The JI
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     
 

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Liu, K.
Right arrow Articles by Abrams, S. I.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Liu, K.
Right arrow Articles by Abrams, S. I.
The Journal of Immunology, 2003, 171: 4164-4174.
Copyright © 2003 by The American Association of Immunologists

Exposure of Human Primary Colon Carcinoma Cells to Anti-Fas Interactions Influences the Emergence of Pre-existing Fas-Resistant Metastatic Subpopulations

Kebin Liu, Elwood McDuffie and Scott I. Abrams1

Laboratory of Tumor Immunology and Biology, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda MD, 20892


   Abstract
Top
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 References
 
Fas, an important death receptor-mediated signaling pathway, has been shown to be down-regulated during human colon tumorigenesis; however, how alterations in Fas expression influence the metastatic process remains unresolved. In mouse models, loss of Fas function was found to be both necessary and sufficient for tumor progression. In this study, we investigated the link between functional Fas status and malignant phenotype using a matched pair of naturally occurring primary (Fas-sensitive) and metastatic (Fas-resistant) human colon carcinoma cell lines in both in vitro and in vivo (xenograft) settings. Metastatic sublines were produced in vitro from the primary tumor cell line by functional elimination of Fas-responsive cells. Conversely, sublines derived from the primary tumor in vivo at distal metastatic sites were Fas-resistant. In contrast, simply disrupting the Fas pathway by molecular-based strategies in the Fas-sensitive primary tumor failed to achieve the same metastatic outcome. Interestingly, both in vitro- and in vivo-produced sublines resembled the naturally occurring metastatic population, based on functional and morphologic studies and genome-scale gene expression profiling. Overall, using this human colon carcinoma model, we: 1) showed that loss of Fas function was linked to, but alone was insufficient for, acquisition of a detectable metastatic phenotype; 2) demonstrated that metastatic subpopulations pre-existed within the heterogeneous primary tumor, and that anti-Fas interactions served as a selective pressure for their outgrowth; and 3) identified a large set of differentially expressed genes distinguishing the primary from metastatic malignant phenotypes. Thus, Fas-based interactions may represent a novel mechanism for the biologic or immunologic selection of certain types of Fas-resistant neoplastic clones with enhanced metastatic ability.


   Introduction
Top
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 References
 
It is generally well accepted that as the neoplastic process becomes more aggressive, resulting subpopulations of tumor cells tend to exhibit a more apoptotic-resistant phenotype. This may include resistance not only to gamma-irradiation or chemotherapeutic agents (1, 2) but also, perhaps, to innate or adaptive host defense immune mechanisms that trigger cell death through common intracellular signaling elements (3, 4, 5). Although considerable information is now being accumulated about the genetic events underlying the transformation of a normal cell or benign lesion into a malignant one, such as in the case of human colon carcinoma (1, 6, 7), much remains to be understood regarding the cellular and molecular bases for the progression of a solid primary neoplasm to a clinically relevant metastasis (8).

The Fas/Fas ligand (FasL) 2 system has been characterized as an important physiologic mechanism for the regulation of homeostasis in both immune and nonimmune cellular compartments (4, 5, 9). In humans, Fas is constitutively expressed at high levels essentially along the entire coloncytic compartment, as determined by immunohistochemistry (10, 11). In colon carcinoma, however, the pattern of Fas expression is altered. In over one-third of colon carcinoma tissues examined, the level of Fas expression was reported diminished, whereas the complete loss of detectable Fas expression was much more frequent in metastatic lesions as compared with primary lesions. These observations suggest that in cases of malignant transformation of normal colonic epithelium, there seems to be a pattern of progressive down-modulation of Fas expression. Nevertheless, the biologic significance of loss of Fas function and the mechanisms regulating these phenotypic or functional outcomes are still unclear.

Thus, in an effort to better explore the link between functional Fas status and malignant phenotype in human colon carcinoma, we took advantage of two naturally occurring primary and metastatic cell lines, termed SW480 and SW620. The SW480 and SW620 tumor cell lines have been previously characterized as primary and metastatic colon adenocarcinoma cell lines, respectively, established from the same patient (12). The SW620 cell line was derived as a lymph node-metastasis identified 6 mo later during disease recurrence. Furthermore, both cell lines were isolated from the patient without any prior chemotherapy (13). Therefore, the availability of a matched pair of primary and metastatic colon adenocarcinoma cell lines allowed us to investigate initially potential differences in their lytic susceptibility to HLA-A2-restricted, Ag-specific CD8+ CTL (14). We found that treatment of both SW480 and SW620 cell lines with IFN-{gamma} was necessary to achieve efficient Ag-specific lysis; however, the mechanisms leading to lysis of these two targets were distinct, which reflected differential levels of sensitivity to the Fas pathway. Whereas IFN-{gamma}-pretreatment rendered SW480 cells sensitive to both Fas- and perforin-based pathways, SW620 cells displayed lytic susceptibility to perforin, but not Fas-mediated death (14).

In this study, we investigated whether such differences or alterations in functional Fas status influenced tumor progression toward a more metastatic phenotype. To address this aim, the approach taken was to produce sublines from the primary tumor in vitro for Fas resistance and in vivo in nude mice for enhanced metastatic competence. Thus, any functional and/or molecular differences observed with such SW480-derived sublines or variants could be compared with the naturally occurring primary and metastatic tumor cell lines. Overall, we established an inverse correlation between functional Fas status and metastatic capacity; albeit, loss of Fas function alone was insufficient for acquisition or detection of a productive metastatic outcome. In fact, in addition to alterations in Fas function, we identified a large set of differentially expressed genes distinguishing the primary from metastatic tumor cell lines. Moreover, we revealed experimentally for the first time that such Fas-resistant metastatic subpopulations already pre-existed within the primary tumor population and that they morphologically, functionally, and molecularly resembled the naturally occurring metastatic cell line. These data thus support the hypothesis that Fas-based interactions mechanistically impose an immunologic or biologic selective pressure favoring the outgrowth of certain pre-existent metastatic subpopulations or clones, perhaps, conceptually analogous to how radiation or chemotherapeutics might select for radio- or chemoresistant neoplastic cells.


   Materials and Methods
Top
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 References
 
Cell lines

The HLA-A2+ SW480 (CCL-228) and SW620 (CCL-227) colon adenocarcinoma cell lines (12) were obtained from the American Type Culture Collection (ATCC; Manassas, VA). As described in the ATCC product sheets, the SW480 and SW620 cell lines were acquired by the ATCC shortly after their establishment from the patient and have been cryopreserved since then. SW480.sel1 was a subline derived from SW480 cells by in vitro selection for Fas resistance in response to serial coculture with an agonistic anti-Fas mAb, clone CH-11 (15) (Immunotech, Westbrook, ME), as described (14). Briefly, SW480 cells were plated at 5 x 105/T-25 flask, and allowed to adhere overnight. IFN-{gamma} (250 U/ml) was added the next day, followed 24 h later by CH-11 (1 µg/ml). At weekly intervals, cells were recovered and recultured under these same conditions for three to four additional cycles of IFN-{gamma} plus CH-11 treatment. The cells (termed SW480.sel1) were then maintained and propagated in the absence of IFN-{gamma} and CH-11. The SW480.spl subline was also derived from SW480 cells, but as a spontaneous distal splenic metastasis in nude mice following a s.c. tumor transplant. In this case, SW480 cells (4 x 106 cells/mouse) were injected into the right flank of female athymic (nu/nu) mice (National Cancer Institute-Frederick Cancer Research Animal Facility, Frederick, MD). Spleens were collected 4–5 wk later when average s.c. tumor volume exceeded 800 mm3. Spleens were processed through a nylon cell strainer (70-µm filter; BD Biosciences, Franklin Lakes, NJ) to produce a single cell suspension for cell culture. Tumor cell colonies were grossly demonstrable within 4–5 wk of seeding in T-150 flasks, and were recovered for subsequent reculturing. All cell lines were Mycoplasma-negative as determined by PCR analysis using the Mycoplasma Detection kit from the ATCC.

Tumor growth in nude mice

Tumor cell lines or sublines (4 x 106 cells/mouse in 0.1 ml HBSS) were inoculated s.c. into athymic (nu/nu) mice in the right flank, as previously described. Tumor growth was monitored by digital caliper in two dimensions, and the volumes were calculated according to the formula of (width-squared x length)/2. Approximately 4–5 wk later, spleens were isolated, and single-cell suspensions were prepared from each mouse separately. One-third to one-half of the splenocyte preparation from each mouse was used for total RNA isolation (see below), and the remainder was cultured in T-75 or T-150 flasks. Adherent cell colonies (SW480.spl) were collected and maintained in cell culture. All animal studies were approved by the appropriate institutional review boards.

Cell surface marker analysis

Tumor cells, either untreated or pretreated overnight (18–24 h) with recombinant human IFN-{gamma} (sp. act. 2.4 x 107 U/mg, 250 U/ml; Biogen, Cambridge, MA), were incubated with mAb directed against Fas (clone DX-2; BD PharMingen, San Diego, CA), followed by washing and incubation with affinity-purified, FITC-conjugated goat anti-mouse IgG (Kirkegaard & Perry Laboratories, Gaithersburg, MD) and analysis by flow cytometry. Isotype-matched Abs served as negative controls.

Measurement of apoptotic cell death

Apoptotic cell death was measured by TUNEL assay (16, 17) and genomic DNA fragmentation analysis. Untreated or IFN-{gamma}-pretreated cells were incubated at 37°C for 18–24 h in the absence or presence of CH-11 (1 µg/ml) or an isotype-matched (IgM) Ab (MOPC-104E; ICN Biomedicals, Aurora, OH) as a negative control, and then washed in buffered saline. For TUNEL staining, an apoptosis detection kit (R&D Systems, Minneapolis, MN) was used according to the manufacturer’s protocol. For DNA fragmentation analysis, genomic DNA was isolated from the cells using the Easy-DNA kit (Invitrogen, Carlsbad, CA) according to the manufacturer’s protocol. Isolated DNA was then separated on a 2.5% agarose gel, stained with an ethidium bromide solution, and visualized under UV light.

RT-PCR analysis

Total RNA was isolated from cells (tumor cells or splenocytes) using RNA STAT-60 reagent (Tel-Test, Friendswood, TX) according to the manufacturer’s instructions and used for the first strand cDNA synthesis using the ThermoScript RT-PCR system (Invitrogen). The cDNA was then used as a template for PCR amplification of human keratin-18, human Fas, or human or mouse {beta}-actin. The following parameters were used: 30 s at 94°C, 30 s at 60°C, and 1 min at 72°C for 30 cycles. The coding sequences of keratin-18 of human and mouse were aligned, and PCR primers were chosen from the regions with the most divergent coding sequences by computer program (MacVector 7.0; Oxford Molecular, Cambridge, U.K.). The PCR primers for human keratin-18 were as follows: forward primer: 5'-GCCTACAAGCCCAGATTGCC-3'; reverse primer: 5'-GGTGGTCTTTTGGATGGTTTGC-3'. The PCR primers for human Fas were as follows: forward primer: 5'-ATTATCGTCCAAAAGTGTTAAT-3'; reverse primer: 5'-TGCATGTTTTCTGTACTTCCTT-3' (18). The PCR primers for human {beta}-actin were as follows: forward primer: 5'-ATGGATGATGATATCGCCGCG-3'; reverse primer: 5'-CTAGAAGCATTTGCGGTGGAC-3'. The PCR primers for mouse {beta}-actin were as follows: forward primer: 5'-ATTGTTACCAACTGGGACGACATG-3'; reverse primer: 5'-CTTCATGAGGTAGTCTGTCAGGTC-3'.

Stable transfection of SW480 cells with virally encoded FLICE-inhibitory protein (vFLIP)

SW480 single cell clones were produced by limiting dilution. A Fas-sensitive clone, termed SW480.C1, was used for transfection with the expression vector pEGFPN1 (Clontech Laboratories, Palo Alto, CA) containing the genes for vFLIP (19) and green fluorescent protein (GFP), kindly provided by Dr. R. Siegel (National Institutes of Health, Bethesda, MD). The vector control plasmid (i.e., lacking vFLIP) and the vector containing the vFLIP coding sequence were then linearized with AflII restriction enzyme and used for transfection. Transfections were performed using LipofectAMINE 200 reagent (Invitrogen) according to the manufacturer’s instructions. The transfected cells were propagated in culture medium containing Geneticin (Invitrogen) at a concentration of 0.75 mg/ml for 7 days, and recovered and recultured under the same conditions for two more passages before being sorted by a FACSVantage SE cell sorter (BD Biosciences) based on GFP intensity. The sorted cells were cultured with Geneticin for another 7 days, and resorted once more to ensure stable retention of GFP-positive cells. The sorted cells were then maintained and propagated under Geneticin selection.

Global gene expression profiling

Total RNA was isolated from cells using RNA STAT-60 reagent (Tel-Test). mRNA was isolated from total RNA by oligo(dT) beads (Dynal Biotech, Lake Success, NY) and used for cDNA probe preparation. In general, mRNA (0.8 µg) was used for each cDNA probe preparation. cDNA probes were synthesized using the FairPlay microarray labeling kit (Stratagene, La Jolla, CA). The cDNA probes made from mRNA isolated from SW620 cells and SW480.sel1 and SW480.spl sublines were then labeled with Cy5 monofunctional reactive dye, and cDNA probe made from mRNA isolated from SW480 was labeled with Cy3 monofunctional dye (both dyes from Amersham Biosciences, Piscataway, NJ). The appropriate Cy3- and Cy5-labeled probes were combined, along with 10 µg Cot-1 DNA (Invitrogen), 4 µg yeast tRNA and 10 µg Poly(dA) in a final volume of 15 µl, and incubated at 98°C for 1 min. The denatured probes were mixed with 15 µl 2x hybridization buffer (50% formamide, 10x SSC, and 0.1% SDS). The hybridization solution and cDNA probe mixtures were added to the processed National Cancer Institute human cDNA microarray slides, which were then placed in hybridization chambers and incubated at 43°C for 16 h. The slides were then washed for 5 min in 2x SSC and 0.1% SDS; for 5 min in 1x SSC; for 5 min in 0.2x SSC; for 5 min in 0.05x SSC, then spin dried. Fluorescence images were captured using a Genepix 4000 (Axon Instruments, Union City, CA). Both image and signal intensity data were loaded into a database supported by the Center for Information Technology of the National Institutes of Health. Cy3:Cy5 intensity ratios from each gene or cDNA clone were calculated and subsequently normalized to ratios of overall signal intensity from the corresponding channel in each hybridization. The normalized data were then extracted from the database as text files and analyzed using computer software JMP (SAS Institute, Cary, NC) to compare the gene expression profiles quantitatively. For clustering analysis, Cluster and TreeView programs (20) were used to analyze the gene expression patterns in a one-dimensional hierarchical clustering to generate gene dendrograms based on the pair-wise calculation of the Pearson coefficient of normalized fluorescence ratios as measurements of similarity and linkage clustering. The clustered data were loaded into TreeView and displayed by the graded color scheme.


   Results
Top
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 References
 
Differential expression of Fas and sensitivity to Fas-mediated apoptosis by the primary and metastatic colon carcinoma cell lines

First, we compared SW480 and SW620 cells for expression of Fas by flow cytometry and RT-PCR analysis (Fig. 1). We found that SW480 cells constitutively expressed higher levels of Fas, as compared with SW620 cells (Fig. 1A, a and b). We also examined the Fas expression levels of these two tumor cell lines after treatment with IFN-{gamma}, because it has been reported that IFN-{gamma} enhances cell surface Fas on neoplastic cells (14, 21). After IFN-{gamma} treatment, both tumor cell lines displayed heightened levels of Fas, although IFN-{gamma}-treated SW480 cells still expressed a higher amount (Fig. 1A, a and b). Similar patterns were observed by RT-PCR analysis of untreated and IFN-{gamma}-treated preparations for the intensity of expression of the Fas transcript (Fig. 1B).



View larger version (42K):
[in this window]
[in a new window]
  		FIGURE 1. Fas expression and function by SW480 cells and SW620 cells. A, Cell surface expression levels of Fas in SW480 (a) and SW620 (b) cells as measured by flow cytometry. The isotype control Ab indicated by the shaded area. Untreated cells (dashed histogram) and IFN-{gamma}-treated cells (solid histogram) are represented. Percentage of positive cells (with mean fluorescence intensity values in parenthesis) are given for: SW480/-IFN-{gamma}, 63% (19); SW480/+IFN-{gamma}, 91% (62); SW620/-IFN-{gamma}, 5% (14); and SW620/+IFN-{gamma}, 88% (27). B, Agarose gel images of Fas transcript levels in SW480 and SW620 cells. Tumor cells, either untreated or pretreated with IFN-{gamma}, were harvested for isolation of total RNA that was then used for cDNA synthesis and PCR analysis using human Fas- and {beta}-actin-specific primers. C, Agarose gel electrophoresis reveals DNA fragmentation of untreated or IFN-{gamma}-pretreated cells after incubation with either CH-11 or an isotype-matched control Ab (1 µg/ml). The cell lines and treatment conditions are indicated at the top of the gel panel. D, Tumor cells were treated as in C, and apoptotic death induced by CH-11 (solid thin line) or an isotype-matched control Ab (shaded area) was measured by TUNEL assay. Untreated tumor cells (left histogram) and IFN-{gamma}-pretreated tumor cells (right histogram) with SW480 (a and b) and SW620 (c and d) are shown.

 
To determine whether there was a correlation between Fas expression levels and sensitivity to Fas-mediated apoptosis, we examined functional Fas status in response to CH-11, and the extent of apoptotic death was analyzed by agarose gel electrophoresis for detection of DNA fragmentation (Fig. 1C) and TUNEL staining (Fig. 1D). In the absence of IFN-{gamma} treatment, SW480 cells, but not SW620 cells, showed a weak response toward CH-11, based on detectable DNA laddering within the gel (Fig. 1C). Furthermore, the extent of DNA degradation of SW480 cells induced by CH-11 was markedly enhanced following IFN-{gamma} treatment. Specificity of the CH-11-mediated effects was revealed by the inability of an isotype-matched control Ab to trigger detectable DNA degradation (Fig. 1C). In contrast to what was observed with SW480 cells, pretreatment of SW620 cells with IFN-{gamma} had little, if any, effect on their subsequent sensitivity to CH-11 (Fig. 1C). Moreover, similar to that observed by DNA agarose gel electrophoresis, pretreatment of SW480 cells with IFN-{gamma}, but not of SW620 cells, led to enhanced Fas-mediated apoptosis in response to CH-11, as determined by TUNEL staining (Fig. 1D). In the absence of IFN-{gamma} treatment, SW480 cells showed a marginal apoptotic response toward CH-11 as compared with exposure to the isotype-matched control Ab (Fig. 1Da). However, the percentage of apoptotic SW480 cells induced by CH-11 was substantially increased after IFN-{gamma} treatment (Fig. 1Db). In contrast to what was observed with SW480 cells, pretreatment of SW620 cells with IFN-{gamma} had no effect on their subsequent sensitivity to CH-11 (Fig. 1, Dc and Dd), despite the fact that Fas expression in SW620 cells was up-regulated by IFN-{gamma} (Fig. 1, A and B).

Generation of primary tumor sublines that morphologically resemble the naturally occurring metastatic cell population

Although the general consensus is that metastatic cells originate from the primary tumor (1, 8), it remains to be fully understood what intrinsic and/or extrinsic selective pressures or factors might influence their outgrowth. Because down-regulation of Fas has been reported as a common occurrence of an advancing colonic neoplastic phenotype (10, 11), it is possible that this may represent not only a potential mechanism of tumor escape, but also a biologic basis of selection and emergence of certain Fas-resistant and more aggressive subpopulations. The observation that the naturally occurring primary and metastatic tumor cell lines were differentially Fas responsive (Fig. 1) allowed us to investigate the potential interrelationship between functional Fas status and malignant phenotype in this model system. Thus, if metastatic subpopulations already pre-exist within the SW480 primary tumor cell line, and if loss of Fas function impacts malignant progression, then Fas-resistant subpopulations with enhanced metastatic potential should emerge from the parental primary tumor population. To address that hypothesis, sublines were produced from the primary tumor in vitro for Fas resistance. Conversely, sublines were produced from the primary tumor in vivo (in nude mice) from sites of spontaneous distal splenic metastases (following a s.c. tumor transplant). Such SW480-derived sublines were then analyzed and compared with the naturally occurring primary and metastatic tumor cell lines for functional and/or molecular similarities and differences. In so doing, we established two SW480 sublines, SW480.sel1 and SW480.spl, by these two approaches, respectively. (In fact, two different in vitro-derived and three different in vivo-derived SW480 sublines were established from three independent mice, although similar results were observed with all sublines.)

The SW480 and SW620 cell lines, as well as the SW480.sel1 and SW480.spl sublines, were examined under the light microscope for gross morphology. In cell culture, the morphology of SW480 and SW620 cells appeared to be largely distinct. SW480 cells were larger and exhibited an epithelial-type appearance (Fig. 2a), whereas SW620 cells were smaller and displayed a rounder, fibroblast or spindle-shaped morphology (Fig. 2b). Interestingly, the SW480.sel1 and SW480.spl sublines appeared to display a smaller, fibroblast-like appearance consistent with that of the naturally occurring SW620 cell line (Fig. 2, d and e). In an effort to better appreciate the transformation or evolution of the SW480 cell line in response to CH-11 selection, Fig. 2c portrays the morphology of those cells just after two (of four) rounds of anti-Fas mAb exposure. The resulting cell population appeared to represent an intermediate phenotype or heterogeneous mixture of both SW480 and SW620 cell types. These observations highlight that CH-11-mediated selection is a dynamic process, impacting the balance and outgrowth of one (or more) cell types over another. By four cycles of CH-11 treatment, the resultant population appeared to be more uniformly SW620-like (Fig. 2d). The fact that SW620-like cells seemed to overrun and eventually consume the composition of the culture under such long-term conditions of anti-Fas exposure further supported the notion that such cells are Fas-resistant. Thus, the in vitro- and in vivo-selected SW480 sublines morphologically resembled SW620 cells more so than the parental SW480 cells.



View larger version (123K):
[in this window]
[in a new window]
  		FIGURE 2. Morphologic analysis and comparison of SW480 cells, SW620 cells, and SW480-derived sublines. SW480 (a) and SW620 (b); SW480 (c) after two rounds of CH-11 selection (as described in Materials and Methods); SW480.sel1 (d) after four rounds of CH-11 selection; and SW480.spl (e) are shown. Magnification was x200.

 
Fas expression and function by the primary tumor sublines

Because the two SW480-derived sublines morphologically resembled the naturally occurring SW620 metastatic cell line, we next compared them to SW620 cells for expression of Fas by flow cytometry and RT-PCR analysis (Fig. 3). In comparison with SW620 cells, we found that both SW480.sel1 and SW480.spl sublines expressed similar levels of the Fas transcript, as well as cell surface Fas with or without IFN-{gamma} treatment (Fig. 3, A and B). To examine their sensitivity to Fas-mediated apoptosis, untreated and IFN-{gamma}-pretreated SW620 cells and the two sublines were incubated with CH-11, and the extent of apoptotic death was determined by agarose gel electrophoresis (Fig. 3C) and TUNEL staining (Fig. 3D). Functionally, the two sublines were essentially resistant to Fas-mediated apoptosis. Thus, in terms of Fas expression and function, these data also supported the notion that such SW480-derived sublines resembled the SW620 cell line more so than the parental SW480 cell line from which they were derived.



View larger version (32K):
[in this window]
[in a new window]
  		FIGURE 3. Comparison of Fas expression and function between SW620 cells and SW480-derived sublines. A, Agarose gel images of Fas transcript levels in SW620, SW480.sel1, and SW480.spl cells. RT-PCR analysis was performed as in Fig. 1. B, Cell surface expression levels of Fas in SW620 (a), SW480.sel1 (b) and SW480.spl cells (c) are shown, as measured by flow cytometry. Staining with the isotype control Ab is indicated by the shaded area. Untreated cells (dashed histogram) and IFN-{gamma}-treated cells (solid histogram) are represented. Percentage of positive cells (with mean fluorescence intensity values in parenthesis) are given for: SW480.sel1/-IFN-{gamma}, 5% (5); SW480.sel1/+IFN-{gamma}, 82% (20); SW480.spl/-IFN-{gamma}, 6% (7); SW480.spl/+IFN-{gamma}, 79% (27); and data for SW620 are the same as in Fig. 1. C, Agarose gel images of DNA fragmentation of untreated or IFN-{gamma}-pretreated tumor cells as performed in Fig. 1. The cell lines and treatment conditions are indicated at the top of the gel panel. D, Tumor cells were treated as in C, and apoptotic death induced by CH-11 (solid thin line) or an isotype-matched control Ab (shaded area) was measured by TUNEL assay. Untreated tumor cells (left histogram) and IFN-{gamma}-pretreated tumor cells (right histogram) with SW620 (a), SW480.sel1 (b), and SW480.spl (c) are shown.

 
Genome-scale gene expression analysis of the various colon carcinoma populations

To better elucidate the genetic relationship of these sublines with SW480 or SW620 cells, cDNA microarray studies were conducted to profile the general gene expression patterns of two SW480 sublines and SW620 cells using SW480 cells as a comparative reference (Fig. 4). Our cDNA microarray chip contained 9128 nonredundant cDNA clones. Three independent experiments for each cell line were analyzed. Reproducibility between replicated experiments was strong, because the R2 value of linear regression for duplicated experiments, for example, was ~0.9 (Fig. 4A).



View larger version (18K):
[in this window]
[in a new window]
  		FIGURE 4. Genome-scale gene expression profiles among the different colon carcinoma cell lines by cDNA microarray analysis. A, Scatter plots, comparing the reproducibility of gene expression intensities between two replicate experiments: 9128 nonredundant cDNA clones were examined in each experiment. The coefficiency of duplicate linear regressions (R2) for each cell line is shown. Each axis represents the log ratio value of each cDNA clone for each cell line, as compared with SW480 cells. The center red lines indicate an identical ratio value for the two replicated experiments. B, Diagram of complete-linkage hierarchical clustering and array clustering of SW620, SW480.sel1, and SW480.spl cells based on their gene expression profiles of 9128 genes. Each colored square represents the mean ratios of each gene in each cell line over SW480 cells taken from three independent measurements. The scale of intensity ratio is in log units, with red indicating an increase and green indicating a decrease. C, The number of genes (or cDNA clones) that were differentially expressed between SW620, SW480.sel1, and SW480.spl cells vs SW480 cells are shown. The genes that were highly expressed in SW620, SW480.sel1, and SW480.spl in all three replicate experiments are represented (+). The genes that were highly expressed in SW480 in all three replicate experiments are represented (-), and the remainder of the genes are indicated (0). The left column indicates all the possible patterns of gene expression between SW620, SW480.sel1, and SW480.spl cells vs SW480 cells. The right column indicates the number of genes (or cDNA clones) in each category. D, The ratio plot of differentially expressed genes between SW620, SW480.sel1, and SW480.spl cells vs SW480 cells is shown. The intensity ratios of genes highly expressed in SW620, SW480.sel1, and SW480.spl cells (n = 1315) and of genes highly expressed in SW480 cells (n = 715) were averaged from three independent experiments. The mean ratios are indicated with dark solid lines.

 
Fig. 4B illustrates both complete-linkage hierarchical clustering and array clustering of the expression patterns of the 9128 genes in SW620, SW480.sel1, and SW480.spl cells based on their gene expression levels as compared with SW480 cells. Overall, the global gene expression patterns of SW620, SW480.sel1, and SW480.spl cells were essentially the same (Fig. 4B). We next analyzed the differentially expressed genes of SW620, SW480.sel1, and SW480.spl cells vs SW480 cells in more detail, using an arbitrary cutoff of 0.3 of the log2 ratio value (1.24-fold) of each comparison. A relatively less stringent cutoff was chosen for this analysis because this cutoff would generate a larger pool of genes and would give a better representation of the general expression patterns or molecular signatures. By using this cutoff criterion, any gene that had a log2 ratio of >=0.3 in all three replicate experiments was considered to be expressed higher in SW620 cells and/or the two SW480 sublines, and conversely, any gene that had a log2 ratio of <=-0.3 in all three replicate experiments was considered to be expressed higher in SW480 cells. Of the 9128 genes analyzed, collectively 2151 (23.6%) were considered differentially expressed between SW480.sel1, SW480.spl, and/or SW620 vs SW480 cells (Fig. 4C). Interestingly, of the 2151 genes, 2027 (94.2%) fell into two sets: 1) 1315 genes that were highly expressed in both SW480 sublines and/or SW620 cells, as compared with SW480 cells; and 2) 712 genes whose expression levels were lower in both SW480 sublines and/or SW620 cells, as compared with SW480 cells. Only 124 genes (5.8%) were differentially expressed among the two SW480 sublines and SW620 cells. Therefore, these data strongly demonstrated that the two SW480 sublines and SW620 cells shared almost identical gene expression patterns (94.2%). Finally, we plotted all 2151 genes differentially expressed by the two SW480 sublines and SW620 cells and compared the average expression intensities of these genes (Fig. 4D). The average intensity ratios were highly similar among both SW480 sublines and SW620 cells. Collectively, our data demonstrated that SW480.sel1 and SW480.spl sublines were more dissimilar to SW480 cells and more ‘similar’ to SW620 cells at both qualitative and quantitative levels of genome-scale gene expression. Thus, these gene expression profiles strongly supported the contention that the SW480.sel1 and SW480.spl sublines were much more SW620-like than SW480-like, further suggesting that such Fas-resistant metastatic subpopulations pre-existed within the primary tumor cell population.

Spontaneous metastatic ability of the various colon carcinoma populations

These experiments explored potential correlations between functional Fas status and metastatic potential of the various colon carcinoma populations in a xenograft nude mouse model. Specifically, we examined whether these in vitro- or in vivo-selected SW480 sublines, when compared with unselected SW480 and SW620 cells, exhibited enhanced metastatic behavior in vivo (Fig. 5). To that end, tumor cells were injected s.c. in the right flank of nude mice; s.c. tumor growth was measured, as well as the potential for spontaneous splenic metastasis. Previous studies comparing SW480 and SW620 cells for s.c. tumor growth in nude mice revealed that both tumor cell lines proliferated similarly and progressively over the course of the experiments (14). Therefore, mice were sacrificed when s.c. tumor growth of each group exceeded 800 mm3, which typically occurred within 4–5 wk of inoculation. Individual spleens were then collected for subsequent molecular and biologic analyses.



View larger version (48K):
[in this window]
[in a new window]
  		FIGURE 5. Differences in metastatic ability of SW480 and SW620 cells and two independent SW480 sublines. A, Agarose gel image showing the specificity of detection of human keratin-18 message by RT-PCR from SW480 and SW620 cells, as well as cells from non-tumor-bearing mouse spleen (shown in the right-end lane). B, Agarose gel image showing the sensitivity of detection of the human keratin-18 transcript by RT-PCR in mixtures of SW480 cells with total splenocytes (from non-tumor-bearing mice). The numbers shown at the top of the gel panel reflect molecular detection of tumor cells expressing keratin-18 message determined on a per spleen basis. C–E, Agarose gel images showing human keratin-18 transcript in total splenocytes of tumor-bearing mice. Each mouse was injected with a given tumor cell line or subline, and spleens were harvested. SW620 (C), SW480 (D), SW480.sel1 (Ea), or SW480.spl (Eb) are presented. Tumor volume ± SEM in C, D, Ea, and Eb were 981 ± 203.1 mm3, 1609 ± 437 mm3, 1155 ± 248 mm3, and 812 ± 164 mm3, respectively. In all experiments, mouse {beta}-actin was used as normalization standard.

 
An RT-PCR assay for detection of human tumor cells of epithelial origin in mouse lymphoid tissue was developed. The spleen was selected as an organ for analysis of spontaneous distal metastasis because it represented a lymphoid compartment consistent with the histologic tropism of the naturally occurring SW620 metastatic cell line (12). Furthermore, because of spleen size, significant amounts of RNA could be isolated for detailed molecular analysis of individual mice. Keratin-18 expression was chosen as a selective tumor marker because of its unique and abundant expression by human tumor cells of epithelial origin against a murine lymphoid background (see Fig. 5A). Initial control experiments examined both the specificity and sensitivity of the RT-PCR assay for detection of SW480 and SW620 cells. We found that both SW480 and SW620 cell lines expressed comparably high levels of message for keratin-18, based on RT-PCR analysis of that transcript at multiple cDNA dilutions (Fig. 5A). In contrast, no keratin-18 message was detectable from non-tumor-bearing splenocytes, revealing specificity of this assay for detection of these human tumor cell lines against this mouse background (Fig. 5A, last lane). To better define the sensitivity of the assay, mouse splenocytes were spiked with various concentrations of SW480 cells, and RT-PCR analysis was performed on total RNA for detection of the keratin-18 transcript (Fig. 5B). Because both SW480 and SW620 cells expressed comparable amounts of keratin-18 message (Fig. 5A), we selected one of the two lines (SW480) for this analysis. Based on the expression of keratin-18 message, a dose-dependent response was observed for detection of 102–107 tumor cells/spleen (~108 splenocytes/spleen, based on two independent experiments). Thus, the limit of detection occurred at ~102–103 tumor cells/spleen (Fig. 5B). Mouse {beta}-actin message was included as a normalization control for the integrity of the RNA samples.

Next, we examined total splenocytes from the various groups of tumor-bearing mice for detection and expression of human keratin-18 message. In mice receiving a s.c. transplant of SW620 cells, splenocytes of seven of eight independent mice were found to be positive for the expression of the keratin-18 transcript, which was detectable in a cDNA titratable fashion (Fig. 5C; no message detectable in mouse 1). Molecular detection of tumor cells by this RT-PCR assay was confirmed at the cellular level by culturing aliquots of these splenic preparations for the outgrowth of SW620 cells. In contrast, in mice receiving a s.c. transplant of SW480 cells, no detectable keratin-18 signal was observed (Fig. 5D, cDNA tested neat), despite the fact that spleens were harvested at a later time point when s.c. tumor growth was larger as compared with SW620 tumor-bearing mice (see Fig. 5). These data thus confirmed disparities in the splenic metastatic behavior of the naturally occurring SW480 and SW620 cell lines. Interestingly, despite the inability to detect tumor cells in the spleens of SW480-bearing mice by this RT-PCR assay, SW480 sublines could be established in cell culture from three independent mice, suggesting that the absolute numbers of tumor cells in the spleen fell below the limit of molecular detection (Fig. 5B). Lastly, in mice receiving a s.c. transplant of SW480.sel1 or SW480.spl sublines, splenocytes of five of five mice from each tumor-bearing group were shown to express keratin-18 message in a cDNA titratable manner (Fig. 5E, a and b). Thus, as with the earlier experiments (Figs. 1–4), the in vitro- and in vivo-selected SW480 sublines also biologically resembled SW620 cells more so than SW480 cells, at least in terms of spontaneous splenic metastatic activity from a s.c. tumor transplant.

Effect of loss of Fas function on metastatic behavior of vFLIP-expressing SW480 cells

To further examine the relationship or link between loss of Fas function and metastatic potential, we undertook another approach, one based on direct molecular disruption of the pathway in SW480 cells as opposed to one based on biologic selection of Fas-resistant subpopulations. To that end, the parental population of SW480 cells was first cloned by limiting dilution; individual clones were selected for the highest level of sensitivity to Fas-mediated apoptosis and then stably transfected with cDNA encoding the vFLIP gene known to block Fas-mediated signaling, and consequently, Fas-mediated apoptosis (19 ,22). Cloning allowed us to start with a more homogeneous population of Fas-responsive SW480 cells for vFLIP transfection, thus avoiding potential contamination of coexisting Fas-resistant SW480 cells within the heterogeneous parental population.

First, we demonstrated that vFLIP-expressing SW480.C1 cells (termed pEGFPN1-vFLIP) were, in fact, resistant to Fas-mediated apoptosis, as determined by the lack of induced DNA fragmentation following IFN-{gamma}/CH-11 treatment (Fig. 6A). In contrast, nontransfected SW480.C1 cells or SW480.C1 cells, transfected with the vector as negative control (termed pEGFPN1), expressed or retained sensitivity to Fas-mediated apoptosis (Fig. 6A). Next, we examined vFLIP-expressing SW480.C1 cells in vivo in nude mice for spontaneous splenic metastatic potential, as compared with SW620 cells (Fig. 6B). As in Fig. 5, splenocytes from appropriate groups of tumor-bearing mice were examined by RT-PCR for expression of the human keratin-18 transcript. In mice receiving a s.c. transplant of vFLIP-expressing SW480.C1 cells, no detectable keratin-18 message was observed in splenocyte preparations of all four mice tested. Similar results were observed in mice receiving vector-transfected SW480.C1 cells (data not shown). In contrast, in mice receiving SW620 cells, keratin-18 message was observed in splenocytes of four of five mice examined (Fig. 6B, cDNA tested neat). These findings were verified by cell culture in which tumor cell colonies emerged from splenocytes of the same four of five SW620-bearing mice, whereas no colonies were detectable from the vFLIP-transfected SW480.C1 cells. Lastly, single cell suspensions were also prepared from fresh s.c. tumor explants of all four vFLIP-SW480.C1 tumor-bearing mice. These cells were then examined immediately for GFP intensity by flow cytometry as a surrogate marker for in vivo transgene expression (Fig. 6C). An aliquot was also placed in culture for two passages and reanalyzed for sensitivity to Fas-mediated apoptosis in vitro. The majority of cells recovered from primary explants of vFLIP-SW480.C1-bearing mice (although likely comprising a mixture of both human and mouse cells at this point) appeared to be GFP-positive based on the shift in fluorescence intensity relative to nontransfected SW480 cells, suggesting that the transgene remained stably expressed. Moreover, cultures re-established from these fresh s.c. explants retained GFP expression and resistance to Fas-mediated apoptosis induced by IFN-{gamma}/CH-11 treatment, as observed morphologically under the microscope. Lastly, despite the fact that vFLIP-expressing SW480.C1 cells were Fas-resistant, these cells were morphologically similar to those of the parental SW480 population (data not shown).



View larger version (28K):
[in this window]
[in a new window]
  		FIGURE 6. Effect of vFLIP transfection of SW480 cells on their sensitivity to Fas-mediated apoptosis and metastatic behavior. A, DNA agarose gel electrophoresis revealing the sensitivity of untreated (-) or IFN-{gamma}-pretreated (+), untransfected SW480.C1 cells or SW480.C1 cells transfected with the vector control (SW480-pEGFPN1) or the vFLIP gene (SW480-pEGFPN1-vFLIP) to Fas-mediated apoptosis induced by CH-11 treatment, as in Fig. 1. B, Agarose gel images show the detection of the human keratin-18 transcript in total splenocytes of the indicated tumor-bearing mice, as in Fig. 5, C–E. Mouse {beta}-actin was used as a normalization standard. C, Single cell suspensions were prepared from fresh s.c. explants of the four vFLIP-SW480.C1 tumor-bearing mice. Cells from each mouse (b–e) were then examined immediately for GFP expression by flow cytometry relative to untransfected SW480 cells (a).

 
Identification of genes differentially expressed in the various colon carcinoma populations

In this study using a human colon carcinoma cell model, we demonstrated that loss of Fas function was associated with the metastatic phenotype, but alone was insufficient for metastasis under these experimental conditions (Fig. 6). Therefore, it was reasonable to postulate that in addition to alterations in Fas function, alterations in the expression of other genes might be required for a more favorable or productive malignant phenotype. To further identify genes that were differentially expressed between the primary and metastatic colon carcinoma cell lines, we selected genes whose expression levels were >=2-fold or <=2-fold in SW620 cells and/or the two SW480 sublines, as compared with SW480 cells. This cutoff was more stringent than the cutoff previously described (Fig. 4), because the intention was to identify a pool or cluster of specific genes potentially linked to the metastatic phenotype, rather than to generate broader molecular profiles or signatures for comparative analyses. Any gene that had a log2 ratio >=0.7 (2-fold) in all three replicate experiments was considered overexpressed in SW620 cells and/or the two SW480 sublines, and conversely, any gene that had a log2 ratio <=-0.7 (0.5-fold) in all three replicate experiments was considered underexpressed in the SW620 and/or the two SW480 sublines. By using this selection criterion, we identified 1112 overexpressed genes and 594 underexpressed genes. We presented the genes with known functions in Fig. 7, and classified them according to their ontology in the databases and known activities in the literature. The five functional categories of genes included tumorigenesis and metastasis-related, signal transduction, apoptosis, cell cycle control, and immune response-related.



View larger version (75K):
[in this window]
[in a new window]
  		FIGURE 7. Identity of overexpressed and underexpressed genes in SW620 cells and the SW480-derived sublines. Genes that were overexpressed by at least 2-fold (indicated in red) or underexpressed by at least 2-fold (indicated in green) in the SW620 metastatic cell line and the two SW480-derived sublines as compared with the SW480 primary tumor cell line are shown. Each colored square represents the ratio of SW620 cells or the two sublines over SW480 cells from one measurement of one gene. Three independent hybridization experiments (top) were conducted. The scale of the intensity ratio is in log2 value, with a red color indicating overexpression and a green color indicating underexpression in SW620 cells and the sublines vs SW480 cells. The Unigene number and gene names are indicated (right side) beside the microarray results.

 
Ras genes encode an extensive family of small GTPase molecules, mediating integral roles in cellular signaling, proliferation, and function. Oncogenic ras activation or altered regulation of components of the ras pathway is one of the earliest genetic events involved in malignant transformation (23, 24). Although both SW480 and SW620 cell lines are known to harbor a ras codon 12 mutation (23), we identified in this study several Ras GTPase family members that were either overexpressed or underexpressed in the metastatic cells or sublines (Fig. 7). Some of these small GTPase molecules belong to the Rho/Rac GTPase family and consist of, for example, Rho guanine nucleotide exchange factor, Rho GDI dissociation inhibitor, Rho GTPase activating protein, Rac GTPase activating protein, and Rho/Rac GEF (Fig. 7). Rho/Rac GTPase family genes regulate cell morphology and actin cytoskeleton structure (25). It has been reported that transformation of mammary epithelial cells into invasive carcinoma is associated with a switch from a normal cell morphology into a more motile invasive one, indicating that the alterations in the expression of these GTPase molecules may play a potentially important role in metastatic development (26). Other GTPase family members, including G protein-coupled receptor kinase and regulator of G protein signaling molecules, control the activity of several crucial signaling pathways, and aberrant expression of these genes has also been suggested to contribute to malignant potential (24). Furthermore, we observed overexpression of type 1 collagen, {alpha}1, and small nuclear ribonucleoprotein F, two genes recently reported to be up-regulated in diverse metastatic tumor types (27).

Another significant feature was the differential expression of a large set of kinase and phosphatase genes (Fig. 7). Some of these identified kinases and phosphatases play pivotal roles in signal transduction, whereas some of them have been shown to be key molecules regulating events of tumorigenesis and metastasis (28). Spleen tyrosine kinase is a Syk kinase whose expression is correlated with a poorer prognosis in patients with breast cancer (29). Nonreceptor protein tyrosine phosphatases are known to be overexpressed in tumor tissues (30). It is also worth noting that two cellular cytoskeleton genes, actin {gamma}2 and vimentin, were overexpressed >=10-fold in the metastatic cells and the sublines as compared with the primary tumor (Fig. 7). Actin {gamma}2 has been proposed to be a signature gene of metastasis (27), whereas the precise role of vimentin in tumor metastasis has not yet been fully elucidated; albeit, studies in prostate carcinoma support elevated levels of vimentin expression with enhanced metastatic potential (31). As previously outlined, several Rho/Rac small GTPase family genes, as well as actin {gamma}2 and vimentin, were found to be overexpressed in the metastatic cells and sublines. Therefore, it is possible that members of the Rho/Rac GTPase family can modulate motility and invasion of metastatic cells through regulating the expression of actin {gamma}2- and vimentin-mediated cellular cytoskeleton reorganization.

In regard to genes associated with apoptosis, we identified 11 such genes that were overexpressed and five that were underexpressed in the metastatic cells and sublines (Fig. 7). Caspases-3 and -8 are key effectors or initiators of the Fas signaling pathway, respectively (32). Interestingly, although caspase-3 was found to be overexpressed, caspase-8 was shown to be underexpressed, which similarly has been observed in neuroectodermal tumors (33, 34). Bcl-2 family members, which exhibit anti-apoptotic activities (35), were also found to be overexpressed in the metastatic cell line and the sublines. Programmed cell death 6 is also involved in the regulation of cell death mechanisms and has been reported to be highly up-regulated in cancerous tissues of different origins (36). Dysfunction of cell cycle control genes often results in uncontrolled cell proliferation, and thus, influences tumor formation. In this study, we identified 17 genes that were overexpressed, and two genes that were underexpressed in the metastatic cell line and sublines (Fig. 7). These genes included the cyclins, cell cycle-related kinases, modulators, and DNA repair molecules. Finally, in regard to immune response-related genes, we identified 12 genes overexpressed and 13 genes underexpressed (Fig. 7). These genes largely reflected those involved in the regulation of cytokine, chemokine, and MHC functions; consequently, altered regulation of any of these genes may impact diverse aspects of the host/tumor interaction and the potential efficacy of cell-mediated antitumor responses.


   Discussion
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
References
 
Neoplastic cells may evade cell-mediated immunity at multiple levels of the effector/target interaction (37, 38, 39, 40, 41, 42), which consequently may impact the malignant process. Tumor escape mechanisms may affect Ag recognition, conjugate formation, or T cell activation or proliferation. What remains to be further understood, however, is whether tumor cells may evade immune attack by acquiring resistance to cytotoxic effector mechanisms. This is important because as the neoplastic process becomes more aggressive, resulting subpopulations of tumor cells are likely to exhibit a more apoptotic-resistant phenotype. Because Fas is an important receptor-mediated signaling pathway for triggering apoptotic death by both innate and adaptive elements of the immune system (4, 5, 9), disruption of such a cell death mechanism in neoplastic cells might confer a selective survival advantage for tumor escape. Several studies in mice support the notion that the Fas/FasL system may be important for the regulation of murine tumor growth (43, 44, 45, 46). Additionally, in a mouse model of melanoma metastasis, it was reported that Fas loss-of-function was both necessary and sufficient for tumor progression (43). Furthermore, appreciable interest has now surrounded the soundness of the "tumor counterattack model" and whether FasL expression by malignant cells contributes to tumor escape as a consequence of apoptosis of infiltrating Fas-bearing effector T cells (41, 42). This model has been recently challenged given the findings that FasL-expressing tumor cells are more easily rejected as compared with Fas-negative parental cells via a proinflammatory-based, neutrophil-mediated antitumor mechanism (41, 42).

In human colon tumorigenesis, immunohistochemical evidence suggests that loss of Fas expression is a common occurrence of an advancing neoplastic phenotype (10, 47, 48, 49); however, the biologic significance and mechanisms regulating these outcomes are still unclear. To begin to analyze the relationship between Fas sensitivity and malignant potential in the area of human colon carcinoma, we made use of a pair of naturally occurring primary and metastatic cell lines. Although functional Fas status appeared to inversely correlate with metastatic phenotype (Figs. 1 and 5), proof-of-principle required experimental disruption of the Fas pathway in SW480 cells. To that end, we undertook an approach based on direct disruption of the Fas pathway in SW480 cells. Interestingly, the use of Fas-resistant SW480 cells generated by vFLIP transfection, however, failed to elicit detectable splenic metastases, suggesting that tumor cells that ultimately became metastatic under these conditions likely harbored additional genetic alterations. Therefore, rendering SW480 cells resistant to Fas-mediated death, at least as accomplished or defined by FLIP overexpression, was not sufficient for acquisition of the metastatic phenotype. Although this study focused on the spleen as a target organ for metastasis, it does not preclude the possibility for tumor spread to other tissues, which warrants further study. In preliminary experiments, however, we found a similar pattern of dissemination in the draining lymph nodes, whereas no evidence was found for tumor infiltration in the liver, based on the RT-PCR assays.

The inability of IFN-{gamma} to endow a Fas-responsive phenotype in SW620 cells or the different SW480 sublines was not likely due to a complete defect in their response to IFN-{gamma}, because these cells efficiently up-regulated cell surface expression of Fas (Figs. 1 and 3), as well as MHC class I (HLA-A2) and ICAM-1 molecules (data not shown). Resistance to Fas-mediated lysis was demonstrated in both short-term and longer-term assay formats (data not shown), and furthermore, substantiated by the observation that the only tumor cell colonies that actually successfully survived and proliferated in response to anti-Fas selection were SW620-like, based on both morphologic assessment and cDNA microarray studies (Figs. 2 and 4). Future studies are warranted to explore in further detail the molecular basis for Fas resistance in SW620 cells, which may occur at multiple levels in the apoptotic signaling pathway (see Fig. 7, for example). DNA sequence analysis of the coding region of the Fas gene in SW620 cells, however, revealed it to be wild type (data not shown). Furthermore, although caspase-8 was underexpressed in SW620 cells as compared with SW480 cells as determined by cDNA microarray analysis (Fig. 7), it is likely that the inability of IFN-{gamma} to sensitize SW620 cells to Fas-mediated death occurred at multiple levels. This is supported by our recent identification of two additional signaling defects in SW620 cells that accounted for, at least in part, resistance to Fas-mediated death following IFN-{gamma} sensitization; one was involved in IFN-{gamma}-mediated signaling, IFN consensus sequence binding protein, and the other involved in Fas-mediated signaling, caspase-1 (50). IFN consensus sequence binding protein and caspase-1 were not identified in this present study as differentially induced genes (Fig. 7) because their expression was IFN-{gamma}-dependent (50). In this study, both cell lines were compared with each other without IFN-{gamma} treatment to identify differentially expressed genes potentially linked to the metastatic phenotype. Therefore, given these additional deficiencies, it is unlikely that caspase-8 underexpression, albeit a potentially important player, would be solely responsible for the failure of SW620 cells to undergo any detectable levels of Fas-mediated apoptosis ensuing IFN-{gamma} sensitization.

Although it is generally well accepted that metastatic cells originate from the primary neoplasm (1, 8), it remains to be fully understood whether: 1) metastatic cells arise as progenitors and must still face additional genetic modifications and environmental influences at varying stages of the metastatic cascade before becoming a terminally committed metastatic cell; 2) terminally committed metastatic cells have already undergone relevant molecular and biologic alterations within the primary mass, and therefore, already pre-exist, albeit at extremely low frequencies; and 3) both possibilities together, as well as others contribute to shaping the metastatic phenotype. In this study, SW480 cells still exhibited a detectable level of metastatic activity in the spleen, based on the rare recovery of SW480 sublines established from the spleen. This observation is thus consistent with the broad notion that cells comprising the primary tumor are heterogeneous in metastatic phenotype. Interestingly, however, such in vivo-derived sublines morphologically, functionally, and molecularly resembled SW620 cells more so than did SW480 cells (Figs. 2–5), which also lends support to the second notion (specified in 2). Furthermore, the observation that SW620-like cells emerged and outgrew from the primary tumor population by functional elimination of Fas-sensitive cells (at least in vitro) implied an important and dynamic role for Fas-based interactions in the generation of subpopulations exhibiting a more malignant phenotype, which thus may account for a novel mechanism of tumor escape and neoplastic progression. In this study, we used a surrogate stimulus of the Fas pathway to specifically focus on our hypothesis on Fas-based interactions, although similar results were observed using human CD4+ T cell-derived soluble FasL (data not shown).

It is important to point out, however, that our findings do not preclude the possibility for the existence of metastatic precursors, subpopulations, or clones within the SW480 primary tumor population expressing phenotypes distinct from those of SW620 cells, which did not emerge or were not detectable under these experimental conditions. We also recognize that future investigations are warranted to extend observations made with the SW480/SW620 paradigm to other naturally occurring pairs of primary and metastatic tumor isolates. To our knowledge, however, no other matched pair(s) of primary and metastatic colon carcinoma cell lines have been reported. Despite that, we have data in a syngeneic mouse model of experimental lung metastasis that conceptually supports the "Fas selection hypothesis" in the generation of a more malignant phenotype from a parental primary tumor population (51).

In summary, we have identified and characterized a previously unrecognized and unique influence of anti-Fas interactions in a human colon carcinoma model of primary and metastatic disease, which is distinct from other studies in murine tumor models (43, 44, 45, 46). Although it was reported in a mouse model of melanoma that loss of Fas function was both necessary and sufficient for metastatic development (43), in our human colon carcinoma model, we found that altered Fas function was characteristic of, but alone was insufficient for, acquisition of a detectable metastatic phenotype. In this study, our data support the idea that such Fas-resistant subpopulations also co-possess additional genetic characteristics requisite for metastatic capability, which yet remain to be fully elucidated (Figs. 7 and 8). It is important to note, however, that differences in invasion and adhesion interactions, as well as other host factors between xenogeneic and syngeneic mouse models may influence metastatic potential or outcome. Yet, if loss of Fas function is a pivotal or hallmark characteristic of more aggressive neoplastic clones, then selection against Fas-sensitive clones by anti-Fas interactions may represent, in part, a mechanistic basis of tumor progression toward a more metastatic phenotype. The observations that SW480 sublines, derived not only in vitro but also in vivo from metastatic sites, resemble the SW620 cell line further strengthen the notion for the pre-existence of such metastatic subpopulations within the heterogeneous primary tumor, and perhaps, other still undefined metastatic clonal populations. Given the capacity and repertoire of genes detectable by our cDNA microarray chips, a large number of differentially expressed genes have been identified (Fig. 7). Detailed molecular and functional analyses of such differentially expressed candidate genes may reveal important insights into the nature and identity of additional genetic elements of tumor progression.



View larger version (11K):
[in this window]
[in a new window]
  		FIGURE 8. Model of Fas-based interactions in tumor progression in a human colon carcinoma system. The SW480 primary tumor population appeared to comprise multiple subpopulations, embracing the gamut of both Fas-sensitive (FasS) and Fas-resistant (FasR) cells, as well as metastatic-incompetent (MetI) and metastatic-competent (MetC) ones, illustrated by the dashed line. In contrast, the SW620 metastatic tumor population appeared to be more homogeneous in terms of expressing a FasRMetC phenotype. FasSMetI cells, although largely characteristic of the parental SW480 population, were also identified based on the isolation of Fas-sensitive, single cell clones. FasRMetC cells were identified within the parental SW480 population, based on the isolation of both in vitro- and in vivo-derived SW480 sublines. A FasRMetI phenotype was produced in the primary tumor, based on vFLIP transfection. The observation that such SW480 sublines harboring a FasRMetC phenotype morphologically and molecularly resembled the naturally occurring SW620 population suggested that such metastatic subsets already pre-existed within the primary tumor and that anti-Fas interactions served as a biologic selective pressure for their outgrowth. However, the idea that Fas status alone was sufficient for this biologic outcome was unlikely, because SW480 cells generated to express a FasRMetI phenotype did not display detectable splenic metastatic behavior under these experimental conditions. Thus, these data identified a FasRMetC phenotype, likely reflecting neoplastic subpopulations that also "co-possessed" additional malignant/metastatic-associated genes and properties.

 


   Acknowledgments
 
We thank R. Siegel for the vFLIP construct, L. Granger of the National Cancer Institute Core Flow Cytometry Facility for assistance with cell sorting, L. Young of the Center for Information Technology/National Institutes of Health for instructional use of the computer programs for cDNA microarray analysis, and D. Weingarten for editorial assistance.


   Footnotes
 
1 Address correspondence and reprint requests to Dr. Scott I. Abrams, Laboratory of Tumor Immunology and Biology, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Building 10, Room 5B46, 10 Center Drive, Bethesda, MD 20892-1402. E-mail address: sa47z{at}nih.gov Back

2 Abbreviations used in this paper: FasL, Fas ligand; vFLIP, viral-encoded FLICE-inhibitory protein; GFP, green fluorescent protein. Back

Received for publication April 11, 2003. Accepted for publication August 14, 2003.


   References
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 References
 

  1. Hanahan, D., R. A. Weinberg. 2000. The hallmarks of cancer. Cell 100:57.[Medline]
  2. Kaufmann, S. H., W. C. Earnshaw. 2000. Induction of apoptosis by cancer chemotherapy. Exp. Cell Res. 256:42.[Medline]
  3. Henkart, P. A.. 1996. ICE family proteases: mediators of all apoptotic cell death?. Immunity 4:195.[Medline]
  4. Nagata, S.. 1997. Apoptosis by death factor. Cell 88:355.[Medline]
  5. Russell, J. H., T. J. Ley. 2002. Lymphocyte-mediated cytotoxicity. Annu. Rev. Immunol. 20:323.[Medline]
  6. Saha, S., A. Bardelli, P. Buckhaults, V. E. Velculescu, C. Rago, B. St. Croix, K. E. Romans, M. A. Choti, C. Lengauer, K. W. Kinzler, B. Vogelstein. 2001. A phosphatase associated with metastasis of colorectal cancer. Science 294:1343.[Abstract/Free Full Text]
  7. Wang, T. L., C. Rago, N. Silliman, J. Ptak, S. Markowitz, J. K. Willson, G. Parmigiani, K. W. Kinzler, B. Vogelstein, V. E. Velculescu. 2002. Prevalence of somatic alterations in the colorectal cancer cell genome. Proc. Natl. Acad. Sci. USA 99:3076.[Abstract/Free Full Text]
  8. Fidler, I. J.. 2002. Critical determinants of metastasis. Semin. Cancer Biol. 12:89.[Medline]
  9. Van Parijs, L., A. K. Abbas. 1998. Homeostasis and self-tolerance in the immune system: turning lymphocytes off. Science 280:243.[Abstract/Free Full Text]
  10. Moller, P., K. Koretz, F. Leithauser, S. Bruderlein, C. Henne, A. Quentmeier, P. H. Krammer. 1994. Expression of APO-1 (CD95), a member of the NGF/TNF receptor superfamily, in normal and neoplastic colon epithelium. Int. J. Cancer 57:371.[Medline]
  11. Krammer, P. H.. 1999. CD95 (APO-1/Fas)-mediated apoptosis: live and let die. Adv. Immunol. 71:163.[Medline]
  12. Leibovitz, A., J. C. Stinson, W. B. McCombs, III, C. E. McCoy, K. C. Mazur, N. D. Mabry. 1976. Classification of human colorectal adenocarcinoma cell lines. Cancer Res. 36:4562.[Abstract/Free Full Text]
  13. Hewitt, R. E., A. McMarlin, D. Kleiner, R. Wersto, P. Martin, M. Tsokos, G. W. Stamp, W. G. Stetler-Stevenson, M. Tsoskas. 2000. Validation of a model of colon cancer progression. J. Pathol. 192:446.[Medline]
  14. Bergmann-Leitner, E. S., S. I. Abrams. 2000. Differential role of Fas/Fas ligand interactions in cytolysis of primary and metastatic colon carcinoma cell lines by human antigen-specific CD8+ CTL. J. Immunol. 164:4941.[Abstract/Free Full Text]
  15. Itoh, N., S. Yonehara, A. Ishii, M. Yonehara, S. Mizushima, M. Sameshima, A. Hase, Y. Seto, S. Nagata. 1991. The polypeptide encoded by the cDNA for human cell surface antigen Fas can mediate apoptosis. Cell 66:233.[Medline]
  16. Gorczyca, W., S. Bruno, R. J. Darzynkiewicz, J. Gong, Z. Darzynkiewicz. 1992. DNA strand breaks occurring in apoptosis: their early in situ detection by the terminal deoxynucleotidyl transferase and nick translation assays and prevention by serine protease inhibitors. Int. J. Oncol. 1:639.
  17. Gorczyca, W., K. Bigman, A. Mittelman, T. Ahmed, J. Gong, M. R. Melamed, Z. Darzynkiewicz. 1993. Induction of DNA strand breaks associated with apoptosis during treatment of leukemias. Leukemia 7:659.[Medline]
  18. Hara, I., H. Miyake, S. Hara, S. Arakawa, S. Kamidono. 2000. Differential involvement of the Fas receptor/ligand system in p53-dependent apoptosis in human prostate cancer cells. Prostate 45:341.[Medline]
  19. Thome, M., P. Schneider, K. Hofmann, H. Fickenscher, E. Meinl, F. Neipel, C. Mattmann, K. Burns, J. L. Bodmer, M. Schroter, et al 1997. Viral FLICE-inhibitory proteins (FLIPs) prevent apoptosis induced by death receptors. Nature 386:517.[Medline]
  20. Eisen, M. B., P. T. Spellman, P. O. Brown, D. Botstein. 1998. Cluster analysis and display of genome-wide expression patterns. Proc. Natl. Acad. Sci. USA 95:14863.[Abstract/Free Full Text]
  21. Bergmann-Leitner, E. S., S. I. Abrams. 2000. Influence of interferon-{gamma} on modulation of Fas expression by human colon carcinoma cells and their subsequent sensitivity to antigen-specific CD8+ cytotoxic T lymphocyte attack. Cancer Immunol. Immunother. 49:193.[Medline]
  22. Siegel, R. M., F. K. Chan, H. J. Chun, M. J. Lenardo. 2000. The multifaceted role of Fas signaling in immune cell homeostasis and autoimmunity. Nat. Immun. 1:469.[Medline]
  23. Bos, J. L.. 1988. The ras gene family and human carcinogenesis. Mutat. Res. 195:255.[Medline]
  24. Downward, J.. 2003. Targeting RAS signalling pathways in cancer therapy. Nat. Rev. Cancer 3:11.[Medline]
  25. Hall, A.. 1998. Rho GTPases and the actin cytoskeleton. Science 279:509.[Abstract/Free Full Text]
  26. Clark, E. A., T. R. Golub, E. S. Lander, R. O. Hynes. 2000. Genomic analysis of metastasis reveals an essential role for RhoC. Nature 406:532.[Medline]
  27. Ramaswamy, S., K. N. Ross, E. S. Lander, T. R. Golub. 2003. A molecular signature of metastasis in primary solid tumors. Nat. Genet. 33:49.[Medline]
  28. Zhang, L., Y. K. Lau, L. Xi, R. L. Hong, D. S. Kim, C. F. Chen, G. N. Hortobagyi, C. Chang, M. C. Hung. 1998. Tyrosine kinase inhibitors, emodin and its derivative repress HER-2/neu-induced cellular transformation and metastasis-associated properties. Oncogene 16:2855.[Medline]
  29. Toyama, T., H. Iwase, H. Yamashita, Y. Hara, Y. Omoto, H. Sugiura, Z. Zhang, Y. Fujii. 2003. Reduced expression of the Syk gene is correlated with poor prognosis in human breast cancer. Cancer Lett. 189:97.[Medline]
  30. Kim, Y. W., H. Wang, I. Sures, R. Lammers, K. J. Martell, A. Ullrich. 1996. Characterization of the PEST family protein tyrosine phosphatase BDP1. Oncogene 13:2275.[Medline]
  31. Lang, S. H., C. Hyde, I. N. Reid, I. S. Hitchcock, C. A. Hart, A. A. Bryden, J. M. Villette, M. J. Stower, N. J. Maitland. 2002. Enhanced expression of vimentin in motile prostate cell lines and in poorly differentiated and metastatic prostate carcinoma. Prostate 52:253.[Medline]
  32. Cohen, G. M.. 1997. Caspases: the executioners of apoptosis. Biochem. J. 326:1.
  33. Zuzak, T. J., D. F. Steinhoff, L. N. Sutton, P. C. Phillips, A. Eggert, M. A. Grotzer. 2002. Loss of caspase-8 mRNA expression is common in childhood primitive neuroectodermal brain tumour/medulloblastoma. Eur. J. Cancer 38:83.
  34. Yang, X., M. S. Merchant, M. E. Romero, M. Tsokos, L. H. Wexler, U. Kontny, C. L. Mackall, C. J. Thiele. 2003. Induction of caspase 8 by interferon {gamma} renders some neuroblastoma (NB) cells sensitive to tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) but reveals that a lack of membrane TR1/TR2 also contributes to TRAIL resistance in NB. Cancer Res. 63:1122.[Abstract/Free Full Text]
  35. Hara, I., S. Hara, H. Miyake, S. Arakawa, S. Kamidono. 2001. Bcl-2 modulates Fas-mediated apoptosis in human renal cell carcinoma cell lines. Int. J. Oncol. 18:1181.[Medline]