**2.1.4 Breast cell lines and metabolism of therapeutic drugs**

#### - Drug metabolism

24 Breast Cancer – Focusing Tumor Microenvironment, Stem Cells and Metastasis

cells (Liscovitch & Ravid 2007), and experiments of Affymetrix SNP array analysis at the Sanger Institute (Cancer Genome Project) and of karyotyping, helped to put in evidence an indisputable resemblance of NCI/ADR-RES with the OVCAR-8 human ovarian carcinoma cell line. The most likely scenario is that the stock of MCF-7 cells from the National Cancer Institute used in 1986 for the development of the lineage, was contaminated with OVCAR-8 cells before the first generation of MCF-7-ADR-r. OVCAR-8 cells are naturally resistant to adriamycin, and the *in vitro* selection probably eliminated the MCF-7 cells and allowed the survival of OVCAR-8 cells (Liscovitch & Ravid 2007). It can be noted that MCF-7-ADRr are

The second misidentification concerns the MDA-MB-435 cell line established by Cailleau and colleagues in 1978. This cell line has been controversial in 2000, further to the results of DNA microarray analysis which suggested that these cells might be of melanocyte origin (Ross & Perou 2001). Some other results, obtained by microsatellite comparison analysis, karyotyping and comparative genomic hybridisation experiments (Rae *et al.* 2007),

However, these two cell lines, MCF-7-ADRr and MDA-MB-235, are still used as breast cancer cell lines for some studies and are used for publications in international journals, while it has been proven that they are not from breast cancer origin (Lacroix 2008). The verification of the origin of a cell line is essential, and a way of ensuring that the cell lines are really from a well-defined origin is to make a short tandem repeat (STR) profiling. This method is used to confirm the identity of a cell line by comparison to a known profile and a periodic re-authentication of cell lines is advisable. Moreover, banks of cell lines such as ATCC guarantee the exact origin of their cells. Several authors suggested to prove the authenticity of the cell lines used for each publications (Burdall *et al.* 2003; Lacroix

It should be noticed that the study of mammary tumours also involves the use of non cancerous cells which were immortalised. These cell lines were derived from healthy breast

 The immortalisation could be the consequence of a particular composition of the growth medium. This is the case for the non-tumourigenic epithelial cell lines MCF-10A (adherent cells) and MCF-10F (floating cells) which were established from the same sample in the nineties (Soule *et al.* 1990). These cell lines were produced by a long-term culture in a special medium containing a low concentration of Ca2+ and no serum addition, which resulted in the apparition of immortalised cells with normal features of

 Two other cell lines were derived from a mammoplastic surgery. These cells named MCF-12A and MCF-12F became spontaneously immortal after unexpected exposition to

 Another cell line, hTERT-HME1 was obtained from the HME1 cells (Human Mammary Epithelial) which were immortalised by infection with the retrovirus pBabepuro+hTER. The immortality feature results from the exogenous expression of the telomerase gene coming from the viral infection (Van der Haegen & Shay 1993;

no longer distributed by the international cell bank ATCC.

**2.1.3 Non cancerous immortalised cells as controls** 

mammalian epithelial cells.

Gollahon & Shay 1996).

tissue, but only few models, obtained by different methods, are available.

high temperatures (45°C during 72 hours, Pauley *et al.* 1991).

2008).

confirmed that MDA-MB-435 cells are in fact M14 melanoma cells.

The metabolic equipment of a cell can explain its sensitivity/resistance to drugs. Indeed, any xenobiotic molecule (therapeutic drugs included) undergoes the same metabolic fate in the cells. Briefly, enzymes of Phase I (essentially cytochromes P450 (CYP) dependent enzymes) ensure a bioactivation of the molecules while enzymes of Phase II conjugate the metabolites issued from Phase I to endogenous molecules (glucuronic acid, glutathione, sulfates…) in order to make them more water-soluble and to facilitate their elimination. Finally, transporters of Phase III are responsible for exporting these last products out of the cells. Each human organ is equipped with these enzymes, but their expression pattern differs quantitatively and qualitatively. The liver is the most efficient organ in metabolising processes, even if we know that some enzymes are more specifically expressed in non hepatic tissues.

When considering the usefulness of breast cell lines as *in vitro* tools to predict sensitivity or resistance to a molecule, it is easy to perform, in first line, simple cytotoxicity tests. However, in order to explain the reasons of these cells behavior, or to predict the metabolism of a new compound, the knowledge of the metabolic equipment of the cells is necessary. As it is impossible, and not very interesting, to decline the results of the literature concerning breast cell lines and assays with the numerous chemical molecules which have been, precisely or not precisely, tested, we chose two examples of therapeutic drugs, used in breast cancer, that need to be bioactivated by CYP before exerting their deleterious effects in the cells: oxazaphosphorines and ellipticine.


The oxazaphosphorines generally used in pharmacology (*i.e.* cyclophosphamide (CPA), ifosfamide (IFO), and trofosfamide) represent an important group of chemotherapeutic agents. However, their use is limited by severe toxic side effects. New oxazaphosphorines derivatives have been developed in order to improve selectivity and to reduce toxicity but they won't be studied here, due to their bioactivation process which is different from that of previous molecules (Zhang *et al.* 2005).

Both CPA and IFO, the most widely used as alkylating agents, are prodrugs whose metabolism involves different cytochromes P450 (CYPs) catalysing 4-hydroxylations leading to acrolein and nitrogen mustards capable of reacting with DNA molecules leading to cell apoptosis and/or necrosis. Another pathway consists in an N-dealkylation whose last product is the toxic chloroacetaldehyde (Figure 1) (Rooseboom *et al.* 2004; Zhang *et al.* 2005). All these metabolites are highly reactive metabolites responsible for urotoxicity, neurotoxicity and nephrotoxicity. As all the mechanisms underlying these toxicities are not

*In Vitro* Breast Cancer Models as Useful Tools in Therapeutics? 27

Fig. 2. Hepatic CYP expression (A) and their contribution to metabolism of clinically-used

Another example is given by ellipticine. This alkaloid compound found in several plants (Ochrosia, Aspidoserma subincanum, Bleekeria vitiensis) is a topoisomerase poison often used in ovarian and breast cancer treatment. It is also a prodrug whose efficiency depends on CYP activation. 13-hydroxy- and 12-hydroxy-ellipticine, responsible for the formation of

Fig. 3. Main pathways of ellipticine metabolism. Reprinted from Stiborova *et al.* 2011, ©2011,

Members of the CYP1 family are usually expressed in extrahepatic tissues and it is not strange to find CYP1A1 in MCF-7 (Androutsopoulos *et al.* 2009; Stiborova *et al.* 2011), in MDA-MB-231 and T47D (Macpherson & Matthews 2010). We already mentioned the presence of CYP3A4 in the three cell lines, but no precise information is available for

This slight overview shows that the three main breast cancer cell lines are able to give interesting information about drugs that have to be bioactivated before exerting their deleterious effects in cancer cells. However, we must keep in mind that polymorphic variants of the genes coding these enzymes, or splicing variants, may influence the

drugs (B). From Wang & Tompkins 2008, permission granted by Bentham Science

DNA adducts, are generated by CYP1A1/2, CYP3A4 and CYP2C9.

Publishers Ltd.


with permission from Elsevier.

CYP2C9.

elucidated, Mesna (Sodium 2-mercaptoethanesulfonate) is often used to limit these side effects (Giraud *et al.* 2010).

Fig. 1. **First phase of metabolism of the oxazaphosphorines by CYPs**: hydroxylation leads to oxazaphosphorine mustards, and N-dealkylation results in chloracetaldehyde formation. From Rooseboom *et al.* 2004 with permission from ASPET.

As already mentioned, several CYPs are involved in these drug metabolism: CYP2B6 (Wang & Tompkins 2008; Mo *et al.* 2009; Bray *et al.* 2010), CYP3A4 (Kivisto *et al.* 1995), but also CYP2A6 (Di *et al.* 2009), CYP2C9, CYP2C19, CYP3A5 (Bray *et al.* 2010) and probably others. Figure 2 below, extracted from Wang & Tompkins 2008, shows the expression of the different human hepatic CYP and their contribution to metabolize clinically-used drugs. No analog study was performed in breast tissue, and *a fortiori* in breast cancer cell lines. However, the literature reports the presence of CYP3A4 (the CYP enzyme the most involved in drug metabolism) in MCF-7, T47D and MDA-MB-231 (Nagaoka *et al.* 2006; Chen *et al.* 2009; Mitra *et al.* 2011), of CYP2B6 in MCF-7 and T47D (Lo *et al.* 2010) whereas this information is not available for MDA-MB-231. While CYP2D6 and splicing variants similar to those found in breast cancer tissues were shown expressed in MCF-7 (Huang *et al.* 1997), no information about this CYP, to our knowledge, was related for T47D and MDA-MB-231.

Fig. 2. Hepatic CYP expression (A) and their contribution to metabolism of clinically-used drugs (B). From Wang & Tompkins 2008, permission granted by Bentham Science Publishers Ltd.


26 Breast Cancer – Focusing Tumor Microenvironment, Stem Cells and Metastasis

elucidated, Mesna (Sodium 2-mercaptoethanesulfonate) is often used to limit these side

Fig. 1. **First phase of metabolism of the oxazaphosphorines by CYPs**: hydroxylation leads to oxazaphosphorine mustards, and N-dealkylation results in chloracetaldehyde formation.

As already mentioned, several CYPs are involved in these drug metabolism: CYP2B6 (Wang & Tompkins 2008; Mo *et al.* 2009; Bray *et al.* 2010), CYP3A4 (Kivisto *et al.* 1995), but also CYP2A6 (Di *et al.* 2009), CYP2C9, CYP2C19, CYP3A5 (Bray *et al.* 2010) and probably others. Figure 2 below, extracted from Wang & Tompkins 2008, shows the expression of the different human hepatic CYP and their contribution to metabolize clinically-used drugs. No analog study was performed in breast tissue, and *a fortiori* in breast cancer cell lines. However, the literature reports the presence of CYP3A4 (the CYP enzyme the most involved in drug metabolism) in MCF-7, T47D and MDA-MB-231 (Nagaoka *et al.* 2006; Chen *et al.* 2009; Mitra *et al.* 2011), of CYP2B6 in MCF-7 and T47D (Lo *et al.* 2010) whereas this information is not available for MDA-MB-231. While CYP2D6 and splicing variants similar to those found in breast cancer tissues were shown expressed in MCF-7 (Huang *et al.* 1997), no information about this CYP, to our knowledge, was related for T47D and

From Rooseboom *et al.* 2004 with permission from ASPET.

MDA-MB-231.

effects (Giraud *et al.* 2010).

Another example is given by ellipticine. This alkaloid compound found in several plants (Ochrosia, Aspidoserma subincanum, Bleekeria vitiensis) is a topoisomerase poison often used in ovarian and breast cancer treatment. It is also a prodrug whose efficiency depends on CYP activation. 13-hydroxy- and 12-hydroxy-ellipticine, responsible for the formation of DNA adducts, are generated by CYP1A1/2, CYP3A4 and CYP2C9.

Fig. 3. Main pathways of ellipticine metabolism. Reprinted from Stiborova *et al.* 2011, ©2011, with permission from Elsevier.

Members of the CYP1 family are usually expressed in extrahepatic tissues and it is not strange to find CYP1A1 in MCF-7 (Androutsopoulos *et al.* 2009; Stiborova *et al.* 2011), in MDA-MB-231 and T47D (Macpherson & Matthews 2010). We already mentioned the presence of CYP3A4 in the three cell lines, but no precise information is available for CYP2C9.

This slight overview shows that the three main breast cancer cell lines are able to give interesting information about drugs that have to be bioactivated before exerting their deleterious effects in cancer cells. However, we must keep in mind that polymorphic variants of the genes coding these enzymes, or splicing variants, may influence the

*In Vitro* Breast Cancer Models as Useful Tools in Therapeutics? 29

of the metabolic activity showed by the measurement of a decrease in the acidification of the

This technique has advantages and drawbacks. The direct culture of tumour fragment has the major advantage of preserving tissue architecture and all the cell populations constituting the human tumour. This method is thus a valuable technique which permits to take into account the whole tumour environment *in vivo*, allowing the investigation of the role of 3-dimensional structures and stromal interactions in tumour. It also allows to study the response of a particular tumour type to environmental stimulations, drugs, and

However, the culture of tumour samples presents limitations that do not allow its widespread use. Obtaining tumour samples is submitted to ethical constraints relative to the use of patient samples for research. In addition, it must be performed under ideal conditions. Thus, the samples have to be prepared very quickly after their excision, which means that the research laboratory should have particular facilities to have a direct access to fresh tissues. Moreover, the samples excised by the surgeon are becoming smaller and smaller, due to early diagnoses, and the major part of the samples is kept for diagnosis. Then, if some sample is still available for research, priority is given to research on biomarkers of the tumour in order to give personalised therapies, and, only after, it is disposable for fundamental research. Additionally to the availability restrictions, the same sample cannot be used for many tests because of the limitations of growth of this tissue *in vitro*. Repetition of assays and comparative measurements are thus more difficult with this

The use of samples from animal models with mammary tumour partially resolves the problem of availability of samples, but it also raises questions on the representativeness of the samples with human breast tumours. High improvements for providing human tissues

The co-culturing represents another way to circumvent the lack of cell diversity found in cell lines and to allow understanding of the tumoural proliferation mechanisms and intercellular interactions within a tumour. It is an indispensable tool to elucidate the regulation

This model can be used by different ways: co-culturing of two cell types with a direct contact or co-culturing with a separating porous membrane between both cell types. The first method implicates to be able to differentiate the two cell types by microscopy. For that the use of fluorescent markers is a valuable tool (see Figure 4 for an example of co-culture of MDA-MB-231 with hASCs (adipose stem cells) respectively stained by the lipophilic tracers DiI (dialkylindocarbocyanines) and DiO (dialkyloxacarbocyanines), Pinilla *et al.* 2009). The second method allows a relative isolation of the two cell types, the porosity of the membrane separating them allowing the exchange of substances. The two techniques give complementary information on the behavior of cells studied, especially the crucial role of

In example, we could cite the co-culture of MDA-MB-231 and MCF-7, which has highlighted the importance of the heterogeneity of tumours for their growth and the role of oestrogen receptors. In this study, the co-culture of MCF-7 and MDA-MB-231 (respectively ER+ and

medium (Mestres *et al.* 2006).

model.

cytokines under well-defined and reproducible conditions.

of good quality will be brought by the emergence of biobanks.

**2.3 Circumventing the lack of diversity: Co-culturing of cell lines** 

of the tumour by epithelial and stromal components surrounding it.

the inter-cellular communication (Cappelletti *et al*. 1991).

pharmacology of any drugs. Very few information about that are available in patients, but no study was performed in breast cancer cells.

BCCL have been created to study tumour development and related mechanisms and to test molecules potentially active. They are inevitable models for many studies. However, their extensive use in all areas of research on breast cancer remains sometimes controversial due to the over simplicity of the model, the instability of the strain, the existence of "false cell lines" and the failures of representativeness of the tumour. Thus, it clearly appears that these models are not sufficient to answer all the questions on breast cancer, and it is essential to turn to complementary models. Consequently, new models were introduced in the late 70s. They were used to a lesser extent than cell lines for a long time, but they tend to be more used now.

#### **2.2 Improving representativeness of the model: Direct culture of tumour fragment**

There are several methods to circumvent the problem of representativeness of BCCL, e.g. the direct culture of tumour fragments. The first attempts in this direction were made in the late 60s from tumours of 1mm3 volume (Matoska & Stricker 1967). However, these cultures were proven difficult due to the high thickness of the samples, preventing the diffusion of nutrients and oxygen to the center of the sample, and thus, avoiding a long-term cultivation *in vitro*. This method has been modified over time, and with the use of microtome, problems associated with diffusion of nutrients have been resolved. The samples are now constituted of extremely thin slices of about 150 to 200 µM thick (Nissen *et al.* 1983).

This type of model was used to study the different inter-tumoural cell interactions and also to test the sensitivity to drugs (Milani *et al.* 2010). The slice tumour model associated with the development of microscopic analysis methods, such as the triple-fluorescence viability assay developed by Van Der Kuip, allowed the study of the cytotoxic effect of Taxol on this breast cancer model (Van Der Kuip *et al.* 2006).

Another example of drug study is the evaluation of the action of cytokines and cytotoxic drugs on animal (MMTV-Neu mice) breast cancer slices, especially the monitoring of apoptosis increase and DNA damage after treatment with interferon-gamma or doxorubicin (Parajuli & Doppler 2009).

The last noticeable example is the use of a tropism-modified oncolytic adenovirus, and a wild-type adenovirus on these slices to treat breast cancer. The results showed that the modified oncolytic adenovirus can infect and replicate in breast cancer tissue slices, suggesting the great potential of this model for evaluating the potential of oncolytic adenovirus constructs (Pennington *et al.* 2010).

This list is not exhaustive and the literature shows that a lot of results were obtained by the slice culturing method, more particularly on the study of drugs effects like tamoxifen or paclitaxel (Conde *et al.* 2008; Sonnenberg *et al.* 2008; Rajendran *et al.* 2011).

Although used since the late 70's, the slice technique evolved over time and was adapted to technological innovations. We may especially underline the use of silicon sensor chips wearing electrodes and sensors as a carrier of culture slice. The samples are deposited on the chip and data concerning the tumour-slice are analysed continuously during its cultivation and during its contact with drugs; measurements are made in real-time by the readout of ionic-sensitive field effect transistors and an oxygen electrode. This model was used to study the effects of Taxol on 200 slices of breast cancer, which revealed a dose-dependent decrease

pharmacology of any drugs. Very few information about that are available in patients, but

BCCL have been created to study tumour development and related mechanisms and to test molecules potentially active. They are inevitable models for many studies. However, their extensive use in all areas of research on breast cancer remains sometimes controversial due to the over simplicity of the model, the instability of the strain, the existence of "false cell lines" and the failures of representativeness of the tumour. Thus, it clearly appears that these models are not sufficient to answer all the questions on breast cancer, and it is essential to turn to complementary models. Consequently, new models were introduced in the late 70s. They were used to a lesser extent than cell lines for a long time, but they tend to

**2.2 Improving representativeness of the model: Direct culture of tumour fragment**  There are several methods to circumvent the problem of representativeness of BCCL, e.g. the direct culture of tumour fragments. The first attempts in this direction were made in the late 60s from tumours of 1mm3 volume (Matoska & Stricker 1967). However, these cultures were proven difficult due to the high thickness of the samples, preventing the diffusion of nutrients and oxygen to the center of the sample, and thus, avoiding a long-term cultivation *in vitro*. This method has been modified over time, and with the use of microtome, problems associated with diffusion of nutrients have been resolved. The samples are now constituted

This type of model was used to study the different inter-tumoural cell interactions and also to test the sensitivity to drugs (Milani *et al.* 2010). The slice tumour model associated with the development of microscopic analysis methods, such as the triple-fluorescence viability assay developed by Van Der Kuip, allowed the study of the cytotoxic effect of Taxol on this

Another example of drug study is the evaluation of the action of cytokines and cytotoxic drugs on animal (MMTV-Neu mice) breast cancer slices, especially the monitoring of apoptosis increase and DNA damage after treatment with interferon-gamma or doxorubicin

The last noticeable example is the use of a tropism-modified oncolytic adenovirus, and a wild-type adenovirus on these slices to treat breast cancer. The results showed that the modified oncolytic adenovirus can infect and replicate in breast cancer tissue slices, suggesting the great potential of this model for evaluating the potential of oncolytic

This list is not exhaustive and the literature shows that a lot of results were obtained by the slice culturing method, more particularly on the study of drugs effects like tamoxifen or

Although used since the late 70's, the slice technique evolved over time and was adapted to technological innovations. We may especially underline the use of silicon sensor chips wearing electrodes and sensors as a carrier of culture slice. The samples are deposited on the chip and data concerning the tumour-slice are analysed continuously during its cultivation and during its contact with drugs; measurements are made in real-time by the readout of ionic-sensitive field effect transistors and an oxygen electrode. This model was used to study the effects of Taxol on 200 slices of breast cancer, which revealed a dose-dependent decrease

paclitaxel (Conde *et al.* 2008; Sonnenberg *et al.* 2008; Rajendran *et al.* 2011).

of extremely thin slices of about 150 to 200 µM thick (Nissen *et al.* 1983).

no study was performed in breast cancer cells.

breast cancer model (Van Der Kuip *et al.* 2006).

adenovirus constructs (Pennington *et al.* 2010).

(Parajuli & Doppler 2009).

be more used now.

of the metabolic activity showed by the measurement of a decrease in the acidification of the medium (Mestres *et al.* 2006).

This technique has advantages and drawbacks. The direct culture of tumour fragment has the major advantage of preserving tissue architecture and all the cell populations constituting the human tumour. This method is thus a valuable technique which permits to take into account the whole tumour environment *in vivo*, allowing the investigation of the role of 3-dimensional structures and stromal interactions in tumour. It also allows to study the response of a particular tumour type to environmental stimulations, drugs, and cytokines under well-defined and reproducible conditions.

However, the culture of tumour samples presents limitations that do not allow its widespread use. Obtaining tumour samples is submitted to ethical constraints relative to the use of patient samples for research. In addition, it must be performed under ideal conditions. Thus, the samples have to be prepared very quickly after their excision, which means that the research laboratory should have particular facilities to have a direct access to fresh tissues. Moreover, the samples excised by the surgeon are becoming smaller and smaller, due to early diagnoses, and the major part of the samples is kept for diagnosis. Then, if some sample is still available for research, priority is given to research on biomarkers of the tumour in order to give personalised therapies, and, only after, it is disposable for fundamental research. Additionally to the availability restrictions, the same sample cannot be used for many tests because of the limitations of growth of this tissue *in vitro*. Repetition of assays and comparative measurements are thus more difficult with this model.

The use of samples from animal models with mammary tumour partially resolves the problem of availability of samples, but it also raises questions on the representativeness of the samples with human breast tumours. High improvements for providing human tissues of good quality will be brought by the emergence of biobanks.

#### **2.3 Circumventing the lack of diversity: Co-culturing of cell lines**

The co-culturing represents another way to circumvent the lack of cell diversity found in cell lines and to allow understanding of the tumoural proliferation mechanisms and intercellular interactions within a tumour. It is an indispensable tool to elucidate the regulation of the tumour by epithelial and stromal components surrounding it.

This model can be used by different ways: co-culturing of two cell types with a direct contact or co-culturing with a separating porous membrane between both cell types. The first method implicates to be able to differentiate the two cell types by microscopy. For that the use of fluorescent markers is a valuable tool (see Figure 4 for an example of co-culture of MDA-MB-231 with hASCs (adipose stem cells) respectively stained by the lipophilic tracers DiI (dialkylindocarbocyanines) and DiO (dialkyloxacarbocyanines), Pinilla *et al.* 2009).

The second method allows a relative isolation of the two cell types, the porosity of the membrane separating them allowing the exchange of substances. The two techniques give complementary information on the behavior of cells studied, especially the crucial role of the inter-cellular communication (Cappelletti *et al*. 1991).

In example, we could cite the co-culture of MDA-MB-231 and MCF-7, which has highlighted the importance of the heterogeneity of tumours for their growth and the role of oestrogen receptors. In this study, the co-culture of MCF-7 and MDA-MB-231 (respectively ER+ and

*In Vitro* Breast Cancer Models as Useful Tools in Therapeutics? 31

These repeatable techniques have permitted to highlight the regulation of mammary tumours by the surrounding stroma and the complex interactions between the cell subtypes

Another particular model allows cells to grow in 3-dimensions, generally with a matrix support (Yuhas *et al.* 1978). This type of culture permits an *in vitro* depiction of tumour tissue more accurate than classical 2-dimensional cultures in monolayers, as this last model does not correctly imitate the architecture and cellular gradients of oxygen and nutrients

Only few cell lines are spontaneously able to establish spheroid architectures under certain culture conditions, but most of the systems require the use of synthetic or non-synthetic matrix. Systems are most often made of agar matrices or collagen support (Kim *et al.* 2004b). The Figure 5 show the growth of a MCF-7 spheroid growth in a hydrogel agarose matrix

Fig. 5. Growth of a MCF-7 tumour spheroid in agarose hydrogel. The pictures represent the spheroid at 2 days old (a), 11 days old (b) and 27 days old (c) (the scale bar represent 50 μm). Reprinted by permission from Macmillan Publishers Ltd: Nature Physics, Fritsch *et al.* 2010,

Co-culturing of multiple cell types on these 3-dimensional systems is often used to study the relationship between cells, while simulating the tumour architecture with the most fidelity. These systems generally implicate the cultivation of tumour cells with other cell types like stromal, endothelial, fibroblasts and immune-competent cells. Moreover, this type of model, structurally like-looking the tumour, can be used quite indefinitely because it relies on the use of immortalised cells lines. This allows circumventing the problem of the lack of

More advanced systems have been derived from this principle; one can cite the microfluidicbased 3-dimensional culturing (Bauer *et al.* 2010) that allows to grow multicellular tumour spheroid on a microchannel support, in order to analyze complex and heterotypic cellular interactions between breast cancer cells and fibroblast from the surrounding stroma. It has many advantages compared to the standard 3D culture: the culture volume and the number of needed cells are smaller than in standard support, the molecules are only distributed by

samples which is the major drawback of the tumour fragment culturing.

diffusion mechanisms and the model is adapted to high throughput screenings.

**2.4 A model with a tumour-like structure and cell diversity: 3-D culture** 

that are found in poorly vascularised regions of the tumour.

of the tumour.

system (Fritsch *et al.* 2010).

©2010. http://www.nature.com/nphys.

ER-) in a membrane separation system, was characterised by an increase of the MCF-7 cells growth rate in comparison with monocultures. This suggests that complex interactions between heterogenous cells population in tumour could explain the variability in tumour progression between different patients and the failure in response to endocrine treatment for some patients with ER+ tumours.

Fig. 4. Human stem cells derived from adipose tissue (hASCs) and breast cancer cells (MDA-MB-231) cultured in a monolayer co-culture system. (a) Direct microscopic observation of the co-culture of MDA-MB-231 and hASCs cells. (b) Overlay of DiO (hASCs), DiI (MDA-MB-231) and DAPI (nucleus) stainings. (c) DiO staining of hASCs derived stem cells (green). (d) DiI staining of MDA-MB-231 breast cancer cells (red). Reprinted from Pinilla *et al.* 2009, ©2009, with permission from Elsevier.

Another example concerns the direct co-culturing of MCF-10A, a non-cancerous breast cell line, with the cancerous one MCF-7. An exposure to hormonal treatment with 17βestradiol was able to inhibit the proliferation of MCF-7 cells in this co-culture, whereas this phenomenon was not observed in a monoculture of MCF-7. This highlighted the complex interactions between ER+ MCF-7 and ER- MCF-10A cells which may reect physiologically relevant mechanisms of the paracrine regulation of cell proliferation (Spink *et al*. 2006).

The co-culture of MCF-7 with fibroblasts derived from normal biopsies or from cancer biopsies also allowed to highlight the crucial role of fibroblasts in breast tumours. The results of two studies, one in direct co-culturing (Samoszuk *et al*. 2005) the second in membrane separated system (Dong-Le Bourhis *et al.* 1997), showed that MCF-7 growth rate was inhibited by fibroblasts issued from non cancerous tissues, but not by fibroblasts issued from tumourous tissues or serum-activated fibroblasts which enhanced MCF-7 growth rate. This suggests that fibroblasts could release some tumour growth inhibiting or activating factors.

The role of tumour-associated macrophages in the proliferation of tumour cells was also studied by co-culturing macrophages with MCF-7 cells in a membrane separated system. This co-culture lead to a significant increase of MCF-7 invasiveness *in vitro* (Hagemann *et al.* 2004).

ER-) in a membrane separation system, was characterised by an increase of the MCF-7 cells growth rate in comparison with monocultures. This suggests that complex interactions between heterogenous cells population in tumour could explain the variability in tumour progression between different patients and the failure in response to endocrine treatment

Fig. 4. Human stem cells derived from adipose tissue (hASCs) and breast cancer cells (MDA-MB-231) cultured in a monolayer co-culture system. (a) Direct microscopic observation of the co-culture of MDA-MB-231 and hASCs cells. (b) Overlay of DiO (hASCs), DiI (MDA-MB-231) and DAPI (nucleus) stainings. (c) DiO staining of hASCs derived stem cells (green). (d) DiI staining of MDA-MB-231 breast cancer cells (red). Reprinted from Pinilla *et al.* 2009,

Another example concerns the direct co-culturing of MCF-10A, a non-cancerous breast cell line, with the cancerous one MCF-7. An exposure to hormonal treatment with 17βestradiol was able to inhibit the proliferation of MCF-7 cells in this co-culture, whereas this phenomenon was not observed in a monoculture of MCF-7. This highlighted the complex interactions between ER+ MCF-7 and ER- MCF-10A cells which may reect physiologically relevant mechanisms of the paracrine regulation of cell proliferation

The co-culture of MCF-7 with fibroblasts derived from normal biopsies or from cancer biopsies also allowed to highlight the crucial role of fibroblasts in breast tumours. The results of two studies, one in direct co-culturing (Samoszuk *et al*. 2005) the second in membrane separated system (Dong-Le Bourhis *et al.* 1997), showed that MCF-7 growth rate was inhibited by fibroblasts issued from non cancerous tissues, but not by fibroblasts issued from tumourous tissues or serum-activated fibroblasts which enhanced MCF-7 growth rate. This suggests that fibroblasts could release some tumour growth inhibiting or activating

The role of tumour-associated macrophages in the proliferation of tumour cells was also studied by co-culturing macrophages with MCF-7 cells in a membrane separated system. This co-culture lead to a significant increase of MCF-7 invasiveness *in vitro* (Hagemann *et al.* 2004).

for some patients with ER+ tumours.

©2009, with permission from Elsevier.

(Spink *et al*. 2006).

factors.

These repeatable techniques have permitted to highlight the regulation of mammary tumours by the surrounding stroma and the complex interactions between the cell subtypes of the tumour.

### **2.4 A model with a tumour-like structure and cell diversity: 3-D culture**

Another particular model allows cells to grow in 3-dimensions, generally with a matrix support (Yuhas *et al.* 1978). This type of culture permits an *in vitro* depiction of tumour tissue more accurate than classical 2-dimensional cultures in monolayers, as this last model does not correctly imitate the architecture and cellular gradients of oxygen and nutrients that are found in poorly vascularised regions of the tumour.

Only few cell lines are spontaneously able to establish spheroid architectures under certain culture conditions, but most of the systems require the use of synthetic or non-synthetic matrix. Systems are most often made of agar matrices or collagen support (Kim *et al.* 2004b). The Figure 5 show the growth of a MCF-7 spheroid growth in a hydrogel agarose matrix system (Fritsch *et al.* 2010).

Fig. 5. Growth of a MCF-7 tumour spheroid in agarose hydrogel. The pictures represent the spheroid at 2 days old (a), 11 days old (b) and 27 days old (c) (the scale bar represent 50 μm). Reprinted by permission from Macmillan Publishers Ltd: Nature Physics, Fritsch *et al.* 2010, ©2010. http://www.nature.com/nphys.

Co-culturing of multiple cell types on these 3-dimensional systems is often used to study the relationship between cells, while simulating the tumour architecture with the most fidelity. These systems generally implicate the cultivation of tumour cells with other cell types like stromal, endothelial, fibroblasts and immune-competent cells. Moreover, this type of model, structurally like-looking the tumour, can be used quite indefinitely because it relies on the use of immortalised cells lines. This allows circumventing the problem of the lack of samples which is the major drawback of the tumour fragment culturing.

More advanced systems have been derived from this principle; one can cite the microfluidicbased 3-dimensional culturing (Bauer *et al.* 2010) that allows to grow multicellular tumour spheroid on a microchannel support, in order to analyze complex and heterotypic cellular interactions between breast cancer cells and fibroblast from the surrounding stroma. It has many advantages compared to the standard 3D culture: the culture volume and the number of needed cells are smaller than in standard support, the molecules are only distributed by diffusion mechanisms and the model is adapted to high throughput screenings.

*In Vitro* Breast Cancer Models as Useful Tools in Therapeutics? 33

Coxsackievirus 21 (CVA21) was intravenously injected in SCID mice xenografted with MDA-MB-231 breast cancer cells. CVA21 virus targets the receptors ICAM-1 and DAF that are overexpressed in breast cancer cells. In this experiment a rapid lysis focused on cancer cells was observed in all mice, making this virus a good candidate for use in systemic therapy (Skelding *et al.* 2009). See Figure 7 illustrating the effect of the virus on xenografted

Fig. 7. **Observation of the oncolytic activity of CVA21 virus in SCID mouse xenografted with MDA-MB-231-luc**. The breast cancer cells were xenografted into the mammary fat pad, mice were then treated with PBS or CVA21. Metastases were detected 3 weeks post-cell injection. The mice on the pictures are representative for bioluminescent observation at day 42 post treatment. From Skelding *et al*. 2009, with kind permission from Springer Science

In the investigation of new treatments, the vitamin D3 receptors constitute good targets as they are present in over 80% of mammary tumours and they are negative growth regulator of both oestrogen-dependent and independent breast cancer cells *in vitro*. In a study published in 1998 it was shown that EB1089, a vitamin D3 analog, was able to highly reduce the growth of tumour in nude mice xenografted with MCF-7 cells (tumours were 4-fold smaller than those in untreated mice). This reduction was resulting from an enhancement of apoptosis and reducing proliferation of tumour epithelial cells, suggesting the great potential of vitamin D3 analogs

This model can also be used to explore new potential targets for anticancer therapies. A good example is the targeting of receptor ERβ. In an experiment, standard T47D ERα+ ERβand modified T47D ERα+ ERβ+ (T47D stably transfected with a plasmid allowing the expression of the receptor ERβ), were xenografted in SCID mice. 17β-estradiol was then injected into mice. The treatment triggered an acceleration of tumour growth in mice xenografted with the native T47D strain, and conversely a regression of tumours T47D ERβ+. These results emphasize the antagonistic role of ER receptors that appear to play an antitumourigenic role, and offered prospects for the development of ER-selective inhibitors.

The targets cited above are non exhaustive. Many other therapeutic targets are tested with xenografts models, as it is the case of the VEGF pathway implicated in tumour angiogenesis (Le *et al.* 2008), or of cell cycle regulating proteins such as CDK kinases (Fry *et al.* 2004). The use of established cell lines for producing xenografts raises several questions about their relevance. The murine model presents considerable differences with the human body, concerning the biochemical and physiological regulation. Moreover, the stroma that will grow surround the tumour will be of murine origin and it will result in a chimeric tumour

such as EB1089 against human breast cancer (VanWeelden *et al.* 1998).

mice, visualised by bioluminescent analysis.

and Business Media B.V.

(Hartman *et al.* 2006).
