Similarities and Differences in the Protein Composition of Cutaneous Melanoma Cells and Their Exosomes Identified by Mass Spectrometry

Simple Summary Proteins transferred by tumor-derived exosomes can contribute to cancer progression and/or constitute novel biomarkers of a given disease. Therefore, this study used shotgun nanoLC-MS/MS to obtain complete protein profiles of four cutaneous melanoma cell lines representing different stages of the disease and exosomes released by them. As a result, 3514 and 1234 unique proteins were identified in melanoma cells and exosomes, respectively. Specific alterations to the proteomic profiles associated with disease stages have also been reported, along with a conserved portion of their proteome that may be used by various tumor cells to promote their growth and dissemination. Such a description of the complex composition of cellular and exosomal protein and their related functions provides a deeper insight into the role of exosomes in melanoma progression. The obtained results also indicate some of the exosomal proteins that should be evaluated as potential biomarkers of circulating melanoma. Abstract Intercellular transport of proteins mediated by extracellular vesicles (EVs)—exosomes and ectosomes—is one of the factors facilitating carcinogenesis. Therefore, the research on protein cargo of melanoma-derived EVs may provide a better understanding of the mechanisms involved in melanoma progression and contribute to the development of alternative biomarkers. Proteomic data on melanoma-derived EVs are very limited. The shotgun nanoLC-MS/MS approach was applied to analyze the protein composition of primary (WM115, WM793) and metastatic (WM266-4, WM1205Lu) cutaneous melanoma cells and exosomes released by them. All cells secreted homogeneous populations of exosomes that shared a characteristic set of proteins. In total, 3514 and 1234 unique proteins were identified in melanoma cells and exosomes, respectively. Gene ontology analysis showed enrichment in several cancer-related categories, including cell proliferation, migration, negative regulation of apoptosis, and angiogenesis. The obtained results broaden our knowledge on the role of selected proteins in exosome biology, as well as their functional role in the development and progression of cutaneous melanoma. The results may also inspire future studies on the clinical potential of exosomes.


Introduction
The most aggressive form of skin cancer is malignant melanoma, which originates from transformed melanocytes, i.e., pigment cells of neuroectodermal origin. It is a multifactorial disease driven by both genetic and environmental factors. Melanoma accounts for about 4% of skin cancer cases, but it is responsible for about 75% of all related deaths. The incidence of primary cutaneous melanoma diagnosed annually worldwide has increased

Isolation of Exosomes and Assessment of the Purity of the Exosome Samples
Before exosome isolation, CM cells were kept for 24 h in FBS-free media. Conditioned media were then collected (approx. 200 mL per exosome sample) and centrifuged. Through centrifugations at 400× g (5 min, 4 • C) and 4000× g (20 min, 4 • C), cells and cellular debris were removed; supernatants were then concentrated by a low-vacuum filtration (LVF) procedure described by Drożdż et al. [21]. Concentrated media (approx. 2 mL) were centrifuged three times: at 7000× g (20 min, 4 • C), 18,000× g (20 min, 4 • C), and 80,000× g (20 min, 4 • C), to remove any larger vesicles. The final centrifugation step was performed at 150,000× g (90 min, 4 • C) to obtain exosome pellets. Finally, exosomes were resuspended in ice-cold PBS or in a LC-MS/MS lysis buffer.
The purity of exosome samples was analyzed by nanoparticle tracking analysis (NTA) on NanoSight LM 10 (Malvern Panalytical, Malvern, UK). Five independent records were collected for each sample (10 µL diluted to 2 mL with PBS). The mean results ± SD are presented on graphs.

LC-MS/MS Proteomics
Exosome lysis, sample preparation for mass spectrometric analysis, and LC-MS/MS analysis were performed as described in [17,23] with minor modifications. Namely, the acetonitrile gradient on the analytical column was 240 min, and the flow rate was 250 nL/min. In addition, the Q-Exactive was operated using the top 12 method. Full-scan MS spectra were acquired with automatic gain control (AGC target) of 1 × 10 6 , and the MS/MS spectra were acquired with an AGC target of 5 × 10 5 . The maximum ion accumulation times for the full MS and the MS/MS scans were 120 ms and 60 ms, respectively.

Analysis of Proteomic Data
Raw mass spectra were processed using MaxQuant 2.0.3.1 [24]. Peak lists were searched against the forward and reverse Swissprot_database restricted to Homo sapiens taxonomy (20,376 sequences; downloaded on 5 May 2022) with the use of the integrated Andromeda search engine. Fully tryptic peptides with a maximum of two missed cleavages and with at least seven amino acids were treated as valid. Cysteine carbamidomethylation was set as a fixed modification, whereas variable modifications included methionine oxidation and protein N-terminal acetylation. The precursor mass tolerance in the first search used for mass recalibration was set to 20 ppm. The main search was performed with precursor and fragment mass tolerances of 4.5 ppm and 20 ppm, respectively. The maximum false discovery rate for both peptide and protein identification were set to 0.01. The MaxLFQ label-free algorithm using a minimum ratio count of 2 was used for relative quantification and normalization. Both razor and unique peptides were used for protein quantitation. Separate batches were used to analyze samples from cells and exosomes. The raw data were deposited via the MassIVE repository to the ProteomeXchange Consortium with the dataset identifier PXD038861.
Further analysis was performed on the Perseus platform (version 2.0.7.0) [24]. All contaminants, the proteins from the decoy database, and proteins identified only by modified peptides were excluded from the study. Quantitative analysis was performed on label-free quantification (LFQ) intensities transformed to the logarithmic scale. The student's t-test was then performed with the permutation-based FDR at 0.01 and 0.05 to reveal changes in the protein abundances between different melanoma cell lines and exosomes. Only proteins with at least three valid LFQ intensity values in both compared groups were considered for statistical analysis. Proteins identified by at least two peptides, score > 10, and with a fold change of at least 1.5 (for comparison between cells) or 1.2 (for comparison between exosomes) were considered as differential. Entire protein lists are provided in Supplementary Data S2. Complete quantitative analysis is provided in Supplementary Data S3.

Bioinformatic Analysis
The final protein lists contained proteins identified by at least two peptides in three or four out of four biological repetitions of cellular or exosome samples. Venn diagrams, including Vesiclepedia protein overlap, and gene ontology (GO) analysis were performed with FunRich 2.0 software using the UniProt (release 2022_11) database as a reference. For each GO term, six categories with the highest protein enrichment were presented as graphs. Additionally, enrichment within selected cancer-related categories (calculated as −log10(p-value)) was compared between CM cells and exosome samples from primary and metastatic cells. Entire protein lists and GO data are included in Supplementary Materials Data S1. Interaction diagrams from Appendix A and Supplementary Data S3 were prepared with the use of https://string-db.org/ (accessed on 2 December 2022) Version: 11.0.

Characterization of CM Exosome Samples
In the present study, a complex exosome isolation protocol based on sequential centrifugation and low vacuum filtration (LVF) was applied [21]. Nanoparticle tracking analysis (NTA) showed that the final 150,000× g exosome pellets contained mostly <100-nmdiameter vesicles (typical range of exosomes), with less than 5% being >100-nm-diameter vesicles ( Figure 1).
Additionally, enrichment of exosomal protein markers (CD9, CD63, and Hsp70) was confirmed in each exosome sample compared to whole-cell protein extracts ( Figure 2). On the other hand, exosome samples were depleted of Arf6, the protein involved directly in ectosome biogenesis but not in exosome biogenesis. The evidence presented allows us to consider the isolated EV samples to be highly enriched in exosomes and the ectosome contamination to be negligible.

Identified Proteins of CM Cells and Exosomes and Their Functional Classification
Protein profiles for four CM cell lines and exosomes released by these cells were obtained using the gel-free nanoLC-MS/MS shotgun proteomic approach. Four biological replicates of each cellular/exosomal sample were analyzed, and only proteins identified in at least three replicates (and with at least two peptides) were considered for further analyses. Variability of protein detection across replicates was presented ( Figure A1). A total of 3514 proteins were identified for all CM cell lines analyzed ( Figure 3A, complete protein lists in Supplementary Data S2). In terms of individual cell lines, a greater number of proteins were identified for primary WM793 and WM115 cells compared to their isogenic metastatic equivalents, i.e., WM1205Lu and WM266-4 cells, respectively. In addition, 2114 proteins (approx. 60% of all identified proteins) were identified for all CM cell lines analyzed ( Figure 3B). More than 70% similarity in protein composition was observed for all possible pairings of cell lines analyzed ( Figure 3C-E). The greatest similarity was observed  Immunodetection of extracellular vesicle (EV) markers in whole-cell (lines C) and exosome samples (lines E). Prior to immunodetection, 50 μg of prot separated by 10% SDS-PAGE and transferred into the PVDF membrane. Immun performed with the use of the following primary antibodies: anti CD9, anti-CD63, anti-Arf6, and anti-mouse IgG-HRP as a secondary antibody. Original blots w Supplementary Data S1. WM115/WM266-4,WM793/WM1205Lu-isogenic ry/metastatic) of CM cell lines that were used for exosome isolation.

Identified Proteins of CM Cells and Exosomes and Their Functional Classific
Protein profiles for four CM cell lines and exosomes released by these tained using the gel-free nanoLC-MS/MS shotgun proteomic approach. F replicates of each cellular/exosomal sample were analyzed, and only prot in at least three replicates (and with at least two peptides) were conside analyses. Variability of protein detection across replicates was presented total of 3514 proteins were identified for all CM cell lines analyzed (Figure protein lists in Supplementary Data S2). In terms of individual cell lines, a ber of proteins were identified for primary WM793 and WM115 cells com isogenic metastatic equivalents, i.e., WM1205Lu and WM266-4 cells, respe dition, 2114 proteins (approx. 60% of all identified proteins) were identif cell lines analyzed ( Figure 3B). More than 70% similarity in protein compo served for all possible pairings of cell lines analyzed ( Figure 3C-E). The gr ty was observed between isogenic pairs, i.e., WM115 and WM266-4 (83 proteins), and WM793 and WM1205Lu (78.3% of shared proteins) cells, w Figure 2. Immunodetection of extracellular vesicle (EV) markers in whole-cell protein extracts (lines C) and exosome samples (lines E). Prior to immunodetection, 50 µg of protein per line was separated by 10% SDS-PAGE and transferred into the PVDF membrane. Immunodetection was performed with the use of the following primary antibodies: anti CD9, anti-CD63, anti-Hsp70, and anti-Arf6, and anti-mouse IgG-HRP as a secondary antibody. Original blots were provided in Supplementary Data S1. WM115/WM266-4,WM793/WM1205Lu-isogenic pairs (primary/metastatic) of CM cell lines that were used for exosome isolation.
A total of 1234 unique proteins were identified in CM-derived exosomes ( Figure 4A, complete lists of proteins are given in Supplementary Data S2). Exosomes released by primary WM793 and WM115 cells were characterized by a greater number of identified proteins than their isogenic metastatic pairs-WM1205Lu and WM266-4, respectively. It is possible that metastatic cell lines retain more proteins (rather than secrete them) due to their overactive metabolism. Exosomes derived from primary WM115 cells had the highest number of proteins identified, 1067, while exosomes derived from metastatic WM1205Lu cells had the lowest number of proteins, 571. The number of unique proteins for a given exosome sample was between 19 (WM1205Lu-derived exosomes) and 264 (WM115 exosomes). In contrast, 417 proteins were identified in exosomes from all CM cell lines ( Figure 4B), accounting for 33.8% of all identified exosomal proteins compared to approximately 60% for cell-derived samples. The highest similarity of the proteome was observed between exosomes derived from two primary cell lines, i.e., WM793 and WM115 (61.2%), and the lowest for WM1205Lu and WM115 exosomes (45.7%) ( Figure 4C-E). In addition, the similarity was high for exosomes of metastatic origin (WM1205Lu and WM266-4), i.e., 57.7%. This suggests that the protein composition of exosomes may be more dependent on the disease stage than on the genetic background of the parental cell (which was observed for CM cell samples). Additionally, all proteins identified in exosome samples were searched against the Vesiclepedia database ( Figure 4F), which collects proteomic data from multiple EV-oriented studies. There was a significant overlap between the proteins identified in the present study and the Vesiclepedia dataset. which supports their vesicular origin instead of being part of cell-derived contamination. A total of 1234 unique proteins were identified in CM-derived exosomes ( Figure  4A, complete lists of proteins are given in Supplementary Data S2). Exosomes released by primary WM793 and WM115 cells were characterized by a greater number of identified proteins than their isogenic metastatic pairs-WM1205Lu and WM266-4, respectively. It is possible that metastatic cell lines retain more proteins (rather than secrete them) due to their overactive metabolism. Exosomes derived from primary WM115 cells had the highest number of proteins identified, 1067, while exosomes derived from metastatic WM1205Lu cells had the lowest number of proteins, 571. The number of unique proteins for a given exosome sample was between 19 (WM1205Lu-derived exosomes) and 264 (WM115 exosomes). In contrast, 417 proteins were identified in exosomes from all CM cell lines ( Figure 4B), accounting for 33.8% of all identified exosomal proteins compared to approximately 60% for cell-derived samples. The highest similarity of the proteome was observed between exosomes derived from two primary cell lines, i.e., WM793 and WM115 (61.2%), and the lowest for WM1205Lu and WM115 exosomes (45.7%) (Figure of the parental cell (which was observed for CM cell samples). Additionally, all proteins identified in exosome samples were searched against the Vesiclepedia database ( Figure  4F), which collects proteomic data from multiple EV-oriented studies. There was a significant overlap between the proteins identified in the present study and the Vesiclepedia dataset. which supports their vesicular origin instead of being part of cellderived contamination.  Furthermore, the proteomes of CM cells and exosomes were compared ( Figure 5). For each cell line, from ca. 17% to almost 30% of their proteins were also identified in exosomes. This suggests that the sorting of protein into exosomes is a highly regulated process involving only a very specific set of proteins. On the other hand, between 16% to 20% of proteins identified in exosome samples were not identified in cells. Since exosomal proteins must be derived from the cell, the lack of identification in CM cells may be due to their low abundance in total cellular proteomes and sufficient enrichment in exosomes. tween exosomes derived from isogenic (D) and primary or metastatic CM cell lines (E). (F) Venn diagram illustrating protein overlap between isolated ectosomes and Vesiclepedia database as a reference. WM115/WM266-4,WM793/WM1205Lu-isogenic pairs (primary/metastatic) of CM cell lines that were used for exosome isolation. Complete protein lists are presented in Supplementary Data S2. Furthermore, the proteomes of CM cells and exosomes were compared ( Figure 5). For each cell line, from ca. 17% to almost 30% of their proteins were also identified in exosomes. This suggests that the sorting of protein into exosomes is a highly regulated process involving only a very specific set of proteins. On the other hand, between 16% to 20% of proteins identified in exosome samples were not identified in cells. Since exosomal proteins must be derived from the cell, the lack of identification in CM cells may be due to their low abundance in total cellular proteomes and sufficient enrichment in exosomes.  Gene ontology (GO) analysis was then performed to group proteins identified in CM cells and exosomes by the cellular compartment, molecular function, and biological processes. The six categories with the highest statistical significance of protein enrichment within the category (p < 0.001) were selected for each analysis. Complete protein lists are shown in Supplementary Data S2. Similar GO patterns were observed for each CM cell line ( Figure 6). The most numerous groups of proteins were associated with the cytoplasm (up to 62.7% of identified proteins). Moreover, up to 34.5% of proteins identified in CM cells were connected to exosomal origin. GO analysis of molecular function showed that the identified proteins were mainly associated with catalytic (up to 6.2%) or transporter activity (up to 5.9%), and RNA binding (up to 6.1%). A significant abundance of structural proteins were also identified. Furthermore, the analysis of the biological function of cellular proteins pointed out that identified proteins are mostly involved in metabolism, especially protein metabolism, and in various energy pathways.  GO analysis of CM-derived exosomal proteins showed that up to 73.6% of the identified proteins had exosomal origin and up to 67.1% were associated with the cytoplasm (Figure 7). This observation is consistent with exosome biogenesis, which involves blebbing of endosomal membranes and creating multivesicular bodies, which later fuse with the cell membrane and give rise to exosomes. In addition, GO analysis for the molecular function revealed the greatest enrichment in categories such as GTPase, chaperone, and translation regulatory activity. CM exosomal proteins function as structural constituents of ribosomal and extracellular matrices. In terms of biological processes, the proteins identified were involved in cell communication (over 23%), signal transduction (over 24%), protein metabolism (over 18%), and cell growth and/or maintenance (over 13%). Metastatic WM266-4-and WM1205Lu-derived exosomes showed higher enrichment in several categories than their primary counterparts, including cell growth and/or maintenance, cell communication, signal transduction, GTPase activity, etc. This may reflect higher prometastatic potential compared to exosomes from primary cells.

Functional Similarities and Differences in the Protein Composition of CM Cells and CM-Derived Exosomes
An increasing number of studies are focusing on the role of EVs in various diseases, including cancer. Tumor-derived EVs facilitate the transfer of biologically active molecules that can regulate the function of the recipient cell at various levels. To gain insight into how CM-derived exosomes might affect recipient cells, GO analysis was performed using the lists of all proteins identified in CM cells (3514 proteins) and CM-derived exosomes (1234 proteins). Selected cancer-related GO categories and their enrichment significance score (i.e., −log10(p-value)) were presented in Figure 8.
Considering categories with p < 0.001, the analyzed CM cells were enriched in proteins related to antigen processing and presentation, negative regulation of apoptosis, platelet aggregation, and glycolysis. All these processes are enhanced in cancer cells, which tend to avoid immune surveillance, escape apoptosis, and contribute to the prothrombotic state, as well as rely on aerobic glycolysis for ATP generation. Numerous other categories for CM cells included several signaling pathways, i.e., the NIK/NF-kappaB pathway, the Wnt pathway, the tumor necrosis factor-mediated pathway, and the vascular endothelial growth factor receptor (VEGFR) signaling pathway. These are all known to be altered in cancer and involved in key steps in the metastatic cascade, such as angiogenesis.
Since exosomes (and other EVs) largely reflect the composition of parental cells, CMderived exosomes isolated in the present study also showed enrichment in selected cancerrelated GO categories. Notably, for most categories, enrichment was more significant for CM-derived exosomes than for CM cells. This includes categories that were not enriched (p > 0.001) for CM cells, such as cell growth, cell adhesion, cell migration, response to hypoxia, blood coagulation, or angiogenesis. In addition, CM-derived exosomes showed enrichment in categories related to immune cell migration, suggesting their involvement in the modulation of the immune response against melanoma tumors. Finally, the GO category "drug response" was enriched for CM-derived exosomes, indicating their possible role in the phenomenon of multidrug resistance.  An increasing number of studies are focusing on the role of EVs in various diseases, including cancer. Tumor-derived EVs facilitate the transfer of biologically active molecules that can regulate the function of the recipient cell at various levels. To gain insight into how CM-derived exosomes might affect recipient cells, GO analysis was performed using the lists of all proteins identified in CM cells (3514 proteins) and CM-derived exosomes (1234 proteins). Selected cancer-related GO categories and their enrichment significance score (i.e., −log10(p-value)) were presented in Figure 8.  To characterize the changes in the cellular and exosomal proteome that occur during disease progression, the proteins identified in CM cells and CM-derived exosomes were qualitatively and quantitively compared within isogenic pairs (same donor) of cell lines (that is, WM793 vs. WM1205Lu and WM115 vs. WM266-4, respectively). Table 1 lists the proteins that were identified by all four replicates in WM793 (79 proteins) but not in 1205 Lu cells and vice versa (38 proteins). Smaller discrepancies in the number of unique proteins were observed for WM115 and WM266-4 cells, i.e., 31 and 40 proteins, respectively ( Table 2). Considering exosomes, 58 proteins were identified in WM793-derived exosomes but not in WM1205Lu-derived exosomes, with only two proteins unique to WM1205Lu-derived exosomes (Table 3). Similarly for the second isogenic pair, exosomes derived from WM115 cells contained more unique proteins (79 proteins) than exosomes derived from WM266-4 (34 proteins) ( Table 4). This suggests that as the disease progresses, tumor cells tend to sort fewer proteins into exosomes, possibly to preserve/increase their own metastatic potential. Another possibility is that metastatic cells sort fewer proteins into exosomes but increase the abundance of proteins crucial for cancer progression.    In addition, the STRING v. 11.0 platform was used to prepare diagrams of functional protein association networks for proteins unique to given CM cell lines. The unique proteins from WM793 cells were shown to be enriched in, among others, the PPAR signaling pathway and vitamin D receptor pathway proteins ( Figure A2). In the case of WM1205Lu cells, a unique protein enrichment in proteins with a Pleckstrin homology domain (PH domain) was observed ( Figure A3). The PH domain is a protein domain found in many proteins involved in intracellular signaling or as components of the cytoskeleton. The unique proteins from WM115 cells were enriched in proteins related to the MHC class II protein complex responsible for presenting tumor antigens to the immune system ( Figure A4). This corresponds to the primary character of WM115 cells, when the cells are not yet completely out of immune surveillance, as in the later stages of the disease. Finally, no enriched categories were found for the unique WM266-4 proteins.
Similar STRING diagrams were prepared for exosomal proteins. Proteins present in WM793-derived exosomes but absent in WM1205Lu-derived exosomes were enriched in proteins related to signaling receptor binding and mRNA processing ( Figure A5). As only two proteins are unique for WM1205Lu, a similar analysis could not be performed. Furthermore, proteins identified in WM115-derived exosomes but absent in WM266-4-derived exosomes showed enrichment in categories related to cell adhesion molecule binding, RNA binding, translation, and peptide metabolic processes ( Figure A6). For proteins identified in WM266-4-derived exosomes but absent in WM115-derived ones, similar generic categories were enriched, i.e., extracellular matrix structural constituent, extracellular matrix organization, glycosaminoglycan binding, and cell adhesion ( Figure A7).
Label free quantification (LFQ) was also performed to determine differentially expressed proteins within isogenic pairs of CM cell lines and CM-derived exosomes. Proteins with fold change > 1.5 were considered upregulated in CM cells (p < 0.001), while for exosomes, a fold change of >1.2 and p < 0.05 were applied as threshold values. In summary, 243 proteins were upregulated in WM793 cells compared to WM1205Lu cells, and 185 were upregulated in WM1205Lu cells compared to WM793 cells. For the second isogenic pair of CM cell lines, WM115 cells contained 116 proteins that were upregulated compared to WM226-4 cells, while 160 proteins were found to be upregulated in WM266-4 compared to WM115 cells. Regarding exosomes, no differences in protein expression were observed between the proteins common to WM793-derived and WM1205Lu-derived exosomes. However, a total of 322 proteins were upregulated in WM115-derived exosomes compared to WM266-4-derived exosomes, while 77 proteins were upregulated in WM266-4-derived exosomes compared to WM115-derived exosomes. In summary, the ten proteins with the highest fold change for CM cell lines and exosomes are listed in Table 5, while the entire LFQ analysis is presented in Supplementary Data S3.  The entire lists of upregulated proteins were later subjected to STRING analysis. For proteins upregulated in CM cell lines, the enriched categories included quite generic terms related to cell metabolism, cell adhesion, or RNA binding (Figures A8-A11). Interestingly, proteins upregulated in primary melanoma WM793 cells (compared to WM1205Lu cells) were enriched in the category of "abnormality of acid-base homeostasis". This may be related to tumor acidosis, which often occurs within growing primary tumors until they develop a vascular network that increases the oxygen supply and restores normal pH levels. Regarding angiogenesis, proteins upregulated in WM266-4 cells were enriched in proteins involved in the vascular endothelial growth factor A/vascular endothelial growth factor receptor 2 (VEGFA-VEGFR2) signaling pathway. Signaling via VEGF induces the migration of endothelial cells during angiogenesis and may enhance microvascular permeability during tumor metastasis. Proangiogenic proteins were also upregulated in WM115-derived exosomes (category "VEGFA-VEGFR2 signaling pathway") and WM266-4-derived exosomes (category "Tube development") ( Figures A12 and A13). This suggests that exosomes released by primary tumor cells may be involved in proangiogenic signaling during the induction of angiogenesis. On the other hand, exosomes released by a metastatic tumor may contain proteins involved in the later stages of blood vessel formation such as sprouting or vessel maturation.

CM-Derived Exosomes Are Enriched in Proteins with Functional Implications in Cancer
An increasing amount of evidence confirms that CM-derived exosomes increase the invasive potential of various recipient cells and drive the metastatic spread of melanoma tumors. An increased invasiveness after treatment with CM-derived exosomes was already demonstrated for melanocytes [25] as well as bone marrow-derived stromal cells [26]. Consequently, in the present study, CM-derived exosomes were more enriched in proteins involved in positive regulation of cell migration than CM cells (Figure 8) based on GO analysis. It suggests that proteins with a promigratory function are preferentially sorted into exosomes by CM cells.
Spreading of CM cells is connected to the activity of matrix metalloproteinases (MMPs), and positive correlation between the concentration of exosomes, the number of lytic enzymes in their cargo, and their pro-invasive capabilities have been demonstrated [27,28]. CM-derived EVs have already been shown to carry several MMPs and their endogenous activator CD147 [29][30][31][32][33]. In the present study, MMP1 was identified only in WM793 cells, while MMP14 was found in WM266-4 cells and all exosome samples except WM1205Luderived exosomes. In contrast, CD147 and other tissue inhibitors of MMPs (TIMP1, TIMP2, TIMP3) were present in all exosome samples, suggesting that CM exosomes may not always transfer MMPs but that they have a regulatory role towards MMP activity.
Regarding proteins involved in CM metastasis, c-Met and Rab27a proteins were shown to be transferred via exosomes [34,35], and the knockdown of Rab27a decreased invasiveness and metastasis of CM cells as well as exosome secretion [34]. A more recent proteomic study revealed enrichment of CM-derived exosomes in proteins belonging to the prometastatic NRAS, SRC, KIT, EGFR, and MET signaling pathways [36]. In the present study, Rab27 was identified in WM115 and WM266-4 cells, but it did not appear in exosome samples. On the other hand, Src kinase, NRAS GTPase, and the epidermal growth factor receptor (EGFR) were identified in all exosome samples, but only Src kinase was identified in CM cells. The abundance of NRAS and EGFR in CM cells may not be high enough for them to be identified among more abundant cellular proteins. Furthermore, NRAS and EGFR are likely to be enriched in CM-derived exosomes, demonstrating their role in providing key players in prometastatic signaling pathways.
In addition, there is substantial evidence that CM-derived exosomes may contribute to organotropisms during CM metastatis. Several in vivo studies have reported the involvement of CM exosomes in the organ-specific formation of secondary tumors in the lungs [37], bone [38], and sentinel lymph nodes [39]. CM-derived exosomes may also contribute to the formation of brain metastasis. Using a biomimetic blood-brain barrier (BBB) model (co-culture of brain microvascular endothelial cells, astrocytes, and microglial cells), CMderived exosomes have been shown to induce endothelial damage, disrupt BBB integrity, and induce glial activation [40]. Finally, CM-derived exosomes upregulated proteins from the MAPK signaling pathway in primary melanocytes to induce epithelial-mesenchymal transition (EMT) and promote the metastatic phenotype [41]. In the present study, MAPK1, MAPK2, MAPK3, and MAPK14 were identified in CM cells, while MAPK1 and MAP4K4 were present in most of the CM-derived exosome samples.
Exosomes also participate in the delivery of proangiogenic factors or increase their expression in recipient cells. A study by Hood et al. showed that CM-derived exosomes induce the formation of endothelial spheroids [42]. Another study showed that CM cells release exosomes enriched in interleukin 8 (IL-8), vascular endothelial growth factor (VEGF), MMP2, and IL-6 [43]. CM-derived exosomes containing the urokinase plasminogen activator receptor (uPAR) were shown to promote angiogenesis in recipient endothelial cells by upregulating VE-Cadherin, EGFR, and uPAR expression, and enhancing ERK1,2 signaling [44]. Alternatively, activation of the JAK-STAT pathway and enhanced angiogenesis were observed in endothelial cells after treatment with CM-derived exosomes [45]. In the present study, CM-derived exosomes were more enriched in proteins involved in angiogenesis than CM cells (Figure 8), including proteins from the VEGF signaling pathway. However, no VEGF was found in any sample. Instead, other proangiogenic factors were identified in both CM cells and CM-derived exosomes such as neuropilin 1, annexin A2, or integrin subunits (α5, αV).
CM-derived exosomes can also induce lymphangiogenesis and thus contribute to lymph-node metastasis [46,47]. Potential mechanisms include ERK kinase induction, nuclear factor (NF)-κB activation, and increased expression of intracellular adhesion molecule (ICAM)-1 expression in lymphatic endothelial cells [47]. Importantly, it has been shown that exosomal transfer of the nerve growth factor receptor is responsible for the aforementioned effects. In the present study, NGFR was identified in WM115 exosomes, supporting the hypothesis that primary melanoma tumors may release NGFR-containing exosomes to enhance formation of lymph-node metastasis.
In addition to endothelial cells, exosomes can modulate the function of other cells present in the tumor microenvironment such as fibroblasts and immune cells. In a recent study, cancer-associated fibroblasts were activated by the CM-derived exosomes to a greater degree than normal fibroblasts for the transcription of genes for proinflammatory cytokines and chemokines, mainly IL-6 or IL-8 [48]. CM exosomes can also induce proinflammatory polarization and activation of macrophages [49,50]. They can also impair the function [51,52] or induce apoptosis [46] of CD8+ cytotoxic T-cells to locally suppress anti-tumor cytotoxicity. On the other hand, CM exosomes can directly activate CD4+ helper T-cells through the transfer of miR690 and Rab27a [53]. Finally, CM exosomes were shown to reduce the differentiation of bone marrow-derived dendritic cells [54]. In the present study, multiple GO categories related to immune response were enriched for CM cells and CM-derived exosomes. Specifically, CM exosomes were enriched in proteins involved in leukocyte and neutrophil migration (Figure 8). This suggests the involvement of CM-derived exosomes in either enhancing or inhibiting the influx of immune cells into melanoma tumors and the subsequent immune response.

Clinical Relevance of Proteins Identified in CM-Derived Exosomes
It has been shown that the number and/or composition of exosome proteins changes when the parental cell undergoes neoplastic transformation. Cancer cells are known to secrete more exosomes than normal cells, so there is an interest in using exosome concentration as a biomarker. A significantly higher concentration of released exosomes was observed for CM cells compared to normal melanocytes [25], and for metastatic CM cells compared to primary CM cells [55]. However, in the present study, NTA showed a rather similar concentration of exosomes in samples from four CM cell lines (approx. 5 × 10 7 particles/mL) after taking equal volumes of conditioned media for exosome isolation. This suggests that sole exosome concentration may not be a good indicator of disease stage, as opposed to being a marker of disease occurrence.
Currently, lactate dehydrogenase A (LDH) or S100 calcium binding protein B (S100B) are the two most important clinical markers of CM. Serum levels of S100B increase with tumor growth and are used to monitor patients in advanced stages of the disease [56,57]. In the present study, S100B was identified in all CM cell samples and in exosomes derived from WM793 and WM115 cells. The lack of S100B in exosomes derived from metastatic WM1205Lu and WM266-4 cells may be potentially useful in discriminating between primary and metastatic CM. Interestingly, in our previous study, S100B was present in ectosomes from all CM cell lines, regardless of disease stage [33]. This suggests that the diagnostic target for S100B protein should not be ectosomes but rather exosomes. Finally, all CM cell and CM exosome samples contained LDH, which is now being used in clinical practice as a predictor of CM patient survival [58].
Recently, EV-oriented studies have moved forward in the CM biomarker field. First, the levels of MET, dopachrome tautomerase (TYRP2), integrin α4β1 (VLA-4), Hsp-90, and Hsp-70 were found to be upregulated in plasma-derived exosomes of CM patients compared to healthy controls [34]. In the present study, Hsp-90, Hsp-70, and α4 integrin subunits were identified in all samples, while TYRP2 was found only in WM115 and WM266-4 cells. Nevertheless, none of these proteins displayed differential expression based on LFQ; therefore, their biomarker potential needs further evaluation. Other studies have shown increased levels of CD63 tetraspanin in exosomes from both CM cell cultures (compared to normal cells) [59] and plasma of CM patients (compared to healthy controls) [60]. However, in the present study, CD63 expression was not significantly different between samples. Exosomes can also be isolated from less obvious sources, such as fluid from lymphatic drainage of melanoma tumors. It was shown that such exosomes are enriched in Braf protein with V600E mutation, and its level was correlated with the risk of disease recurrence [61]. Here, B-raf protein was identified in samples from all CM cells, but it was not detected in CM-derived exosomes, possibly due to the source of isolated EVs.
Exosomal proteins are also potential biomarkers of treatment response. In melanoma patients responding positively to treatment with nivolumab and pembrolizumab (combination of antibodies against programmed cell death protein 1/programmed cell death protein ligand 1 (PD1/PD-L1)), a significant decrease in exosomal PD-L1 expression was observed [62,63]. In the present study, neither PD1 nor PD-L1 was identified in CM cells and exosomes. However, several immunosuppressive proteins from the family of programmed cell death protein were identified in CM cells, such as PD3, PD5, PD6, PD10, PD11, and PD4-a potent inhibitor of neoplastic transformation. On the other hand, CM-derived exosomes (besides the WM115-derived sample) contained only PD6 and PD10, suggesting that this group of proteins is not preferentially carried by exosomes. In addition, exosomal CD73 ectonucleotidase (producing adenosine-suppressor of T-cell function) has been found to increase in CM patients unresponsive to treatment with anti-PD-1 agents [64]. In addition, Pietrowska et al. identified 75 proteins upregulated in plasma-derived exosomes from CM patients with progressive disease compared with patients showing no evidence of CM after therapy. Programmed cell death 6-interacting protein (PDCD6IP) showed the highest upregulation in exosomes from patients with disease progression, while contactin-1 (CNTN1) was upregulated in exosomes from patients in remission [65]. CD73 and PDCD6IP were also present in all exosome samples in the present study, demonstrating their potential as a treatment response biomarker in CM.
Finally, a qualitative and quantitative comparison of CM cells and CM-derived exosomes in isogenic pairs of cell lines (WM793 vs. WM1205Lu and WM115 vs. WM266-4) revealed many more potential targets for protein biomarkers than mentioned in Section 4. This includes unique proteins (Tables 1-4) or proteins with differential expression (Table 5). Focusing on CM-derived exosomes, differentially expressed proteins were identified only between WM115-derived and WM266-4-derived exosomes. Among proteins with the greatest fold change (Table 5), there are several with well-known implications in melanoma progression. For example, annexin A1 (upregulated in WM115-derived exosomes) is a known promoter of primary melanoma tumor dissemination, mainly by inhibiting E-cadherin expression [66,67]. Additionally, A-kinase anchor protein 12 (upregulated in WM115-derived exosomes) shifts PKA-mediated protein phosphorylation to increase CM cell migration and metastasis [68]. Regarding proteins upregulated in WM266-4-derived exosomes, tenascin mediates protective signals in therapy-resistant melanomas by downregulation of multiple ATP-binding cassette transporters [69]. Therapeutically, upregulation of adipocyte enhancer-binding protein 1 (also upregulated in WM-266-4-derived exosomes) has also been shown to contribute to resistance to BRAF (V600E) inhibition in the treatment of melanoma [70]. Finally, semaphorin 5A is known to regulate melanoma cell migration and invasion, and angiogenesis [71,72].
Nevertheless, larger preclinical and clinical trials are required to further validate any novel CM biomarker identified in exosomes in vitro. The clinical potential of exosomes is based on the development of efficient isolation protocols. Such protocols must provide highly concentrated, uncontaminated EV samples, optimally preserving the native form and function. Based on the promising results from the present study, the search for biomarkers for CM should continue. Similar high-throughput proteomic approaches could help overcome the current lack of effective diagnostic and prognostic tools.

Conclusions
Despite advances in diagnostics and treatment, patients with CM still face a poor prognosis. Specific alterations to the proteomic profiles of CM cell lines representing different stages of disease and exosomes derived from those cell lines were reported in the present study. In addition to these unique features of CM cells and exosomes, a conserved part of their proteome was also described, which can be used by various tumor cells to promote their growth and dissemination. Our description of the complex composition of cellular and exosomal proteins and their related functions provides deeper insight into the role of exosomes in CM progression.
Our results also point to some of the exosomal proteins to be evaluated as potential circulating CM biomarkers. This includes the most commonly used CM markers such as LDH and S100B. Additionally, all unique proteins (Tables 3 and 4) or proteins with the most differential expression between exosomes from primary and metastatic CM cells (Table 5)-such as annexin A1 [66,67], A-kinase anchor protein 12 [68], tenascin [69], adipocyte enhancerbinding protein 1 [70], and semaphorin 5A [71,72]-should be evaluated. In addition, some proteins with biomarker potential assessed by other groups were also detected in CM exosomes in the present study. This includes proteins with diagnostic potential (α4β1 integrin, Hsp-90, and Hsp-70 [34]) or those that can discriminate patients at different stages (GTPase HRas, cofilin-2, hypoxia upregulated protein 1, Hsp90B1, Hsp90AB1, and HspA5 [55]). Finally, exosomal CD73 [64] and PDCD6IP [65] appear to be promising potential treatment response biomarkers in CM.        Figure A7. Diagram of functional protein association networks prepared with the use of STRING v. 11.0 software for proteins identified in WM266-4-derived exosomes but not found in any replicates of WM115-derived exosomes. The selected, strongly represented pathways are presented as an interactome. A complete analysis is presented in Supplementary Data S3.  Figure A10. Diagram of functional protein association networks prepared with the use of STRING v. 11.0 software for proteins upregulated in WM115 cells vs. WM266-4 cells. The selected, strongly represented pathways are presented as an interactome. Proteins with fold change > 1.5 were considered upregulated in WM115 cells (p < 0.001). A complete analysis is presented in Supplementary Data S3. Figure A11. Diagram of functional protein association networks prepared with the use of STRING v. 11.0 software for proteins upregulated in WM266-4 cells vs. WM115 cells. The selected, strongly represented pathways are presented as an interactome. Proteins with fold change > 1.5 were con-