A Targeted Quantitative Proteomic Method Revealed a Substantial Reprogramming of Kinome during Melanoma Metastasis

Kinases are involved in numerous critical cell signaling processes, and dysregulation in kinase signaling is implicated in many types of human cancers. In this study, we applied a parallel-reaction monitoring (PRM)-based targeted proteomic method to assess kinome reprogramming during melanoma metastasis in three pairs of matched primary/metastatic human melanoma cell lines. Around 300 kinases were detected in each pair of cell lines, and the results showed that Janus kinase 3 (JAK3) was with reduced expression in the metastatic lines of all three pairs of melanoma cells. Interrogation of The Cancer Genome Atlas (TCGA) data showed that reduced expression of JAK3 is correlated with poorer prognosis in melanoma patients. Additionally, metastatic human melanoma cells/tissues exhibited diminished levels of JAK3 mRNA relative to primary melanoma cells/tissues. Moreover, JAK3 suppresses the migration and invasion of cultured melanoma cells by modulating the activities of matrix metalloproteinases 2 and 9 (MMP-2 and MMP-9). In summary, our targeted kinome profiling method provided by far the most comprehensive dataset for kinome reprogramming associated with melanoma progression, which builds a solid foundation for examining the functions of other kinases in melanoma metastasis. Moreover, our results reveal a role of JAK3 as a potential suppressor for melanoma metastasis.

proteomic data collected by analyzing tryptic digestion mixtures from multiple human cell lines 12 . The library encompasses 1050 tryptic peptides representing 478 kinase proteins, and 395 of them are protein kinases 12,16-18 . In this work, we applied the recently established PRM-based targeted proteomic method 12 , in combination with stable isotope labeling by amino acids in cell culture (SILAC), to examine the differential protein expression of kinases in 3 pairs of matched primary/metastatic melanoma cells. We were able to quantify the relative expression levels of approximately 300 unique kinase proteins in each pair of cell lines. Among them, JAK3, MAP4K4 and CCNH were consistently down-regulated in the metastatic lines of all three pairs of melanoma cells. Analysis of publicly available data also showed that elevated JAK3 expression is correlated with improved survival of melanoma patients. We further demonstrated that JAK3 suppresses the migratory and invasive abilities of melanoma cells in vitro, which involves diminished enzymatic activities of secreted matrix metalloproteinases.

Results
Quantitative assessment of the differential expression of kinase proteins in primary and metastatic human melanoma cells. Our primary objective was to explore the potential roles of kinases in melanoma metastasis. Toward this end, we applied a PRM-based targeted proteomic method 12 , in combination with SILAC 19 , to examine the differences in kinase protein expression in 3 pairs of primary/metastatic melanoma cells (i.e. IGR-39/IGR-37, WM-115/WM-266-4, and WM-793/1205Lu) (Fig. 1a). In this vein, WM-115 and WM-266-4 cells were derived from the primary and skin metastatic sites of the same melanoma patient, respectively 20 , and IGR-39 and IGR-37 cells were initiated respectively from the primary and groin metastatic sites of another melanoma patient 21 . On the other hand, 1205Lu cells were originated from a lung metastasis in an immune-deficient mouse through tail-vein injection of WM-793 human melanoma cells, which were initiated from a superficial spreading melanoma 22 .
The proteomic results led to the quantification of approximately 300 unique kinases in each pair of melanoma cells, including 237-252 protein kinases, along with many lipid, carbohydrate and nucleotide kinases (Figs. S1-S3, Tables 1 and S1). All 4-6 PRM transitions used for the quantification of every kinase peptide displayed the same retention time and exhibited a dot product (dotp) value of >0.7 (Fig. S4) 23 . In addition, over 90% of the quantified kinase peptides showed consistent trends in forward and reverse SILAC labeling experiments (Fig. 1b,c, Table S1, and representative results for AK1 are shown in Fig. S5b). Along this line, owing to the complex matrices for the samples employed for the LC-PRM analysis (i.e. the tryptic digestion mixtures of whole-cell protein lysates), a somewhat modest R 2 value (0.50) was found for the correlation between the ratios obtained from forward and reverse SILAC labeling experiments (Fig. 1c). Based on the range of signal intensities detected for the kinase peptides, we estimated that the PRM method exhibits a dynamic range of 4 orders of magnitude (Fig. 1d).
A comparison of the kinome quantification results revealed that the IGR-39/IGR-37and WM-115/WM-266-4 pairs of melanoma cells displayed many similar attributes in metastasis-associated alterations in expression levels of kinases. However, the results for these two pairs differed substantially from those obtained for the WM-793/1205Lu pair ( Fig. 2 and Table S1). This might be attributed to the facts that melanoma is a highly heterogeneous disease 24 and that the metastatic lines from the first two pairs were derived from metastasis in melanoma patients, whereas that of the last pair was from experimental metastasis in a mouse 22 .
We also validated the protein expression levels for six quantified kinases (AK1, JAK3, PAK1, PAK2, SCYL3 and STK26) by Western analysis and the results are all in agreement with what we obtained from PRM measurement (Fig. 3), supporting that the LC-PRM method, in conjunction with SILAC labeling, afforded robust quantification of kinase proteins. In this vein, it's worth noting that the PRM-based LC-MS/MS quantification method relies on unique peptide sequences of kinases; hence, the method can distinguish different kinase isoforms. For instance, the method allowed for the independent quantifications of PAK1 and PAK2 (Fig. 3c). The PAK antibody that we used for Western blot analysis can recognize all three isoforms of PAK (i.e. PAK1, PAK2, and PAK3). While the peptides derived from PAK3 were below the detection limit of the PRM method, the peak intensity observed for the PAK2 peptide is approximately 10-fold higher than that for the PAK1 peptide, which is consistent with the Western blot results. These observations, in conjunction with the similar molecular weights of PAK1 and PAK3, allow us to assign the higher-molecular weight band observed in Western blot exclusively to PAK1.
We further performed KEGG pathway analysis 25 on the basis of those kinases that are up-regulated in the metastatic WM-266-4, IGR-37 and 1205Lu cells over the corresponding primary melanoma cells, and we found that many pathways, including those of MAPK signaling, cancer, focal adhesion and purine metabolism, were up-regulated (Fig. S6). Additionally, pathways in melanoma is up-regulated based on those kinases that are commonly up-regulated in the metastatic WM-266-4 and IGR-37 cells over the corresponding primary melanoma cells (Fig. S6).
Potential roles of JAK3 in melanoma progression. To identify potential drivers and suppressors for melanoma metastasis, we compared the 130 perturbed kinases obtained from the PRM results of WM-115/ WM-266-4 pair of cells with the TCGA data collected for more than 400 melanoma patients 26 . We found that the mRNA expression levels of 19 differentially expressed kinases are correlated with the overall survival of melanoma patients (Table S2, Logrank p-value < 0.05). Among these kinases, 6 (RPS6KB2, DSTYK, TBRG4, CDK4, POLR2E, FASTKD5) and 4 (STK26, PAK1, EPHB2 and JAK3) were consistently up-or down-regulated, respectively, in the metastatic lines of at least two pairs of melanoma cells (Figs. 4b and S7). For instance, JAK3 displayed consistently lower levels of expression in all three metastatic lines of the paired melanoma cells (Fig. 3c,d). The same trend held for the mRNA expression levels (Fig. 4a).
JAK3, a tyrosine kinase involved in signal transduction by receptors 27 , was previously shown to assume important roles in multiple types of cancer. For instance, JAK3 was demonstrated to promote breast cancer metastasis 28 . In addition, JAK3 exhibits high level of expression in leukemia (Fig. S8) and it was found to be frequently mutated in T-cell acute lymphoblastic leukemia 29 Table S1.

WM-115/WM-266-4 IGR-39/IGR-37 WM-793/1205Lu
Protein kinases 252 237 242 Lipid kinases 10 11 10 Nucleotide kinases 27 27 28 Carbohydrate kinases 12 13 12 Other kinases 5 6 6  www.nature.com/scientificreports www.nature.com/scientificreports/ control cell migration, invasion and apoptosis 30 . The roles of JAK3 in melanoma progression, however, remain unclear. To address this, we first investigated whether JAK3 gene is expressed at different levels in metastatic and primary melanoma tissues by analyzing the data retrieved from the Gene Expression Omnibus (GEO) and CCLE datasets. In particular, an analysis of the GSE8401 dataset, which included 83 melanoma patients (31 primary and 52 metastatic) 31 , revealed a significant down-regulation of JAK3 gene in the metastatic relative to primary melanoma tissues (Fig. 4c). Analysis of the CCLE dataset also unveiled the diminished mRNA level of JAK3 gene in metastatic melanoma cell lines over primary melanoma lines (Figs. 4d and S8). Thus, our quantitative proteomic results, combined with TCGA data and other datasets, suggest that JAK3 may suppress melanoma metastasis.

JAK3 suppresses the invasion of cultured melanoma cells by regulating the activities of secreted metalloproteinases (MMps).
We next studied the potential function of JAK3 in melanoma metastasis by asking how the capacities in migration and invasion of cultured melanoma cells are influenced by the expression levels of kinase proteins 32 . Our results from transwell migration and invasion assay showed that the WM-115 primary melanoma cells exhibited a modest, yet statistically significant increase in migratory capacity, and a marked elevation in invasive ability upon siRNA-mediated knock-down of JAK3 (Figs. 5a,c,d and S9). Reciprocal experiment showed that the overexpression of JAK3 in WM-266-4 metastatic melanoma cells suppressed their motility and invasion (Figs. 5b-d and S9).
Given that MMP-2 and MMP-9 play important roles in degrading extracellular matrix proteins and promoting cancer metastasis 33 , we next investigated whether the enzymatic activities of MMP-2 &-9 could be modulated  Table S1 for ratios obtained from individual biological replicates). Only those kinases that were commonly quantified in all three pairs of melanoma cells were plotted. Red and green boxes represent the upand down-regulated kinases in metastatic over primary melanoma cells, respectively. Arrows indicate those kinases whose relative levels of expression in the 3 matched pairs of melanoma cells were validated by Western blot analyses (see Fig. 3). (2020) 10:2485 | https://doi.org/10.1038/s41598-020-59572-5 www.nature.com/scientificreports www.nature.com/scientificreports/ by JAK3 expression levels. We found, from gelatin zymography assay 34 , that the activities of secreted MMP-2 MMP-9 were heightened in WM-115 primary melanoma cells upon siRNA-mediated knock-down of JAK3 (Figs. 5e,g,h and S9). Reciprocally, ectopic overexpression of JAK3 in WM-266-4 metastatic melanoma cells led to diminished enzymatic activities of secreted MMP-2 &9 (Figs. 5f-h and S9). Together, the above results suggest that JAK3 impedes the invasion of melanoma cells in vitro, at least in part, by regulating the activities of secreted MMP-2 and MMP-9.
The epithelial-to-mesenchymal transition (EMT) is often accompanied with invasion of cancer cells, particularly those of epithelial origin 35 , and JAK3 was shown to inhibit EMT by strengthening the adherens junctions 36 . Hence, we also examined the levels of expression of N-cadherin, an EMT marker, and our result showed that the N-cadherin protein was expressed at a higher level in WM-115 primary melanoma cells than WM-266-4 metastatic melanoma cells (Fig. S10). Furthermore, siRNA-mediated knockdown of JAK3 resulted in diminished expression of N-cadherin (Fig. S10). Thus, EMT does not appear to contribute to elevated invasive capacity of the WM-266-4 melanoma cells, and the JAK3-mediated suppression of invasive capacity of melanoma cells is not attributed to EMT inhibition. This is in keeping with the previous finding that melanoma cells are not of epithelial origin 37 . Listed are the PRM traces of IGQPTLLLYVDAGPETMTQR from AK1 (y 5 , y 6 , y 7 , y 8 , y 9 , y 10 ), SCSPSAEFLR from JAK3 (y 3 , y 5 , y 6 , y 7 , y 8 ), NTSTMIGAGSK from PAK1 (y 5 , y 6 , y 7 , y 8 , y 9 ), MTDEEIMEK from PAK2 (y 3 , y 4 , y 5 , y 6 , y 7 , y 8 ), VILPQVLLGLR from SCYL3 (y 4 , y 5 , y 6 , y 7 , y 8 , y 9 ) and SIAVAEAACPGITDK from STK26 (y 3 , y 6 , y 7 , y 8 , y 9 , y 11 ). The y ions listed in the parentheses were the fragment ions observed in the MS/MS of the corresponding peptides that were chosen for peptide quantification. www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
Kinase-and phosphatase-mediated reversible phosphorylation of proteins and small molecules represents one of the most crucial and best characterized cell signaling pathways 1,2 . Thus, high-throughput kinome profiling constitutes a powerful approach for the systematic interrogation of kinase-modulated molecular events forged by extracellular cues and intracellular signaling. In this work, we applied a PRM-based targeted proteomic method to assess kinome reprogramming during melanoma metastasis, which led to the identification of novel kinases functioning as potential promoters or suppressors for melanoma metastasis. To our best knowledge, this is so far the most comprehensive kinome profiling dataset for melanoma metastasis.
We found that approximately half of the ~300 quantified kinases exhibited differential expression between primary and metastatic melanoma cells (Table S1). In combination with TCGA data, we discovered that JAK3 plays important roles in melanoma progression. Furthermore, the mRNA expression of JAK3 was reduced in the metastatic over primary tumor tissues of melanoma patients, and in metastatic over primary melanoma cell lines (Fig. 4). Therefore, we further investigated JAK3's roles in melanoma metastasis in vitro. Our results showed that reduced JAK3 expression is correlated with increased migration and invasion of cultured melanoma cells (Fig. 5). www.nature.com/scientificreports www.nature.com/scientificreports/ In addition, the activities of secreted MMPs are negatively correlated with the expression levels of JAK3 gene in primary and metastatic melanoma cells (Fig. 5), illustrating that JAK3 modulates melanoma cell invasion by regulating the activity of secreted MMPs. Along this line, protein interactome analysis demonstrated that JAK3 www.nature.com/scientificreports www.nature.com/scientificreports/ interacts with BRMS1, which was shown to suppress melanoma metastasis 38 , and JAK3 co-expresses with another melanoma suppressor, EDAR (Fig. S11) 39 . Therefore, JAK3 may function together with BRMS1 and EDAR to suppress the migration and invasion of melanoma cells.
In summary, our targeted kinome profiling method provides a systematic assessment about kinome reprogramming during melanoma metastasis and our results provides important lines of evidence to support JAK3 as a potential suppressor for melanoma metastasis. experimental procedures cell culture. IGR-39/IGR-37 (generous gifts from Prof. Peter H. Duesberg, with BRAF V600E and P53 C229fs ) 40 , WM-793/1205Lu (The Wistar Institute, with BRAF V600E , P53 WT and CDK4 K22Q ) cells were cultured in Dulbecco's modified eagle medium (DMEM). WM-115/WM-266-4 cells (ATCC, with BRAF V600D and P53 WT ) were cultured in Eagle's minimum essential medium (EMEM). All culture media were supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA) and penicillin (100 IU/mL). The cells were kept in a humidified atmosphere with 5% CO 2 at 37 °C. Approximately 2 × 10 7 cells were harvested, followed by washing with cold PBS for 3 times. The cells were then lysed by incubating on ice for 30 min with CelLytic M cell lysis reagent (Sigma) with 1% protease inhibitor cocktail. After a 30-min centrifugation at 9,000 g and at 4 °C, the resulting supernatants of cell lysates were collected. For SILAC labeling, the cells were cultured in SILAC medium containing unlabeled lysine and arginine, or [ 13 C 6 , 15 N 2 ]-lysine and [ 13 C 6 ]-arginine for at least five cell doublings. The initial passage numbers for the melanoma cells were: WM-115 (p9), WM-266-4 (p6), IGR-39 (p4), IGR-37 (p7), WM-793 (p16), 1205Lu (p70). plasmids and siRnAs. The sequences for siJAK3 were 5′-GGGUCCUUCACCAAGAUUU-3′ and 5′-CCAUGGUGCAGGAAUUUGU-3′ (Dharmacon Inc) 41 , and RNAiMAX (Invitrogen) was employed as the transfection reagent following the supplier's recommended protocol, where non-targeting siRNA (Dharmacon, D-001210-02-20) was the control. The MIG and the MIG-JAK3 (wild-type) plasmids were kindly provided by Dr. Kara Johnson at Oregon Health and Science University. tryptic digestion of whole-cell protein lysates, and Lc-pRM analyses. A previously reported filter-aided sample preparation (FASP) protocol was used to generate tryptic peptides for LC-PRM analysis 12,42 . Four replicates (2 forward and 2 reverse SILAC labeling experiments) of lysates from the WM-115/WM-266-4 pair and 2 replicates (1 forward and 1 reverse SILAC labeling experiments) each of lysates from the IGR-39/IGR-37 and WM-793/1205Lu pairs were prepared for LC-PRM analyses. Peptide mixtures (500 ng) were subsequently dried in a Speed-vac, desalted with OMIX C18 pipette tips (Agilent Technologies), and analyzed by LC-MS and MS/MS on a Q Exactive Plus quadruple-Orbitrap mass spectrometer (Thermo Fisher Scientific) in the PRM mode. The mass spectrometer was coupled with an EASY-nLC 1200 system, and detailed LC-MS/MS conditions were described elsewhere 12 .
All raw files were processed using Skyline (version 3.5) for plotting the extracted-ion chromatograms and for peak integration 43 , where a previously described Skyline PRM kinome library was employed for PRM data acquisition and analysis 12 . Up to four most abundant distinct peptides for each kinase with at most one tryptic mis-cleavage site were included in the library. For peptide identification and quantification, we selected 6 most abundant y ions found in MS/MS obtained from shotgun proteomic analysis, where a ≤20 ppm mass accuracy was imposed in Skyline for fragment ions during peptide identification. We manually inspected all targeted peptides to make sure that the chromatographic profiles for fragment ions derived from the light and heavy forms of the same peptide can be overlaid. We also ensured that the distributions of relative intensities of multiple transitions associated with the same precursor ion are correlated with the theoretical distribution in the kinome MS/ MS spectral library entry, where we imposed a dot product (dotp) value 23 of at least 0.7. The sum of peak areas from all transitions of light or heavy peptides was used for quantification [44][45][46] , and no other adjustments were made. The relative standard deviations (RSD) for kinase protein quantification were 17.2%, 16.9% and 16.4% for WM-115/WM-266-4, IGR-39/IGR-37 and WM-793/1205Lu pairs of cell lines, respectively (Table S1). Hence, we imposed a 1.5-fold cutoff when considering the expression levels of a kinase to be significantly different in the paired primary/metastatic melanoma cells.
tcGA, ccLe and Geo data analysis. OncoLnc was employed for the analysis of The Cancer Genome Atlas (TCGA) data to reveal the correlation between patient survival and kinase gene expression 26 , where the top and bottom quartiles of expression values were considered as high and low groups, respectively. The Log-rank p-values for each kinase are calculated individually from the survival distributions of patients with high (top quartile) and low (bottom quartile) expression, and were not adjusted for multiple testing. Survival differences with logrank p-values of <0.05 were considered significant.
Scatter plots were generated for the mRNA expression in melanoma cell lines and tissues based on the data retrieved from the Cancer Cell Line Encyclopedia (CCLE) (https://portals.broadinstitute.org/ccle) 47 , and the GSE8401 dataset from the NCBI Gene Expression Omnibus (GEO). Box-and-whisker plot showing the mRNA levels of JAK3 gene in different types of cancer cells was obtained from CCLE database.
The detailed experimental conditions for Western blot, migration and invasion assay, and gel zymography assay are described in the Supplementary Materials.

Data availability
All the raw files for LC-PRM analyses of kinases were deposited into PeptideAtlas with the identifier number of PASS01179 (http://www.peptideatlas.org/PASS/PASS01179).