ANTICANCER RESEARCH 36: 1507-1518 (2016)

Establishment and Characterization of a Pair of Patient-derived Human Non-small Cell Lung Cancer Cell Lines from a Primary Tumor and Corresponding Lymph Node Metastasis

FLORIAN GEBAUER1,2*, DANIEL WICKLEIN2*, MICHAEL TACHEZY1, TOBIAS GROB3, DORIS STEINEMANN4, GEORGI MANUKJAN4, GUDRUN GÖHRING4, BRIGITTE SCHLEGELBERGER4, HANNA MAAR2, JAKOB R. IZBICKI1, MAXIMILIAN BOCKHORN1 and UDO SCHUMACHER2

1Department of General, Visceral and Thoracic Surgery, 2Institute of Anatomy and Experimental Morphology, University Cancer Center Hamburg, and 3Institute for Pathology, University Medical Center Hamburg-Eppendorf, University of Hamburg, Hamburg, Germany; 4Institute of Human Genetics, Hannover Medical School, Hannover, Germany

Abstract. Background: Non-small lung cancer is the leading epithelial–mesenchymal transition. Due to accelerated tumor cause of cancer-related mortality worldwide. For a deeper growth in vivo by the PT cells, a shortened overall survival understanding of tumor biology, we established a pair of cell was seen (60 vs. 101 days, p=0.005). Conclusion: We provide lines derived from a primary tumor and a corresponding detailed analysis of a cell line derived from a primary tumor lymph node metastasis. Material and Methods: The cell line and a corresponding LN metastasis. This unique feature BC4323 from the primary tumor (PT) and a mediastinal lymph allows further investigative analysis of the differences and node metastasis (LN) were derived from an adenocarcinoma regulatory processes underlying the metastatic process during (pT2, pN2, G3, UICC stage IIIa) in a 47-year-old female tumor progression in non-small cell lung cancer. patient. Comparative characterization was performed by in vitro analysis. A murine xenograft was established for analysis Lung cancer is the leading cause of cancer-related mortality of in vivo behavior. Results: Chromosomal aberrations were worldwide (1). Non-small cell lung cancer (NSCLC) accounts detected in multiple chromosomal sections throughout the for 80% of all lung cancer-related deaths. Patients with entire genome, with only a few differences between PT and LN NSCLC often experience relapse and develop metastases after cells. High-level Kirsten ras oncogene homolog (KRAS) initial surgery; this results in an approximate 5-year survival mutation and amplification were seen based on a rate of only 15% (2). NSCLC commonly metastasizes to chromosomal translocation and novel assembled . distant organs such as brain, adrenal glands, liver and bones, In contrast to the genomic level, at the mRNA and and this is generally considered a sign of dismal prognosis and levels, multiple differences were detectable, in particular in poor clinical outcome (3). Important driver mutations in the markers for cell adhesion [e.g. epithelial cell adhesion carcinogenesis of NSCLS have been identified that have molecule (EpCAM), CD44, P- binding, epidermal radically altered treatment pathways for NSCLC (4, 5). New growth factor receptor (EGFR) and integrin alphaV] and the drugs have been developed that specifically target key mutated such as epidermal growth factor receptor (EGFR), and kirsten ras oncogene homolog (KRAS) and anaplastic ALK This article is freely accessible online. lymphoma kinase ( ) mutations (6-8). However, the overall survival (OS) of patients is still unsatisfactory as the *These Authors contributed equally to this work. aforementioned agents often do not have the expected success in clinical practice. These aspects show the importance of an Correspondence to: Florian Gebauer, MD, Department of General, improved understanding of the pathophysiological processes Visceral and Thoracic Surgery, University Medical Center Hamburg- in tumor cells that promote local aggression and recurrence in Eppendorf, Martinistr. 52, 20246 Hamburg, Germany. Tel: +49 NSCLC, and in particular, the need for understanding the 40741050152, Fax: +49 40741054995, e-mail: [email protected] metastatic process, which despite its clinical importance is still Key Words: Non-small cell lung cancer, cell line, lymph node poorly understood (9). Cell lines established from NSCLC are metastases, chromosomal aberrations, KRAS, EpCAM, CD44, P- widely used for in vitro and in vivo analyses of carcinogenesis, selectin, EGFR, , epithelial mesenchymal transition. as well as in the development of new targeted therapies and

0250-7005/2016 $2.00+.40 1507 ANTICANCER RESEARCH 36: 1507-1518 (2016) conventional chemotherapy (10). However, these cell lines are exceeded 2 cm3 or ulcerated the mouse skin. The mice were derived from primary tumors in the majority of cases and terminally narcotized and sacrificed by cardiocentesis. therefore the obtained findings are of limited clinical The left lungs were homogenized in a sample disruptor relevance, since a comparison between the primary tumor and (TissueLyser II; Qiagen, Hilden, Germany) and subjected to DNA- isolation (QIAamp DNA Mini Kit; Qiagen). Bone marrow was a corresponding metastasis is not possible. collected by flushing the left femoral and tibial bones with 1 ml of Herein, we established a pair of cell lines from a NSCLC 0.9% NaCl. Two hundred microliters of blood and the bone marrow from the primary tumor and a corresponding LN metastasis suspensions were subjected to DNA isolation using QIAamp DNA and characterized these cells extensively in vitro and in vivo Blood Mini Kit. with respect to differences in expression and the DNA concentrations in all the samples were quantified using a biological behavior of the cells, providing a tool for a NanoDrop spectrophotometer (Peqlab, Erlangen, Germany). As the deepened analysis of the metastatic process. content of detectable Alu-sequences in the subsequent quantitative polymerase chain reaction (qPCR) would have been affected by varying the DNA concentration, all lung and bone marrow DNA Materials and Methods samples were normalized to 30 ng/μl using AE buffer (Qiagen). The concentrations of blood DNA were quite similar in all samples Patient and surgical procedure. A female patient was operated on in (approx. 10 ng/μl) and were therefore not normalized. qPCR was September 2004 at the Department for General, Visceral and Thoracic performed with established human-specific Alu-primers (12). Two Surgery at the University Medical-Center Hamburg Eppendorf. The microliters of total DNA (i.e. 60 ng lung or bone marrow DNA; 20 patient was 47 years old at the time of the operation and underwent a ng blood DNA) were used for each qPCR. Numerical data were right-sided pneumectomy with locoregional and mediastinal determined against a standard curve as described elsewhere (12). lymphadenectomy. The patient presented with a history of 60 pack- For each sample, analyses were performed in duplicate and as years of smoking before surgery and a distant metastasis was found 8 independent experiments at least twice. weeks after the initial operation. She received four cycles of palliative chemotherapy with carboplatin/vinorelbine and finally died 5.5 Flow cytometry. Cultured cells were trypsinized, washed and stained months after initial diagnosis. The final histopathological examination on ice with either phycoerythrin (PE)-, or fluorescein-isothocyanate revealed an adenocarcinoma of the lung, pT2, pN2, G3, UICC stage (FITC)-, or allophycocyanin (APC)-conjugated primary antibodies IIIa. Written-informed consent by the patient for investigational without prior fixation. The following antibodies were used purposes was obtained before surgery. according to the manufacturers’ recommended concentrations: activated leukocyte cell adhesion molecule (AlCAM; 1:100 IgG1 PE; eBioscience, Frankfurt, Germany), epithelial cell adhesion Establishment of cell lines BC4323PT and BC4323LN. Small molecule (EpCAM; 1:100 IgG1 PE; R&D Systems, Minneapolis, fragments of the tumor (BC4323PT) and the mediastinal LN MN, USA), CD44 (1:100 IgG1 FITC; R&D Systems), CXC-motif- metastasis (BC4323LN) (1 mm3) were enzymatically disaggregated chemokine receptor 4 (CXCR4; 1:100 IgG2a, APC; R&D Systems), after incubation with 0.5% collagenase IV (Sigma Aldrich, carbohydrate-antigen 19-9 (CA19-9; 1:100 IgM APC; R&D Steinheim, Germany) at 37˚C in a rotary shaker for 45 min. After Systems), CD15s (1:100 IgM APC; R&D Systems), P-selectin centrifugation at 700 × g for 5 min, the cell pellet was collected, (1:100 P-Selectin chimera APC; R&D Systems), E-selectin (1:100 washed twice in cell-culture medium (RPMI-1640; Invitrogen, New E-selectin chimera APC; R&D Systems), epidermal growth factor York, NY, USA) and re-suspended in cell culture medium (RPMI- receptor (EGFR; 1:100 IgG2b PE; eBioscience), vascular 1640 supplemented with 10% fetal calf serum (FCS) and 200 IU/ml endothelial growth factor (VEGF; 1:100 rabbit polyclonal IgG PE; penicillin-streptomycin) then plated in 75 cm2 cell-culture flasks eBioscience), CD24 (1:100 IgG1 PE; R&D Systems), CD133 (1:100 (Sarstedt, Darmstadt, Germany). The cells were cultured under IgG2b PE; eBioscience), cell adhesion molecule (L1-CAM; standard conditions at 37˚C with 5% CO . Cell-culture medium was 2 1:100 IgG2a PE; eBioscience), -related replaced every 3-5 days, depending on the confluence of the cells. cell adhesion molecule (CEACAM1; 1:100 IgG2b PE; R&D The cells were cultivated for 25 passages to eliminate fibroblasts Systems), CEACAM6 (1:100 IgG2a APC; R&D Systems), integrin and confirm malignant immortal cell growth of the cells. For alpha V (1:100, IgG2a FITC; Santa Cruz, Dallas, TX, USA), passaging, the cells were treated with 0.05% Trypsin-EDTA (1:100 IgG1 FITC; Santa Cruz), and (Invitrogen, Darmstadt, Germany) for 3 minutes at 37˚C; the (1:200, IgG1, Alexa Fluor e660; eBioscience). Corresponding enzymatic reactions were stopped with cell-culture medium and the isotype controls were used as negative controls. cells were transferred into new culture flasks. For intracellular protein detection, cells were fixed with 2% formaldehyde for 20 min at room temperature. After washing with Animal experiments. Animal experiments were conducted according PBS, cells were permeabilized with 1% saponin (Sarstedt) for 20 min to the United Kingdom Co-Ordinating Committee on Cancer at room temperature, after which the cells were again centrifuged Research for the welfare of animals in experimental neoplasia (11). and stained with the antibody dilutions mentioned above. The experiment was approved and supervised by the Institutional Animal Welfare Officer and approved by the local licensing Immunostaining. Standard indirect immunoperoxidase procedures authority. were performed on paraffin-embedded tissue with the peroxidase For the subcutaneous tumor model, 1×106 BC4343PT and method according to the manufacturer's protocol (HRP-AEC BC4323LN tumor cells were injected into the right scapula region in System, Cell and Tissue Staining Kit; R&D Systems). For 8- to 12-week-old C57BL/6N Pfp−/−/Rag2−/− double-knockout immunohistochemical detection, the antibodies to the following mice, respectively. Animals were sacrificed when primary tumors were used: VEGF, EGFR (clone E30; Dako, Jena, Germany),

1508 Gebauer et al: Establishment of NSCLC Cell Lines intercellular adhesion molecule 1 (ICAM1; clone 8439; Santa Cruz), RNA isolation and PCR analysis. RNA was extracted from cell-culture VCAM (clone 1360; Santa Cruz), integrin beta 4 (clone 439-9B; grown BC4323PT and BC4323LN cells using an RNA isolation Abcam, Cambridge, United Kingdom) CAECAM5, CEACAM6, (RNeasy Mini Kit; Qiagen). RNA concentrations were quantified CD44 (clone G44-26; BD Bioscience, San Jose, CA, USA), S100 using a NanoDrop spectrophotometer and 1 μg was processed to calcium-binding protein A 4 (S100 A4); clone A5114; Dako), HIF- cDNA with the RT2 First Strand Kit (SA Biosciences, Hilden, 1α (H1α67; Sigma-Aldrich, Hamburg, Germany), zinc finger E- Germany). Quantitative PCR analysis of 84 human extracellular matrix box-binding homeobox 1 (ZEB1; polyclonal, Sigma-Aldrich), ZEB2 and epithelial-mesenchymal transition (EMT) was assessed with (polyclonal; Sigma-Aldrich), twist-related protein 1 (TWIST; RT2 Profiler PCR Arrays (SA Biosciences) and RT2 SYBR Green polyclonal; Abcam), and Ki67 (clone MIB-1; Dako). qPCR master mix (SA Bioscience) in a LightCycler 480 (Roche, Mannheim, Germany). Five housekeeping genes and internal controls Array-based comparative genomic hybridization, KRAS and EGFR (beta-actin, beta-2-microglobulin, glyceraldehyde-3-phosphate exon analysis. Array-based comparative genomic hybridization dehydrogenase, hypoxanthine phosphoribosyltransferase 1, large (CGH) was performed using the Agilent Microarray ribosomal protein) for genomic DNA contamination, RNA quality, and Kit 2×400K (Agilent Technologies, Santa Clara, CA, USA), a high general PCR performance were also included. resolution 60-mer oligonucleotide-based microarray with a median overall probe spacing of about 5.3 kb. Prior to DNA extraction, the Cell proliferation and chemosensitivity assays. Cell proliferation cells were treated with 100 nM Torin1 (R&D Systems, Wiesbaden- was assessed using the colorimetric XTT assay (Roche Diagnostics, Nordenstadt, Germany) for 5 h to provide reliable profiles of rapidly Basel, Switzerland) according to the manufacturer's instructions. dividing cells in the subsequent array-CGH analyses. Labelling and Cells were plated in 96-well plates at 4,000 cells/well. After 48 h hybridization of genomic DNA was performed according to the to allow for cell adherence, cells were incubated with colorimetric protocol provided by Agilent. Briefly, 0.75 μg of DNA was labelled substrates. Colorimetric changes were measured in a multiwell by random priming using the Agilent Genomic DNA Labelling Kit spectrophotometer (DynaTech MR 5000; Denkendorf, Germany). Plus test DNA with Cy3-dUTP and reference DNA (pool of five sex- To determine the drug sensitivity of BC4323PT and BC4323LN matched human control DNAs) with Cy5-dUTP. Labeled products cells, cells were prepared as described aove and incubated for 24 h were purified over Amicon Ultra 30k filters (Millipore, Billerica, with standard cell-culture medium. After 24 h, increasing MA, USA), combined and then mixed with human COT1-DNA (50 concentrations of gemcitabine (Merck, Darmstadt, Germany), μg), 10× Agilent blocking agent and 2× Agilent hybridization buffer. oxaliplatin (Actavis, Munich, Germany), cisplatin (Hexal, This solution was then hybridized to Agilent’s 2×400K Human Holzkirchen, Germany) and docetaxel (Betapharm, Augsburg, Genome CGH microarray at 65˚C with 20 rpm rotation for 24 h. Germany) were added for another 24 h and cell proliferation was Washing steps were performed according to the Agilent protocol. measured with XTT as described above. Microarray slides were scanned immediately using an Agilent microarray scanner G2505B at a resolution of 2 μm. For image Data analysis. Data are representative of three or more independent analysis, default CGH settings of Feature Extraction Software repetitions. Data are presented as the mean with standard deviation (Agilent Technologies, Waldbronn, Germany) were applied. Output (SD) or median with range. Statistical analysis was performed with files from Feature Extraction were subsequently imported into Student t-test or Wilcoxon test, respectively. A p-value less than Agilent’s CGH data-analysis software, Genomic Workbench. The 0.05 for differences was considered statistically significant. Aberration Algorithm ADM2 was applied with the Aberration Filters set to: threshold 5.0, at least 10 probes with mean log2 ratio of −0.3 Results for deletion and at least 10 probes with mean log2 ratio of 0.3 for gain/amplification, leading to a resolution of approx. 50 kb. Two bronchial adenocarcinoma cell lines designated For EGFR and KRAS exon mutation analysis, 100 ng of DNA BC4323PT and BC4323LN were generated from a resected extracted from PT and LN cells were subjected to PCR reactions PT and a corresponding mediastinal LN metastasis. Within using AmpliTaq GOLD PCR master mix (Applied Biosystems). the first 2 weeks, tumor cell clusters adhered to the surface EGFR exons 18, 19, 20 and 21 were analyzed with respect to single of the cell-culture flasks and gradually formed cell colonies. nucleotide polymorphisms. The primer sequences used have been Initially, contaminating fibroblastic cells proliferated and published before (13) For KRAS exon analysis, two primers surrounded the tumor cell colonies. During serial passages, (forward 5’-GCCTGCTGAAAATGACTGAA-3’ and reverse 5’- AGAATGGTCCTGCACCAGTAA-3’) were applied. PCR products the number of fibroblasts gradually decreased and they were were subjected to sequencing with the BigDye Terminator Cycle replaced entirely by tumor cells. Tumor cells exhibited Sequencing Kit (Applied Biosystems) and analyzed on an ABI adherent growth at the bottom of the cell-culture flasks and PRISM 3100 genetic analyzer (Applied Biosystems). presented a monolayer growth pattern (Figure 1A and B). At first sight, cells derived from the primary tumor did not differ Cytogenetic and fluorescence in situ hybridization (FISH) analysis. in shape, size or growth pattern compared to those derived Chromosomal preparation and fluorescence-R banding were from the LN metastasis. performed as previously described (14). Karyotypes were described BC4323PT cells exhibited an increased cell proliferative according to the International Standard of Cytogenetic Nomenclature (15). FISH of interphase nuclei and metaphases was capability compared to BC4323LN cells, as determined by performed according to standard procedures for the KRAS locus XTT assays (Figure 1C). Treatment with different (BueFISH probe, RP11-707G18;BlueGnome Ltd, Illumina Inc, San chemotherapeutic agents showed, in general, a higher Diego, CA, USA) (16). chemosensitivity for BC4323LN cells than BC4323PT cells,

1509 ANTICANCER RESEARCH 36: 1507-1518 (2016)

Figure 1. Primary tumor (BC4323PT) (A) and lymph node metastasis (BC4323LN) (B) cells show adherence to the bottom of cell culture flasks. BC4323PT cells exhibited increased proliferative ability compared to BC4323LN cells as determined by the XTT assay (C). Cytotoxic proliferation tests showed, in general, greater sensitivity to cytotoxic drugs for BC4323LN cells than BC, 4323PT cells (D-G). *Difference between cell lines significant at p<0.05 by the Wilcoxon test.

1510 Gebauer et al: Establishment of NSCLC Cell Lines

Figure 2. Graphical overview of genomic copy-number changes in primary tumor (BC4323PT) (A) and lymph node metastasis (BC4323LN) (B) cells. green bars: deleted chromosomal regions; red bars: gained chromosomal regions.

that was reflected by decreased proliferation at equivalent PCR expression analysis. BC4323PT and BC4323LN cells concentrations of chemotherapy agents (Figure 1D-G). The were analyzed with respect to 180 genes correlated with largest differences between BC4323PT and BC4323LN cells EMT and extracellular matrix proteins. A total of 42 genes were found for treatment with cisplatin and docetaxel. were found to be differentially regulated between the two cell lines. The highest up-regulation [fold-change (FC) >5] Cytogenetics and FISH. As analyzed by array-CGH, in BC4323LN compared to BC4323PT was found for chromosomal amplifications and deletions were found in desmoplakin (FC=30.7, p<0.001), osteopontin (FC 25.0, multiple chromosomal sections throughout the entire genome p=0.04), pathway inhibitor 2 (FC=22.0, (Figure 2). There were only minor differences in chromosomal p<0.001), collagen type XIV alpha 1 (FC=12.1, p=0.029), aberrations between the BC4323PT and BC4323LN cells, as guanine nucleotide binding protein (G protein) gamma 11 depicted in Figure 3A. Chromosomal gains in BC4323PT cells (FC=11.2, p<0.001) and tenascin C (FC=9.8, p<0.001). The compared to BC4323LN were seen in 1q and 2p. most up-regulated (FC>5) genes in BC4323PT compared to Losses were observed more frequently (4q, 4p, 8p, 9p, 11q and BC4323LN were collagen type XV alpha 1 (FC=33.9, Xq). Amplifications found in BC4323LN were in 5q, 5p and p=0.013), (FC=13.3, p<0.001), collagen 12q, while deletions were not detectable in either cell line. type I alpha 2 (FC=12.6, p=0.016) and ZEB2 (FC=5.6, High-level amplifications were found in both BC4323PT and p<0.001). BC4323LN cells on chromosome 12p (Figure 3A). Array-CGH showed a massive chromosomal gain of the p-arm of Flow cytometry. Differences in expression of surface chromosome 12. The KRAS gene is located within the proteins, as analyzed by flow cytometry were detectable amplified gene region, as shown in Figure 3A. Classical between the primary tumor and LN metastasis for EpCAM, banding analysis showed a tetraploid complex karyotype with CD44, P-selectin binding, EGFR and integrin alpha V. The numerical and structural aberrations. An additional FISH cells were highly positive for Alcam, CD24, CD44, integrin analysis for the KRAS locus on metaphases showed four regular alpha1, beta1 and integrin alpha V, while a moderate surface signals for KRAS located in 12p12, as well as an amplified presentation was found for numerous proteins (Figure 4). signal for KRAS in a derivative chromosome 4 and a high-level amplification in a marker chromosome (Figure 3B and C). Murine xenograft model. In order to assess the in vivo growth KRAS mutation analysis revealed a mutation in exon 2, type of the tumor cells, BC4323PT and BC4323LN cells were c.35G>C p. (G12A). The mutation was found to be probably injected subcutaneously in immunodeficient Pfp−/−/Rag2−/− homozygous. No mutations were found in any of the analyzed mice. Tumor take was seen in every single mouse (N=10 per EGFR exons. Responses to tyrosine kinase inhibitors were not group); however, mice with BC4323PT cells showed expected and therefore not tested for (data not shown). accelerated tumor growth that was reflected in a decreased

1511 ANTICANCER RESEARCH 36: 1507-1518 (2016)

Figure 3. A: Genomic profiles of chromosome 12 from lymph node metastasis (BC4323LN) (upper profile) and primary tumor (BC4323PT) (lower profile) cells showing a high level 12p intrachromosomal amplification (iamp 12p). The region of iamp12p is magnified to show gene resolution, showing kirsten rat sarcoma viral oncogene homolog (KRAS) within a region of highest amplification in BC4323PT and BC4323LN cells. Karyotyping (B) and fluorescence in situ hybridization analyses with a specific KRAS probe (C) showing high level amplification of the KRAS gene in a novel assembled marker chromosome (M1).

1512 Gebauer et al: Establishment of NSCLC Cell Lines

Figure 4. Flow cytometric data for numerous surface proteins. Black line shows the corresponding isocontrol, blue line the primary tumor cells and red line the lymph node metastasis cells. Activated leukocyte cell adhesion molecule (AlCAM), epithelial cell adhesion molecule (EpCAM), CXC- motif-chemokine receptor 4 (CXCR4), carbohydrate-antigen 19-9 (CA19-9), epidermal growth factor receptor (EGFR), vascular endothelial growth factor (VEGF), L1 cell adhesion molecule (L1CAM), carcinoembryonic antigen-related cell adhesion molecule (CEACAM) 1, 5 and 6.

1513 ANTICANCER RESEARCH 36: 1507-1518 (2016)

Figure 5. Survival data from the murine xenograft model shows prolonged survival for animals bearing tumors from lymph node metastasis (BC4323LN) (60 vs. 101 days, p=0.005) (A). Differences in total tumor weight (B) were not observed. There were no differences revealed in the number of circulating tumor cells (C) nor pulmonary metastasis (D) as determined by alu-quantitative polymerase chain reaction.

OS in this group compared to the BC4323LN-injected group on H&E staining. Both patient and mouse-grown tumors (Figure 5A). The median time of growth from tumor-cell presented considerable central necrosis, with only a small injection until reaching the termination criteria was 60 days rim of vital tumor tissue at the border between the tumor and in the BC4323PT group and 101 days in the BC4323LN normal tissue. The tumor cells exhibited a high level of group (log- test p=0.005). There were no differences in proliferative activity, represented by numerous mitotic tumor weight between the groups (Figure 5B). figures within the entire vital tumor area. The disseminative potential of BC4323 cells was assessed Immunohistochemical staining was performed on mouse- by PCR analysis for specific human sequences in different grown tumors for numerous cell-surface proteins. Differences compartments. Circulating tumor cells (CTCs) in the blood in antigen expression levels between tumors from BC4323PT- were detected in 4/10 animals in the BC4323PT-injected and BC4323LN-injected mice as revealed by either qPCR or group and 5/10 mice in the BC4323LN-injected group. The flow cytometry were validated by immunohistochemistry median number of CTCs did not differ significantly (Figure (Figure 7). Different expression levels were found for CD44, 5C). Similar results were found for lung metastases as tenascin C, TWIST and ZEB2. determined by PCR, with 4/10 mice showing evidence of lung metastasis in the BC4323PT-injected group and 5/10 Discussion mice in the BC4323LN-injected group (Figure 5D). No correlation was found between a high number of CTCs and Lung cancer remains one of the major causes of cancer- a higher copy number of human DNA sequences in the lung related mortality in the Western world (1). The mortality rate (data not shown). The total number of lung metastases was of lung cancer is high, even when novel multimodal rather low, and microscopically detectable metastases were therapeutic strategies are employed. Despite aggressive local not seen. Human DNA sequences representing bone marrow tumor growth, the individual prognosis of the majority of metastases were not detected in any group (data not shown). patients is limited by the development of local recurrence and distant metastasis (17). Microscopy and Immunohistochemistry. As depicted in In the present study, we characterized a pair of cell lines Figure 6, the primary patient tumor and the subcutaneous from a primary tumor and its corresponding LN metastasis. tumor developed in the murine xenograft were highly similar As distant and LN metastasis are one of the major challenges

1514 Gebauer et al: Establishment of NSCLC Cell Lines

Figure 6. Hematoxylin and eosin staining of the primary patient tumor as primarily resected (5- and 20-fold magnification) (A). Mouse-grown tumors of primary tumor (BC4323PT) (B) and lymph node metastasis (BC4323LN) cells (C) exhibit similar growth patterns, with central necrosis and vital tumor tissue only in the outer tumoral regions (5- and 20-fold magnification).

Figure 7. Immunohistochemical staining for CD44, tenascin C, twist-related protein 1 (TWIST) and zinc finger E-box binding homeobox 2 (ZEB2) of mouse-grown primary tumor (BC4323PT) and lymph node metastatic (BC4323LN) tumors.

1515 ANTICANCER RESEARCH 36: 1507-1518 (2016) in treating NSCLC, understanding the biology of the considered as a major driver mutation in carcinogenesis and underlying processes is of major importance. We focused on might explain the highly aggressive behavior of this particular whether we could find patterns that identified metastatic tumor in the clinical course of the patient (8, 27). The same processes or differences between the phenotypes of cells. Both result was found in both the primary tumor and LN metastasis. cell lines had the capacity to establish primary tumors, as This finding was unexpected, as KRAS mutations in both examined in a murine xenograft model. However, there were primary tumors and LN metastases are rather infrequent, as essentially no distant metastases seen, although small numbers reported by Sun and colleagues analyzing 80 patients with of human tumor cells were detected by qPCR. The results NSCLC (28). In the majority of patients, KRAS mutations have to be interpreted cautiously, as we were unable to were acquired during the metastatic process but not found in confirm the presence of micrometastases or even disseminated the primary tumor. Mutations in both the primary tumor and tumor cells microscopically. Irrespective of the metastatic LN metastasis were found in only one out of 80 patients status and as already shown for ovarian cancer and other (1.3%) and this underlines their infrequent occurrence. tumor entities, patient-derived cell line xenograft models Karyotyping including array-CGH, banding analysis and FISH recapitulate the biological processes of tumor progression in showed a complex tetraploid karyotype due to a greatly a quite consistent way and can therefore increase our increased genomic instability in the tumor cells, probably in understanding of carcinogenesis, therapy resistance and combination with defects in double-strand repair mechanisms metastatic progression (18-20). However, xenograft models as reported for numerous tumor entities (29, 30). PCR KRAS have limitations with respect to vascular, inflammatory and exon analysis showed a rather uncommon homozygote stromal interactions of murine tissue and the tumor cells (21). mutation (type p.G12A) that can be explained by a breakage- Additionally, as also seen in our model, spontaneous fusion-bridge that leads to amplification of unstable metastases are rare events and different means of application chromosomal regions as previously described (31). (e.g. intravenous, intracranial, intraperitoneal injection) are Other gene regions showed slight differences between the often required to provoke distant metastasis (22, 23). primary tumor and the LN metastasis, with inconclusive The question arises as to why non-metastatic cell lines interpretation. Whether this is an effect of general genomic were established from a clinically malignant tumor instability of the tumor genome or specific mutations associated with the onset of widespread metastases shortly promoting metastasis is not clear. As reported by Hoang and after surgery. As the two cell lines were genetically colleagues, differences in the mRNA profiles of NSCLC remarkably similar (see below), epigenetic changes could be primary tumors and their corresponding LN metastases are the key to the different metastatic behavior of the cell lines in much less frequent than initially expected (32). Samples from comparison to the tumor in the patient. Nestor et al. showed 11 patients revealed only minor differences in the expression rapid reprogramming of epigenetic and transcriptional profiles, making it difficult to identify a metastatic pattern profiles just 3 days after mouse fibroblasts had been put into explaining the lymphatic metastatic process. However, cell culture (24). The observed hypermethylation of the according to our results, cells from the primary tumor and LN genome resulted in an up-regulation of genes associated with metastasis exhibited a high level of homology that makes it cell adhesion and extracellular matrix organization. Hence, almost impossible to differentiate between cells derived from these epigenetic changes could account for the non- the primary tumor and those from the LN metastasis. During metastatic phenotype observed after initial cell culture of the the metastatic process, tumor cells generally undergo several carcinoma cells. This phenomenon has already been morphological changes. In the so-called EMT, tumor cells observed in murine Lewis lung carcinoma, which does form change their phenotype from epithelial in the primary tumor to lung metastases after subcutaneous implantation (25). a mesenchymal phenotype (33). We therefore analyzed whether Culture of tumor cells from subcutaneous primary tumors or this pattern existed in the LN metastasis but, as shown, typical artificial lung metastases resulted in two cell lines from a EMT markers (e.g. TWIST, and ZEB2) did not differ in such a total of 405 cell lines that were tumorigenic but not way that we can predict whether the LN cells exhibited a more metastatic, while the rest were both tumorigenic and mesenchymal phenotype. This leads to the conclusion that metastatic. Interestingly, the metastatic phenotype was tumor cells may have already undergone transition in the restored in these two cells lines by treating them with the reverse direction (MET), which means that the cells at the site demethylating agent 5-azacytidine (26). It is tempting to of metastasis had already re-established their epithelial speculate that similar processes were at play in our two cell phenotype. Another explanation could be that the lymphatic lines, which are an ideal pair to investigate this question. metastatic process differs from hematogenous distant Interestingly, only minor differences were detectable at the metastasis (e.g. brain, and adrenal gland) in NSCLC. As genomic, RNA and protein-expression levels. Array-CGH and lymphatic vessels are generally considered ‘leaky’ without tight FISH analysis of the genome revealed a highly amplified junctions between the lymphatic endothelium, in contrast to genomic region involving the KRAS gene, which is generally blood vessels, the lymphatic system can be entered easily once

1516 Gebauer et al: Establishment of NSCLC Cell Lines a tumor cell leaves the compact tumor formation. Therefore, 6 Kobayashi S, Boggon TJ, Dayaram T, Janne PA, Kocher O, the invasive and EMT programs of that particular tumor cell Meyerson M, Johnson BE, Eck MJ, Tenen DG and Halmos B: should differ between lymphatic and distant metastasis. EGFR mutation and resistance of non-small-cell lung cancer to gefitinib. N Engl J Med 352: 786-792, 2005. However, the cells derived from LN metastasis are, in this 7 Gridelli C, Peters S, Sgambato A, Casaluce F, Adjei AA and particular case, generally not that robust compared to the PT Ciardiello F: ALK inhibitors in the treatment of advanced cells tumor, as seen in the chemosensitivity and proliferation NSCLC. Cancer Treat Rev 40: 300-306, 2014. assays, and the xenograft model. PT cells exhibited increased 8 Timar J: The clinical relevance of KRAS gene mutation in non- proliferation, both in vitro and in vivo. Possibly, cells that are small-cell lung cancer. Curr Opin Oncol 26: 138-144, 2014. able to establish distant metastases in LNs exhibit a high 9 Valastyan S and Weinberg RA: Tumor metastasis: molecular level of aggressiveness that is not reflected in high insights and evolving paradigms. Cell 147: 275-292, 2011. proliferative and growth rates but rather more in the way 10 Gazdar AF, Girard L, Lockwood WW, Lam WL and Minna JD: Lung cancer cell lines as tools for biomedical discovery and these cells have the ability to endure over a long period of research. J Natl Cancer Inst 102: 1310-1321, 2010. time in a somewhat hostile environment. 11 Workman P, Aboagye EO, Balkwill F, Balmain A, Bruder G, In summary, we provide a unique tool for analyzing the Chaplin DJ, Double JA, Everitt J, Farningham DA, Glennie MJ, ongoing processes during LN metastatic formation in Kelland LR, Robinson V, Stratford IJ, Tozer GM, Watson S, NSCLC in a cell line model derived from patient material Wedge SR and Eccles SA: Guidelines for the welfare and use that could help improve our understanding of the metastatic of animals in cancer research. Br J Cancer 102: 1555-1577, process in NSCLC. 2010. 12 Nehmann N, Wicklein D, Schumacher U and Muller R: Comparison of two techniques for the screening of human tumor Conflicts of Interest cells in mouse blood: quantitative real-time polymerase chain reaction (qRT-PCR) versus laser scanning cytometry (LSC). Acta None. Histochem 112: 489-496, 2010. 13 Grob TJ, Hoenig T, Clauditz TS, Atanackovic D, Koenig AM, Acknowledgements Vashist YK, Klose H, Simon R, Pantel K, Izbicki JR, Bokemeyer C, Sauter G and Wilczak W: Frequent intratumoral heterogeneity The Authors thank Tobias Gosau, Christine Knies and Susanne of EGFR gene copy gain in non-small cell lung cancer. Lung Feldhaus for experimental support for the mouse models. Yogesh Cancer 79: 221-227, 2013. Vashist performing the initial surgery on the patient. 14 Wegner R-D: Diagnostic Cytogenetics. Berlin; New York: Springer, 1999. References 15 Mitelman F and Franck U: ISCN 1991: Guidelines for Cancer Cytogenetics: supplement to An international system for human 1 Siegel RL, Miller KD and Jemal A: Cancer statistics, 2015. CA cytogenetic nomenclature: Recommendations of the Standing Cancer J Clin 65: 5-29, 2015. Committee on Human Cytogenetic Nomenclature - Subcommittee 2 Sun S, Schiller JH and Gazdar AF: Lung cancer in never on Cancer Cytogenetics. Basel; New York: Karger, 1992. smokers – a different disease. Nat Rev Cancer 7: 778-790, 16 Gohring G, Hanke C, Kratz C, Kontny U, Steinemann D, 2007. Niemeyer CM and Schlegelberger B: Fluorescence in situ 3 Wakelee HA, Bernardo P, Johnson DH and Schiller JH: Changes hybridization using the subtelomeric 11q probe as a diagnostic in the natural history of nonsmall cell lung cancer (NSCLC)-- tool for congenital thrombocytopenia. Ann Hematol 85: 883-885, comparison of outcomes and characteristics in patients with 2006. advanced NSCLC entered in Eastern Cooperative Oncology 17 Xie Y and Minna JD: Non-small-cell lung cancer mRNA Group trials before and after 1990. Cancer 106: 2208-2217, expression signature predicting response to adjuvant 2006. chemotherapy. J Clin Oncol 28: 4404-4407, 2010. 4 Sequist LV, Martins RG, Spigel D, Grunberg SM, Spira A, Janne 18 Gebauer F, Wicklein D, Stubke K, Nehmann N, Schmidt A, PA, Joshi VA, McCollum D, Evans TL, Muzikansky A, Salamon J, Peldschus K, Nentwich MF, Adam G, Tolstonog G, Kuhlmann GL, Han M, Goldberg JS, Settleman J, Iafrate AJ, Bockhorn M, Izbicki JR, Wagener C and Schumacher U: Engelman JA, Haber DA, Johnson BE and Lynch TJ: First-line Selectin binding is essential for peritoneal carcinomatosis in a gefitinib in patients with advanced non-small-cell lung cancer xenograft model of human pancreatic adenocarcinoma in harboring somatic EGFR mutations. J Clin Oncol 26: 2442- pfp–/–/rag2–/– mice. Gut 62: 741-750, 2013. 2449, 2008. 19 Schumacher U, Adam E, Dietl J and Horny HP: Immunophenotype 5 Seto T, Kato T, Nishio M, Goto K, Atagi S, Hosomi Y, of human ovarian malignancies (cystadenocarcinomata and mixed Yamamoto N, Hida T, Maemondo M, Nakagawa K, Nagase S, mullerian tumor) established in SCID mice. Exp Mol Pathol 64: Okamoto I, Yamanaka T, Tajima K, Harada R, Fukuoka M and 103-113, 1997. Yamamoto N: Erlotinib alone or with bevacizumab as first-line 20 Schumacher U, Adam E, Horny HP and Dietl J: Transplantation therapy in patients with advanced non-squamous non-small-cell of a human ovarian cystadenocarcinoma into severe combined lung cancer harbouring EGFR mutations (JO25567): an open- immunodeficient (SCID) mice – formation of metastases without label, randomised, multicentre, phase 2 study. Lancet Oncol 15: significant alteration of the tumour cell phenotype. Int J Exp 1236-1244, 2014. Pathol 77: 219-227, 1996.

1517 ANTICANCER RESEARCH 36: 1507-1518 (2016)

21 Gazdar AF, Gao B and Minna JD: Lung cancer cell lines: 29 Ciccia A and Elledge SJ: The DNA damage response: making it Useless artifacts or invaluable tools for medical science? Lung safe to play with knives. Mol Cell 40: 179-204, 2010. Cancer 68: 309-318, 2010. 30 Curtin NJ: DNA repair dysregulation from cancer driver to 22 Oie HK, Russell EK, Carney DN and Gazdar AF: Cell culture therapeutic target. Nat Rev Cancer 12: 801-817, 2012. methods for the establishment of the NCI series of lung cancer 31 Zakov S, Kinsella M and Bafna V: An algorithmic approach for cell lines. J Cell Biochem Suppl 24: 24-31, 1996. breakage-fusion-bridge detection in tumor genomes. Proc Natl 23 Wistuba, II, Bryant D, Behrens C, Milchgrub S, Virmani AK, Acad Sci USA 110: 5546-5551, 2013. Ashfaq R, Minna JD and Gazdar AF: Comparison of features of 32 Hoang CD, Guillaume TJ, Engel SC, Tawfic SH, Kratzke RA human lung cancer cell lines and their corresponding tumors. and Maddaus MA: Analysis of paired primary lung and lymph Clin Cancer Res 5: 991-1000, 1999. node tumor cells: a model of metastatic potential by multiple 24 Nestor CE, Ottaviano R, Reinhardt D, Cruickshanks HA, genetic programs. Cancer Detect Prev 29: 509-517, 2005. Mjoseng HK, McPherson RC, Lentini A, Thomson JP, Dunican 33 Wellner U, Schubert J, Burk UC, Schmalhofer O, Zhu F, DS, Pennings S, Anderton SM, Benson M and Meehan RR: Sonntag A, Waldvogel B, Vannier C, Darling D, zur Hausen A, Rapid reprogramming of epigenetic and transcriptional profiles Brunton VG, Morton J, Sansom O, Schuler J, Stemmler MP, in mammalian culture systems. Genome Biol 16: 11, 2015. Herzberger C, Hopt U, Keck T, Brabletz S and Brabletz T: The 25 Hergueux J, Bischoff P and Donner M: Resistance and EMT-activator ZEB1 promotes tumorigenicity by repressing susceptibility to pulmonary metastasis of Lewis lung carcinoma stemness-inhibiting microRNAs. Nat Cell Biol 11: 1487-1495, in different mouse strains. Exp Cell Biol 51: 181-191, 1983. 2009. 26 Olsson L and Forchhammer J: Induction of the metastatic phenotype in a mouse tumor model by 5-azacytidine, and characterization of an antigen associated with metastatic activity. Proc Natl Acad Sci USA 81: 3389-3393, 1984. 27 Chiosea SI, Sherer CK, Jelic T and Dacic S: KRAS mutant allele-specific imbalance in lung adenocarcinoma. Mod Pathol 24: 1571-1577, 2011. 28 Sun L, Zhang Q, Luan H, Zhan Z, Wang C and Sun B: Comparison of KRAS and EGFR gene status between primary non-small cell Received January 28, 2016 lung cancer and local lymph node metastases: implications for Revised March 17, 2016 clinical practice. J Exp Clin Cancer Res 30: 30, 2011. Accepted March 28, 2016

1518