Flavonoids increase melanin production and reduce proliferation, migration and invasion of melanoma cells by blocking endolysosomal/melanosomal TPC2

Two-pore channel 2 (TPC2) resides in endolysosomal membranes but also in lysosome-related organelles such as the melanin producing melanosomes. Gain-of-function polymorphisms in hTPC2 are associated with decreased melanin production and blond hair color. Vice versa genetic ablation of TPC2 increases melanin production. We show here an inverse correlation between melanin production and melanoma proliferation, migration, and invasion due to the dual activity of TPC2 in endolysosomes and melanosomes. Our results are supported by both genetic ablation and pharmacological inhibition of TPC2. Mechanistically, our data show that loss/block of TPC2 results in reduced protein levels of MITF, a major regulator of melanoma progression, but an increased activity of the melanin-generating enzyme tyrosinase. TPC2 inhibition thus provides a twofold benefit in melanoma prevention and treatment by increasing, through interference with tyrosinase activity, the synthesis of UV blocking melanin in melanosomes and by decreasing MITF-driven melanoma progression by increased GSK3β-mediated MITF degradation.

www.nature.com/scientificreports/ and redheads have a greater risk for melanoma than black or brown haired individuals (by a factor of 2-4). The pheomelanin/eumelanin ratio accounts for some of this risk 3 . Ambrosio et al. (2016) have recently shown that knockout of two-pore channel TPC2 in human MNT-1 melanoma cells elicits a strong increase in pigment content and that this effect can be reversed by transient overexpression of TPC2-GFP 4 . Vice versa, Sulem et al. (2008) have shown that certain TPC2 polymorphisms, rs35264875 (encoding TPC2M484L) and rs3829241 (encoding TPC2G734E) are associated with reduced pigmentation and a higher probability for blond hair in humans 5 . Chao et al. (2017) investigated the functional effects of these variations on the channel properties using endolysosomal patch-clamp electrophysiology and found that both polymorphisms are gain-of-function (GOF) variants 6 . Furthermore, it has been shown recently that pharmacological or siRNA mediated inhibition of TPC2 abrogates migration of cancer cells and the formation of metastases 7 . However, a more detailed mechanistic understanding of how TPC2 activity and expression in melanosomes on the one hand and endolysosomes on the other hand affect melanoma cells is lacking. Here, we used MNT-1 human melanoma cells to assess the effect of genetic ablation or pharmacological inhibition of TPC2 on proliferation, migration, and invasion as well as melanin production. Surprisingly, we found that melanoma cell proliferation, migration, and invasion are inversely correlated with TPC2-dependent melanin production as loss of TPC2 increases melanin content but decreases proliferation/migration/invasion. This is possible due to independent mechanisms: via regulation of MITF (microphthalmia-associated transcription factor) protein levels through interference with endolysosomal activity of TPC2 and endolysosomal GSK3β degradation on the one hand and on the other hand via MITF-independent regulation of tyrosinase activity by TPC2 in melanosomes. Similar effects were found with novel, flavonoid based inhibitors of TPC2, corroborating a new treatment option for melanoma using TPC2 as a pharmacological target. Flavonoids, previously proposed anti-cancer agents thus emerge as direct inhibitors of TPC2 and the higher risk for blond haired individuals or individuals with light pigmentation to develop melanoma may be directly correlated to TPC2 activity and GOF variation.

Human TPC2 −/− melanoma cells show reduced proliferation, migration, and invasion but increased melanin production and tyrosinase activity. Genetic ablation or inhibition of TPC2
has been demonstrated before to affect cancer cell migration and the formation of metastases 7,8 as well as neoangiogenesis 9 . We show here that knockout of TPC2 in MNT-1 human melanoma cells results in significant decrease in melanoma cell proliferation, migration, and invasion ( Fig. 1a-f). Ambrosio et al. (2016) have recently reported that TPC2 knockout elicits a strong increase in pigment content in MNT-1 cells 4 . Consistently, siRNA-mediated knockdown of TPC2 was also found to cause a substantial increase in melanin content in MNT-1 cells and primary human melanocytes 4 . We recapitulated these findings and confirmed the strong increase in melanin production in TPC2 −/− cells (Fig. 1g,h). We further found a significant increase in the activity of tyrosinase, which is the key enzyme responsible for efficient melanin production with a pH optimum at 6.8 ( Fig. 1i). At the same time, we also found increased protein levels of tyrosinase in TPC2 −/− MNT-1 cells while transcript levels were unchanged (Fig. S1a-d and S1g). In contrast to tyrosinase, expression levels of other proteins involved in melanogenesis or regulated by MITF such as Dct (dopachrome tautomerase, TYRP2), Rab27a, or PMEL (premelanosome potein) were unchanged ( Fig. S1c-f).

Flavonoids from a Southeast Asian plant extract affect melanin production in melanoma cells.
In an unbiased approach to identify compounds that affect melanin production in melanoma cells, we screened several plant extracts in B16F10 mouse melanoma cells. An extract prepared from Dalbergia parviflora (D. parviflora), also called Akar Laka which is mainly found in lowland tropical areas, in particular in Myanmar, Thailand, Malaysia, Indonesia, and the Philippines showed the strongest effect alongside Kaempferia parviflora (K. parviflora), also called Thai ginseng or Thai black ginger, a native plant of Thailand (Fig. 2a). Subsequently, 44 different flavonoid compounds were isolated from D. parviflora (Fig. S2) and tested for their effects on melanin production in B16F10 mouse melanoma cells ( Fig. 2b; Fig. S3a and 3b). Among the top five hits with the strongest effect on melanin generation were MT-8, an O-methylated isoflavone, also called pratensein, and UM-9, a tri-O-methylated isoflavan, also called duartin. Based on the results obtained in the melanin content assay, the following chemical features were identified to be optimal for the isoflavone group (e.g. MT-8): Aryl ring A should be dihydroxy-substituted at positions 5 (-OH) and 7 (-OH), and the aromatic ring B should be methoxysubstituted at position 4′. For the isoflavan group (e.g. UM-9), ring A should be one hydroxy-substituted at position 7 and one methoxy-substituted at position 8, with ring B being dimethoxy-substituted at positions 4′ and 2′ (or 3′). Subsequently, the screening was replicated in human melanoma cells (MNT-1), where UM-9 and MT-8 were again found to be among the top five hits (Fig. 2c). The effects of UM-9 and MT-8 were found to be concentration and time dependent. The most prominent effects were seen at day 3-4 when using a concentration of 10-20 µM in B16F10 cells (Fig. 2d,e; Fig. S3c-f). Similar optimal concentrations were found for MNT-1 cells (Fig. S3g-h).

TPC2-inhibiting flavonoids reduce melanoma cell proliferation, migration, and invasion in a TPC2-dependent manner.
An anti-cancer potential of flavonoids has long been claimed [15][16][17] . Proposed anticancer mechanisms for flavonoids are inhibition of proliferation, inflammation, invasion, metastasis, and activation of apoptosis 15   MITF protein levels are reduced in TPC2 knockout melanoma cells while GSKβ levels are increased. MITF is a major regulator of melanoma proliferation and progression, and mutations in MITF are associated with Tietz albinism-deafness syndrome, Waardenburg syndrome type 2A, and melanoma development 18,19 . Several pathways are involved in the regulation of MITF such as the RAS/RAF/MEK/ERK pathway, the PI3K/AKT, and the Wnt/GSK3β/β-Catenin signalling pathways 20 . Melanin formation is also triggered by melanocyte-stimulating hormone (MSH), a peptide hormone encoded by the proopiomelancortin gene (POMC). MSH binding to MC1R results in the induction of MITF via CREB (cAMP response element-binding protein) 20 . We performed Western blot experiments to assess protein levels of MITF and several key proteins involved in the regulation of MITF expression. We found that in TPC2 −/− MNT-1 cells MITF protein levels were strongly decreased compared to WT cells. We confirmed this result by using three different antibodies against MITF (Fig. 6a,b and Fig. S4). We next assessed the expression levels of CREB, ERK, Akt, and GSK3β ( Fig. 6c-j,  Fig. S4, and Fig. 7). While ERK, Akt, and CREB showed no significant differences, the expression of GSK3β was significantly increased. GSK3β is a negative regulator of MITF expression and can target MITF for proteasomal degradation. Activation of Wnt signalling can prevent this process by increasing the endolysosomal destruction of GSK3β 21 . An increased level of GSK3β suggests reduced degradation of the GSK3β containing destruction complexes in endolysosomes, resulting in increased GSK3β-dependent MITF degradation.
To further corroborate this hypothesis we assessed the MITF protein stability in WT and TPC2 −/− MNT-1 cells by using cycloheximide (CHX). CHX is an inhibitor of protein synthesis and can be used to determine the longevity of proteins 22 . In Western blot experiments we found an increased sensitivity of TPC2 −/− compared www.nature.com/scientificreports/ to WT MNT-1 cells to CHX treatment, resulting in a faster and significantly stronger degradation of MITF in TPC2 −/− MNT-1 cells (Fig. 6k-l and Fig. S4). To demonstrate that the differences in degradation of MITF were dependent on proteasomal activity we used the proteasome inhibitor MG-132. MG-132 was found to reestablish WT MITF levels within 5 h after treatment to TPC2 −/− MNT-1 cells (Fig. 6m-n), indicative of proteasomal degradation causing the reduction in MITF levels.

Discussion
We show here a TPC2-dependent inverse correlation between melanin generation and melanoma proliferation, migration, and invasion by using both genetic and pharmacological approaches. While genetic ablation or pharmacological inhibition increase melanin production in a TPC2-dependent manner, proliferation, migration, and invasion of melanoma cells are reduced. We further found that the flavonoids MT-8 (pratensein) and UM-9 (duartin) efficiently reduce melanoma cell proliferation, migration, and invasion in a TPC2-dependent manner, thus providing a molecular rationale for previously suggested anti-cancer effects of flavonoids [15][16][17] . Besides MT-8 and UM-9, the related compound MT-16 which corresponds to NAR was confirmed as a TPC2 inhibitor. NAR had been demonstrated before to impair VEGF-induced vessel formation 9,10 in a TPC2 dependent manner. Flavonoids thus emerge as anti-cancer drugs acting through the endolysosomal/melanosomal cation channel TPC2. Melanin content and the development of melanoma have previously been suggested to be correlated 12 and it is well established that melanin is one of the major protective factors against UV radiation mediated DNA damage that results in melanoma development. TPC2 now emerges as a critical regulator of both melanin generation and melanoma proliferation, migration, and invasion in human melanoma cells due to its dual functions in melanosomes and endolysosomes.
Through regulation of genes related to invasiveness, migration and metastasis MITF can promote melanoma progression 19 . We found here that knockout of TPC2 results in a strong decrease in MITF protein abundance, suggesting that the reduction in MITF levels is likely causative for the observed reduced effects on proliferation/ migration/invasion in TPC2 −/− MNT-1 cells. Wnt signalling is known to stabilize MITF protein levels in melanoma cells. Ploper et al. (2015) have shown that Wnt signalling can also regulate MITF at the protein degradation level, underscoring the importance of misregulated endolysosomal biogenesis and trafficking in Wnt signalling and cancer 21 . Wnt inhibits GSK3β and promotes sequestration of destruction complexes containing GSK3β into endosomes and MVBs, thus stabilizing MITF protein levels (Fig. 7). In the absence of Wnt and Wnt signalling GSK3β phosphorylates MITF, targeting MITF for proteasomal degradation. In TPC2 −/− cells endolysosomal trafficking and degradation are impacted as shown before for, e.g. LDL, EGF/EGFR, or PDGFR trafficking and degradation [23][24][25] , suggesting a possible impact on the sequestration and degradation of the destruction complex containing GSK3β. Consequently, we found increased GSK3β protein levels and an increased proteasomal degradation of MITF. Of note, flavonoids have been reported before to affect MITF expression through interference with Wnt signalling. Thus, e.g. Syed et al. (2011) showed that human melanoma cell growth inhibition by flavonoids was associated with disruption of Wnt signalling and decreased MITF levels 26 . Such data further support a connection between flavonoids, TPC2, Wnt signalling and MITF expression.
A role of TPC2 and TPC2 variation in melanoma development is also supported by genome-wide association studies. Thus, Kosiniak-Kamysz et al. (2014) examined 33 candidate polymorphisms located in 11 pigmentation genes and the vitamin D receptor (VDR) gene in a population of 130 cutaneous melanoma patients and 707 healthy controls 27 . In the final multivariate analysis with genetic interactions included and after adjustment for age and skin colour, five epistatic effects remained significant, i.e. interactions between MC1R and TYR, SLC45A2 and VDR, HERC2 and VDR, OCA2 and TPC2. The identified TPC2 variant rs3829241 (P = 0.007) is one (TPC2 G734E ) of the two variants shown to be GOF variants 6 and to be associated with blond hair color in humans 5 .
Taken together, our data provide strong evidence not only for a role of TPC2 in cancer proliferation, migration, and invasion in general but specifically a twofold role for TPC2 in melanoma development by affecting on the one hand MITF protein abundance and on the other hand melanin production independently of MITF through direct interference with tyrosinase activity in melanosomes. Thus, melanoma cell proliferation, migration, and invasion are inversely correlated with TPC2-dependent melanin production as reduction of TPC2 expression increases melanin content but decreases proliferation/migration/invasion. This is possible due to independent mechanisms: via regulation of MITF protein levels through interference with endolysosomal activity of TPC2 and endolysosomal GSK3β degradation on the one hand and on the other hand via regulation of tyrosinase activity in melanosomes which likewise express TPC2 28 . In a very recent study, D' Amore et al. (2020) have also investigated the role of TPC2, but in a model of human amelanotic melanoma: CHL1. In CHL1 cells, TPC2 was surprisingly found to increase the metastatic traits of this amelanotic melanoma cell line by a mechanism involving store-operated calcium entry and the Hippo signalling pathway that negatively regulates YAP/ TAZ activity. Clearly, these differences between amelanotic melanoma cells (CHL1) on the one hand and highly pigmented melanoma cells (MNT-1) but also a range of other cancer cells (e.g., HUH7, T24, 4T1) 7 on the other hand regarding TPC2 need to be further elucidated in future studies.

Melanin screening in B16F10 mouse melanoma cells. Melanin content determination was per-
formed as described previously with some modifications 31 . In brief, B16F10 cells at density of 5 × 10 3 cells/well in 96-well plate were cultured and incubated with various plant extracts or flavonoids at a concentration of 20 µg/ ml or 20 µM, respectively, for 4-5 days. Melanin content was measured using a microplate reader (Anthros, Durham, NC, USA) and calculated based on the OD ratio between treated and untreated cells.
Cell proliferation assay. Cell proliferation assay was performed in 96-well, flat-bottom microtiter plates (Sarstedt), in triplicates, and at a 5 × 10 3 cell density per well. Cells were seeded overnight, including cells measured as day zero control. Proliferation rate was assessed by incubation with CellTiter-Blue (Ctb, Promega, Mannheim, Germany) reagent for 3 h. Fluorescence was measured using a microplate reader at 560Ex/600Em (Tecan, Infinite M200 PRO).
Wound healing/migration assay. Wound healing assay was performed using 12-well plates (Sarstedt) at a density of 120,000 cells/well. Cells were incubated overnight, and a scratch was performed using a yellow pipet tip. Pictures were taken at 0, 24, 48, and 72 h with an inverted microscope (Leica DM IL LED) and using a www.nature.com/scientificreports/ formed. The lower compartment contained the chemotactic gradient, medium with 10% FBS. Cells were allowed to migrate for 24 h, and were then fixed and stained with crystal violet containing methanol. Non-invaded cells were removed with Q-tips and pictures were taken of the bottom side of the membrane using an inverted microscope (Olympus CKX41) and an Olympus SC50 camera (Olympus). The number of invaded cells was quantified using ImageJ 1.52a software.

Western blotting.
Western blot experiments were performed as described previously 32 . Briefly, cells were washed twice with 1 × PBS and pellets were collected. Total cell lysates were obtained by solubilizing in TRIS HCl 10 mM pH 8.0 and 0.2% SDS supplemented with protease and phosphatase inhibitors (Sigma). Protein concentrations were quantified via Bradford assay. Proteins were separated via a 10% sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE; BioRad) and transferred to polyvinylidene difluoride (PVDF; BioRad) membranes. Membranes were blocked with 5% bovine serum albumin (Sigma) or milk diluted in Tris Buffered Saline supplemented with 0.5% Tween-20 (TBS-T) for 1 h at room temperature (RT), then incubated with primary antibody at 4 °C overnight. Then, membranes were washed with TBS-T and incubated with horseradish peroxidase (HRP) conjugated anti-mouse or anti-rabbit secondary antibody (Cell Signaling Technology) at RT for 1 h. Membranes were then washed and developed by incubation with Immobilon Crescendo Western HRP substrate (Merck) and by using an Odyssey imaging system (LI-COR Biosciences). Quantification was carried out using unsaturated images on ImageJ 1.52a software. The blots were cropped prior to hybridisation with antibodies against vinculin, GAPDH, or actin. www.nature.com/scientificreports/ and COSY spectra. Inverse-detected heteronuclear correlations were measured using HMQC (optimized for 1 J C-H = 145 Hz) and HMBC (optimized for 3 J C-H = 8 Hz) pulse sequences with a pulsed field gradient. FABMS spectra were obtained on a JEOL JMS-700 using a m-nitrobenzyl alcohol matrix. Optical rotation was measured on a JASCO DIP-360 digital polarimeter. Column chromatography (CC) was performed with powdered silica gel (Kieselgel 60, 230-400 mesh, Merck KGaA, Darmstadt, Germany) and styrene-divinylbenzene (Diaion HP-20, 250-800 µm particle size, Mitsubishi Chemical Co., Ltd.). Precoated glass plates of silica gel (Kieselgel 60, F254, Merck Co., Ltd., Japan) and RP-18 (F254S, Merck KGaA) were used for TLC analysis. The TLC spots were visualized under UV light at a wavelength of 254 nm and sprayed with dilute H 2 SO 4 , followed by heating. HPLC separation was mainly performed with a JASCO model 887-PU pump, and isolates were detected by an 875-UV variable-wavelength detector. Reversed-phase columns for preparative separations (Develosil Lop ODS column, 10-20 µm, 5 × 50 × 2 cm; Nomura Chemical Co. Ltd., Aichi, Japan; flow rate 45 mL/min with detection at 205 nm) and semi-preparative separations (Capcell Pak ODS, 5 µm, 2 × 25 cm, Shiseido Fine Chemiacls Co. Ltd, Tokyo, Japan; flow rate 9 mL/min with detection at 205 nm) were used.