phosphorylation on tyrosine Y397, which is important for the activation of the downstream survival pathways. Inhibiting the pro-survival pathway, means promoting the apoptotic one, trying to drive the tumorigenic cells to death.

For the ILK inhibition, we have the same inhibition process in which the steric hindrance by the inhibitor block the ATP site.

To further block the reorganization of the surviving cells during doxycycline withdrawal, we decided to use an inhibitory antibody for E-cadherin to block the formation of the adherent junctions. In this way the cells are not able to establish the cell-cell contact, the polarity, and consequently are driven to apoptosis.

The problem for the inhibition of cytoplasmatic molecules such as FAK and ILK, is that, external molecules that go into the cells are often rejected throughout the ABC pumps. Highly proliferating cells with high expression of ABCG2, have been found in tumors resistant to chemotherapy treatment [97]. This cell defense system is often causing drug treatment resistance.

The expression profile of ILK and FAK, tested by immunohistochemistry (IHC), (and confirmed by qPCR in normal, induced and regressed in vitro growing cells) showed a defined signal in all tumorigenic, normal and regressed mammary gland. The expression of the p-ILK and p-FAK is expected to be in the cytoplasm, where the FAK and ILK receive the signal from the Integrin receptors and become phosphorylated. The pan form, indeed, is expected to be in the cytoplasm when the signal from the outside basement membrane through the integrins receptors is absent, and in the nucleus where portions of the ILK and FAK have been found. Interestingly in the nucleus they perform antiapoptotic and genome expression regulation function. These findings in our system are still under investigation.

The most interesting expression is the one of pILK, localized exactly in the first layer of luminal epithelial cells closest to the basal membrane, the layer that we hypothesize to be resistant to dox withdrawal. This expression pattern is present in both normal and tumorigenic tissue, and marks the cells that most probably will receive the survival signals from the basement upon dox withdrawal.

We proposed that inhibiting the phosphorylation of ILK and FAK would inhibit important pathways for cells survival during the targeted treatment of shutting down oncogenic signals.

We could also argue that blocking the phosphorylation of these molecules a larger number of

non-phosphorylated ILK and FAK are available for the nuclear translocation, performed only by

the non-phospho molecules. These are only assumptions that need a biochemical investigation in

order to be better understood. For this reason we started performing western blots on normal

cells, after a nuclear fractionation, in order to separate the nuclear and the cytoplasmatic pool of

ILK and FAK pan and phosphorylated, in normal and tumorigenic tissue and in vitro growing cells. Unfortunately, the presence of a large amount of fat and in case of 3D gels of matrigel present in our samples, prevented a good results. We are trying to improve the protocol. The same problem but due to the presence of a lot of protein from the gels, occurred when we wanted to test the absence of p-ILK and pFAK in the inhibited cells..

Going back to the in vitro treatment, we decide to inhibit, together with the phosphorylation of ILK and FAK, the establishment of adherence junctions. Indeed we wanted to try to destabilize the tumorigenic structures by inhibiting the formation of these junctions that anchor the cells among them in a tissue. For this reason we used an E-cad inhibitory antibody.

We added to this treatment the ABCG2 pump inhibitor, to avoid the pumping out of the inhibitors used for the treatment.

During the test of different dilutions for the inhibitors, in fact, the cells didn’t seem to have any survival problem even with very high concentrations, supposed to be toxic. We repeated the tests with the ABCG2 inhibitor and there we saw toxicity effects. After choosing the right dilutions, we set up the treatment timetable.

As shown in the Results, in the graphic for the percentage of surviving cells after treatment, the E-cadherin had a toxic effect in normal cells treated with E-cad, alone or combined with other inhibitors. Nevertheless, the right E-cad dilution have not only been mentioned by the literature, but also tested. Strangely, in most of the tests for establishing toxic concentrations of the anti E- cad antibody, we did not see a relevant decrease in the number of surviving cells. We are still investigating this aspect.

We decided to start the treatment the day before the oncogenes are shut down, so efficacy of the treatment is secured. This way we wanted to target oncogenes, using the principle of oncogene addiction (see above), and at the same time apply a broader “chemotherapeutic” shock.

According to the hypothesis of the oncogenic dependence, the cell death following the oncogene inactivation, is not a passive response of the tumor cells to the absence of the oncogenic signal, but it is an active emanation of long-lived pro-apoptotic signals from the oncoprotein, after its inactivation [98].

The analysis with bright-field microscope of the normal and treated cells, showed a slight difference in terms of “microscopic” changes in the number of survival cells. Due to the high amount of debree present in the gels after the whole kinetic of induction and de-induction of oncogenes, we couldn’t really distinguish between the effects of the long term dox administration and the inhibitors treatment.

Nevertheless we could argue that at least in presence of E-cad inhibition, it seems that there are much less cells regressed forming normal acini, and much more cells that look like dying.

Before starting a deeper analysis on the effects of the treatments, we performed IF staining on control and treated cells to be sure that the inhibitors were working and in presence of ILK and FAK inhibitors there was complete absence of pILK and pFAK signals. The results is an absence of signal in the treated cells.

Afterwards, to better understand the effects of the treatments, we processed the treated and the control cells and we counted the number of surviving cells after different treatment compared with the control not treated regressed cells.

The counting process in order to determine the number of cells surviving to the treatment, didn’t give the results expected, with a strong reduction in survival cells only in presence of E-cad.

Nevertheless this is probably not the best way to analyze the survival profile, due to the amount of work necessary to get it and the inevitable technical problems behind it. A FACS approach is probably the next step to improve the accuracy of the results.

Nevertheless, after the counting process, we partly used these cells for transplantations and partly

for re-seeding. The aim was to check the final capacity of treated and control cells, to eventually

recapitulate mammary glands growth in cleared fat pad of Rag-/- or even to potentially form

tumor. The transplantation assay have been done replicating the injections of all the treated and

the control cells in two cohorts of mice, one of which have been put in cages with normal food

and the other one with food containing dox. The same have been done with the re-seeded cells.

Half of them have been left growing with normal media and the other half, after 1 week for them to recover, with media with dox.

These experiments have never been done on inhibitor treated cells before and it was hard to predict, if these cells would be able to grow in a cleared fat pad like untreated regressed acini (100% repopulation efficiency of 10000 injected cells [56]). While we were not sure if our controls would repopulate a cleared mammary fat pad, we expected the cells transplanted into mice on doxycycline diet to give rise to tumors. The first results showed a mammary gland regrowth only in two cleared fat pad transplanted with ILK+ABCG2 inhibitors. Indeed, for the in vitro part, the re-seeded cells induced with dox, formed really few tumorigenic solid structures only in FAK+ABCG2 and ILK+ABCG2 inhibited cells and nothing in the On/off dox and Never on dox control.

If this would be confirmed by the in vivo results, it would mean that the inhibition of ILK and FAK increase the malignancy of the cells surviving dox withdrawal, instead of driving the cells to an apoptotic pathway, or decreasing their capacity to form relapses.

Regarding the results of the transplantation in the two cohorts of mice, one fed with food with

dox and the other without dox, all the transplanted mice fed with food with dox developed

tumors. The transplantations in mice were performed twice , but only one cohort of the two

doxycycline induced populations developed tumors, probably due to the lack of time for the more

recently transplanted mice. The whole mount mammary glands are still under investigation to

check the eventual regrowth of mammary gland or the presence of micro-tumors. At the same

time deeper investigations will be performed to analyze the molecular and cellular origins of the

tumors.