From Proteomic Mapping to Invasion-Metastasis-Cascade Systemic Biomarkering and Targeted Drugging of Mutant BRAF-Dependent Human Cutaneous Melanomagenesis

Simple Summary Despite the recent advances in human malignancy therapy, metastasis and chemoresistance remain the principal causes of cancer-derived deaths. Given the fatal forms of cutaneous metastatic melanoma, we herein employed primary (WM115) and metastatic (WM266-4) melanoma cells, both obtained from the same patient, to identify novel biomarkers and therapeutic agents. Through state-of-the-art technologies including deep proteome landscaping, immunofluorescence phenotyping, and drug toxicity screening, we were able to describe new molecular programs, oncogenic drivers, and drug regimens, controlling the invasion-metastasis cascade during BRAFV600D-dependent melanomagenesis. It proved that proteomic navigation could foster the development of systemic biomarkering and targeted drugging for successful treatment of advanced disease. Abstract Melanoma is classified among the most notoriously aggressive human cancers. Despite the recent progress, due to its propensity for metastasis and resistance to therapy, novel biomarkers and oncogenic molecular drivers need to be promptly identified for metastatic melanoma. Hence, by employing nano liquid chromatography-tandem mass spectrometry deep proteomics technology, advanced bioinformatics algorithms, immunofluorescence, western blotting, wound healing protocols, molecular modeling programs, and MTT assays, we comparatively examined the respective proteomic contents of WM115 primary (n = 3955 proteins) and WM266-4 metastatic (n = 6681 proteins) melanoma cells. It proved that WM115 and WM266-4 cells have engaged hybrid epithelial-to-mesenchymal transition/mesenchymal-to-epithelial transition states, with TGF-β controlling their motility in vitro. They are characterized by different signatures of SOX-dependent neural crest-like stemness and distinct architectures of the cytoskeleton network. Multiple signaling pathways have already been activated from the primary melanoma stage, whereas HIF1α, the major hypoxia-inducible factor, can be exclusively observed in metastatic melanoma cells. Invasion-metastasis cascade-specific sub-routines of activated Caspase-3-triggered apoptosis and LC3B-II-dependent constitutive autophagy were also unveiled. Importantly, WM115 and WM266-4 cells exhibited diverse drug response profiles, with epirubicin holding considerable promise as a beneficial drug for metastatic melanoma clinical management. It is the proteome navigation that enables systemic biomarkering and targeted drugging to open new therapeutic windows for advanced disease.


Introduction
Melanoma represents the deadliest form of skin cancer, with~100,000 new incidents and~7000 deaths estimated to have occurred for 2020 in the United States [1][2][3]. Its financial burden for clinical treatment remains cumbersome, since the approximate annual cost per patient ranges from $2000-$15,000 for stage I and $35,000-$155,000 for stage IV of the disease [2]. Among its strongest risk factors are family history, fair skin, multiple moles, immunosuppression, and ultraviolet radiation exposure [1,4]. For cutaneous thin melanoma, at the radial growth phase, surgical removal can afford a cure, whereas patients carrying tumors of intermediate thickness (Breslow depth; Clark level) (2-4 mm; from the upper epidermis layer to the innermost invasion depth) may eventually succumb to recurrence at regional or distant tissues [5,6]. Tumor thickness of 4 mm presents a high risk of metastasis and a median survival of 6-9 months [5]. Cells at the vertical growth phase feature growth-factor and anchorage independence, and presage distal metastasis [6]. Thickness-dependent metastasis is associated with genetic heterogeneity and therapy resistance of melanoma cell (sub-)populations [5,7].
Metastasis represents the end product of a multistep cellular process termed the invasion-metastasis cascade (IMC) [31]. IMC is defined by the dissemination of "skillful" cancer cells from a primary tumor and their subsequent colonization in distant tissues [31][32][33]. This sequence of events involves cancer cell intravasation into the circulatory system, survival during hematogenous transit, arrest, extravasation through vascular wall into distant tissue parenchyma, micro-metastatic colony formation, and clinically (macroscopically) detectable, metastatic lesion growth (colonization) [31,33]. Hitherto, no gene mutation has proven to be characteristically associated with progression to metastasis. This indicates the need for prompt development of advanced systemic biomarkering platforms typifying IMC.
For high-scale and deep proteomics analysis, massive cultures of WM115 (primary tumor) cutaneous melanoma cells were harvested (on ice) via mild scrapping, and after three washes with (ice-cold) 0.9% NaCl, they were centrifuged (at +4 • C) for 10 min at 750× g. Supernatants were aspirated and cell pellets were stored at −20 • C for further proteomics processing. WM115 (primary) (this study) and WM266-4 (metastatic) [35] melanoma cells were treated and processed as similarly as possible to each other regarding culturing, harvesting, storage, and proteomics protocols, respectively applied. All cell culture media and related reagents were provided by Merck Millipore/Biochrom AG (Merck KGaA, Darmstadt, Germany).

Protein Extraction-Tryptic Peptide Generation
Technical protocols were employed as previously described [35,48]. Briefly, cell pellets derived from~10 7 cultured cells were suspended in lysis buffer containing 1.5 M Tris-HCl (pH 7.6), 3.5 M urea, 0.1 M SDS, and 3.2 mM DTE, and they were disrupted by tip sonication. Lysates were centrifuged at 13,000 × g for 20 min and total protein concentration was measured in each supernatant using the Bradford assay.
Protein lysates (~150 µg) were reduced and alkylated via treatment with 0.1 mM DTE in Tris-HCl (pH 6.8) at 56 • C for 30 min. Proteins were alkylated by the addition of 0.05 mM iodoacetamide at room temperature for 30 min in the dark. Samples were digested by trypsin (Roche/F. Hoffmann-La Roche Ltd., Basel, Switzerland) at a protein-totrypsin ratio of 100:1, and trypsinization was terminated by the addition of 5% acetic acid. Peptide-containing solutions were vacuum-dried for 60 min, and powder preparations were reconstituted in 0.1% formic acid, for nano liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis.

LC-MS/MS-Data Analysis
Protocols for LC-MS/MS analysis were performed as previously described [35,48,49]. Briefly, each tryptic peptide mixture was separated in a linear gradient of 2-30% solution containing 99.9% acetonitrile and 0.1% formic acid, at a flow rate of 300 nL/min, on a C-18 column (75 µm × 50 cm, 100 Å, 2 µm bead-packed Acclaim Pepmap RSLC; Thermo Fischer Scientific, Waltham, MA, USA). An Ultimate-3000 System (Dionex/Thermo Fisher Scientific, Bremen, Germany) was coupled (on-line) to an Orbitrap Elite Instrument and mass spectra were collected in data-dependent acquisition mode using XCalibur TM v.2.2 SP1.48 software (Thermo Fisher Scientific, Bremen, Germany). Full-scan data were obtained in the 300-2000 m/z range with a 60,000 resolution value and 100 milli-second maximum injection time. Data-dependent MS/MS for the 20 most intense ions per survey scan was performed, with HCD (higher-energy collision dissociation) fragmentation on the Orbitrap at a collision energy of 36 NSE% and a resolving power of 15,000. Fragments were analyzed on the Orbitrap, while the MS/MS spectra were acquired with a maximum injection time of 120 ms and a resolving power of 15,000. All measurements were carried out using m/z 445.120025 as the lock mass and dynamic exclusion was engaged within 45 s to prevent repetitive selection of the same peptide.
Raw data were processed via engagement of Proteome Discoverer (Thermo Fisher Scientific, Bremen, Germany), while protein identification was achieved by employment of the Homo sapiens proteome of reference derived from the UNIPROT database using the Sequest-HT v.28.0 algorithm (Thermo Fisher Scientific, Bremen, Germany). Search parameters were chosen as following: (a) two maximum missed cleavages for Trypsin; (b) oxidation of methionine as variable modification; (c) 0.05 ppm fragment ion tolerance; and (d) 10 ppm peptide mass tolerance. PSMs (peptide spectral matches) were validated using a percolator based on q values at 1% FDR (false discovery rate). Six amino acid residues were chosen as the minimum length of acceptable identified peptides.

Lysosomal Staining
Lysosomal staining was performed using the LysoTracker-Red reagent (Invitrogen/Thermo Fisher Scientific, Waltham, MA, USA) in order to assess lysosomal acidification. WM115 and WM266-4 melanoma cells were seeded onto µ-slide 8-well IBIDI plates (Ibidi GmbH, Martinsried, Germany) until they reached~80% cell confluency. Next, growth medium was removed, cells were washed three times with 1× PBS, and incubated for 30 min with 50 nM LysoTracker-Red at 37 • C in the dark. Subsequently, cells were washed again three times with 1× PBS, covered with mounting medium, and immediately observed under a NIKON Digital Eclipse C1 CLSM (Nikon Corporation, Tokyo, Japan).

Cell Viability-MTT Assay
Melanoma cells were seeded onto 48-well plates and treated with different doses of each indicated drug (or the LY-364947 inhibitor) for 24 h, unless stated otherwise. Cells were incubated with MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] solution for 4 h, and the formazan crystals produced were dissolved in pure isopropanol. Spectrophotometric absorbance was measured in a TECAN Infinite F50 absorbance microplate reader (Tecan Group Ltd., Männedorf, Switzerland) at 550 nm, using 630 nm as the wavelength of reference. Each cell viability (MTT) assay was repeated three times, using three wells per condition (e.g., absence or presence of drug/inhibitor) [66].

Western Blotting
Whole cell protein extracts (~50 µg) were separated in 12% SDS-PAGE gels and subsequently (electro-)transferred onto nitrocellulose membranes (Whatman-Schleicher & Schuell GmbH, Dassel, Germany). Membranes were blocked in 1× TBS-T (tris-buffered saline-Tween-20) containing 5% NFM (non-fat milk) for 2 h at room temperature. Primary antibodies were added at 1:1000 dilution for 2 h at room temperature and subsequently for 16 h at +4 • C. IgG-HRP (anti-rabbit, or anti-mouse) secondary antibodies were diluted 1:2000 and used for 2 h at room temperature, while the immuno-reacting protein bands were visualized by ECL reactions, following the manufacturer's instructions. Pan-actin and β-tubulin served as proteins of reference (control) [66].

Molecular Modeling
Three dimensional (3D) predictions were generated by using the I-TASSER (Iterative Threading ASSEmbly Refinement) online server that has been designed for automated protein structure and function predictions [68,69]. Structural models of protein sequences were constructed from multiple threading alignments and iterative structural assembly simulations. Comparison of the produced models with other known protein structures can provide insights for the function of proteins being investigated [70].
The resulted molecular model of the KIAA0930-vimentin hybrid monomer protein was subjected to advanced docking bioinformatics processing, in order to examine its in silico ability to form homodimers via suitable employment of the automated protein docking server ClusPro [71][72][73][74]. ClusPro has proved able to yield energetically acceptable structural models in many rounds of the community-wide experiments called the CAPRI (Critical Assessment of PRedicted Interactions). CAPRI was designed to test protein docking algorithms in blind predictions of protein-protein complex structures [75] and to therefore identify the best and near-native conformations [73]. ClusPro consists of a search algorithm (Piper) that provides 1000 low-energy results to the clustering program. The 10 generated models form the central structure of the clusters that contain members within a 9 Å C-alpha RMSD radius. These are ranked initially according to the size of the cluster and subsequently according to the lowest energy due to the balanced scoring function of the program. Therefore, the top-ranked conformation of the complex is finally selected because of the good ranking performance of the program in the CAPRI challenge [75].
Images containing structural models were prepared by the PyMol Molecular Visualization System.
The insulin-degrading enzyme (IDE) represents another protein that is differentially expressed between the two cell types herein examined. IDE (5.82) is solely identified in WM266-4 metastatic melanoma cells (Table S3), and according to its vital role in insulin degradation and clearance [84], it may serve as a metastatic driver of human melanoma by suppressing the presumable ability of insulin to inhibit cell invasion and metastasis. Interestingly, leptin and insulin can decrease the invasiveness of colon cancer cells, with insulin (or/and leptin) resistance likely increasing the metastasis risk [85]. Since IDE was not recognized in the WM115 proteomic map (Table S1), primary melanoma cells (e.g., WM115) could activate an IDE-independent, distinct, pathway to achieve "insulin resistance" (n = 29) ( Figure 1B) required for their future metastatic fate. The IGFBP3/IBP3 (16.79) and IGFBP5/IBP5 (3.51) members of the insulin-like growth factor-binding protein family are also presented with a WM266-4-specific expression patterning (Table S3), likely deregulating the bioavailability and signaling power of their cognate insulin-like growth factors (IGFs). IBP3 upregulation has been tightly associated with brain metastasis in lung adenocarcinoma [86], tumor metastasis in nasopharyngeal carcinoma [87], and TGF-βdependent colorectal cancer cell migration and invasion [88]. Hence, it seems that IBP3 may act as principal metastatic driver in BRAF V600D melanoma cell environments, with its targeted drugging presumably offering new opportunities for successful management of the advanced disease.
Surprisingly, WM266-4 but not WM115 cells were shown to carry high levels of the DAP/DAP1 proteomic component (96.65) (Table S3), a small, proline-rich, cytoplasmic protein, which belongs to the death-associated protein family [89,90]. Activated (de-phosphorylated) DAP1 negatively regulates autophagy, while its elevated expression increases the risk of lymph-node metastasis in squamous cell carcinoma of the oral cavity [91][92][93]. Thereby, DAP1 could be endowed with a property to promote melanoma advancement from primary (e.g., WM115) to the metastatic (e.g., WM266-4) stage by downregulating the autophagic machinery whose functionality may differ between WM266-4 [35] and WM115 cells ( Figure 1C; n = 47) (see also Figure 8G). Given that DAP1 has been previously suggested to act as a positive mediator of IFN-γ-induced programmed cell death [90], an IFN-γ-based, novel, drug-cocktail scheme might prove beneficial for metastatic melanoma patients in the clinic. Intriguingly, PGAM5 (10.39), a mitochondrial serine/threonine-protein phosphatase that functions at the convergence point of several necrotic/necroptotic death pathways [94], is also presented with a WM266-4-specific pattern of expression (Table S3), therefore indicating a surprising role of necrosis/necroptosis in melanoma metastasis. Indeed, it has been previously reported that necroptosis of cancer cells can cause tumor necrosis and can promote tumor metastasis in certain oncogenic settings [95]. It is likely, and really astonishing, that the major cell death programs such as apoptosis, autophagy, and necroptosis (and ferroptosis) can critically contribute to the metastatic process of melanoma [35] ( Figure 1C) and other tumors through yet undiscovered, novel, cross-talks among tumor cell sub-populations and their micro-environments.
In contrast to E-cadherin, the other member of the family, CDH2/CADH2/N-cadherin, which serves as a major mesenchymal marker [37][38][39][40][41]113], exhibited positive staining in a significant number of WM115, but not WM266-4 cells ( Figure 2A). However, N-cadherin could be recognized in both WM115 (27.00) (Table S1) and WM266-4 (19.84) [35] proteomes (see also Table S10), thereby dictating the metastasis-induced antigenicity re-modeling (e.g., via post-translational modifications, or/and mRNA splice variants production) of N-cadherin in human melanoma. A rather inverse immunofluorescence patterning was observed for the ZEB2 transcription factor, with its nuclear expression levels being notably higher in WM266-4 than WM115 melanoma cells ( Figure 2A). The transcription factors ZEB1, ZEB2, SNAI1/SNAIL, and SNAI2/SLUG act as principal regulators of EMT through repression of the epithelial state and induction of the mesenchymal state. For example, ZEB and SNAI gene family products can directly bind onto the CDH1 promoter, causing its cognate gene silencing [36,37,[39][40][41]45,111,112,114,115]. Remarkably, the strong nuclear immunofluorescence signal of ZEB1 in both WM115 (primary) and WM266-4 (metastatic) melanoma cells (Figure 2A) is mechanistically associated with their inability to transcriptionally activate the CDH1 gene ( Figure 2A). Despite its absence from both proteomic collections (Table S1 [35]), SLUG protein presented a clear and intense nuclear compartmentalization only in WM115 cells (Figure 2A), showing a cell type-specific and rather opposite distribution pattern compared to the ZEB2 protein. It seems that primary (WM115) and metastatic (WM266-4) melanoma cells employ different combinations (and intracellular quantities/topologies) of transcription factors to direct their respective EMT programs.
To more accurately determine the hybrid EMT/MET states (early or late [38]), WM115 and WM266-4 cells were processed for immunofluorescence detection of PDGFRβ, a signaling membrane receptor that typifies the late-hybrid EMT state [38]. Remarkably, many WM115 (primary) and WM266-4 (metastatic) melanoma cells proved positive for PDGFRβ immunostaining, carrying diverse-sized cytoplasmic "specks" ( Figure 2B). Hence, it seems that distinct WM115 and WM266-4 cell sub-populations have entered a PDGFRβ-mediated late-hybrid EMT program. Different tumor (e.g., melanoma) cell sub-populations may be associated with different EMT stages, ranging from completely epithelial (full MET) to completely mesenchymal (full EMT) ones, and passing through intermediate hybrid (EMT/MET) states that confer distinct cell plasticity, invasiveness, and metastatic potential [39,113]. Hybrid EMT/MET phenotypes likely provide navigating tumor (e.g., melanoma) cells with survival advantages in adverse micro-environments such as blood and lymphatic vessels, and secondary tumor sites [36]. Mesenchymal properties are required for the intravasation of cells from primary tumor (e.g., melanoma) and their survival in blood circulation, whereas epithelial traits seem to be indispensable for metastatic colonization at distant sites [33,36,114]. Interestingly, spontaneous metastasis of primary tumors has been recently associated with MET program activation [119], although MET-independent mechanisms may significantly contribute to the higher metastatic capacities of hybrid EMT/MET-undergoing cell sub-populations [39,113]. Epithelial (MET) and mesenchymal (EMT) cell traits probably need to be co-expressed within individual tumor (e.g., melanoma) cells, rather than in exchanging signal distinct (either EMT or MET) cells for efficient tumorigenesis (e.g., melanomagenesis) including metastasis to unfold. Although both WM115 (primary) and WM266-4 (metastatic) BRAF V600D melanoma cells have activated hybrid EMT/MET programs, WM266-4 cells seem to have acquired a rather stronger epithelial (MET) character, thus revealing their higher propensity to melanoma outgrowth and distant metastasis.
Strong evidence has emerged for the critical role(s) of lysyl oxidase (LOX) family members in promoting metastasis [120] with LOX, an extracellular matrix-modifying enzyme, having been proved essential for hypoxia-induced metastasis [121,122]. Thereby, via western blotting, LOX cellular contents were examined in both WM115 and WM266-4 cells. Interestingly, both cell types presented similar LOX (dimer) expression levels ( Figure 2C), indicating the proclivity of primary melanoma cell sub-populations toward LOX-mediated metastasis. Given that the lysyl oxidase homolog 2 (LOXL2) upregulation seems to cause tumor progression and metastasis through re-modeling of the tumor microenvironment [120,123], LOXL2 was investigated by immunofluorescence ( Figure 2B) and western blotting ( Figure 2C) protocols in the two cell types. Strong expression was detected in either WM115 (primary) (2.33; Tables S1 and S2) or WM266-4 (metastatic) melanoma cells, with WM115 carrying higher (cytoplasmic) protein levels compared to the WM266-4 respective ones ( Figure 2B,C). It seems that, similar to LOX, LOXL2 may also foster primary (e.g., WM115) melanoma cells to precociously gain metastatic traits, phenotypically resembling melanoma "mature" metastases (e.g., WM266-4). Importantly, LOX and LOXL2 enzymes have been reported to mediate a HIF1-dependent EMT program induction in response to hypoxia [120,124]. Taken together, primary (e.g., WM115) and metastatic (e.g., WM266-4) melanoma cells likely engage LOX family members to activate EMT repertoires and trigger time-specific metastases.

TGF-β Signaling Controls WM115 and WM266-4 Cell Motility In Vitro
The TGF-β signaling pathway leads to activation of EMT program(s) through several distinct mechanisms, with the TGF-β-induced SMAD complexes transcriptionally turning on mesenchymal genes (e.g., VIM) or turning off epithelial genes such as CDH1, via, among others, upregulation of the EMT-specific transcription factors SNAIL, SLUG, and ZEB1. Thereafter, critical EMT regulators (e.g., SNAIL) can activate an autocrine TGF-βdependent signaling, creating a positive feedback loop that helps cells to maintain their EMT repertoire(s) once established [37,41,45,111,112,[125][126][127][128]. Hence, we next examined the expression profile of phosphorylated SMAD2 (p-SMAD2) transcription factor, a major mediator of TGF-β signaling [37,125,128,129], in WM115 and WM266-4 cells ( Figure 3A). Both primary (WM115) and metastatic (WM266-4) melanoma cells presented with similar immunofluorescence patterns, being mainly featured by few nuclear "specks" (of diverse size and shape) per positive (immunoreacting) cell, thus indicating the TGF-β/SMAD(2) axis functional engagement in both primary and metastatic melanoma environments. Altogether, an autocrine TGF-β signaling route and a p-SMAD2-dependent induction of EMT program(s) are strongly suggested to successfully operate both in primary and metastatic forms of human BRAF V600D melanoma.
To investigate the role(s) of TGF-β signaling in melanoma cell motility and migration, two fundamental EMT features [36][37][38]45,111,112,128], a wound healing assay, in the presence or absence of LY-364947 that serves as a specific TGF-β signaling inhibitor [126,130], was suitably conducted in vitro. In contrast to WM115, WM266-4 cells proved able to carry strong motility capacities with a closing gap completion time between 24 and 48 h ( Figure 3B,D). The significantly lower migration speed of WM115 cells ( Figure 3B,D) likely reflects their primary tumor (e.g., melanoma) character, while the comparatively higher speed of WM266-4 cells to fill the gap in vitro ( Figure 3B,D) proclaims their strong migratory, invasive, and metastatic abilities.
Remarkably, exposure of WM115 (primary) and WM266-4 (metastatic) melanoma cells to the LY-364947 inhibitor compelled both cell types to completely cease migration toward the artificial gap-closure ( Figure 3C,E), thus evidencing the strong dependence of primary and metastatic melanoma cell motility/migration processes by TGF-β (autocrine) signaling in vitro. Of note, many cell death incidents could be specifically recognized in the LY-364947-treated (72 h) WM115 cell (sub-)populations, whereas several large-sized and shapeelongated cells were rather exclusively observed inside the respective gaps of WM266-4 culture monolayers in response to LY-364947 administration for 72 h ( Figure 3C). Clonal isolation, molecular characterization, and chemical targeting of metastatic melanoma cell escapers generated during LY-364947 treatment may open new windows for the successful management of advanced disease.
Since podocalyxin (PODXL) behaves as an EMT-induced protein, mediates cancer cell extravasation, and directly interacts with ezrin [160], an F-actin-binding protein [161], we next examined its immunofluorescence patterning in WM115 and WM266-4 cells via utilization of a mouse monoclonal antibody against TRA1-60(S) that most likely recognizes PODXL [162,163]. Surprisingly, in contrast to WM266-4, a relatively small number of WM115 cells were characterized by extremely strong TRA1-60(S)/PODXL (PODXL) expression and filamentous-like organization, with several invadopodia being enriched by PODXL accumulation, specifically in WM266-4 cell (sub-)populations ( Figure 5C). Interestingly, SOX2-like DNA-binding motifs could be in silico recognized in the (distal) respective promoters (or enhancers) of ARPIN and PODXL human genes (data not shown), thereby dictating the presumable ability of SOX2 to transcriptionally (over-)activate the ARPIN and PODXL genes in WM115 primary melanoma cells (see also Figure 4). It seems that PODXL represents a novel EMT marker in primary melanoma (e.g., WM115) with its downregulated levels in metastatic melanoma (e.g., WM266-4) pointing out the activation of a hybrid EMT/MET program in the advanced disease. In accordance with a previous report [160], PODXL may promote extravasation and thus subsequent transition from primary to metastatic state(s) of melanoma cells via ezrin-mediated rearrangement(s) of the actin cytoskeleton. Of note, since PODXL can be antigenically modified in retinoic acid (RA)-treated cells [162], we herein suggest the engagement of RA-dependent antigenic aberration(s) of PODXL protein in WM266-4 cells ( Figure 5C). A retinoic-acid receptor RXR-gamma (RXRG) signaling axis has recently been shown to drive the emergence of a (mutant BRAF) melanoma cell population that confers treatment resistance, while targeting of RXR signaling holds strong promise for delaying or obviating melanoma relapse [164]. Remarkably, the nuclear receptor RXRG (5.70) was exclusively detected in the WM266-4 proteomic map (Table S3, [35]), likely indicating the RXRG-mediated antigenic compromise and thereby functional loss of PODXL protein in metastatic (e.g., WM266-4) melanoma cells. Pharmacological inhibition of RXR activities by the selective and potent RXR antagonist HX531 [164,165], in combination with clinically approved chemotherapy schemes, may prove beneficial for the successful management of metastatic and chemoresistant BRAF V600D melanoma.
Given the critical role of PCM1, a major component of centriolar satellite(s) (CS), in the assembly of centrosomal proteins and MT network organization [171,172], we next examined its immunofluorescence patterning in WM115 and WM266-4 cells ( Figure 5C). Both cell types proved positive for strong PCM1 immunostaining, with WM115 featuring a notably more scattered cytoplasmic profiling compared to the WM266-4 respective one ( Figure 5C). Large-sized "specks" neighboring cell nuclei (in WM115 and WM266-4) may derive from a PCM1 polymerization process, whereas small-sized "specks" dispersed in the cytoplasm (in WM115) can presumably result from a PCM1 oligomerization, or even dimerization, mechanism. The lack of small-sized (and low signal) scattered "specks" in WM266-4 cytoplasm ( Figure 5C) seems to serve as a novel metastatic biomarker for BRAF V600D human melanoma. Likewise, the absence of PCM1 and β-tubulin (MT) colocalization patterns (yellow coloring) in WM266-4 cells ( Figure 5C) could also be used as a valid indicator for metastatic melanoma development. It may be the different PCM1 posttranslational modifications (e.g., phosphorylation) or diverse transcript splice variants that control the distinct PCM1 topology and interactivity, and CS localization between primary (WM115) and metastatic (WM266-4) melanoma cells. Accordingly, it previously described the formation of "PCM1 granules" being (besides concentrated around centrioles) scattered throughout cell cytoplasm and directed by PCM1 self-aggregation that is regulated in a cell cycle-dependent manner [172].
Since lysosomal motilities are mechanistically associated with MT dynamics and homeostasis [173][174][175][176], we subsequently studied the intracellular topology profiles of lysosomes in WM115 (primary) and WM266-4 (metastatic) melanoma cells ( Figure 5C). LysoTracker-Red-stained lysosomes seem to undergo a significant re-organization of their topology during the transition process from primary (e.g., WM115) to metastatic (e.g., WM266-4) melanoma state(s). Notably, the number of cells featuring a "cortical", peripheral (plasma membrane underlying), localization, and distribution of (acidified) lysosomes is markedly increased in WM266-4 (metastatic) compared to WM115 (primary) melanoma cells ( Figure 5C). This indicates the mechanistic coupling of MT-dependent lysosome trafficking/positioning to hybrid EMT/MET-driven metastasis during BRAF V600D -positive melanomagenesis. Although peripheral scattering, as opposed to perinuclear clustering, of lysosomes has been associated with alterations in mTORC1-kinase activities, the mechanism that links lysosome positioning to mTOR signaling still remains elusive [173]. Altogether, the anterograde (toward cell periphery) movement of lysosomes may promote the IMC program in BRAF V600D melanoma environments.

Molecular Modeling of a KIAA0930-VIM Gene-Fusion Product Exclusively Identified in WM266-4 Cells
Since vimentin has herein been proven as the most abundantly expressed (EMT) proteomic component in both WM115 (Table S1) and WM266-4 [35] cells (see also Table S10), its mutation-driven aberrant function(s) may critically contribute to the IMC process in BRAF V600D -dependent melanoma. Hence, by engaging a computational platform derived from the "Cancer Dependency Map" project (Cancer Cell Line Encyclopedia) of Broad Insti-tute (MIT-Harvard University; Cambridge, MA, USA) [62][63][64][65], we systemically compared the mutational signatures of WM115 (primary) with WM266-4 (metastatic) melanoma cell ones (access day: 14 September 2020). WM115 cells were presented to contain 535 gene mutations (e.g., single/double nucleotide polymorphisms, insertions, and deletions) (Table S4), whereas WM266-4 cells were shown to carry 531 mutations (of similar type) (Table S5) with 114 and 110 of them being exclusively detected in WM115 and WM266-4 melanoma cells, respectively (Tables S6 and S7) (see also Figure 10). Furthermore, 15 fused genes were recognized in WM115 (Table S8), while 132 fused genes could be described in the WM266-4 (Table S9) cells (see also Figure 10), thus indicating the major importance of multiple gene-fusion product functionalities to the transition process from primary to metastatic melanoma state(s).
Surprisingly, WM266-4 (metastatic) but not WM115 (primary) melanoma cells proved to bear two VIM-related gene fusions: (a) a KIAA0930-VIM and (b) a VIM-OGFOD3, with KIAA0930-VIM being the only one to presumably result in functional protein product(s) (data not shown). KIAA0930 (K0930/C22orf9) represents a hitherto, uncharacterized protein (UniProtKB-Q6ICG6-K0930_Human) whose cognate coding gene has been recently suggested to serve as a novel candidate for lung cancer risk [177]. Employment of the bioinformatics tool "MOTIF: Searching Protein Sequence Motifs-Genome Net" (MOTIF Search) revealed that the human KIAA0930 protein isoform of 409 amino acid residues ("aa") derived from the KIAA0930-201 (Ensembl Transcript ID: ENST00000251993.11) transcript splice variant possesses a DUF2045 (PF09741.10) domain that embraces the "E- x-x-[S/T]" (x: any "aa") novel motif, which can only be detected (via MOTIF Search) in the KIAA1712/CEP44 centrosomal protein ( Figure S2). CEP44 plays important roles in centrosome cohesion and linker (holds the duplicated centrosomes together) assembly [178], while it ensures the formation of centriole wall required for centriole-to-centrosome conversion [179].
Hence, we next attempted via an advanced computational approach to construct structural models for human vimentin, KIAA0930, and KIAA0930-vimentin hybrid protein.
Although experimental data has been reported only for vimentin, these concern small fragments of the protein and not its complete "aa" sequence. Molecular models were herein built by the I-TASSER server, providing as input the sequence of each examined protein. Quality assessment of each protein model (by I-TASSER) was performed by calculation of the C-score, the template modeling score (TM-score), and the root mean square difference (RMSD). C-score is a confidence score, whose value typically ranges from −5.00 to +2.00. A high value of C-score indicates high confidence in the model. The TM-score represents a scale for measuring the structural similarity in between two proteins with different tertiary structures. A value of TM-score over +0.50 indicates the correct topology of a predicted model, while a value below +0.17 points out random similarity.
In the case of vimentin, two different molecular models were herein built. The first one corresponded to its full-protein sequence (1-466 "aa"), with the C-score, TM-score, and RMSD obtained values being measured as −2.90, 0.38 ± 0.13, and 14.3 ± 3.8 Å, respectively ( Figure 6A), clearly indicating that the predicted model was not structurally accurate and valid. Notably, model inspection (and I-TASSER results) foresaw conformational disorders at both amino-("N") and carboxyl-("C") termini of full-length vimentin ( Figure 6A). However, an in silico truncated form of vimentin (81-450 "aa") that was missing its presumably disordered parts produced a model with −1.34, 0.55 ± 0.15, and 9.7 ± 4.6 Å C-score, TM-score, and RMSD measured values, respectively, thus indicating the model's structural reliability and accuracy ( Figure 6B) compared to the full-length protein-derived one ( Figure 6A). Interestingly, the 81 vimentin 450 modeled protein seems to adopt a fibrillar structure, which nicely justifies its ability to self-polymerize and generate long-length fibers (filaments) in vivo. Similar to the full-length vimentin ( Figure 6A), the KIAA0930 (1-409 "aa") molecularly modeled protein was also characterized by unsatisfied measurements with the C-score, TM-score, and RMSD being calculated at −2.93, 0.38 ± 0.13, and 14.0 ± 3.9 Å, respectively ( Figure 6C), strongly suggesting uncertainty for the model's fidelity and trust. Surprisingly, the KIAA0930-vimentin hybrid protein, which in silico embraces both the full-length KIAA0930 (1-409 "aa") and vimentin (1-466 "aa") protein sequences, led to a notably satisfactory molecular model, with the C-score, TM-score, and RMSD parameters obtaining −0.22, 0.68 ± 0.12, and 9.1 ± 4.6 Å measured values, respectively ( Figure 6D), thereby dictating the model's accuracy and reliability. Next, via employment of the automated protein docking server ClusPro, docking experiments were suitably conducted to examine the homodimerization capacity of the KIAA0930-vimentin hybrid monomer protein. Indeed, as illustrated in Figure 6E, KIAA0930-vimentin proved capable of self-dimerizing through a molecular interface that contained the KIAA0930specific LZ-like motif "A -6-L -6-L -6-S -6-A -6-V".
An unexpected and surprising finding of the present study is the ability of KIAA0930 and vimentin proteins to obtain structurally robust conformations exclusively at their hybrid state context, thus indicating that KIAA0930-vimentin likely acquires novel properties concerning intracellular functionality, interactivity, regulation, and topology. Since KIAA0930/K0930 (KIAA0930) (3.88) could be solely identified in the WM266-4 proteomic map (Table S3, [35]), the KIAA0930-VIM gene-fusion product(s) may critically contribute to the IMC process of BRAF V600D -dependent human melanomagenesis. Given that the

DUF2045-accomodated "E-x-x-C-V-x-L-x-x-x-D-x-x-x-[S/T]-x-x-[G/I]-[V/I]-x-[F/Y]-x-x-[S/T]" novel motif, besides KIAA0930
, can also be recognized in the CEP44 centrosomal protein ( Figure S2), it can presumably serve as a dimerization signal for KIAA0930 and CEP44 specific interactions. If so, KIAA0930-vimentin (only in its structured conformation) may direct CEP44 away from its target, the centrosome, likely compromising the fidelity of mitotic division and promoting incidents of chromosomal heterogeneity. Intriguingly, LRRC45, a component of centrosomal linker [184], carries a new motif of regularly spaced "L", the "L -12-L -12-L -12-L -12-L" sequence (data not shown), which through its putative interaction(s) with the LZ-like motifs of structured KIAA0930-vimentin hybrid protein may be depleting centrosome(s) from LRRC45, further deregulating cell-division and chromosomal-segregation processes. Alternatively, KIAA0930-vimentin could recruit centrosomal components (or even ventrosomes) onto vimentin cytoskeleton filaments, probably facilitating invadopodia formation in metastatic melanoma (e.g., WM266-4) cells via coordinated and synergistic activities of vimentin-and MT-based networks. Accordingly, it was previously reported that the PCM (pericentriolar material) of a centrosome appears to possess attachment sites for vimentin intermediate filaments (VIFs) and perinuclear VIFs can infiltrate the PCM [185]. Our inability to detect the putative KIAA0930-vimentin hybrid protein in a western blotting-derived vimentin profiling (Figure 2) suggests the major antigenic alteration(s) of the hybrid protein or/and the extremely low rate(s) of genefusion and hybrid protein synthesis events. The systemic examination of tumor biopsies derived from patients having been affected by primary or metastatic BRAF V600D -positive melanoma(s) (before and after drug treatment) for KIAA0930-VIM gene-fusion incidents (Table S9) may prove beneficial for the successful management of advanced disease in the clinic. In each monomer, residues 22-99 are colored in red and blue, respectively, and are located at the interface of interacting monomers. "M": Methionine. "E": Glutamic Acid. "Q": Glutamine. "R": Arginine. "T": Threonine.
Likewise, given the importance of YAP transcriptional actions to cancer chemoresistance [206,209,210], its nuclear compartmentalization in both primary (e.g., WM115) and metastatic (e.g., WM266-4) melanoma cells strongly suggests the HIPPO Off /YAP On signaling axis as a major contribution to genetically directed chemoresistance in the advanced disease. Since the protein products of AMOTL1 and AMOTL2, two typical YAP-target genes [205,206], could both be identified in the WM115 (2.74; 0.00) (Table S1) and WM266-4 (0.00; 0.00) [35] proteomic collections (see also Table S10), a YAP-specific (HIPPO Off ) molecular signature seems to be tightly associated with the development of BRAF V600D -positive melanomagenesis. The significant heterogeneity of YAP-positive nuclear localization pattering (e.g., YAP + and YAP − nuclei) herein observed solely in WM266-4 cells ( Figure 7A) mechanistically reflects the HIPPO signaling heterogeneity (e.g., HIPPO Off and HIPPO On , respectively) and dictates the critical role(s) of HIPPO Off /YAP On signaling pathway in the IMC program during BRAF V600D -dependent melanoma formation. Given the principal implication of YAP activity in the transcriptional control of EMT [205,206,211], and the WM266-4-specific nuclear YAP profiling heterogeneity ( Figure 7A), it seems that a YAPdependent hybrid EMT/MET program is required for IMC in mutant BRAF (e.g., V600D) human cutaneous melanoma(s).
Sustained activation of endoplasmic reticulum (ER)-stress sensors can endow tumor cells with greater metastatic capacity [212][213][214][215]. The unfolded protein response (UPR) adaptive mechanism that is induced to restore ER homeostasis essentially contributes to multiple steps along IMC including EMT [212,214,216]. Interestingly, it seems that the PERK-eIF2α-ATF4 signaling branch of UPR can be selectively and constitutively activated by cancer cells having undergone EMT [212,216], with PERK kinase inhibition or ATF4 gene silencing dramatically reducing tumor (lung) metastasis ability in vivo [212,216,217]. Hence, we herein examined the immunofluorescence patterning of the ATF4 transcription factor, and its downstream CHOP gene-target product [212,214] in WM115 and WM266-4 cells ( Figure 7B). Both ER-stress/UPR-related transcription factors ATF4 and CHOP were shown to be excluded from the nucleus, and specifically compartmentalized in the cytoplasm ( Figure 7B), thus indicating the PERK-ATF4-CHOP signaling-independent transition process from a primary (e.g., WM115) to metastatic (e.g., WM266-4) melanoma state(s).
Metabolic re-programming is tightly associated with the tumor's metastatic potential. Since the MCT1 lactate transporter (importer) plays diverse critical roles in the metastasis process [218][219][220][221][222][223], we next analyzed the cellular localization and distribution pattern of the MCT1 (SLC16A1/MOT1) transporter in WM115 and WM266-4 (2.70) (Table S3, [35]) melanoma cells ( Figure 7C). Surprisingly, through immunofluorescence imaging, MCT1 was presented to strongly accumulate in the melanoma nucleus and not the cell membrane (as expected) for both cell types ( Figure 7C). Likewise, MCT4 (SLC16A3/MOT4), another critical lactate transporter (exporter) of the family [220,224], also exhibited a nuclear compartmentalization pattern and an unexpected absence from the cell membrane area in both WM115 (primary) (2.55) (Tables S1 and S2) and WM266-4 (metastatic) melanoma cells ( Figure 7C). Intriguingly, nuclear localization of MCT1 has also been observed in prostate cancer cells [225], soft-tissue sarcomas [226], and endometrial cancer [227]. Similarly, MCT4 was previously described to reside in the cell nucleus throughout mouse embryo pre-implantation development [228], while a weak signal profile could be detected in the nucleus of pancreatic cancer cells [229]. Given that MCT1 can promote tumor metastasis independently of its lactate-transporter activity [220,221], novel molecular functions of the MCT1 and MCT4 family members likely remain to be discovered. Thereby, the nuclear topology of MCT1 and MCT4 in primary (e.g., WM115) and metastatic (e.g., WM266-4) melanoma cells may indicate new and transporter-independent properties of these proteins, causing them to act as orchestrators of nuclear architecture or/and regulators of gene transcription during BRAF V600D -dependent melanomagenesis. Most importantly, nuclear MCT1 (nMCT1) and MCT4 (nMCT4) patterning could serve as a clinically reliable and valid (diagnostic) biomarker for the disease.
Aside from metabolic adaptation, hypoxic adaptation that involves hypoxia-inducible (transcription) factors (HIFs) is also a survival mechanism for tumor cells [220,230,231]. Hypoxia has proved able to activate the SLC16A3 gene transcription in a HIF1-mediated fashion [220,232], thereby re-modeling the metabolite profiling of hypoxic cells. Depletion of oxygen (O 2 ) likely leads to oxidative stress, which can cause detrimental protein misfolding that induces ER-stress and adaptive UPR [231,233]. Most importantly, hypoxia has been mechanistically linked to EMT, with HIF1 playing a major role in this oncogenic association [234][235][236]. Therefore, we herein examined the expression patterning of HIF1α transcription factor in WM115 and WM266-4 (0.00) (Table S3, [35]) melanoma cells via employment of both immunofluorescence ( Figure 7D) and western blotting ( Figure 7E) experimental platforms. Remarkably, in accordance with WM115 and WM266-4 proteomic maps (Tables S1-S3 [35]), HIF1α was exclusively detected in WM266-4 (metastatic), but not WM115 (primary) melanoma cells ( Figure 7D,E), strongly suggesting its principal contribution to the IMC program during BRAF V600D -dependent melanomagenesis. Since many WM266-4 cells have presented with a strong nuclear compartmentalization pattern of HIF1α ( Figure 7D), it seems that metastatic melanoma cell sub-populations can activate a HIF1α-mediated transcriptional program independently of their normoxic growth setting(s). Despite their normoxic culturing in vitro, WM266-4 melanoma cells can retain their tumor-derived hypoxic adaptations, engaging HIF1α to orchestrate this "hypoxic memory" process. Alternatively, a hypoxia-independent mechanism such as the formation of genetic alterations [237] might be activated for the localization of HIF1α in the metastatic melanoma nucleus.
Similarly, cisplatin (via immunofluorescence) proved able to induce strong phosphorylation of the H2AX histone (p-H2AX) in both WM115 (primary) and WM266-4 (metastatic) melanoma cells ( Figure 8C). Nevertheless, WM266-4 presented markedly lower numbers of p-H2AX-negative cells compared to the WM115 respective ones in response to cisplatin, indicating the critical contribution of p-H2AX-emanated signaling to the IMC of mutant BRAF human melanoma(s) during chemotherapy. Since p-H2AX serves as a sensitive marker for DNA double-strand break (DSB) formation and a key factor for DNA-damage repair [247,249,264], its genotoxicity-sensing proficiency can generally en-hance chromosomal stability and reduce phenotypic heterogeneity in metastatic state(s) of BRAF V600D -positive melanoma(s). However, it may be that the few p-H2AX-negative WM266-4 cells can efficiently survive and propagate in the presence of cisplatin, leading to new chemoresistant and heterogenic melanoma cell sub-populations with strong invasion and metastatic activities. Hence, restoring H2AX DNA-repair function in drug-treated cell escapers may open a new therapeutic window for the successful management of advanced melanoma(s) in the clinic.
TRAIL belongs to the ligands that activate the caspase repertoire to drive tumor cells to apoptosis. However, tumor, and especially melanoma, cells can evade TRAIL-directed apoptotic death via engagement of diverse molecular mechanisms including downregulation of its cognate TRAIL-R2 (DR5) death receptor [271,[276][277][278][279]. Thereby, the DR5 immunofluorescence profiling was herein imaged in WM115 and WM266-4 melanoma cells ( Figure 8E). Surprisingly, DR5 was presented with a strong cytoplasmic, but not membranous (as expected), compartmentalization patterning for both cell types with the obtained staining being asymmetrically accumulated in close to nucleus areas resembling the ER/Golgi apparatus ( Figure 8E), markedly underpinning previously reported observations [277,278,280]. It seems that primary (e.g., WM115) and metastatic (e.g., WM266-4) melanoma cells can re-program their DR5-specific trafficking routes to exclude DR5 from its typical cell membrane niche, thus prohibiting TRAIL-DR5 signaling axis activation and promoting resistance to TRAIL-triggered apoptosis. Interestingly, the DR4/DR5emanated apoptotic signaling can be selectively attenuated by the EMT program. Binding of E-cadherin (a major epithelial marker) specifically to the ligated DR4/DR5 death receptors causes augmentation of the downstream apoptotic signaling (e.g., caspase-8 activation), while depletion of E-cadherin significantly reduces sensitivity to cell death induction by TRAIL [281]. Therefore, the absence of E-cadherin from WM115 and WM266-4 cells (Figure 2) may attenuate the TRAIL-DR5-dependent apoptosis, likely begetting BRAF V600D melanoma resistance to TRAIL administration. Besides its apoptotic signaling impotence, cytoplasmic DR5 (cDR5) clustering may be critically implicated in the initiation and progression of BRAF V600D -dependent cutaneous melanoma disease. Indeed, the TRAIL-DR5 pathway, instead of triggering apoptosis, can induce the metastatic potential of mouse melanoma cells [282]. Likewise, oncogenic KRAS can convert death receptors into metastasis-promoting receptors in colorectal cancer cells [283][284][285][286]. Taken together, we herein suggest that clustered cDR5 (in the absence of TRAIL) may act as an IMC regulator in BRAF V600D -positive melanoma micro-environments.

Targeted Drugging Efficacy against WM115 and WM266-4 Melanoma Cells Depends on Their IMC States
Given the ability of LY-364947, a specific TGF-β signaling inhibitor [126,130,[313][314][315][316], to remarkably inhibit the migratory (and invasion) capacity of both WM115 (primary) and WM266-4 (metastatic) melanoma cells in vitro (Figure 3), we next investigated its cytotoxic potency against the two cell types via employment of a MTT survival assaybased technology (Figure 9). Notably, 48 h treatment with LY-364947 proved unable to drive WM115 and WM266-4 cells to death ( Figure 9A), indicating the distinct TGF-β signaling roles in controlling melanoma cell survival (Figure 9) versus motility/migration activity ( Figure 3) in a time-course of 48 h. Nevertheless, many cell death incidents were observed exclusively in WM115 cells upon their exposure to LY-364947 (100 µM) for 72 h (Figure 3), thereby revealing the inhibitor's power to impair both migratory/invasion and survival/growth abilities of BRAF V600D primary melanoma cells in response to its (LY-364947) long-term administration (e.g., 72 h). Since WM266-4 cells have been subjected to a structural MT network re-modeling during the IMC process ( Figure 5), vinblastine (Vinca alkaloid), a specific MT-destabilizing agent [317][318][319][320][321], was subsequently analyzed for its capacity to kill melanoma cells (Figure 9). Interestingly, WM266-4 presented a significant tolerance to vinblastine exposure for 24 h compared to WM115 at all administered doses, with the higher ones (50 and 100 µM) exhibiting the most important and clinically relevant differences ( Figure 9B), which could be beneficially exploited for advanced disease therapeutic management. It may be that the re-organized, metastatic, MT cytoskeleton architecture renders WM266-4 cells (partly) resistant to vinblastine. Remarkably, TUBB4A/TBB4A (348.32), a brain-specific (tubulin) isotype [317], is exclusively detected in WM266-4 proteomic map (as its most abundantly expressed component) (Table S3 [ 35]), and thus it is herein suggested to essentially contribute to the acquisition of a metastatic MT network and a vinblastine (semi-)tolerance. Of note, downregulation of βIVa-tubulin (TBB4A) seems to be implicated in the increased sensitivity of lung cancer cells to the tubulin-binding agent vincristine (Vinca alkaloid) [322]. Hence, it could be the upregulation of TBB4A that endows BRAF V600D metastatic melanoma cells with significant resistance to vinblastine via critical re-modeling of the MT cytoskeleton.
The capacity of cisplatin, a platinum-based drug that is widely used in cancer chemotherapy [260,261,323] to induce melanoma cell type-specific apoptotic responses (Figure 8) prompted us to next examine the survival profiling of WM115 and WM266-4 cells in response to 24 h cisplatin exposure (Figure 9). Both cell types were presented with almost identical viabilities for all administered drug doses. Of note, the higher cisplatin concentrations (50 and 100 µg/mL) proved highly efficient for either WM115 (primary) or WM266-4 (metastatic) melanoma cell elimination ( Figure 9C), indicating their presumable utilization for clinical treatment of the disease. Intriguingly, despite its inability to trigger a strong (activated) caspase-3-mediated apoptotic program in WM266-4, but not WM115 cells (Figure 8), cisplatin at 50 and 100 µg/mL could markedly reduce the survival of both WM115 (primary) and WM266-4 (metastatic) melanoma cells in a rather identical manner ( Figure 9C). This suggests the implication of non-apoptotic sub-routine(s) in cisplatin-induced death of BRAF V600D primary and metastatic melanoma cells. Accordingly, cisplatin ototoxicity has been mechanistically associated, besides apoptosis, with activation of necroptosis, autophagy, and pyroptosis [324,325]. Thereby, agents that trigger autophagy (or necroptosis) could likely synergize with cisplatin to confer strong chemotherapeutic actions against LC3B-II Low and BRAF V600D -positive metastatic melanomas.
The tolerance of WM266-4 cells to (activated) caspase-3-dependent apoptotic capacity of cisplatin (Figure 8) prompted us to further examine the presumable cytotoxic responses of WM115 and WM266-4 melanoma cells to epirubicin, an Anthracycline family member that blocks the catalytic activity of topoisomerase II and stabilizes DNA breaks, initiating cell death [340][341][342][343][344][345][346]. Surprisingly, WM266-4 melanoma cells presented a notable sensitivity to (24 h) treatment with epirubicin compared to the WM115 respective ones. Even at the lower drug doses of 1, 5, and 10 µM, WM266-4, but not WM115, cell viability was significantly reduced in a dose-dependent manner ( Figure 9F). Altogether, it seems that epirubicin can specifically damage BRAF V600D -positive metastatic (e.g., WM266-4) melanoma cells at clinically low concentrations, thus opening a new therapeutic window for the advanced disease. Since autophagy inhibition augments the anti-cancer effect of epirubicin in breast cancer cells, while autophagy can protect them from epirubicininduced apoptosis [340,347,348], it may be the compromised constitutive autophagy (LC3B-II Low ) (Figure 8) that renders WM266-4 metastatic melanoma cells notably vulnerable to low doses of epirubicin ( Figure 9F). Hence, combination of autophagy inhibitors (e.g., hydroxy-chloroquine) with epirubicin could prove therapeutically beneficial for (LC3B-II High ) melanoma-suffering patients. Strikingly, besides elevated autophagy, increased levels of the SOX2 (NC) stem cell marker are also associated with anthracycline(s) resistance [340,[349][350][351][352]. Given that WM115 cells carry elevated contents of nuclear SOX2 transcription factor (Figure 4), it may be the SOX2 upregulation that critically contributes to primary (e.g., WM115) melanoma cell resistance to low doses of epirubicin ( Figure 9F). Altogether, systemic biomarkering for LC3B-II Low (e.g., reduced constitutive autophagy), SOX2 Low , and BRAF V600D+ , and targeted drugging with low doses of epirubicin can likely emerge as a novel and promising approach for metastatic melanoma therapeutic management in the clinic.

Conclusions
The entire IMC process is exceptionally inefficient, with only a very small percentage of cells leaving primary tumor site(s) to form macroscopic metastasis(es) [37]. To better understand the IMC principles, it is imperative to promptly unveil the biological similarities and differences between primary tumors and their metastatic descendants.
Hence, by using the powerful pre-clinical platform of the primary melanoma cell line WM115 and the metastatic melanoma cell line WM266-4, both derived from the same patient, we have herein described the critical, proteomic-based, systemic biomarkering, and targeted drugging profiles to develop novel strategies for the successful management of advanced disease in the clinic. From the 812 and 3538 proteomic components uniquely identified in the WM115 and WM266-4 human melanoma cells, respectively (n = 3143 common) ( Figure 10A and Tables S2, S3, and S10), 36 and 193 (n = 234 common) are recognized as "EMT program members" (Figure 10B and Tables S11-S13), 68 and 342 (n = 439 common) are categorized as "Cancer-metastasis proteins" (Figure 10C and Tables S13-S15), and 42 and 130 (n = 93 common) are classified as "Stemness-associated components" ( Figure 10D and Tables S13, S16, and S17) (with the first and second, in a row, numerical value of each couple corresponding to WM115 and WM266-4 cells, respectively). Regarding their mutational loads, WM115 cells were specifically presented with 114 "Gene-mutation" and 11 "Gene-fusion" events, while the WM266-4 ones proved to exclusively carry 110 "Mutations" and 128 "Fusions" (n = 421 and n = 4 common, respectively) ( Figure 10E ,F and  Tables S6, S7, S13, S18, and S19), strongly suggesting the major importance of "Gene-fusion" incidents to the IMC process in BRAF V600D human melanoma(s). Similarly, WM266-4 cells were shown to contain 159 major chromosomal abnormalities including "Inversion-like", "Deletion-like", and "Duplication-like" incidents (Table S19) ("Cancer Dependency Map"-"Cancer Cell Line Encyclopedia"-Broad Institute; access day: 7 April 2021), thereby denoting the critical role(s) of chromosomal instability in IMC during BRAF V600D human melanomagenesis.
Phosphorylation-dependent signaling homeostasis is reflected on the 48 "Kinase" and nine "Phosphatase" activity-bearing proteins in WM115, and on the 200 "Kinase" and 78 "Phosphatase" activity-carrying family members in the WM266-4 cells (n = 182 common kinases and n = 58 common phosphatases) ( Figure 10G,H and Tables S13 and S20-S23), thus indicating the essential contribution of major signaling network(s) to melanoma cells undergoing IMC. Transcriptional activity is also upregulated in WM266-4 compared to WM115, since WM266-4 metastatic melanoma cells are shown to contain 199 "Transcription factor activity proteins", while WM115 primary melanoma cells are presented with 42 respective proteomic components (n = 91 common) ( Figure 10I, and Tables S13, S24, and S25). Remarkably, 6 and 20 "Pseudogene-derived (putative) proteins", with presumably new functions, are uniquely recognized in the WM115 and WM266-4 cells, respectively (n = 2 common) ( Figure 10J, and Tables S13, S26, and S27), revealing the pivotal roles of novel proteome members that are encoded by human pseudogenes in the IMC process during BRAF V600D -dependent melanomagenesis.
Most importantly, we herein demonstrate that IMC in mutant BRAF-positive melanoma environments seems to be typified by: (a) hybrid (intermediate) EMT/MET programs, (b) specific (NC-like) stemness sub-routines, (c) re-modeling of cytoskeleton architecture (to support migratory, invasion and colonization actions of tumor cells), (d) elevated constitutive activities of multiple signaling pathways, (e) HIF1α-driven gene expression, (f) attenuated (LC3B-II Low -driven) constitutive (basal) autophagy, (g) (partial) tolerance to genotoxicity, (h) (partial) resistance to (activated Caspase-3-directed) apoptosis, and (i) differential vulnerability to drug-induced cytotoxicity, with epirubicin exhibiting the strongest elimination of BRAF V600D metastatic melanoma cells, dictating the necessity of its prompt utilization in advanced disease therapeutic strategies. Taken together, we herein propose an IMC-dependent protein immunophenotype signature ( Table 1) that seems to hold strong biomarkering and drugging promise for mutant BRAF melanoma management in the clinic.