Alterations in Fas Expression Are Characteristic of, But Not Solely Responsible for, Enhanced Metastatic Competence

Considerable interest is dedicated to understanding the complex sequence of events underlying the metastatic process of solid tumors. This aspect of tumor biology represents both a major limitation and challenge confronting the development of more effective treatments for patients with advanced neoplastic disease. From an immunotherapeutic perspective, neoplastic cells may elude cell-mediated immunity at multiple stages of the effector/target interaction (1, 2, 3), which, consequently, may impact the malignant process. For example, tumor escape mechanisms may affect Ag recognition, conjugate formation, or T cell activation or expansion. However, what remains to be further understood is whether tumor cells may evade immune attack by acquiring resistance to cytotoxic effector mechanisms. This is an important consideration 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 inducing apoptotic death by host immune and nonimmune defense mechanisms (2, 3, 4), disruption of such a cell death pathway in neoplastic cells might confer a selective survival advantage for tumor escape.

Ordinarily, regulation of the Fas pathway is crucial for homeostasis in a number of normal and pathologic conditions (2, 3, 4). However, dysregulation of the Fas pathway has been implicated in tumor progression (5). Although the exact reasons for this association remain to be fully elucidated, it has been reported in mouse models that loss of Fas function is both necessary and sufficient for tumor growth as a consequence of tumor escape from Fas/Fas ligand (FasL)²-dependent interactions (6, 7, 8, 9). Nonetheless, how alterations in Fas expression or function influence metastatic behavior remains to be fully understood. In this study, we examined the following: 1) whether alterations in Fas expression alone were sufficient for tumor progression in a mouse model of lung metastasis, and 2) the relationship between Fas expression and malignant phenotype in matched pairs of primary and metastatic cell lines established from transgenic mice with spontaneously arising mammary carcinoma, a model that better resembles the pathogenesis of tumor progression.

Female BALB/c (H-2^d) mice were obtained from the National Cancer Institute, Frederick Cancer Research Animal Facility (Frederick, MD). Female FasL-deficient CPt.C3-Tnfsf6gld mice on a BALB/c background were obtained from The Jackson Laboratory (Bar Harbor, ME). All mice were used at >6 wk of age. Transgenic mice with spontaneously arising primary and metastatic mammary carcinoma (10), now backcrossed in a C57BL/6 (H-2^b) background, were kindly provided by S. Gendler (Mayo Clinic, Scottsdale, AZ) via J. Schlom (National Institutes of Health, Bethesda, MD). This transgenic mouse model was originally produced by expression of the polyomavirus middle T Ag via germline introduction of the middle T oncogene under the transcriptional control of the mouse mammary tumor virus promoter/enhancer (10).

The CMS4 sarcoma (11) was kindly provided by A. Deleo (University of Pittsburgh, Pittsburgh, PA). The in vivo-derived metastatic CMS4 subline (CMS4-met) was produced from the parental population by one in vivo passage in the lungs of normal BALB/c mice, as described (12). Briefly, CMS4 cells (2.5 × 10⁵ cell/mouse) were injected i.v. into the lateral tail vein. Mice were sacrificed 14 days later, and lungs were removed and digested for 4–6 h at room temperature with an enzyme mixture containing hyaluronidase (0.1 mg/ml), collagenase (1 mg/ml), and DNase I (30 U/ml), all obtained from Sigma-Aldrich (St. Louis, MO). Tumor cells that grew out from these lung digests, termed CMS4-met, were then maintained in culture. The in vitro-derived Fas-resistant CMS4 subline (CMS4.sel) was selected from the parental line in vitro following six successive passages in the presence of anti-Fas stimuli. Briefly, CMS4 cells were first treated with recombinant mouse IFN-γ (100 U/ml; R&D Systems, Minneapolis, MN) and TNF-α (100 U/ml; R&D Systems) overnight, followed by culture with anti-mouse Fas mAb (10 μg/ml, clone Jo2; BD PharMingen, San Diego, CA) and protein G (10 μg/ml; Sigma-Aldrich) to maximize cross-linking of anti-Fas at approximately weekly intervals for a total of four cycles. These cells then underwent two additional cycles of IFN-γ/TNF-α exposure plus recombinant human sFasL (100 ng/ml; Alexis, San Diego, CA).

The primary and metastatic tumor cell lines were isolated from a transgenic mouse colony, as mentioned earlier. Three matched pairs of primary and metastatic (to the lung) cell lines were established from three independent mice >120 days of age. The primary tumor was resected from a progressively growing mammary lesion, whereas the metastatic tumor from the same mouse was established from lung digests in a similar manner to that described above for the isolation of CMS4-met.

Cells were immunostained with FITC-conjugated anti-Fas mAb (BD PharMingen) or an isotype-matched hamster IgG and analyzed by flow cytometry.

Cell death was measured by propidium iodide (PI) staining; albeit, similar results were observed by TUNEL assays. Briefly, untreated or cytokine-treated cells were incubated with anti-Fas stimuli (i.e., sFasL) for 20–24 h. Cells were then collected and stained with PI for 10 min at room temperature, according to the manufacturer’s instructions (R&D Systems). After staining, the cells were washed and immediately analyzed by flow cytometry.

For immunohistochemical detection of Fas, 5-μm sections were cut from formalin-fixed, paraffin-embedded tissues taken from both primary (mammary gland) and metastatic (lungs) sites of tumor growth of three separate transgenic mice and mounted onto glass slides. Mice were >170 days of age, at which time a progressively growing primary mammary tumor was resected along with the lungs. Before staining, the sections were deparaffinized and rehydrated with PBS. Specimens were trypsinized for 30 min at 37°C using a 0.02% trypsin solution and then rinsed in PBS. Endogenous peroxidase activity was blocked using a 2% hydrogen peroxide/methanol solution for 15 min at room temperature, followed by rinsing in PBS. The Vectastain Elite ABC and Rabbit IgG kits were used for the blocking, secondary Ab, and immunoperoxidase steps, as described by the manufacturer (Vector Laboratories, Burlingame, CA). Briefly, specimens were blocked with normal goat serum (1.5%), then incubated with an affinity-purified, rabbit anti-mouse Fas polyclonal Ab (M-20, 1/250 dilution; Santa Cruz Biotechnology, Santa Cruz, CA) for 30 min, followed by rinsing and staining with a goat anti-rabbit biotinylated Ab (1/2,000) for another 30 min. After incubation with the secondary Ab, the slides were rinsed and incubated with ABC reagent. Color was developed by incubation with 3′3-diaminobenzidine solution (Sigma-Aldrich), followed by rinsing and counterstaining with hematoxylin. In the negative control samples the primary Ab was omitted. Images were acquired and processed by a computer equipped with a microscope (Zeiss Axiophot; Zeiss, Oberkochen, Germany) and a digital camera (Photometrics Cool Snap FX; Roper Scientific, Trenton, NJ).

Total RNA was isolated from tumor cells as described previously (13, 14) and used for the first strand cDNA synthesis using the ThermoScript RT-PCR system (Invitrogen, San Diego, CA). The cDNA was then used as templates for PCR amplification of mouse Fas and mouse β-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 PCR primers for mouse Fas were as follows: forward primer: 5′-ATGCTGTGGATCTGGGCT-3′; reverse primer: 5′-TCACTCCAGACATTGTCC-3′. The PCR primers for mouse β-actin were as follows: forward primer: 5′-ATTGTTACCAACTGGGACGACATG-3′; reverse primer: 5′-CTTCATGAGGTAGTCTGTCAGGTC-3′.

CMS4 cells were transfected with the mammalian expression plasmid pEGFPN1 (Clontech Laboratories, Palo Alto, CA) or pEGFPN1 containing both the vFLIP gene (15) and green fluorescent protein (GFP), kindly provided by R. Siegel (National Institutes of Health). The expression plasmid and the plasmid containing the vFLIP coding sequence were then linearized with AflII restriction enzyme and used for transfection. Transfections were performed using LipofectAMINE 2000 reagent (Invitrogen), according to the manufacturer’s instructions. The transfected cells were propagated in culture medium containing Geniticin (Invitrogen) at a concentration of 0.75 mg/ml for 7 days, recovered, and recultured under the same conditions for two more passages before being sorted by a FACSVantage SE cell sorter (BD Biosciences, Mountain View, CA) based on GFP intensity. The sorted cells were cultured with Geniticin 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 Geniticin selection.

The various groups of CMS4 cells were resuspended in HBSS and injected i.v. into the lateral tail vein (2.5 × 10⁵ cells in 100 μl). Fourteen days later mice were sacrificed; lungs were inflated with a 15% solution of India ink, resected, and fixed in Fekete’s solution, as described (12). The number of pulmonary nodules were enumerated in a single-blinded fashion under a dissecting microscope. Values exceeding 250 lesions were considered too numerous to count accurately, and therefore were reported as >250.

Where indicated, data were reported as the mean ± SD. Statistical analysis was determined using an unpaired, two-sided t test, with p values <0.05 considered statistically significant.

The CMS4 sarcoma is a solid tumor of BALB/c (H-2^d) origin (11) that grows aggressively in naive, syngeneic hosts following transplant s.c. (12). Although the parental line is poorly metastatic for growth in the lungs following i.v. administration, we produced a highly metastatic subline from lung digests of those mice (12). We first examined Fas expression in the following three groups of CMS4 tumor cells (Fig. 1): the parental line (termed CMS4), the in vivo-selected metastatic subline (termed CMS4-met), and a CMS4 subline selected in vitro from the parental population by serial culture with agonistic anti-Fas mAb and sFasL (termed CMS4.sel).

RT-PCR analysis showed that CMS4 cells expressed the Fas transcript at a higher level, compared with CMS4-met and CMS4.sel (Fig. 1,A, lanes 1, 5, and 9). The proinflammatory cytokines IFN-γ and TNF-α have been reported to enhance Fas expression on both normal and malignant cells in other cell systems (16, 17, 18). CMS4 cells treated with either IFN-γ, TNF-α, or a combination of both cytokines showed an enhanced level of Fas transcription compared with untreated cells (Fig. 1,A, lanes 1–4). However, treatment of CMS4 cells with both cytokines resulted in the highest level of Fas mRNA expression compared with these same cells treated with either cytokine separately. Furthermore, under these same cytokine treatment conditions, similar hierarchal patterns of Fas mRNA expression were observed with both CMS4-met and CMS4.sel (Fig. 1,A, lanes 5–8 and 9–12, respectively). Comparison of both CMS4-met and CMS4.sel with CMS4 cells for Fas mRNA revealed that CMS4 cells expressed the highest level of the Fas transcript under these various cytokine treatments (Fig. 1 A).

Flow cytometric analysis revealed that although the three groups of CMS4 cells were Fas⁺ to varying degrees, the absolute levels of Fas were low, based on their mean fluorescence intensity (MFI) values (Fig. 1, Ba and C, untreated conditions). As with the RT-PCR assays for detection of the Fas transcript, treatment of CMS4 cells with either IFN-γ, TNF-α, or both cytokines increased the percentage of Fas⁺ cells compared with the untreated controls (Fig. 1,B, a–d, left column). However, treatment of CMS4 cells with both cytokines resulted in the highest level of Fas expression based on their MFI values compared with these same cells treated with either cytokine separately (Fig. 1, B, a–d, left column, and C). Again, under these same cytokine treatment conditions, similar hierarchal patterns of Fas expression were observed with both CMS4-met and CMS4.sel (Fig. 1,B, middle and right columns, and C). Although the MFI values were similar among the three CMS4 groups, whether they were untreated or treated with either cytokine separately, comparison of these populations after treatment with both cytokines revealed that CMS4 cells displayed the highest level of Fas expression, which was significantly different from both CMS4-met and CMS4.sel (Fig. 1 C).

Functionally, the three groups of CMS4 cells were relatively resistant to Fas-mediated death (Fig. 2,Aa) in response to recombinant sFasL. Although pretreatment with IFN-γ or TNF-α each independently up-regulated Fas (Fig. 1), this led to only a marginal increase in sensitivity to Fas-mediated death (Fig. 2, A and B). In contrast, the combination of both IFN-γ and TNF-α sensitized the three groups of CMS4 cells to Fas-mediated death; albeit, based on the percentages of cell death induced by sFasL, CMS4 cells displayed the highest level, followed by CMS4-met and CMS4.sel (Fig. 2, Ad and B). Thus, under these cytokine-inducible conditions, these results revealed a correlation between Fas expression levels and Fas-mediated death among the three CMS4 groups, although the correlation was more significant with CMS4.sel, which was produced under anti-Fas selection conditions.

To determine the consequences of altered Fas expression and function on metastatic behavior, we made use of an experimental lung metastasis model. As expected, in mice receiving CMS4 cells, only a few lesions were detectable, whereas in mice receiving CMS4-met, a large number of nodules were visible (Fig. 3, A and B). Interestingly, in mice receiving CMS4.sel, a large number of lesions were found; however, the size of the individual nodules was much smaller compared with CMS4-met nodules (Fig. 3, A and B). Nonetheless, in this mouse model these data showed that biologic selection against Fas-responsive cells in a heterogeneous tumor population (as shown in vitro) can favor the outgrowth of a more malignant phenotype in vivo.

If loss of, or alterations in, Fas function is an important characteristic of a progressive neoplastic phenotype, then spontaneously arising primary and metastatic tumors should express differential Fas expression patterns. Accordingly, we extended our study from an experimental metastasis setting to a transgenic model system, reflecting spontaneous mammary carcinoma development with accompanying metastases to the lung (10). First, we examined tissue sections from both primary (mammary) and metastatic (lungs) sites of tumor growth from three independent transgenic mice for Fas expression by immunohistochemistry. We found that Fas was highly expressed in all three mammary tumors (Fig. 4, 1a, 2a, and 3a). No qualitative or quantitative difference was noted among the three mice. In contrast, Fas expression was considerably less in metastatic foci in the lung when compared with primary mammary gland carcinoma (Fig. 4, 1b, 2b, and 3b). Again, no qualitative or quantitative difference was noted among the three mice. Fas expression was observed in bronchiolar and alveolar epithelium, as previously reported (19). These observations thus revealed an inverse correlation between Fas expression and metastatic phenotype in a mouse model of mammary carcinoma in vivo by immunohistochemistry.

Furthermore, three matched pairs of primary and metastatic cell lines were established from three independent transgenic mice and then analyzed for Fas expression at molecular, phenotypic, and functional levels. RT-PCR analysis showed that all three primary tumor cell lines expressed higher levels of Fas mRNA compared with their metastatic counterparts (Fig. 5,A). In the D4387 cell line pair, cell surface Fas expression was observed in the primary tumor cell line, but not in its matched metastatic counterpart (Fig. 5,B, a1 and b1). Similarly, after treatment with both IFN-γ and TNF-α, cell surface Fas increased on the D4387B primary tumor cell line, but not on its metastatic counterpart, D4387L (Fig. 5,B, a2 and b2). In regard to the K6625 cell line pair, cell surface Fas expression was weakly detectable on both primary and metastatic cell lines without any cytokine treatment (Fig. 5,B, c1 and d1). However, following treatment with both cytokines, the percentage of Fas⁺ cells increased more with the K6625B primary tumor cell line compared with its metastatic counterpart, K6625L (Fig. 5,B, c2 and d2). In regard to the J1717 cell line pair, the pattern of cell surface Fas expression was similar to that of D4387 in that a higher percentage of Fas⁺ cells was associated with the primary tumor cell line compared with its metastatic counterpart (Fig. 5,B, e1 and f1). Similarly, after treatment with both IFN-γ and TNF-α, cell surface Fas increased more so with the J1717B primary tumor cell line compared with its metastatic counterpart, J1717L (Fig. 5 B, e2 and f2).

Functionally, the primary and metastatic tumor cell lines of all pairs displayed little to no susceptibility to Fas-mediated death (Fig. 5, C, a1–f1, and D). However, pretreatment with both IFN-γ and TNF-α partially sensitized the primary tumor cell line D4387B, but not its metastatic counterpart, to Fas-mediated death (Fig. 5, C, a2 and b2, and D), similar to the pattern seen with the CMS4 system. In contrast, cytokine pretreatment of the second pair, K6625, led to a smaller increase in responsiveness to Fas-mediated death, with no significant difference in susceptibility observed between the primary and metastatic cell lines (Fig. 5, C, c2 and d2, and D). Lastly, cytokine pretreatment of the third pair, J1717, resembled the functional pattern seen with K6625 more so than with D4387, in that the J1717 primary tumor cell line remained weakly responsive to Fas-mediated death (Fig. 5, Ce2 and D). Although the percentages of Fas-mediated cell death were not extensive under these conditions, the extent of death observed with the primary tumor cell lines D4387B as well as J1717B was significantly different from that observed with their respective metastatic counterparts (Fig. 5 D).

Having shown that altered Fas expression and/or function was associated with metastatic phenotype, we next sought to determine whether Fas alone was sufficient for acquisition of metastatic activity. The approach taken was to uncouple or disrupt the Fas/FasL process in the CMS4 model. First, CMS4 cells were modified by ectopic expression of the vFLIP gene to render them Fas-resistant. Both cellular and virally encoded forms of FLIP have been shown to block Fas-mediated apoptosis by inhibition of the signaling pathway at the level of complex formation with pro-caspase-8 (15, 20). The prediction was that if loss of Fas function was sufficient to promote metastatic behavior, then rendering the parental line Fas-resistant should confer demonstrable metastatic ability. Another strategy involved the direct examination of metastatic capacity of CMS4 cells in syngeneic FasL (gld)-deficient mice. Similarly, if the Fas/FasL interaction was the sole factor that controlled metastatic potential, then the loss of host-derived FasL would enable even Fas-sensitive CMS4 cells to metastasize as efficiently as CMS4-met or CMS4.sel did in wild-type mice.

Indeed, the ectopic expression of vFLIP in CMS4 cells resulted in resistance to Fas-mediated death (Fig. 6,Ab), whereas CMS4 cells transfected with the expression vector without the vFLIP gene retained sensitivity to Fas-mediated death (Fig. 6,Aa). However, i.v. injection of vFLIP-expressing CMS4 cells to normal BALB/c mice did not result in an increase in detectable lesions compared with CMS4 cells (or the vector control, data not shown), which contrasted with what was observed with CMS4-met and CMS4.sel (Fig. 6, B and C vs Fig. 3). Interestingly, when CMS4 cells were injected into syngeneic gld mice, an increase in detectable lung nodules, albeit not significant, was noted in some mice compared with groups receiving the same tumor cells (p = 0.084) or the vFLIP-transfectants (p = 0.116) administered to wild-type mice (Fig. 6, B and C vs Fig. 3). Furthermore, the increase under those conditions in gld mice was still highly significantly lower than that observed with CMS4-met cells transplanted into either wild-type (p = 1.6 × 10⁻⁶) or gld mice (p = 1.0 × 10⁻⁷) (Fig. 6, B and C vs Fig. 3). Permissive tumor growth in gld mice also suggested that the microenvironment of the gld mouse did not negatively impact metastatic formation of these cells. Additional control experiments revealed that the administration of either CMS4.sel or vFLIP-expressing CMS4 cells to gld mice resulted in the same patterns of lung metastases as that seen in wild-type mice (Figs. 3 and 6). In fact, the number of detectable lung nodules formed by CMS4.sel (p = 0.45) or the CMS4-vFLIP-expressing cells (p = 0.98) in wild-type vs gld mice was not significantly different. Taken collectively, these data suggested that although Fas expression or status was an important determinant in tumor progression (Figs. 3 and 6), it was not the sole factor regulating the extent or severity of metastatic behavior.

In this study, we investigated the link between Fas expression and metastatic phenotype in two distinct syngeneic mouse tumor models. In a mouse model of experimental lung metastasis, we showed for the first time the following: 1) that loss of, or alterations in, Fas function was linked to, but alone was insufficient for, full acquisition of the metastatic phenotype; and 2) that anti-Fas interactions, however, did serve as a selective pressure for the outgrowth of low Fas-expressing or resistant neoplastic subpopulations from the parental line, which apparently copossessed metastatic capability. The notion that functional Fas status was not solely responsible for this biologic outcome was demonstrated by dissociating the Fas/FasL process, either biologically, molecularly, or genetically.

First, in mice receiving CMS4.sel, although a large number of lesions was found, the size of the individual nodules was much smaller compared with those seen in mice harboring CMS4-met (Fig. 3, A and B). Although the exact reasons for differences in lesion size remain unclear, this may reflect in part the notion that CMS4 subpopulations selected under anti-Fas conditions may consist of clones that harbor diverse or intermediate metastatic phenotypes. Consequently, this may have resulted in a quantitatively and/or qualitatively different metastatic outcome compared with that achieved by CMS4-met. Nonetheless, these data revealed for the first time that biologic selection against Fas-responsive cells in a heterogeneous tumor population (as shown in vitro so that the effects can be specifically targeted against Fas) can influence the outgrowth of a more malignant phenotype in vivo. Secondly, experimental or molecular modification of the poorly metastatic parental tumor line to express a Fas-resistant phenotype did not change their metastatic behavior (Fig. 6). Thirdly, in a converse fashion, administration of the parental tumor line to FasL-deficient mice failed to achieve the same degree or magnitude of metastatic activity as observed with CMS4-met, the in vivo-derived subline. Thus, a Fas-resistant/metastatic-competent phenotype likely consisted of neoplastic subpopulations that also coexpressed additional intrinsic malignant characteristics, which yet remain to be fully elucidated (Fig. 7). Given the outcome that CMS4.sel emerged from the parental tumor line via Fas-based selective pressures in vitro, it is likely that single-cell cloning studies of the parental tumor line would also reveal the existence of these subpopulations. Additionally, although our hypothesis addressed the relationship between Fas expression and metastatic phenotype, this does not preclude the possibility for the existence of metastatic precursors or clones within the parental CMS4 tumor line expressing phenotypes distinct from those of CMS4-met or CMS4.sel, which did not emerge or were not detectable under these experimental conditions. Another consideration of this paradigm is the potential role of proinflammatory cytokines in vivo to modulate tumor-cell sensitivity to Fas-mediated death. The process of tumor growth in vivo may initiate or provoke a localized inflammatory response, perhaps consisting of IFN-γ and TNF-α production. Consequently, the relative absence or presence of appropriate proinflammatory cytokines as well as survival or growth factors may influence tumor-cell responsiveness to diverse biologic processes, including death receptor signaling.

Additionally, we extended the inverse correlation between Fas expression and metastatic phenotype in transgenic mice with spontaneously arising primary and metastatic mammary carcinoma, a model system that better mirrored the natural process of tumor progression. In those studies we did so by immunohistochemistry of tissues isolated in vivo from sites of primary and metastatic tumor growth as well as with tumor cell line pairs established from those sites (Figs. 4 and 5). However, at the functional Fas level, cytokine pretreatment did not sensitize the primary tumor cell lines of the K6625 and J1717 pairs as efficiently as compared with the D4387 pair to Fas-mediated death, despite the observation that Fas expression increased on all three distinct primary tumor cell lines (Fig. 5, B–D). These results suggested that the primary tumor cell lines of these two pairs may harbor additional defects in the Fas signaling pathway, and thus, may reflect a more advanced stage of the neoplastic process that warrants further investigation. Future studies also are necessary to examine the role of anti-Fas interactions in tumor progression in this transgenic mouse mammary carcinoma model. Nonetheless, from a translational standpoint to a human system, we now have data in a primary and metastatic colon carcinoma model that conceptually support the “Fas selection hypothesis” in the generation of a Fas-resistant/metastatic-competent phenotype from the heterogeneous primary tumor population.

Previously, it was reported that Fas-sensitive, non- or poorly metastatic murine melanoma clones form lung metastases in gld mice as efficiently as their Fas-insensitive metastatic clonal counterparts in wild-type mice (6). These data suggested that endogenous Fas/FasL interactions in the host played a direct role in the regulation of metastatic formation. Therefore, loss of Fas function alone was characterized as both necessary and sufficient for tumor progression in that model. As mentioned before in our study, when CMS4 cells were rendered Fas-resistant (by vFLIP transfection), it was insufficient for the growth of lung metastases. In a reciprocal fashion, the administration of CMS4 cells or the CMS4 sublines to syngeneic gld mice did result in an increase in lung metastases in at least some of the recipients (Figs. 3 and 6), which partially was consistent with earlier findings (6). Presently, in our study it remains unclear why the metastatic outcome was not identical regardless of whether the Fas/FasL interaction was disrupted at the level of Fas on the tumor vs FasL by the host. Despite the exact reasons for this disparity or variability, it is important to emphasize that the extent of metastasis seen with the parental CMS4 population in the gld group overall was not significantly different from that seen in the wild-type mice (or with the vFLIP-transfectants) and was still significantly lower compared with that achieved by CMS4-met. These observations reinforced the notion that the Fas/FasL system played an important, but not exclusive, role in determining the full magnitude of the metastatic outcome.

Overall, the results reported in this study and elsewhere (6) suggested that, depending upon the cellular and molecular composition of a given tumor population, the regulation or dysregulation of the Fas pathway may influence the neoplastic process via multiple mechanisms. In the CMS4 model, these data revealed a previously unrecognized and novel contribution of the Fas pathway in tumor progression and suggested that Fas-based interactions impose an immunologic or biologic selective pressure favoring the emergence of such metastatic subpopulations. However, our findings in the CMS4 model also introduce the hypothesis that the contribution of other tumor-associated genetic events in combination with alterations in Fas expression or function are required for an optimally productive metastatic phenotype (Fig. 7). In that sense, the evolution of malignant potential likely reflects the relative contributions of not only host-derived factors, but also intrinsic characteristics of an advancing neoplasm. Thus, detailed studies focused on appropriate comparisons of tumor cell lines/sublines from these model systems by cDNA microarray techniques, for example, may serve to identify additional relevant molecular features or events that associate with or perhaps underlie the metastatic phenotype. Identification of such additional tumor-associated gene(s) will not only improve our understanding of the pathogenesis of the neoplastic process, but will also aid in identifying potential molecular targets for clinical intervention.

We thank R. Siegel for the vFLIP construct, S. Gendler and J. Schlom for transgenic mice, J. DiPietro for tumor cell lines from the transgenic mice, L. Granger of the National Cancer Institute Core Flow Cytometry Facility for assistance with cell sorting, K. Rogers and D. Haines for assistance with immunohistochemistry analyses, E. McDuffie for technical assistance, S. Caldwell for assistance with tumor injections, and D. Weingarten for editorial assistance.