Tumorigenicity of the Mixed Spindle-Epithelioid SP6.5 and Epithelioid TP17 Uveal Melanoma Cell Lines Is Differentially Related to α5β1 Integrin Expression

Abstract

purpose. It has been suggested that the epithelioid morphology and high aggressiveness that is typical of the uveal melanoma cell line TP17 is dependent on the loss of α5β1 integrin expression at the cell surface. The purpose of the current study was to test this hypothesis in the TP17 cell line and investigate the role this integrin may play in the tumorigenicity of the SP6.5 cells, a mixed spindle–epithelioid culture-type human uveal melanoma that shows tumorigenic properties clearly distinct from that of TP17 cells.

methods. Expression of the α5 integrin subunit was restored in the α5-TP17 cell line by stably transfecting the cells with a recombinant plasmid encoding the integrin subunit. Flow cytometry and adhesion assays on fibronectin (FN)-coated culture plates were used to monitor α5 expression in the cells. The effect of α5 expression on both tumorigenicity and cell proliferation was evaluated in vivo in nude mice. In vitro growth properties of the α5⁺ TP17 cells was evaluated by cell counting and compared with that of the α5 parental TP17 cell line. The influence exerted by the α5 integrin subunit on the tumorigenic and proliferative properties of the SP6.5 cells was evaluated in vivo in nude mice by exposing the cells to increasing doses of a blocking antibody directed against theα 5-subunit before subcutaneous injection, and compared with the results obtained with untreated SP6.5 cells.

results. Expression of the α5 integrin subunit in the α5-TP17 cells was successfully achieved, as evidenced by both flow cytometry and adhesion assays on FN-coated culture plates. Restoring expression of α5 in TP17 cells enhanced epithelioid cell morphology and increased the growth properties of this cell line in vivo. The ability of the SP6.5 cells to yield subcutaneous tumors was found to be concentration dependent and was reduced in a dose-dependent manner when the cells were exposed to the anti-α5 blocking antibody.

conclusions. Restoring expression of α5 in the α5-negative TP17 uveal melanoma cell line influenced the proliferative properties of these cells but did not alter its tumorigenic potential. In contrast, the ability of the SP6.5 cells to yield tumors in vivo in nude mice appeared to be related to expression of this integrin.

Uveal melanoma represents the most common intraocular tumor in the adult population ^{1 2} and is considered the paradigm of hematogenous invasion of tumor cells, because no lymphatic circulation is present in the eye. ³ The metastatic ability of uveal melanoma is particularly high and specific to the liver. The estimated survival time after 5, 10, and 15 years of treatment are 72%, 59%, and 53%, respectively, and the disease is 100% lethal when liver metastasis is involved. ⁴ Among the prognostic criteria used to evaluate the malignancy of uveal melanoma, cell morphology is one of the most revelatory of aggressiveness. Whereas pure spindle cell tumors are less aggressive than pure epithelioid cell tumors, mixed-cell tumors that still are spindle cell–containing behave more like epithelioid than spindle cell tumors. ^{5 6 7}  

Integrins, such as the fibronectin (FN)-binding, membrane-bound receptor α5β1, are heterodimers made up of an α- and aβ -subunit held together through noncovalent interactions. ^{8 9} They represent the main class of cell adhesion receptors for the various components of the extracellular matrix (ECM). ¹⁰ Considering that adhesive interactions are thought to be required for cell guidance during development and in the maintenance of the body’s structure, it is therefore not surprising that integrins also play important roles during cancer progression. ¹¹ Many studies have shown that α5β1 and its corresponding ligand (FN) are frequently associated with tumorigenicity. ^{11 12} The production of a cell-surrounding, FN-insoluble matrix is mainly mediated by theα 5β1 integrin. ^{13 14} This specific property of α5β1 could explain why a decrease in α5β1 expression in Chinese hamster ovary (CHO) cells also leads to an increase in their tumorigenic ability. ¹⁵ Expression of α5β1 expression is also frequently altered in most tumor cells. ¹¹ Human colon carcinomas HT-29, which do not express α5β1, lose their aggressive phenotype when α5 expression is restored through stable transfection of an appropriate α5-expression vector in these cells. ¹⁶ Restoring expression of α5 in CHO cells has also been reported to affect negatively the tumorigenicity of this cell line. ^{15 17} On the contrary, many cell lines derived from human colon carcinoma exhibit an aggressive phenotype that correlates with an increased expression of the α5 integrin subunit at the cell surface. ¹⁸  

These contradictions can be reconciled if we take into consideration the anoikis phenomenon. Anoikis, meaning homelessness, is derived from ancient Greek and is used to define the apoptosis related to the absence of cell adhesion or to an adhesion mediated by the wrong adhesion molecules. ^{19 20} Thus, the presence of a specific integrin and the absence of its corresponding ligand may lead to growth arrest and, most likely, to apoptosis. ²¹ One other major function of α5β1 is the critical function this integrin plays in the formation of an organized actin microfilament bundle. ⁹ In turn, formation of such a structure is also a prerequisite for the further formation of focal contacts. Strong evidence suggests that one of the major functions played by these structures is to bring together molecules that account for cytoskeletal organization and signal transduction—cellular fate being dependent on both processes. ^{8 9}  

We recently characterized two cells lines derived from primary human uveal melanoma: the SP6.5 and the TP17 cell lines. ²² Both were derived from primary tumors essentially made up of spindle and epithelioid cell types, respectively. The morphology of these cell lines in vitro also reflects their in vivo origin. We reported that integrin α5β1 was expressed in human uveal melanocytes as well as in SP6.5 cells in vitro. However, no membrane-bound α5β1 was detected in the highly tumorigenic TP17 cell line. The present study was designed to determine how expression of the α5β1 integrin at the cell surface of these melanoma cell lines is related to their respective tumorigenic and proliferative properties, both in vitro and in vivo. 

Materials and Methods

This study was conducted according to our institution’s guidelines and the Declaration of Helsinki, and the protocols were also approved by the institution’s Committee for the Protection of Human Subjects. In addition, all experiments conducted in nude mice strictly adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Cell Culture

The melanoma cell lines (SP6.5 and TP17) were obtained from primary tumors extracted from the eyes of patients with diagnosed uveal melanoma, as previously detailed. ²³ All tumor cell lines were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS) under 5% CO₂ at 37°C. Gentamicin was added to all media at a final concentration of 15 μg/ml. Geneticin (G418; Wisent, Montréal, Québec, Canada) was also added at a final concentration of 100 μg/ml to the culture medium of those TP17 cells stably transfected with the α5 integrin subunit cDNA. 

Stable Transfection

The α5-subunit subcloned into the Tet-off vector system (Clontech, Palo Alto, CA), along with the neomycin-encoding plasmid pUHD15-1neo were both kindly provided by Michael G. Brattain (Department of Biochemistry and Molecular Biology, Medical College of Ohio, Toledo, OH). TP17 cells were transfected using the calcium phosphate precipitation procedure. ²⁴ Cells were plated at a density of 3 × 10⁵ cells per 60-mm petri dish and incubated at 37°C for 24 hours before they were transfected. Transfected cells received 20 μg of the α5 expression plasmid and 2μ g of the neomycin-encoding plasmid, both previously linearized and further incubated for 18 hours before being washed with PBS. Cells were maintained in DMEM for 24 hours before the addition of geneticin in the culture media at a concentration of 300 μg/ml. Selection pressure was maintained for 2 weeks. Individual clones that expressed theα 5-subunit at varying levels were sorted out by flow cytometry (Epics XL; Coulter Electronics, Miami, FL). Wild-type cells transfected solely with the neomycin-encoding plasmid were used as the control. 

Assays of Adhesion and Inhibition of Adhesion

Adhesion assays were performed as previously described ¹⁶ with 96-wells plates previously coated with doubling dilutions (0.08–10 μg) of human FN (Boehringer Mannheim, Laval, Québec, Canada) in triplicate. Inhibition of cell adhesion was assayed as described, ²² by exposing cultured cells to twofold serial dilutions (from 250 to 1 ng) of a monoclonal antibody (mAb) directed against the α5 integrin subunit (IIA1) before their loading into FN-coated culture wells (10 μg/ml FN). The cells selected for these assays were plated in FN-coated culture wells at a density of 5 × 10⁴ cells per well and allowed to adhere for 1 hour at 37°C. After incubation, wells were washed with adhesion medium (37°C) and twice with PBS. The cells were fixed and stained as described. ²² Absorbance was determined at 570 nm. 

Determination of Cells Growth Properties

Approximately 1.0 × 10⁴ cells from each of the cell lines used (TP17α5/30X, TP17α5/90X, and TP17Neo) were plated into a 24-well plate. Every 24 hours and up to 7 days, three cell samples were harvested from these plates and counted with a cell counter (ZM; Coultronics, S.A., Margency, France). The cell counts were plotted against time to obtain a growth curve for each cell line. 

Flow Cytometry

All flow cytometry analyses were conducted on the human uveal melanoma cell lines, as recently described ²² using mAbs directed against both the human α5 integrin subunit (IIA₁; PharMingen, Mississauga, Ontario, Canada) and the C-terminal end of the DNA-binding domain of bovine poly(ADP-ribose)polymerase (PARP; C-2-10 ²⁵ ; used as a negative control) as the primary antibodies and an FITC-conjugated secondary antibody (Sigma, St. Louis, MO). Cells were resuspended and analyzed using a flow cytometer (Epics XL; Coulter Electronics). 

Tumorigenicity Assays

Exponential cell cultures of both the TP17Neo and the TP17α5/30X cells were isolated from the culture plates with PBS-EDTA 0.02%. After centrifugation, 4 × 10⁵ and 8 × 10⁵ cells from each cell line were resuspended in 0.1 ml serum-free DMEM. TP17α5/30X cells were injected subcutaneously (SC) using a 1-ml syringe and a 23-gauge needle on the anterior part of the right flank of Crl:CD1-nuBR nude mice (Charles River, St. Constant, Québec, Canada), whereas TP17Neo was inoculated as a growth control on the left flank of each mouse. Ten 7-week-old female mice were used for each transfected cell line. Mice were housed in vinyl cages equipped with air-filter lids that were kept in laminar air-flow hoods and maintained under pathogen-limiting conditions. Cages, bedding, food, and water were autoclaved before use. Mice were examined at 3, 5, 7, 10, 13, 17, and 19 days after injections, and tumors, when apparent, were measured on the horizontal and vertical axes with slide calipers. Volume was determined by the equation V = (L × W ²) × 0.5 where V is volume, L is length, and W is width. 

When the tumors reached the approximate size of 15 × 15 mm, mice were killed and tumors excised. The experiment with the blocking antibody directed against the α5 integrin subunit was assayed as previously described. ²⁶ Briefly, 5 × 10⁶ SP6.5 cells were resuspended in 0.1 ml DMEM without sera and incubated 2 hours at 4°C with 0.1 μg, 0.5 μg, and 5.0 μg of the anti-α5 blocking antibody (IIA₁; PharMingen) before being injected SC in the anterior part of both the right and left flanks of the nude mice. Approximately 5 × 10⁶ cells were also incubated with 5.0 μg anti-entamoeba (kindly provided by Jacques Hébert, Unit of Rheumatology and Immunology, Center Hospitalier Universitaire de Laval, Center Hospitalier Universitaire de Québec, Canada) as a control in 0.1 ml DMEM without sera. Five 7-week-old female mice were used for each condition. 

In Vivo and In Vitro Assessment of FN Secretion in SP6.5 Cells

Immunofluorescence analyses were also performed on SP6.5 cells plated at 5 × 10⁴ cells per 113-mm² glass slide (Bellco Glass, Vineland, NJ), allowed to grow at 37°C under 5% CO₂ for 2 days, and then fixed with 70% ethanol for 10 minutes at −20°C. Each cell-containing slide was then covered with 50 μl of a 1:50 dilution (in PBS containing 1% BSA of a mouse anti-human FN monoclonal antibody; Chemicon, Temecula, CA). Incubation proceeded at room temperature for 45 minutes in a moist chamber. Slides were then rinsed with PBS and further covered with 50 μl of a 1:200 dilution (in PBS-BSA 1%) of a rhodamine-coupled secondary antibody (goat anti-mouse IgG, tetrarhodamine isothiocyanate [TRITC] conjugate; Sigma). Incubation was performed in the dark for 30 minutes, as before. Slides were then washed with PBS and mounted with coverslips in mounting medium (ProLong; Molecular Probes, Eugene, OR). 

Each uveal melanoma tumor generated by the SC injection of SP6.5 cells in nude mice was embedded in optimal cutting temperature (OCT) tissue-embedding medium (Tissue-Tek; Miles, Inc., Elkhart, IN), frozen in liquid nitrogen, and stored at −80°C until use. An indirect immunofluorescence assay was performed on acetone-fixed (10 minutes at− 20°C) cell-containing glass slide as previously reported. ^{27 28} Sections (4 μm thick) were incubated with anti-human FN as the primary antibody for 45 minutes, followed by the appropriate conjugated antibody for 30 minutes. Cell nuclei were also labeled with Hoechst 33258 reagent (Sigma) after immunofluorescence staining. Slides were viewed with an epifluorescence microscope (Diaphot 300; Nikon, Tokyo, Japan) and photographed with a ×20 objective. All slides were then observed under a microscope (Optiphot; Nikon), equipped with epifluorescence, and photographed (Tmax 400 ASA film; Eastman Kodak, Rochester, NY). 

Statistical Analyses

The Kruskal-Wallis test was used to compare the distributions of the tumor volumes yielded by the SC injection in nude mice of the SP6.5 cell line exposed to increasing doses of the α5Ab within each of the four experimental groups tested. SD has also been provided where indicated. 

Results

Restoring the Expression of the α5 Integrin Subunit in the TP17 Melanoma Cell Line

To investigate whether the absence of α5β1 expression in TP17 cells is involved in its tumorigenic properties, derivatives of this cell line in which a recombinant expression vector encoded high levels of the human α5 integrin subunit were produced through stable transfection. Two clones, 30 and 90 times more fluorescent than the TP17Neo negative control, were isolated and designated TP17α5/30X and TP17α5/90X, respectively (Fig. 1A) . To verify the presence of a functional α5β1 integrin in the TP17α5/90X cell line, we performed adhesion assays on culture plates coated with increasing amounts of FN. The TP17α5/90X cells adhered twice as much as TP17Neo cells on FN (Fig. 1B) . This improved adhesion of the TP17α5/90X cell line on FN-coated culture wells was further validated by performing assays of inhibition of adhesion. As expected, adhesion of the TP17Neo control cell line, which expresses no detectable membrane-bound α5β1, was not significantly affected by the presence of increasing amounts of the α5 antibody in this assay, even when used at 250 ng (Fig. 1C ; results shown are compared with those obtained when these cells are seeded on noncoated wells, referred to as 100% for comparison purposes). On the contrary, adhesion of TP17α5/90X cells was inhibited by approximately 40% when exposed to concentrations of α5Ab of 62 ng and more, suggesting that its ability to bind FN-coated culture wells is in part dependent on the presence of the α5 integrin subunit. We therefore concluded that expression of a functional α5β1 integrin was restored in the TP17α5/90X cells and that the adhesive properties of these cells for FN were substantially improved. 

Influence of the α5 Integrin Subunit on Morphology, Proliferation, and Tumorigenicity of the TP17α5-Positive Cells

We postulated that the absence of α5β1 at the cell surface of TP17 cells might account in part for the morphologic changes that are typical of these cells. Indeed, observation under light microscopy of the TP17α5⁺ cell lines revealed an enhancement of epithelioid cell morphology when compared with that of TP17Neo (Figs. 2B 2A , respectively). We then determined the rate of cell proliferation for both the TP17α5/30X and TP17Neo cell lines to evaluate whether the aggressiveness of the former was somehow altered by the de novo expression of the α5-subunit. As shown on Figure 3A , the proliferation rate measured for TP17α5/30X was practically identical with that measured for TP17Neo. Similar results were obtained with the TP17α5/90X cell line (data not shown). However, TP17α5⁺ cells (30X and 90X) clearly were more resistant to growth arrest than TP17Neo when they reached 6 or 7 days of culture under complete medium culture conditions (Fig. 3A) , which corresponds to the moment the cell cultures reach confluence. We then examined whether restoring expression of α5β1 in TP17 cells would improve the TP17 cells’ ability to proliferate under stress condition, as has been reported in other tumor cells, ²⁹ by repeating the proliferation experiment under serum-free conditions. Under such culture conditions, TP17Neo stopped proliferating 48 hours after they were plated, whereas TP17α5/30X efficiently proliferated for up to 5 days (Fig. 3B) . Identical results were also obtained with the TP17α5/90X clone (data not shown). 

To investigate whether re-expressing the α5-subunit would confer on TP17α5⁺ cells (in this case, the TP17α5/30X) tumorigenic properties distinct from those of TP17Neo, both cell lines were injected SC in nude mice, and tumor formation was monitored. No significant difference in tumor size was observed when either the TP17α5/30X or the TP17Neo cells were injected (Fig. 4) . Although TP17α5⁺ clones were as aggressive as TP17Neo, the measurement of growth rates revealed subtle differences. It is interesting to point out that at specific times (such as 10 days after SC injections), tumors yielded by TP17Neo were clearly much smaller than those resulting from TP17α5/30X (48 and 189 mm³, respectively, at day 10). We therefore conclude that restoring expression of α5 in TP17 cells both enhanced the epithelioid morphology and increased the cells’ resistance under stress conditions as suggested by a reduced cell mortality at confluence and the ability to grow under serum starvation. 

In Vivo and In Vitro Secretion of FN by TP17 and SP6.5 Cells

To evaluate the requirement for the secretion of FN in the tumorigenicity of melanoma in relation to expression of the α5β1 integrin, we investigated the expression of FN in SP6.5 and TP17 cells. We have demonstrated by RT-PCR that the FN gene is transcribed in SP6.5 but not in TP17 cells. ²² To determine whether this pattern of expression also translates into the presence or absence of secreted FN, indirect immunofluorescence was performed on in vitro (in cell monolayers, shown in Fig. 5A for SP6.5 cells as they appear in phase contrast microscopy) and in vivo (in SC tumors from nude mice, shown in Fig. 5E , for tumors yielded by SP6.5 as they appear in phase-contrast microscopy) samples of SP6.5 and TP17 cells. Both in vitro and in vivo samples from TP17 were negative for FN staining (data not shown). However, a clear network of FN could be observed for SP6.5 in vivo (compare Fig. 5G with its corresponding negative control in Fig. 5H ). Only weak, but significant, FN staining was detected in the in vitro sample (compare Fig. 5C with its corresponding negative control in Fig. 5D ). In the in vivo tumor sample, FN expression was observed outside the cells and surrounded them to form isolated aggregates of cells (Fig. 5G) , as suggested by the absence of overlap between Hoechst staining of the cells nuclei (Fig. 5F) and FN fluorescence (Fig. 5G) . However, both the Hoechst and FN stainings perfectly overlapped in the SP6.5 in vitro samples (compare Figs. 5B and 5C ). We therefore concluded that SP6.5 but not TP17 cells possess the ability to organize complex FN networks in vivo. 

Influence of a Blocking Antibody against the α5 Integrin Subunit on SP6.5 Cell Tumorigenicity

To further investigate the relationship between the α5β1 integrin and its ligand (FN) on the tumorigenic ability of the SP6.5 cell line, cells previously incubated with increasing amounts of a blocking antibody against the α5-subunit were injected SC into nude mice. The influence of the α5Ab appeared to be dose dependent (Fig. 6) . Incubation of SP6.5 cells with 0.1 μg antibody had no statistically significant effect on average tumor size (304 vs. 320 mm³) whereas incubation with 5 μg antibody resulted in significant reduction (81%; Kruskal-Wallis, P < 0.05) in the tumor volume (62 vs. 320 mm³). Although much smaller on average than the control group (214 vs. 320 mm³), tumors yielded by SP6.5 cells exposed to 0.5 μg α5Ab were not found to be statistically different from tumors in both the control group (no Ab) and the 0.1-μg α5Ab group. The specificity of the α5 antibody’s influence on tumor growth was provided by the absence of any comparable effect when the antibody was replaced with 5 μg of an anti-entamoeba (data not shown). 

Discussion

We recently characterized four spontaneously transformed cell lines derived from primary human uveal melanoma. ²² Among them, the SP6.5 cell line, a spindle–epithelioid mixed type culture, was obtained from a primary tumor that comprised essentially spindle-like cells, whereas the highly invasive TP17 cell line was derived from a tumor mainly composed of epithelioid-like cells. Both these cell lines displayed properties that are typical of their morphology, such as aggressiveness and proliferative ability. Expression of the α5β1 integrin was not observed at the cell surface of TP17 cells in contrast to the remaining cell lines (SP6.5, SP8.0, and TP31, as well as primary cultured human uveal melanocytes). ²² We therefore suggested that the absence ofα 5β1 could account, at least in part, for the aggressiveness and epithelioid morphology that are typical of the TP17 cells. ²² To test this hypothesis, we stably transfected a recombinant plasmid that allowed expression in these cells of high levels of the human α5 integrin subunit. The TP17α5⁺ cells exhibited a slight morphologic change characterized chiefly by an enhancement of the epithelioid morphology and a more compact growth pattern. Furthermore, subtle changes were observed with the TP17α5⁺ cells when they were grown to confluence. Indeed, they were clearly more resistant to cell senescence than the wild-type TP17 cells from which they were derived. TP17α5⁺ cells were also more resistant to the growth arrest that normally occurs after serum withdrawal from the culture medium. However, these changes were not sufficient to significantly reduce the in vivo tumorigenicity of TP17α5⁺ cells when injected in nude mice. 

In any case, these changes could be compared with those observed in the HT-29 cell line derived from a human colon carcinoma that has also been stably transfected with the α5 full-length cDNA. ¹⁶ Although re-expression of the α5-subunit virtually eliminated the tumorigenic properties of HT-29 cells, the in vitro doubling time was not affected by the presence of the α5-subunit when complete culture medium was used, as for TP17α5⁺ cells. Furthermore, very similar results were obtained regarding the cells’ resistance to stress conditions. Indeed, in HT-29 as well as in TP17 cells, restoring α5 expression reduced cell proliferation ¹⁶ and increased the threshold for triggering apoptosis under stress conditions. ²⁹ However, in TP17α5⁺ transfectants, it appeared that the former effect was, in a large part, eclipsed by other proliferative and tumorigenic mechanisms. Under the experimental conditions used in this study, α5 expression in TP17 cells had a beneficial effect on cell survival and no apparent effect on tumorigenic potential. 

The melanoma cell line SP6.5 was clearly more affected by the expression of the α5-subunit. This cell line expresses α5 endogenously and has reduced tumorigenicity compared with TP17 cells. We demonstrated the negative influence that a blocking antibody against the α5-subunit has on the tumorigenic potential of these cells when injected SC into nude mice. This effect was found to be specific, because it could not be reproduced with the control antibody. Indirect immunofluorescence confirmed the endogenous expression of FN and the potential of these cells to organize it into an insoluble network, both in vivo and in vitro. The α5β1 integrin is well known to play an active role in the elaboration of the FN network, and its absence drastically impairs the cells’ ability to organize such a structure. ^{13 14} Interaction of this integrin with its ligand enhances cell proliferation and protects them from apoptosis. This relationship between anchorage dependency and apoptosis is best defined by the term anoikis in which loss of adhesion, or adhesion mediated by the inappropriate integrin, triggers apoptosis. ¹⁹ Expression of bcl-2, a well-known anti-apoptotic molecule, is induced by the binding of FN to theα 5β1 integrin and was believed to account for the protective effect of cell adhesion against anoikis. ³⁰  

The mechanism by which re-expression of α5 into HT-29 and CHO cells affects tumorigenicity seems to correlate with the absence of FN, which also results in a modification of the adhesion pattern of these cells. Gong et al. ¹⁸ isolated many cell lines from human colon carcinoma and organized them into specific phenotypes related to their corresponding aggressiveness. Among them, cells from group I, which were also the most aggressive, expressed both molecules (α5β1 and FN) whereas those from group III, the least tumorigenic, expressed FN but not α5. Restoring expression of α5 through stable transfection in a cell line (GEO) from group III drastically increased their aggressiveness in vivo, therefore providing evidence that the effect ofα 5β1 on cell tumorigenicity depends on the endogenous expression of FN. ¹⁸ Hence, the effect of blocking the α5β1 integrin present on the cell surface of the SP6.5 melanoma with the anti-α5 antibody mimics the cell condition of HT-29α5⁺ in which α5β1 is present but not the ligand, leading to a reduction of the cell’s tumorigenic potential. 

The experiments presented in this study confirmed that the spindle–epithelioid-like tumor-derived SP6.5 cells still retained their ability to respond to the ECM through their membrane-bound integrin α5β1. In contrast, epithelioid-like tumor-derived TP17 cells were totally independent of such interactions. Taking into account the previously reported studies, ^{15 16 17} we expected a reduced tumorigenicity for TP17α5⁺ cells. The fact that cell tumorigenicity was barely altered by restoring the expression of α5 in these cells suggests that one (or a few) other mechanism(s) contribute to this physiological property. This (or these) mechanism(s) more profoundly affect the interaction between cells and the ECM in a manner that bypasses the information transduced by integrins. The results obtained with the TP17 cell line also highlights the dichotomy of the α5-subunit’s effect on apoptosis. On the one hand, the absence of α5β1-mediated adhesion triggered apoptosis; on the other hand, expression of α5 increased the apoptosis threshold triggered by stress conditions. Indeed, TP17α5⁺ turned out to be more resistant to stress conditions such as serum starvation or high cell confluence. Although, we do not know at the moment whether such resistance is related to apoptosis in these cells, anoikis is unlikely to be involved in that phenomenon, because proliferation occurred despite the absence of anchorage to FN in vivo. We believe this phenomenon may be dependent on the action of an anchorage-independent mechanism. 

The loss of α5β1 does not seem to be involved in the morphologic changes observed in uveal melanoma. This drastic morphologic alteration implies profound reorganization of the cell cytoskeleton. Becauseα 5β1 is involved in the focal contact assembly, we expected to see a major reorganization of the cell cytoskeleton related to that function. Whatever the level of α5 expression that we restored in TP17 cells, no clear morphologic changes occurred, apart from the enhancement of the epithelioid morphology mentioned earlier. Uveal melanoma is not the only type of cancer in which morphologic changes correlate with cell aggressiveness. Studies reporting alterations in the expression of intermediate filaments (IFs) are emerging. ^{31 32} These components of the cell cytoskeleton are used as markers to confirm the origin of a cell line. ³³ Epithelial and endothelial cells express exclusively keratin heteropolymers, whereas cells from the stromal compartment express vimentin. ^{34 35} It is possible that the morphologic changes that are typical of the TP17 cell line were not solely related to integrins but rather to altered expression of such intermediate filaments. Indeed, and unlike normal uveal melanocytes, all uveal melanoma appear to express vimentin. ^{36 37} Most of all, uveal melanoma cell lines originating from primary tumors made up of mixed populations of spindle, intermediate, and epithelioid cells, that coexpress both vimentin and keratin types 8 and 18 are many times more invasive in vitro than normal uveal melanocytes or uveal melanoma that express only vimentin. ³⁶ Among the many properties that are typical of tumor cells that coexpresses vimentin and keratins 8 and 18 are an increased ability to migrate and invade ECM in vitro, a switch in the pattern of expression of certain integrins, and an increase in tyrosine phosphorylated proteins that colocalize with β1 integrins (to which belong α5β1) on the leading invasive edge. ^{38 39 40}  

Taken together, these observations suggest that the metastatic ability of human uveal melanoma may depend on alterations in the expression of both IFs and integrins. They also provide a rationale for conducting additional studies to develop therapeutic intervention strategies that would help in preventing the two major problems with which the clinician is faced in the treatment of patients with uveal melanoma: death caused by metastatic uveal melanoma and lost of sight. 

 Supported by the Réseau de Recherche en Vision du Fonds de la Recherche en Santé du Québec (FRSQ), the Fondation des Maladies de l’Oeil, and the Cancer Research Society. AB and PC are recipients of Studentships from the Fondation de l’Université Laval and from the FRSQ, respectively. SLG is an FRSQ scholar.

 Submitted for publication January 3, 2001; revised June 14, 2001; accepted July 18, 2001.

 Commercial relationships policy: N.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked“ advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Influence of overexpression of α5 on TP17 cell adhesion to FN-coated culture dishes. Cell surface expression of the α5 integrin subunit was monitored by flow cytometry (A) in two derivatives from the TP17 cell line (TP17α5/30X: thick tracing; and TP17α5/90X: thin tracing) that expressed various amounts of this integrin, as well as in the α5-negative TP17Neo cell line, using the mAb IIA₁ (data not shown). As a negative control, a monoclonal antibody directed against bovine PARP was used as the primary antibody (dotted tracing). The relative fluorescence is shown as a logarithmic scale in the x-axis and the cell number as a linear scale in the y-axis. Data are from one of three similar experiments. Because the fluorescence curve obtained for TP17Neo overlapped that obtained with the PARP Ab used as the negative control, it was not included on the figure, for simplicity. (B) The TP17α5/90X derivative and the TP17Neo cell line were plated on FN-coated culture plates (0–10 μg) and allowed to adhere before they were quantified. SD is provided for each value. (C) TP17α5/90X and TP17Neo cells were exposed to twofold serial dilutions of an antibody (IIA1) directed against the α5 integrin subunit and then plated on FN-coated culture plates (10 μg/ml). The percentage of adhering cells was plotted against the amount of α5 antibody used (in nanograms). Data are the average of five experiments, each conducted in triplicate.

Figure 2.

 

Phase contrast images of uveal melanoma cell lines. Phase-contrast micrographs of monolayer cultures of the following human uveal melanoma cell lines: neoplasic TP17 (A) cells and its derivative TP17α5/90X (B) in which expression of the α5 integrin subunit was restored by stable transfection of the α5 expression plasmid. Magnification, ×200.

Figure 3.

 

In vitro growth properties of the TP17Neo and TP17α5/30X cells. TP17Neo and TP17α5/30X cells were seeded onto a 96-well plate and maintained in DMEM containing 10% FBS (A) or in serum-free medium (B). The cell number is a logarithmic scale in the y-axis and the time of incubation is a linear scale in the x-axis.

Figure 4.

 

In vivo proliferative properties of the TP17Neo and TP17α5/30X cells. Both the TP17Neo and the TP17α5/30X cell lines were injected into nude mice and tumor formation was monitored. The tumor volume (in cubic millimeters ± SD) for each individual measurement is on a linear scale in the y-axis and the time of incubation (in days) is in the x-axis.

Figure 5.

 

Immunofluorescence analyses of FN secretion by SP6.5 cells in vitro and in vivo. Secretion of FN by SP6.5 cells grown onto tissue culture dishes in vitro or by the SC tumors they yield after their injection in nude mice in vivo was examined by indirect immunofluorescence. Phase-contrast micrographs show an SP6.5 monolayer culture (A) or the cryostat section of the SC tumor they produced in nude mice (E). Hoechst staining showed in the cell nuclei on both the SP6.5 monolayer culture (B) and the SP6.5 tumor section (F). Immunofluorescence analysis was conducted with the anti-human FN antibody on the tissue-cultured SP6.5 cells (C) or on the SP6.5 tumor section (G). In negative control cultures, the specific (anti-FN) but not the secondary antibody was omitted on both the monolayer culture (D) and the tumor section (H). Magnification, ×20.

Figure 6.

 

Influence of antibody inhibition of the α5 integrin subunit on the tumorigenic properties of SP6.5 cells. SP6.5 cells incubated with either no or increasing concentrations (0.1, 0.5, and 5.0 μg) of an anti-α5 antibody (IIA1) were injected SC into athymic nude mice. All mice were killed 22 days after the injection and the volume of each tumor determined. Horizontal bars: mean value for each individual condition examined.

The authors thank Guy Pelletier and Alain Rousseau for expert advice on uveal melanoma; Christian Salesse for critically reviewing the manuscript; and Jean Morissette, Oncology and Molecular Endocrinology Research Center, Centre Hospitalier Universitaire Québec-Centre Hospitalier Universitaire Laval, Québec, Canada, for statistical analyses.