Development of a 3D Skin Model for Studying Melanoma Progression

Abstract

Despite advances in the treatment of cutaneous melanoma, there is still a high percentage of patients who fail to respond or develop resistance to treatment. Establishing robust in vitro melanoma models will enable mechanism-based drug screening while reducing animal testing. In this work, a three-dimensional (3D) melanoma skin model (3DMSM) was developed on a porous scaffold. The culture of three melanoma cell lines (SKMEL-1, A375, and G361) in co-culture with human fibroblasts, melanocytes, and keratinocytes allowed the formation of the dermis, and stratified epidermis. Tumors were established in this model using two methodologies: adding previously formed melanoma cell aggregates (CA) or seeding melanoma cells directly into the dermis (CD). In this model, melanoma cells remain in their original microenvironment and, after proliferation, invade the basal layer. The model recapitulates correct melanocyte localization, epidermal disruption, extracellular matrix (ECM) remodeling, including collagen deposition, and epithelial-to-mesenchymal transition (EMT). Additionally, the cytokine profiles studied indicate that the model could mirror the inflammatory and immune-evasive traits of melanoma. Overall, 3DMSM provides a useful tool for understanding the mechanisms of melanoma progression and invasion, and for developing personalized medicine strategies through the implementation of a patient-derived model.

1. Introduction

Human skin is a complex organ that comprises three main layers (the epidermis, dermis, and hypodermis) and a diverse array of specialized cells, each contributing to its protective, sensory, and regulatory functions. In the epidermis, keratinocytes account for over 80% of epidermal cells [1]. They originate in the basal layer and migrate upward, synthesizing keratin, a structural protein that fortifies the skin barrier against trauma, dehydration, and pathogens. Another important cell type in the epidermis is the melanocyte, which produces melanin and transfers it to neighboring keratinocytes, where it protects against ultraviolet (UV) radiation [2,3]. Melanocyte proliferation and spread can lead to the formation of benign clusters (nevi), which can serve as precursors to melanoma. Although melanocyte proliferation and clustering can give rise to benign nevi, it is now well established that only 20–40% of melanomas originate from pre-existing nevi. The majority of melanomas arise de novo from clinically inapparent or histologically subtle precursor lesions. This distinction is important for understanding melanocyte behavior and melanoma initiation [4]. Mutations in melanocyte’s critical growth regulatory genes, the loss of adhesion receptors and the production of autocrine growth factors allow melanocytes to escape from the tight regulation of keratinocytes [3]. Nevi can also undergo local microinvasion of dermis and progress into the radial growth-phase (RGP) melanoma. These RGP melanoma cells can progress to vertical-growth phase (VGP), which is characterized by the metastatic potential of these cells to invade the dermis and consequently migrate and invade other organs, advancing to metastatic malignant melanoma [5]. Not all melanomas transit to these individual phases, since both RGP and VGP can be developed directly from isolated melanocytes or nevi, and both can progress directly to metastatic malignant melanoma, highlighting the complexity of these tumors [6].

Although it is not the most prevalent type of skin cancer, cutaneous melanoma is the deadliest form- due to its high metastatic potential, accounting for over 330,000 new cases every year (according to World Health Organization) [2]. Regarding treatment, localized melanomas can be surgically removed. However, for metastatic melanomas, while targeted therapies against specific signaling cascades and immunotherapies have shown remarkable clinical responses, a significant number of treated patients fail to respond to therapies or later develop resistance [4,7].

In cancer research, therapy development presents a low rate of reproducibility when comparing results obtained from animal models versus patients [8]. To overcome this is-sue, reduce the use of animals in research, and support the precision medicine approach to human cutaneous melanoma formation and treatments, improved systems are necessary. During the past few years human cell-based cancer cultures as monolayer or three-dimensional (3D) models, such as spheroids, organoids or microtissues, have been used for high-throughput screening [9,10,11]. These types of models are very important for the first steps of cancer research, by helping to understand the cells’ behavior, potential inhibitors, and serve as platforms for drug tests. However, they present limitations re-producing tumor microenvironment (TME), essential to understanding cancer growth and spreading. TME is a heterogeneous network of cellular and acellular components, including endothelial cells, immune cells, and cancer-associated fibroblasts (CAFs), that are critical for producing the extracellular matrix (ECM), which provides a highly supportive scaffold [12].

Recently, tissue-engineered skin models that allow the production of 3D tissues and better reproduce TME have been developed [13,14]. Most of them use synthetic or animal-derived scaffolds to mimic ECM, as collagens from rodents or porcine sources [13,14]. Therefore, a model that recreates the natural human cutaneous melanoma microenvironment using human stromal cells to produce ECM, allowing cell-ECM matrix interactions and the secretion of important factors as cytokines, is fundamental to explore cutaneous melanoma progression.

Here we aimed to develop an in vitro cutaneous melanoma model, that allows the investigation of the early melanoma invasion with vertical and horizontal growth phases. The seeding of melanoma cells onto the dermis allows these cells to remain in their original micro-environment and invasion of the basement membrane occurs only after proliferation at the dermis-epidermis junction. Our 3D melanoma skin model (3DMSM) has the advantage of being constructed only with human cell lines: human fibroblast that ensure a human derived-ECM, and establishing the dermis; three different human melanoma cell lines; human melanocytes and human keratinocytes, allowing the epidermis establishment. In addition, it can be developed in a shorter culture time and at lower costs, when compared to other methods [15]. Importantly, by direct comparison with patient-derived melanoma sections, we showed that this model precisely mimics the melanoma TME. Altogether, this model will provide a useful tool to allow a better understanding of the molecular mechanisms affecting melanoma progression and invasion, improving the knowledge of melanoma aggressiveness, which could reveal new potential druggable targets. At the same time the 3DMSM offers a platform for testing new therapeutic approaches, contributing for a patient specific care.

2. Materials and Methods

2.1. Cell Lines

Three melanoma cell lines (A375, G361, and SKMEL-1) from the American Type Culture Collection (ATCC) panel (ATCC^® TCP-1013™, Manassas, VA, EUA) were used for the establishment of the 3DMSM. These three melanoma cell lines harbor the BRAF V600E oncogenic driver mutation. However, they differ in additional genomic mutations (according to the Sanger COSMIC database) and cellular phenotypes.

The primary cells used for the establishment of the 3DMSM include neonatal human dermal fibroblasts (HDFn, CellnTec; Stauffcherstr, Switzerland), neonatal human epidermal melanocytes darkly pigmented (HEM-DP, GIBCO ThermoFisher Scientific, Waltham, MA, USA), and neonatal human epidermal keratinocytes (HEKn, GIBCO ThermoFisher Scientific, Waltham, MA, USA). All the experiments for the reconstructed human skin model were performed only with primary cells.

2.2. Cell Culture Growth

HDFn and three melanoma lines (A375, SKMEL-1 and G361) were cultured in high glucose Dulbecco’s modified eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin solution (Pen-Strep). HDFn, A375, and G361 are adherent cell lines, and the SKMEL-1 cell line was the only one cultured in suspension.

HEM-DP were cultured in medium 254 (M254) supplemented with 1% human melanocyte growth supplement (HMGS) and 1% of an antibiotic mixture: penicillin (10,000 U/mL) and streptomycin (10 mg/mL). HEKn were cultured in EpiLife^TM medium (Gibco, Thermo Fisher Scientific, Waltham, MA, USA), supplemented with 0.06 mM CaCl₂, 1% human keratinocyte growth supplement (HKGS) and 1% pen-strep. HDFn, HEM-DP, and HEKn were used until passage 13, 20 and 3, respectively. When a confluency of 70% was reached, cells were washed with 1X phosphate-buffer saline (PBS), pH 7.4 (Gibco, Thermo Fisher Scientific, Waltham, MA, USA). HDFn, melanoma and HEK cells were detached with Trypsin-Ethylenediaminetraacetic Acid (EDTA) (0.05%) (Trypsin-Ethylenediaminetraacetic Acid (EDTA) (0.05%)), and HEM-DP cells were detached with TrypLE^TM (Gibco, Thermo Fisher Scientific, Waltham, MA, USA). The different cell lines were maintained at 37 °C in a humidified atmosphere with 5% CO₂. All previously mentioned reagents for cell growth were supplied by Gibco/Thermo Fisher Scientific, MA, USA.

Cell numbers and viability were measured using Trypan blue (Sigma Aldrich, St. Louis, MO, USA) dye exclusion with a Burker counting chamber (VWR, Radnor, PA, USA).

2.3. 3D Melanoma Skin Model

The 3DMSM culture comprised three main developmental phases: (i) formation of a dermal equivalent layer (day 0–7), (ii) submerged culture of dermis and epidermis with fibroblasts, melanocytes and melanoma cells (days 7–12) and (iii) the culture of the different layers at the air-liquid interface (ALI) were the stratified layers are now exposed to the air and culture media is only provided by the bottom of the scaffold, to mimic the in vivo environment. (days 12–25), as shown in Figure 1. The human skin equivalent was prepared on a 12-well format Alvetex^® inert porous polystyrene scaffolds (REPROCELL Europe Ltd., Sedgefield, UK) with void sizes of 33–55 mm and 90% porosity, previously treated with 70% ethanol and a solution of PBS in a 6-well plate (VWR, Radnor, PA, USA).

An amount of 1.0 × 10⁶ HDFn cells were seeded onto the scaffold in 100 μL of fibroblast growth media A (

Table 1. Media components for the development of 3D melanoma skin model.

Media	Components	Concentration
A	High glucose Dulbecco’s Modified Eagle medium Fetal Bovine Serum Penicillin-streptomycin solution	- 10% 1%
B	Media A 2-Phospho-L-ascorbic acid trisodium salt (Vit. C, Sigma, 49752)	- 100 μg/mL
C	Medium 254 Human Melanocyte Growth Supplement Penicillin-streptomycin solution	- 1% 1%
D	Media A Media C	50% 50%
E	EpiLifeTM medium Human Keratinocyte Growth Supplement Penicillin-streptomycin solution Calcium chloride	- 1% 1% 0.06 mM
F	EpiLifeTM medium Human Keratinocyte Growth Supplement Penicillin-streptomycin solution Calcium chloride	- 1% 1% 1.5 mM
G	Media F Media A Media C	60% 20% 20%
H	Media G Recombinant human keratinocyte growth factor (BioLegend, FGF-7) 2-Phospho-L-ascorbic acid trisodium salt	- 0.01% 50 μg/mL

) and incubated at 37 °C in a humidified atmosphere of 5% CO₂ for 1.5 h and afterwards added 9 mL of media B (Table 1) to flood the insert before incubation for 7 days. The media was changed every 2 days to allow the formation of the dermis. On day 7, the epidermis formation began with the addition of 5.0 × 10⁴ HEM-DP cells in 100 μL media C (Table 1) onto the scaffold. After 3 h of incubation at 37 °C and 5% CO₂, 9 mL media C (Table 1) were applied to flood the insert. After 24 h (day 8), the melanoma cells (SKMEL-1, A375 or G361) were added onto the scaffold by seeding previously formed cell aggregates (1) or adding melanoma cells directly (2) on the scaffold.

1.

To induce the formation of melanoma cell aggregates, depending on the cell line, the following were plated: (i) SKMEL-1—5 × 10², 7.5 × 10², 1 × 10³, and 1.5 × 10³, (ii) A375—5 × 10¹, 1 × 10² and 2.5 × 10² and (iii) G361—2.5 × 10², 5 × 10² and 7.5 × 10² with 100 uL media A in round-bottomed, ultra-low-attachment 96-well microplate (Costar, Corning, VA, USA). The plate was then centrifuged at 250 g for 3 min and incubated for five days at 37 °C. After that time, four aggregates were added to each scaffold, and the plate was incubated for 3 h at 37 °C. Then, 9 mL of media D (Table 1) was added to each well.

2.

For the assays that include cells seeded directly in the dermis, four different numbers of melanoma cells were used: 2.5 × 10³, 5.0 × 10³, 5.0 × 10⁴ and 1 × 10⁵ cells/mL wit exception for SKMEL-1 where 1 × 10⁵ cells/mL was not used.

On day 9. 5 × 10⁵ HEKn cells were applied over the dermis in 100 μL media F (Table 1) and incubated at 37 °C for 3 h. After this, 9 mL of media G (Table 1) was applied to each well to flood the insert before further incubation at 37 °C for 3 days, replacing the medium every day. On day 12 the medium in each well was removed and the Air Liquid Interphase (ALI) begins; 4 mL of media H (Table 1) were applied only in the outer side of the scaffold. The media H was replaced every 2 days for 14 days. After a total of 25 days of culture, the 3DMSM was obtained.

2.4. Cell-Viability Assay of Melanoma Cells Aggregates

The cell viability of melanoma cells aggregates was assayed using the cell counting kit-8 (CCK8, Tebu-bio, Le Perray-en-Yvelines, France). The assay was performed 5 days after seeding the melanoma cell lines into 96-well, round-bottom, ultra-low-attachment microplates (Costar, Corning, VA, USA). To each well, 10 μL of CCK8 was added and the plate was incubated in a humidified incubator at 37 °C with 5% CO₂. After approximately 20 min, the absorbance was measured at 450 nm using a microplate reader (SpectraMax 340, Molecular Device LLC, San Jose, CA, USA).

2.5. 3DMSM Histological Analysis

After 25 days of culture, 3DMSM samples were fixed immediately in 10% neutral buffered formalin (Bio-Optica, Milan, Italy) for a minimum of 24 h at room temperature, before paraffin embedding (Gulbenkian Institute for Molecular Medicine, Oeiras, Portugal). Sectioning was performed microtome (Leica RM2155; Leica Biosystems, Deer Park, IL, USA) to obtain 3–5 μm thick tissue sections that were used for different applications.

2.5.1. Hematoxylin and Eosin (HE) and Masson’s Trichrome Staining

HE staining was performed in the samples at the Pathological Anatomy facility of the Instituto Português de Oncologia de Lisboa Francisco Gentil E.P.E. (IPOLFG) using a standard protocol [16]. Images of the stained samples were acquired with a digital microimaging device (Leica DMD108, Leica Microsystems GmbH, Wetzlar, Germany). Masson’s trichrome is a fast and effective method for detecting morphological changes associated with fibrosis, scarring and dermal structure disorders in skin, which was carried out according to the manufacturer’s instructions [17].

2.5.2. Immunohistochemistry

Immunohistochemistry (IHC) was performed in an automated platform BenchMark ULTRA system (Ventana Medical Systems (Roche), Tucson, AZ, USA) using the detection kits UltraView and OptiView Universal DAB. For the IHC technique, 3–4 µm thick paraffin sections were obtained and placed on TOMO microscope adhesion slides (Matsunami Glass Ind., Ltd., Kishiwada, Osaka, Japan). The target temperature for the sections to adhere to the slide was 60 s, at 62 °C. The slides were visualized with DAB chromogen and nuclei were counterstained with hematoxylin and mounted in a suitable mounting medium to obtain a final mount. The automated BenchMark system puts the slides through a series of user-defined deparaffinization and antigen retrieval steps prior to antibody staining. The antibodies used are listed in

Table 2. List of antibodies.

Reagent or Resource	Source
E-cadherin (36 Ms monoclonal) Pre-diluted	Ventana Medical Systems (Roche), Tucson, AZ, USA
Vimentin V9 (36 Ms monoclonal) Pre-diluted	Ventana Medical Systems (Roche), Tucson, AZ, USA
SOX-10 (SP267 Rb monoclonal) Pre-diluted	Cell Marque Corporation, Rocklin, CA, USA
Collagen type I (COL1A1) (E8F4L Rb monoclonal) 1:100	Cell Signaling Technology, Danvers, MA, USA
KI-67 (30-9 Rb monoclonal)	Ventana Medical Systems (Roche), Tucson, AZ, USA
S100 Pre-diluted	Ventana Medical Systems (Roche), Tucson, AZ, USA
Tyrosinase	Ventana Medical Systems (Roche), Tucson, AZ, USA
Fibronectin (1:100)	Sigma Aldrich, St. Louis, MO, USA
Anti-involucrin (1:200)	Sigma Aldrich, St. Louis, MO, USA
Anti-cytokeratin 10 (1:10)	PROGEN Biotechnik GmbH, Heidelberg, Germany
Anti-cytokeratin 14 (1:1000)	Abcam, Cambridge, United Kingdom
Anti-cytokeratin 15 (1:100)	Sigma Aldrich, St. Louis, MO, USA
Anti-BRAF (1:400)	Roche, Basel, Switzerland
Anti-Rabbit IgG F(ab) A488 (1:500)	Invitrogen, Carlsbad, CA, USA
FITC Goat Anti-Mouse IgG (1:200)	Sigma Aldrich, St. Louis, MO, USA
Wheat Germ Agglutinin Conjugate (WGA)	Invitrogen, CA, USA

. For human melanoma samples, the selection of the paraffin blocks of interest was made by observing the respective HE slides, under the optical microscope, with the help of an expert pathologist. The samples used in this project were obtained from Egas Moniz Hospital at PD ULSLO and were approved by “Comissão Ética para a Saúde” of the Hospital.

2.5.3. Immunofluorescence

For immunofluorescence analysis, 5 μm thick paraffin 3DMSM sections mounted on slides were deparaffinized using HistoChoice^® (Sigma Aldrich, St. Louis, MO, USA) incubated in a dry chamber at 65 °C for 30 min or xylene for 15 min at room temperature. Then, 3DMSM sections were hydrated for 5 min in a gradient of each ethanol solution (100%, 96% and 70%). The antigen retrieval was performed in citrate pH 6 for 5 min in the microwave with 1 min of cooling, another 3 min in the microwave and finally 20 min of rest for cooling. Sections were washed with a solution of 1x PBS three times for 10 min and blocked with PBS-BSA 1% for 15 min. Immunofluorescence staining of sections was performed using the primary and secondary antibodies described on the Table 2. Primary antibodies were diluted in a solution with PBS-BSA 1% or diluted in goat serum in TBST 1:50 and incubated overnight for detection, secondary antibodies were diluted in goat serum in TBST 1:50 for 2–4 h at room temperature. Nuclei were stained with 1 μg/mL of 4,6-diamidino-2-Phenylindole (DAPI; Invitrogen (Thermo Fisher Scientific), Carlsbad, CA, USA) and slides were mounted with Vectashield (Ventana Medical Systems (Roche), Tucson, AZ, USA). Raw images were obtained using a fluorescence microscope (Nikon Eclipse TE2000-S, Nikon Corporation, Tokyo, Japan) and they were merged with the ImageJ software version 1.54f (National Institutes of Health, Bethesda, MD, USA).

2.5.4. Measurement of Cytokine Secretion in Culture Supernatant

Cytokine secretion was measured during melanoma growth in 3DMSM generation in culture supernatants. For the immunological assay, the samples were collected at various time points throughout the 3DMSM development to monitor cytokine levels in the media, as depicted in Figure 2.

The first sample was collected prior to HEM seeding (day 0). The second sample was obtained before A375 seeding (day 1). The third sample was taken before HEKn seeding (day 2). On days 3 and 4, two media exchanges were conducted, and the used media was collected. The fifth sample was taken before the initiation of the ALI phase (day 5). Samples were collected before media exchanges on days 7, 9, 11, 12, 14, and 16 throughout the ALI phase.

The enzyme-linked immunosorbent assays (ELISA) for the detection of IL-6, IL-8 and IL-10 were performed following the manufacturer’s protocol (BioLegend, San Diego, CA, USA). Supernatants from at least three independent experiments were used for ELISA. The absorbance was read at 450 and 570 nm within 30 min using a multi-well plate reader (SpectraMax 340PC Microplate Reader, Molecular Devices, LLC, San Jose, CA, USA).

2.6. Statistical Analysis

Statistical analyses were performed using GraphPad Prism (version 10.6.1; GraphPad Software, San Diego, CA, USA). Data are presented as mean ± standard deviation (SD) unless otherwise indicated. Comparisons between two independent groups were conducted using an unpaired two-tailed t-test. A p-value < 0.05 was considered statistically significant.

3. Results

3.1. Development and Characterization of Melanoma Skin Model (3DMSM)

3.1.1. In Vitro 3D Full Thickness Human Skin Model (3DFTSM)

To generate an in vitro 3DMSM, melanocytes were introduced into a full-thickness 3D skin model (3DFTSM) derived from primary human cells, following a previously published protocol with adaptations (Figure 3) [18]. Melanocytes were located along the basement membrane, reflecting normal human skin architecture. Figure 3 represents hematoxylin-eosin staining of the 3DFTSM. Immunohistochemistry for tyrosinase and S100 confirmed the presence of melanocytes (Figure 3). Furthermore, Ki67 staining, a marker of cellular proliferation, evidenced active proliferation within the basal layer of the epidermis at this stage (Figure 3). This observation demonstrates cellular turnover consistent with healthy epidermal maintenance, with the generated 3DFTSM serving as a reference for comparison with the developed 3DMSM.

3.1.2. Optimization of the Melanoma Cells’ Integration into the Skin Microenvironment

A 3DMSM was generated by incorporating melanoma cells into an in vitro 3D FTS model previously described (Section 3.1). Three melanoma cell lines (SKMEL-1, A375, and G361) were used, and various cell concentrations were evaluated to better replicate in vivo conditions. To better mimic the in vivo tumor progression, melanoma cells were introduced into a 3D FTS using two distinct approaches:

(1)

Pre-formed cell aggregates were added to the previously established 3DFTSM (3DMSM-CA; CA- cells aggregates). To determine the optimal initial number of tumor cells for aggregate formation in 3DMSM-CA, various cell concentrations were tested across cell lines. Cells were cultured for 5 days as described in the literature [8] (Figure 4A–C). Viability was assessed using the CCK8 assay and showed that aggregates were viable under all tested conditions (Figure 4D–F). Increased absorbance, reflecting higher proliferation, was seen in larger aggregates of SKMEL-1 and A375 (Figure 4D,E), while only minor differences were noted for G361 (Figure 4C). Aggregates formed under different conditions were then added to the 3DMSM to evaluate tumor growth by aggregate size and proliferation capacity.

SKMEL-1 Cell Lines Capacity to Form 3DMSM

Cell aggregates of SKMEL-1 were generated using an initial number of cells of 5.0 × 10², 7.5 × 10², 1.0 × 10³ and 1.5 × 10³ cells at day 0 (Figure 4A). These aggregates were then transferred into the 3D skin model for the establishment of 3DMSM models (Figure 5). In 3DMSM-CA, tumors were formed in 2 out of 4 conditions (Figure 5A). No correlation was observed between the number of added melanoma cells and tumor formation. Tumors were detected only with 5.0 × 10² and 1.0 × 10³ cells as indicated by red arrows in Figure 5A. This was further supported by regions that were simultaneously positive for Ki67, a marker commonly used to evaluate tumor cells proliferation, and S100, a classical melanoma marker (Figure 5B). The Ki67 marker, a nuclear protein required for cellular proliferation, is generally accepted as the most reliable marker of proliferating cells and is expressed in all phases of the cell cycle, except G0 [19]. Ki67 labeling is restricted to the basal and parabasal layers. Anti-S100 protein stains all components of benign compound nevi [20]. Tyrosinase is a copper-containing oxidase that catalyzes the formation of pigments such as melanin and is found in the melanosome membrane [21]. In the 3DMSM-CA, melanocytes were aligned in the basal layer, as observed by tyrosinase staining. However, tumor formation resulted in a loss of this melanocyte organization, as well as a decrease in tyrosinase activity and melanin production above the tumor region (Figure 5B) [22].

Representative images of the models obtained by seeding 2.5 × 10³, 5.0 × 10³, 5.0 × 10⁴ SKMEL-1 melanoma cell lines directly on the scaffold (3DMSM-CD) are shown in Figure 5C,D. In these conditions, tumor formation was not achieved.

The number of tumors was counted in each 3DMSM sample and divided into two categories: tumors only present in the epidermis and tumors present in both the dermis and the epidermis (Figure 6A). Similar to the histological analysis (Figure 5A), there was no correlation between the increase in the number of tumors and the number of added melanoma cells (Figure 6A) or the depth of the tumor formed and the number of cells either added directly (Figure 6(B2,C2)) or as an aggregate (Figure 6(C2)). Although the tumor depth does not show a significant difference between the two populations of aggregates seeding (5 × 10² and 1 × 10³; Figure 6(C1)), the tumor area shows a significant difference between the same two populations (p ≤ 0.05; Figure 6(B1)).

Immunolabeling (Figure 7) revealed that the presence of the SKMEL-1 tumors does not affect the formation of the dermis with normal production of collagen IV, a component of the ECM of normal human skin (Figure 7). In immunofluorescence studies of melanoma, keratin 10, keratin 14, involucrin, and collagen IV are used as diagnostic markers. Keratin 10, keratin 14, and involucrin highlight epidermal differentiation and help to distinguish keratinocytes from melanoma cells, since melanoma cells do not express these proteins [23]. Collagen IV marks the basement membrane and ECM, and its integrity indicates non-invasive lesions, while its disruption signals tumor invasion [24]. In the SKMEL-1 melanoma model the epidermis showed alterations with the development of the tumor. In the presence of the tumor cells, the epidermis is thinner than that of normal human skin, and its stratification does not occur correctly. In the SKMEL-1 melanoma model, with a higher number of tumor cells, there was a decrease, or even an absence, in the production of keratin 10, keratin 14, and involucrin in both methods of melanoma cells seeding.

A375 Cell Lines Capacity to Form 3DMSM

Next, we investigated whether the inability to consistently form tumors with this protocol was characteristic of SKMEL-1 or due to additional protocol adjustments. For that, we conducted a similar analysis using the A375 melanoma cell line. In contrast to SKMEL-1, A375 formed tumors under all conditions tested. The results obtained with 3DMSM-CA and 3DMSM-CD are shown in Figure 8A and 8C, respectively.

In all tested conditions, epidermal stratification was compromised by the rapid growth of these melanoma cells, both in 3DMSM-CA and 3DMSM-CD. A tendency to develop larger tumors was observed when the number of melanoma cells per sample was higher, for both seeding methods (Figure 8). Nonetheless, in 3DMSM-CA, there seems to be a greater tendency to form large tumors that proliferate more and reach the dermis.

Histology showed increased staining for S100-positive cells confirming higher melanoma proliferation compared with the with SKMEL-1 cell line. Tyrosinase only marks the melanocytes that are present in the basal layer of the epidermis, but with the increase in tumor’s size, there was a loss of alignment of these melanocytes, as well as a decrease in their presence above the tumor region (Figure 8B,D).

An increased number of melanoma cells correlated with an increased likelihood of tumor formation in both models, as observed in the 3DMSM A375 melanoma model (Figure 9A). Specifically, at higher cell concentrations, 3DMSM-CA showed fewer tumors that reached the dermis than 3DMSM-CD (Figure 9A). It should also be noted that the number of tumors was lower in the 3DMSM-CA model, in which cells were introduced as tumor aggregates, and caution is needed when comparing tumor numbers across models due to differences in initial cell numbers. Regarding the average area and the depth of invasion of the tumors formed, significant differences were observed with a higher number of cells in melanoma aggregates compared to the other conditions. In 3DMSM-CA, the depth of invasion and the area of tumors developed were not as homogeneous as when comparing different cell numbers of tumor aggregates in relation to 3DMSM-CD (Figure 9(B1,C1)). For the 3DMSM-CD, the statistical analysis did not provide sufficient evidence to demonstrate an effect of cell number on tumor average area or depth of invasion. Specifically, the p-value was greater than 0.05, indicating the results were not statistically significant (Figure 9(B2,C2)).

Immunofluorescence analysis of A375 3DMSM-CA (Figure 10A) showed that the absence of expression of keratin 10 and involucrin was only observed when a higher number of melanoma cells was introduced in aggregates (2.5 × 10²). A minimal decrease in fluorescent signal for keratin 10 expression and a lack of keratin 14 expression were observed in 3DMSM-CD, where melanoma cells are dispersed throughout the entire skin, in contrast to 3DMSM-CA (Figure 10B), indicating interactions between melanoma cells and epidermal stratification.

G361 Cell Lines Capacity to Form 3DMSM

To further support the usefulness of this model, we used a third melanoma cell line, G361, which is also adherent, and similar to A375. Results obtained with the G361 cell line were identical to those previously reported for the A375 melanoma cell line (Figure 11). For both models, 3DMSM-CA and 3DMSM-CD, an increase in the tumor size was observed, coinciding with a higher number of melanoma cells initially introduced (Figure 11A,C). Along with increased tumor size, the epidermal thickness in 3DMSM-CA was also affected (Figure 11A). The epidermal structure in 3DMSM-CD was altered, as evidenced by changes in melanocyte positioning within the basal layer and a reduction in their number in association with larger tumors (10 × 10⁴) (Figure 11B).

In G361 3DMSM-CD, there was a clear tendency to an increase in the number of tumors formed in all samples, when the number of melanoma cells was higher (Figure 12A). Consequently, the quantity of tumors presented both in the epidermis and in the dermis also increased. On the other hand, in 3DMSM-CA, this proportionality was not present. However, the number of tumors in the epidermis and dermis remained high compared with thuse restricted to the epidermis (Figure 12A). The introduction of the tumor aggregates previously formed in the cutaneous microenvironment was advantageous for the development of isolated tumors, preventing the growth of a single tumor that would spread across the skin model. Despite this increase in the number of tumors, their size was similar (Figure 12B,C). In 3DMSM-CA, there was a significant increase in tumor area and depth in three conditions tested (2.5 × 10², 5.0 × 10², and 7.5 × 10² cells) (Figure 12(B1,C1)). With 7.5 × 10² cells, both the average area and depth of invasion of the tumors formed were approximately twice those of the remaining tumor aggregate conditions tested (Figure 12(B1,C1)), thereby explaining the presence of fewer tumors (Figure 12A). In 3DMSM-CD, the average area and depth of invasion of G361 tumors were similar across melanoma cell concentrations (Figure 12(B2,C2)). As tumors increase in area and depth of invasion, they tend to progress from the epidermis into the dermis in both methods tested (Figure 12).

In Figure 13, it is possible to observe a collagen structure that supports the dermis in all samples of the G631 melanoma model. In the 3DMSM-CA, the tumors were larger, leading to no or very faint signals for keratin 10, keratin 14, and involucrin in most samples (Figure 13A). Epidermal markers, such as keratin 10 and involucrin, were less expressed in the presence of larger tumors in 3DMSM-CD (Figure 13B). In samples containing 5.0 × 10⁴ and 1.0 × 10⁵ G361 melanoma cells, staining these markers was very faint, or absent. A similar observation was performed with models using the SKMEL-1 and A375 cell lines, reinforcing the idea that melanoma cells negatively impact epidermal formation when in contact with keratinocytes [25,26].

3.2. ECM Components Are Altered in 3DMSM Tissues

In the skin, ECM is the largest component of dermis layer mainly composed of collagens I and III and its main function include wound healing and tissue regeneration [27]. ECM is crucial for the maintenance of homeostasis in normal tissues. However, in solid tumors, abnormal ECM structure is associated with tumor progression, increased risk of metastasis, and poor survival [28,29]. To analyze the ECM components on the 3DMSM tissues, histological sections stained for ECM proteins of interest were compared between 3DMSM tissues with three different melanoma cell lines (SKMEL-1, A375 and G361), with control skin model (without melanoma) and a patient-derived melanoma sample, which also included normal adjacent skin. We started the characterization by broadly analyzing collagen fibers using Masson’s Trichrome staining (Figure 14). This technique highlights the deposition of collagen fibers and, using three different dyes, simultaneously differentiates muscle fibers (red), collagen (green) and nuclei (black) [30,31]. Upon observation, collagen was present in all samples (Figure 14; white arrow), consistent with what is typically observed in the human skin (Figure 14; left column). To obtain more detailed analytical information on ECM, we proceeded with the identification of COLI, the most abundant protein found in the skin [29]. The staining was particularly intense in non-tumoral areas, mimicking what was observed for the patient’s samples (Figure 14; highlighted by the yellow arrow). Next, we analyzed the presence of other important ECM components, the fibronectin, and evaluated its expression pattern. Fibronectin was shown to form scaffolds for other ECM proteins, promoting ECM maturation and attachment, mechano-signaling, and cell migration [32]. Interestingly, there is an absence of fibronectin staining in the tumor areas, contrasting with the non-tumoral areas, as observed in the patient samples (Figure 14; highlighted by the purple arrow).

Further analysis with epithelial-mesenchymal transition EMT markers showed a reduction in E-cadherin expression in the tumor region, while its expression remained more prominent in the epithelium, particularly in the patient-derived melanoma samples and the tissue-engineered skin models without melanoma (control) (Figure 14; green arrow). Conversely, vimentin displayed a pattern indicative of stronger staining in tumor-associated areas and cells (Figure 14; pink arrow), with a more moderate expression in the superficial layers. This pattern was observed in all 3DMSM tissues except for the A375 model, where the vimentin expression is detected, consistent with the absence of E-cadherin staining in this model (Figure 14).

In melanoma diagnosis, the challenge relies on differential diagnosis in relation to other types of tumors. The Sry-related HMg-Box gene 10 (SOX10) is a key nuclear transcription factor in the differentiation of neural crest progenitor cells to melanocytes. Since melanoma arises from melanocytes, it typically expresses SOX10 [33,34]. Therefore, SOX10 positivity in a tumor context strongly suggests melanocytic origin, especially when correlated with clinical and histopathological findings. These results show SOX10 staining in the 3DMSM tissues, which is equally intense as in the human samples (Figure 14) reinforcing the similarity between 3DMSM and in vivo tissue.

Overall, these results confirm that this in vitro melanoma model to mimic key histological and molecular features of human skin.

3.3. 3DMSM Tissue Allows Cytokine Secretion

TME varies between tumor types but generally includes immune cells, stromal cells, secreted factors, and the ECM [13]. Cytokines are small polypeptides that induce autocrine and paracrine signaling in tumor cells and other cell types, mediate leukocyte infiltration into tumors, reshape TME, and are drive of metastatic progression [35]. In cutaneous melanoma, interleukin 6 (IL-6) supports melanoma maintenance and plasticity [36], interleukin 8 (IL-8) has been associated with aggressive growth and metastasis [37] and interleukin 10 (IL-10) expression has been associated to the transition from radial to vertical growth of melanoma, potentiating metastatic competence [38,39]. Therefore, we decided to analyze cytokine secretion (IL-6, IL-8 and IL-10) overtime during melanoma establishment in the 3DMSM with A375 cell line as the study model.

A375 is frequently selected for studies of cytokine expression in melanoma skin models due to its high immunogenic, comprehensive characterization, and strong cytokine responses. These attributes are less pronounced in SK-MEL-1 and G361 cell lines. SK-MEL-1 is often referenced for its genetic mutations and suspension growth properties rather than for cytokine secretion or immune responsiveness. G361 melanoma cells are not ideal for cytokine studies because they exhibit low invasiveness, reduced immunogenicity, and limited cytokine/chemokine expression, which makes them less representative of the inflammatory microenvironment seen in advanced melanoma [35,36].

Results showed that IL-6 and IL-8 levels were below 5 pg/mL, during the initial days, and IL-10 was absent (Figure 15). This corresponds to the stages when the model contains only HDFn (day 0), HDFn and HEM (day 1), and finally, HDFn, HEM, and A375 cells (day 2). By day 3, the model included all cell types, and A375 cells had been present for 48 h (Figure 2). At this point, the cells could potentially interact and trigger immune responses, which may explain the sharp increase in IL-8 concentration, peaking at 168 pg/mL (Figure 15). The IL-6 reached its highest value around day 4, 17 pg/mL (Figure 15). At the onset of the air-liquid interface (day 5), all interleukins showed low initial concentrations. It is important to note that there was no detectable IL-10 until day 5. During 12 days of the ALI phase, all interleukins followed different patterns. IL-6 levels remained consistent across all time points.

It is noteworthy that medium changes occurred every 48 h, leading to interleukin accumulation over this period (compared to the 24-h intervals previously used) (Figure 2). On days 11 and 12, a medium change (media H, Table 1) reduced the accumulation period to 24 h, but it did not appear to affect IL-6 concentration. IL-8 levels tend to rise, peaking around day 11, suggesting a response to the introduction of new nutrients with a medium change (Figure 15). Unlike IL-6, IL-8 showed a reduction in concentration from day 12 to 16, suggesting that during the early stages of differentiation, cells can produce higher amounts of IL-8. As differentiation progresses and cells settle into their roles and functions, they may either be less capable of or be less influenced by IL-8 production. Regarding IL-10, there has been a steady increase in its concentration in the medium since the start of the ALI phase. With each medium change, the measured levels continued to rise, reaching a peak of 17 pg/mL on day 16 (Figure 15). This indicates that as the skin model undergoes stratification and cellular differentiation, IL-10 production increases accordingly.

These results show that 3DMSM can resemble key tumor features, such as cytokine secretion in the melanoma TME, which is important for modulating the melanoma spread [39].

4. Discussion

This work arises from the need to develop a 3D melanoma model that closely mimics in vivo cutaneous melanoma. The model should represent its development, enabling the study of its mechanisms of aggressiveness and advances in therapeutic assays as well as a deeper comprehension of the dynamics of melanoma cell interactions in the TME. Improved skin models support translation of lab results to human conditions and help to observe melanoma cell behavior as seen in patients. When reflecting tumor heterogeneity, they would assist in personalized medicine.

The structure of the dermal ECM could trigger melanoma cell invasion, promoting melanoma cell proliferation toward blood vessels and distant organs [29]. When keratinocytes contact melanoma cells, their behavior changes. Melanoma cells also alter interactions between melanocytes and keratinocytes, increasing melanocyte-fibroblast contact [31].

Furthermore, the model presented here offers several advantages: a full development time of 25 days, melanoma formation in both the epidermis and the dermis, melanocytes present in the epidermal basal layer, and the exclusive use of human cells. In this model, fibroblasts are induced to produce their own ECM components within a neutral polymeric matrix, eliminating animal factors and improving skin barrier function [28,31].

A porous polystyrene scaffold was used to generate the model. Inert in nature, the scaffold’s material does not interfere with cell behavior or proliferation [29]. While existing collagen hydrogel-based models can limit cell growth, the porous nature of this scaffold provides strong support for proliferation and reduces skin contractibility [8,13]. Additionally, its use yields more durable models that can withstand prolonged culture, enabling longer studies [29].

To construct a viable 3D melanoma skin model, it was first necessary to examine the morphology of the established models. It was also important to determine the number of melanoma cells to introduce and the best introduction method (Figure 5, Figure 6, Figure 7, Figure 8, Figure 9, Figure 10, Figure 11, Figure 12 and Figure 13). This ensured tumor development without destroying the entire skin, justifying the different melanoma cell concentrations used in the 3DMSM-CD and 3DMSM-CA models. In 3DMSM-CD, the number of melanoma cells was estimated based on the number of keratinocytes to be seeded later, enabling the study of melanoma-keratinocyte interactions. Several melanoma-to-keratinocyte ratios were tested, considering those suggested in the literature [40,41,42]. In 3DMSM-CA, exponential aggregate growth was monitored over 5 days to check if tumor size matched the number of cells initially introduced. This model also allowed assessment of whether pre-aggregated melanoma cells offered an advantage. Bourland et al. developed a similar model using melanoma spheroids generated by the hanging-drop method, cultured for longer periods, using between 5 × 10² and 4 × 10³ cells [40]. Here, in both tested 3DMSM models, using all different melanoma cell lines, tumor formation is observed. However, two lines (A375 and G361) showed stronger tumor-forming ability in this 3D model (Figure 8, Figure 9, Figure 10, Figure 11, Figure 12 and Figure 13). In contrast, the SKMEL-1 cell line grows in suspension, limiting cell-cell interactions and reducing its proliferation during development (Figure 5, Figure 6 and Figure 7). Recent work has shown that cell aggregates give an advantage over individual cells surviving harsh conditions, evading the immune system, and proliferating faster [36,37].

In this study, a more physiologically relevant tumor model was observed in 3DMSM-CA. Compared with similar studies reported in the literature, the model presented here formed tumors in a more restricted area of the skin (Figure 4, Figure 5, Figure 6, Figure 7 and Figure 8), reproducing the early stages of melanoma development and allowing a more accurate epidermal development [27,29]. Tumor size was consistently proportional to the initial number of melanoma cells added, confirming that melanoma cells cultured as aggregates for 5 days benefit from cell-cell and cell-ECM interactions, resulting in a more structured cluster [28]. Additionally, ECM proteins already secreted and present, not only provide physical and mechanical support but also facilitate cell-ECM communication, activating important signaling pathways that lead to efficient tumor formation and invasion in the 3DMSM-CA model [43].

S100 and tyrosinase markers were used to identify melanocytes in our skin model and to confirm their correct localization in the epidermis and the basal lamina. These molecular markers are currently used, for example, by pathologists to stain melanocytes and to evaluate melanoma progression [37]. Regarding tumor growth, tyrosinase staining was reduced or absent in tumors formed by A375 and G361 cells, as these melanoma cells produce little melanin [43,44]. The absence of tyrosinase can also be related to tumor-induced metabolic changes. Increased glucose uptake by melanoma cells can lead to extracellular acidification and, consequently, decrease tyrosinase production, which may explain the absence of tyrosinase in the with SKMEL-1 model [45]. In these melanoma models, tyrosinase was only present in melanocytes. To confirm the presence of melanoma in 3DMSM models, an important marker used in clinics for melanoma diagnosis, SOX10, was compared to patient-derived melanoma samples. Results showed similar staining in 3DMSM tissues when the 3DMSM generated with the different melanoma cell lines tested were compared to the patient-derived tissues. In all our models (3DFTSM as the reference model and 3DMSM), upon staining with anti-Ki67, cell proliferation was observed mainly in the basal cell layer, showing the correct formation of the epidermis, as in the tumors, indicating the proliferative stage of melanoma cells (Figure 3).

A comparative analysis of the two models showed that 3DMSM-CA exhibited a thicker epidermis, with alterations in keratinocyte organization observed exclusively in the epidermal region situated above tumors.

Examination of epidermal marker expression in the absence of tumor indicated no detectable impact, supporting the preservation of normal epidermal development and stratification (Figure 3). Upon tumor development, a reduction in the expression of keratin 10 and keratin 14 was observed. These keratins are implicated in the differentiation of suprabasal layers and the proliferative capacity of keratinocytes, respectively [25,26,46]. The observed reduction in keratin production suggests that the tumor interferes with appropriate epidermal formation [26]. Notably, the epidermis underwent fewer structural alterations in instances where the tumor was smaller.

Given that three-dimensional models more accurately recapitulate the architectural and cellular heterogeneity of tumor tissues [44], the 3DMSM tissues were characterized with respect to components of the TME. The results confirmed that 3DMSM tissues reflect the in vivo environment with respect to essential ECM constituents and organized collagen deposition. Comparative analysis of patient-derived samples and 3DMSM tissues revealed reduced COLI and the absence of fibronectin in tumor regions in both 3DMSM and patient-derived samples (Figure 14). Additionally, 3DMSM tissues proved suitable for detecting pivotal tumor progression processes, such as EMT, as evidenced by the decrease in E-cadherin levels and increase of vimentin levels in tumor regions, as observed in patient-derived samples (Figure 14). Collectively, these findings underscore the relevance of 3DMSM in skin cancer research; the capacity to differentiate discrete skin layers renders this model highly effective for investigating basement membrane disruption, tumor evasion, and metastatic processes, thus providing valuable insights into melanoma progression [47,48].

We also analyzed cytokine secretion in 3DMSM models by measuring cytokine levels in supernatants during 3DMSM development. Low cytokine concentrations on days 0 and 1 likely reflect that fibroblasts and melanocytes were not yet stimulated to trigger inflammation. Because melanoma cells were seeded atop the HEM layer, they likely needed time to establish before tumor progression began. The absence of IL-6 and IL-8 suggests a stable microenvironment with minimal stress or tumor activity at day 2. Introduction of HEK cells likely triggered a pro-inflammatory response, leading to increased IL-8. For example, after UVB irradiation, keratinocytes initiate an inflammatory response by releasing IL-1β, IL-6, IL-8, MCP-1, and the anti-inflammatory IL-10 [49]. Similarly, interaction between melanoma and HEK cells could trigger the release of IL-6 and IL-8. Melanoma cells secrete IL-8 to promote progression and angiogenesis, explaining the day 3 spike [50]. The IL-6 peak on day 4 reflects heightened inflammation, likely due to interactions among melanoma, fibroblasts, and HEK cells [50]. As ALI is established, the system reorganizes and triggers cell differentiation. The decline in cytokine levels indicates a shift from a pro-inflammatory to a regulated environment, setting the stage for maturation. Steady IL-6 suggests ongoing inflammation, possibly linked to tumor survival and growth [51,52,53]. The IL-8 peak at days 9–11 indicates an active inflammatory phase, which may drive invasion [54]. The decline after day 11 may signal resolution as inflammation subsides and tissue stabilizes, or it may signal that the tumor becomes immune-evasive. A gradual increase in IL-10 suggests a shift toward an immunosuppressive state, promoting survival by dampening immune responses [54]. The IL-10 peak at day 16 likely reflects full differentiation, while the tumor microenvironment fosters immune evasion.

The release of cytokines is strongly influenced by the tumor environment [35]. Cancer-associated fibroblasts are common in the tumor stroma. CAFs are involved in melanoma progression, metastasis, and drug resistance. They exert these effects through cell–cell interactions and secretion of ECM components, growth factors, and cytokines. In vitro co-culture experiments demonstrated that CAFs enhance the migration and invasiveness of melanoma cells. This process depends on the secretion of IL-6 and IL-8 [54]. For example, Jobe et al. developed a spheroid-based model using A2058 human melanoma cells. Their analysis showed that CAFs and fibroblasts were the main producers of IL-6 within the TME [55]. Furthermore, fibroblasts co-cultured with invasive melanoma cell lines showed elevated levels of cytokines, including IL-1β, IL-8, and IL-6. IL-1β may promote invasiveness by increasing expression of pro-inflammatory molecules, such as IL-6 [51]. In addition to paracrine signaling, exosomes released by melanoma cells stimulated IL-6 production in CAFs. In turn, this enhanced melanoma cell migration in vitro within heterogeneous spheroids composed of melanoma cells and CAFs within 3D collagen gels [56,57]. These findings highlight the key role of stromal-tumor cell interactions in melanoma. Given the interactions between HDFn and melanoma cells in the 3DMSM, and the observed spikes in inflammatory cytokines such as IL-6 and IL-8, it is likely that CAFs have developed within this environment (Figure 15). If CAFs have emerged, they could be influencing cytokine release and contributing to the progression of melanoma-like characteristics in the model. Additionally, keratinocytes facilitate immune evasion in melanoma’s, contributing to its immunosuppressive environment. Itakura et al. [38] also demonstrated that IL-10 expression in melanoma is associated with immune evasion, tumor progression, and metastatic potential. This highlights the importance of understanding melanoma biology and developing therapeutic strategies. Therefore, the rise in IL-10 levels during the ALI phase could reflect maturation and stratification of the model, indicating a more developed microenvironment that promotes immune evasion. The distinct patterns of IL-6, IL-8, and IL-10 in the melanoma model offer valuable insights into tumor microenvironment dynamics. These cytokine profiles indicate that the model could mirror inflammatory and immune-evasive traits of melanoma, making it a promising tool for studying tumor progression and exploring potential therapies.

5. Conclusions

The proposed in vitro cutaneous melanoma model was successfully established and included melanocytes and three different melanoma cell lines. The model demonstrated robustness and correct epidermal differentiation and stratification. The tumor spread depends on the type and concentration of melanoma cell lines used in both 3DMSM-CS and 3DMSM-CD. Considering the results, the model with cell aggregates (3DMSM-CA) appears to generate a melanoma model with greater control over tumor size and better dermal and epidermal formation.

Both 3DMSM models effectively replicate critical ECM characteristics and tumor-associated alterations observed in patient samples, including altered collagen I, reduced fibronectin, and shifts in EMT-related markers. The observed cytokine patterns (IL-6, IL-8, IL-10) during melanoma growth indicate that the model exhibits an immunoreactive response. More studies are needed to examine the link between immune response and melanoma growth. Also, comparing these results with tests performed under the same protocols but without melanoma cells (using 3DFTSm) could specifically reveal how melanoma cells alter cytokine dynamics in this model. Future research will monitor immune responses over time and develop a 3D melanoma skin model (3DMSM) with basal immune perfusion for preclinical studies.

3DMSM tissues’ ability to distinguish skin layers and replicate tumor–stroma interactions make them a strong and relevant platform for studying melanoma invasion, basement membrane disruption, and metastatic progression.

This model can be improved by introducing cells involved in the tumor immune response, enabling the study of interactions between melanoma cells and immune cells. Moreover, it can be used as a platform to test new therapeutic drugs towards the development of new therapeutics for cutaneous melanoma.