Main

The phagocytic-like behavior of tumor cells was first observed a century ago (Steinhaus, 1891) and confirmed by many other authors who described phagocytic activity of various murine and human tumor cells against both dead cells and undefined particles (DeSimone et al, 1980; Marin-Padilla, 1977). More recently, many morphologic aspects associated with phagocytosis have been identified in tumors (Banerjee and Harris, 2000), and the phagocytic-like activity of these cells has been associated with invasiveness (Coopman et al, 1998; Montcourrier et al, 1994). Moreover, other histopathologic signs of tumor malignancy, recalling an in vivo phagocytic-like behavior of tumor cells, have been described. Between the above mentioned signs there are the so-called “signet-ring cells,” mostly typical of adenocarcinoma (Kuroda et al, 1999) and characterized by the presence of intracytoplasmic mucin displacing the nucleus. However, signet-ring cells can also be found in nonepithelial tumors, such as malignant lymphomas and melanomas (Tsang et al, 1992). The nature of these cells, particularly in nonepithelial tumors, is not known, but it is conceivable that an aberrant scavenging activity underlies this phenomenon (Banerjee and Harris, 2000; Kuroda et al, 1999). Altogether these reports suggest that active phagocytosis occurs in human tumor cells, but the phagocytic apparatus of tumor cells has never been defined and the mechanism(s) underlying this phenomenon have not been investigated.

To address the phenomenon of phagocytosis, its underlying mechanism(s), and its possible role in tumor aggressiveness, we used human melanoma cells as a prototypic model. In fact, it has been shown that melanosomes are specialized lysosomal-like organelles (Dell’Angelica et al, 2000), able to fuse with phagosomes (Le Poole et al, 1993). This evidence is supported by the presence of an acidic pH, hydrolases, and acidic phosphatases in these organelles, as well as by their expression of some lysosomal-associated proteins (Lamp) (Dell'Angelica et al, 2000). Moreover, during melanogenesis, melanoma cells (similar to professional phagocytes) increase their amount of tyrosinase-produced superoxide anion (Koga et al, 1992).

Some recent data, obtained in murine macrophages, suggest a crucial role of the actin cytoskeleton in different phases of the phagocytosis process, such as actin assembly on secondary phagosome membranes (Defacque et al, 2000) and vesicular trafficking (Tauton, 2001), both caused by an association between cellular membranes and the actin cytoskeleton through ERM (ezrin, radixin, and moesin). Moreover, we have recently shown that the bacterial-toxin–induced activation of some actin-regulatory proteins (ie, ρ-GTPase) results in the induction of a genuine phagocytic activity in a human epithelial cell line (Fiorentini et al, 2001), further suggesting a crucial role of the actin cytoskeleton in this process.

On the basis of these considerations, the goals of this study were as follows: (i) to evaluate the in vivo occurrence of phagocytosis in human melanoma; (ii) to assess the presence of phagocytosis in human melanoma cells and characterize the phagocytic machinery of melanoma cells and its possible relationship with the phagocytic activity; (iii) to evaluate both the ability of melanoma cells to ingest and phagocyte amorphic material (latex beads), yeasts, and dead cells and the relationships between phagocytic activity and the malignancy of melanoma; and (iv) to understand the key mechanism(s) underlying this phagocytic activity. Our results showed that human melanoma cells deriving from metastatic lesions, but not from primary tumors, possess an efficient phagocytic machinery responsible for a macrophage-like activity against latex beads, yeast, and apoptotic cells and that ezrin holds a key role in the phagocytic activity of metastatic melanoma cells.

Results

In Vivo Occurrence of Phagocytic Behavior in Human Melanoma

We first investigated the occurrence of phagocytosis in samples of human melanomas. Histologic analysis of melanoma tumors revealed the presence of in vivo phagocytosis in human melanoma (Fig. 1A). Moreover, immunohistochemistry showed the presence of apoptotic (Tunel positive) cells within cytoplasmic vacuoles of in vivo phagocyting cells (Fig. 1B). This first set of results showed that phagocytic activity against apoptotic cells occurred in vivo in human malignant melanoma.

Figure 1
figure 1

In vivo autophagocytosis in melanoma tumors. A, Histologic preparation of a metastatic melanoma sample. A large epithelioid tumor cell contains an (apparently) intact tumor cell (arrow). The nuclear features of the “ingested” cell are the same as those present in the rest of the sample. [Hematoxylin and eosin (HE): magnification, × 350]. B, Histologic preparation of a metastatic melanoma sample. In the center of the field, a Tunel-positive (apoptotic) cell within a bigger, Tunel-negative cell is shown. (HE: magnification, × 350).

Full size image

Expression and Distribution of Lysosomal Proteins and Spontaneous Phagocytic Activity of Human Melanoma Cells

Thus, we investigated the expression of phagocytic markers in human melanoma cells as compared with monocyte-derived macrophages. The results showed that 100% of melanoma cells stained for all of the endocytic (Rab5), lysosomal (Lamp-1, Lamp-2, CD63), and phagocytic markers (CD68 and Rab7) tested (Fig. 2). Moreover, the immunocytochemical analysis showed that the staining for the phagocytic markers, such as Lamp-1, Rab7, and CD68 (Fig. 2, A, B, and C, respectively), was mostly localized on perinuclear vesicles measuring 100 to 300 nm (Fig. 2A, inset) (the images were electronically acquired by a camera scanner for light microscopy application and electronically elaborated through a modular image processing and analysis system; see “Materials and Methods”). This was consistent with the distribution of the same proteins in professional macrophages (data not shown). The immunocytochemical data were consistent with the Western blot analysis inasmuch as both confirmed the expression of the phagocytic markers by melanoma cells and showed comparable molecular weights of these proteins between melanoma cells and professional phagocytes (Fig. 2D). Figure 3(A to D) shows additional findings suggesting a possible phagocytic sequence of the ingested autologous cells (early contact and internalization: Figure 3, A and B; and total degradation: Fig. 3, C and D), occurring in vacuoles stained for a phagocytic marker (ie, CD68). These results clearly show that the phagocytic markers were mostly localized on vacuoles containing cells in various stages of degradation, suggesting the occurrence of a genuine phagocytosis in human melanoma cell cultures. Moreover, these data definitively demonstrate that a genuine macrophage-like phagocytic behavior underlies the frequent in vivo findings of autophagocytic figures in histopathologic specimens of human melanoma (Banerjee and Harris, 2000).

Figure 2
figure 2

Expression and distribution of phagocytic markers in melanoma cells. Lysosomal-associated protein-1 (Lamp-1) (A), Rab7 (B), and CD68 (C) stainings in a representative melanoma cell line. The inset in A shows an electronic magnification (× 3500) of the area indicated by the arrowhead (bar = 0.5 μm). Immunophenotyping of fixed cells was performed using the peroxidase anti-peroxidase (PAP) method. AEC was used as chromogen and Mayer’s hematoxylin for the counterstaining. Final magnification, × 1500. (D) Western blot analysis of Rab5, Rab7, Lamp-1, and Lamp-2 proteins on total protein extract of four (1, 2, 3, 4) melanomas as compared with human primary macrophages (m). The relative molecular weights are indicated.

Full size image
Figure 3
figure 3

Phagocytic markers and autophagocytosis in melanoma cells. A to D, Possible sequence of an autophagocytic process occurring in homotypic culture of human melanoma cells: early internalization (A) and progressive degradation (B to D) of the internalized cell. The red staining is for the phagocytic marker CD68. Immunophenotyping of fixed cells was performed using the PAP method. AEC was used as chromogen and Mayer’s hematoxylin for the counterstaining. Arrowheads indicate the vacuoles containing the apoptotic cells in various states of degradation. Final magnification, × 2500.

Full size image

In Vitro Phagocytic Activity of Human Melanoma Cells

To assess the phagocytic activity of human melanoma cells, we measured the ability of 8 primary and 11 metastatic human melanoma cell lines to ingest either latex beads or apoptotic cells, as compared with human primary macrophages. In the first set of experiments, latex beads were added to human melanoma cell lines or human primary macrophages. After 2 hours the cultures were stopped and analyzed by immunocytochemistry for the expression and distribution of the lysosomal marker Lamp-1. Melanoma cell lines derived from either primary or metastatic lesions were analyzed to investigate the potential role of this phenomenon in tumor aggressiveness. The results showed that all of the melanoma cells derived from metastatic lesions absorbed the latex beads within 2 hours after their addition to the cell cultures (Fig. 4, B to D), at levels comparable to that of professional phagocytes (Fig. 4A). Notably, cell lines derived from primary melanomas did not absorb the particles (not shown). Of interest, as shown in Figure 4, the phagocytic markers were mostly distributed on the membrane of the vesicles containing latex beads. We quantified the level of phagocytosis by calculating the phagocytic index (P.I.; see “Materials and Methods”) in the melanoma cell lines as compared with human primary macrophages. The results demonstrated a comparable P.I. between human primary macrophages from different donors and melanoma cell lines derived from metastatic lesions, whereas melanoma cells from primary tumors did not show phagocytosis (Table 1).

Figure 4
figure 4

Latex beads and apoptotic cell endocytosis and their relationship to phagocytic markers in human melanoma cells. Internalization of latex beads in well-defined cytoplasmic vacuoles of human macrophages (A) and three human melanoma cell lines (B–D) is shown. Note the staining for Lamp-1 that nicely outlines the bead-containing vacuoles. Immunophenotyping of fixed cells was performed using the PAP method. AEC was used as chromogen and Mayer’s hematoxylin for the counterstaining. The internalization of apoptotic cells by human primary macrophages (E) and melanoma cells (G) is shown. F and H are electronically obtained magnifications of the insets of E and G, respectively. Mayer’s hematoxylin counterstaining. Final magnifications: A–D and F–H, × 2500; E–G, × 1500.

Full size image
Table 1 Phagocytic Activity of Melanoma Cell Lines and Macrophages
Full size table

In further evaluating the phagocytic activity of melanoma cells, we cultured either melanoma cells or macrophages with apoptotic cells of different origins (ie, primary melanoma, metastatic melanoma, human primary monocytes, macrophages, dendritic cells, lymphoblastoid, and promonocytic cells), using the same method as that for latex beads. The data were highly consistent with those obtained using latex beads, showing that the various melanoma cell lines derived from metastatic lesions, as well as the single donor-derived human primary macrophages, exerted variable levels of apoptotic cell phagocytosis, whereas cell lines from primary tumors did not show detectable phagocytic activity (Table 1). Moreover, because of the larger cellular size, melanoma cells in some cases were able to internalize up to 20 apoptotic particles (Fig. 4, G and H), whereas the number of apoptotic cells absorbed from human primary macrophages ranged from two to seven per cell (Fig. 4, E and F). The phagocytic activity of melanoma cells against apoptotic cells was also evaluated by electron microscopy. The results showed the presence of the apoptotic cells within well-defined cytoplasmic vacuoles (Fig. 5A), in which ingested cells underwent degradation (Fig. 5B). Thus, as suggested by immunocytochemical and electron microscopy appearance, a genuine phagocytic process occurred in human melanoma cells, leading to endocytosis and progressive degradation of the ingested apoptotic cells.

Figure 5
figure 5

Electron microscopy analysis of phagocytic-like activity in human melanoma cells. A, Transmission electron microscopy (TEM) micrograph showing the early engulfment of a U937 cell undergoing apoptosis in a big vacuole beneath the membrane of a melanoma cell. Note that the apoptotic cell does not show features of degradation at this stage (bar = 2 μm). B, TEM micrograph showing an apoptotic cell undergoing partial degradation inside an intracytoplasmic vacuole of a melanoma cell. The arrow points to the phagocytic vacuole, where the inner and the outer membrane sites are clearly detectable. Note the chromatinic structures of the apoptotic cell clearly visible within the vacuole (bar = 0.2 μm).

Full size image

To further quantify the phagocytic activity of melanoma cells, we modified a previously described method (Kaminski et al, 1985) that exploits Congo Red-stained yeast as targets of phagocytosis. In evaluating melanoma cell phagocytosis toward the stained yeast, we used both spectrophotometric and FACS analysis. The analysis by the spectrophotometer was markedly impaired by the baseline absorbance at 510 nm of melanoma cells, mostly because of the variable content of melanin (not shown). Thus, we used the FACS analyses to reliably measure the level of internalization of the stained yeast within the melanoma cells at 1.5, 2.5, and 3.5 hours after the addition of the yeasts. We found that, similar to human primary macrophages, each melanoma cell line derived from metastatic lesions was able to internalize yeasts and that this phagocytic activity increased in a time-dependent manner (Fig. 6B). Again melanoma cells derived from primary tumors did not show yeast internalization (Fig. 6B). Both human peripheral blood lymphocytes and a murine fibroblast cell line (3T3), used as negative controls, failed to ingest the Congo Red-stained yeasts (not shown). Our results (obtained by analyzing only cells included in “region 1”; Fig. 6A) demonstrated that most of the fluorescence intensity observed in melanoma cells was a result of internalized yeast particles. In fact, when a phagocytic test was performed at 4° C (a temperature that inhibits internalization of yeasts but not their adhesion on the cell surface), a small percentage of red-positive melanoma cells (from 0.5 ± 0.2 to 1.2 ± 0.9%) was revealed.

Figure 6
figure 6

Melanoma cell phagocytic activity against yeast cells as evaluated by FACS. Row A, Physic parameters and red fluorescence emission of Congo Red-stained yeast particles and unstained melanoma cells. Row B, Analysis of cell fluorescence emission, performed only in the cells included in region R1 to exclude yeast particles not associated with melanoma cells. Numbers reported in the figure represent the percentage of two melanoma cell lines derived from metastatic lesions (M mel), two melanoma cell lines derived from primary lesions (P mel), and human primary macrophages (macrophages) as positive controls, containing the Congo Red-stained yeast particles.

Full size image

We thus performed time-course experiments with the purpose of comparing the phagocytic activity of primary and metastatic melanoma cells at early (1–3 hours) and late (24–48 hours) time points. The results summarized in Figure 7 show that melanoma cells derived from metastatic tumors increased their phagocytic activity over time, reaching impressive values at 48 hours of culture, whereas melanoma cells from primary melanomas showed a slight but detectable phagocytic activity only at the later time points. Notably, data obtained with the FACS analysis of ingested Congo Red-stained yeast (Fig. 7A) were highly consistent with those obtained with latex beads (Fig. 7B).

Figure 7
figure 7

Time course of phagocytic activity of human melanoma cells. A, Kinetics of melanoma cell phagocytic activity against yeast cells as evaluated by FACS analysis. Histograms are the mean ± sem of the percentages of either primary or metastatic melanoma cells containing the Congo Red-stained yeast particles at the various time points. B, Kinetics of melanoma cell phagocytic activity against latex beads calculated as phagocytic index (P.I.). Histograms are the mean ± sem of the P.I. of either primary or metastatic melanoma cells containing the latex beads at the various time points.

Full size image

The Role of Cytoskeleton and Ezrin in the Phagocytic Activity of Melanoma Cells

On the basis of previous studies showing a crucial role of the actin cytoskeleton in the phagocytic function of professional phagocytes, and in particular of the proteins that link the membrane of phagosomes to actin (ie, ERM) (Defacque et al, 2000), we explored the possible involvement of ERM in the phagocytic activity of melanoma cells. We first investigated the involvement of actin and ERM in the phagocytic activity of melanoma cells. We treated melanoma cells with (i) a noncytotoxic concentration (0.2 μm per 1 hour) of CytB and (ii) antisense phosphorothioate oligonucleotides (PONs) complementary to the internal sequence of the coding regions of human ERM (Luciani et al, 2002; Parlato et al, 2000), assaying phagocytosis by FACS analysis. We preliminarily assessed the effect of the various PONs on ERM protein synthesis. Consistent with previous experiments performed on human cell lines of different origin (Luciani et al, 2002), the 60-hour treatment of melanoma cells induced a marked reduction of the ERM protein levels (Fig. 8A, left panel), without detectable cytotoxic effects (not shown), whereas the 72-hour treatment, although abolishing ERM protein synthesis (Fig. 8A, right panel), showed significant cytotoxic effects (not shown). On the basis of these results, we treated melanoma cells with the various PONs for 60 hours, to avoid the toxic effects shown with long-lasting treatment with the ERM antisense PONs. The results showed that treatment with CytB completely abolished the phagocytic activity of metastatic melanoma (Fig. 8B, upper panels). The results of the pretreatment with ERM antisense PONs showed that only the ezrin antisense PONs and a mix of all of the ERM antisense PONs (Fig. 8B, central panels) markedly inhibited the phagocytic activity of metastatic melanoma; the moesin antisense PONs only slightly inhibited this activity of melanoma cells (Fig. 8, lower panels). The ERM sense PONs, used as an internal control, did not affect the phagocytic function of melanoma cells (Fig. 8B, lower panels).

Figure 8
figure 8

The role of cytoskeleton in the phagocytic activity of melanoma cells. A, Western blotting on membrane/cytoskeleton fraction of sense (s) or antisense (as) phosphorothioate oligonucleotide (PONs)-treated or untreated (ctr) melanoma cells. Actin and gp-100 were used as control proteins. One of five representative experiments is shown. B, FACS analysis of the effect exerted by cytochalasin B and PONs to ERM (ezrin, radixin, and moesin) on the phagocytic activity of melanoma cells. In the upper panels, metastatic melanoma cells without yeasts, 3.30 hours after the addition of the yeasts, and the effect of cytochalasin B pretreatment on yeast internalization, respectively, are shown. In the middle panels, the effect of treatment with the antisense PONs to ezrin and a mix of the antisense PONs to all of the ERM are shown. In the lower panels, the effect of treatment with the antisense PONs to moesin and a mix of the sense PONs to all of the ERM are shown. One representative experiment is shown.

Full size image

To further investigate the involvement of ezrin in the phagocytic activity of melanoma cells, we evaluated the ezrin distribution in melanoma cells undergoing phagocytosis of plastic beads or apoptotic cells. The results showed that ezrin was clearly expressed on phagocytic vacuoles containing the beads (Fig. 9A). Moreover, we performed triple fluorescence experiments showing that ezrin was coexpressed with lysosomal antigens on phagocytic vacuoles of melanoma cells containing apoptotic cells (Fig. 9B). In the same experiments, we compared the ezrin distribution in melanoma cells undergoing phagocytosis to that occurring in human adenocarcinoma cells. The results showed a lysosomal antigen and ezrin distribution on the phagocytic vacuoles of adenocarcinoma cells similar to that of melanoma cells (Fig. 9C).

Figure 9
figure 9

ERM localization on phagocytic vacuoles of human melanoma cells. A, Progressive internalization of latex beads in human melanoma cells stained for ezrin. Note the early process of endocytosis, where ezrin concentrate into the internal site of the endocytic compartment (left panel, magnification × 3500), the internalization of the bead within a vacuole nicely surrounded by the ezrin staining (central panel, magnification × 2500), and the definitive internalization of three beads outlined by the ezrin staining (right panel, magnification × 2500). Immunophenotyping of fixed cells was performed using the PAP method. AEC was used as chromogen and Mayer’s hematoxylin for the counterstaining. B, Immunofluorescence analysis of the internalization of an apoptotic cell by a human melanoma cell. Triple fluorescence was performed on fixed cells with polyclonal anti-ezrin, with mAb anti-human LAMP-1, and with Hoechst 33258. Note the colocalization (merge) of ezrin and LAMP-1 on cytoplasmic vesicles of a melanoma cell, containing an apoptotic cell (Magnification, × 3500). C, Immunofluorescence analysis of the internalization of an apoptotic lymphocyte by a human breast adenocarcinoma cell line. Triple fluorescence was performed on fixed cells with polyclonal anti-ezrin, with mAb anti-human LAMP-1, and with Hoechst 33258. Note the colocalization (merge) of ezrin and LAMP-1 in cytoplasmic vesicles of a carcinoma cell, containing an apoptotic cell (Magnification, × 3500).

Full size image

Discussion

This study shows that human melanoma cells, similarly to professional phagocytes, clearly expressed both lysosomal-associated proteins (LAMP-1, LAMP-2, CD68, CD63) and markers of early endosomes (Rab5) and late phagosomes (Rab7). These markers clearly outline macrophage-like phagosomes, in which a progressive degradation of autologous nuclei does occur, strongly suggesting that a genuine phagocytosis underlies the in vivo phagocytosis we have shown in melanoma tumors. This supports a previous finding showing features of autophagocytosis in human melanomas (Banerjee and Harris, 2000). Our study definitively demonstrates that both the presence of a complete phagocytic machinery and the phagocytosis are consistent with the ability of melanoma cells to exert a macrophage-like activity. In fact, only melanoma cells derived from metastatic lesions were able to internalize latex beads and progressively digest yeasts and apoptotic cells.

We further showed that the actin cytoskeleton had a key role in the phagocytic activity of melanoma cells. In fact, ezrin was coexpressed with lysosomal markers on phagosomes of melanoma cells, and treatments with either cytochalasin or antisense oligonucleotides against ERMs, a family of proteins that link actin to the cell membranes (Bretscher et al, 1997; Gautreau et al, 2002), markedly inhibited the phagocytic activity of melanoma cells. In particular, we showed that ezrin was the most involved protein in this process. This last finding was highly consistent with a recent study showing the importance in macrophages of the linkage between actin filaments and the membrane of the secondary phagosomes through the ERMs, and particularly through ezrin, for the occurrence of efficient phagocytosis (Defacque et al, 2000). Moreover, recent findings suggest that ezrin is involved in the intracellular traffic of NK granules (Ramoni et al, 2002), which together with melanosomes belong to a lysosomal-like vesicle family (Dell'Angelica et al, 2000). The ability of cytochalasin B to entirely abrogate the phagocytic activity of melanoma cells further suggests a key role of the actin cytoskeleton in this phenomenon.

We have recently shown that specific activation of ρ-GTPase family proteins through bacterial toxins leads to a genuine phagocytic activity in a human epithelial cell line that normally does not endocyte or phagocyte either latex beads or apoptotic cells (Fiorentini et al, 2001). Dysregulation of the ezrin activation/phosphorylation may represent a major mechanism in promoting this potent phagocytic activity of metastatic melanoma cells, rather than the different level of expression of the ezrin protein between metastatic and primary tumors. On one hand, both recent reports (Ohtani et al, 1999, 2002; Makitie et al, 2001) and our preliminary data (not shown) did not show a clearcut difference in the levels of ezrin between human tumors with different degrees of malignancy. On the other hand, a correlation has been hypothesized between overexpression of proteins of the ρ-GTPase family, known as major ERM phosphorylation regulators (Yonemura et al, 2002), and tumor cell metastatic potential (Clark et al, 2000).

Our findings are highly consistent with some reports showing that phagocytic behavior is an indicator of high-grade malignancy and increased turnover in tumors of different histotypes (Banerjee and Harris, 2000; Brouwer et al, 1984; Coopman et al, 1998; Montcourrier et al, 1994). However, it is highly conceivable that the phagocytic potential of human melanocytes (Dell’Angelica et al, 2000; Koga et, 1992; Le Poole et al, 1993) underlies the potent phagocytic activity of human metastatic melanoma cells that we have shown in this study. In fact, in the present report we compare phagocytic activity of metastatic melanoma cells with the phagocytic activity exerted by professional phagocytic cells, ie, by human primary macrophages. Strikingly, results obtained did not show any significant difference between the two cell types in terms of phagocytic ability.

Phagocytic activity of human normal melanocytes has been shown using long-term assays and opsonized beads (Le Poole et al, 1993), whereas in this study plain plastic beads, apoptotic nuclei, and Congo Red-stained yeasts were used, to assess the phagocytic activity at early time points (1–3 hours). This approach has been suggested as the most suitable to compare the phagocytic activity of melanoma cells to that of professional phagocytes (Hu et al, 2000). In these experimental conditions only melanoma cells deriving from metastatic melanomas exerted a phagocytic activity comparable to that of human primary macrophages. In fact, the experiments performed at 24 and 48 hours showed that the metastatic cells increased their phagocytic activity over time, whereas the primary tumor-derived cells showed barely detectable phagocytosis, of both yeasts and latex beads, even at the later time points. However, it is highly conceivable that, given the ability of normal melanocytes to phagocyte, even though in strained experimental conditions (Le Poole et al, 1993), and the lysosomal nature of melanosomes (Dell’Angelica et al, 2000), malignant melanoma cells may hijack a physiologic activity exerted by melanocytes in normal tissues. This may include other advantages that tumor cells may acquire during malignant differentiation, such as the multivesicular body-derived exosomes (Andreola et al, 2002).

Altogether, our findings define an important mechanism that may be exploited by human melanoma, as well as by other human tumors, to invade tissues through an uncontrolled feeding of surrounding dead cells and extracellular materials. Moreover, the present study shows that the phagocytic behavior of human metastatic melanoma cells was always comparable to that exerted by human professional phagocytes, in phagocyting apoptotic cells from different human histotypes. It is therefore conceivable that melanoma cells, through phagocytosis of dead cells, may compete with the scavenging activity of both macrophages and professional APC, leading to the reduction of antigenic material to the professional APC. This mechanism may in turn interfere with the activation of the immune response against tumors, thus contributing to the development of tumor immune escape. This is further supported by our evidence that phagocytic activity was exerted by melanoma cells deriving from metastatic lesions, whereas cells obtained from primary melanomas did not show detectable phagocytic activity.

Because phagocytic activity was exclusively exerted by metastatic melanoma cells, we can speculate that phagocytic activity is a powerful property of highly aggressive tumors. The novelty of our data is the demonstration that this phagocytic activity does not merely account for the ability of malignant tumor cells to nibble the extracellular matrix or cell debris (Banerjee and Harris, 2000; Kuroda et al, 1999), but to actively engulf, ingest, and digest apoptotic cells. This mechanism may contribute to the typical ability of malignant tumor cells to survive in the lack of trophic supply, both in vivo and in vitro. Thus, the ex vivo assessment of phagocytic activity may represent a new tool in discriminating the level of aggressiveness of human malignant melanoma, as well as of malignant tumors of other histotypes. Further investigation needs to characterize the specific role of ezrin in the development of the potent phagocytic activity of metastatic melanoma cells, but our data provide clear evidence for a major role of the ezrin-mediated connection to the actin cytoskeleton in this mechanism. Data obtained with murine professional phagocytes (Defacque et al, 2000) suggest that tumor phagocytic activity may occur through the ezrin-mediated assembly of the actin filaments on the lysosomal membranes. This supports the use of new antitumor strategies aimed at inhibiting the tumor phagocytic activity, through the use of actin inhibitors or specific activators/inhibitors of the actin regulatory proteins.

Materials and Methods

Cells

Human melanoma cell lines used were as follows: (i) metastatic melanomas: Me30631, Me26258, Me14783, Me10249, Me17781, Me501, Me624.38, Me15392, Me8621 (obtained from melanomas of patients surgically resected at the Istituto dei Tumori, Milan), Me116, Me160 (kindly provided by Dr. Maio, Centro di Riferimento Oncologico, Aviano); (ii) primary melanomas: Me5810P, Me9923P, Me10538P, Me1007P, Me20842P, Me4405P (obtained from melanomas of patients surgically resected at the Istituto dei Tumori, Milan), WM904, and WM 743 (a gift from Dr. Meenhard Herlyn, Wistar Institute, Philadelphia, Pennsylvania).

Primary and metastatic melanoma cell lines were derived from histologically diagnosed primary (cutaneous) and secondary (subcutaneous, lymphonodular, or visceral) melanoma lesions, respectively, as previously described (Lupetti et al, 1996; Jonjic et al, 1992; Mortarini et al, 1991, 1992; Pietra et al, 2001; Rodolfo et al, 1988). All melanoma cell lines expressed melanoma antigens such as MART-1, gp100, and tyrosinase and members of the MAGE family. Additionally, all lines were found to bear mutations or deletions of the p16 gene (not shown). Peripheral blood mononuclear cells were isolated by Ficoll-Hypaque (Pharmacia, Uppsala, Sweden) density separation of buffy coats from healthy donors. To obtain monocytes, the peripheral blood mononuclear cell fraction was separated on 46% Percoll (Biochrom KG, Berlin) density gradients as described (Parlato et al, 2001).

To obtain macrophages, human monocytes were cultured under the same conditions as melanoma cells but with 15% fetal bovine serum and left to differentiate for 1 week at 37° C. To obtain dendritic cells, blood-derived monocytes were plated at a concentration of 1 to 2 × 106 cells/mL in RPMI 1640 (Gibco BRL, Gaithersburg, Maryland) supplemented with 10% lipopolysaccharide-free FCS. Cultures were added with granulocyte macrophage-CSF (500 U/mL) or IL-4 (25 ng/mL; specific activity, 107 IU/mg) (R&D Systems, Minneapolis, Minnesota), as described (Parlato et al, 2001). Human U937 cell line and Jurkat lymphoblastoid CD4+ T cells and the mouse NIH-3T3 fibroblast cell line were cultured in RPMI 1640 containing 10% fetal bovine serum, supplemented with 100 IU/ml penicillin, 100 μg/ml streptomycin (Gibco), and 2 mm glutamine (Gibco) in a 5% CO2 environment at 37° C.

Antibodies, Immunocytochemistry, Immunofluorescence, and Immunohistochemistry

Antibodies

mAbs used were as follows: anti-Lamp-1, anti-Lamp-2, anti-CD63 (PharMingen International), anti-CD68 (Dako, Denmark), anti-ezrin (Biogenesis, United Kingdom), anti-moesin (Transduction Laboratories). Polyclonal antibodies used were as follows: anti-rabbit Rab5 and anti-goat Rab7 antibodies (Santa Cruz Biotechnology, Inc., Santa Cruz, California); and anti-ezrin, anti-moesin, and anti-radixin (Santa Cruz).

Immunocytochemistry

Melanoma cells were attached on glass chamber slides (Labtek, Naperville, Illinois), fixed and stained by immunocytochemistry using either the alkaline phosphatase anti-alkaline phosphatase (Dako, Denmark) or the peroxidase anti-peroxidase (Dako, Denmark) method, in single staining, as appropriate (Parlato et al, 2001). Images were collected by a camera scanner for universal light microscopy application (AxioCam, Carl Zeiss Vision, Germany) and electronically elaborated through a modular image processing and analysis system (AxioVision, Carl Zeiss Vision) allowing electronic magnification of the acquired images and measurements of any recognizable cellular structures, as described (Andreola et al, 2002).

Immunofluorescence

Cells were fixed with 4% paraformaldehyde in PBS for 30 minutes at room temperature. After washing in the same buffer, cells were permeabilized with 0.5% Triton X-100 in PBS for 5 minutes at room temperature. For localization of ezrin and LAMP-1, samples were incubated at 37° for 30 minutes with the polyclonal antibodies to ezrin and the mAbs to LAMP-1. Cells were then incubated with anti-rabbit IgG FITC-conjugate (Sigma Chemical Company, St. Louis, Missouri) or anti-mouse IgG TRITC-conjugate (Sigma) for detection of ezrin and LAMP-1, respectively. After washing, all samples were incubated for 10 minutes with Hoechst 33258 (10 μg/ml) and mounted with glycerol-PBS (2:1) and observed with a Nikon Microphot fluorescence microscope. Samples were analyzed by intensified charge-coupled device video microscopy (IVM). Images were captured by a color chilled 3CCD camera (Zeiss). Normalization and background subtraction were performed for each image. Figures were obtained by merging ezrin-FITC (green), LAMP-1-TRITC (red), and Hoechst 33258 (blue) images by using IAS 2000 software for image analysis (Deltasistemi, Milan, Italy).

Histologic and Immunohistochemical Evaluation

For histologic analysis, tumor tissues were fixed in 7% neutral-buffered formalin and embedded in paraffin wax. Four-micrometer-thick sections of embedded tissues were cut and stained with hematoxylin and eosin. Tunel stain was performed on formalin-fixed paraffin-embedded tissue using the Apoptosis Detection Kit (R&D Systems).

Electronmicroscopy

For transmission electron microscopy, melanoma cells were fixed with 2.5% glutaraldehyde and post-fixed in 1% OsO4, dehydrated through graded ethanols, and embedded in epoxy resin (EPON 812). Thin sections were stained with uranyl acetate and lead citrate and observed with a Philips EM208S transmission electron microscope.

Phagocytosis Assays

Phagocytosis was assayed by adding 5 × 105 latex beads (Sigma) (Fiorentini et al, 2001) or UVB irradiation-induced apoptotic cells of the various origins (see below) (Kaminski et al, 1985) to Chamber slides (Falcon) containing either 5 × 104 human melanoma cells or 2 × 105 primary macrophages. The slides were incubated for 1.5, 2.5, 3.5, 24, and 48 hours at 37° C, washed five times with PBS, and stained with Mayer's hematoxylin (BDH, England). Phagocytosis was evaluated by counting 200 melanoma cells or macrophages per well at × 40 magnification. Results were expressed as phagocytic index (P.I.), which was generated by multiplying the percentage of cells containing at least one bead or apoptotic cell by the mean number of absorbed materials per melanoma cells (Hu et al, 2000).

Yeast cells were stained by adding 3 ml of Congo Red solution, 0.87% w/v in PBS, to 1.5 g of yeast (Saccharomyces cerevisiae) (Kaminski et al, 1985). The slurry was mixed and left for 15 minutes at room temperature. Then distilled water was added to the slurry of stained yeast cells. The solution was autoclaved for 15 minutes to kill yeast cells and sterilize the solution. These cells were incubated 60:1 with either melanoma cells or primary macrophages at 37° C and 5% CO2. Phagocytosis was stopped at 1.5, 2.5, 3.5, 24, and 48 hours by washing away the excess yeast cells with PBS and adding a PBS solution containing trypsin (1.5 g/L)-EDTA (0.4 g/L). The analysis of phagocytosis was performed by spectrophotometry (Amersham, Pharmacia), evaluating the absorbance of stained yeast cells within melanoma cells or macrophages at 510 nm. A standard curve has been prepared by using a series of standard solutions containing known concentrations of yeast cells and melanoma cells or a macrophage control. Alternatively, phagocytosis of stained yeast cells was analyzed on a FACScan flow cytometer (Becton Dickinson) equipped with a 488-nm argon laser. At least 20,000 events have been acquired. Data were statistically analyzed (Student's t test) on a Hewlett Packard computer using the Lysys II Software.

Antisense Oligonucleotides

Melanoma cells were treated with ERM antisense PONs and submitted to FACS phagocytosis assay. The sequences 5′AGACGGGTCCTCCAGTCCTTC3′ and 5′TACGGGTTTTGCTAG3′ were chosen for ezrin/radixin and moesin, respectively (Parlato et al, 2000). The PONs were suspended in serum-free medium and added to the culture medium every 12 hours at a concentration of 40 mm for 60 or 72 hours. Sense PONs of the same region were used as controls.

Cell Lysis and Western Blot Analysis

Subconfluent melanoma or macrophage cultures were pelleted and washed with PBS and then lysed in an extraction buffer (AKT buffer) containing 20 mm Tris-HCl (pH 7.5), 150 mm NaCl, 10% glycerol, 1% NP-40, and both phosphatase and protease inhibitors (phenylmethylsulfonyl fluoride 2 mm, Na3VO4 10 mm, and 10 mg/ml of aprotinin) for 20 minutes at 4° C. Proteins were resuspended in SDS sample buffer, denatured by boiling, and separated on 8% SDS-PAGE gel and electroblotted on nitrocellulose (Protran BA85; Schleicher & Schuell). Lamp-1, Lamp-2, and CD63 mAbs and Rab5 and Rab7 polyclonal Abs were used at 2 μg/ml. Immunodetection of proteins was visualized after incubation with peroxidase anti-Ig followed by the enhanced chemiluminescence method (ECL; SuperSignal West Dura Extended Substrate, Pierce, Illinois). ERM levels were detected by Western blot as described (Luciani et al, 2002).