**3. The establishment of human breast cancer cell lines**

The first human breast cancer cell line, BT-20, was established by Lasfargues and Ozzello in 1958 from an explant culture of a tumor slice from a 74 year old caucasian woman (Lasfargues and Ozzello, 1958). These cells are estrogen receptor alpha (ER) negative, progesterone receptor (PR) negative, Tumor Necrosis Factor alpha (TNF-α) positive, and epidermal growth factor receptor (EGFR) positive (Borras et al., 1997). While BT-20 is the oldest established breast cancer cell line, it is not the most commonly used line. By far, the most widely used breast cancer cell line worldwide is the MCF-7 cell line (Table 1 and Figure 1)(Burdall et al., 2003). Established in 1973 by Soule and colleagues at the Michigan Cancer Foundation, from where it derives its name, MCF-7 cells were isolated from the plural effusion of a 69 year old woman with metastatic disease (Soule et al., 1973). Since its establishment, MCF7 has become the model of ER positive breast cancer (Lacroix and Laclercq, 2004). Establishment of other cell lines has followed, including ones from other breast cancer types such as BRCA mutant, triple negative, HER2 overexpressing, and those derived from normal mammary epithelial cells such as MCF-10A cells (Soule et al., 1990) (Table 2).

Cell line use in labs is ubiquitous and continues to increase. From 2000 - 2010, the publication of manuscripts using the 10 most commonly used cell lines has almost tripled (2.8% increase) (Figure 2). Clearly demonstrating that the importance of, need for, and use of breast cancer cell lines will not diminish in the near future. Evaluation of the existing lines indicates that most breast cancer cell lines in use are derived from metastatic cancer and not other breast cancer phenotypes (Borras et al., 1997). Indeed, the overall success rate of establishing a cell line is only 10%. Most of the cell lines that exist today have been derived from pleural effusion instead of from primary tumors and are primarily ER - lines (Table 2 and reviewed in (Lacroix and Laclercq, 2004). This is surprising since ER - breast cancer is detected in only 20 - 30% of all primary tumors, whereas ER + tumors are detected 55-60% of the time (Ali and Coombes, 2000; McGuire et al., 1978). The reason for this discrepancy remains unknown, however it has been postulated that this could be because ER - cells are easier to establish in culture than ER + or that as cells are grown in culture, the epithelial like phenotype is lost while more mesenchymal traits are retained, therefore cells in culture appear to undergo a endothelial to mesenchymal transition (EMT) *in vitro* which is associated with the ER - phenotype (Lacroix and Laclercq, 2004). This suggests that culture systems are a model of metastatic disease that can grow in isolation and not a model the

transient. Primary culture refers to the period of time the primary tissue/organ fragment is kept in culture *in vitro* prior to the first passage or subculturing of cells, at which time they are referred to as a cell culture. This could range from days to a few weeks at most

Cell lines are primary cultures that have been subcultured or passaged and can be clonal, terminal or immortalized cells (McAteer and Davis, 2002). Clonal cell cultures are created by selecting a single cell that will proliferate to establish a single population. Terminal cell lines are able to grow in culture for a few generations before senescence occurs and the cell line can no longer survive in culture media. Immortalized cell lines are able to grow in culture forever. These immortalized cell lines can occur naturally, such as HeLa cells, or through transformation events, such as Epstein-Barr Virus transformation. All types of *in vitro* cell

The first human breast cancer cell line, BT-20, was established by Lasfargues and Ozzello in 1958 from an explant culture of a tumor slice from a 74 year old caucasian woman (Lasfargues and Ozzello, 1958). These cells are estrogen receptor alpha (ER) negative, progesterone receptor (PR) negative, Tumor Necrosis Factor alpha (TNF-α) positive, and epidermal growth factor receptor (EGFR) positive (Borras et al., 1997). While BT-20 is the oldest established breast cancer cell line, it is not the most commonly used line. By far, the most widely used breast cancer cell line worldwide is the MCF-7 cell line (Table 1 and Figure 1)(Burdall et al., 2003). Established in 1973 by Soule and colleagues at the Michigan Cancer Foundation, from where it derives its name, MCF-7 cells were isolated from the plural effusion of a 69 year old woman with metastatic disease (Soule et al., 1973). Since its establishment, MCF7 has become the model of ER positive breast cancer (Lacroix and Laclercq, 2004). Establishment of other cell lines has followed, including ones from other breast cancer types such as BRCA mutant, triple negative, HER2 overexpressing, and those derived from normal mammary epithelial cells such as MCF-10A cells (Soule et al., 1990)

Cell line use in labs is ubiquitous and continues to increase. From 2000 - 2010, the publication of manuscripts using the 10 most commonly used cell lines has almost tripled (2.8% increase) (Figure 2). Clearly demonstrating that the importance of, need for, and use of breast cancer cell lines will not diminish in the near future. Evaluation of the existing lines indicates that most breast cancer cell lines in use are derived from metastatic cancer and not other breast cancer phenotypes (Borras et al., 1997). Indeed, the overall success rate of establishing a cell line is only 10%. Most of the cell lines that exist today have been derived from pleural effusion instead of from primary tumors and are primarily ER - lines (Table 2 and reviewed in (Lacroix and Laclercq, 2004). This is surprising since ER - breast cancer is detected in only 20 - 30% of all primary tumors, whereas ER + tumors are detected 55-60% of the time (Ali and Coombes, 2000; McGuire et al., 1978). The reason for this discrepancy remains unknown, however it has been postulated that this could be because ER - cells are easier to establish in culture than ER + or that as cells are grown in culture, the epithelial like phenotype is lost while more mesenchymal traits are retained, therefore cells in culture appear to undergo a endothelial to mesenchymal transition (EMT) *in vitro* which is associated with the ER - phenotype (Lacroix and Laclercq, 2004). This suggests that culture systems are a model of metastatic disease that can grow in isolation and not a model the

(MacDonald, 2002).

(Table 2).

cultures are used in breast cancer research.

**3. The establishment of human breast cancer cell lines** 

wide heterogeneity of disease that is detected clinically. Although current cell lines are derived form only a subset of primary cancers, overall these lines are a reliable model to study the fundamental questions concerning cell growth, death, and the basic biology of breast cancer. Indeed, many advances in breast cancer biology have been made using cell culture systems and should not be dismissed because of these concerns.


Table 1. List of commonly used cell lines, the number of citations and their origin

#### **3.1 Breast cancer cell lines as models of primary tumors**

Using breast cancer cell lines clearly hold advantages over use of animal or human models. Beyond the ethical implications of animal or human use, the advantages to using cell lines include the ease of obtaining cell lines (can be purchased from commercial sources), the ease of harvesting large numbers of cells (can be grown in culture for long periods of time to accumulate the necessary concentration), and the ability to test an individual cell type without confounding parameters such as other cell types or local microenvironment (to date, no two cell lines can grown simultaneously in culture for extended periods). Conversely, much debate has circulated concerning the applicability of the data derived from isolated cell lines to the predicted outcomes in humans. One area that this debate has been most contentious has been regarding the importance of the immune system in cancer development. Clearly, the microenvironment and infiltrating immune cells contribute to development and progression of disease, therefore individual cells grown in isolation will lack the influence of other neighboring cells (Voskoglou-Nomikos et al., 2003). Genetic, epigenetic and cytotoxicity studies that focus on outcomes in breast cells clearly benefit from use of cell culture systems. The fundamental understanding of the underlying genetic or molecular pathways involved in breast cell growth and its response to cytotoxic agents are best understood in isolated cell culture systems (Voskoglou-Nomikos et al., 2003).

Breast Cancer Cell Line Development and Authentication 9

Fig. 2. The total number of publications using breast cancer cell lines from 2000 through

Fig. 3. Number and percent of papers published using MDA-MB-435 cells from 2000 - 2010. The tumor type that gave rise to MDA-MB-435 cells has been controversial since 2000. In 2004, STR profiling confirmed that MDA-MB-435 was not a breast cell line but rather has been contaminated with the M4 melanoma cell line. There has been a subsequent drop in the use and publication of these cells. Shown is the total number of papers published using MDA-MB-435 cells (green bars) and the percent of the total number of publications use MDA-MB-435 cells (blue circles). Arrow denotes when MDA-MB-435 were identified as

M14 melanoma cells.

2010. Use of breast cancer cell lines has steadily been rising since 2000.

Fig. 1. The total number of publications per breast cancer cell line from 2000 through 2010. The most commonly used cell line is the ER+ MCF7 cell line, followed by ER - MDA-MB-231 cell lines. Many other cell lines are in use, however the number of publications using these models is quite small. A. Total number of publications using breast cancer cell lines. B. Each breast cancer cell line as a percentage of the total breast cancer cell lines used per year.

Fig. 1. The total number of publications per breast cancer cell line from 2000 through 2010. The most commonly used cell line is the ER+ MCF7 cell line, followed by ER - MDA-MB-231 cell lines. Many other cell lines are in use, however the number of publications using these models is quite small. A. Total number of publications using breast cancer cell lines. B. Each breast cancer cell line as a percentage of the total breast cancer cell lines used per

year.

Fig. 2. The total number of publications using breast cancer cell lines from 2000 through 2010. Use of breast cancer cell lines has steadily been rising since 2000.

Breast Cancer Cell Line Development and Authentication 11

HCC1395 1994 primary tissue -/- HCC1419 1994 primary tissue -/- HCC1428 1995 pleural effusion ? HCC1500 1995 primary tissue +/+ HCC1569 1995 primary tissue -/- HCC1806 1995 primary tissue -/- HCC1937 1995 primary tissue -/- HCC1954 1995 primary tissue +/+ HCC2157 1995 primary tissue -/+ HCC2158 1996 primary tissue -/? HCC1599 1998 primary tissue -/- AU565 1998 pleural effusion ?

Table 2. Commercially available cell lines, their establishment date, and hormonal receptor

Debate has also centered on whether cell lines grown in culture maintain the same genotypic/phenotypic changes that are detected in the primary tissues from which they are derived. Characterization of breast cancer cell lines has been ongoing since their establishment in 1958. In general, breast cancer cell lines are representative models of the primary breast tumors they are derived from (Kao et al., 2009). Initial characterization including karyotyping and comparative genomic hybridization (CGH) demonstrate that, when created and propagated in culture, cell lines maintain the same mutations and chromosomal abnormalities as their primary tumor samples (Lacroix and Laclercq, 2004). While new mutations and chromosomal instability develop in cultured cell lines, overall the genotype remains generally consistent between primary cells and cell lines (Lacroix and Laclercq, 2004). Due to differences in the *in vitro* environment, lack of surrounding naturally occurring microenvironment, and selection pressures, differentiation in culture can occur (Kao et al., 2009; Lacroix and Laclercq, 2004; Voskoglou-Nomikos et al., 2003). Because cancer cells are inherently unstable, differences between same cell line grown in different labs under different environments, even if the growth conditions are the same, are evident (Lacroix and Laclercq, 2004; Osborne et al., 1987). This impacts experimentation as data derived from one lab may not be reproducible in another lab, even is using the same cell line. Caution must be taken when relying on one or two cell

Use of more modern molecular techniques to characterize cell lines has revealed that while differences between primary cells and cell lines do exist. These techniques do confirm, however, that cell lines maintain the molecular distinction found the primary tumors. Gene expression changes detected in primary tumors are not dramatically different to those found in culture systems, even when cultures are grown directly on plastic in 2D cultures or in reconstituted 3D cultures (Vargo-Gogola and Rosen, 2007). Direct comparison of primary tissue to cultured cells revealed "close similarities" between molecular profiles (Dairkee et al., 2004). Indeed, even epigenetic changes found in primary cancers are similarly detected

**established origin ER/PR** 

**status** 

**cell line year** 

status

lines to draw conclusions.


BT-20 1958 primary tissue -/? SK-Br-3 1970 pleural effusion +/+ SW13 1971 ? ? MDA-MB-134-VI 1973 pleural effusion +/- MDA-MB-157 1973 pleural effusion ? MDA-MB-175-VII 1973 pleural effusion ? MDA-MB-231 1973 pleural effusion -/- MDA-MB-361 1973 brain metastasis ? MDA-MB-330 1973 pleural effusion ? MDA-MB-415 1973 pleural effusion ? MDA-MB-436 1973 pleural effusion ? MDA-MB-453 1973 pleural effusion -/- MDA-MB-468 1973 pleural effusion -/- MDA-MB-157 1974 pleural effusion ? MCF7 1974 primary tissue +/+ CAMA-1 1975 pleural effusion ? SW527 1977 ? ? Hs578Bst 1977 non-tumorigenic breast tissue -/- Hs578T 1977 primary tissue -/- ZR-75-1 1978 ascites +/+ ZR-75-30 1978 ascites ? BT483 1978 primary tissue ? DU4475 1979 primary tissue ? T47D 1979 pleural effusion +/+ MCF10A 1984 non-tumorigenic breast tissue -/- MCF10F 1984 non-tumorigenic breast tissue -/- MCF10-2A 1984 non-tumorigenic breast tissue -/- 184A1 1985 normal mammoplasty (transformed) ? 184B5 1985 normal mammoplasty (transformed) ? UACC-812 1986 primary tissue -/- UACC-893 1987 primary tissue -/- HCC38 1992 primary tissue -/- HCC70 1992 primary tissue -/- HCC202 1992 primary tissue -/- HCC1008 1994 lymph node -/- HCC1143 1994 primary tissue -/- HCC1187 1994 primary tissue ?/-

**established origin ER/PR** 

**status** 

**cell line year** 


Table 2. Commercially available cell lines, their establishment date, and hormonal receptor status

Debate has also centered on whether cell lines grown in culture maintain the same genotypic/phenotypic changes that are detected in the primary tissues from which they are derived. Characterization of breast cancer cell lines has been ongoing since their establishment in 1958. In general, breast cancer cell lines are representative models of the primary breast tumors they are derived from (Kao et al., 2009). Initial characterization including karyotyping and comparative genomic hybridization (CGH) demonstrate that, when created and propagated in culture, cell lines maintain the same mutations and chromosomal abnormalities as their primary tumor samples (Lacroix and Laclercq, 2004). While new mutations and chromosomal instability develop in cultured cell lines, overall the genotype remains generally consistent between primary cells and cell lines (Lacroix and Laclercq, 2004). Due to differences in the *in vitro* environment, lack of surrounding naturally occurring microenvironment, and selection pressures, differentiation in culture can occur (Kao et al., 2009; Lacroix and Laclercq, 2004; Voskoglou-Nomikos et al., 2003). Because cancer cells are inherently unstable, differences between same cell line grown in different labs under different environments, even if the growth conditions are the same, are evident (Lacroix and Laclercq, 2004; Osborne et al., 1987). This impacts experimentation as data derived from one lab may not be reproducible in another lab, even is using the same cell line. Caution must be taken when relying on one or two cell lines to draw conclusions.

Use of more modern molecular techniques to characterize cell lines has revealed that while differences between primary cells and cell lines do exist. These techniques do confirm, however, that cell lines maintain the molecular distinction found the primary tumors. Gene expression changes detected in primary tumors are not dramatically different to those found in culture systems, even when cultures are grown directly on plastic in 2D cultures or in reconstituted 3D cultures (Vargo-Gogola and Rosen, 2007). Direct comparison of primary tissue to cultured cells revealed "close similarities" between molecular profiles (Dairkee et al., 2004). Indeed, even epigenetic changes found in primary cancers are similarly detected

Breast Cancer Cell Line Development and Authentication 13

contained all ER + cell lines, both Basal A and B consisted of all ER - cell lines. HER2+ cell lines were grouped into the luminal. Basal A contained the HCC cells and BRCA1 mutant cells, whereas basal B genotype contained non-tumorigenic lines including MCF10A cells

Cell lines are merely a model of breast disease that aim to provide clinical predictability of outcomes in humans. To directly test the applicability of breast cancer cell lines, xenograft cancer models, and mouse breast cancer models to clinical outcome, Voskoglou-Nomikos et. al. compared outcomes *in vitro* to those in xenograft models, to mouse models and phase II clinical trails (Voskoglou-Nomikos et al., 2003). In these comparisons, a general correlation between relative risk (predictive value of a drug in cell line) and the phase II human trial (tumor/control ratio) existed for *in vitro* cell lines. A general predictive value when using xenograft models to predict outcome to chemotherapy was detected, however this was dependent on the drug tested and the grade/type of tumor analyzed (Voskoglou-Nomikos et al., 2003). Overall, Vaskoglou-Nomikos et. al. concluded that cell lines and xenograft models were good predictors of clinical phase II trial outcomes, but are reliable predictors only when testing cytotoxic drugs and when using the correct model system. These models generally were not predictive of human outcomes when testing non-cytotoxic drugs (Voskoglou-Nomikos et al., 2003). Taken together, these studies emphasize the critical need to establish more breast cancer cell lines that model the heterogeneity of breast cancer and to employ many *in vitro* and xenograft model systems using multiple cell lines per experiment

Overt contamination of cell lines, such as bacterial, fungal or yeast infections, is readily detectable merely by altered appearance of the culture and can be rectified without impacting the quality or reproducibility of the data. Less overt contamination, such as mycoplasma and cell line cross-contamination, can occur undetected and can seriously jeopardize experimental findings. While it is well recognized that periodic testing for mycoplasma is a necessary requirement when using cell lines, cross-contamination with other cell lines is less recognized as a problem and therefore and cell authentication

Cell line cross-contamination is most evident in the case of MDA-MB-435 cells. When Ross et. al. published the molecular profiles of breast cancer cell lines in 2000, the MDA-MB-435 cell line consistently fell outside the range of profiles of the other breast cancer cell lines and clustered with melanoma cell lines (Ross et al., 2000). This sparked great debate about the authenticity of the this line. Derived in 1976 from the pleural effusion of a 31 year old patient with metastatic adenocarcinoma of the breast, initial debate suggested that this was still a breast cancer cell line, but had been derived from a patient who may have also had undiagnosed melanoma (Cailleau et al., 1978). Data indicating that MDA-MB-435 cells expressed a mixture of both melanoma and epithelial markers fueled this debate, however the overwhelming belief was the these were indeed breast cancer cells (Chambers, 2009; Sellappan et al., 2004)(Figures 2 and 3). Indeed, early characterization of the cell line indicated that they were highly metastatic and secrete milk proteins, findings consistent with those of breast cancer cells (Howlett et al., 1994; Price, 1996; Price et al., 1990; Price and Zhang, 1990; Sellappan et al., 2004; Suzuki et al., 2006; Welch, 1997). Confusingly, MDA-MB-435 cells also expressed the melanocyte markers tyrosinase, melan A and S100 (Ellison et al.,

(Kao et al., 2009). This highlights that breast cancer cell lines are a model of disease.

to reliably predict clinical outcome.

**4. Contamination** 

practices are not routine.

in cell lines (Lacroix and Laclercq, 2004). This suggests that cell lines are an appropriate model of primary disease and, depending on the research focus, cell lines will faithfully reflect the processes of primary tissues.

Since cell lines generally remain faithful in terms of the molecular and genetic profiles of the primary tumor from which they are derived, it is critical to consider the correct model system. While ER/PR status of primary tumors leans predominantly toward ER+ expression (55-60%), most breast cancer cell lines have been derived from ER - tumors or pleural effusions (McGuire et al., 1978)(Table 2). Therefore it is of utmost importance to select the proper model to answer the experimental question. A detailed analysis of the applicability of cell lines to accurately model primary breast tumors revealed that overall breast cancer cell lines as a whole do model primary tumors, however on an individual basis, one specific cell line does not accurately mirror a primary breast tumor, even with the same gene expression profile. Since variability in cell lines exist, it is generally thought that to more accurately predict outcomes in primary tissue, a panel of breast cancer cell lines rather than just 1 or 2 individual lines should be tested. Using panels more accurately reflects primary breast tumors and will help translate findings from *in vitro* studies to *in vivo* therapeutic options (Dairkee et al., 2004).

Microarray analysis clearly defined primary breast tumors and breast cancer cell lines at the genetic level. Perou and others have conducted detailed studies using microarray platforms and determined a molecular signature of gene expression changes found in primary breast cancer tumors (Alizadeh et al., 2001; Perou et al., 1999b; Perou et al., 2000b; Ross et al., 2000; Sorlie et al., 2001). These signatures are used to understand the molecular basis of breast cancer and to define different subtypes of cancer that occur naturally in humans. It was also developed as a diagnostic tool to detect breast cancer tumors earlier and to facilitate proper treatment based on a gene signature. Based on these studies, 5 molecular signatures and types of primary breast tumors have been identified. These are luminal A, luminal B, basallike, HER2+, and normal-like profiles (Perou et al., 1999a; Perou et al., 2000a; Ross et al., 2000; Sorlie et al., 2001). Prior to establishment of these molecular signatures, diagnosis was determined by receptor expression status, i.e. ER/PR/HER2, and treatment regimes assigned accordingly. Using this molecular approach, luminal A and luminal B tend to also be ER + expressing tumors, basal-like encompasses ER - tumors, HER2+ incorporate those HER2+ expressing tumors, and normal-like have similar expression patterns to noncancerous cells (Perou et al., 1999a; Perou et al., 2000a; Ross et al., 2000; Sorlie et al., 2001). Such molecular characterization will lead to providing more personalized therapy to patients. Efficacy of drugs in different subtypes will be easily determined and accurately assigned to patients expressing a similar molecular profile. While such personalized medicine may be still in the future, some current breast cancer treatment options that exist today are based on the molecular profile of the tumor. For example, tumors expressing the estrogen receptor are treated with selective estrogen receptor modulator (SERM) or other similar anti-estrogen compound whereas tumors lacking ER do not receive the same therapy. Similarly, HER2+ tumors are susceptible to trastuzumab because of HER2 expression. In the future as molecular characterization improves and new chemotherapeutics are developed, more personalized options will be available.

Do cell lines reflect the molecular signature of primary tumors? In a direct comparison of the molecular profiles from cell lines and primary tumors, Kao et. al. found that instead of the 5 breast cancer subtypes identified in primary breast tumors, cell lines can be divided into three main groups, luminal, basal A, or basal B phenotypes (Kao et al., 2009). Luminal cells contained all ER + cell lines, both Basal A and B consisted of all ER - cell lines. HER2+ cell lines were grouped into the luminal. Basal A contained the HCC cells and BRCA1 mutant cells, whereas basal B genotype contained non-tumorigenic lines including MCF10A cells (Kao et al., 2009). This highlights that breast cancer cell lines are a model of disease.

Cell lines are merely a model of breast disease that aim to provide clinical predictability of outcomes in humans. To directly test the applicability of breast cancer cell lines, xenograft cancer models, and mouse breast cancer models to clinical outcome, Voskoglou-Nomikos et. al. compared outcomes *in vitro* to those in xenograft models, to mouse models and phase II clinical trails (Voskoglou-Nomikos et al., 2003). In these comparisons, a general correlation between relative risk (predictive value of a drug in cell line) and the phase II human trial (tumor/control ratio) existed for *in vitro* cell lines. A general predictive value when using xenograft models to predict outcome to chemotherapy was detected, however this was dependent on the drug tested and the grade/type of tumor analyzed (Voskoglou-Nomikos et al., 2003). Overall, Vaskoglou-Nomikos et. al. concluded that cell lines and xenograft models were good predictors of clinical phase II trial outcomes, but are reliable predictors only when testing cytotoxic drugs and when using the correct model system. These models generally were not predictive of human outcomes when testing non-cytotoxic drugs (Voskoglou-Nomikos et al., 2003). Taken together, these studies emphasize the critical need to establish more breast cancer cell lines that model the heterogeneity of breast cancer and to employ many *in vitro* and xenograft model systems using multiple cell lines per experiment to reliably predict clinical outcome.
