Background
Breast cancer comprises a complex and diverse group of neoplasms that are stratified into various subtypes to guide treatment [
1]. The three-marker diagnostic system of estrogen receptor alpha (ER), progesterone receptor (PR), and HER2 remains the cornerstone of clinical decision-making with prognostic and predictive value. In addition, the WHO classifies invasive carcinomas into a dozen more common histological subtypes, plus several rare subtypes, based on architectural and morphological features [
2]. Among these, invasive ductal carcinomas (IDC) comprise the vast majority of cases (> 70%). However, it is increasingly recognized that other histological subtypes such as invasive lobular carcinomas (ILC) have unique morphologic and clinical attributes [
3,
4]. Molecular classifications that emerged in the 2000s use gene expression analysis to define subtypes with different recurrence and survival patterns, for example, the PAM50 classifier [
5], or a specific triple-negative breast cancer (TNBC) classifier [
6]. Breast cancers can also be characterized by gene mutation status such as germline BRCA1/2 and somatic PIK3CA mutations [
7‐
9], which can indicate eligibility for targeted inhibitors to PARP or PIK3CA, respectively, or by progressive mutation in the gene for ER (ESR1) which indicates loss of sensitivity to some endocrine therapies [
10]. Existing human breast cancer cell lines, patient-derived xenografts (PDX), patient-derived organoids (PDO), and murine models with intact immune systems capture many of these unique histopathological, molecular, and genetic subtypes. Here we describe generation of six PDX-derived breast cancer cell lines including three ER+ lines, one with ESR1 mutation, that add to the collection of research models available to investigators.
In vitro propagating cell lines are the most long-standing and heavily used models for conducting basic breast cancer research. The first breast cancer cell line was generated from a primary breast IDC by Lasfargues et al. over 60 years ago named BT-20, a basal-like TNBC [
11]. In the 1970s, the first ER+ lines were developed including two from late stage pleural effusions (MCF7 [
12] and T47D [
13]), and two from primary IDCs (BT-474, also HER2 amplified, and BT-483 [
14]). Larger collections were described over a 20-year period including 19 from the MD Anderson (MDA) [
15], 18 from the Hamon Cancer Center (HCC, UT Southwestern) [
16], and 12 from the University of Michigan (SUM) series [
17]. There are numerous other descriptions of generating single or several cell lines within focused research articles. Collectively, these endeavors have produced approximately 51 human breast cancer cell lines available for research [
18]. More recently, breast cancer cell lines have been developed from circulating tumor cells [
19,
20]. A survey of PubMed shows > 40,000 original research articles utilizing MCF7 cells and > 14,000 for MDA-MB-231 cells. These well-characterized cell lines are commonly manipulated to over or under express specific genes, test specific drugs, and perform genome-wide functional genetic and drug screens. This research has led to the discovery of a multitude of (mostly) cell intrinsic behaviors. Despite their historic importance, the evolving emphasis on personalized therapies necessitates continued generation of newer models that capture tumor diversity found in the population.
We previously reported the generation of a collection of PDX with an emphasis on ER+ models [
21]. These provide a continuously propagating in vivo resource for testing therapeutics in a heterogeneous context, and there are > 500 such models developed by groups worldwide [
22]. PDO offer an ex vivo approach to screen different patients’ tumors for drug sensitivity and can potentially include other cell types contained in the microenvironment [
23]. However, PDX and PDO are time-consuming and molecular biology approaches including shRNA knockdown or CRISPR/Cas9 gene editing remain difficult in such models. We sought to generate cell lines from PDX that harbor underrepresented phenotypes and genotypes that allow basic molecular manipulations and mechanistic studies and that can be used in parallel with in vivo studies. We have created six such cell lines that capture several important clinical features and subtypes and present here their initial characterization.
Methods
Hormones
Hormones and drugs 17β-estradiol (E2), progesterone (P4), dihydrotestosterone (DHT), 4-hydroxy-Tamoxifen (4-OH-Tam), and ICI 118,551 (Fulvestrant (Fulv)) were purchased from Sigma-Aldrich (St. Louis, MO).
Cell line generation
Derivation and propagation of PDX was described previously [
21,
24]. Generation of cell lines was essentially as described by Jambal et al. [
25]. Briefly, tumors were excised from animals, partitioned into ~ 5 mm chunks, and mechanically dispersed using a cell dissociation sieve and glass pestle (Sigma-Aldrich, St. Louis, MO). Cells were seeded into 6-well dishes, allowed to adhere and grown at 37 °C under 5% CO
2 in Dulbecco’s modified minimum essential medium (DMEM)/F12 containing L-glutamine (365 mg/L) buffered with sodium bicarbonate (1200 mg/L) and HEPES (3575 mg/L) with 10% fetal bovine serum, cholera toxin (100 ng/mL), hydrocortisone (1 μg/mL), insulin (10
−9 M), and pen strep. Cells were fed every 3–4 days until they emerged from crisis into passageable cell lines (6–18 months). Control breast cancer cell lines T47D and MDA-MB-468 were obtained from the University of Colorado Cancer Center Tissue Culture core. Short tandem repeat (STR) profiling and analysis was performed by the University of Arizona Genetics Core (University of Arizona, Tucson, AZ). PDX-derived cell lines were compared to their PDX of origin and other cell lines in the database. Cell lines were routinely tested for mycoplasma using the MycoAlert detection kit (Lonza, Basel, Switzerland).
Immunochemistry and microscopy
For immunocytochemistry (ICC), 1–2 × 10
5 cells were seeded onto glass coverslips in 6-well plates in regular media. For assessing PR, cells were treated with 10 nM E2 for 48 h prior to collection. Cells were washed twice with PBS and fixed with ice-cold 70% acetone/30% methanol for 5 min. Fixed cells were blocked with 10% normal goat serum (Vector Labs, Burlingame, CA) in 0.05% TBS-T for 30 min. Primary antibodies were as follows: ERα (SP1, 1:200, Thermo-Fisher, Waltham, MA), PR (1294, 1:50, [
26], CK5 (ab75869, 1:400, Abcam, Cambridge, MA), CK8/18 (NCL-L-5D3, 1:2000, Leica Biosystems), and Vimentin (5741, 1100, Cell Signaling, Danvers, MA) for 1 h. Secondary antibodies were A11029 (green) and/or A11037 (red) (Invitrogen, 1:200) for 30 min followed by counterstaining with 0.1 μg/mL DAPI. Phase contrast images were captured using a Nikon TiE microscope (Nikon, Melville, NY) equipped with a digital camera and NIS Elements 4.6 software. Fluorescent images were captured using an Olympus BX40 microscope equipped with a digital camera and cellSens Standard 1.13 software. Adobe Photoshop CC 2019 was used to perform minimal linear adjustments to brightness/contrast and to assemble pictures into multipanel figures. Immunohistochemistry (IHC) of PDX was performed with the same antibodies for the indicated markers as previously described [
21].
Immunoblotting
Cells were treated with vehicle (ethanol), 10 nM E2, 100 nM P4, E2 plus P4, or 10 nM DHT for 48 h. Cell lysates were harvested in lysis buffer (50 mM Tris pH 7.4, 140 nM NaCl, 2 mM EGTA, 1% Tween-20) plus protease/phosphatase inhibitors. Primary antibodies were as follows: ERα (6F11, Santa Cruz Biotechnology (sc-56,836), 1:1000), PR (1294, 1:500), AR (441, 11,000, DAKO-Agilent, Santa Clara, CA), and α-tubulin (ST1568, 1:30,000, Sigma-Aldrich). Secondary antibodies were as follows: IRDye 800CW Goat-Anti-Mouse IgG and IRDye 680RD Goat-Anti-Rabbit IgG (all from Li-Cor Biosciences, Lincoln, NE). The Odyssey CLx Infrared Imaging System and Image Studio 5.2 (Li-Cor Biosciences) were used to capture images.
Fluorescent tagging of cell lines and proliferation assay
Lentiviral particles were produced by co-transfecting HEK293FT cells with the structural plasmids VSV-G, d8.9, and pLJMGFP-GFP-3xNLS-puromycin constructs using TransIT-LT1. Media containing viral particles was harvested after 48 h. UCD cell lines were plated at ∼ 60,000 cells per 6 cm dish in 5 mL of complete medium. Approximately 24 h post plating, the medium was replaced with medium containing lentiviral particles. Polybrene was added at a final concentration of 8 mg/mL. Cells were transduced for 48 h, medium replaced with fresh complete medium and allowed to recover for 24 h, and selection performed over 7–10 days.
Real-time imaging (IncuCyte, Sartorius, Ann Arbor, MI) was used to measure proliferation of nuclear-GFP labeled cells at × 10 magnification. UCD4 cells were plated at 15,000 cells/well, UCD65 at 20,000 cells/well, and UCD12, UCD46, UCD115, and UCD178 cells at 10,000 cells/well, all in 96-well plates. For treatments, cells were given vehicle, 10 nM E2, 100 nM 4-OH-Tam, or 10 nM Fulv on day one. Green count was taken immediately after treatment and subsequently taken every 4 h for 6 days. Counts were calculated as fold change. Significance was assessed by one-way ANOVA/Tukey at the final time point.
Growth of cell lines in vivo
Animal experiments were performed under an approved University of Colorado Institutional Animal Care and Use Committee protocol. Tumor xenografts were developed by injecting 1 × 106 cells in 90% Cultrex Basement Membrane Extract (R&D Systems, Minneapolis, MN) into the fourth mammary fad pad of ovariectomized female NOD-scid IL2Rgammanull (NSG). Silastic pellets containing 17β-estradiol (1 mg) were implanted subcutaneously at time of tumor cell injection. Tumors were measured weekly using a digital caliper, and tumor volume estimated using the formula (lw2)/2.
RNA sequencing
Cell lines were plated in 60 mm plates in regular media. When cells were ~ 70% confluent, cells were washed with PBS and lysed with 0.7 mL Qiazol (Qiagen, Germantown, MD) immediately. RNA was prepared using miRNeasy mini columns (Qiagen) and treated with RNase-free DNase. RNA concentration was measured using a Nanodrop 2000 (Thermo-Fisher), and integrity analyzed using an Agilent 2100 Bioanalyzer and the RNA 6000 Nano kit. Libraries were prepared using the Illumina TruSeq Stranded mRNA Library Prep kit and samples sequenced using the Illumina Novaseq System. Paired-end 150 nt reads were aligned to the human genome version GRCh38.p13 using STAR 2.6. The downstream expression analysis was done using Cufflinks 2.2.1. PAM50 scores were assigned from the expression values (TPM) using the R package genefu 2.18.1 [
27].
Targeted oncogenic driver analysis
Samples of each of the six cell lines were submitted to the Virginia Commonwealth University Molecular Diagnostics Laboratory (MDL) for testing using the Oncomine Comprehensive Assay v3 panel (Thermo-Fisher). This assay uses an NGS platform to detect relevant SNVs, CNVs, gene fusions, and indels from 161 unique cancer driver genes (
https://assets.thermofisher.com/TFS-Assets/LSG/brochures/oncomine-comprehensive-assay-v3-flyer.pdf). Data were annotated by the MDL according to the ASCO/AMP guidelines. All samples were confirmed to be human in origin.
Discussion
The innate intertumoral heterogeneity among breast cancers and an increasing emphasis on individualizing therapies necessitate that we continue to generate research models to meet this challenge. Advances in measuring CTCs and circulating tumor DNA further facilitate real-time monitoring of disease progression and personalized care. Our group and others have derived collections of breast cancer PDX that can be utilized for pre-clinical drug testing [
22]. However, some fundamental research questions still require novel human disease models that can be more feasibly engineered. Here we describe the generation of six PDX-derived passageable breast cancer cell lines that are amenable to manipulations such as viral transduction. These complement existing models, with well-annotated oncogenic driver mutations and expression profiles, to provide depth in conducting basic and translational research on breast cancer.
One of our primary goals was to increase the number of workable ER+ breast cancer cell lines, which are relatively underrepresented compared to their clinical predominance. The primary “workhorse” ER+ breast cancer models include IDC subtypes (MCF7, T47D, ZR75-1, and the ER+HER2+/amplified BT474) with several ER+ ILC cell lines seeing increased use (MDA-MB-134, MDA-MB-330, SUM44, and BCK4) [
25,
30]. Several additional cell lines are reported to have ERα mRNA transcripts [
18]; however, ER protein expression has not always been documented. PR is expressed only in UCD4 and UCD65; UCD65 cells have some constitutive expression of PR in the absence of estrogen likely due to their naturally high ER level. AR is present in all three cell lines to some degree but is highest in UCD4 cells where it is stabilized with DHT. A drawback of these ER+ cell lines is their relatively long doubling times compared to long-term cultured ER+ cell lines. UCD65 has the longest doubling time, which is typical of the slower growing luminal A subtype breast cancers. A slow proliferation rate is also typical of newly developed breast cancer cell lines [
17] and may more accurately reflect growth rates in ER+ patients.
It is now recognized that up to 30% of advanced breast cancer patients contain somatic genetic anomalies in the ER gene (ESR1), prospectively driven by long-term estrogen deprivation with aromatase inhibitors (AIs) [
31]. Existing breast cancer cell lines do not harbor ESR1 mutations naturally (
cbioportal.org), even though some were derived from metastatic patients, prospectively because these patients were either untreated or treated prior to standard use of AIs in the 2000s. To functionally study the mutant ERs, laboratories have used exogenous expression or generated CRISPR knock-in models of the D538G and Y537S ESR1 mutations, or forced mutations by long-term endocrine treatment of ER+ breast cancer cells [
31‐
34]. Some PDX models contain ESR1 mutations, notably in the WHIM collection [
35]. To our knowledge, a cell line has not been derived from a specimen with a natural ESR1 mutation, without the potential unintended off-target effects of CRISPR-derived clonal cell lines, and that likely contain other anomalies that co-occur with ESR1 mutation. The UCD4 cell line is homozygous for the ESR1 D538G mutation and is Fulv but minimally Tam responsive (Fig.
3). UCD65 cells are the first cell line, to our knowledge, that is luminal A and expresses high levels of ER comparatively to the other cell lines. UCD12 (formerly PT12 cells) are ER+ have been used to study mucin secretion dynamics, anti-AR therapies, and endocrine resistance in diet-induced obesity models of breast cancer [
28,
36,
37]. UCD4 and UCD12 contain BRCA2 mutations that are likely somatic although the germline status of the patients is unknown. All three cell lines contain amplification of the FGF-FGFR signaling axis including FGFR1 (UCD12 and UCD65), FGFR3 (UCD4), and FGF3 and FGF19 in UCD65. This underscores the significance of FGF signaling as driving ER+ breast cancers, and here we note these anomalies in early (primary tumor, UCD12), intermediate (lymph node, UCD65), and advanced (pleural effusion, UCD4) disease settings. Thus, the ER+ cell lines provide additional platforms for testing FGFR inhibitors [
38].
The three ER−PR− cell lines (by our limits of detection) were derived from PDX which each contained rare ER+ cells (Supplemental Figure
1). However, the patient tumors were defined as ER−PR+ (UCD46), ER−PR− (UCD115), and ER+ at original diagnosis but untested in the recurrent pleural effusion (UCD178). Total loss of ER upon tumor progression occurs in only 10–20% of patients [
39], and we speculate this could be the case for UCD178. We have observed that most TNBC specimens contain rare ER+ cells as PDX, prospectively due to estrogen used at initial implantation or a change in microenvironment. We speculate that the predominant ER− populations were likely selected for in culture. All three ER− cell lines also have alterations associated with impaired P53 function, either mutation in the TP53 gene (UCD115, UCD178) or amplification of its negative regulator MDM2 (UCD46). Basal-like UCD46 cells additionally have amplification of cyclin E (CCND2), PIK3CA, and CREBBP. UCD115 is peculiar as its parent PDX is epithelial-like with CK5+ and CK18+ cells, whereas the cells that grew in culture may have undergone a partial epithelial-mesenchymal transition (EMT), evidenced by loss of CKs, gain of vimentin, and increase in transcripts for EMT transcription factors (Fig.
1, Supplemental Figure
5, Supplementary Table
1). The UCD178 line was derived from a patient with ILC at the time of recurrence in the lungs (pleural effusion). Although this cell line aligned with the luminal B molecular subtype, it has low ER transcript expression and lack of ER protein. RNA-seq data for UCD178 shows expression of both E-cadherin and N-cadherin transcripts, which we confirmed by ICC (not shown). We presently describe the UCD178 cell line as mixed ductal/lobular, although further characterization is needed to define their unusual histological type. UCD178 also contains a SEC16A-NOTCH1 fusion protein occasionally found in breast cancers [
40]. Thus, the three ER− cell lines share some common features of TNBC cell lines (i.e., mutation in the P53 signaling axis, MYC amplification) and harbor some less common mutations (i.e., SEC16A-NOTCH fusion) and histological features (i.e., ILC/IDC subtype).
Open AccessThis article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit
http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (
http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.