The human melanoma proteome atlas—Defining the molecular pathology

Abstract The MM500 study is an initiative to map the protein levels in malignant melanoma tumor samples, focused on in‐depth histopathology coupled to proteome characterization. The protein levels and localization were determined for a broad spectrum of diverse, surgically isolated melanoma tumors originating from multiple body locations. More than 15,500 proteoforms were identified by mass spectrometry, from which chromosomal and subcellular localization was annotated within both primary and metastatic melanoma. The data generated by global proteomic experiments covered 72% of the proteins identified in the recently reported high stringency blueprint of the human proteome. This study contributes to the NIH Cancer Moonshot initiative combining detailed histopathological presentation with the molecular characterization for 505 melanoma tumor samples, localized in 26 organs from 232 patients.


HIGHLIGHT
• Protein expression combined with indepth histopathological characterization is determined for a broad spectrum of primary and metastatic melanoma tumors isolated from multiple body locations.
• Mass spectrometry analysis provides identification of more than 15,500 proteoforms subsequently annotated to chromosomal and subcellular localization.
• The presented melanoma tumor protein blueprint cover 72% of proteins currently identified in the human proteome.

Graphical Abstract
The MM500 study is an initiative to map the protein levels in malignant melanoma tumor samples, focused on in-depth histopathology coupled to proteome characterization. The protein levels and localization were determined for a broad spectrum of diverse melanoma tumors originating from multiple body locations. More than 15,500 proteoforms were identified by mass spectrometry, from which chromosomal and sub-cellular localization was annotated within both primary and metastatic melanoma.

INTRODUCTION
Metastatic malignant melanoma carries a poor prognosis, however, surgical intervention of the primary melanoma is curative in most patients, which underlines the importance of early diagnosis. In 2020, Globocan reported 324,635 new cases and 57,043 deaths (https://gco.iarc.fr) 1 from melanoma worldwide. In many European countries, melanoma is increasing at a rate of 3-7% and this figure is expected to rise further. [2][3][4] Approximately 120,000 new cases in the United States were expected to be diagnosed in 2020, with about 7000 patients dying from the disease. In Sweden, melanoma now ranks number 5-6 among cancers (incidence: 38/100,000 in 2019) and the 10-year survival approaches 90% as the proportion of thinner melanomas with an extremely good prognosis, increases. 5,6 Metastatic melanoma (MM) used to have an extremely poor prognosis until 10 years ago, since then the development of modern drugs modulating the immune response or targeting specific cellular signaling events has prolonged survival from months to years for many patients. [7][8][9] However, some patients with MM do not fully respond or develop drug resistance to the novel treatments why there is still need for more mechanistic disease knowledge. [10][11][12] Clonal diversity and genetic inter-and intratumor heterogeneity in MM are in particular, poorly understood. [13][14][15][16] Currently, there is only one clinically validated predictive marker (BRAFV600E) for molecular targeted therapy. [17][18][19][20] Clinically useful markers are lacking for deciding on sequential administration of drugs in firstline management (targeted vs. immune checkpoint therapy). [21][22][23] Cutting-edge technologies (such as proteomics, single-cell transcriptomics, spatial imaging, machine learning, etc.) have been developed to identify novel biomarkers and will contribute to building clinical prognostic and predictive useful information. [24][25][26][27] For example genomic research has led to major improvements in the diagnosis and treatment of cancer and the foundation of precision oncology. Nevertheless, the challenge today is still an unmet need of information guiding the daily treatment of patients. The proteome, which is defined as the entire set of proteins of a cell, tissue, organ, or body fluid in time, space as well as in a stimuli-dependent stage is composed of proteoforms that are expressed in response to specific intrinsic and extrinsic signals. Moreover, proteins are the true functional molecular species in the cell, important both for disease modulation and drug interactions. Proteomics is defined as the characterization of proteins encoded by the genome of a given organism at a given time in a given state. 28 By mass spectrometry (MS) a sensitive analysis of complex mixtures of proteins and peptides is enabled. MS-based techniques provide a means of identifying and quantifying proteins, studying post-translational modifications, protein interactions, and localization. Thus, identification of protein biomarkers could be used to evaluate the clinical course of the disease and identify protein signatures related to the effects of treatment. [29][30][31] For example, some recent studies revealed that heterogeneous expression of the BRAF V600E mutated protein in melanoma patients was related to survival. 15,32 In the present study, the proteomic map of primary and MM was determined and combined with in-depth pathophysiological and clinical characterization of the tumors. The MM500 study (505 samples from 294 tumors obtained from 232 patients) utilizes samples from both Sweden (BioMel biobank, SUS University Hospital Lund) and Hungary (Semmel study, Semmelweis University Hospital and Melszeg study, Szeged University Hospital). The

HIGHLIGHT
• Protein expression combined with in-depth histopathological characterization is determined for a broad spectrum of primary and metastatic melanoma tumors isolated from multiple body locations. • Mass spectrometry analysis provides identification of more than 15,500 proteoforms subsequently annotated to chromosomal and subcellular localization. • The presented melanoma tumor protein blueprint cover 72% of proteins currently identified in the human proteome.
overall aim of the MM500 study is to identify and define the overall melanoma proteome including the intracellular localization of expressed proteins. By combining this information with genome and transcriptome sequencing, digital pathology (DP), and clinical information, this study will serve as a roadmap to find new markers for prognostic and predictive use in malignant melanoma. The milestone delivery of the melanoma proteome will enable open resource bioinformatics tools to be applied in an interactive way with computational science.

RESULTS AND DISCUSSION
The present publication belongs to a series of two on the Human Melanoma Proteome published by Clinical and Translational Medicine. Both are integral parts of the MM500 study. The other publication is entitled "The Human Melanoma Proteome Atlas-Complementing the Melanoma Transcriptome." It provides a map with the abundance of identified proteins, a comparison with the TCGA melanoma transcriptome data, analysis of posttranslational modifications, along with the detection of missing proteins, melanoma driver mutations at the protein level, and the analysis of protein expression in the blood plasma of melanoma patients.

Clinicopathological data of the MM500 cohort
Histopathological characterization of the tumor samples was made, followed by a quality assessment by detailed pathological analysis of the fresh frozen and formalin-fixed and paraffin-embedded (FFPE) tissue samples. The body sites of the metastatic tumor specimens are outlined in Figure 1A. The cohort included 294 melanoma tumors and from most of them multiple pieces were collected and analyzed which made a total number of 505 tumor samples. The melanomas were stratified according to the recent AJCC (8th edition, 2017) staging system 27 . A summary of the patient clinical information, and histopathological data is presented in Table 1. Table  2 shows the clinic sites from which the tumors were obtained, the number of tumor samples analyzed, and the corresponding melanoma disease stages of donor patients.
The cohort had a preponderance of male individuals (62%) and 63% of all patients were older than 60 years. Among patients with available overall survival (OS) information, nearly 47% survived less than 5 years. Several factors could have influenced this outcome, for instance, 56 patients had thicker primary tumors (Breslow > 1 mm), which are associated with short survival. 6 In addition, 48 of these patients were diagnosed before BRAF/MEK targeted treatment and immune checkpoint inhibitor therapies were introduced, which have improved the management of MM disease and patient survival. 33 (Table 1) Overall, the 505 analyzed tumor samples included 13% primary tumors, 64% regional lymph node metastases, and 22% of other metastases.

Heterogeneity of melanoma tissues in MM500
Both melanoma tissue and tumor-derived stroma were collected for the study. Representation of the classical variants of melanoma subtypes is shown in Figure 2A-J. Primary tumors and metastatic tissue samples were characterized by histopathology including: percentage of (1) tumor cell, (2) stroma, (3) immune cells within the tissue, (4) necrosis and (5) adjacent tissue ( Figure 2N). Melanoma cell content varied significantly (ANOVA test, FDR < 0.05) between primary tumors and both lymph nodes and other metastases. Immune cells presence in the samples was statistically higher (ANOVA test, p-value < 0.05) in regional lymph node metastases compared to other tumor types, which were below 20%. Stromal components showed a significantly lower content (ANOVA test, p-value < 0.05) in lymph node metastases than in primary tumors and metastases from other locations. In contrast, there were no significant differences between lymph node and other metastases in the contents of necrosis (t-test p-value = 0.193) and adjacent tissue (t-test p-value = 0.119).
The intratumor cell heterogeneity observed within the specimens is exemplified in the image from one of the melanoma patients ( Figure 2G-M).
DP is slowly gaining acceptance in routine diagnostics and can now be utilized both for frozen sections and paraffin tissue blocks. 24,36,37 The utility of deep neural networks is approaching a level of accuracy similar to or even higher than analysis made by dermatologists and dermatopathologists. [38][39][40] To further characterize the study tumor samples, a method for visualizing the heterogeneous melanoma morphology by DP was created, utilizing recent developments in machine learning. The combination of deep learning-based single-cell segmentation and phenotyping now for the first time enables digital dissection of melanoma samples at the resolution of a single cell ( Figure 3A and B) and phenotype assignment to individual cells ( Figure 3C).
Investigation of the intratumoral heterogeneity of primary melanomas revealed that the clonal evolutionary pathways of the manifested metastases are very diverse; not just at the histological, but also at the molecular level. [41][42][43] This is exemplified by four metastases from one patient, all displaying high tumor contents (≥70%) and low percentages of adjacent tissue (0-10%). Here protein profiles showed marked differences across the metastases, with expressed proteins enriched in distinct biological processes ( Figure 4). Interestingly, despite of the histological evaluation, the liver metastasis showed enrichment in processes that can be related to the tumor location.

Protein identification and melanoma proteome dynamics
All tumor samples were submitted to the slicing procedure where up to 30 slices were selected for proteomics, while the first and last slices were saved for histological confirmation of melanoma tumor tissue. The strategies for sample processing, protein digestion, and MS data acquisition were summarized in Figure 5, where several state-of-theart procedures were optimized and/or implemented in our laboratory. [44][45][46][47][48] For protein quantification mainly two approaches were adopted, the label-free and the isobaric labeling using tandem mass tag TMT-11plex. The samples analyzed by the TMT approach were fractionated after labeling. Furthermore, for MS data acquisition two strategies were applied, the classical data dependent acquisition (DDA) approach and a data independent acquisition (DIA) strategy consisting of variable wide windows selected for MS/MS together with a spectral library for peptides and proteins identification. Samples were analyzed under a system suitability test designed to ensure the best performance of the LC-MS instruments. Here, a standard peptide mixture from HeLa cells total protein digestion was used as a quality control sample to monitor different metrics ( Figure 6). In total, 1549 liquid chromatography-tandem mass spectrometry (LC-MS/MS) experiments were conducted with both melanoma tumor, and cultured cell line samples. These generated more than 51 million MS/MS spectra and resulted in approximately 13.6 million peptideto-spectrum-matches (PSMs) that successfully annotated 15,531 proteoforms, which included canonical proteins and isoforms (Supporting Information Table S1). Altogether we identified 12,878 protein-coding genes (Supporting Information Table S2), represented by 3,66,172 peptides.
In cultured melanoma cell lines 10,511 proteoforms were identified and quantified, representing 8657 proteincoding-genes ( Figure 7A, Supporting Information Table  S3). Interestingly, 135 (1.5%) protein-coding-genes were exclusively identified in melanoma cell lines despite having significantly fewer analyzed samples, LC-MS/MS experiments, and a lower number of identifications. We did not find any particular enrichment in pathways or Gene Ontology terms for this group of proteins. Out of the 135 proteins, 119 were found as transcripts in the TCGA database for melanoma, and 13 of the remaining proteins were detected in melanoma tumors either by immunohistochemistry or at mRNA level according to the Human F I G U R E 5 General scheme comprising the proteomic workflows used throughout the MM500 study. Upper panel) Overall results from the melanoma proteome as studied by the European Cancer Moonshot Lund Center, were more than 500 melanoma tissue samples and four cultured cell lines were analyzed. Sample preparation) Proteins were extracted in the presence of urea or SDS with the aid of a Sonifier or a Protein Atlas (HPA) (Supporting Information Table S5). Overall, there were only three proteins for which no evidence of expression was found in both databases and the analyzed tumor samples. Several factors might have influenced the result such as false-positive identifications (1% in our study) or protein-specific expression as an artifact due to cell culture conditions. In contrast, 4221 proteins were uniquely identified in the tumor samples, which can be attributed to the large diversity of cell types in the tissue, and the intra-, and intertumor heterogeneity. (Figure 7A). The KEGG pathways enrichment analysis showed that most of these proteins are involved in immune system related pathways, or they are part of the extracellular space/vascular system ( Figure 7A). These findings highlight the importance of studying the molecular profiles of melanoma in the context of the tumor tissue and its microenvironment, in order to better understand the mechanisms responsible for development and progression of the disease. Although established melanoma cell lines to some extent will mimic the disease, the changes imposed by interaction with the tumor microenvironment is very limited. However, the proteins commonly identified in both cell lines and tumors showed a significant correlation (Pearson r = 0.733, p-value = 10e-323) in their abundance profiles ( Figure 7B), indicating that, melanoma cell lines also show tumor-like relative protein levels. The differential proteome analysis between tumors and cell lines concludes that cell lines mainly upregulate pathways related to cell proliferation and in contrast, tumors mainly upregulate proteins involved in cell-cell communication and endocytosis ( Figure 7C).
The MM500 study included the analysis of melanoma samples with large variation in the cellular composition, which was evident from the histopathology characterization ( Figure 2N). In particular, a subset corresponded to samples with low tumor cell content and more than 70% of stroma. The differences between this group and samples with the highest tumor cell content (>70%) were explored through multiple t-test analysis (FDR 0.01). In total, 1266 proteins were regulated in relation to the cellular Bioruptor. Manual or automatic enzymatic digestion protocols were carried on the protein mixtures depending on the agent used for protein extraction. Global proteomicsDDA-MS) DDA data were generated from labeled or labeled-free peptides. We analyzed TMT 11-plex labeled peptides after high pH RP-HPLC fractionation. Labeled-free peptides were directly analyzed (shotgun proteomics), fractionated by SCX     Figure 7D). The functional annotation analysis showed that samples with high stroma content were, as expected, enriched in proteins involved in the extracellular matrix and its interaction with the cells, in addition to elements of the complement and coagulation cascade. Oppositely, tissue samples with high tumor cell content were significantly enriched in proliferation related pathways including metabolic pathway linked to rapid energy production ( Figure 7E). These examples highlight the diversity in the disease presentation in melanoma, which can be captured by combining histopathology and proteomics.
The BRAF V600E mutation is present in 40-60% of melanomas. 49,50 To investigate the differences in the proteome profiles on BRAF mutated and wild-type (WT) melanoma tumors, we compared the protein expression of 48 samples with BRAF V600E mutation and 70 WT BRAF samples that were processed and measured under similar conditions. A statistically significant difference (ttest, FDR 0.05) between the two groups was seen for only two proteins. A recent study, although with fewer samples, only found 17 differentially expressed proteins when compared the protein expression between BRAF mutated and WT tumors. 51 The results suggest that tumors with the BRAF V600E mutation does not display a specific proteomic profile when compared to WT BRAF tumor. Nevertheless, we recently reported that the level of expression of the mutation does cause a different proteome profile within BRAF V600E positive tumors, which also correlates with differences in tissue morphology and patient outcome. 15

Protein localization: cell and chromosome
In the present study, we go one step deeper for the spatial localization of proteins, a critical step to understand biological functions. 52,53 An outline of the gene products of the 12,878 identified protein-coding genes was constructed and organized to chromosome and subcellular localization ( Figure 8).
Interestingly, most chromosomes (C) were evenly covered by the identified proteins (55to 68% in chromosomes, and 80% in mitochondrial genome) with the exceptions of chromosome 21 (48%) and the Y chromosome (19%) (Figure 8A). The chromosomal p arm has low, or no, gene density of expressed proteins for C13 to C15 and C21, and C22. A low density of expression was also observed in the q arm close to the centromere of C1, C9, and C16.
The identified proteins were distributed and categorized according to the Cell Atlas database from the HPA. From the interrogated melanoma data (Supporting Infor-mation  54 the total annotation for proteins identified in our study equates to 12,860 localizations. This includes 361 proteins with localization information confirmed by immunofluorescence images, and confocal microscopy and directly derived from SK-MEL-30 melanoma cultured cells. (Supporting Information Table S7).
The melanoma proteome was annotated into 26 different cellular localization comprising organelles, structures, and substructures. Figure 8B shows the main localization of proteins sorted by cellular compartments. The largest melanoma proteome localized to the cytoplasm with 4654 proteins, followed by 4351 secreted and 3396 nuclear proteins, which account for 83, 58, and 68%, respectively when considering the corresponding proteins of the human proteome annotated to these organelles.
The lower coverage of secreted proteins might be explained by the significant number of missing proteins in the human proteome that are classified as secreted. Currently, these proteins lack sufficient experimental data from mass spectrometry or other direct protein identification methods. 55,56 As detailed in Figure 8C, the proteins of 13 out of 17 cytoplasmic organelles and structures were covered, from 58 to 100%. The cytosol was the most populated with 3210 proteins (83%), and in the rest of the compartments, albeit with lower percentages, just a few dozens of proteins were not identified. A similar panorama was found for the nucleus where more than 68% of proteins from the six subcellular locations were identified. Within this organelle, nuclear body proteins were less represented with only 20 proteins that were not identified.
Interestingly, the plasma membrane accounts for the highest number of proteins (4411) among all the cellular compartments, and at the same time, it exhibited one of the lowest percentages of protein coverage in our melanoma protein data (49%, 2036 proteins). These proteins could be peripheral or integral membrane proteins, with the later known to possess hydrophobic transmembrane domains. Therefore, the grand average of hydropathicity (GRAVY) index was investigated for plasma membrane proteins from the melanoma proteome as well as for those classified as missing proteins ( Figure 8D). For the different data sets, similar patterns were obtained, which indicated that the MM500 tissue analysis protocols covered almost the whole hydrophobicity spectrum of proteins. However, a decrease in frequency for the more hydrophobic proteins (GRAVY-index ≥0.6) was seen in the melanoma dataset. Missing proteins in the secretory component were highly represented in the plasma membrane (31% of all missing proteins) and particularly towards the high end of the hydrophobicity range. Consequently, the missing proteins were a determinant factor in the lower coverage of plasma membrane proteins within the melanoma proteome. Altogether, the proteins identified here were classified and assigned to 26 of 28 distinct subcellular localization categories, representing 67% of the human proteome.

CONCLUSIONS AND FUTURE PERSPECTIVES
The histopathological characterization of more than 500 melanoma tumor samples showed a very heterogeneous disease, not only intertumor but also intratumor. A significant set of melanoma samples showed low levels of tumor cells, which in most cases correlated to high stromal content. Primary lesions presented a relatively lower tumor cell content compared to regional or distant metastases.
This MS melanoma proteome study identified 15,500 proteoforms, covering 65% of the total human proteome. Also, among the large number of cytoplasmic proteins identified more than 2000 membrane proteins were detected, including receptors and ligands usually present in low copy numbers. Combining the detailed molecular characterization of expression with histopathology contributes to the pathologist definition of tumor samples. The presented outline of the melanoma proteome can now form a basis for hypothesis building within the diagnosis, treatment, drug discovery, and drug development. For example, the annotations of specific key regulating proteins and their respective functions, cellular localizations, whether intracellular, or nuclear, can be associated with the mode of action of possible therapeutic agents. Information of the abundance and origin of expressed proteins; for example, in primary tumors, or metastasis within the lymph nodes, lung, brain and other organs, coupled to detailed clinical phenotyping, complement the investigation. Taken together, the reported melanoma proteome data will be vital for generating a better understanding of the disease and the mechanisms that drive the progression.

Tissue specimens
A total of 505 patient tumor samples ( Figure 1 and Table 1 & 2) obtained from University clinics in Sweden and Hungary, was included in the study. All studies were approved by the local ethical committees, including the Regional Ethical Committee at Lund University, Southern Sweden ( Fresh-frozen tissues collected at Lund University Hospital were stored in the Melanoma biobank, BioMEL, Region Skåne, Sweden. Tissues collected at sites in Hungary were stored at the respective biobanks of Semmelweis University and the University of Szeged. Samples transportation from Hungary was carried out in liquid nitrogen. In the case of FFPE samples, the fixation of tumor tissues was performed right after surgery with 4% buffered formaldehyde. Samples were then dehydrated in xylene/alcohol solution and embedded into paraffin. FFPE samples were transferred and stored at room temperature. Sections of 10 μm were used for further analysis. The study has been performed in compliance with GDPR. All tumors were processed with integrated Biobanking consolidations within all involved medical centers. The workflow was built according to Swedish biobanking laws and best practices and guidelines provided by the BBMRI-ERIC, ESBB, and ISBER (https://www. bbmri-eric.eu/services/quality-management). 57 The process flow enabled rapid sample handling whereby collected tissues were stored at an ultra-low temperature in a biobank at a cycle time of approximately 20 min. Using the same data management system and database reconnaissance, sample integrity was ensured via electronic surveillance. The patient and sample processing workflow and protocols for the Semmelweis and Szeged cohorts were transferred and interfaced with the RedCap database, (https://www.project-redcap.org/). 58

Cell cultures
All melanoma cell lines (SK-MEL-2, SKMEL28, and VMM1) were purchased from ATCC and cultured and maintained at standard conditions and recommendations by the manufacturer. In detail, SK-MEL-2 and SKMEL28 were maintained with DMEM (Dulbecco's modified Eagle's medium) supplemented with 10% fetal bovine serum (FBS) and penicillin-streptomycin (P/S). VMM1 were maintained with RPMI-1640 supplemented with 10% FBS and P/S. All cells were maintained at 37 • C in a humidified 5% CO 2 incubator.

Histopathological analysis
Stepwise sectioning of the tissues was performed using a cryostat and setting the slices thickness at 10 μm. On average, three sections were evaluated. Tissue sections were placed on glass slides, stained with hematoxylin and eosin, and then placed in an automated slide scanner system (Zeiss Mirax, Jena, Germany). The slides were then evaluated for tissue content: tumor, necrosis, stroma, tissue, and adjacent background tissue-mostly lymphatic cellslymph node area. 52

Statistical analysis
One-way analysis of variance (ANOVA) followed by Tukey post-hoc analysis for multiple comparisons and paired ttest, were used to determine the statistical significance of histological parameters between the different tumor types.

Deparaffinization of FFPE tissue
The FFPE tissue sections were incubated with 1 mL of 1:50 diluted EnVision™ FLEX Target Retrieval Solution High pH (Agilent Dako) at 97˚C for 10 min (500 RPM). Incubation was followed by a brief centrifugation at 14,000 g at 4˚C for 3 min, removal of the EnVision solution and the paraffin. These steps were repeated until complete paraffin removal as previously described. 59

Protein extraction
For fresh-frozen tissues, the lysis buffers contained 100 mM ammonium bicarbonate or 100 mM Tris pH 8.6 and up to 6 M Urea or 2% SDS. Lysates were generated by sonication in an ice batch using a Branson Sonifier 250 (output 4, 10% duty cycle) or using the Bioruptor plus, model UCD-300 (Dieagenode) for 40 cycles. Each cycle consisted of 15 s at high power and 15 s without sonication at 4 • C. Samples were centrifuged at 10,000 g and 4 • C for 10 min and the supernatants were transferred into a new tube and the pellet was discarded.
In the case of FFPE tissue samples, the protein extraction was performed by adding 100 mM TEAB containing 25 mM DTT and 10w/v% SDS pH 8. The samples were incubated at 99˚C for 1 h with shaking (500 RPM) and sonicated in the Bioruptor R Plus UCD-300 (Diagenode) for 40 cycles (15 s on and 15 s off) at 4˚C, followed by centrifugation at 20,000 g at 18˚C for 10 min.

4.5.3
Protein determination The protein in each one of the samples was determined using a colorimetric micro BCA Protein Assay kit (Thermo Fisher Scientific, Rockford, IL) following the manufacturer's instructions.

Protein digestion
Proteins were reduced with 10 mM DTT for 1 h at 37 • C and alkylated with 40 or 50 mM iodoacetamide for 30 min, in the dark, at room temperature. Proteins were digested overnight with trypsin or Lys-C and trypsin using published and optimized protocols including buffer exchange 47,60 or urea in-solution digestion 46 which comprised automated sample handling. 44 SDS was removed from the samples by the MED-FASP method 48 or by ethanol precipitation. 61 The later was followed by protein solubilization in 50 mM Ambic with 0.5 SDC (Sodium deoxycholate) and trypsin digestion. For acetylation analysis, the samples were processed and digested as previously described. 61

Peptide determination
The peptide quantity in each sample and fraction was determined using the Pierce Quantitative Colorimetric Peptide Assay according to the instructions provided by the manufacturer.

Peptide and protein identification and quantitation in DDA-MS experiments
Raw DDA-LC-MS/MS files were analyzed with the Proteome Discoverer™ Software (Thermo Scientific™) against Uniprot Human dataset to which were added Fasta format protein sequences of known driver mutations of Melanoma disease. 63 The search engine Sequest HT was used for peptide identification. Carbamidomethylation was set as a static modification as well as TMT 6plex (+229.1629 Da) at peptide and lysine N-termini as well as lysine ε-amino for labeling experiments. Oxidation of methionine residues and acetylation at protein N-termini were selected as dynamic modifications. Precursor and fragment mass tolerance was set as 20 ppm and 0.02 Da, respectively, and two missed cleavages were allowed for peptides. The Minora node was included in the search for identification using retention time alignment and the match-between-runs features. For label-free experiments, the quantification was carried out using the TOP3 method where the protein abundance is reported as the mean of the three highest peptides (unique and razor) areas measured for each protein. For TMT labelling experiments, protein abundances were calculated as the summed areas of reporter ions considering unique peptides. Identification and sorting of unique peptides were carried using the neXtProt tool "Peptide uniqueness checker" (https://www. nextprot.org/tools/peptide-uniqueness-checker). 64

4.7.2
Peptide and protein identification and quantitation in DIA-MS experiments A Global proteomics spectral library was generated from DDA experiments as described above. Raw files were converted to HTRMS files with a special converter provided by Biognosys AG and searched in the Spectronaut X platform (Biognosis AG) against the Homo sapiens database from Uniprot containing isoforms. Dynamic retention time prediction was selected to enable nonlinear alignment of precursor retention times between the (iRT, normalized retention time) spectral library and the DIA-MS data by segmented regression. The following parameters were used: cysteine carbamidomethylation (+57.0215 Da) as fixed modification and methionine oxidation (+15.9949 Da), N-terminal acetylation (+42.0105 Da) as dynamic modifications. A maximum of two missed cleavages were accepted. Precursor mass tolerance was set to 10 ppm and for the MS/MS fragments it was set to 0.02 Da. Filtering was performed at a 1% false discovery rate (FDR) for all the peptides and proteins that were used to construct the spectral library. The resulted library containing identified spectra for 220,360 peptides representing 12,293 proteins. The software computed MS1 peptide abundance as the summed precursor XIC (Extracted-Ion Chromatogram, from the monoisotopic precursor ion plus isotopic envelope). The protein abundance resulted from the average of the top three most intense precursor ions corresponding to unique and razor peptides.

Localization of melanoma proteins and chromosomal distribution
The complete melanoma dataset of 12,878 proteins (Supporting Information Table S1) was confronted on 2020/09/18 with the UniProt database (https://www.uniprot.org, UniProt release 2020_03) 65 for the initial spatial localization of proteins. Proteins not localized in the UniProt database were taken to a second search on the same date in the Human Proteome Atlas databank version 19.3 (https://www.proteinatlas. org/humanproteome/cell/organelle) 66 for complementary subcellular location. All UniProt levels of evidence, for example, protein, transcript, inferred from homology and predicted, as well as the immunohistochemistry reliabilities of HPA, namely, Enhanced, Supported, Approved, and Uncertain were accepted for search, considering that identification by MS confirmed the presence of the gene product. Both databanks were searched for main subcellular localization only. Structures and organelles were grouped according to the HPA Human Proteome nomenclature and Thul et al. 52 Based on genomic gene annotation and its chromosomal position, the R package RIdeogram 67,68 was used to map gene density heatmap on the chromosomes and its coverage percentages, respectively.

Protein normalization and analysis
The results from protein identification and quantification were imported into Perseus software version 1.6.14.0. 69 Data were normalized by log2 transforming the protein intensities, and standardization was performed by subtracting individual values by the median in each sample. The proteins showing less variability across all batches that were identified in 100% of the samples were used to correct the abundance differences between batches. To do that, individual protein intensities in each batch were subtracted by the median abundance of selected proteins in the specific batch. After correction, the median abundance for each protein across all samples was calculated and reported as the relative abundance in our melanoma proteome. The statistical analyses were also performed using the Perseus software.

D ATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are openly available in ProteomeXchange at http://www. proteomexchange.org/, reference numbers PXD001725, PXD001724, PXD009630, PXD017968, and PXD026086 and will be complemented by the addition of data from the study. Tables 1 and 4 of Supporting Information are available at https://github.com/rhong3/TCGA_melanoma/ tree/master/Supporting%20Information%20tables.