Encyclopedia
Please note this is a comparison between Version 1 by Ata Saba Moshiri and Version 2 by Jason Zhu.
Melanoma is a deadly skin cancer with rapidly increasing incidence worldwide. The progression from melanomagenesis to metastasis is known to differ between the various subtypes of melanoma, which are defined by their disparate clinical, histopathologic, and genetic features.
- Melanoma
- Genetics
- Transcriptomics
1. Introduction
Melanoma is now known to have genetically and epigenetically heterogenous clonal populations among both primary and metastatic tumors [1][2][63,64]. Most melanoma tumors are recognized to contain heterogenous cell populations with only small subpopulations that have metastatic potential [3][65]. Melanoma metastases were initially thought to disseminate from a primary tumor to regional lymph nodes and then to distant sites in a serial fashion [4][9]. However, this model has been questioned since circulating tumor cells can be present before regional or any metastasis at all, and there are patterns suggesting parallel progression in phylogenetic analyses [4][9]. Shain and Bastian propose that reseeding could occur over time from multiple cells to form melanoma metastases [4][9]. This would explain why regional metastases are often larger and appear to occur earlier. With reseeding, there is an increased probability for disseminating cells to land in nearer sites to the primary tumor as compared to distant ones, therefore resulting in a larger size [4][9]. It is well known that the risk of cutaneous melanoma metastasis correlates with Breslow depth of invasion and ulceration of the primary lesion, but a concrete set of metastatic molecular changes have yet to be elucidated [5][7]. Recently, a common metastatic evolution pathway has been elucidated for uveal melanoma and new research has shed light on the evolution of cutaneous metastatic melanoma.
2. Pathways of Uveal Melanoma Metastasis
A recent retrospective study by Shain et al. elucidated a common genomic evolution pathway that applied to majority of their metastatic uveal melanomas [6][53]. As mentioned before, Gαq pathway mutations are the earliest mutations to undergo selection, followed by gain of chromosomal arm 8q and secondary driver mutations in BAP1, SF3B1, or EIF1AX [6][53]. GNAQ loss of heterozygosity, additional chromatin remodeling mutations, and further ramp-up of 8q copy number occur at later points in the metastatic progression cascade [6][53]. At intermediate points in the evolutionary pathway, copy number alterations affecting chromosomes 16q, 8p, 1p, 6p, and 6q undergo selection [6][53]. Chromosome 8q gain and 3, 8p, and 16q loss together were found to be 85.9% predictive of uveal melanoma liver metastasis [7][24]. The tertiary driver mutations identified by Shain et al. include loss of function of CDKN2A, PBRM1, PIK3R2, and PTEN, gain of function of EZH2, PIK3CA, and MED12, and loss of heterozygosity in the GNAQ gain of function mutation [6][53]. Though this is the usual pathway of metastatic progression, there were 3 cases in their cohort of 35 where metastatic dissemination preceded 8q gain and 2 cases where metastatic dissemination potentially preceded biallelic loss of BAP1 [6][53]. This leaves room for metastatic dissemination to occur before the uveal melanoma has developed the full set of mutations [6][53].
3. Pathways of Cutaneous Melanoma Metastasis
Compared to uveal melanoma, the genomic evolution of metastasis in cutaneous melanoma is less clear. Soo et al. propose that the genetic drivers of melanomagenesis (e.g., BRAF, NRAS, and KIT) are first acquired, followed by RB pathway changes [8][66]. As the cutaneous melanoma progresses from a radial growth phase (RGP) to a vertical growth phase (VGP), mutations inhibiting apoptosis (e.g., TP53 or PTEN loss) emerge [8][66]. TERT promoter mutations then occur, conferring metastatic potential [8][66]. Birkeland et al. found that whole genome duplication (WGD) then occurs prior to most copy number alterations (CNAs) [9][67]. This is consistent with the recent findings of Vergara et al. that the acquisition of CNAs and aneuploidy due to both WGD and loss of chromosomes and chromosome arms are characteristic of metastatic cutaneous melanoma [10][68]. That study also elucidated that allele imbalances tended to occur on specific chromosomes. The authors hypothesized that with those genomic changes, specific alleles were being selected for and against [10][68]. In contrast with the role acquiring genetic variants plays in early melanomagenesis, mutational accumulation deviates from being UV-induced and is very low during metastasis [9][10][67,68]. Metastasis-specific genetic variants in cutaneous melanoma have yet to be elucidated in detail, but several genes have recently been implicated.
3.1. Novel AKT Pathway Metastasis Genes
3.2. Sex-Linked Metastasis Genes
Another study discovered that loss of DDX3X (DEAD-Box Helicase 3 X-Linked) is associated with metastasis formation and decreased distant metastasis-free survival [12][70]. DDX3X was found to regulate MITF protein levels and mutations were present in 5.8% of their melanoma cohort [12][70]. Females with melanoma are known to have a survival advantage, a lower risk of progression, and a decreased probability of nodal and visceral metastases [13][14][12,71]. One study found that the median time to metastasis was approximately 7 months greater in female patients as compared to males [15][72]. DDX3X may play a role in this disparity and perhaps future studies will identify other sex-chromosome-specific genes affecting survival and melanoma metastasis [12][70].
3.3. Germline Genes and Metastasis
The germline APOE4 variant was recently discovered to reduce progression and improve survival as compared to APOE2, despite the deleterious effect of APOE4 in Alzheimer’s disease development [16][73]. This study indicates the propensity for future research to identify other germline variants that may be used as prognostic melanoma biomarkers.
3.4. Database Candidate Metastasis Genes
A bioinformatic study using the TCGA and GEO databases identified nine candidate genes as being differentially expressed among primary and metastatic melanomas [17][74]. Their increased expression was also associated with worse prognosis in the TCGA SKCM cohort [17][74]. The previously identified genes in melanoma progression (AURKA, BUB1, and PRKCA) were also candidate genes in this study [17][18][19][20][74,75,76,77]. Another recent study using those databases as well as a transcriptome analysis of a genetically engineered mouse model (GEMM) identified 43 genes that could accurately predict patient prognoses and were upregulated in stage III/IV melanoma samples as compared to stage I/II ones [21][78]. From this study, GULP1, DAB2, P4HA2, and KDELR3 had their roles in promoting metastasis validated by siRNA knockdown [21][78]. A mechanistic analysis was further performed on KDELR3 and it was found to relieve ER stress in melanoma, an important function in the survival of metastases [21][78]. Several other recent studies have also used the TCGA and GEO databases to identify differentially expressed candidate genes in melanoma metastases [22][23][79,80]. They include those known to function in desmosomes, the keratinocyte cornified envelope, UV protection, leukocyte chemotaxis, WNT/β-catenin signaling, MAPK/ERK signaling, PI3K/AKT signaling, TGF-β signaling, VEGF signaling, and EMT-like program signaling [24][22][23][41,79,80].
4. Transcriptomics of Melanoma Metastasis
4.1. Transcription and Translation Factors
Transcription and translation factors regulate many important pathways in melanoma metastasis. EMT-like phenotype switching is important for metastatic melanoma progression, and it is orchestrated by MITF, AXL, and EMT-inducing transcription factors (EMT-TFs), such as TWIST, ZEB, and SLUG [25][81]. Melanocyte inducing transcription factor (MITF) (also known as microphthalmia-associated transcription factor) physiologically plays important roles in melanoblast survival, melanocyte development, and melanosome export. AXL is a receptor tyrosine kinase that is involved in inflammatory processes [25][26][81,82]. The high ZEB1/TWIST1, low MITF, high AXL state promotes an invasive, dedifferentiated phenotype in melanoma while the high ZEB2/SLUG, high MITF, low AXL state promotes an anti-invasive, proliferative, differentiated phenotype [25][27][81,83]. This is similar to the EMT/MET balancing observed in carcinomas and these factors likely act to promote different phenotypes to adapt to the specific stage and context of the melanoma [25][81]. A recent study demonstrated that activation of the invasive high ZEB1/AXL program through downregulation of SMAD7 also maintained proliferative ability and high MITF expression [28][84]. Patients with low SMAD7 expression in that study were found to have significantly decreased overall survival as compared to those with high expression [28][84]. Endothelin 1 (EDN1) has recently been identified as a regulator of phenotype switching heterogeneity. Through the GPCR endothelin receptor B (EDNRB), EDN1 supports MITF high populations while through endothelin receptor A (EDNRA), it supports AXL high populations [29][85]. The GPCR melanocortin receptor 1 (MC1R) is also known to regulate and increase MITF levels. Interestingly, it is rarely mutated in melanomas, though germline mutations predispose individuals to developing melanoma [1][30][63,86]. Tumors can contain cells expressing both high MITF and low MITF states. Rather than two distinct phenotypes, melanoma remains on a heterogenous spectrum of invasive and proliferative states, which allows for optimization to context and immune evasion [27][83]. This tumoral heterogeneity contributes significantly to metastasis and patient survival.
The translation-affecting proteins eIF4E and sequestosome 1 have also been implicated in metastasis progression. Carter et al. found that reduced survival and risk of metastasis are both correlated with increased eIF4E and phospho-eIF4E expression [31][97]. Phospho-eIF4E disproportionally promotes the translation of mRNAs associated with cancer progression, such as CCND1 and VEGF [31][97]. Sequestosome 1 also promotes the translation of prometastatic mRNAs, including FERMT2 [32][33][98,99]. It stabilizes these mRNAs via interactions with RNA-binding proteins and IGF2BP1 [32][98]. Though mRNA expression profiles play a significant role in melanoma progression, little is known about the role of RNA binding proteins in metastasis. Future research should continue to explore the role RNA binding proteins and other related factors have in the metastatic progression of melanoma.
4.2. Non-Coding RNA
Recent research has indicated the potential role that long non-coding RNA (lncRNA) and microRNA play in melanoma metastasis and their utility as biomarkers. lncRNAs tend to promote melanoma metastasis through increasing the expression of a transcript by acting as a decoy for its inhibitory microRNA. The lncRNAs that have been associated with melanoma metastasis and their corresponding microRNA and upregulated transcripts include KCNQ1OT1/miR-153/MET [34][100], LINC00518/miR-204-5p/AP1S2 [35][101], LINC00520/miR-125b-5p/EIF5A2 [36][102], and UCA1/hsa-miR-125b-1/AKT1 [37][103]. LINC00518 and LINC00520 were each found to be independent risk factors for metastatic melanoma patient survival and the network motif UCA1/hsa-miR-125b-1/AKT1 has an 88% chance of correctly identifying a TCGA sample as metastatic melanoma [35][36][37][101,102,103]. These studies demonstrate not only how lncRNA and miRNAs function in melanoma metastasis but also their potential to be prognostic and diagnostic biomarkers. Stark et al. found that a panel of seven serum microRNAs were able to detect metastatic melanoma with a high sensitivity (93%) and specificity (≥82%) [38][104]. This panel performed better at predicting overall survival, melanoma progression, and recurrence than serum LDH and S100B [38][104]. Armand–Labit et al. also found that plasma miR-1246 and miR-185 had a high sensitivity (90.5%) and specificity (89.1%) at identifying metastatic melanoma [39][105]. A small study found 25 miRNAs that were differentially expressed between pre and post-surgical metastatic melanoma plasma samples [40][106]. There is great potential and utility for lncRNA and miRNA, particularly non-invasive plasma miRNA, to be used as new biomarkers of melanoma metastasis.
5. Tumor Microenvironment
Though subclinical melanoma metastases have been found in many parts of the body, clinically detectable distant metastases are usually limited to skin, lung, brain, liver, bone, and intestine [3][65]. This suggests that tumor microenvironments play a significant role in melanoma metastasis. Further supporting this is the observation that some patients developed melanoma metastases after receiving organ transplants from supposedly disease-free individuals with a history of melanoma, ostensibly resulting from reactivation of the donor’s dormant metastatic disease [3][65]. A recent study by Tirosh et al. using single-cell RNA sequencing to analyze the transcriptome of melanoma tumors found that the tumor microenvironment is shaped by the intra- and interindividual spatial, functional, and genomic heterogeneity of melanoma and the associated tumor components [2][64]. A later study by Thrane et al. using in situ transcriptome analysis showed intra- and intertumoral gene expression heterogeneity in lymph node metastases. They also found that lymphoid tissue gene expression programs were potentially influenced by their distance from tumor cell clusters. In lymphoid tissues far away from tumor cell areas expression of the immune-related gene CD74 was observed while in tissues in close proximity to tumor cell areas, expression of the immune-related gene IGLL5 was seen. In the tumor cell cluster, PMEL and SPP1 were overexpressed [41][107]. It is thought that components of the adaptive immune system inhibit metastasis while innate immune system cells, such as macrophages, assist in mediating metastasis [3][65]. Overexpression of αvβ3/αIIbβ3 integrins on melanoma cells, which is mediated by EMT-TFs, is known to be associated with metastasis and the switch from RGP to VGP [25][42][81,108]. αvβ3-integrin has also been found to mediate extracellular matrix degradation and regulate the expression PD-L1, therefore promoting immune evasion [25][81]. Autocrine motility factor (AMF), a cytokine that is secreted from tumor cells, has been shown by Tímár et al. to be linked with in vivo spontaneous metastatic potential and greater β3 integrin expression [43][109]. CD44, a cell surface adhesion receptor that participates in lymphocyte activation as well as other immune functions, has also been linked with melanoma metastasis. Döme et al. found that the five year survival of patients with melanomas greater than 1mm thickness was significantly lower for those positive for the CD44v3 splice variant than those that were negative [44][110]. Other metastasis-associated factors on melanoma cells that affect the tumor microenvironment include MCAM/MUC18, L1-CAM, α4β1-integrin, ECM remodeling molecule FN1, and glypican-6 [25][81]. Melanoma is known to evade the immune system by increasing expression of PD-L1 (ligand for the T cell PD-1 receptor) on melanoma cells and CTLA-4 (an immune suppressive protein) on T cells. Their increased expression promotes immunosuppressive activity, such T cell anergy, Treg differentiation, and tumor infiltrating lymphocyte apoptosis. Progression is associated with the presence of Treg cells due to their ability to help melanoma avoid immune surveillance through the secretion of cytokines and chemokines (CXC) with immunosuppressive actions, such as IL-10, IL-35, and TGF-β. Melanoma also suppresses the immune system by expressing immune inhibiting miRNAs, depriving T cells of arginine and tryptophan, inhibiting natural killer cell and macrophage function through high levels of adenosine, inhibiting natural killer cell activity and APC immunogenicity through high levels of kinurenine, and the secretion of exosomes that modulate the tumor microenvironment. The immune evasion strategies have been covered in greater detail in previous reviews [45][111]. NRAS, KIT, EGFR, MET, RAB27A, and other pro-progression proteins have all been found within exosomes released by melanomas [46][112]. Melanoma exosomes are able to promote invasion, angiogenesis, and metastasis in receiving tumor cells [46][112]. They are also able to create premetastatic niches in the body [46][112]. Melanoma metastases are known to favor colonizing areas with higher exosomes and sentinel lymph nodes were found to have very high levels [46][112]. Lymph metastases are known to often precede the discovery of those in the blood and distant sites. A recent finding is that lymph prevents ferroptosis and reduces oxidative stress in metastasizing melanoma cells [47][113]. This is due to the reduced levels of free iron and elevated levels of glutathione and oleic acid in the lymph as compared to the blood [47][113]. They also discovered that melanoma cells that were previously exposed to lymph survived better in blood than those that had no exposure since they incorporated the antioxidants from the lymph [47][113]. Melanoma also has the ability to perform platelet and vasogenic mimicry. Platelet mimicry is characterized by the expression of the megakaryoctic factors and the GPCR PAR1 (protease activating receptor 1) is associated with this [48][114]. Future studies should continue elucidating novel factors affecting the tumor microenvironment and promoting metastasis.
6. Melanoma Progression during Targeted and Immune Therapies
The introduction of targeted and immunotherapies for melanoma has resulted in great improvements in patient survival. The identification of melanoma driver mutations in the MAPK/ERK pathway resulted in the development of BRAF small molecule inhibitors (vemurafenib and dabrafenib) and MEK small molecule inhibitors (trametinib and cobimetinib). Immune checkpoint inhibitors are also commonly used and include the cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4) inhibitor (ipilimumab) and the programmed cell death protein 1 (PD-1) inhibitors (pembrolizumab and nivolumab). Despite these advances in treatment, long-term success is still rare for late stage melanoma patients due to the development of drug resistance [26][82].
There is a great amount of genetic and epigenetic heterogeneity within melanoma tumors comprised of different clonal populations [1][63]. Therapeutic resistance could therefore arise from “Darwinian” selection of subclones that are able to survive treatment or from a “Lamarckian” selective pressure for tumors to acquire changes that allow them to persist [49][26][23,82]. Melanomas with NRAS-mutant cells and BRAF-mutant cells have greater resistance to BRAF inhibitors, likely due to selection of NRAS mutant populations. HOXD8 and RAC1 mutations have also been shown to give melanomas primary resistance to targeted therapies [1][63].
In patients taking BRAF and MEK inhibitors, MAPK/ERK pathway reactivation occurs in up to 80% of tumors through mechanisms such as alterations in MEK, NRAS, PTEN and NF1, and BRAF allele amplification and/or splice variants [1][50][63,115]. There can also be increased expression of alternate MAPK/ERK pathway activators, such as CRAF (C Rapidly Accelerated Fibrosarcoma Serine/Threonine Protein Kinase) and MAP3K8 (Mitogen-Activated Protein Kinase Kinase Kinase 8) [51][52][116,117]. Outside of restoring MAPK/ERK signaling, the PI3K/AKT pathway is commonly activated to compensate [1][63]. Note that multiple different mechanisms of resistance can occur in synchronous tumors of the same melanoma. One study found that 20% of melanoma patients after initiation of BRAF-inhibitor therapy developed two or more resistance mechanisms [1][53][63,118].
Drug resistance is also mediated by the tumor microenvironment and through transcriptomic changes. Stromal cell secretion of growth factors has been shown to promote MAPK/ERK and PI3K/AKT pathways [54][55][119,120]. Other reviews have discussed in more detail the ways the tumor microenvironment promotes resistance [56][121]. Multiple miRNAs have also been implicated in drug resistance [57][58][122,123]. One example is the upregulation of miR-211-5p and miR-204-5p via MITF and STAT3 (signal transduce and activator of transcription 3), which has been shown to reactivate MAPK/ERK and PI3K/AKT signaling, conferring therapeutic resistance [59][124]. Transcriptionally-regulated phenotype switching plays a major role in the ability of melanoma to evade therapies. The heterogeneity of melanoma having cells in both MITF high and low states and their ability to transition from one state to another is thought to confer drug resistance. Cells in the MITF low, AXL high, invasive, dedifferentiated state are slow-proliferating, and this allows them to avoid current therapies which primarily target fast-proliferating cells. In addition, AXL has been shown to promote the MAPK/ERK and PI3K/AKT pathways and this may also contribute to therapeutic evasion of these cells [26][82]. As a result, resistance to several targeted therapies is predicted by a low MITF/AXL ratio [60][125].
