Loss of SR-BI Down-Regulates MITF and Suppresses Extracellular Vesicle Release in Human Melanoma

Melanoma is a skin tumor with a high tendency for metastasis and thus is one of the deadliest cancers worldwide. Here, we investigated the expression of the scavenger receptor class B type 1 (SR-BI), a high-density lipoprotein (HDL) receptor, and tested for its role in melanoma pigmentation as well as extracellular vesicle release. We first analyzed the expression of SR-BI in patient samples and found a strong correlation with MITF expression as well as with the melanin synthesis pathway. Hence, we asked whether SR-BI could also play a role for the secretory pathway in metastatic melanoma cells. Interestingly, gain- and loss-of-function of SR-BI revealed regulation of the proto-oncogene MET. In line, SR-BI knockdown reduced expression of the small GTPase RABB22A, the ESCRT-II protein VPS25, and SNAP25, a member of the SNARE complex. Accordingly, reduced overall extracellular vesicle generation was detected upon loss of SR-BI. In summary, SR-BI expression in human melanoma enhances the formation and transport of extracellular vesicles, thereby contributing to the metastatic phenotype. Therapeutic targeting of SR-BI would not only interfere with cholesterol uptake, but also with the secretory pathway, therefore suppressing a key hallmark of the metastatic program.


Introduction
The scavenger receptor class B type 1 (SR-BI, gene name SCARB1) has emerged as a novel tumor marker, since its expression in different types of tumor cells has been associated with progression of the disease [1]. Studies showed reduced cell proliferation upon inhibition of SR-BI in human neuroblastoma, prostate carcinoma, clear-cell renal cell carcinoma and breast cancer cell lines [2][3][4][5]. The analysis of patient samples from prostate, breast and nasopharyngeal carcinoma highlighted a correlation between high SR-BI expression and worsening of the disease [6][7][8][9].
SR-BI is a so-called high-density lipoprotein (HDL) receptor [10][11][12] and known for its role in reverse cholesterol transport, namely the selective transfer of the cholesterol moiety to the hepatocytes.
In contrast to the low-density lipoprotein receptor pathway, in which the endocytosed particle is completely dismantled, the HDL particles are not degraded, but transported to multivesicular bodies from which they can be exocytosed again [13,14].
Thus, the question arises as to why tumor cells express this specialized receptor and the related functional effects.
Among all tumors, melanoma represents a highly aggressive tumor entity and SR-BI plays an important role in maintaining the metastatic phenotype in this tumor [15]. Specifically, this includes the process of the epithelial-to-mesenchymal transition (EMT), migration, invasion and increased cellular glycosylation. Therefore, besides its role in mediating selective cholesteryl ester uptake from HDL particles to cells, SR-BI facilitates tumor promotion.
The secretory pathway is a key cellular process determining melanoma aggressiveness and has a major impact on clinical outcome. It involves membrane trafficking at the endoplasmic reticulum and Golgi as well as the sorting of proteins to the cytoplasm membrane. Here we investigated the involvement of SR-BI in the secretory phenotype of melanoma. By testing a large patient cohort, we identified the co-expression of SR-BI with the MITF transcription factor. During mammalian development MITF promotes the transition of precursor cells to melanoblasts [16]. In adult melanocytes, MITF is mainly known to regulate melanosome formation and melanin production [17,18]. MITF governs the melanocyte-specific melanin synthesis pathway with tyrosinase as an essential enzyme [19], and in melanoma it is one of the main drivers of proliferation and invasiveness [20]. Interestingly, melanoma cells display varying amounts of MITF, which can be used to identify different subpopulations within melanoma lesions [21][22][23]. A well-known MITF target is the proto-oncogene MET, which we found to be regulated by SR-BI knockdown as well as over-expression. MET receptor has been shown to correlate with the metastatic phenotype [24]. Additionally, MET governs exosome release in melanoma leading to enhanced metastatic colonization [25,26]. Accordingly, we investigated whether SR-BI is also involved in the generation of extracellular vesicles.

SR-BI Expression Pattern in Melanoma Is Associated with the Pigmentation Pathway
When primary melanoma metastasizes, the most proximal sites are regional cutaneous metastases, followed by lymph node metastatic sites and finally distant metastatic sites. In order to display the distribution of SR-BI expression at each location, we used data derived from the Cancer Genome Atlas (TCGA) consortium and found highest SR-BI expression in primary lesions as well as in distant metastatic sites ( Figure 1A). Interestingly, sorting for SR-BI expression intensity from highest to lowest revealed that MITF and its target tyrosinase (gene name TYR) segregated accordingly ( Figure 1B). The correlation analysis confirmed this result ( Figure 1C). To test whether SR-BI is necessary for the pigmentation phenotype in melanoma we used the melanoma cell line VM21, derived from early growth phase melanoma, and WM3060, derived from metastatic melanoma, to investigate the effect of SR-BI siRNA treatment. Control or SR-BI-depleted human melanoma cells were pelleted and displayed diminished brown pigmentation ( Figure 1D, upper panel). Accordingly, the MITF protein was highly expressed in WM3060 cells and showed a strong reduction after SR-BI knockdown. VM21 cells, which expressed only moderate amounts of MITF, displayed only a slight reduction after knockdown of SR-BI ( Figure 1D, lower panel).

SR-BI Controls the Proto-Oncogene and Vesicle-Release-Driver MET
MITF is a driver of melanin synthesis and can activate the transcription of MET, an important receptor tyrosine kinase involved in tumor cell growth and release of extracellular vesicles. We used two well-characterized human melanoma cell lines derived from metastatic sites and performed SR-BI knockdown. mRNA harvested 48 hours after siRNA treatment showed a reduction of SR-BI and down-regulation of the MET receptor ( Figure 2A). Protein analysis showed a strong reduction of SR-BI and a concomitant MET reduction in both cell lines ( Figure 2B). Over-expression experiments for SR-BI indicated that SR-BI is able to positively regulate MET expression ( Figure 2C). Since the release of extracellular vesicles contributes to the malignant phenotype in metastatic melanoma, we next quantified the mRNA levels of regulatory proteins in the secretory pathway. SR-BI removal led to the diminished expression of RAB22a, SNAP25 and VPS25, which suggested reduced generation and secretion of tumor-derived extracellular vesicles ( Figure 2D).

SR-BI Controls the Proto-Oncogene and Vesicle-Release-Driver MET
MITF is a driver of melanin synthesis and can activate the transcription of MET, an important receptor tyrosine kinase involved in tumor cell growth and release of extracellular vesicles. We used two well-characterized human melanoma cell lines derived from metastatic sites and performed SR-BI knockdown. mRNA harvested 48 hours after siRNA treatment showed a reduction of SR-BI and down-regulation of the MET receptor ( Figure 2A). Protein analysis showed a strong reduction of SR-BI and a concomitant MET reduction in both cell lines ( Figure 2B). Over-expression experiments for SR-BI indicated that SR-BI is able to positively regulate MET expression ( Figure 2C). Since the release of extracellular vesicles contributes to the malignant phenotype in metastatic melanoma, we next quantified the mRNA levels of regulatory proteins in the secretory pathway. SR-BI removal led to the diminished expression of RAB22a, SNAP25 and VPS25, which suggested reduced generation and secretion of tumor-derived extracellular vesicles ( Figure 2D).

Extracellular Vesicle Secretion Is Altered after SR-BI Inhibition
We isolated extracellular vesicles from the supernatant of control cells and SR-BI depleted cells of two metastatic human melanoma cell lines. In Figure 3A,B (upper panel) we show the knockdown efficiency as demonstrated by immunocytochemical staining for SR-BI. In Figure 3A,B (lower panel), we calculated the vesicle distribution as quantified by Nanosight of the isolated vesicles in both melanoma cell lines. The analysis revealed that the vesicle count per volume decreased by 20% and 50%, according to the cell model used ( Figure 3C). Since vesicles are known to be the carriers of tumor promoting factors, we ruptured harvested vesicles by repeated freeze-thaw cycles and measured VEGFA by ELISA. Equal amounts of vesicles derived from control treated or SR-BI siRNA treated cells were used and a significant reduction of VEGFA was detected ( Figure 3D). Finally, to test whether vesicles could attach to cells we labeled isolated vesicles with a fluorescent dye. Melanoma cells were incubated with labeled vesicles and 30 minutes after exposure the cells were formaldehyde fixed. Confocal image acquisition showed that green fluorescent vesicles of a very small size could be detected at the cellular membrane and within the cellular cytoplasm ( Figure 3E).

SR-BI and MET Expression Correlates in Patient Data Sets
To strengthen our finding of SR-BI regulating MITF as well as MET, we made use of two published melanoma whole genome expression studies. There we found a positive correlation between SR-BI and MET (Figure 4A,B). Additionally, we divided the data according to the median expression of MITF into low and high expression groups, indicated by blue and red dots, respectively. The data demonstrate that only the samples that expressed high amounts of SR-BI and MET also displayed high MITF. This is of particular interest since MET is a discriminator of patient survival, as can be seen in two mRNA data sets GSE19234 ( Figure 4C) and SKCM-TCGA ( Figure 4D) by Kaplan-Meier analysis.
Our data indicate that high mRNA levels of MET led to a worsened survival outcome for patients, associated with high MITF and SR-BI expression.  Values were calculated using a log-rank test.

Discussion
We have shown that SR-BI in human melanoma is important for the pigmentation phenotype driven by the MITF transcription factor. Furthermore, SR-BI regulated the expression of the MET proto-oncogene, and the loss of SR-BI resulted in the down-regulation of molecules involved in the secretory pathway. Measurements of extracellular vesicle size and density revealed that without SR-BI, vesicle size increased and vesicle generation per donor cell was reduced. Importantly, the released vesicles contained less of the pro-tumorigenic factor VEGFA when derived from SR-BI knockdown cells. Hence, we could demonstrate that SR-BI plays an important role in metastatic melanoma by enhancing the secretory process encompassing extracellular vesicles as well as tumor-derived growth factors.
Melanin synthesis driven by MITF depends on the mammalian secretory pathway and is highly sensitive to defects in the glycosylation pathway as well as defects in ER, Golgi and Post-Golgi transport [27]. We have shown that the loss of SR-BI reduces the ability to glycosylate proteins at the ER-Golgi, which resulted in the reduced secretion of VEGFA [15]. Hence our data presented here further link SR-BI to the secretory pathway, which includes the pigmentation process, but which also includes pro-tumorigenic extracellular vesicle formation and release.
Interestingly, MITF represents the major regulator of the pigmentation pathway, and at the same Values were calculated using a log-rank test.

Discussion
We have shown that SR-BI in human melanoma is important for the pigmentation phenotype driven by the MITF transcription factor. Furthermore, SR-BI regulated the expression of the MET proto-oncogene, and the loss of SR-BI resulted in the down-regulation of molecules involved in the secretory pathway. Measurements of extracellular vesicle size and density revealed that without SR-BI, vesicle size increased and vesicle generation per donor cell was reduced. Importantly, the released vesicles contained less of the pro-tumorigenic factor VEGFA when derived from SR-BI knockdown cells. Hence, we could demonstrate that SR-BI plays an important role in metastatic melanoma by enhancing the secretory process encompassing extracellular vesicles as well as tumor-derived growth factors.
Melanin synthesis driven by MITF depends on the mammalian secretory pathway and is highly sensitive to defects in the glycosylation pathway as well as defects in ER, Golgi and Post-Golgi transport [27]. We have shown that the loss of SR-BI reduces the ability to glycosylate proteins at the ER-Golgi, which resulted in the reduced secretion of VEGFA [15]. Hence our data presented here further link SR-BI to the secretory pathway, which includes the pigmentation process, but which also includes pro-tumorigenic extracellular vesicle formation and release.
Interestingly, MITF represents the major regulator of the pigmentation pathway, and at the same time amplification of the MITF locus was shown to be associated with melanoma metastasis [28]. Therefore, MITF harbors a wide range of functions including melanocytic lineage restriction, cell cycle regulation, survival and migration. Since we were able to show that MITF expression depends on the presence of SR-BI, it is interesting to note that SR-BI could be considered an upstream regulator of MITF. This regulation might occur via the AKT signaling pathway, which has been shown to be induced upon cholesterol transport by SR-BI in different cellular models [29][30][31]. Studies in melanocytes have shown that AKT activation is necessary for GSK3beta inhibition leading to MITF expression [32]. Additionally, the application of excess cholesterol, which can be transported by SR-BI [33], induced pigmentation and upregulation of MITF in human melanocytes and melanoma cells [34]. High MITF activity leads to expression of its target genes. Among these, SR-BI has been identified [35]. Taken together, this implies that SR-BI represents a potent inducer of MITF and that MITF itself is connected to SR-BI by a positive feedback loop, which could explain the strong correlation we observed in the patient samples.
Another important downstream target of MITF is the MET receptor [36]. MET expression together with dopachrome tautomerase (TYRP2), which is part of the melanin synthesis process, is part of a melanoma signature and prognostic for tumor progression [25]. Thus, the pigmentation process is linked to vesicle generation and release, which is crucially involved in tumor progression.
The budding of the plasma membrane during ectocytosis requires Rab GTPases, the endosomal sorting complex (ESCRT) and possibly also SNARE proteins. The HDL components phosphatidyl-inositols and phosphatidyl-ethanolamine also play an important role in extracellular vesicle budding [37]. We detected significantly fewer extracellular vesicles, characterized by a diameter bigger than 90 nm, upon the depletion of SR-BI. In line, we were able to identify the down-regulation of RAB22A, VPS25 and SNAP25. RAB22A is important for early endosomal sorting and its over-expression promotes melanoma growth [38]. VPS25 is a member of the ESCRT-II complex and is able to recruit ESCRT-III, which executes vesicle budding and scission [39]. SNAP25 is involved in the regulation of neurotransmitter release and is localized in the presynaptic plasma membrane. Since melanoma cells show a neuronal phenotype it could be concluded that SNAP25 is important for neuronal-like vesicle generation. We suggest that extracellular vesicles released by melanoma cells could lead to the autocrine stimulation of tumor cells or even be involved in preparing a metastatic niche by fusing with proximal or distal stromal cells.
Taken together, SR-BI influences vesicle generation and release at multiple levels. This includes the role of SR-BI in supplying cancer cells with cholesterol as well as its role in ER-Golgi-associated protein glycosylation. Here we identified MITF, the MET receptor as well as several members of the secretory pathway as being regulated by SR-BI in human melanoma cells. Moreover, patient sample analysis confirmed the correlation of SR-BI with MITF and MET in metastatic melanoma lesions. Our data pinpoint SR-BI as an important functional node for enabling not only melanin synthesis and potential melanosome formation, but also for generating tumor promoting extracellular vesicles. Therefore, our results show that SR-BI represents an important player in disease progression in melanoma. We suggest that the inhibition of SR-BI will interfere with the metastatic phenotype and in combination with state of the art treatment might lead to regression of the disease.

Cell Culture and Melanin Measurement
Patient-derived melanoma cell model MCM1 has been characterized previously [24]. Human melanoma cell lines 1205Lu was purchased from ATCC and WM3060 from Coriell. The cell line VM21 was a gift from Walter Berger and has been described previously [40]. All cells were cultivated under standard conditions and maintained in melanoma isolation medium (MIM) supplemented with 2% FCS, except the cell line VM21, which was grown in RPMI (Lonza, Basel, Switzerland) containing 10% FCS. MIM contains 80% MCDB153, 20% Leibovitz s L-15, 1.23% sodium bicarbonate solution, 0.5 ng/mL epidermal growth factor, 5 µg/mL human insulin, 1.68 mM calcium chloride (all from Sigma-Aldrich, St. Louis, MO, USA), 50 mg/l streptomycin sulfate, 30 mg/L penicillin (both from GE Healthcare, Buckinghamshire, United Kingdom), and 2 µg/mL ciprofloxacin (Sigma). Pelleted cells were imaged by a color camera. The pixel density of the identical pellet area was calculated in the blue color channel at 450 nm.

Transient SR-BI Overexpression
Human melanoma cells were seeded in a 6-well plate and transduced with PEI (Polysciences Inc., Warrington, PA, USA) and 3 µg wildtype plasmid or plasmid containing human SR-BI (pcDNA™5/FRT Vector, Invitrogen, Paisley, UK). Overexpression efficiency was tested by Western blot analysis.

siRNA Knockdown
Reagents for siRNA knockdown were obtained from Dharmacon (GE Healthcare). Cells were transfected with DharmaFECT 1 and 5 nM ON-TARGETplus SMARTpool siRNA against SCARB1 (L-010592) or non-targeting siRNA (D-001810) for 24 hours followed by mRNA isolation (72 h post-transfection) and protein isolation (96 h post-transfection). Knockdown efficiency was tested by RT-PCR and Western blot analysis.

RNA Isolation and Real-Time PCR
RNA was isolated according to the manufacturer's instructions (peqGOLD total RNA Kit). cDNA synthesis and polymerase chain reaction (PCR) were performed. Primer pairs are listed in Table 1. Values were normalized to human ACTB mRNA expression. Results were quantified using the Delta C(T) method.

Western Blotting
Cells were lysed with RIPA buffer (Cell Signaling, Frankfurt am Main, Germany) on ice, and SDS-PAGE and Western blotting was performed according to standard protocols. Nitrocellulose membranes (Biorad, Munich, Germany) were incubated with the antibodies listed in Table 2. Semi-quantitative densitometric analyses of the detected bands were performed with the corresponding software from Biorad according to standard procedures.

Immunocytochemistry and Immunofluorescence
Cells grown in chamber slides were fixed with 4% formaldehyde, washed with PBS, treated for 5 min with 1% Triton X-100, washed, blocked with 1% bovine serum albumin and then stained with specific antibodies at 4 • C overnight (see Table 2). IF images were taken with a Leica TCS SP8 microscope (Leica Microsystems, Wetzlar, Germany).

Vesicle Isolation
The 60 × 10 6 human melanoma cells were grown in serum-free MIM for 48 h. Collected supernatant fractions were filtered with a 0.2 µm filter (Sarstedt, Nümbrecht, Germany). Vesicles were pelleted by ultracentrifugation at 34800 rpm for 70 min at 4 • C (Beckman Coulter 55.2 Ti rotor, Krefeld, Germany). The pellet, containing vesicles, was resuspended in 100 µl PBS. Vesicle concentration and size were measured using the Nanosight technology equipped with a blue laser (405 nm) for real-time characterization of the vesicles.

Statistical Analysis
All values are given as means ± standard error of the mean (S.E.M.), except when noted otherwise. Figures were analyzed and drafted with GraphPad Prism ® 6. Western blots and RT-PCRs were repeated at least two times with at least two replicates. Kaplan-Meier plots and scatter plots were generated from the publicly available data sets GSE19234 [41], GSE7553 [42], GSE3189 [43] and the SKCM-TCGA study on cutaneous melanoma [44]. Pearson correlation analysis was performed to calculate the P-values in displayed scatter blots. Differences in gene expression were assessed for statistical significance by an unpaired two-tailed t test. Log-rank (Mantel-Cox) test was performed for Kaplan-Meier analysis. P < 0.05 was defined as statistically significant and P-values are displayed as follows: * P < 0.05; ** P < 0.01; *** P < 0.001; and **** P < 0.0001.