Melanocyte expression of Survivin promotes development and metastasis of UV-induced melanoma in HGF-transgenic mice

Abstract

We previously found the apoptosis inhibitor Survivin to be expressed in melanocytic nevi and melanoma, but not in normal melanocytes. To investigate the role of Survivin in melanoma development and progression, we examined the consequences of forced Survivin expression in melanocytes in vivo. Transgenic (Tg) mouse lines (Dct-Survivin) were generated with melanocyte-specific expression of Survivin, and melanocytes grown from Dct-Survivin mice expressed Survivin. Dct-Survivin melanocytes exhibited decreased susceptibility to UV-induced apoptosis but no difference in proliferative capacity compared to melanocytes derived from non-Tg littermates. Induction of nevi in Dct-Survivin and non-Tg mice by topical application of DMBA did not reveal significant differences in lesion onset (median 10 wks) or density (4 lesions/mouse after 15 wks). Dct-Survivin mice were bred with melanoma-prone MH19/HGF-B6 Tg mice and all progeny expressing either individual, neither, or both (Survivin/HGF) transgenes were UV-treated as neonates and then monitored for 43 wks. Melanocytes in neonatal Survivin⁺/HGF⁺ mouse skin were less susceptible to UV-induced apoptosis than those from Survivin⁻/HGF⁺ mice. Onset of melanocytic tumors was earlier (median 18 vs. 24 wks, p = .01, log-rank test) and overall tumor density was greater (7.7 vs. 5.2 tumors/mouse, p = .04) in Survivin⁺/HGF⁺ compared to Survivin⁻/HGF⁺ mice. Strikingly, melanomas arising in Survivin⁺/HGF⁺ mice demonstrated a greater tendency for lymph node (35% vs. 0%, p = .04) and lung (53% vs. 22%) metastasis, and lower rates of spontaneous apoptosis, than those in Survivin⁻/HGF⁺ mice. These studies demonstrate a role for Survivin in promoting both early and late events of UV-induced melanoma development in vivo.

Introduction

While the incidence of other cancers appears to be stabilizing, melanoma is increasingly more common with approximately 60,000 new cases expected in the U.S. in 2005 (1). Cutaneous melanoma arises in the skin from isolated melanocytes or from melanocytic neoplasms termed nevi, and can be influenced by multiple genetic and environmental factors. Various genetic models based on melanocyte expression of oncogenes (i.e. Ras) or growth factors (i.e. hepatocyte growth factor, HGF), or deficiency of tumor suppressors (i.e. Ink4a/Arf), demonstrate increased susceptibility to melanoma that may be enhanced by exposure to topical carcinogens or UV radiation (2). Patients with advanced melanoma respond poorly to conventional therapies, largely due to acquired apoptosis resistance in tumor cells (3). Although multiple mechanisms underlying apoptosis resistance in melanoma have been identified (3), including dysfunctional death receptor signaling (4), transcriptional repression of pro-apoptotic regulators (5) and upregulation of apoptotic inhibitors (6, 7), the role of apoptosis inhibition in melanoma development is not well understood.

Survivin is an IAP inhibitor of apoptosis (8) widely expressed in cancer, and appears to be an attractive target for cancer therapy (9). Survivin has been implicated in mitotic control (10) and cytokinesis (11), and more recently shown to be an important component of a complex regulating chromosomal alignment during mitosis (12). However, transgenic expression of Survivin in skin keratinocytes was not associated with increased keratinocyte proliferation or epidermal hyperplasia (13) and over-expression of Survivin in HeLa cells did not affect cell cycle progression (14). These experiments, and more recent work examining Survivin promoter activity in tumors (15), suggest that Survivin expression in tumors may be more important for inhibition of apoptosis than driving proliferation. Like other IAPs, which function as caspase inhibitors (8), Survivin can protect cells against apoptosis induced by multiple stimuli (16). The mechanism(s) of apoptotic protection by Survivin, however, may vary with cell type and apoptotic stimulus. For example, its capacity to inhibit caspase-3 has not been observed consistently (17), and while Survivin interacts with caspase-9 and prevents its activation in HeLa cells (18), it appears to protect against caspase-independent apoptosis in other cell types (19, 20).

Our previous studies on Survivin suggested it may be particularly important in melanoma development. First, unlike other apoptotic inhibitors, Survivin is not expressed in normal melanocytes (21). By contrast, Survivin was detected in >85% of localized melanomas and all metastatic melanomas examined (7). Its expression in nevi (7, 22), on the other hand, suggested it may also be a marker of neoplastic melanocytes and play an important role in melanoma initiation. We and others have shown that blocking Survivin in melanoma cells is sufficient for induction of apoptosis (7, 23), sensitization to chemotherapy (23, 24), and inhibition of melanoma tumor growth in vivo (23, 24). Finally, in patients with metastatic melanoma, Survivin expression in tumor cells correlates with poor survival (25, 26).

To investigate the role of Survivin in nevus/melanoma formation and metastasis in vivo, we generated a transgenic mouse with melanocyte-specific expression of Survivin. We found that constitutive Survivin expression in melanocytes was associated with increased formation and metastasis of UV-induced tumors, suggesting a role for Survivin in promoting both early and late events in melanoma development.

Materials and Methods

Dct-Survivin transgenic mice

The Survivin transgene was constructed using pTRP2-pA/StI (27) containing 1.7 kb of the Dct (dopachrome tautomerase, formerly known as TRP2) promoter and 0.8 kb of SV40 polyadenylation sequences. This expression vector was kindly provided by Dr. Paul Overbeek (Baylor College of Medicine), and has been used previously to confer melanocyte-specific expression in mice (27). A cDNA encoding the 140 amino acids and stop codon of mouse Survivin (28) was cloned into the XbaI and ClaI sites of pTRP2-pA/StI between the Dct and SV40 sequences, respectively, and the 3 kb Dct/Survivin/SV40 fusion product was released with KpnI and NotI. This fragment was sequentially purified by agarose gel electrophoresis, electroelution and ion-exchange chromatography, and dialyzed against injection buffer (10 mM Tris, 0.1 mM EDTA, pH 7.4) as described previously (13). Correct orientation and absence of mutations in the Survivin sequence were confirmed by routine DNA sequencing. Finally, the fragment was diluted in injection buffer (1 ng per μL) and microinjected into F1 (C57BL/6 × CBA) embryos, which were implanted into F1 (C57BL/6 x CBA) pseudopregnant females.

Three mice (out of 23 potential founders screened) were Tg (transgenic) by tail DNA genotyping, and all transmitted the transgene to offspring in Mendelian fashion. One founder (#15) did not express the transgene, while the two remaining founders (designated #8, #21) demonstrated comparable expression levels by in situ hybridization (see below). Separate lines were established from these two founders by mating with C3H/HeN mice (Charles River Laboratories, Wilmington, MA). Animals from second and third backcrosses were further propagated by repeatedly mating Tg animals with non-Tg littermates.

Genotyping

Potential founders and littermates were screened for the transgene by PCR. Genomic DNA was prepared from tails using the DNeasy Tissue Kit (Qiagen, Valencia, CA). PCR reactions were performed using primers corresponding to Dct (5′-GTGCACACGATCCTCTAG-3′) and SV40 (5′-GCAGGAATTCGATATCAAGC-3′) sequences, and the 0.5 kb product was visualized on ethidium bromide-stained agarose gels as described previously (13). For negative reactions, PCR was repeated using primers corresponding to diacylglycerol kinase-delta (5′-CTGCAAGTGGACCACATTGGCC-3′ and 5′-CCACTGGTGAGGCATGGCAATC-3′) to confirm DNA integrity.

In situ hybridization

Given the paucity and predominantly follicular distribution of melanocytes in mouse skin (29), and the poor suitability of skin for the antigen retrieval procedure required for Survivin staining (7), we found that Survivin expression in Tg animals was best detected by in situ hybridization. Ten-micron sections were cut from paraffin tissue blocks and adhered to slides using Vectabond (Vector Laboratories, Burlingame, CA). A full-length mouse Survivin antisense riboprobe was prepared by in vitro transcription from a T7-based plasmid using a biotin RNA-labeling kit (Roche Applied Science, Indianapolis, IN) as described elsewhere (30). We did not employ a sense riboprobe as a control given the possibility of reactivity with effector cell protease receptor-1, which is oriented in the opposite direction of the Survivin gene (31). Hybridization, washing and detection were performed with minor modifications as described previously (32). Briefly, the riboprobe was applied in 50% formamide, 2X SSC, 10% dextran sulfate, 0.01% herring sperm DNA and 0.02% SDS at 55°C for 5 hours, followed by successive washing in 2X SSC overnight, 50% formamide/1X SSC at 55°C, and 1X SSC and TBS at room temperature. Sections were blocked in 10% normal sheep serum and 0.03% Triton X-100, and then reacted with alkaline phosphatase-conjugated anti-biotin antibody overnight at 4°C. The hybridized probe was visualized using BCIP/NBT substrate and sections were counterstained with Nuclear Fast Red.

Melanocyte isolation and characterization in vitro

Skin from 1-day-old pups was excised, placed in 70% ethanol for 10 sec followed by PBS, then residual subcutaneous tissue was dissected away. Following incubation in dispase II (25 mg/ml, Roche Applied Science) overnight at 4°C, the epidermis was lifted off and placed in 0.05% trypsin-EDTA (Invitrogen, Carlsbad, CA) for 10 min at 37°C with occasional vortexing. Residual tissue fragments were removed, and serum-containing media was added, cells were pelletted, and resuspended in Opti-MEM medium containing 7% FCS, 7% horse serum, 1% penicillin-streptomycin-glutamine, 0.05 μg/ml Fungizone (all from Invitrogen), 50 ng/ml phorbol 12-myristate 13-acetate, 1 μM sodium vanadate, and 0.2 μM α-melanocyte stimulating hormone (all from Sigma, St. Louis, MO). The medium was changed twice weekly, and fresh phorbol 12-myristate 13-acetate was added each week. Western analysis for Survivin was performed as described previously (7). Proliferative responses were assessed by incorporation of ³H-thymidine. Cells (2.5 × 10⁴ cells per well) were incubated in a 96-well flat bottomed plate with 1 μCi ³H-thymidine (PerkinElmer Life and Analytical Sciences, Wellesley, MA) per well for 16 hrs, then lysed in water, and DNA was collected on A/E glass fiber filters (Gelman Sciences, Ann Arbor, MI) and cpm were detected in Opti-fluor (PerkinElmer) in a scintillation counter. Apoptotic responses were determined by assessing nuclear morphology of cells stained with 6.5 μg per ml 4,6-diamidino-2-phenylindole, as described previously (7).

Chemical-induced nevi

Melanocytic nevi were induced using a modification of a previously described technique (33). Briefly, Dct-Survivin mice and non-Tg littermates aged 3 weeks were weaned and shaved to expose an approximately 1.5 × 3 cm patch of dorsal skin. Four days later, 50 μg freshly prepared 7,12-dimethylbenz[a]anthracene (DMBA, Sigma) was applied topically (100 μL of 0.05% solution in acetone). An additional four applications of DMBA were applied at weekly intervals, and mice were monitored weekly for a total of 15 weeks for development and growth of pigmented lesions.

UV-induced tumors

Dct-Survivin mice were mated with HGF-Tg (BL6 MH19) mice (34), kindly provided by Glenn Merlino (National Cancer Institute). All offspring, representing four possible genotypes (Survivin⁻/HGF⁺, Survivin⁺/HGF⁺, Survivin⁺/HGF⁻, Survivin⁻/HGF⁻), were irradiated unrestrained 1–2 days after birth in open cages at a dose of approximately 4000 J/m² (at a rate of 4 J/m² per sec) using a bank of 4 fan-cooled unfiltered sun lamps (FS20T12-UVB, National Biological Corporation, Twinsburg, OH). These bulbs emit wavelengths between 250 and 420 nm (72.6% UVB, 27.4% UVA, 0.01% UVC), with peak emission at 313 nm, according to the manufacturer. Dosimetry was determined using a UVB-500C meter (National Biological Corp.). Following irradiation, mice were returned to their respective cages, weaned at 3 weeks of age, and monitored weekly thereafter for an additional 40 weeks until euthanized. Animals developing tumors ≥1 cm in diameter prior to 40 weeks were also euthanized.

Tissue analysis

All skin lesions ≥3 mm in diameter were excised from animals reaching experimental endpoints. A subset of animals were necropsied, and pieces of lung, liver, spleen, and regional (usually inguinal) lymph nodes were removed. Tissues were processed in 10% formalin and embedded in paraffin blocks. Sections were cut for routine H&E staining. Lymph node metastasis was interpreted as >50% effacement of node by pigmented cells, or definitive presence of malignant cells (resembling those of primary tumor) in the node. Lung metastasis was associated with the presence of either macroscopic dark dots on lung tissue or large collections of malignant cells seen microscopically. In some cases, sections were bleached to remove melanin prior to H&E staining to remove melanin, permit better visualization of cell morphology, and allow for special staining. For melanin bleaching, sections were deparaffinized, hydrated, and incubated in 0.25% potassium permanganate (Sigma) for 30 min. After washing in water, sections were treated with 5% oxalic acid (Sigma) for 5 min, and then washed again in water. Some slides were stained with the Dct-specific rabbit antibody PEP8 (kindly provided by Vince Hearing, NCI) at 1/300 dilution for 1 hr at 37 °C, and developed with a rabbit Vectastain ABC kit (Vector Laboratories) and counterstained with hematoxylin. All slides were examined by a dermatopathologist (S.R.F.) who did not know the genotype. In some cases, fresh tissue was placed into Karnovski’s fixative overnight and then subjected to standard processing, embedding and sectioning for transmission electron microscopy.

Apoptosis in tissues

For measurement of apoptosis, tumor sections were stained by a TUNEL technique incorporating fluorescein into fragmented DNA using the In Situ Cell Death Detection kit (Roche Applied Science) according to the manufacturer’s instructions. Slides were viewed on a fluorescent microscope, and positive cells were counted within 4–5 representative fields. Fields containing abundant inflammatory cells were excluded from examination. For examination of melanocyte apoptosis in vivo, deparaffinized sections were first TUNEL-stained, followed by staining with PEP8. After incubation with PEP8 antibody and washing with PBS, 1/200 dilution of Alexa Fluor-conjugated goat-anti-rabbit antibody (Invitrogen) in PBS was added to slides which were incubated for 1 hr at 37 °C. After washing with PBS, slides were viewed on a fluorescent microscope, and >300 PEP8-staining cells were scored for TUNEL-positivity.

Statistics

Data derived from multiple animals or determinations was subjected to an unpaired t test using Welch’s correction (Prism, Graphpad software, San Diego, CA). The p-values for some comparisons were obtained using contingency tables in Prism. P values ≤ .05 were considered statistically significant.

Results

Dct-Survivin mice have normal melanocyte distribution and pigmentation

To investigate Survivin function in melanocyte neoplasia and melanoma formation and progression in vivo, Tg mice were generated expressing wild-type mouse Survivin under the control of a Dct promoter. This cassette (Fig. 1A) has been used previously to direct expression of transgenes to melanocytes (27). Two independent Tg lines were generated, demonstrating Survivin expression in hair follicles (Fig. 1B) where melanocytes in mouse skin are known to reside (29). Melanocytes were specifically identified by PEP8 staining and found in and around the hair follicles, but were not increased in number (Fig. 1C) or present in the interfollicular epidermis (not shown) of Dct-Survivin mice. Consistent with their similar skin histology, Dct-Survivin animals from each line were indistinguishable in appearance from non-Tg littermates (not shown). Thus constitutive Survivin expression in melanocytes was not associated with aberrant melanocyte expansion or migration, or altered pigmentation in vivo.

Figure 1.

Construction and characterization of Dct-Survivin mice. (A) The 0.5 kb mouse Survivin cDNA was cloned into the XbaI (X) and ClaI (C) sites of pTRP2-pA/StI between the Dct and SV40 sequences, respectively, and the 3 kb Dct/Survivin/SV40 fusion product was released with KpnI (K) and NotI (N) for microinjection. (B) Sections of skin from non-Tg and Dct-Survivin mice were incubated with a full-length biotinylated mouse Survivin riboprobe, then reacted with alkaline phosphatase-conjugated anti-biotin. Binding was visualized by addition of BCIP/NBT substrate (blue staining), and sections were counterstained with Nuclear Fast Red. Original magnification ×400. (C) Sections of skin from adult non-Tg and Dct-Survivin mice were stained with PEP8 antibody (red) and counterstained with hematoxylin. Non-specific staining of sebaceous glands indicated by arrows. Original magnification ×400.

Transgenic Survivin increases apoptosis resistance but not melanocyte proliferation

Melanocytes isolated from Dct-Survivin mice maintained Survivin expression in culture (Fig. 2A), and were assessed for potential differences in proliferative capacity and susceptibility to apoptosis. There was not a significant difference (p = .57) in cellular proliferation between melanocytes derived from Dct-Survivin and non-Tg animals (Fig. 2B). By contrast, melanocytes from Dct-Survivin mice demonstrated reduced susceptibility to UV-induced apoptosis compared to non-Tg melanocytes (Fig. 2C, p = .02). Thus while transgenic expression of Survivin in melanocytes did not confer a proliferative advantage, it was associated with increased resistance to apoptosis.

Figure 2.

Functional analysis of Dct-Survivin melanocytes. (A) Melanocytes were isolated from Dct-Survivin and non-Tg mice as indicated, and subjected to SDS-PAGE followed by Western blotting for Survivin and Actin. (B) Cellular proliferation of melanocytes isolated from Dct-Survivin (open bar) and non-Tg mice (filled bar), as assessed by incorporation of ³H-thymidine. Error bars indicate SEM of quadruplicate determinations. *, p = .57. (C) Melanocytes from Dct-Survivin (open bars) and non-Tg mice (filled bars) were treated with 480 J/m² UV, then 24 hrs later assessed for apoptosis by nuclear morphology. Arrows indicate apoptotic cells. Error bars indicate SEM of triplicate determinations. *, p = .02. (D) UV-induced melanocyte apoptosis in vivo. Two-day-old Survivin⁻/HGF⁺ (filled bar) and Survivin⁺/HGF⁺ (open bar) mice were UV-irradiated, and 24 hrs later dorsal skin was harvested and subjected to PEP8 and TUNEL staining. Percentages of melanocytes that were TUNEL-positive are indicated. Error bars represent SEM of values from 4 mice in each group. *, p = .02. Color panels show representative staining of melanocytes (PEP8, red) and apoptotic cells (TUNEL, green). Arrows indicate apoptotic melanocytes.

Survivin expression does not increase susceptibility to chemical-induced nevi

To investigate whether Survivin expression predisposes melanocytes to neoplasia, Dct-Survivin mice and non-Tg littermates were treated topically with the chemical carcinogen DMBA and monitored for the appearance and growth of pigmented lesions over a 15-week period. While standard skin carcinogenesis protocols employing the combination of topical DMBA plus phorbol 12-myristate 13-acetate yield keratinocytic neoplasms (35), application of DMBA without phorbol ester produces melanocytic neoplasms (nevi) in C3H/HeN mice (33). Animals treated with DMBA developed darkly-pigmented lesions (Fig. 3A), histologically characterized by intradermal collections of melanized cells (Fig 3B). Approximately 90% of the animals developed nevi (Supplementary Table I), with a mean time to lesion onset of approximately 10 weeks for both Dct-Survivin mice and non-Tg littermates (Fig. 4A, p = .31), and subset analyses of both transgenic lines of Dct-Survivin mice yielded similar results (not shown). Untreated mice observed for up to 9 months did not spontaneously develop nevi. There were not significant differences in lesion density (p = .89), mean lesion size (p = .41), or development of large lesions (p = .78) between Dct-Survivin mice and non-Tg littermates (Supplementary Table I). Thus constitutive expression of Survivin in melanocytes did not increase susceptibility to chemical-induced nevi.

Figure 3.

Melanocytic lesions in Survivin-Tg mice. (A) DMBA-induced pigmented nevi on Dct-Survivin mouse. (B) Nevus, microscopic, H&E stained, ×100. Inset, ×400. (C) UV-induced melanoma (arrow) on Survivin⁺/HGF⁺ mouse. (D) Melanoma, microscopic, ×100. Inset ×400. (E) Enlarged lymph nodes (arrows) from Survivin⁺/HGF⁺ mouse. (F) Lymph node effaced by melanoma cells, microscopic, ×100. Left inset, PEP8 staining of melanoma cells (red) in context of normal lymphocytes (blue) in lymph node. Right inset, electron micrograph demonstrating premelanosome structures. (G) Lung tissue from Survivin⁺/HGF⁺ mouse with dark metastatic foci (arrows). (H) Lung with metastasis, microscopic, x200. (I) prominent vascular involvement in melanoma from Survivin⁺/HGF⁺ mouse, ×400. (J) melanoma infiltrating lymphatic wall, ×400.

Figure 4.

Survivin expression does not affect development of DMBA-induced nevi, but accelerates UV-induced melanoma. (A) Dct-Survivin (n = 31, open circles) and non-Tg (n = 37, filled circles) mice were treated with DMBA and monitored for 15 weeks. Survival curves depicting percentage of animals without melanocytic lesions are charted. P value for comparison of non-Tg and Dct-Survivin mice is .31. (B) Survivin⁺/HGF⁻ (n = 36, filled triangles), Survivin⁻/HGF⁻ (n = 29, open squares), Survivin⁻/HGF⁺ (n = 34, filled circles), and Survivin⁺/HGF⁺ (n = 42, open circles) mice were UV-irradiated and monitored for 43 weeks. Survival curves depicting percentage of animals without melanocytic tumors are charted. A total of 11 animals in the original groups died prior to 43 weeks without tumors (Survivin⁺/HGF⁻,1; Survivin⁻/HGF⁺,5; Survivin⁺/HGF⁺,5) and were not included in these analyses. Dashed lines indicate median time to first tumor formation. P value for comparison of Survivin⁻/HGF⁺ and Survivin⁺/HGF⁺ groups is .01.

Reduced UV-induced melanocyte apoptosis and enhanced tumor formation in Dct-Survivin mice

To examine the impact of Survivin expression on melanocytic tumor formation, we placed the Dct-Survivin transgene on a melanoma-prone genetic background. Dct-Survivin animals were crossed with MH19/HGF-B6 Tg mice, which express an HGF transgene previously shown to predispose to melanoma development following a single neonatal UV exposure (36). Newborn litters from these crosses consisting of animal expressing either individual, neither, or both (Survivin/HGF) transgenes were UV-irradiated, then monitored for 43 weeks for the development of melanocytic skin tumors. In a separate experiment, Survivin⁻/HGF⁺ and Survivin⁺/HGF⁺ neonates were UV-irradiated and the skin was examined for induction of melanocyte apoptosis in vivo by two-color fluorescence microscopy for PEP8 and TUNEL. As shown in Fig. 2D, melanocytes in neonatal Survivin⁺/HGF⁺ mouse skin were significantly (p = .02) less susceptible to UV-induced apoptosis than those from Survivin⁻/HGF⁺ mice. Animals not carrying an HGF transgene (Survivin⁻/HGF⁻, Survivin⁺/HGF⁻) remained tumor-free for the duration of the observation period, while in singly Tg HGF (Survivin⁻/HGF⁺) mice tumors developed with a mean onset of 24 weeks (Fig. 4B). In compound Tg Survivin⁺/HGF⁺ mice, tumor development was accelerated with a significantly earlier tumor onset (median 18 vs. 24 weeks, p = .01, log-rank test) compared to Survivin⁻/HGF⁺ mice (Fig. 4B). Subset analyses of mice derived from separate Dct-Survivin founder lines demonstrated similar results (not shown). Melanocytic tumors were clearly visible on the skin (Fig. 3C), and histologically revealed infiltration of large melanized cells (Fig. 3D). Tumor density was significantly greater (7.7 vs. 5.2 lesions per mouse, p = .04) in Survivin⁺/HGF⁺ compared to Survivin⁻/HGF⁺ mice (Table I). Thus melanocytes in Survivin⁺/HGF⁺ mice demonstrated increased resistance to UV-induced apoptosis and a predisposition towards UV-induced melanocytic tumor formation.

Melanoma metastasis in Dct-Survivin mice

A subset of animals with larger (≤5 mm) tumors was necropsied and lymph nodes, liver, spleen, and lung were examined both grossly and microscopically for metastasis. Metastasis to liver or spleen was not seen in any of the animals examined. However, while none of the lymph nodes examined from Survivin⁻/HGF⁺ mice (0/10) demonstrated metastasis, lymph nose metastasis was observed in 9/26 (35%, p = .04) Survivin⁺/HGF⁺ mice (Fig. 3E,F and Fig. 5A). Melanocytic identity of these infiltrating cells was confirmed by PEP8 staining and visualization of melanosomal structures by electron microscopy (Fig. 3F, insets). Significantly higher rates of lung metastasis (53% vs. 22%, p = .14) were also seen in Survivin⁺/HGF⁺ mice compared to Survivin⁻/HGF⁺ animals (Fig. 3G,H and Fig. 5A). The higher rates of lung metastasis compared to lymph node metastasis, seen in both Survivin⁺/HGF⁺ and non-Tg mice, may reflect hematogenous tumor spread in some cases. Indeed, prominent involvement of blood vessels (Fig. 3I) as well as lymphatics (Fig. 3J) by melanoma cells was frequently noted. A group of melanomas from Survivin⁺/HGF⁺ mice that metastasized was compared to a group (matched for depth; 5.4 mm +1.6 vs. 4.3 mm +1.6, p = .64) of melanomas from Survivin⁻/HGF⁺ mice that did not metastasize for tumoral involvement of blood vessels, lymphatics, and muscle. Although tumoral involvement of these structures was seen in melanomas arising in Survivin⁻/HGF⁺ animals, it was a less common finding than in tumors from Survivin⁺/HGF⁺ mice (Fig. 5B). Finally, melanomas that metastasized in Survivin⁺/HGF⁺ mice contained a lower percentage of apoptotic (TUNEL-staining) cells than matched non-metastasizing tumors from Survivin⁻/HGF⁺ animals (Fig. 5C), consistent with the increased apoptosis resistance seen in melanocytes from Dct-Survivin mice (Fig. 2C). Thus melanomas arising in Survivin⁺/HGF⁺ mice exhibited lower rates of spontaneous apoptosis which was associated with a greater tendency for tissue invasion and ultimately metastasis.

Figure 5.

Survivin promotes melanoma metastasis. (A) A subset of Survivin⁻/HGF⁺ (filled bars) and Survivin⁺/HGF⁺ (open bars) mice were examined for both lymph node and lung metastasis. Number of mice examined indicated in parentheses above bars. P values for comparisons of Survivin⁻/HGF⁺ and Survivin⁺/HGF⁺ mice are .04 (*) and .14 (**). (B) A group of melanomas from Survivin⁻/HGF⁺ mice (n = 9, filled bars) that did not metastasize was compared to a group (matched for depth) of melanomas from Survivin⁺/HGF⁺ mice (n = 10, open bars) that did metastasize for tumoral involvement of blood vessels, lymphatics, and muscle. P values for comparisons of tumors from Survivin⁻/HGF⁺ and Survivin⁺/HGF⁺ mice are .26 (*) and .12 (**). (C) TUNEL staining of tumors in (B) for apoptotic cells. Error bars indicate SEM. P value for comparison of tumors from Survivin⁻/HGF⁺ and Survivin⁺/HGF⁺ mice is .009 (*). Representative fields shown in right panels.

Discussion

Although previous studies have established Survivin as an important viability factor in melanoma cells (7, 23, 24), and correlated its expression in melanoma tumors with poor clinical outcome (25, 26), its role in melanocyte transformation has not been characterized in vivo. Thus the importance of Survivin expression at particular stages of melanoma development and progression is unknown. In this study, we engineered a mouse with melanocyte-specific expression of Survivin that allowed us to investigate its potential role in normal melanocyte function (pigmentation), melanocyte neoplasia (nevus formation), and melanoma development and metastasis.

Although Survivin is potentially involved in both mitotic progression and apoptosis inhibition in malignant cells (37), our previous studies in K14-Survivin mice (with keratinocyte-specific expression of Survivin) revealed reduced susceptibility of keratinocytes to apoptosis (UV-induced) but no effect on keratinocyte proliferation in vivo (13). In other studies we found that forced expression of Survivin in human melanocytes in vitro conferred protection against UV-induced apoptosis, reducing both caspase activation and mitochondrial AIF release (38). Here, constitutive Survivin expression in melanocytes did not cause melanocyte hyperplasia, or affect melanocyte localization or melanin production, as indicated by normal histology, PEP8 staining, and pigmentation of the skin and hair in Dct-Survivin mice. Thus despite its pro-mitotic role in malignant cells (10) and expression in benign nevi (7, 22), Survivin expression in melanocytes was not sufficient to promote their spontaneous proliferation or neoplasia (nevus formation). A lack of histologic changes in association with transgenic expression of Survivin was similarly seen in the skin of untreated K14-Survivin mice (13), as well the bladder of untreated UPII-Survivin mice with uroepithelial Survivin expression (39).

Chemical (DMBA)-induced nevi harbor Ras mutations (40), shown to be sufficient for DMBA-initiated squamous cell carcinoma (35). We found that addition of Survivin in this setting of oncogene-driven melanocyte neoplasia did not accelerate the development or increase the multiplicity of lesions. It is worth noting that in most cases over-expression of inhibitors of apoptosis is not oncogenic per se, in that they are not sufficient for induction of tumorigenesis which requires additional factors (41). Although Survivin expression alone had little effect on normal melanocyte function in vivo, it did confer increased resistance to UV-induced apoptosis. We observed an impact on both early (melanocytic tumor formation) and late events (metastasis) in UV-induced melanocytic tumorigenesis when the transgene was placed on the melanoma-prone HGF background. Survivin Tg animals demonstrated an acceleration of tumor onset, and increased tumor burden, and lymph node and lung metastasis. These results are consistent with studies in UPII-Survivin mice which similarly exhibited increased susceptibility to chemical-induced bladder carcinogenesis (39). While both chemical- and UV-induced skin tumors in K14-Survivin mice demonstrated an elevated rate of malignant conversion to squamous cell carcinoma (42, 43), overall tumor formation was paradoxically reduced due to the role of apoptosis in promoting clonal expansion of premalignant keratinocytes (43). Thus the carcinogenic effects of Survivin expression may differ with cell type and tissue compartment, depending on the consequences of modulating apoptosis at various stages in tumor development.

The mechanisms for enhanced metastatic activity of melanomas arising in Survivin Tg mice are not entirely clear. Apoptotic inhibition may promote metastasis (44), and we did observe lower rates of spontaneous apoptosis in tumors that metastasized in Survivin Tg mice compared to size-matched tumors in non-Tg mice that did not metastasize. It is also possible that melanocyte expression of Survivin in this model may promote metastasis through additional mechanisms distinct from apoptosis. Salz et al. (39) found that Survivin expression UPII-Survivin mice triggered dysregulated expression of multiple extracellular matrix and inflammatory genes. The potential role of all these molecules in tumor metastasis has not been fully investigated, but may represent indirect mediators of metastasis that are downstream of Survivin expression. The capacity of Survivin to promote both early and late events in melanoma development underscores its potential value as a therapeutic target in cancer. If Survivin-expressing melanocytes in nevi are susceptible to apoptosis induced by Survivin inhibition, then targeting Survivin might also be potentially useful as a preventive strategy in patients with numerous nevi and dysplastic nevi, who are at increased risk for melanoma (45).