Background
Wnt signaling proteins are a family of highly conserved proteins that are important during developmental processes. They have also been implicated in several diseases, such as cancer and diseases with an inflammatory component [
1,
2]. Canonical Wnt signaling, exemplified by WNT3A, is the most well characterized Wnt signaling pathway, leading to activation of β-catenin [
1,
3]. Non-canonical signaling via WNT5A can inhibit β-catenin signaling but also activate distinct signaling pathways, independent of β-catenin. Activation of non-canonical WNT5A signaling can lead to several different outcomes, such as activation of Ca
2+ signaling, activation of small Rho-GTPases (Cdc42, Rac1 and RhoA), Calmodulin-Kinase II, PKC, PKA, and JNK [
3‐
5].
Malignant melanoma is a highly aggressive cancer form which once spread, has a 5-year survival rate of 5% [
6]. For the tumor to spread to distant sites, the formation of new vessels is required, a process known as angiogenesis. In malignant melanoma, angiogenesis is correlated to the transition of the tumor from the radial growth phase to the more invasive vertical growth phase [
7]. Several secreted factors regulate angiogenesis such as VEGF, IL-6, Matrix metalloproteinase 2 (MMP2), IL-8 and FGF [
8,
9]. Some of these factors are also important in immunomodulation and the over-expression of these, by either the malignant melanoma cells or by infiltrating immune cells, can lead to enhanced metastasis due to induction of a local or systemic immunosuppression that is beneficial for the tumor cells escape from immune recognition and eradication [
10]. The later stages of melanoma including the spread to distant sites and the formation of metastasis have been shown to be promoted by an increased non-canonical WNT5A signaling. In line with this, a high WNT5A expression was also correlated to a poor prognosis in melanoma patients [
11]. This could partly be explained by the observations that; in vitro
, WNT5A increases migration and invasion of malignant melanoma cells [
12] and in vivo, WNT5A signaling increases the spread and tumor formation of lung metastasis [
13].
Exocytosis, or cytokine secretion, is a process with important implications in most tissues and cellular systems. Despite being widely studied, there are still questions to be answered regarding the molecular mechanisms behind this process [
14]. Briefly, activation of specific receptors causes an immediate release of preformed mediators from secretory granules. Regulated exocytosis pathways that are not constitutive in mode of action, are generally induced by an increased intracellular Ca
2+-signal. This signal causes a complex reorganization of the filamentous actin (F-actin) that is facilitated by cellular mediators such as the small Rho GTPases Cdc42 and Rac1 and the Synaptic soluble NSF attachment protein receptors (SNAREs). Among these are the proteins syntaxins, Soluble NSF Attachment Proteins (SNAPs) and vesicle-associated membrane proteins (VAMPs). The VAMPs can be divided into tetanus neurotoxin (TeNT)-sensitive and -insensitive VAMPs [
15]. Questions regarding the specific function and regulation of the actin cytoskeleton in secretory processes have been raised. However, an increase in intracellular calcium is necessary for cortical F-actin disassembly and its subsequent reorganization. Cdc42 and Rac1 have previously been shown to regulate the basolateral exocytosis and secretion of cytokines in polarized epithelial cells [
16]. It was also shown that the polarization of cytolytic effectors in immune cells was regulated by Cdc42 [
17].
Exosomes are 30–90 nm non-plasma membrane-derived vesicles that are formed in endosomal compartments called multivesicular endosomes and are released by a wide range of mammalian cells [
18,
19]. They contain various molecules ranging from endosomal markers (e.g. hsp70 and CD63) to signaling proteins (IL-6, IL-8, VEGF, Tissue inhibitor of metalloproteinases (TIMP-1/2) and FGFα) and mRNAs. The released exosomes merge with and empty their content into other cells, thus contributing to an intercellular communication. Tumor cells are known to have an exacerbated exosome secretion that has been linked to angiogenesis, metastatic spread and immunosuppression [
18]. Exosome secretion can be constitutive or regulated by for instance growth factors. The molecular mechanism involves tetraspanins (e.g. CD63), activation of the Rab family of proteins and probably also certain SNAREs (e.g. Rab5b) [
18‐
20]. Regulated exosome secretion can be Ca
2+-induced and dependent on cytoskeletal reorganization [
18,
19]. It has previously been shown that the exosome dependent protein Rab35, can mediate the transport of Cdc42 and Rac1 to the plasma membrane to remodel the actin-structures [
21]. We have previously shown that WNT5A induces an intracellular Ca
2+-increase in human malignant melanoma and breast cancer cells [
5,
22]. We have also shown that WNT5A induced a specific activation of Cdc42 and to some extent Rac1 in human breast cells [
5]. WNT5A has previously been shown to activate Cdc42 and induce cytoskeletal changes in fibroblasts [
23].
Here we show, that malignant melanoma cell lines treated with recombinant (r)WNT5A induces a prominent, immediate release of immunomodulatory and pro-angiogenic factors IL-6, IL-8, VEGF and MMP2, while transcriptional activation of these genes remained unaffected. The release was inhibited by calcium chelation and expression of a dominant negative Cdc42. Neither Brefeldin A nor TeNT inhibited the WNT5A-induced release of the soluble mediators. Instead we show that WNT5A induces release of exosomes containing IL-6, IL-8, VEGF and MMP2. Using gene expression data of 223 primary malignant melanomas from the study by Harbst et al. [
24], we further revealed a correlation between WNT5A expression and the angiogenesis marker ESAM. We also show that knock-down of WNT5A in malignant melanoma cells induced a decrease in endothelial cell branching in co-culture experiments with melanoma cells
in vitro, suggesting that WNT5A might have an effect on tumor progression in malignant melanoma, through induction of a broad release of soluble mediators.
Discussion
Previous studies have correlated high WNT5A expression to poor prognosis in melanoma patients [
11]. Angiogenesis and expression of immunomodulatory cytokines also correlate to poor prognosis in melanoma patients [
8]. Here we show that, in melanoma cells with low endogenous WNT5A expression, the stimulation with rWNT5A induces a rapid release of exosomes containing the immunomodulatory cytokine IL-6 and the pro-angiogenic factors IL-8, VEGF and MMP2. We also show that, when cells with a high endogenous WNT5A expression were depleted of WNT5A, the release was reduced, suggesting that WNT5A is functioning in an auto- or paracrine manner. These changes were not accompanied by a change in mRNA expression excluding a possible transcriptional induction of these genes upon WNT5A signaling. We instead show that the WNT5A-induced effects on the release of these pro-angiogenic and immunomodulatory factors, are due to a WNT5A/Ca
2+-regulated release of exosomes containing these mediators. In good agreement with this, the inhibitor of cAMP dependent Kinase (PKA), previously implicated in cytokine exocytosis [
41,
42], also inhibited the WNT5A-induced release of exosomes.
Changes in calcium signaling have been shown to induce a change in F-actin organization that can allow for secretory granules to reach the cell membrane and activate secretion [
43]. We and others have previously shown that WNT5A induces an intracellular calcium signal in melanoma cells [
22,
44] and breast cells [
5]. Also in this study, the Ca
2+-chelator Bapta, inhibited the effects of WNT5A. Reorganization of F-actin from the prominent cell cortex towards the cytoplasmic region has previously been reported to be part of the Ca
2+-dependent exocytosis mechanisms [
14]. This has lately been correlated to activation of the small RhoGTPases Cdc42 and Rac1 [
14,
16,
45] although the RhoGTPases probably mediate the vesicular trafficking primarily [
14]. WNT5A was also shown to activate both Cdc42 and to a lesser extent Rac1 in breast cells [
5]. Indeed, we could show that WNT5A induced activation of Cdc42 also in malignant melanoma Mewo cells and introduction of DN-Cdc42 and -Rac1 inhibited the WNT5A induced release.
Exosomes are produced by a wide range of mammalian cells and contribute to intercellular communications [
18,
19]. The exosomes may contain various molecules, also soluble mediators such as cytokines (eg. TGFβ and PGE2) [
19,
37], chemokines [
46] and angiogenic factors (eg. VEGF and MMP2) [
47]. It is known that exosomes act immunomodulatory on myeloid cells [
18,
48]. In immune cells, IL-6 and IL-8 mRNA expression has previously been connected to exosome signaling. Due to the fact that IL-6 mRNA levels remained unaffected in our study we believe that the mechanism behind WNT5A induced IL-6 release in malignant melanoma cells is Myd88/TLR2 independent [
38,
49]. Indeed, TLR2 was not expressed in the cell lines used in this study. We recently showed that also rWNT5A possess immunomodulatory effects on human monocytes. The findings in this study might explain why we observed a WNT5A specific induction of IL-6 and IL-10 mRNA in primary human monocytes that express TLRs at high levels specifically [
27]. We now show, that in malignant melanoma cells, the induction of immunomodulatory and pro-angiogenic mediators are not caused by transcriptional activation, but by exosome-release of already formed proteins. Tumor cells generally have an increased exosome secretion that has been linked to angiogenesis, metastatic spread and immunosuppression [
18]. A previous study even showed that the tumor microenvironment was able to specifically promote sorting of immunosuppressive factors into exosomes [
37].
Exosome secretion is dependent on cytoskeletal reorganization and although it has previously been shown that the exosome dependent protein Rab35, can mediate the transport of Cdc42 to the plasma membrane to remodel the actin-structures [
21], Cdc42 itself has not been connected to exosome release in mammals. In yeast however, Sec4p a Rab5b related protein, was shown to interact with Cdc42 to induce exocytosis [
50]. Just as in the WNT5A induced malignant melanoma exosomes, Rab5b was expressed in plasma-derived exosomes from malignant melanoma patients [
51]. Also, Rab5b was recently shown to participate in exosome-formation in malignant melanoma cells [
20]. We propose that the mechanism behind the WNT5A induced exosome-release is Ca
2+- and small RhoGTPase-regulated, affecting downstream proteins such as Rab5b and related Rab-family proteins [
18‐
20,
51]. It should not be excluded that also other Ca
2+-regulated proteins, affecting cortical F-actin disassembly, could affect the exosome release [
52,
53]. In this study, even the canonical WNT3A protein induced an increase in exosome-release. We suggest that an independent signaling pathway, distinct from the Ca
2+-induced, non-canonical WNT5A-pathway, causes this release. Strengthening this hypothesis, the exosomes produced by canonical WNT3A displayed a different content as compared to those produced by non-canonical WNT5A signaling. Or it could be explained by the recent finding that indeed Wnt proteins (WNT3A and WNT5A) are secreted on exosomes [
54,
55]. It has long been known in the Wnt field that WNT5A can induce expression of it self. The mechanism behind has not been explained and the data in this study together with the mentioned studies might partly explain how this loop could work.
Although numerous articles describing the effects of WNT5A on intracellular signaling proteins have been published, few studies concerning WNT5A and transcriptional regulation are available. We also show that WNT5A does not affect the factors analyzed in this study at the transcriptional level.
Methods
Cell culture and treatments
All malignant melanoma cell lines were purchased from the American Type Tissue Collection (ATCC, Manassas, VA). MS1 murine endothelial cells [
56] were a gift from Professor Kristian Pietras, (Lund University, Sweden). Recombinant proteins (rWNT5A (0.2 μg/ml) or rWNT3A (0.05 μg/ml) were from R&D systems, Minneapolis, MN. All chemicals and inhibitors were purchased from Sigma Aldrich (St Louis, MO.) unless otherwise noted and used at concentrations: 10 μM Bapta-AM or DMSO control, 5 μM PKA inhibitor H89, 10 ng/ml Tetanus Toxin or 10 μg/ml Brefeldin A. The antibodies used were Goat anti-WNT5A (R&D systems, Minneapolis, MN), anti-CD63, anti-Rab5B and anti-GAPDH (Santa Cruz biotechnologies), mouse anti-β-actin (MP biomedicals). For BD Cytometric Bead Array (CBA) the Human Inflammation Kit was used (BD Biosciences). For hematoxylin and eosin cells were fixed in 4% PFA and embedded in paraffin and then stained. F-actin cytoskeleton was stained using Alexafluor546-coupled Phalliodin. For ELISAs Quantikine Human IL-6 ELISA kit, Quantikine Human MMP2 ELISA kit and Quantikine Human VEGF ELISA kit from R&D systems (Minneapolis, MN) was used. Exosome Elisa (ExoELISA) CD63 ExoELISA
TM from System Biosciences (Uden, The Netherlands) was used.
RNA extraction and RT-QPCR
mRNA expression was analyzed using RT-QPCR. RNA was extracted using RNeasy kit (Qiagen, Hilden, Germany). For primer sequences, see Additional file
1: Figure S4 and [
57].
Transfection and knockdown experiments using siRNA
For overexpression of CA-Cdc42, DN-Cdc42 and Rac1 (N17), Mewo cells were transfected using 1 μg plasmid (a kind gift from Dr Pontus Aspenström, (Karolinska Institutet, Sweden) per 24-well. 2 μg CA-Cdc42, DN-Cdc42 or pcDNA3.1 was used per 6-well for ExoELISA experiments. For knockdown experiments 150000 HTB63 cells were transfected with Silencer Select siRNA against WNT5A (10nM, Applied Biosystems Carlsbad CA) and the appropriate concentrations of scrambled siRNA as control.
MS1 Co-culture experiments
Co-culture experiments were performed as previously published [
58]. For exosome stimulation experiments, MS1 cells were seeded in gelatine-coated 12-well plates and stimulated with exosome-enriched or –depleted fractions from carrier or rWNT5A (0.3 μg/ml, 3 h) stimulated Mewo cells (see below). The primers used for angiogenesis specific Q-PCR of the MS1 cells have been published previously [
57].
Exosome isolation
Mewo cells were cultured over night in serum free Eagle’s Minimum Essential Medium supplemented with penicillin/streptomycin that had been Ultracentrifuged for 2 h at 100,000 g using Beckman Optima TLX Ultracentrifuge. The medium was replaced by fresh Ultracentrifuged medium and cells were stimulated with carrier, rWNT3A or rWNT5A as indicated for 3 h. The supernatants were collected and exosome isolation was carried out by differential centrifugation. Briefly, the supernatants were centrifuged for 300 g for 10 min to remove debris. The supernatants were then centrifuged once at 2,000 g for 15 min, once at 10,000 g for 30 min and finally exosomes were pelleted at 100,000 g for 1,5-2 h.
Exosome identification by electron microscopy
The exosome enriched pellet was diluted in 50 μl PBS and kept at 4°C until EM analysis. A drop of the exosome sample was placed on a carbon coated copper grid and was let to adhere for 1 min. The sample was contrast stained by adding a drop of 2% uranyl acetate to the sample on the grid. Excess liquid was removed by gently using an absorbing paper, before positioning the grid on a paper with the coated side up and was let to air dry for 5 minutes. The preparation was examined using an electron microscope FEI Tecnai spirit at 100 KV.
microRNA (miRNA) expression analysis
MicroRNA expression analyses were performed using Affymetrix miRNA-3_0 arrays according to the manufacturer’s instructions (Affymetrix, Santa Clara, CA). These array experiments were performed at Swegene Centre for Integrative Biology (SCIBLU) at Lund University. Microarray data were initially pre-processed and normalized using Robust Multi-array Analysis (RMA) method [
59]. These analyses were performed using Affymetrix Expression Console Software v1.1.2. To identify significantly differentially expressed miRNAs between carrier stimulated and rWNT5A stimulated Mewo-exosomes, we used significance analysis of microarrays (SAM) method [
60]. SAM analyses were performed using TMEV v4.0 software.
Statistical analysis
All data was analyzed using Graphpad Prism software and is visualized as mean with error bars representing standard deviation (SD) or standard error of the mean (SEM). Statistical significance was calculated using ANOVA or Student’s
t-test as indicated in the Figure legends. Spearman’s rank test was used for the statistical analysis of the gene expression data set. The malignant melanoma mRNA data set has previously been published [
24] and approved by the local ethics committee of Lund University (Dnr 191/2007). The mRNA was prepared from paraffin embedded samples of 223 primary malignant melanomas.
Competing interests
T.A. is a shareholder of WntResearch and is also part-time CSO of the company. The other authors declare no conflict of interest.
Authors’ contributions
EJE performed the majority of the experiments, designed experiments and wrote the manuscript together with KL and TA, CB, VB, and FS performed experiments. EC performed the electron microscopy. GJ was responsible for the microarray analyses. KL was responsible for designing the study and writing the final manuscript. All authors read and approved the final manuscript.