Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Lung cancer is the most common and lethal disease in the world. Although the total annual number of cases has declined, probably due to the decreased trend in cigarette consumption, the incidence and mortality rate of lung cancer have increased at an alarming rate in the female population and in developing countries. The overall cure rate of lung cancer is only about 13%, and most patients in whom therapy fails have distant metastases. During the neoplastic progression of the disease, lung cancer cells can invade surrounding tissue and then metastasize to regional and mediastinal lymph nodes. Ultimately, the cancer cells spread to distant sites including liver, bone, brain, and other organs (1).

Tumor-cell invasion and metastasis are the most difficult problems faced by oncologists. Unfortunately, no major clinical breakthrough that can efficiently prevent tumor-cell invasion has been made. Invasion, the most crucial step in the metastatic cascade, occurs via a series of biologic activities involving the interaction of tumor cells with the surrounding environment. This process can be subdivided into three steps: (1) attachment of tumor cells to the surrounding extracellular matrix (ECM); (2) production of matrix-degrading enzymes; and (3) migration of tumor cells through the degraded matrix (2). Tumor cells must initiate and successfully complete all of these steps in order to enter the circulatory and lymphatic systems to form distant metastases. In order to study mechanisms accompanying each invasive step, several in vitro invasion models have been developed and can be used to assess the invasive activity of various tumor cells (3).

Widespread metastasis is a common phenomenon in non-small-cell lung cancer. However, the behavior of lung adenocarcinoma is different from that of squamous lung cancer in that metastasis in the former type of tumor often occurs at an early stage, whereas that in the latter occurs relatively late. Therefore, it is clinically more important to elucidate the metastatic nature of lung adenocarcinoma than that of squamous carcinoma. Unfortunately, no good lung-tumor metastasis model is available for the mechanistic analysis of the metastatic potential of lung cancer cells. Metastasis has been viewed as a highly selective, competitive process that favors the outgrowth and survival of a metastatic subpopulation that preexists within the heterogeneous primary tumor (7). It is very likely that during subsequent cancer development, some tumor cells acquire favored phenotypes that give them an advantage in disseminating from the primary tumor and invading other areas. Therefore, comparative studies of phenotypes expressed in nonmetastatic and metastatic cancer cells will be helpful for the identification of genes responsible for tumor metastatic behavior (8, 9). On the basis of these considerations, we used the Transwell (Costar, Inc., Cambridge, MA) invasion chamber to isolate progressively metastatic cell subpopulations from a clonally established human lung adenocarcinoma cell line, CL1, which was developed in our laboratory several years ago (10). In the present study, we correlated the in vitro selection of several progressively invasive sublines and their metastatic potential in vivo. Preliminary characterization of these sublines demonstrated the enhanced expression of 92-kD gelatinase and keratins 8 and 18 in these progressively invasive sublines.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Cell Culture

The human lung cancer cell line CL1 was established from a 64-yr-old man with a poorly differentiated adenocarcinoma (10). The cells were maintained in RPMI-1640 supplemented with 10% fetal bovine serum (FBS) (Gibco BRL, Gaithersburg, MD). The CL1 cell line has been cloned and passed on for more than 60 generations. The xenograft experiment in the present study demonstrated tumor formation in nude mice, and histology of xenograft tumors also showed characteristics of adenocarcinoma.

Selection of Invasive Cells in Transwell Invasion Chamber

Subpopulations from CL1 cells were selected according to their differential invasiveness, using Transwell plates. Briefly, the polycarbonate membranes (containing 8-µm pores) of the Transwell inserts were coated with a reconstituted basement-membrane gel (Matrigel; Collaborative Research, Bedford, MA). Cells were resuspended in RPMI containing 10% NuSerum (Collaborative Research) and seeded into the wells. Following incubation for 72 h at 37°C, the inserts were removed. The cells that migrated through the membranes and attached to the lower-chamber compartments were harvested aseptically and expanded for second-round selection. The subline of the first-round selection was designated as CL1-1, and sublines from 2, 3, 4, and 5 rounds of selection were designated as CL1-2, -3, -4, and -5, respectively.

In Vitro Invasion Assay

The membrane invasion culture system (MICS) (6) was used to measure a cell line's invasive ability. A polycarbonate membrane containing 10-µm pores (Nucleopore Corp., Pleasanton, CA) was coated with a mixture of laminin (50 µg/ml; Sigma Chemical Co., St. Louis, MO), type IV collagen (50 µg/ml; Sigma), and gelatin (2 mg/ml; Bio-Rad, Hercules, CA) in 10 mM glacial acetic acid solution. The membrane was placed between the upper- and lower-well plates of the MICS chamber. Subsequently, cells were resuspended in RPMI containing 10% NuSerum and seeded into the upper wells of the chamber (5 × 10⁴ cells/ well). After incubation for 24 h at 37°C, cells that had invaded the coated membrane were removed from the lower wells with 1 mM ethylene diamine tetraacetic acid (EDTA) in phosphate-buffered saline (PBS), and dot-blotted on a 3-µm polycarbonate membrane. After fixation in methanol, blotted cells were stained with Liu stain (Handsel Technologies, Inc., Taipei, Taiwan) and the cell number in each blot was microscopically counted.

Soft Agar Colony-forming Assay

Exponentially growing cells were suspended in complete growth medium containing 0.3% Bacto-agar (Difco Laboratories, Detroit, MI) and overlaid on 1% agarose in 100-mm tissue-culture dishes (10³ cells/dish). The dishes were maintained at 37°C in a humidified incubator with 5% CO₂/95% air for 2 wk. The number of visible colonies was subsequently scored from quadruplicate replicates in order to determine mean values for colony-forming efficiency.

Tumorigenicity in SCID Mice

Six-week-old SCID mice were housed at four mice per cage in an isolator, and were ad libitum fed with autoclaved food. Tumor xenografts were grown by subcutaneously inoculating the dorsal region of each animal with 5 × 10⁶ cells/0.2 ml Hank's balanced salt solution (HBSS) (Gibco/BRL). Each tumor cell line was injected into five mice. These mice were then observed for 8 wk for the development of tumors.

Experimental Metastasis

Cells were washed and resuspended in HBSS. Subsequently, 4- to 6-week-old SCID mice were injected in the lateral tail vein with a single-cell suspension containing 10⁶ cells in 0.2 ml Hank's buffer. Mice were killed after 8 wk. All organs were examined for metastasis formation. The lungs were removed and fixed in 10% formalin fixative. The number of lung-tumor colonies was counted under a dissecting microscope. The representative lung tumors were removed, fixed, and embedded in paraffin, which was then sectioned into 4-µm layers and stained with hematoxylin and eosin (H&E) for histologic analysis.

Zymography for Gelatinase

Zymographic analysis of gelatinase activity in secreted medium was performed in 10% sodium dodecyl sulfate (SDS)-polyacrylamide gels containing 0.1% gelatin, as originally described by Heussen and Dowell (11). Cells were cultured in a 24-well tissue-culture plate (5 × 10⁵ cells/well) in RPMI containing 10% FBS. After 2 h, cells were washed extensively and changed to serum-free RPMI. After an overnight incubation, media were collected and mixed with Laemmli's SDS sample buffer (without -mercaptoethanol) for electrophoresis (12). The gels were then incubated for 30 min in 50 mM Tris-HCl containing 2.5% Triton X-100, followed by an incubation with metalloproteinase buffer (50 mM Tris-HCl, 150 mM NaCl, 10 mM CaCl₂, and 0.05% NaN₃) at 37°C for 24 h. After the gels were stained with Coomassie blue, they were destained in a 10% (vol/vol) methanol/10% (vol/vol) acetic acid solution until the transparent bands were shown on the blue background.

Immunofluorescence Staining

Cultured cells were grown on glass coverslips to 70% confluence. Samples were then washed three times with PBS and fixed in cold methanol-acetone (1:1, vol/vol) for 10 min. The fixed cells were washed with PBS, treated with 5% bovine serum to block any nonspecific binding, and labeled with either CK5 monoclonal antibody, which recognizes human keratin 18 (Sigma), or with control antibody. After incubation at 37°C for 1 h, cells were washed three times with PBS and then incubated with fluorescein-conjugated goat antimouse IgG (Organon Teknika Corp., Durham, NC) as the secondary antibody. One hour later, slides were washed at 37°C and examined under a Zeiss fluorescence microscope (Carl Zeiss, Jena, Germany). The CK5-positive cells were counted in three separate fields, and > 100 cells/field were counted for each slide.

Western Blot Analysis of Intermediate Filaments

For cytoskeletal protein isolation, cultured cells were washed twice with PBS and rinsed with cold lysis buffer (10 mM Tris, pH 7.6, 140 mM NaCl, 5 mM EDTA, and 0.5% Triton X-100). Following incubation with cold high-salt buffer (10 mM Tris, pH 7.6, 140 mM NaCl, 1.5 M KCl, 5 mM EDTA, and 0.5% Triton X-100) for 30 min, the cells were removed from the tissue-culture plates with a rubber policeman. The cell lysates were homogenized and centrifuged at 5,000 rpm for 40 min. The insoluble cytoskeletal pellets were washed with PBS and extracted with 0.5% SDS and 1% NP-40 in 10 mM Tris buffer. The protein concentration was determined with the D_C Protein Assay Kit (Bio-Rad). To detect intermediate filaments, equal amounts of cytoskeletal proteins from each cell line were separated with 12.5% SDS-polyacrylamide gel electrophoresis (PAGE) and transferred to an Immobilon membrane (Millipore, Bedford, MA). Proteins were detected with either anti-vimentin V9 (DAKO Corp., Santa Barbara, CA), CK5 (Sigma), or mouse antihuman cytokeratin 8 (DAKO Corp.) antibodies followed by use of a biotinylated antimouse secondary antibody (Vector Laboratories, Burlingame, CA). The protein bands were chromagenically stained with avidin and horseradish peroxidase (Vectastain ABC Kit, Vector Laboratories).

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Although the CL1 cancer cell line was initially established from a single-cell clone, it has probably become heterogeneous after long-term culture in vitro owing to the genetic instability of cancer cells. Thus, it is possible to isolate populations that have different characteristics from this cell line. To establish a lung-cancer metastasis cell model, CL1 cells were seeded onto Matrigel-coated Transwell-membrane(s). After a 72-h incubation period, the cells that had invaded Matrigel were collected as CL1-1, signifying one passage through the basement-membrane matrix. Subsequently, these cells were regrown and repeatedly passed four more times through the invasion-selection procedure. The cells harvested from each subsequent round of selection were designated CL1-2, CL1-3, CL1-4, and CL1-5, respectively. Figure 1 shows the morphologic changes in cells as seen under a phase-contrast microscope after invasion selection. The parental CL1 cells had a very typical epithelial-like cell morphology; they were flat with a triangular cell shape in a monolayer culture (Figure 1A). However, by the second selection, the cells were smaller and more rounded in culture as compared with CL1 cells (Figure 1B). In addition, the selected cells seemed less adhesive, and could be removed easily from tissue-culture dishes.

View larger version (72K): [in this window] [in a new window]   Figure 1.   Phase-contrast light microscopy of CL1 cells before (A) and after (B) invasion selection. Original magnification: ×900.

We measured the difference in invasiveness associated with each selection. The invasive potential was determined on the basis of cells' ability to invade a matrix barrier containing mainly laminin and type IV collagen, the major components of the basement membrane. For this study, the membrane was coated with purified basement-membrane components instead of Matrigel, since the latter contains impure contaminants that may interfere with the assay. Interestingly, the results obtained from five independent experiments showed that the invasive potential had increased by 4- to 6-fold for invasion-selected CL1 sublines as compared with the parental cells (Figure 2). Furthermore, data from these five independent experiments indicated that the invasive potential of these sublines was continuously maintained in the culture throughout this study.

View larger version (52K): [in this window] [in a new window]   Figure 2.   In vitro invasion activity of CL1 sublines. The invasion activity of each clone was measured in vitro with the MICS chamber as described in MATERIALS AND METHODS. Cells (5 × 104) were seeded on the matrix, and invaded cells were harvested after 24 h. Bar graphs represent average number of cells harvested ± SE. The numbers were: CL1: 202 ± 16; CL1-1: 777 ± 86*; CL1-2: 960 ± 136*; CL1-3: 808 ± 108*; CL1-4: 986 ± 111*; and CL1-5: 1,219 ± 243*. *P < 0.05 as compared with the data for CL1, analyzed by one-way ANOVA and the Student-Newman-Keuls test.

To further assess the tumorigenicity of the selected sublines, various in vitro and in vivo tumorigenic parameters were used. As shown in Table 1, anchorage-independent growth on soft agar was enhanced by 5- and 20-fold, respectively, in the CL1-2 and CL1-5 sublines as compared with the parental CL1 line. The in vivo tumorigenicity and metastatic potential of these selected sublines were also increased. It was shown that tumors formed rapidly when CL1-5 cells were injected subcutaneously into SCID mice, and within 3 wk the average tumor diameter was larger than 1 cm (Figure 3). Tumor growth in mice injected with CL1-2 cells was slower than in those injected with CL1-5 cells, but much faster than in mice injected with the parental CL1 cells. These tumorigenic characteristics also correlated well with the cells' metastatic potential in vivo. As shown in Table 2, the parental CL1 cells showed no evidence of producing lung metastasis at 8 wk after injection, whereas two of five and five of six mice injected with CL1-2 and CL1-5 cells, respectively, had obvious lung-tumor colonies. No metastatic tumors were found in other organs than the lung in this experimental metastasis study. Histologic analysis showed that lung metastases formed by CL1-5 cells had a typical adenocarcinoma tumor morphology in lung-tissue sections (Figure 4), resembling that of metastases formed by CL1 cells in our previous study (10).

View larger version (11K): [in this window] [in a new window]   Figure 3.   Tumorigenicity of lung-cancer cells in SCID mice. Six-week-old SCID mice were subcutaneously inoculated with 5 × 106 cells. Four mice were injected with each cell subline, and the average size of tumors formed was measured periodically for over 7 wk.

View larger version (138K): [in this window] [in a new window]   Figure 4.   Histologic analysis of CL1-5 lung metastasis. The lungs were removed, fixed in formalin, and embeded in paraffin. H&E staining was then performed on 4-µm-thick sections of lung tissue. A metastatic tumor is seen within the lung parachyma. Original magnification: ×40.

To further elucidate the phenotypic difference between these sublines and the parental one, we performed both zymographic and keratin protein analyses. Zymographic analysis was used to assess whether the invasive nature of the sublines correlated with their gelatinase activity. A 92-kD gelatinase activity could be observed with the sublines, with the strongest expression observed for gelatinase secreted by CL1-5 cells (Figure 5). In contrast, the parental CL1 cells did not show significant 92 kD gelatinase activity. Intermediate filaments have also been shown to play a role in cells' invasion and migration processes (13). Interestingly, SDS-PAGE revealed that all of the sublines and parental cells expressed vimentin (Figure 6A). However, only the CL1-2 to CL1-5 sublines expressed synthesis of keratin-8 and -18 (Figure 6B); Western blotting revealed no keratin-8 or -18 in either CL1 or the first-round selected subline CL1-1 cells. To further assess the biochemical data, we performed immunofluorescent staining with anti-keratin-18 antibody. Table 3 summarizes the quantitative data for keratin-18-positive cell populations in these cultures.

View larger version (88K): [in this window] [in a new window]   Figure 5.   Zymographic analysis of secreted gelatinase activity. The analysis of secreted gelatinase in the culture medium was performed as described in MATERIALS AND METHODS. Media were collected from CL1 (Lane 1), CL1-1 (Lane 2), CL1-2 (Lane 3), CL1-3 (Lane 4), CL1-4 (Lane 5), and CL1-5 (Lane 6), and from RPMI (Lane 7) as a negative control.

View larger version (38K): [in this window] [in a new window]   Figure 6.   Intermediate-filament protein levels in CL1 subclones. Cytoskeletal proteins were prepared with high salt buffer as described in MATERIALS AND METHODS. Equal amounts (3 µg) of isolated proteins from each cell clone were separated on 12.5% SDS-PAGE and transferred to an Immobilon membrane. Proteins were detected with anti- vimentin-V9 (A) and CK5 plus antihuman keratin-8 (B) antibodies followed by an antimouse secondary antibody. Lane 1: CL1; Lane 2: CL1-1; Lane 3: CL1-2; Lane 4: CL1-3; Lane 5: CL1-4; and Lane 6: CL1-5.

The preceding results indicate that the enhanced expression of keratin-8 and -18 seems to be associated with the invasive behavior of lung-tumor cells. To further characterize this correlation, 50 single-cell colonies were isolated from CL1-1 culture by plating cells at a very low cell density. Immunofluorescence staining of cells from each colony indicated that 10 of 50 colonies were keratin-18-positive, and that the remaining colonies were all keratin-18-negative. Three colonies from each population were randomly picked for further study. Figure 7A through C shows that colonies 7, 19, and 30 exhibited no localization for keratin-18 filaments, whereas colonies 23, 33, and 46 exhibited positive immunofluorescence, (as seen in Figure D through F). Interestingly, when their in vitro invasive ability was measured, keratin-18-negative cells (colonies 7, 19, and 30) demonstrated much lower invasiveness than the keratin-18-positive cells (colonies 23, 33, and 46; Figure 8).

View larger version (96K): [in this window] [in a new window]   Figure 7.   CL1-1 subclones stained for keratin-18 filaments. C-7 (A), C-19 (B), C-30 (C), C-23 (D), C-33 (E) and C-46 (F ) cells were labeled with antibody against human keratin-18, with fluorescein-conjugated goat antimouse IgG as a second antibody. Original magnification: ×1,730.

View larger version (47K): [in this window] [in a new window]   Figure 8.   Invasive abilities of CL1-1 subclones. The invasive ability of each clone was measured in vitro with the MICS chamber as described in MATERIALS AND METHODS. Cells (5 × 104) were seeded on the matrix, and invaded cells were harvested after 24 h. Bar graphs represent average number of cells harvested ± SE. The numbers of cells were: C-7: 113 ± 17; C-19: 93 ± 16; C-30: 107 ± 41; C-23: 848 ± 10; C-33: 828 ± 71; and C-46: 593 ± 112.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

Tumor-cell metastasis is a major problem in cancer therapy. Unfortunately, little is known regarding what factors may trigger tumorigenic cells to initiate further invasive and metastatic processes. Some tumor cells have high invasive potential whereas others do not. There may be one or more controls in the invasive tumors that serve as signals to initiate the cascade of invasion and metastasis during tumor progression. Comparison of the phenotypic differences associated with the differential invasive and metastatic ability of tumor cells may enhance our understanding of the mechanism(s) underlying cancer metastasis. The ultimate goals of the present study were to identify potential markers that could predict the possibility of lung-tumor invasion and metastasis, and to develop methods that might lead to the prevention of tumor metastasis on the basis of this information. For this goal, we first tried to develop an in vitro method for the isolation of invasive and metastatic subpopulations from a human lung adenocarcinoma cell line. In vitro selection of a highly metastatic cell population has been developed, but its application is mainly for melanoma cells (9). However, most lung cancers are of epithelial origin, and none of these studies are related to human lung carcinoma. Using Transwell chambers, we were able to systematically select invasive and metastatic subpopulations from the human lung adenocarcinoma cell line CL1, a clonal cell line derived from human lung adenocarcinoma tissues (10). After 60 passages, this cell line continues to maintain tumorigenic characteristics when implanted into nude mice. With several passages of CL1 adenocarcinoma cells through the Transwell chamber, we isolated several variants with differences in metastatic potential. Evidence for the metastatic potential of these sublines in vivo was demonstrated in experiments with SCID mice. Interestingly, this selection also coordinately isolated highly anchorage-independent subpopulations with enhanced expression of 92-kD gelatinase activity and keratins 8 and 18. Thus, this in vitro selection process provides a useful approach to isolating cell variants with different degrees of malignancy, which can be used to assess properties associated specifically with lung-tumor invasion and metastasis.

Of further interest is that this in vitro invasion selection process concomitantly identified both more invasive cells and cells with increased expression of 92-kD gelatinase and keratin-8 and -18 intermediate filaments. Using a zymography gel, we were able to observe a 92-kD gelatinase activity associated with the first selected variant cell line, CL1-1. In contrast, this gelatinase activity was not observed in the parental CL1 culture. This result is consistent with the notion that the invasiveness of cancer cells depends on the development of proteases that are capable of digesting the components of the ECM. The invasion chamber coated with Matrigel is apparently useful in the selection of these variants with high gelatinase activity. A similar concept, with different coating components, may apply to the selection of cell variants with different metastatic characteristics.

Tumor invasion and metastasis are very complex processes. A variety of proteins, including integrins, metalloproteinases, motility factor receptor, and ECM components, have been implicated in regulation of the invasive and metastatic potential of a cancer cell type (16). The present study has demonstrated a correlation between enhanced keratin-8 and -18 protein expression and the invasiveness of variant sublines of the CL1 cell line of human lung adenocarcinoma. We have observed that most parental CL1 cells did not show keratin-18 filaments upon immunofluorescence staining, and that only a very few cells (1.98% of the total cell population) yielded positive localization. After one round of selection from the in vitro invasion chamber, the keratin-18-positive cell population increased to 15.86%. This cell population increased 90 to 100% in the subsequent round of selection. The enrichment of keratin-18-positive cells associated with selection suggests that lung-cancer cells containing keratin-18 filaments may constitute a more invasive tumor population. To further corroborate this association, clonal cultures of keratin-18-positive and -negative cells were prepared from the CL1-1 culture, and all positive cells again showed higher invasive rates than did the negative cells.

The expression of intermediate filaments in normal cells is in general highly regulated. Therefore, these filaments are usually used as markers for typing cell-lineage and -differentiation stages. In this study, we observed the expression of two different intermediate filaments (vimentin and keratin) in the CL1 adenocarcinoma cell line and its more invasive sublines. The expression of vimentin in epithelial cell lines could have been an artifact of the cell-culture conditions. However, the expression of keratin-8 and -18 is highly regulated and was related to the invasiveness of the cells we investigated. Normally, keratins-8 and -18 are paired, and their expression serves as a marker specific for simple epithelium. However, several reports have revealed the anomalous expression of this keratin pair in premalignant and malignant epithelial lesions (17). Recently, Schaafsma and coworkers (20) found that keratin-8 and -18 positive tumor cells of mucosal squamous cell carcinoma and urinary tract carcinoma are more likely to localize at the tumor invasion front. These results are consistent with the findings in our study, suggesting a potential role of keratin-8 and -18 in the development of invasive cancer cells. Schaafsma and coworkers (21) further suggested that the acquisition of keratin-8 and -18 by the invasive tumor cells may represent an adaption to a more flexible and motile cell phenotype. This hypothesis is supported by the observation that keratin-8 and -18 are coordinately expressed in umbrella cells when these cells become flexible during expansion of the urinary bladder. Additionally, Robey and coworkers (23) observed the association of keratin-18 expression with the migration of human retinal pigment epithelial cells. We have demonstrated enhanced invasion by mouse L cells (15) and human melanoma cells (14) after their transfection with vector DNAs that express both keratin-8 and -18. The increased invasiveness of transfected cells is apparently related to an increase in their motility and a decrease in cell spreading on ECM; two important events in the invasion cascade. These in vitro and in vivo results collectively indicate that keratin-8 and -18 may play an important role in tumor-cell invasion by increasing cell flexibility and motility.

It is necessary to point out that the parental CL1 cell line used in our study is not highly metastatic. Despite the clonal nature of this cancer cell line, our results demonstrated that subpopulations of metastatic and highly tumorigenic cells could be developed from it after 60 passages. This change in metastatic and tumorigenic nature is probably related to the intrinsic genetic instability associated with many cancer cells. Yet even with very few metastatic cells in CL1, our in vitro invasion selection procedure was able to isolate these variants. It is possible that a similar application to identifying metastatic variant cells can be extended to cells derived from primary tumors. Such an approach would be clinically important in assessing the malignancy of tumors from biopsy specimens (16).