Background
The origin of intratumoral heterogeneity in melanoma is not yet clearly defined. However, studies are revealing that many cancers, including melanoma, are hierarchically organized and driven by a particular subpopulation of cancer cells that have stem cells properties, known as cancer stem cells (CSCs) [
1]. CSCs have been defined by their high self-renewal capacity and tumorigenic potential leading to tumor initiation, metastasis, therapeutic resistance and tumor recurrence. Single or multiple intra- or extra-cellular markers have been used to identify CSCs in different types of tumors [
1,
2]. In melanoma, different markers have been employed to characterize the putative melanoma CSC and, particularly, CD133, CD271, CD146, ALDH, ABCB5, ABCG2, Nestin and CD117 were considered [
3,
4]. Although the specificity of these markers are still under debate, the stem cell marker CD133 has been widely used to characterize and isolate the putative melanoma CSC in both in vitro and in vivo studies [
4,
5]. In fact, CD133 was shown to be expressed in human melanoma biopsies but hardly detected in normal skin sections [
3,
5]. Furthermore, it was revealed that CD133
+ melanoma cells have an enhanced capability to initiate primary tumors and metastasis in NOD/SCID mice, while showing higher self-renewing and migration capacity and differentiation potential into mesenchymal lineages in in vitro studies [
4,
6,
7]. Moreover, the CD133 marker has been associated with promotion of vasculogenic mimicry and angiogenesis, and was found to inhibit tumor cell apoptosis by interacting directly with the vascular endothelial growth factor (VEGF) [
8,
9]. Intriguingly, recent studies revealed that CD133+ melanoma cells are highly resistant to chemotherapy, indicating that they are also responsible for tumor recurrence [
2,
10]. Likewise, monoclonal antibodies directed against CD133 protein induced a specific dose-dependent cytotoxic effect in metastatic melanoma cells, suggesting CD133 as a potential target for immunotherapy [
11]. Recent evidence also indicated knocking down CD133 in NRASQ
61R/BRAF
WT mutant melanoma renders cells more sensitive to clinically employed-MEK/BRAF inhibitors [
12].
Beyond the notion that melanoma growth and progression is sustained by stem cell-like cancer cells, it is a well-evidenced fact that melanoma is driven by a diverse group of genetic and epigenetic lesions and signaling pathways that are imposed by change in microenvironmental conditions [
13,
14]. Nevertheless, in melanoma, as well in other type of cancers, protein coding gene mutations are no longer considered as sole drivers of the disease. For these reasons, a growing interest is now focused on the role of mobile genetic elements in tumor initiation and progression. These mobile genetic elements, primarily retroelements, comprise nearly half of the human genome. Retroelements, unless tightly regulated, may cause insertional mutagenesis or transcriptional dysregulation that potentially induces cellular transformation and tumorigenesis [
15,
16].
Human endogenous retroviruses (HERVs) belong to retroelements and are remnants of ancient retrovirus infections that actually constitute about 9% of the human genome [
16]. Most HERVs are defective but there are few active HERVs that are dynamically expressed and epigenetically regulated in a stage-dependent manner during early embryonic development. In somatic tissues, some members of HERV family remain transcriptionally active and display tissue-specific expression [
16‐
18]. So far, HERVs have been implicated in both biological and pathological processes such as in cancer, diabetes and neuropsychiatric diseases, but their specific pathophysiological roles are still poorly defined [
19‐
22]. Among HERV families, HERV-K is the youngest and most active family that maintains most of the open reading frames (ORFs) analogous to many active retroviruses [
15,
22]. HERV-K viral particles have been identified and characterized at tissue, serum and cell lines level in association with different types of tumors, including ovarian, breast and prostate cancer, teratocarcinoma, lymphomas, leukemia, sarcomas and in more recent years with melanoma [
23,
24]. The oncogenic mechanisms of HERV-K may depend from the expression of gene products potentially carcinogenic or immune escape causative, or from the regulatory role of their long terminal repeats (LTRs) sequences for the nearby (proto-) oncogenes or growth factors [
15,
16,
22].
Several studies have shown that HERV-K is aberrantly activated during melanoma progression, contributes to cell malignant transformation and promotes immune escape during metastasis formation [
23‐
28]. HERV-K
Rec and
Np9 accessory proteins, described as putative oncogenes, have been associated with carcinogenesis by interacting with proteins involved in cellular transformation [
29‐
31]. Likewise, HERV-K env protein may increase the risk of melanoma cancer by disrupting normal intracellular redox potential resulting in rise of toxic free radicals [
32]. Furthermore, HERV-K proteins have been shown to suppress the host immune system [
33,
34]. Recent studies also suggested the env protein of HERV-K might be a key mediator, at least partly, in the constitutive activation of the RAS-RAF-MEK pathway, which is aberrantly activated in over 80% of all cutaneous melanomas [
34‐
36].
Previously we demonstrated for the first time that HERV-K activation induced melanoma cell malignant transformation and reduced the immunogenicity of melanoma cells that favors tumor immune escape [
26]. Herein, we show that melanoma cells exposed to stem cell media were compelled to undergo phenotype-switching towards greater malignancy and increment of stem cell related features concomitant to HERV-K activation. These phenomena are reversible and promoted by HERV-K activation. Moreover, this study revealed that HERV-K activation is strictly required to sustain CD133+ melanoma cells with stemness features during microenvironmental modifications.
Methods
Cell lines and culture conditions
In this study the human melanoma primary tumor derived WM-115 cell line, and its metastasis derived counterpart WM-266-4, the malignant human melanoma cell lines G-361, A375 (all from ATCC, Manassas, VA, USA) and the human melanoma TVM-A12 cell line, stabilized in our laboratory, were used [
37]. TVM-A12-CD133
+ cells were sorted and isolated from TVM-A12 cell line. All cell lines were cultured as adherent cells in RPMI-1640 medium supplemented with 10% (
v/v) heat-inactivated fetal bovine serum (FBS), L-glutamine (2 mM), Penicillin-Streptomycin (100 mg/ml) at 37 °C in a humidified 5% CO
2 atmosphere and serially passaged twice weekly after detachment with 0.05% trypsin and 0.02% EDTA solution in PBS (all reagents from Sigma-Aldrich, St Louis, MO, USA). To assess cellular plasticity and malignant phenotype switching under microenvironmental modifications, cells were cultured in the serum-free stem cells medium X-VIVO™ 15 (Lonza, Verviers, Belgium) supplemented with Penicillin-Streptomycin (100 IU/ml).
Flow cytometry analysis
For cytofluorimetric analysis adherent cells were trypsinized, then washed twice in PBS, incubated with 5 μl of fluorochrome-conjugated antibodies for 30 min at 4 °C (5 × 105cells/FACS tube), fixed with 1% formaldehyde solutions for 5 min at 4 °C, centrifuged for 5 min at 1600 rpm and washed once with 1 ml of PBS for 5 min at 1600 rpm. CD133/2 (293C3) (Miltenyi Biotec, Bergisch Gladbach, Germany), HLA-I, NGF-R, ICAM-1, CD20, CXCR4, CD10, c-Kit antibodies (all from BD Biosciences, Franklin Lakes, NJ, USA) conjugated with different fluorescent dyes were used; the unconjugated antibodies Nestin (Novus Biologicals, Minneapolis, USA) and Melan-A/MART-1 (Santa Cruz Biotechnology, Dallas, TX, USA) were used in combination of the goat-anti-mouse IgG-FITC (BD Biosciences) as secondary antibody and used after cell permeabilization. For apoptosis analysis cells were trypsinized, washed in PBS, fixed with 70% ethanol for 45 min at 4 °C, washed in PBS and stained with propidium iodide (50 μg/ml diluted in PBS) and RNAase (250 μg/ml), then stored for at least 3 h at 4 °C before analysis. Flow cytometer analysis was performed by BD FACScan™ System using CellQuest Pro software on a minimum of 5000 events for each sample.
Microscopic and side population analyses
Morphological analysis was carried out by phase-contrast microscopy, using the Motic AE31 Trinocular inverted microscope (Motic Asia, Hong Kong).
The differential ability of melanoma cells to efflux the Hoechst dye [
38] was evaluated by Hoechst 33342 extrusion test. In detail, TVM-A12 cells were grown in X-VIVO medium to induce cellular aggregates. The obtained cellular aggregates were grown on coverslips at a concentration of 10
6 cells/ml in X-VIVO medium in presence of 2% FBS, stained at 37 °C for 90 min with 4 μg/ml Hoechst 33342 (Sigma-Aldrich), and analyzed without fixing under the ZEISS Axioplan fluorescence microscope (Oberkochen, Germania) equipped with a digital camera.
For side population (SP) analysis by flow cytometry, we adopted the original SP protocol [
39] and performed the analysis as described before [
40]. Briefly, TVM-A12 cells cultured in RPMI 10% FBS medium or in X-VIVO for 72 h were adjusted to 10
6 cells/ml concentration and incubated with 5 μg/ml Hoechst 33342 nucleic acid stain for 90 min at 37 °C. To reduces efflux of the Hoechst dye and confirm the SP phenotype, Verapamil (50 μM; Sigma-Aldrich) was added 30 min before Hoechst stain in control samples. For further SP characterization, after Hoechst staining, cells were maintained in ice and immunostained with CD133 or isotype antibodies (Miltenyi Biotec). Finally, cells were resuspended in ice-cold staining buffer (PBS + 2%FBS) for subsequent analysis by flow cytometry using a FACSCanto II (BD Biosciences).
Flow cytometry sorting
For fluorescence-activated cell sorting, TVM-A12 cells, cultivated in X-VIVO, were enzymatically dissociated and stained by incubation with PE-conjugated monoclonal antibody against CD133 and/or PE- isotype mouse IgG2b at 4 °C for 1 h. After staining, samples were washed twice with PBS, resuspended at 2 × 106 cells/ml in PBS and filtered (50 μm, Partech). Live cell sorting experiments were performed using BD FACSAria II (Becton Dickinson Immunocytometry Systems) with 130 μm nozzle and sort gates were defined on a dot plot of CD133 (PE). PE fluorescence of CD133 was determined by a 488 nm excitation line and detected by 585/42 nm filters. Sorted cells were collected in PBS medium. The samples were analyzed using the FACSDiva software (Becton Dickinson).
A total of 500 single viable cells (TVM-A12 and TVM-A12-CD133+) were seeded into 48-well tissue culture plates coated with 0.5 mg/ml of poly-2-hydroxyethyl methacrylate (Poly-HEMA, Sigma) in a 500 μl of serum-free DMEM/F12 (1:1) (Sigma) basal medium supplemented with L-glutamine (2 mM), Penicillin-Streptomycin (100 mg/ml), 20 ng/ml human epidermal growth factor (EGF) 20 ng/ml, human fibroblast growth factor-2 (FGF-2) (ProSpec, Rehovot, Israel), and 1:50 B-27 supplement (Gibco, Life Technologies, Carlsbad, CA, USA) and cultured at 37 °C in a humidified 5% CO2 for 10 days to form melanospheres. For serial passage, these melanospheres were counted using a manually prepared “quadrant grid” under microscopic observation, harvested and centrifuged at 1000 rpm for 5 min, trypsinized to dissociate in to single cell, counted and viable cells reseeded in the Poly-HEMA coated 48-well plates for subsequent passages.
Migration and invasion assays
For migration assay, cells were maintained in serum-free RPMI-1640 medium for 18 h, harvested and resuspended in the same medium, and seeded into Bio-Coat cell migration chambers with 8 μm membrane pore sizes (BD Biosciences) at 1 × 105 cells in 250 μl per chamber. The chambers were then inserted into the wells of a 24-well plate containing 750 μl of RPMI-1640 medium with 20% FBS alone or 20% FBS with 40 ng/ml human hepatocytes growth factor (HGF) (ImmunoTools, Friesoythe, Germany) and incubated at 37 °C with 5% CO2 for 48 h. After this period, the cells remaining on the upper surface of the membrane were removed with cotton swab and the cells adhered to the lower surface were fixed with 70% ethanol for 15 min, stained with Giemsa (Sigma) for further 15 min, then washed twice with water. After drying, the membrane was removed with surgical blade and mounted on glass slides. Pictures were taken with Olympus BX50 microscope from five random microscopic fields at 200x magnification, and cells counted with image analysis software (ImageJ, NIH). For invasion assay, the same procedure was followed except for the coating of the migration chambers with 50 μl (0.3 mg/ml) of Matrigel matrix (BD Biosciences) before of the cells seeding.
RNA extraction and real-time PCR
Total cellular RNA was extracted using NucleoSpin RNA II kit (Machery-Nagel GmbH & Co. KG, Düren, Germany) following the manufacturer’s instructions. Reverse transcription was performed using the ImProm-II™ Reverse Transcription System Kit (Promega, Madison, USA) following the manufacturer’s instructions. The Real-time PCR gene-specific primers for HERV-K env, CD133, and the housekeeping gene beta-glucuronidase (GUSB) were purchased from Invitrogen Thermo Fisher Scientific (Waltham MA, USA). The PCR primers for each specific genes were as follows: env forward, 5′-GCCATCCACCAAGAAAGCA-3′; env reverse, 5′-AACTGCGTCAGCTCTTTAGTTGT-3′ (AF164614); CD133 forward, 5′-TTTCAAGGACTTGCGAACTCTCTT-3′; CD133 reverse, 5′-GAACAGGGATGATGTTGGGTCTCA-3′ (NM_001145848.1); Oct4 forward, 5′-TATGCAAAGCAGAAACCCTCGTGC-3′; Oct4 reverse, 5′-TTCGGGCACTGCAGGAACAAATTC-3′ (NM_002701); Nanog forward, 5′-TCCAGCAGATGCAAGAACTCTCCA-3′; Nanog reverse, 5′-CACACCATTGCTATTCTTCGGCCA-3′ (NM_024865); GUSB forward, 5′-CAGTTCCCTCCAGCTTCAATG-3′; GUSB reverse, 5′-ACCCAGCCGACAAAATGC-3′ (NM_000181). Real-time PCR was performed in CFX96 Real-time System using SsoAdvanced™ Universal SYBR Green Supermix (Bio-Rad Laboratories, Hercules, CA USA) amplification detection method. Samples were analyzed in duplicate and for each experiment no template controls (NTC) and GUSB were used as internal control and to determine the amplification efficiency, respectively. The threshold cycle (Ct) comparative method was used to analyze the relative changes in gene expression of each sample compared with the reference sample, calculating as follows:2−[ΔCt(sample) − ΔCt(calibrator)] = 2−ΔΔCt, where ΔCt (sample) = [Ct (HERV-Kenv) – Ct (GUSB)], and ΔCt (calibrator) was the mean of ΔCt of TVM-A12 cells maintained in RPMI 10%FBS.
RNA interference
HERV-K (AF164614) expression was down-regulated by infecting TVM-A12 cells with a retroviral vector as previously described [
26,
27]. Briefly, Phoenix packaging cells were transfected with 9 μg of construct DNA using lipofectamine (Gibco-BRL, 18324–012). 48 h later, cell culture media were collected and filtered through 0.45 μm Millipore filter and used to infect TVM-A12 cells in the presence of 4 μg/ml polybrene (Sigma) for 12 h, after which the medium was changed. Infected cells were selected on 0.5 μg/ml puromycin (Sigma) for 6 days. After selection, the cells were grown in RPMI-1640 medium supplemented with 10% FBS and maintained under antibiotic selection (0.2 μg/ml).
Reverse transcriptase inhibitors treatments
TVM-A12 and TVM-A12-CD133 cells were cultured for 24 h in RPMI-1640 medium with 10% FBS at 37 °C and 5% CO2 to ensure cells form a monolayer before starting treatment. Then, cells were cultivated in RPMI-1640 with 10% FBS, or in serum-free stem cell media (X-VIVO™ 15) to induce the grape-like cellular aggregates formation. The non-nucleoside reverse transcriptase (RT) inhibitors (NNRTIs) nevirapine (NVP) or efavirenz (EFV) were added to the cultures at 350 μM and 15 μM respectively. Dimethyl sulfoxide (DMSO) was used as diluent for the drugs and was referred as vehicle in control condition. After 72 h of treatment cells were detached (trypsin used only for adherent cells) and back cultured with the same fresh media for the next 24 h and retreated with NVP or EFV at half concentration of initial used. Finally cells from each condition were collected and processed for Real-time PCR and flow cytometry analysis.
Statistical analysis
Data analysis was performed using the SPSS statistical software system (version 17). Comparison of means was carried out using Bonferroni post-hoc multiple comparison Anova test. Statistical probabilities were expressed as p ≤0.050 (*) or p < 0.001 (**).
Discussion
In different types of tumors CSCs have been shown to possess an inherent characteristic of cellular plasticity to resist and survive in harsh microenvironment conditions, encountered during tumor progression [
2]. In melanoma, CSCs plasticity has been related to intratumoral heterogeneity, metastasis formation, resistance to chemotherapy and tumor recurrence [
7].
In the present study TVM-A12 human melanoma cell line, originated from a primary lesion, showed peculiar plasticity features upon the modification of microenvironment by cultivation in a stem cell medium, switching from an adherent phenotype towards an anchorage-independent grape-like cellular aggregates formation. This morphological transition was accompanied by the acquisition of more aggressive and invasive characteristics such as significantly reduced expression of cell surface markers HLA-I, MelanA/MART-1 and ICAM-1, all involved in the host immune response [
42‐
44]. This is in agreement with our previous paper, in which we described the contribution of HERV-K in reducing the immunogenicity of melanoma cells [
26] and therefore favoring the evasion of the tumour from the host immune responses [
1]. In addition, herein increased expression of stem cells and metastasis related markers CD133, nestin, CXCR4, and NGF-R along with continued high-level of CD10 expression were observed, which usually delineates progression to advanced stage of melanoma and metastatic tumour formation [
4,
45]. Moreover, the survival of TVM-A12 cells in non-adherent condition implied these cells may have developed a mechanism to resist and thereby survive after detachment. Likewise, metastatic cancer cells are able to survive after detachment from their primary site by developing resistance to a specialized form of cell death called anoikis [
46,
47].
It was described that HERV-K expression may support cancer cells to go through fast gene plasticity to adopt new microenviromental changes [
16,
48]. Hence, the phenotype-switching of TVM-A12 cells, occurring in the stem cell medium, and the concomitant HERV-K activation, both reversible after serum addition, highlight that TVM-A12 cells harbor a high capacity to adapt to the microenvironment modifications, while maintaining an inherent ductility to reestablish the original features. Intriguingly, silencing HERV-K in TVM-A12 cells attenuated their plasticity towards invasive malignant phenotype, which seems to affirm that in melanoma cells the aberrant activation of HERV-K is associated with the process of invasive phenotype-switching, occurring in response to microenvironmental modification. This finding corroborate previous reports demonstrating melanoma cells undergo the process of phenotype-switching to promote metastatic state while maintaining the stem cell-like identity [
14,
49].
In this study the expansion of the CD133+ subpopulation and the concomitant up-regulation of HERV-K were observed during the morphological transition of TVM-A12 cells towards the non-adherent malignant phenotype. Interestingly, we found that these phenomena are not limited to TVM-A12 melanoma cell line, rather they are seemingly common traits shown with different extent by other type of commercially available melanoma cell lines. Indeed, all the cell lines tested showed various basal expressions of CD133 marker and aptitude to expand this subpopulation, concomitantly with HERV-K activation, when maintained in stem cell medium. Remarkably, blocking of HERV-K expression through siRNA significantly abolished the expansion and maintenance of CD133+ subpopulation within TVM-A12 cells in the stem cell medium. It is also well documented that the main defining characteristics of putative CSCs are self-renewing and metastatic potential [
50] as well as the expression of pluripotent stem cells transcription factors, such as Nanog, Oct4, and Sox2 [
51]. From this perspective, this study showed that the TVM-A12-CD133+ cells, which are dependent on HERV-K activation, are endowed with the defining characteristics of CSCs as evidenced by their enhanced self-renewal ability, higher migratory and invasive capacities and expression of the stem cell markers Oct4 and Nanog. This study also showed CD133+ subpopulation is partly present in the Hoechst dye-effluxing SP cells, which possess many CSCs properties [
40]. These findings seems to indicate that HERV-K is also involved in the stemness genomic networks of the CSCs model in melanoma tumor. In support to this hypothesis, compelling evidences suggest that expression of HERVs have a direct participatory role for the maintenance of human embryonic stem cells (hESCs) and induced pluripotent stem cells (iPSCs), and, their activation or re-activation have been evinced as a marker of pluripotency [
16,
31]. Likewise, HERV-K has been associated with cellular transformation and carcinogenesis [
16,
23,
25]. Of note, very recently has been demonstrated the essential role of HERV-K env protein activation for tumorigenesis and metastasis of breast cancer cells [
52]. In this regard, our study presents the first evidence that demonstrate the fundamental role of HERV-K in the expansion and maintenance of CSCs in melanoma tumor during phenotype switching under microenvironment modifications.
It has been shown that microenvironmental stress conditions influence the expression of HERVs in cancer cells through epigenetic modification, most likely through DNA hypomethylation of relevant retroviral genes [
23]. Moreover, it was reported that DNA hypomethylation is accounted as an important determining factor of CD133 expression in ovarian and glioma cancer cells [
53,
54]. Global DNA hypomethylation is a hallmark of many cancers [
55] and HERV-K (HML-2) hypomethylation in particular has been reported in melanoma cell lines [
56]. Hence, the aberrant activation of HERV-K in melanoma cells could depend on the epigenetic modification induced by change in tumor microenvironment that affects the DNA methylation state. Taken together, this study revealed that the expansion of CD133+ melanoma cells with stemness features are dependent on HERV-K activation during microenvironment modifications, likely mediated by stress condition.
We previously reported that inhibition of retroelements-encoded RT by NNRTIs or by RNA interference, resulted in anti-proliferative, pro-differentiating effects in melanoma cells and reduction of melanoma tumor growth in mice models [
27,
41,
57]. Interestingly, here we found that efavirenz and nevirapine were able to halt the expansion and maintenance of CD133+ melanoma cells inducing high levels of apoptosis and were effective to restrain the activation of HERV-K during microenvironmental modification. Therefore, the high sensitivity of TVM-A12-CD133+ cells to NNRTIs suggests a specific requirement of HERV-K expression to sustain this subpopulation in melanoma cells during microenvironmental modifications. This result confirms our previous findings that retroelements are key players in melanoma tumor progression [
41,
57‐
59], demonstrating for the first time the specific role of HERV-K in the generation and survival of CD133+ melanoma CSCs. These findings open new perspectives for targeting the putative CSCs that express high-level of HERV-K and for developing new possible therapeutic strategies based on tumor cells differentiation.
Acknowledgments
We express our sincere gratitude to Dr. Corrado Spadafora for the helpful discussion and comments. We also thank Dr. Martino Tony Miele for his assistance in English language editing.