Differentiation of Human Malignant Melanoma Cells that Escape Apoptosis After Treatment with 9-Nitrocamptothecin In Vitro

Abstract

After in-vitro exposure to 0.05 µmol/L 9-nitrocamptothecin (9NC) for periods of time longer than 5 days, 65% to 80% of the human malignant melanoma SB1B cells die by apoptosis, whereas the remaining cells are arrested at the G₂-phase of the cell cycle. Upon discontinuation of exposure to 9NC the G₂-arrested cells resume cell cycling or remain arrested depending on the duration of 9NC exposure. In contrast to cycling malignant cells, the cells irreversibly arrested at G₂ exhibit features of normal-like cells, the melanocytes, as assessed by the appearance of dendrite-like structures; loss of proliferative activity; synthesis of the characteristic pigment, melanin; and, particularly, loss of tumorigenic ability after xenografting in immunodeficient mice. Further, the expression of the cyclin-dependent kinase inhibitor p16 is upregulated in the 9NC-treated, G₂-arrested, but downregulated in density G₁-arrested cells, whereas the reverse is observed in the expression of another cyclin-dependent kinase inhibitor, p21. These results suggest that malignant melanoma SB1B cells that escape 9NC-induced death by apoptosis undergo differentiation toward nonmalignant, normal-like cells.

Introduction

Currently available chemotherapies for the treatment of malignant melanoma rarely induce complete remissions, and there has been very slow progress in the development of drugs with effective antimelanotic activity (reviewed in Refs. (1–3)). An alternative approach to drug-induced death of malignant melanoma cells is differentiation chemotherapy of these cells, i.e., drug-induced differentiation to melanocytes. Several studies have demonstrated the possibility of differentiation-chemotherapy by exposing melanoma cells to various agents including phorbol esters (reviewed in Ref. (4)); the pyridine analogue bromodeoxyuridine (5); the fatty acids phenylacetate and phenylbutyrate (6); the soybean isoflavone; genistein (7–9); retinoic acid analogues (10); benzodiazepine receptor ligands (11); and the combination of interferon-β and the compound mezerein (12). The differentiated melanoma cells form dendrite-like structures, have increased levels of the pigment melanin and tyrosinase activity, and either grow slower than undifferentiated melanoma cells or are arrested at the G₂ phase (4,6,7,13).

Cessation of cell proliferation, tumor suppression, and terminal differentiation have been correlated with changes in the expression of proteins that determine cell cycle progression, including the p16 and p21 inhibitors of cyclin-dependent kinases. For example, the upregulation of p21 (WAF1/CIP1/SDI1) in vitro has been implicated in various cellular mechanisms and events including the after: arrest of the cell cycle at the G₁ and G₁/S phases in cells treated with agents that damage DNA (14–16); accumulation of cells, in addition to G₁, near the G₂/M-phase boundary (17–19); arrest of hepatoblastoma HepG₂ cells at G₂ after treatment with the drug 9-nitrocamptothecin (9NC; 20); induction of differentiation of cells of diverse tissue origin, including myocytes (21–23), keratinocytes (24), colon carcinoma cells (25), neuroblastoma cells (26), hepatoma cells (27), melanoma cells (12), erythroleukemia cells (28), monocytes (29),(30), granulocytes (31),(32), and megakaryocytes (33); polyploidization of megakaryocytes (33); and suppression of the ability of melanoma cells to induce tumors after xenografting in immunodeficient nude mice (34). Also, upregulation of p21 has been observed in human melanoma cells that grow to high saturation densities near growth arrest (12), and genistein-induced differentiation of human melanoma cells arrested at G₂ (7). Further, there are reports on downregulation in the expression or cessation of p21 synthesis in terminally differentiated primary mouse keratinocytes (35) and differentiated human hepatoma cells that are arrested at G₁ (36), whereas other studies have correlated upregulation of p21 with arrest of cells either at G₁ (28,29) or concurrently at G₁ and G₂ (19), regardless of subsequent cell differentiation.

The cell-cycle regulator p16 (p16^INK4) is a major inhibitor (i.e., negative regulator) of the cyclin-dependent kinase CDK4 (37,38). Binding of p16 to and thus inactivating the cyclin D-CDK4 (or CDK6) complex ultimately results in cell-cycle arrest at the G₂ restriction point (reviewed in Ref. (39)). However, p16-dependent inhibition of cyclin D/CDK activity may not be sufficient to cause G₁ arrest in actively proliferating tumor cells, and p16-dependent inhibition of cyclin E-dependent kinases is required for p16-dependent growth suppression (40). Functional inactivation of p16 may account for a p16-dependent G₂ cell-cycle checkpoint in the development of melanoma (41). Minor perturbations in the p16 primary structure can lead to loss of its inhibitory activity, thus contributing to malignancy in numerous cell lines (42). In general, studies in human tumors, cell lines, melanoma-prone families and knockout mice have established p16 as a tumor suppressor (43) and reviewed in Ref. (44). In this regard, p16 has been extensively implicated in tumor cell proliferation and progression or suppression of malignant melanoma (45–47). Also, p16 has been implicated in cell differentiation. Thus, p16 upregulation may be part of a differentiation program that is turned on in senescent cells and is essential for maintenance of senescent cell-cycle arrest at G₁ (48). Restoration of p16 into melanoma cells in vitro has resulted in cell-cycle arrest and appearance of morphological features of mature melanocytes (49). Finally, studies with mouse melanocytes that used a combination of gene disruption of p16 or p21 and ectopic expression of the nuclear factor E2F1 have suggested that mechanisms other than those involving p16 and p21 may play an important role in development of malignant melanoma (50).

The anticancer drug 9NC, a semisynthetic analogue of the natural product camptothecin (CPT), has demonstrated multiple capabilities against cells and tumors derived from solid tissues, leukemias, and HIV-infected cells. Thus, 9NC is a potent inducer of differentiation of myeloid cells in vitro (51), inhibits replication of HIV in latently infected lymphoid (52) and freshly infected monocytoid (53) cells, demonstrates exceptional ability to inhibit growth of human cancer cells in culture, and induces regression of various human tumors established as xenografts in immunodeficient nude mice (reviewed in Refs. (54,55)). We have recently investigated the therapeutic efficacy of 9NC and other water-insoluble CPT analogues against human malignant melanoma xenografts established in nude mice, and the results showed that the antitumor effectiveness and toxicity depend on the CPT analogue, dose administered, mode of administration, and scheduling of drug administration (56). Treatment with 9NC ultimately resulted in complete regression of human melanoma in absence of induction of apparent toxicities in mice with or without tumors (55,57). These observations are in agreement with findings in vitro that 9NC induces programmed cell death (apoptosis) in human malignant melanoma cells, but not in their normal counterparts, melanocytes (57,58). Of interest, low 9NC concentrations of 0.015 to 0.025 µmol/L are more effective than higher drug concentrations of 0.05 to 0.08 µmol/L in the ability to induce apoptosis in melanoma cells in vitro (58). Further, low 9NC concentrations induce cell cycle arrest at G₂ and concomitant upregulation of p21 mRNA in human hepatocytes in vitro; moreover, the regulation of this mRNA is dependent on the extent of accumulation of cells in the G₂ phase and depletion of the G₁ cells (20).

In this report, we demonstrate that human malignant melanoma cells continuously exposed to 0.05 µmol/L 9NC can differentiate to melanocyte-like cells and, upon discontinuation of exposure to 9NC and depending on the duration of exposure, the differentiated cells may reverse to the parental malignant phenotype or remain terminally differentiated.

Materials and Methods

Cells

The human malignant melanoma SB1B cell line was established by Verschraegen et al (59). Briefly, several cell clones were derived from a primary human cutaneous melanoma and demonstrated distinct differences in tumorigenic and metastatic ability after xenografting in immunodeficient nude mice. A clone, designated SB1B, showed a characteristic ability to metastasize rapidly to the brain, whereas the other clones were less tumorigenic and showed little ability to metastasize to a specific organ (59). In the studies of this report, SB1B cells were propagated in RPMI 1640 media supplemented with 10% bovine serum in a 5% carbon dioxide atmosphere at 37°C. Cell cultures that reached 70% to 80% confluence were split to subcultures or were treated with 9NC. The cells used for the studies of this report were at passages 20 to 29.

Microscopy and Flow Cytometry

Cells attached to the plastic culture container were stained with Wright-Giemsa dye and then examined under a Zeiss (Göttingen, Germany) or Nikon (Garden City, NY) microscope and photomicrographed on Kodak 200 film. The percentage (i.e., fraction) of attached cells in the various cell cycle phases and in apoptosis were determined by analysis of the relative DNA content in the cells with flow cytometry (Epics-Elite Laser Flow Cytometer, Coulter, Hialeah, FL). For this, trypsin-detached cells were rinsed in phosphate-buffered solution (PBS), and then suspended in the lysis-staining solution (Coulter) which contained detergent and the DNA-binding dye propidium iodide. The cells remained in this solution for 10 minutes at room temperature before they were subjected to DNA determination. Approximately 15,000 cells were used per determination. A suspension of chicken erythrocyte nuclei (provided by Coulter) served as a reference. Cell fractions were calculated with the aid of the Multicycle program (Phoenix Flow Systems, San Diego, CA). Flow cytometry analysis of relative DNA content indicates the cell fractions in G₀+G₁ (i.e., practically all cells in this fraction are in G₁), G₂+M (i.e., practically all cells in this fraction are in G₂), and S phases. It also indicates the apoptotic fraction that includes intact cells with DNA content less than G₁-cells and the hyperdiploid fraction that includes cells with DNA content larger than cells in G₂.

Western Blot Analysis

The methodology for Western blot analysis has been described previously (60,61). Briefly, cells were washed in PBS, then suspended in 20 mmol/L Tris-HCl (pH 7.5), 5 mmol/L MgCl₂, and immediately heated at 100°C for 10 minutes. DNAse I was added to the lysate (10 µg/10⁷ cells), which was then incubated at 37°C for 10 minutes. Subsequently, the incubate was mixed with an equal volume of a solution containing 7 mol/L urea-30% glycerol-3% sodium dodecyl sulfate (SDS), and heated at 100°C for 10 minutes. Protein concentration of the lysate was determined by using detergent-compatible protein assay reagents (Bio-Rad Hercules, CA), and sample volumes containing 70 µg of total protein were subjected to electrophoresis in an SDS-12% polyacrylamide gel. The antibodies to p16 and p21 were from Santa Cruz Biotechnology (Santa Cruz, CA) and Oncogene Science (Manhasset, NY), respectively. Western blots were analyzed by using the Amersham (Piscataway, NJ) chemiluminescence kit.

Tumorigenic Ability of Cells

The tumorigenic ability of cells was investigated in immunodeficient nude mice as described (57,58,62). Briefly, 2x10⁷; cells were subcutaneously injected in NIH-1 nude mice, and then the mice were examined daily for appearance of palpable tumors. Tumor-bearing animals were sacrificed when the tumor size reached a volume of approximately 3 cm³. Mice that were injected with cells but did not develop tumors continued to be monitored up to 6 months. Four to 6 mice were used per each cell sample. Animal care and handling were in accordance with guidelines of the Institutional Animal Care and Use Committee.

Results

Treatment of Cells with 9NC

The 9NC concentration of 0.05 µmol/L was chosen for the treatment of the cells in these studies after several pilot experiments were conducted in which, SB1B cells were exposed for various periods of time to 9NC concentrations ranging from 0.02 to 0.1 µmol/L. Concentrations of 0.02 to 0.03 µmol/L resulted in cytostasis in 72 hours and cell death in a small cell fraction, whereas longer periods of treatment resulted in detachment and death of about 100% of the cells in culture as assessed by a count of trypsin-detached cells, microscopy, and flow cytometry. On the other hand, 9NC-concentrations of 0.08 to 0.1 µmol/L were highly toxic for the SB1B cells, which eventually died in 24 to 48 hours. However, prolonged treatment of the cultures for 5 to 7 days with 0.05 µmol/L 9NC resulted in detachment and death of 65% to 80% of the cells in the culture. No further decrease in the number of attached cells was observed after 7 days of 9NC treatment. The cells that remained attached, i.e., surviving cells, did not proliferate and started forming dendrite-like structures. Therefore, for the studies described in this report, the cells were routinely (four experiments) exposed to 0.05 µmol/L 9NC.

Morphology of Cells

SB1B cells exposed to 0.05 µmol/L 9NC were observed by microscopy for morphological changes as follows. Several cultures were seeded with equal number of cells and incubated at 37°C until they received 9NC for various periods of time up to 40 days. The cells of one culture were counted on the day the 9NC treatment was initiated (0-day), and this cell number was taken to be 100%. Subsequently, every 2 to 3 days, one culture was used to collect trypsin-removed cells, which were counted and subjected to flow cytometry studies. The cell number in each culture was converted to percent of the cell number scored on 0-day. Concurrently, a second culture was stained with Wright-Giemsa or methylene blue dye for photomicrography with the same magnification. Photomicrographs of untreated and 9NC-treated SB1B cells are shown in Figure 1. Exponentially growing untreated cells have epithelioid appearance, most of them are dipolar or tripolar, and each cell contains one nucleus with distinct nucleoli that vary in number and size (Figure 1A). Cells exposed to 0.05 µmol/L 9NC for 24 hours were not dividing, were mostly spindle-shaped, and occasionally contained two nuclei. There was an apparent increase in the size of the cells, nuclei, and nucleoli. In general, the nucleoli were more heterochromatic in the drug-treated than the untreated cells (see Figure 1A and B). Longer exposure of the cells to 9NC for 7 days resulted in even larger cells, nuclei, and nucleoli (Figure 1C). Some of the cells were binucleated, but these cells were present at a lower frequency than the cells in Figure 1B. The nucleoli had become more prominent. Many cells looked flatter, had lost the spindle shape and were rather polygonal with short dendritic-like structures. These cells were not dividing. Cells treated with 9NC for 14 days (Figure 1D) were smaller than the cells in Figure 1B and C. The ratio of nucleus to cytoplasm was higher in the cells shown in Figure 1D than in Figure 1A, B, and C. All cells shown in Figure 1D are polygonal and have numerous short and long dendritic structures (indicated by arrows). No further significant morphological changes were apparent in cells exposed to 0.05 µmol/L 9NC for periods of time longer than 14 days and up to 40 days.

Figure 1.

Photomicrographs of SB1B melanoma cells exposed to 9NC. SB1B malignant melanoma cells were exposed to 0.05 µmol/L 9NC for various periods of time before they were stained and photomicrographed. The cells received no treatment (A) or were treated with 9NC for 24 hr (B), 7 days (C), and 14 days (D). (Arrows indicate dendritic extensions of the cells. Bar equals 100 µm in D).

After exposure of the cells to 9NC for 10 to 12 days, the media were replaced in some cultures with fresh media without 9NC, and incubation continued at 37°C. The cells of these cultures, termed released, were also observed under the microscope and photomicrographed. In general, no visible changes were observed in the 9NC-released cells after discontinuation of 9NC exposure up to 48 hours. Three days after 9NC release, the cultures contained mixed cell populations consisting of cells morphologically similar to those shown in Figure 1A through D, as well as dying cells (discussed later). Ten days after 9NC release, no cells with dendritic morphology were observed, but only cells that had the morphology of untreated SB1B cells. It was evident that the released cells were actively dividing, and the cultures eventually became confluent 17 to 20 days after 9NC release.

Flow Cytometry of 9NC-Exposed Cells

Relative changes in the cell cycle of SB1B cells exposed to 9NC were monitored by flow cytometry analysis. Only trypsin-detached cells were collected and subjected to analysis. Results (histograms) of this analysis are shown in Figure 2. Exponentially grown untreated SB1B cells generated the representative histogram A, which shows that 50%, 39%, and 11% of the attached cells were in the G₁, S and G₂ phases, respectively, whereas no apoptotic cells were present.

Figure 2.

Flow cytometry analysis of SB1B cells treated with 9NC. Untreated (A) and 9NC-treated (B–J) cells were subjected to flow cytometry analysis of the cell cycle. The treatment period was 6 hr (B), 12 hr (C), 24 hr (D), 3 days (E), 7 days (F), 14 days (G), 21 days (H), 30 days (I), and 37 days (J). Only trypsin-detached cells were subjected to analysis. (G₀ = G₀+G₁ cells; G₂ = G₂+M cells; Ap, apoptotic cells. Arrowheads indicate histogram fractions representing hyperdiploid cells).

Treatment of the cells with 0.05 µmol/L 9NC for 6 hours induced accumulation of cells in early S phase with a concomitant decrease in the cell fraction in the G₁ phase (histogram B). At 12 hours of treatment, more cells had accumulated in early S phase, whereas a dramatic decrease in the G₁ fraction was apparent (histogram C). At 24 hours of treatment all cells were in late S phase (histogram D). At 3 days of treatment about 90% of the cells were arrested in the G₂ phase, and a small number of cells appeared in the apoptotic fraction (histogram E). At 5 days, the majority of cells remained arrested in G₂, but there was an increase in the apoptotic fraction (histogram not shown). A further increase in the apoptotic fraction was detected at 7 days of treatment (histogram F), whereas at 10 days, detached (dead) cells were detected in the culture media. There were also cell fragments, so that any estimate by flow cytometry would be an underestimate of the number of dead cells. Conversely, the number of attached cells was continuously decreasing as the culture remained exposed to 9NC. However, there was an increased presence of hyperdiploid cells as 9NC exposure continued (histograms G, H, I, J). It is of interest that many G₂-arrested SB1B cells remained alive and attached to the substrate for a very prolonged period of time up to 40 days.

Flow Cytometry of 9NC-Released Cells

After treatment for 12 days, the media of several cultures of SB1B cells were replaced with fresh media that did not contain 9NC. At desired times after 9NC release, cells were collected and subjected to cell cycle analysis by flow cytometry. Time-dependent changes in the cell cycle of the 9NC-released cells are shown in the histograms of Figure 3. Six hours after release, over 90% of the cells were still accumulated in G₂, with a small cell fraction of hyperdiploid cells, and even a smaller apoptotic fraction present (histogram A). Twelve hours after release, there was an increase in the hyperdiploid and apoptotic fractions (histogram B), and at 24 hours, some proliferation activity was detected in the culture (see presence of small G₁ and S fractions in histogram C). In the subsequent days, that is 2 to 5 days after 9NC release, there was a progressive increase in proliferation activity as well as in apoptosis. Histogram D is indicative of the cell cycle changes taking place in the culture. For example, in this histogram the G₁, S, G₂ and apoptotic fractions were estimated to be 12%, 27%, 37%, and 24%, respectively. Ten days after 9NC release, there were a significant number of detached (dead) cells in the culture, but the attached cells appeared to have resumed cycling (compare histograms in Figures 2A and 3E). Thus, in histogram E, 46%, 28%, 18%, and 8% of the attached cells were in the G₁, S, G₂ and apoptotic fractions, respectively. These percentages are similar to those calculated for the cell cycle fractions of the untreated cells (Figure 2, histogram A). Eventually, the attached cells stopped dividing and they were virtually all (92%) arrested in G₁ as the culture reached confluence at 21 days post-release from 9NC exposure (histogram F) as assessed by microscopy observations and cell counting.

Figure 3.

Cell cycle analysis of 9NC-released cells. Cells were treated with 0.05 µmol/L 9NC for 10 days, then transferred to fresh media in absence of 9NC. After release from 9NC treatment, cells were collected at desired times and subjected to cell cycle analysis by flow cytometry. The 9NC-released cells were collected at 6 hr (A), 12 hr (B), 24 hr (C), 3 days (D), 10 days (E), and 21 days (F). For designations of histogram cell fractions, see Figure 2.

Production of Melanin

During collection of 9NC-treated cells for flow cytometry studies, we observed that the pellets of cells exposed to 9NC for a long time were much darker than cells exposed to 9NC for a short time. Pellets of cells treated with 9NC for various periods of time were collected and stored in glass tubes containing 3% formalin, then placed side by side and photographed (Figure 4). No significant color differences were observed between untreated SB1B cells (tube A) and cells treated with 9NC for 1, 3, and 7 days (tubes B, C and D, respectively). The cells appeared darker at 14 days of drug exposure (tube E), and were even darker after 21, 30, and 37 days (tubes F, G, H). It can also be seen that the dark pigment leached out of the cells into the formalin solution in which the cells were fixed and stored. Apparently, this dark pigmentation is due to melanin synthesized by the cells after extensive exposure to 0.05 µmol/L 9NC. Subsequent to discontinuation of 9NC exposure, synthesis of melanin progressively decreased by the cells (tubes I, J, K). Finally, after prolonged 9NC release, the cells synthesized little or no melanin (tube L). It should be noted that untreated SB1B cells synthesize no or very little melanin and that synthesis was initiated or became evident after the cells had stopped dividing and developed the multidendritic morphology shown in Figure 1. In other words, 9NC induced differentiation of the malignant melanoma cells to melanocyte-like cells that apparently produce melanin.

Figure 4.

Photography of cell pellets. Trypsin-detached cells were pelleted by gentle centrifugation, stored in formalin, and photographed. The pellets shown were from cultured cells that received no treatment (A), those cells continuously exposed to 9NC for 24 hr (B), 3 days (C), 7 days (D), 14 days (E), 21 days (F), 30 days (G), and 37 days (H); and those cells that were exposed to 9NC for 21 days, then released from 9NC treatment for 24 hr (I), 4 days (J), 14 days (K), and 21 days (L).

Detection of p16 and p21

SB1B cells treated with 9NC and 9NC-released cells were analyzed for expression of the cyclin kinase inhibitors p16 and p21 that have been associated with regulation of progression of cells through the G₁ and G₂ phases of the cell cycle. Expression of p16 is upregulated in the 9NC-treated cells, that is, as the cells accumulate in the G₂ phase (Figure 5). Maximal p16 expression was detected at 21 days of 9NC treatment, and similar levels of p16 were expressed in the cells after treatment with 9NC for periods longer than 21 days and up to 31 days (results not shown). Further, p16 expression decreased after release of the cells from 9NC exposure. In fact, there was a very good correlation between decreased p16 expression, decreased G₂ cell fraction and concomitantly increased G₁ cell fraction. No p16 expression was detected at 21 days of 9NC release, that is, when 92% of the cells were accumulated in the G₁ phase. Figure 5 also shows that the quantitative changes in p21 correlated with arrest of the cells in G₁ or G₂ after 9NC treatment. In contrast to p16, p21 was highly expressed in the untreated cells, whose majority was in the G₁ fraction (see Figure 2, histogram A). Expression of p21 was downregulated within 24 hours after the cells were exposed to 9NC, and the extent of downregulation was dependent on the duration of treatment. The lowest p21 expression was detected at 21 days of 9NC treatment, when the majority of cells was arrested in G₂, and a small fraction became hyperdiploid (see Figure 2, histogram G). Discontinuation of exposure to 9NC was followed by upregulation of p21 expression in direct correlation with the percentage of cells in G₁. Similar patterns of p16 and p21 expression were observed in two other independent experiments by use of cell samples collected at similar schedules of 9NC treatment and “release” from 9NC-treatment.

Figure 5.

Detection of p16 and p21 by Western blot analysis. Whole cell extracts were prepared, and samples containing equal protein concentrations were subjected to Western blot analysis. Presence of p16 and p21 was visualized by chemiluminescence. The extracts were prepared from SB1B cells that received no treatment, or were continuously exposed to 0.05 µmol/L 9NC for 1, 3, 14, and 21 days (lanes under the indication: days of 9NC treatment) or after 12 days of 9NC treatment were transferred to and remained in 9NC-free medium for 1, 4, 14, and 21 days (lanes under the indication: days after 9NC release).

In conclusion, the results presented here demonstrate that p16 downregulation and p21 upregulation correlate with cell arrest in G₂ and G₁, respectively, and therefore, there is an inverse relationship in the regulatory mechanisms of p16 and p21 expression in 9NC-treated SB1B cells.

Tumorigenicity of SB1B Cells

SB1B cells exposed to 0.05 µmol/L 9NC for various periods of time were detached by trypsinization and injected into nude mice as described (52). Four to six mice were injected with equal number of each type of cells. Tumors appeared in all mice (i.e., 100% “take”) that were inoculated with dividing untreated (control) SB1B cells and nondividing SB1B cells exposed to 0.05 µmol/L 9NC for 21 or less days, but no tumors appeared in mice inoculated with cells exposed to 9NC for 30 days. Appearance of tumors on mice at the sites of injection indicated that cells treated with 9NC up for 21 days or less, like untreated SB1B cells, were tumorigenic in nude mice. Although the lagging period varied, i.e., 6 to 10 days, for the appearance of tumors derived from SB1B and SB1B/9NC cells, in general, the 9NC-treated cells consistently required longer period of time than untreated cells to induce visible tumors. Tumor establishment and absence of spontaneous regression were assessed by allowing the tumors to reach a size of 2.5 to 3 cm³ before the animals were sacrificed. The average measurements of tumor growth and body weight as well as observations on tumor sections were very similar to those reported for other human malignant melanoma cells xenografted in nude mice (56,57). No other tumor growth characteristics were studied. In contrast, cells treated with 9NC for 30 days did not generate tumors within 60 days after injection into the animals, indicating that these cells had lost their tumorigenic ability.

Discussion

In this report we have shown that cytotoxic concentrations of the anticancer drug 9NC can induce differentiation of human malignant melanoma cells to melanocyte-like cells. Of the total melanoma cell population in the culture exposed to 9NC, 65% to 80% of the cells died by apoptosis, but the remaining cell population remained strongly attached to the substrate and differentiated. Apparently, the SB1B cell line consists of a major subpopulation of cells susceptible to 9NC-induced apoptosis, and a minor subpopulation of cells resisting apoptosis and susceptible to 9NC-induced differentiation. In this regard, we have shown that 9NC can induce irreversible differentiation of human leukemia U-937 cells to more mature cells along the monocytic lineage; however, these cells have remained differentiated even after discontinuation of exposure to 9NC (51). In contrast, the SB1B cells that were converted to melanocyte-like cells, remained differentiated as long as 9NC was present in the culture. Unlike the differentiated U-937 cells that continued to divide, the differentiated melanoma cells were not dividing and were arrested at the G₂ phase of the cell cycle. Further, the nondividing G₂-arrested cells had developed dendritic morphology and apparently synthesized melanin; that is, the 9NC-generated cells exhibit properties of physiological human melanocytes (4). Normal melanocytes have a finite life span, do not grow in soft agar, and are nontumorigenic (63).

Removal of 9NC from the media of the differentiated, G₂-arrested, melanin-producing cells was followed by cell responses that were dependent on the duration of the 9NC treatment. Thus, the melanocyte-like cells de-differentiated, i.e., reverted to malignant melanoma cells if the exposure to 9NC was for a period up to 21 days. Remarkably, these G₂-arrested cells induced tumors after they were inoculated in nude mice indicating that exposure to 9NC for 21 days resulted in changes in the morphology (i.e., appearance of dendrites), physiology (i.e., melanin synthesis), and state of growth (i.e., arrest at the G₂ phase of the cell cycle) but failed to abolish the tumorigenic ability of the cells. In other words, these 9NC-treated cells were differentiated by several criteria, but their malignant state was only masked and became apparent after 9NC was removed. Our results also indicate the presence of cells with DNA content (ploidy) higher than 2N in cultures treated with 9NC for a period of 14 days or longer (see Figure 2). Apparently, these hyperdiploid cells resulted from DNA endoreduplication and have been previously observed in 9NC-treated cultures of melanoma and ovarian, breast, and prostate carcinoma cells (20,57,58, 62,64,65).

Of interest also is the observation that about 100% of the SB1B cells were killed in presence of 9NC concentrations of 0.02 to 0.03 µmol/L applied for periods longer than 5 days, whereas 0.05 µmol/L 9NC allowed a portion of the cells to survive. This observation is in agreement with a previous observation that low 9NC concentrations are more effective than high concentrations in inducing apoptosis (58). At that time, we hypothesized that this paradoxic finding was probably due to specific versus general poisoning by low versus high 9NC concentrations, respectively. In other words, low 9NC concentrations allow the cells to advance to late stages of the S phase, where formation of DNA-topoisomerase I complexes is extensive and therefore available to specifically interact with the drug molecules. On the other hand, high 9NC concentrations result in formation of DNA-topoisomerase I-9NC complexes in cells present in the S phase but also interfere with components and/or factors of cells in other phases, thus not allowing these cells to enter the S phase and consequently resulting in reduced number of apoptotic cells. However, this hypothesis remains to be proven. Further, we suggested that these observations could be of clinical importance. Indeed, a recent report on clinical trials of various CPT-derivatives indicates that low doses of these drugs administered continuously are more effective than large doses administered on intermittent schedules (66).

Further, exposure of the SB1B cells to 9NC for 30 days and subsequent 9NC removal was not followed by de-differentiation; that is, unlike the 21 day-treated, the 30 day-treated cells retained the dendritic morphology and remained melanin-pigmented at the G₂ for a long time in absence of 9NC until they died in the culture. Moreover, these cells were unable to induce tumors in nude mice. In conclusion, treatment of the SB1B cells with 9NC for 21 days and 30 days resulted in reversible and irreversible differentiation, respectively. Further, reversible or irreversible differentiation appeared to correlate with the ability or loss of ability of the cells to induce tumors in vivo. It should be emphasized that the correlation of reversibility/irreversibility of the differentiated cells to the duration of 9NC exposure was reproducible in three independent experiments indicating that the transition from the reversible to irreversible state occurred at a time between 20 and 30 days of 9NC exposure. The reversibly differentiated cells required a longer time than the undifferentiated SB1B cells to induce tumors in mice. This delayed induction of tumors was readily apparent upon inoculation of equal number of cells at the same sites of the animal body. The delay in tumor induction was probably due to the presence of 9NC in the culture of the reversibly differentiated cells, which was suppressing the tumorigenic ability of the cells, and subsequent transfer of these cells to a 9NC-free environment, i.e., mouse body, lifted the blockade of their tumorigenic ability. In contrast, the irreversibly differentiated cells had also irreversibly lost their tumorigenic ability, even after removal of 9NC.

Because of these findings, we hypothesize that the presence of 9NC may activate and sustain the expression of genes and proteins associated with the differentiated, nonmalignant state of the cells, or that 9NC downregulates so-called malignant genes and proteins, whereas concomitantly upregulating so called nonmalignant genes and proteins so that the balance of activities within the cell favors the differentiated state. Accordingly, the suppressed malignant state becomes again dominant upon removal of 9NC. To investigate possible roles of cell cycle regulators in 9NC-induced differentiation of the SB1B cells, in this study we monitored changes in the expression of the cyclin-dependent kinase inhibitors, p21 and p16. Upregulation or overexpression of p21 has recently been correlated with differentiation of cells of diverse tissue origin (12,21–33); arrest of differentiated cells at G₁ (36) or G₂ (7); cell cycle arrest at G₁ (28),(29) or G₁ and G₂ (19), regardless of induction of differentiation; and synthesis of melanin and suppression of the tumorigenic ability of human melanoma cells in vivo (34). Our results indicate that p21 expression was markedly declined when the 9NC-treated cells accumulated at the G₂ phase and upregulated when the cells were density-arrested at the G₁ phase of the cell cycle (Figure 5). It has been shown that introduction of the p21 gene into human malignant melanoma cells elicits morphological changes, increased adherence, growth arrest at the G₁ phase, and suppression of tumor growth in animals inoculated with the transfected cells (34). These findings led to the proposition that p21 is sufficient to arrest tumor cell growth by blocking cell-cycle progression through the induction of terminal differentiation and by growth arrest (19). Our results partially agree with this report, that is, p21 is expressed in G₁-arrested melanoma cells, but decreased expression of p21 correlates with terminal differentiation and loss of tumorigenicity in vivo. One possibility for this discrepancy may be due to the mechanism of G₁-arrest of melanoma cells, that is, transfection versus density-induced growth arrest. It has been postulated that in response to p21 increase, cells positive for functional retinoblastoma protein (pRb) are arrested at G₁, whereas, pRb-negative cells are arrested at G₂ and have the tendency to undergo endoreduplication (17). Again, our results partially agree with this postulation; that is, polyploidy appears concomitantly with or after 9NC-induced cell arrest at G₂ when p21 has decreased to the minimal level (Figures 1 and 5). At any rate, the cells with polyploid DNA content do not appear to be tumorigenic, because no tumors appeared in mice injected with inocula containing various numbers of polyploid cells. Further, our results showed that p16 expression is upregulated in SB1B cells arrested at the G₂ phase and substantially declined in the G₁-arrested cells. Upregulation of p16 expression is associated with both reversible and irreversible differentiation of the SB1B cells regardless of loss of the tumorigenic ability. In this regard, it has been suggested that upregulation of p16 may be part of the differentiation program and essential for the maintenance of G₁-arrested senescent cells (48). Our results extend these suggestions to show that p16 upregulation may be important for the differentiation and maintenance of the 9NC-treated SB1B cells, which are arrested at G₂. However, p16 upregulation alone may not be adequate or sufficient to lead to melanocyte differentiation, because other events/factors may also be required to concomitantly occur, such as downregulation of cyclins, inhibition of specific cyclin/CDK kinases, and activation/inactivation of other cooperative proteins. For example, overexpression of cyclin D1 has been demonstrated in familial melanoma (67), and therefore, it is possible that a decrease in this cyclin may facilitate differentiation. On the other hand, mutated CDK4, found in both sporadic and familial melanoma, has lost its capacity to bind p16 (68). Also, cyclin E may play a prominent role in melanoma differentiation, because p16 can inhibit cyclin E-dependent kinases (40). Finally, inactivation of the Ras oncoprotein is another molecular event that may also be important for melanoma differentiation, because concurrent activation of Ras and loss of p16 can accelerate the development of melanoma (69). These and other possibilities are currently under investigation at our laboratory. However, the findings reported by others and in this study collectively suggest that upregulation of p16 expression may be a prerequisite upstream event in the mechanism that eventually results in irreversible differentiation and loss of tumorigenicity of the SB1B cells. Further, the results described in this report clearly demonstrate a reverse reciprocity in, and perhaps mutual exclusion of, the mechanisms involving p16 and p21, thus supporting and extending previous suggestions that these two tumor suppressors function via different pathways (70–72). Finally, 9NC induction of terminal differentiation in malignant melanoma cells indicates the feasibility of so-called differentiation chemotherapy in this type of cancer with 9NC or other congeners and provides a system to study involvement of cell-cycle regulators in cell differentiation and tumorigenicity.