p53 is the central member of a critical tumor suppressor pathway in virtually all tumor types, where it is silenced mainly by missense mutations. In melanoma, p53 predominantly remains wild type, thus its role has been neglected. To study the effect of p53 on melanocyte function and melanomagenesis, we crossed the ‘high-p53’Mdm4+/− mouse to the well-established TP-ras0/+ murine melanoma progression model. After treatment with the carcinogen dimethylbenzanthracene (DMBA), TP-ras0/+ mice on the Mdm4+/− background developed fewer tumors with a delay in the age of onset of melanomas compared to TP-ras0/+ mice. Furthermore, we observed a dramatic decrease in tumor growth, lack of metastasis with increased survival of TP-ras0/+: Mdm4+/− mice. Thus, p53 effectively prevented the conversion of small benign tumors to malignant and metastatic melanoma. p53 activation in cultured primary melanocyte and melanoma cell lines using Nutlin-3, a specific Mdm2 antagonist, supported these findings. Moreover, global gene expression and network analysis of Nutlin-3-treated primary human melanocytes indicated that cell cycle regulation through the p21WAF1/CIP1 signaling network may be the key anti-melanomagenic activity of p53.
Melanocytes are programmed to survive within the highly mutagenic environment of the skin. It is thought that the same survival pathways are responsible for resistance of melanomas to classical chemotherapy. Conventional chemotherapy targets p53 in many tumor types; however, melanocytes and melanoma cells are resistant to p53-dependent apoptosis. Thus, activation of p53 has been discounted as a potential therapeutic objective in this tumor type. Our data show that p53-dependent cell cycle arrest is the predominant mode of action in normal melanocytes and melanoma cells. Furthermore, increased endogenous p53 may well prevent progression of nevi to malignant and metastatic melanoma. Thus, specific activation of p53 in nevi on individuals who are at high risk for melanoma may be an effective strategy to mitigate some of this risk. Our findings are also in agreement with reports that in human melanomas, increased p53 levels predict a better prognosis and indicate that therapeutic targeting of p53 in melanoma needs to be revisited.
Melanocytes are the pigment-producing cells situated within the skin that can give rise to melanoma, a particularly aggressive cancer. Melanocytes are programmed to survive in the highly mutagenic skin environment, and their major function is to produce melanin, a key ingredient of the protective tanning response that is principally induced by UV exposure. Environmental agents such as UV that induce cellular damage activate the p53 tumor suppressor. While p53 activation results in p53-dependent programmed cell death (apoptosis) in many cell types, melanocytes are resistant to UV-induced apoptosis suggesting that p53 activity is somehow blocked (non-functional p53), a state shared with melanoma cells (Perlis and Herlyn, 2004) that are resistant to conventional modes of chemotherapy that aim to stimulate p53-dependent apoptosis. While p53 is found mutated in most malignancies, a low frequency of p53 mutations (0–10%) was reported in melanoma by several groups (Albino et al., 1994; Castresana et al., 1993; Hartmann et al., 1996; Hocker and Tsao, 2007; Lubbe et al., 1994; Montano et al., 1994; Papp et al., 1996; Sparrow et al., 1995b; Volkenandt et al., 1991). Together, these data suggested that p53 is not a significant player in melanocyte transformation (Box and Terzian, 2008). However, some data suggest that up to 20% of primary and 50% of metastatic melanomas have heterozygous deletions at the p53 locus, indicating that p53 may well play a role in melanoma progression (Hussein, 2004). In keeping with this data, studies using the transgenic Tyr-HrasV12G melanoma-prone mouse model together with p53 deletion identified a clear propensity for melanoma formation (Bardeesy et al., 2001) that was confirmed by the accelerated melanoma incidence in the recent transgenic BrafV600E mouse in the presence of a p53 null genetic background (Goel et al., 2009). Accelerated melanoma incidence was also observed in transgenic BrafV600E zebrafish when p53 was deleted (Patton et al., 2005). More recently, Yu et al. (2009) showed that BRAFV600E transduced human melanocytes with impaired p53 expression and developed melanoma in situ in artificial skin reconstructs. Evidence of p53 pathway disruption including over-expression of p53 (Essner et al., 1998; Florenes et al., 1995; McGregor et al., 1993; Sparrow et al., 1995a,b; Weiss et al., 1995) and its main negative regulators, Mdm2 and Mdm4 (Berger et al., 2004; Polsky et al., 2001, 2002), and loss of expression or deletion of apoptotic target genes such as APAF-1 and PUMA (Karst et al., 2005; Soengas et al., 2001) was also found in melanoma. Thus, contrary to previously held views, p53 has an important role in melanocyte transformation, although its mechanism of action remains unknown. To uncover when and how p53 exerts its tumor suppressive activity, we have followed the progression of pigmented lesions from nevus to melanoma in TP-rasV12G transgenic mice with elevated p53. TP-ras0/+ mice were crossed with mice that carry a deletion of one allele of Mdm4, the negative regulator of p53. Together with chemical activation of p53 in primary melanocytes and melanoma cell lines by Nutlin-3 treatment (specific inhibitor of Mdm2-p53 interaction), we were able to further dissect the mode of action of p53 in primary melanocytes and melanoma.
Mice and DMBA treatment
TP-ras0/+ mice (Powell et al., 1995) and Mdm4+/− mice (provided by Dr G. Lozano (Parant et al., 2001) were maintained on a C57BL6J background and intercrossed. Seven-week-old TP-ras0/+:Mdm4+/− and TP-ras0/+ progeny were shaved, and back skins were treated with 100 μg DMBA (Sigma, St. Louis, MO, USA) dissolved in acetone, once per week for 5 weeks (Broome Powell et al., 1999). After treatment, mice were monitored weekly for tumor formation. Increase in tumor size was measured every 10 days with a caliper. Mice were sacrificed for histopathology when appearing sick or moribund, or with tumors exceeding 20 mm2 in diameter.
Tissue staining and histopathology
Harvested mouse tumors were fixed for 24 h in 4% formalin, processed, embedded in paraffin, and cut into 5-μm sections. Sections were then stained with hematoxylin and eosin (HE) for histopathology, and immunostaining was performed as described previously (Liu et al., 2007; Terzian et al., 2007). Briefly, tissue sections were deparaffinized in xylene followed by rehydration in a series of ethanol/H20 washes and blocked for 1 h at room temperature (RT) with 10% bovine serum albumin (BSA)–2% normal goat serum solution. Sections were incubated with primary antibodies in block solution overnight at 4°C. The antibodies used were rabbit p53 (CM5, 1:400 dilution; Vector Laboratories Inc., Burlingame, CA, USA), mouse S100 (1:5000 dilution; Abcam, Cambridge, MA, USA) and rabbit Ki67 (1:1000; Vector Laboratories Inc.). Sections were then washed and incubated for 1 h at RT with the secondary antibodies goat anti-rabbit and goat anti-mouse conjugated to Alexa Fluor 488 and 594, respectively (Invitrogen, Carlsbad, CA, USA), prepared in block solution. Quantification of p53 and Ki67 positive cells was performed using an eyepiece with a grid covering a surface of 0.5 mm2 surface (20×).
To quantitate pigmentation in HE sections of pigmented lesions, a five-point scale (1–5) was used that represents the inherent pigmentation variation from the most to least pigmented: 1 = 100%, 2 = 75%, 3 = 50%, 4 = 25%, and 5 = 0–10% pigmented areas). The size of pigmented areas within each tumor was then measured using an eyepiece (1 mm2 grid at 10×) and matched with the five-point pigmentation reference scale. Thereafter, a composite pigmentation score, designated Melanin index (Ml), was determined for each tumor based on the size and pigmentation level of the component areas of each tumor. As an example, a 1.7 mm2 nevus may have two areas of pigmentation, with 1.5 mm2 matching the most pigmented pattern (reference level 1 or 100% pigmentation) and 0.2 mm2 of intermediate pigmentation (reference level 3 or 50% pigmentation). Thus, 88% of the tumor was 100% pigmented, and 12% of the tumor was 50% pigmented, equivalent to 6% of the tumor area at the 100% pigmentation level. Therefore, the tumor will score at 94% pigmentation, which could then be recalculated to compare to the original reference scale. Here, 94% will equal a melanin score of 1.24 where 0.24 represents the difference between 1 (100%) and 2 (75%) on the original pigmentation scale. To assess the amount of pigmentation proteins, we lysed A04 melanoma cell line and primary low-passaged human melanocytes in the RIPA extraction buffer (50 mM Tris–HCl pH 8, 150 mM NaCl, 0.5% sodium deoxycholate, 1% NP-40, 0.1% SDS, and protease inhibitors mix). Total proteins were then extracted and separated on 8% polyacrylamide gel by SDS–PAGE electrophoresis as described by standard methods. Precision Plus Protein Standards Kaleidoscope (Bio-Rad, Hercules, CA, USA) was used for molecular weight reference. Briefly, after extraction, proteins were transferred in Tris/Glycine Buffer (Bio-Rad) into 0.45-μm nitrocellulose membranes. Membranes were blocked in Tris-buffered saline (TBS) (pH 7.6)/5% non-fat dry milk for 1 h at RT. Membranes were then incubated with mouse anti-Tyrosinase (TYR; clone T311; 1:200 dilution; Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA) or mouse anti-Tyrosinase-related protein 1 (TYRP1, clone TA99; 1:200 dilution; Abcam) and mouse anti-Actin (1:5000 dilution; Sigma) primary antibodies diluted in the blocking buffer overnight at 4°C. The following day, membranes were washed in TBS and incubated with horseradish peroxidase (HRP)-conjugated anti-mouse secondary antibody (1:2000 dilution; Sigma) diluted in TBS/ 2% non-fat dry milk for 1 h at RT. Protein bands were visualized by using SuperSignal West Pico Chemiluminescent Substrate (Pierce, Rockford, IL, USA).
Tissue culture and Nutlin-3 treatment
Primary human melanocytes (Cascade Biologics, Portland, OR, USA) were cultured in M254 media (Invitrogen) according to supplier’s instructions. Human melanoma cell lines were maintained in RPMI media (Sigma) with 10% fetal bovine serum (Sigma) and antibiotics. The mouse metastatic melanoma cell line from the TP-ras0/+ tumor (Mouse Met) was maintained in Dulbecco’s modified Eagle’s media (DMEM)/F-12 supplemented with 2.5 μg/ml insulin (Sigma), 25 μg/ml transferrin (Sigma), 250 ng/ml epidermal growth factor (EGF, Gibco BRL, Carlsbad, CA, USA), and 10% fetal bovine serum and antibiotics. All cells were maintained at 37°C in a 5% CO2 incubator. Racemic Nutlin-3 (Cayman Chemical Company, Ann Arbor, MI, USA) was dissolved in DMSO (Sigma) and diluted in media just before use.
Viability, proliferation and senescence
Sensitivity of cells to Nutlin-3 treatment was assessed by apoptosis, proliferation, and senescence assays performed in triplicate. For apoptosis/cell death assay, floating and adherent cells were harvested and double stained with YO-PRO-1 (Invitrogen) and 20 μg/ml propidium iodide (PI; Sigma) dyes according to manufacturer’s recommendation. CountBright absolute counting beads (Invitrogen) containing a calibrated suspension of brightly fluorescent microspheres (50 μl) were added before flow cytometric analysis. The counting beads permitted us to accurately quantify cell populations in each sample avoiding separate measurements. Flow cytometry was performed using 488 nm excitation with green fluorescence emission for YO-PRO-1 dye and red fluorescence emission for PI. Cells were separated into three populations: live cells showing low levels of green fluorescence, apoptotic cells showing high levels of green fluorescence, and dead cells showing both red and green fluorescence. Viability was then assessed by deducting the number of dead and apoptotic cells from the total number of cells. For cellular proliferation, after 24 h of Nutlin treatment, adherent cells were incubated with 10 μM BrdU (Amersham Biosciences, Pistacaway, NJ, USA) for 1 h at 37°C, harvested, and fixed with 70% ethanol/PBS. DNA is denatured with 2 N HCl/0.1% Triton for 40 min at RT. After a couple of washes with PBS- 0.1% Triton, cells were stained with anti-Brdu antibody conjugated to Alexa Fluor 488 (1:40; Invitrogen) and 20 μg/ml PI in the presence of 50 μg/ml RNase A (Invitrogen) and 50 μl of CountBright absolute counting beads. Samples were then submitted to flow cytometric analysis. All flow cytometry was performed at the UC Denver to the Stem Cell Biology Flow core (Aurora, CO, USA). For senescence, we treated human primary melanocytes with Nutlin-3 for 72 h and 1 week. We then measured senescence-associate β-galactosidase activity (SA-β-gal) as described (Bandyopadhyay et al., 2005).
Six thousand cells per well (1 cm2) were added to an 8-well chamber slide containing 60 μl of Matrigel matrix (BD Biosciences, Franklin Lakes, NJ, USA) and 400 μl of DMEM supplemented with 10% FBS. Colony growth was observed for 72 h, and colonies with ≥50 cells were counted. Assays were performed in triplicate.
RNA isolation, microarray analysis, and gene expression validation by qRT-PCR
Total RNA was extracted using RNA STAT-60™ (Tel-Test Inc, Woodlands, TX, USA) from 0-, 18- or 72-h Nutlin-3-treated primary melanocytes (passage 4 for each time point) grown to 70% confluence. A total of 4 replicates were used for each time point. Whole-genome expression analysis was performed on extracted RNA using the Illumina HT12 BeadChip platform (Ilumina Inc., San Diego, CA, USA). cRNA labeling, hybridization, and data acquisition were performed at the Mind Research Network microarray core facility (Albuquerque, NM, USA). The HT12 gene expression array platform interrogates 31 000 annotated genes and was selected as it offers almost complete coverage of known and predicted genes within the human genome. Array data were analyzed using Genespring GX11 (Agilent Tech, Santa Clara, CA, USA) and Ingenuity pathway analysis tools (http://www.ingenuity.com). Gene ontology annotation was analyzed using the web-based david software (http://www.david.abcc.ncifcrf.gov; Dennis et al., 2003; Huang et al., 2009). Microarray data were validated by qPCR analysis using a Roche Lightcycler 480 system (Roche Diagnostics, Indianapolis, IN, USA) as previously described (Torchia et al., 2009). All TaqMan assays MDM2 (Hs010669 30_m1), cell cycle regulator CDKN1A (Hs00355782_m1), apoptotic genes FAS (Hs00163653_m1), APAF1 (Hs00559441_m1) and PIG3 (Hs00153280_m1), KITLG (Hs00241497_m1), AKT1 (Hs00920503_m1), CDK2 (Hs01548894_m1), C-KIT (Hs00174029_m1), TYR (Hs00165976_m1), TYRP1 (Hs00167051_m1), MITF (Hs01117294_m1), POU3F2 (Hs00271595_s1), HMMR (Hs00234 864_m1) and CYR61 (Hs00155479_m1) used in this study were purchased from Applied Biosystems (Carlsbad, CA, USA). The relative expression levels of these genes were normalized to 18S mRNA (Applied Biosystems).
Statistical and survival analysis
Graphpad Prism (version 5, La Jolla, CA, USA) software was used. To compare means, we used the unpaired T-test analysis. Log-rank test and Kaplan–Meier analysis were performed to compare survival of two cohorts. Statistical difference was considered significant if it had a two-sided P-value of <0.05.
The TP-ras0/+ mouse as a model of pigmented lesions
To understand the impact of the p53 pathway on the molecular mechanisms underlying melanoma initiation and promotion, we chose a well-characterized model of melanoma, the TP-ras0/+ mouse (Powell et al., 1995), that expresses activated HRASV12G under a mouse tyrosinase promoter. This mouse is an excellent model of progression because a topical treatment of DMBA induces all the stages of melanomagenesis from benign nevus to cutaneous melanoma and distant metastasis (Figure 1) (Broome Powell et al., 1999).
Age of onset of pigmented lesions in DMBA-treated TP-ras+/0 and TP-ras+/0: Mdm4+/− mice
We crossed TP-ras0/+ mice to Mdm4+/− mice [high p53 model (Parant et al., 2001; Terzian et al., 2007)] to produce TP-ras0/+ and TP-ras0/+: Mdm4+/− progeny. We treated a cohort of these two genotypes topically with DMBA and followed daily for the appearance of pigmented lesions (nevi and melanoma; Figure 1A). Tumor sizes were measured every 10 days from emergence. As reported, DMBA-treated TP-ras0/+ mice (Broome Powell et al., 1999) developed papillomas, nevi, and cutaneous melanomas that metastasized to liver, lung, and lymph nodes (Figure 1 and data not shown). Papillomas were excluded from the analysis. In total, 25 of 28 TP-ras0/+ mice (89.2%) developed at least one pigmented lesion, versus 11 of 17 TP-ras0/+: Mdm4+/− mice (65%). Furthermore, TP-ras0/+ mice had more pigmented lesions per mouse, with 4.0 tumors/mouse compared to 2.8 tumors/mouse in TP-ras0/+: Mdm4+/− mice. While treated TP-ras+/0: Mdm4+/− mice developed their first pigmented lesion at the same rate as TP-ras0/+ mice (Figure 2A), the latency of the emergence of their melanomas was delayed (T-test P = 0.0299; Figure 2A). The mean age of onset of all individual tumors was also significantly extended (T-test P = 0.0012; Figure 2A). Thus, a small but measurable effect of high p53 in a DMBA-treated TP-ras0/+ background was seen on the appearance of melanomas and on the overall onset of all lesions. A summary of data presented here is shown in Table S1.
Progression of pigmented lesions in DMBA-treated TP-ras0/+ and TP-ras0/+: Mdm4+/− mice
Next, we examined the impact of elevated p53 on tumor progression defined by tumor growth and histopathology. As mentioned previously, 89.2% of TP-ras0/+ mice developed pigmented lesions versus 65% of TP-ras0/+: Mdm4+/− mice (Figure 2B). The majority of these cutaneous tumors showed limited growth and were <5 mm2 in size. Therefore, we classified these as nevi (95/113 in TP-ras0/+ and 36/47 nevi/pigmented lesions in TP-ras0/+: Mdm4+/− mice). Pigmented lesions of TP-ras0/+ (16%) and TP-ras0/+: Mdm4+/− mice (23%) that continued growing beyond 5 mm2 were considered melanoma. To assist with the definition of nevi versus melanoma, histopathology and Ki67 staining were used (Figures S1, S2). Nevi had a very low Ki67 proliferation index (<3 cells/field) with no evidence of histologic signs of melanoma (Figure S2). On the other hand, melanomas had a high Ki67 staining index (average 158 cells/field), and some or all histologic signs of melanoma (Figure S2). In TP-ras0/+ mice, eight of these lesions (7%) progressed rapidly to reach 10 mm2 and 4 (3.7%) reached 20 mm2. All tumors >10 mm2 metastasized. In contrast, no large TP-ras0/+: Mdm4+/− melanomas (>10 mm2) or metastatic lesions were observed (Figure 2B). Thus, p53 effectively prevented the conversion of small benign tumors to malignant and metastatic melanoma.
Measuring the growth of several individual tumors from each genotype, every 10 days from emergence, showed a rapid progress for many TP-ras0/+ melanomas versus a slow but steady increase in size for all examined TP-ras0/+: Mdm4+/− tumors (Figure 3A, B). We also compared the growth properties of nevi and melanomas that appeared in both groups, especially that nevi and melanomas appeared to grow differently, and a sizeable difference in melanoma growth patterns was observed as well. When the average size of emergent lesions was compared, nevi were considerably smaller than melanomas: 1.8 versus 2.6 mm2, respectively, for TP-ras0/+ and 1.7 versus 2.4 mm2 for TP-ras0/+: Mdm4+/− lesions. Thus, high p53 had no apparent effect on the average emergent size of both nevi and melanoma. Additionally, nevi of both groups grew at an identical rate of 0.015 mm2/day (Figure 3C). The impact of p53 was most evident in the explosive growth phase of 5–10 mm2, averaging 0.224 mm2/day (n = 10) for TP-ras0/+ versus 0.040 mm2/day (n = 5) for TP-ras0/+: Mdm4+/− melanomas (Figure 3C). Together, these data emphasize the role of p53 in limiting melanoma growth, and ultimately metastasis.
Tumor spectrum and overall survival of DMBA-treated TP-ras0/+ and TP-ras0/+: Mdm4+/− mice
A number of TP-ras0/+ mice (32%) were sacrificed because of large melanomas (>10 mm2) that metastasized, among which two mice suffered from lung adenocarcinoma (7%) and two others had lymphomas (7%). Additionally, some TP-ras0/+ mice died of unknown causes (29%) or were sacrificed at study completion (39%) (Figure 3D). Conversely, no TP-ras0/+: Mdm4+/− mouse developed large melanomas that required sacrifice. Large squamous cell carcinomas were observed in 12% of TP-ras0/+: Mdm4+/− mice, while others died unexpectedly (18%) or were sacrificed at study completion (70%) (Figure 3D). Furthermore, no sign of metastasis was found at necropsy or by histopathological examination. Interestingly, despite a striking effect of high p53 on tumor progression and metastasis, a mild but significant increase in the overall survival of TP-ras0/+: Mdm4+/− mice was observed by Kaplan–Meier analysis (median survival = 257 days for TP-ras0/+ versus 330 days for TP-ras0/+: Mdm4+/− mice; P = 0.0441 Log-rank test; Figure 3D).
p53 staining, pigmentation, and progression
To determine whether p53 levels correlated with progression, we performed p53 IF on TP-ras0/+ and TP-ras0/+: Mdm4+/− lesions. Low positive staining was observed in most TP-ras0/+ nevi (average 4 positive cells/field; Figure 4A; Figure S1) consistent with previous reports (Hacker et al., 2006). In rapidly growing TP-ras0/+ melanomas, a tendency toward increased p53 positivity (average 6.5 positive cells/field) was detected. On the other hand, in TP-ras0/+: Mdm4+/− nevi and melanomas, p53 immunopositivity was markedly higher (average 32 and 28 positive cells/field, respectively) than in TP-ras0/+ lesions (Figures 4A; Figure S1).
We also noticed an association between pigmentation levels and progression. A melanin (Ml) index based on the relative sizes of differently pigmented areas (mm2) within each tumor in comparison with a five-point scale of pigmentation from most to least pigmented was used (Figure 4B). It was apparent that TP-ras0/+ nevi were heavily pigmented (Ml = 2.03), with progressive reduction in pigmentation in intermediate melanomas (5–10 mm2; Ml = 2.74) and in large melanomas (>10 mm2; Ml = 4.26; Figure 4B). In comparing all TP-ras0/+ nevi (Ml = 2.03) to all melanomas (Ml = 3.67), the reduction in pigmentation was very significant (T-test, P < 0.0001). Additionally, TP-ras0/+: Mdm4+/− nevi (Ml = 1.34) were significantly darker than TP-ras0/+ counterparts (Ml = 2.03, P < 0.0376), and TP-ras0/+: Mdm4+/− melanomas (Ml = 1.63) were darker than TP-ras0/+ counterparts (Ml = 3.67, P < 0.0009; Figure 4B).
Chemical activation of p53 in melanocytes and melanoma cells
The impact of p53 on melanoma initiation and progression through its anti-proliferative activity was evident in our mouse model. To further examine the mechanism of action of p53, we used a low passage human neonatal melanocyte culture (MC), seven other well-characterized human cell lines: A04, BL, D08, MM329, MM604, SKMEL13, and WM35, and a mouse melanoma cell line derived from a TP-ras0/+ metastatic tumor (Mouse Met). The genetic characterization of the human cell lines is represented in Table S2. p53 activation was induced by treating cells with the very specific Mdm2 inhibitor, Nutlin-3 (Vassilev et al., 2004), and its impact on cell viability, proliferation, and apoptosis was assessed. We also examined senescence in Nutlin-treated primary melanocytes.
Viability, proliferation, and senescence
We used a wide range of Nutlin-3 concentrations (0, 0.1, 0.5, 1, 5, 10, 20, 30, and 40 μM) and tested cell viability after 72 h of treatment of human normal melanocytes and melanoma cell lines by subtracting dead and apoptotic cell number (YO-PRO-1/PI assay) from the total cell numbers (Figure 5A and data not shown). A noteworthy decrease in viability was only observed beyond 10 μM that accentuated with higher doses (>20 μM). Nutlin-3 is commonly used at ∼10 μM to induce apoptosis in many non-melanoma cancer cell lines (Vassilev et al., 2004). Earlier time points (24 and 48 h after treatment) showed that melanocytes and melanoma cells underwent cell cycle arrest with minimal apoptosis at doses starting from 0.1 to 10 μM and as early as 24 h (Figure 5B and data not shown). Beyond 20 μM, a significant increase in apoptosis was observed (data not shown). Moreover, p53 wild-type cells were variably sensitive to Nutlin-3 (Figure 5A), while others that carry heterozygous mutations of p53 (SKMEL13 and BL) were particularly resistant and needed doses higher than 30 μM to be affected (Figure 5A). Interestingly, normal melanocytes had an intermediate sensitivity to Nutlin-3 (Figure 5A). In summary, at doses lower than 20 μM, Nutlin-3 mainly caused cell cycle arrest in melanoma cells and beyond this dose, it mainly induced apoptosis. We also checked for senescence in Nutlin-treated melanocytes using the senescence-associated β-galactosidase (SA-β-gal) assay. No increase in SA-β-gal positivity was observed in 10 or 20 μM treated cells (24–72 h and 1 week) compared to untreated cells indicating that the treatment did not induce senescence (data not shown). However, induction of a senescence program from a chronic treatment of Nutlin cannot be ruled out using our experimental conditions.
Effect of high p53 on clonogenicity of melanoma cell lines
To check whether high p53 influences the clonogenic potential of melanoma cells, we performed a standard assay that measures the ability of single cells to form colonies. Thus, 6000 cells of A04, DO8, MM504, and TP-ras0/+ metastatic mouse melanoma were suspended in media with or without 10 or 30 μM Nutlin-3 and were added in triplicate to an 8-well chamber slide coated with Matrigel matrix to allow colony growth. After 72 h of culture, the number of colonies containing more than 50 cells was counted and plotted. All untreated cell lines were able to form colonies with the exception of WM35 (Figure 6A and data not shown). At 10 μM Nutlin-3, clonogenicity of A04, DO8, MM504, and Mouse Met cell lines was completely abrogated (Figure 6A). On the other hand, for MM329, it was diminished at 10 μM and strongly reduced at 30 μM. BL and SKMEL13 mutant p53 cell lines were as expected resistant to Nutlin-3 treatment and demonstrated no difference in clonogenicity (Figure 6A and data not shown).
Effect of high p53 on pigmentation
p53 has been implicated in differentiation (Lin et al., 2005; Meletis et al., 2006). Therefore, we examined 10 μM Nutlin-3-treated melanocytes and melanoma cells for signs of differentiation such as changes in morphology or pigmentation. Although we did not observe a strong effect on dendricity of the cells, we detected a marked darkening of treated melanocytes and A04 cells that intensified at 48 h and persisted throughout a week of treatment (Figure 7A). After 10 μM Nutlin-3, A04 cells were the most pigmented, followed by primary melanocytes and Mouse Met cells (Figure 7A and data not shown). The remaining treated melanoma cell lines pigmented less or not at all. Thus, p53 can significantly up-regulate pigment production in some melanoma cell lines. To investigate this further, we performed Western blots for two key melanogenic enzymes, tyrosinase (TYR) and tyrosinase-related protein 1 (TYRP1), on protein extracts from primary melanocytes and A04 melanoma cells untreated and treated with 10 μM Nutlin-3 for 24 and 72 h. It was evident that TYR protein levels increased in treated melanocytes and A04 cells. Furthermore, the heavily glycosylated TYR form (top band) was increased in treated cells (Figure 7B). On the other hand, no changes in TYRP1 protein levels were observed at 24 h after treatment and were significantly down-regulated in 72 h of Nutlin-3 treated melanocytes and A04 cells (Figure 7B).
Transcriptional effect of p53 activation in Nutlin-3-treated melanocytes
To determine the global transcriptional changes associated with p53 induction, we profiled primary human melanocytes treated with 10 μM Nutlin-3 for 0, 18, and 72 h using Illumina HT12 gene expression BeadChips. A total of four replicates were used for each time point. We chose the 18-h time point to assess the immediate p53 transcriptional targets and the 72-h time point to assess the effect on downstream events. At 18 h, anova analysis identified 1492 up-regulated genes, while 385 genes were down-regulated relative to 0 h (P < 0.05, Table S3). At 72 h, 1026 genes were found up-regulated and 488 genes down-regulated (P < 0.05; Table S4), suggesting a pattern of mild compensation. Analysis of Gene Ontologies (GO) for the up-regulated genes at 18 h using DAVID revealed significant enrichment (FDR P < 0.05) of GO terms associated with apoptosis (109), cell adhesion (110 genes), and intracellular transport (69 genes). In contrast, genes associated with cell cycle progression (57 genes), DNA replication (37 genes), and chromosomal maintenance (18 genes) terms were repressed at this time point. With persistent Nutlin-3 treatment (72 h), there appeared to be some habituation to p53 signaling, with no up-regulated genes reaching the FDR P = 0.05 significance cutoff. However, genes associated with cell cycle progression (47) remained down-regulated.
Using the Ingenuity Pathway Analysis tools, we examined the status of canonical p53 target genes at 18 h. Consistent with the GO term analysis, apoptotic genes including FAS, DR4, APAF1, PIDD, PIG3, TEAP, and STAG were up-regulated (Figure 8A). While many apoptotic genes were induced, melanocytes and melanoma cells were resistant to apoptosis at 10 μM Nutlin-3 treatment. Thus, we examined genes and pathways that may be important for melanocyte survival or resistance to apoptosis. Increased expression of KITLG, AKT1, SHC, N-RAS, PP2A, and integrins (ITG) was observed (Table S3), indicating a potential increase in signaling of the PI3K-AKT pathway, thus promoting cell survival (Hay, 2005). Moreover, CDKN1A (p21WAF1/CIP1) and NOTCH 1 were both increased, and are both thought to play a role in melanocyte cell survival (Gorospe et al., 1997; Moriyama et al., 2006). No expression changes were observed in the pro-survival MITF-BCL2 family signaling pathway (McGill et al., 2002) (Figure 8A; Table S3).
In addition to CDKN1A, the cell cycle regulator GADD45A was also up-regulated (Figure 8A). Network analysis of differentially regulated genes revealed the top-ranked network centered on CDKN1A at 18 h (Figure 8B), as well as at 72-h post-Nutlin-3 treatment (data not shown). CDKN1A was found 21-fold over-expressed at 18 h by qPCR analysis (Figure S3). Examination of this network revealed that CDK2 and interacting partner CCNA2 were down-regulated, while CCNE1 was up-regulated, indicating that cell cycle progression is de-regulated with arrest at the G1–S phase transition (Figure 8B).
It is known that p53 activation by UV or DMBA induces the expression/activity of DNA repair or oxidative stress detoxification genes (Abdel-Malek et al., 2010; Marrot et al., 2005; Smith et al., 2000; Zhao et al., 2006). Our microarray detected up-regulation of genes such as ERCC3 and 5 and OGG1 involved in these pathways. Also, as part of the UV response, melanocytes are recruited to the burned skin area after a sunburn event (Yamaguchi et al., 2008). Consistent with this migratory effect, 110 cell adhesion and migration genes were up-regulated after 18 h of Nutlin-3 treatment. These genes included KITLG, CDH2, ITGB1, ITGAB1, and ITGA2, 3, 5, and 6. KITLG and integrins have well-known roles in melanocyte migration (Grichnik, 2006; Scott et al., 1997). On the other hand, CDH1 (E-Cadherin) and CDH3 (P-Cadherin) were both down-regulated (Table S3), suggesting loss of anchorage to the basement membrane (Haass and Herlyn, 2005).
As increased pigmentation was observed in Nutlin-3-treated primary melanocytes and melanoma cells, we examined the expression pattern of known pigmentation genes. At 18 h, we observed up-regulation of autocrine KITLG and POU3F2 (∼11-fold by qPCR; Figure S3) and down-regulation of pigmentation genes including KIT, SLC24A4, GRP143, HPS4, and DCT. At 72 h, SILV was added to this list. KITLG has a strong pro-pigmentary effect, while down-regulation of the other pigmentation genes may moderate pigment production. Surprisingly, we did not observe any changes in gene expression (by microarray or by qPCR analysis) for the pigmentation genes TYR and TYRP1 (Figure S3; Table S3), previously implicated as p53 target genes in melanocytes (Khlgatian et al., 2002; Nylander et al., 2000).
In this report, the TP-ras0/+: Mdm4+/− mouse model was used to show that the major role of p53 in melanoma is to restrict the rapid growth associated with the final stages of melanomagenesis, thereby limiting metastatic spread (Figure 8C). In addition, chemical activation of p53 in primary human melanocytes and melanoma cell lines using Nutlin-3 demonstrated that these cells were by far more sensitive to cell cycle arrest than to apoptosis. Indeed, gene expression profiling identified the CDKN1A signaling network as the predominant player in Nutlin-3-treated primary melanocytes. We propose that p53 successfully controls melanoma growth by its anti-proliferative activity rather than its apoptotic ability. Thus, it is this powerful activity of p53 that rapidly growing TP-ras0/+ tumors need to overcome to continue their progression to metastasis, while having stabilized wild-type p53.
A longitudinal analysis of murine TP-ras0/+ and TP-ras0/+: Mdm4+/− pigmented lesions enabled us to examine the impact of high p53 on activated Ras melanocytes, the development of melanocytic nevi and melanoma progression. Melanomas of both genotypes had a larger emergent size than nevi and maintained a higher proliferative rate indicating that DMBA mutagenesis introduced the necessary molecular changes to progress to melanoma, whereas nevo-melanocytes may have only acquired a subset of these mutations. Subsequently, these tumors behaved differently: TP-ras0/+ melanomas passed through a nevus-like phase, growing at a slow pace followed by an explosive growth with metastatic conversion, comparable to human melanomas arising in preexisting nevi (Weatherhead et al., 2007), while TP-ras0/+: Mdm4+/− melanomas remained small and never metastasized. These data highlight the crucial role of p53 in hindering melanoma progression. While some anti-nevogenic activity of p53 was present, with slightly fewer and later appearing pigmented lesions in TP-ras0/+: Mdm4+/− mice, p53-dependent inhibition of progression from a nevus to malignant melanoma was far more significant. The reduced proliferation in TP-ras0/+: Mdm4+/− tumors is in close agreement with the fact that shRNA knockdown of p53 promoted proliferation of BRAFV600E transduced primary human melanocytes and dramatically increased their colony-forming ability (Yu et al., 2009). Our data go further to suggest that antagonizing p53 signaling strongly enhanced the progression from nevus to malignant and metastatic melanoma. A low level of p53 was detected by IF in TP-ras0/+ nevi, and it increased in large and rapidly growing melanomas, with some heavily stained for p53. Consistent with this observation, immunopositive wild-type p53 was also identified in up to 40% of human melanomas (Box and Terzian, 2008). We conclude that wild-type p53 is stabilized at a later stage in TP-ras0/+ tumor progression, perhaps after an initial malignant conversion, but before any metastatic spread. As these tumors continued their typical progression, it is apparent that they must have found a way to counteract the anti-proliferative activity of p53. It is possible in a carcinogenesis model such as ours that secondary mutations are already in place that blunt the anti-proliferative activity of wild-type p53, or that melanoma cells may undergo an adaptive response without the requirement to reduce wild-type p53 levels. Interestingly, previous studies on TP-ras0/+ or other melanoma-prone mouse models have shown that DMBA or UV-induced p53 mutations/deletions are rare in melanomas (Gause et al., 1997; Hacker et al., 2006; Sotillo et al., 2001). Additionally, the low frequency of p53 mutations/deletions in mouse melanomas with deficient Cdkn2a may provide an explanation for the low frequency of p53 mutations in human tumors (Bardeesy et al., 2001; Gause et al., 1997). Nevertheless, it may not be possible to accurately determine the role of p53 in p19Arf-deficient mouse models. In the TP-ras0/+: Mdm4+/− mice, the fewer and delayed appearance of pigmented lesions with increased growth inhibition suggests that p53 is more active from the start and remains active throughout progression. It is possible that this increased p53-dependent growth inhibition seen early on in TP-ras0/+: Mdm4+/− tumor progression may have hindered the activation of the switch that stabilizes wild-type p53.
The early effects of high endogenous p53 on melanomagenesis may be indicative of increased DNA repair activity, oxidative stress detoxification, or even the result of other key tumor suppressor functions such as cell cycle arrest or apoptosis. In agreement, our microarray data identified up-regulation of key p53 target genes involved in these processes. On the other hand, Christophorou et al. (2006) demonstrated that the p53-dependent DNA damage response contributes minimally to its tumor-suppressive activity in vivo, and recently, enhanced cyclobutane dimer detoxification was found to have no effect on tumor formation in UV-irradiated CDK4R24C/R24CNRASQ61R melanoma-prone mice (Hacker et al., 2010). However, an enhanced DNA repair or oxidative stress detoxification capacity in melanocytes programmed to live a lifetime seems critical to minimize the deleterious effects of UV damage, therefore preventing unrestrained pigmented lesion formation and alleviating melanoma risk (Cotter et al., 2007; Schoppy et al., 2010).
The decision to induce cell cycle arrest (survival) or apoptosis after p53 activation revolves around p21WAF1/CIP levels (Garner and Raj, 2008). It was clear that Nutlin-3-treated melanocytes and melanoma cells with wild-type p53 underwent substantial cell cycle arrest in preference to apoptosis at a low dose. Thus, p53 activation in primary melanocytes induced a strong increase in CDKN1A and GADD45A expression. Moreover, a network of CDKN1A interacting cell cycle regulatory genes dominated the p53 transcriptional response. It is known that at a similar dose, Nutlin-3 induced heavy apoptosis resulting in decreased viability of wild-type p53 cancer cell lines (Vassilev et al., 2004). Interestingly, Nutlin-3-treated primary melanocytes also had a robust induction of many apoptotic genes suggesting that apoptosis is subjected to robust post-transcriptional silencing. In support of our data, previous studies where a melanoma cell line was transduced by an adenovirus expressing wild-type p53 (Gorospe et al., 1997), or treated with Nutlin-3 (Smalley et al., 2007), showed that cell cycle arrest was the preferential mode of action of p53. p21WAF1/CIP1 has been implicated as key survival promoter, and CDKN1A knockdown in a melanoma cell line resulted in sensitivity to apoptosis following p53 activation (Gorospe et al., 1997). However, these reports had limited perspective as they were performed in one or two melanoma cell lines. It has been proposed that melanocytes have a strong survival program to support a life-long presence in the high mutational environment of the skin, and resistance to p53-dependent apoptosis is a key aspect contributing to the chemotherapy resistance of melanoma (Soengas and Lowe, 2003). This resistance to various agents, including UV light, that normally induce p53-dependent apoptosis in other cell types (Bae et al., 1996) is comparable to melanocyte and melanoma resistance to Nutlin-3.
On the other hand, TP-ras0/+: Mdm4+/− tumors were distinguished by darker pigmentation compared to TP-ras0/+ tumors. In fact, loss of pigmentation, a feature of tumor progression in humans (Henning et al., 2007; ACS 2010, http://www.cancer.org/Cancer/SkinCancer-Melanoma/DetailedGuide/melanoma-skin-cancer-detection), was preserved in TP-ras0/+ tumors. Moreover, we observed increased pigmentation in Nutlin-3-treated primary melanocytes and some melanoma cell lines. Together, these data support a role of p53 in melanogenesis, consistent with reports involving p53 as a differentiation promoter (Hong et al., 2009; Lin et al., 2005; Meletis et al., 2006), as increased pigmentation is synonymous with melanocyte differentiation. The major pro-melanogenic p53 target gene identified by microarray analysis of Nutlin-3-treated primary human melanocytes was KITLG, a cytokine that is responsible for paracrine stimulation of pigmentation and melanocyte migration (Grichnik, 2006). In contrast, no changes in expression of TYR or TYRP1, previously implicated as pro-melanogenic p53 targets (Khlgatian et al., 2002; Nylander et al., 2000), were observed. This discrepancy could be explained by the possible variation in p53-dependent gene expression with cellular context, mode of activation (e.g. UV versus Nutlin-3), and dosage of the activating agent (Yoon et al., 2002). Nevertheless, we observed an increase in total and glycosylated TYR protein levels in Nutlin-3-treated cells. TYR is normally subject to a high level of post-translational modification that is required for optimal activity (Halaban et al., 2001; Wang and Hebert, 2006), and this includes glycosylation. In contrast, TYRP1 protein levels, while unchanged at 24 h of treatment, were significantly reduced at 72 h, perhaps in response to the cellular stress because of prolonged p53 activation. These data suggest that TYR, possibly by post-translational modification, may well be a p53 target for pigmentation in stressed cells. Consistent with the fact that p53 contributes to the protective tanning response (Cui et al., 2007; Goukassian et al., 1999), p53-null mice are not reported to have a remarkably altered pigmentation phenotype, confirming that p53 acts on melanocytes largely in its damage response capacity, rather than through a role in normal development. Taken together with our data, we speculate that p53 activation in chronically exposed skin may contribute to the paradoxic findings that melanoma incidence is reduced compared to intermittently exposed skin (Elwood, 1996; Gandini et al., 2005), linking the role of p53 in tanning to its tumor-suppressive activity.
In conclusion, our high p53 melanoma model provides in vivo evidence that specific activation of p53 in individuals with many pigmented lesions may be effective in diminishing the risk of progression to metastatic melanoma. We also demonstrated, both in vivo and in vitro, that the anti-proliferation arm of p53 is the preferential mode for its tumor-suppressive action in normal and transformed melanocytes. Therefore, p53 activation in individuals with strong melanoma risk factors such as high nevus counts by topical agents may be a more convenient preventive measure over a costly and burdensome surgical approach. Moreover, these findings underscore the importance of a personalized medicine geared toward knowing the unique clinical, genetic, genomic, and molecular information of each patient with melanoma and guiding the clinician toward the best therapeutic strategy.
This study was supported by a Dermatology Foundation RCDA and an American Skin Association CDA to NB. MBP and DRR were supported by NIH (RO1CA120526 and PO1CA27502) and NCI (CA52607 and CA105491) grants, respectively. We thank Karen Helm and Christine Childs for expert technical assistance in flow cytometry. We are grateful to Drs Farinaz Arbab and James Fitzpatrick for histopathological evaluation and advice. We also thank Dr Rick Sturm and Dr Graeme Walker for providing advice and helpful discussions. We are thankful to Dr Yvonne Berg for critical reading of the manuscript.