ADT-OH exhibits anti-metastatic activity on malignant melanoma through inhibition of FAK/Paxillin signaling pathways

Background: Melanoma is a highly aggressive cancer, and its high metastasis results in a high lethality. Hydrogen sulde (H 2 S) is now widely recognized as the third endogenous gas delivery substance and may play a key role in cancer biological processes. The present study was designed to evaluate the anti-metastatic effect of H 2 S-donor ADT-OH on melanoma cells and the underlying mechanism. Methods: Firstly, the effect of ADT-OH on the migration of melanoma cells was explored in vivo by the mouse footpad injection model and the mouse tail vein metastasis model. Tumor xenograft growth and tumor tissue H&E analyses were also measured in vivo. Then, the migration inhibitory effect and the underlying mechanism of ADT-OH on B16F10, B16F1 and A375 melanoma cell lines were evaluated by wound healing, transwell, western blot and immunouorescence analyses. Results: Our data showed that ADT-OH inhibited the migration and invasion of melanoma cells signicantly in vivo in three different animal models. Further research showed that ADT-OH signicantly suppressed the migratory, invasive and adhesive properties of A375 and B16F10 cells as measured by the wound-healing and transwell assays. Mechanistically, ADT-OH treatment suppressed the EMT processes and reduced the enzymatic activity of FAK and Paxillin. Moreover, abnormal CSE/CBS and AKT signaling pathways in A375 and B16F10 cells were notably observed following ADT-OH treatment. Additionally, ADT-OH at higher concentration signicantly inhibited the proliferation of highly metastatic melanoma A375 and B16F10 cells. Conclusions: Our results suggest that ADT-OH exerts anti-metastasis activity in melanoma cells by suppressing the EMT process through the CSE/CBS and FAK signaling pathways, and it might be used as an agent against metastatic melanoma in future. proteomics analysis. by western blot with indicated antibodies. b The mRNA levels of FAK, E-cadherin, N-cadherin and Vimentin in B16F10 under different treatment conditions were examined by RT-qPCR. c After transfected with pRK5-FAK or pRK5 plasmid, B16F10 cells were exposed to NaHS (2 mM) or ADT-OH (12 μM) for 24 h. Immunouorescence staining for FAK and F-actin was performed and representative images are shown. d Transwell assay. B16F10 cells were transfected with pRK5-FAK or pRK5 plasmid and then treated with NaHS (2 mM) or ADT-OH (12 μM). After 24 h, the cells that migrated to the inferior membrane were stained and counted in 5 elds at a magnication of ×100. n = 5, bar = 100 μm. The experiments were carried out in triplicate and representative data are shown. *p < 0.05, **p < 0.01.

cytometer with CellQuest software (BD Biosciences, CA, USA). The level of intracellular reactive oxygen species was determined based on the uorescence intensity of the FL1 channel.
Luciferase reporter assay B16F10, B16F1 and A375 cells were cotransfected with pGL3-basic, FAK-promoter luciferase (pGL3-FAK) and control pRL-SV40 reporter for 24 h. Luciferase activities were measured consecutively by using Dual-Luciferase assays (Promega, USA). All measurements were normalized for Renilla luciferase activity to correct the variations in transfection e ciencies.
Cell proliferation assay CCK8 assay was used to measure cell proliferation. Cells in the exponential growth phase were seeded into a 96-well plate at a density of 5000 cells per well. The cells were incubated with ADT-OH (0.8-100 μM) for 24 h before adding 10 μl of CCK8 (Sigma, Milan, Italy), and cell viability was measured by using a microplate spectrophotometer (Titertek Mul-tiskan MCC / 340) equipped with a 450 nm lter. Each experiment was performed in quadruplicate and repeated at least three times.

Cell migration assay
The cell migration assay was performed using transwell inserts (8.0 mm pore size, Millipore, Billerica, MA, USA). The cells were starved for 12 h prior to the experiment. Cells were then harvested and resuspended in a cell suspension diluted to 5 × 10 5 cells / mL with serum-free DMEM containing 1% BSA. 200 μL of cell suspension were pipetted into the upper chamber, and the lower chamber was lled with 600 μL of 10% FBS supplemented medium. After incubation at 37°C for 10 h or 16 h, cells on the upper surface of the membrane were removed. The migrant cells attached to the lower surface were xed in 4% paraformaldehyde at room temperature for 30 min, and stained for 30 min with a solution containing 1% crystal violet and 2% ethanol in 100 mM borate buffer (pH 9.0). The number of cells migrating to the lower surface of the membrane was photographed in ve elds under a microscope with a magni cation of × 100. The chamber was then purged with 33% HAC (100 μL). After the crystal violet was completely dissolved and the cells were evenly distributed in the HAC solution, the assay was performed at 570 nm using a microplate reader (TECNA, Switzerland) and quantitative analysis was performed using GraphPad Prism 8.0 software.

Wound-healing assay
Cells were plated in 6-well culture plates to form cell monolayer (near 90% con uence). A sterile P-200 micropipette was used to scrape off the cells to make wounds. Then the wells were washed three times with PBS to remove non-adherent cells. The progress of wound closure was monitored with a ×10 microphotographs taken by the light microscope (Carl Zeiss Axioplan 2) at the beginning and the end of the experiments after washing with PBS. Image J software was used to calculate the scratch area of the melanoma cells treated with or without ADT-OH at 0 h and 24 h, and then the migration rates were calculated using the following formula Migration rate = (S1-S2)/S1, where S1 represents the scratch area at 0 h; S2 represents the scratch area at 24 h.

Colony-forming assay
Colony-forming assay was performed as previously described (14,15). Brie y, about 300 cells in log phase were plated into 60 mm tissue culture Petri-dish (Greiner) in triplicate with 3 mL of culture medium and grown at 37°C with 5% CO 2 . 48 h later, cells were rinsed with fresh medium, and ADT-OH was added at different concentrations (0, 3.2, 6.3, 12.5, 25, 50 μM). After incubation for 48 h, the cells were washed twice with PBS and then incubated in drug-free medium. The medium was changed every 5 days. After culturing for additional 10-14 days, the medium was discarded and each dish was washed twice with PBS carefully. The cells were xed with methanol for 15 min and stained with a 1:10 dilution of Giemsa regent (Merck, Germany) for 10 min. Any grouping of cells containing 30 or more cells was counted as a colony. Colony numbers were determined from triplicate plates. Colony growth was related to the control value without any treatment.
Immuno uorescence assay (IFA) The immuno uorescence assay was performed as previously described (16). B16F10, B16F1 and A375 cells were treated with ADT-OH at different concentrations. After 48 h, cells were harvested and immunostained with anti-E-cadherin and vimentin antibody, respectively. The cells were then incubated with Alexa Fluor 488-labeled (#A21202, Thermo Fisher Scienti c Inc.) and Alexa Fluor 594-labeled (#A21207, Thermo Fisher Scienti c Inc.) secondary antibody for 1 h at room temperature. After the cell nucleus was counterstained with DAPI, images were obtained using the ZEN 2009 Light Edition software (Carl Zeiss) through an inverted Zeiss LSM710 confocal microscope (40x lens) (Carl Zeiss).
Development of mouse tail vein injection model B16F10 and A375 cells (2×10 6 ) suspended in 200 μL of PBS were injected into the tail vein of the C57BL/6 mice or nude mice. Beginning on the second day, the mice were administered with vehicle, 17.5 mg/kg ADT-OH, 37.5 mg/kg ADT-OH, 75 mg/kg ADT-OH, 1.4 mg/kg NaHS, 2.8 mg/kg NaHS and 5.6 mg/kg NaHS ve times a week for a total of 3 weeks with at least 8 mice per group. The mice were then repeatedly imaged for metastatic tumor spreading to distant organs. At the end of study, mice were euthanized and the lungs and livers were harvested, xed in 10% formalin, and para n embedded for pathological examination of H&E slides.
In vivo footpad injection model B16F10 cells (2×10 5 cells in 50 μl PBS) were implanted into the right hind footpads of C57BL/6 mice using a Hamilton syringe and 25-gauge needle. The detailed treatment was described in Fig. 1b. All mice were euthanized 21 days after tumor inoculation in order to assess the number of spontaneous metastases in the lungs. The following formula was used to measure and calculate the volume of the tumor in the sole of the foot: 0.5236×L1×(L2) 2 , where L1 is the long axis of the tumor and L2 is the short axis of the tumor.

H&E assays
In a parallel animal assay (totally three groups, and three mice per group), the tumor establishment and drug treatment are the same as described above. On the 21 th day, mice were euthanized. Tumors were collected, xed with 4% formaldehyde and then embedded in para n. Tissue sections (5 μm in thickness) were prepared according to standard protocols for hematoxylin/eosin (H&E) staining. Apoptotic cells in tumor sections (two sections per mouse, three mice in total) were visualized by the TUNEL technique according to the manufacturer's instruction (Merck).

ADT-OH inhibits melanoma invasion in vivo
H 2 S is produced in mammalian cells by three major enzymes, including CSE, CBS, and 3-MST. Numerous studies have shown that expression of these enzymes has changed in various types of cancer cell processes, especially CSE and CBS (3,(17)(18)(19). According to Oncomine, an online cancer transcriptome database, CSE and CBS were observed to be overexpressed in human skin cutaneous melanoma (SKCM). In addition, Kaplan-Meier survival analysis of SKCM patients showed that cases with higher expression of CSE or CBS exhibited poorer overall survival (Fig. 1a). Furthermore, we found that the overall survival rate of patients exhibiting high CSE and CBS expression with lung adenocarcinoma, acute myeloid leukemia, bladder urothelial carcinoma, and adrenocortical carcinoma is also signi cantly reduced based on The Cancer Genome Atlas (TCGA) datasets (Fig. 1a, Supplementary Fig. S1 and Supplementary Fig. S2). In our study, after treatment with ADT-OH, the protein level of CBS and CSE in melanoma cells was found to be signi cantly down-regulated ( Supplementary Fig. S3). Taken together, these results discover that H 2 S plays a vital role in the molecular biological characteristics of cancer, which may be a potential drug target for treatment of poor prognosis. Therefore, the palm transfer model was implemented. As is shown in Fig. 1b and c, intraplantar injection of mouse melanoma cells (5 × 10 5 cells in 20 μl) into C57BL/6 mice led to time-dependent tumor growth in a hind paw, showing a 3 to 4-fold increase in paw volume 3 weeks after melanoma cell implantation. After treatment with ADT-OH, the paw volume of melanoma-bearing mice was reduced from 252.128 mm 3 to 105.2557 mm 3 . Moreover, H&E staining analysis showed that ADT-OH treatment signi cantly reduced the melanoma nodules in the lungs of mice, especially when treated with a dose of 37.5 mg/kg (Fig. 1d). Furthermore, mice passive lung metastasis model was also established to investigate the effects of ADT-OH on melanoma cancer metastasis in vivo (Fig. 2a). Lung metastasis nodules were found in the lungs and counted macroscopically (Fig. 2b). Compared with control group, the lung metastasis was restrained in ADT-OH-and NaHS-treated groups, especially in the ADT-OH (37.5 mg/kg)-treated group. Likewise, H&E staining results in Fig. 2c showed that ADT-OH and NaHS treatment could signi cantly decrease lung metastases of B16F10 cells and reduce lung nodule volume. We also performed the same experiment on female BALB/c nude mice using A375 cells. As shown in Fig. 2d and e, lung metastases of A375 cells and lung nodule volumes were markedly decreased after ADT-OH treatment. Taken together, our study suggests that ADT-OH could inhibit the migration and invasion of melanoma cells signi cantly in vivo.

ADT-OH suppresses melanoma cell migration and invasion in vitro
Migration is an essential step in tumor cell metastasis process. Since effective therapeutics against metastatic melanoma are absent in the clinical practice, we assessed the anti-migration effect of ADT-OH on the metastatic melanoma cells A375 and B16F10. The wound-healing assay showed that A375, B16F10 and B16F1 cells displayed high migrated capabilities as indicated by being able to completely heal the wound scratch in the absence of ADT-OH (Fig. 3). However, ADT-OH signi cantly inhibited the activity of migration of melanoma A375, B16F10 and B16F1 cells in a dose-dependent manner.

ADT-OH inhibits melanoma cell invasion in vitro
In addition, tumor cell invasion is a vital step in the cancer metastasis process. Cell invasion was usually examined by Matrigel-coated transwell chambers assay. To investigate whether ADT-OH can suppress the invasion of A375, B16F10 and B16F1 cells, we performed an invasion assay following ADT-OH treatment. In the absence of ADT-OH (control group), A375, B16F10 and B16F1 cells showed highly invasion capabilities by being able to completely penetrate the Matrigel-coated lters ( Fig. 4a-c). However, the invasion capability of A375, B16F10 and B16F1 cells were effectively suppressed by ADT-OH in a dosedependent manner (Fig. 4a-c). To further visualize the effect of ADT-OH on the actin cytoskeletons of melanoma cells, a phalloidin immuno uorescence staining assay was performed. Elevated ADT-OH concentration obviously inhibited lopodia production and spear elongation, especially at a concentration of 100 μM ( Fig. 4d-f). Taken together, these data indicate that ADT-OH might reduce the migration and invasion capability of melanoma cells.

ADT-OH inhibits cell migration by regulating FAK signaling pathway
To further explore the speci c role of ADT-OH on cell migration and which signaling pathway it might be involved in, we performed high throughput proteomic approach to compare protein expressions between ADT-OH-treated and -untreated cell lines. There are 2217 differential proteins in all groups were screened by a t-test (Additional le 3). The MetaCore TM pathway mapping tool clustered the differential networks from DEPs results, among which the cell adhesion and migration pathway show a high score ( Fig. 5a and  b). Moreover, as is shown in Supplementary Fig. S4, the intracellular protein changes under the action of 2 μM and 50 μM ADT-OH are quite different, of which only 165 proteins have the same changes. Next, further analysis of the process network of the two groups found that the 2 μM group is more related to tumor metastasis. As shown in Supplementary Fig. S4C, the fourth cell pathway under 2 μM ADT-OH treatment is cell adhesion. It is well known that the FAK signaling pathway plays a crucial role in the metastasis and invasion of a variety of tumor cells (20)(21)(22). Here, we found that FAK is involved in multiple signaling pathways including cell migration following ADT-OH treatment ( Fig. 5c and d). As we can see in Fig. 5d, ADT-OH might directly affect cell migration through the PTEN/FAK/Paxillin pathway. Therefore, the changes in protein and mRNA levels in this signaling pathway after ADT-OH treatment were next detected by Western blotting and qPCR analyses. As shown in Fig. 6a and b, both total FAK and its phosphorylation level were greatly decreased in response to ADT-OH treatment in melanoma cells. After ADT-OH treatment, the protein level of Paxillin, the downstream response protein of FAK, was also signi cantly reduced. In addition, tumor cells with EMT phenotype changes are often involved in the loss of epithelial properties and acquisition of mesenchymal characteristics, exhibiting enhanced motility, and invasive abilities. To determine whether the anti-metastatic activity of ADT-OH also involves the regulation of EMT, typical markers of EMT including E-cadherin, N-cadherin and Vimentin were next detected using Western blot assay. The results showed that ADT-OH reversed EMT changes in A375 and B16F10 cells, causing the re-induction of E-cadherin and the re-inhibition of N-cadherin and Vimentin expression as compared to the control in a dose-dependent manner, indicating that ADT-OH has inhibitory effects on melanoma cell EMT ( Fig. 6a-b, Supplementary Fig. S5 and Supplementary Fig. S6). Furthermore, the change pattern in mRNA levels of FAK, Paxillin, E-cadherin and N-cadherin was similar to that in their protein levels ( Supplementary Fig. S7). Immuno uorescence assay was also performed to show that ADT-OH-treated melanoma cells exhibited suppressed epithelial markers, E-cadherin, and overexpressed mesenchymal markers, vimentin, in comparison with the control (Fig. 6c, d).
ADT-OH affected the protein level of FAK mainly through decreasing the stability of FAK.
As is shown in Fig. 7a, ADT-OH signi cantly reduces the FAK mRNA levels in B16F10, B16F1 and A375 cells. To explore the mechanism of ADT-OH on FAK, we constructed the FAK luciferase reporter gene plasmid and performed experiments in B16F10 melanoma cells. The detection of dual luciferase reporter gene showed that ADT-OH had no effect on the transcriptional activity of FAK promoter (Fig. 7b). However, mRNA stability experiments showed that ADT-OH signi cantly reduced the stability of FAK mRNA (Fig. 7c). Therefore, we speculate that ADT-OH might inhibit cell migration by reducing the level of FAK mRNA and then reducing its protein expression level.
Then, we further determined whether FAK overexpression could reverse the inhibitory effect of ADT-OH on melanoma migration. B16F10 cells were used for transfection of FAK-overexpression vector (Fig. 8). We found that FAK mRNA and protein expression levels increased signi cantly in the transfected cells (Fig.  8a, b). Compared with negatively transfected cells, overexpression of FAK signi cantly promoted the protein and mRNA expression of N-Cadherin and Vimentin. In contrast to the cells that were not treated with ADT-OH or NaHS, the expression levels of Vimentin and N-Cadherin signi cantly increased and FAK, p-AKT, and AKT signi cantly decreased in B16F10 cells after ADT-OH or NaHS treatment, along with the microvilli and pseudopodia; however, these effects and phenotype were signi cantly abolished after overexpression of FAK (Fig. 8a-c). As shown in Fig. 8c, the level of F-actin in melanoma cells was decreased, the cell structure was disorganized, and the cell morphology became round after treatment with ADT-OH in control cells. However, cells overexpressing FAK showed increased levels of F-actin protein, and the cells elongated and grew fusiform, and the barbed ends of the cells were more prominent. After NaHS or ADT-OH treatment, this phenotype of melanoma cells did not change obviously. These results indicate that, by reducing FAK expression, ADT-OH signi cantly inhibited the EMT process of melanoma, while this ADT-OH-induced effect was signi cantly reversed after FAK overexpression. The migrational potency of cells is inhibited once the EMT process is suppressed in melanoma cells. The results of transwell experiment further showed that the migration ability of B16F10 cells was signi cantly inhibited under ADT-OH treatment, but it was restored after overexpression of FAK (Fig. 8d).

ADT-OH inhibits the viability of A375 and B16F10 cells in vitro
To determine whether ADT-OH exhibits cytotoxicity, we next analyzed the inhibitory effects of ADT-OH on the proliferation of A375, B16F10, and MEF cells by CCK-8 assay. As shown in Supplementary Fig. S7, ADT-OH inhibited the viability of A375 cells and B16F10 cells in both a time-and concentration-dependent manner, but had a slight effect on MEF. After pretreatment with ADT-OH for 24 h, ADT-OH exhibited an IC 50 value of 11.67 μM against A375 cells and an IC 50 value of 5.653 μM against B16F10 cells, respectively, while the IC 50 value against MEF cells was 32.37 μM. Furthermore, we found that 0-6.4 μM of Fig. S8). Besides, CCK8 assay showed that ADT-OH could also signi cantly inhibit the proliferation of a variety of tumor cells in addition to melanoma (Supplementary Fig. S9). Next, the longterm cell viability of ADT-OH-treated melanoma cells and control cells were tested by colony formation assay ( Supplementary Fig. S10). The results showed that ADT-OH signi cantly inhibited the colony formation capacity of melanoma cells. Moreover, we studied whether ADT-OH had any effect on B16F10, B16F1 and A375 cell cycle. As shown in Supplementary Fig. S10, ADT-OH arrested the tumor cells in the G2/M phase, while the effect of NaHS treatment on cell cycle was not signi cant. Analysis of apoptosis further con rmed that ADT-OH promoted apoptosis of B16F10, B16F1 and A375 cells ( Supplementary  Fig. S11). Additionally, ADT-OH also increased the content of ROS in melanoma cells to further promote tumor cell death (Supplementary Fig. S12). These results indicate that higher concentrations of ADT-OH can not only inhibit the migration of melanoma cells, but also inhibit the cell activity by inhibiting the melanoma cycle and proliferation, thereby achieving a more effective treatment of melanoma.

Discussion
Melanoma is a kind of highly aggressive cancer and the metastasis accounts for the majority of patient deaths. Many cancer patients exhibit metastasis by the time of diagnosis. Cancer metastasis is a complicated process that involves multiple sequential and interlinked steps including detachment, migration, invasion, and adhesion (23). Until now, effective drugs against melanoma metastasis are absent in clinical practice. Nowadays, various H 2 S donors including ADT-OH have been con rmed to inhibit proliferation and development of melanoma. There is an increasing number of research studies focused on elucidating the molecular mechanisms of H 2 S's anti-cancer effects, and several critical signaling pathways have been reported to be in uenced by H 2 S donors' treatment. The multiple functional effects of H 2 S administration in cancer cells involve the induction of apoptosis, the inhibition of proliferation and the prevention of metastasis and tumor angiogenesis (10,11,(24)(25)(26). But unfortunately, none of the H 2 S donors could be approved for the treatment of melanoma metastasis.
Therefore, H 2 S donors gaining insights into the mechanisms of inhibition melanoma metastasis will provide new therapeutics for metastasis of melanoma. In addition, the slow-releasing H 2 S donor 5-(4hydroxyphenyl)-3H-1,2-dithiocyclopentene-3-thione (ADT-OH) is known as potent therapeutics with chemopreventive and cytoprotective properties due to its 3H-1,2-dithiole-3-thione group (dithiolethiones), which is also one of the most widely used moiety for synthesizing slow-releasing organic hydrogen sul de donors (24,27). Previously, we have found that ADT-OH is a potential anti-tumor agent by enhancing FADD-dependent extrinsic apoptosis of melanoma cells (16). Here, we further broaden the potential anti-tumor effect of ADT-OH by demonstrating its ability in inhibiting melanoma metastasis both in vitro and in vivo.
A375 is a human melanoma cell line while B16F10 is a mouse melanoma cell line. In the current study, these two cell lines were selected to investigate whether ADT-OH's anti-metastasis effect have species selectivity. We rstly found that ADT-OH signi cantly inhibited melanoma cell migration in vivo by the palm transfer model and mice passive lung metastasis models. Similarly, ADT-OH signi cantly suppressed migration, invasion, and adhesion in vitro. Further research found that ADT-OH inhibited the proliferation of melanoma cells A375 and B16F10, but exhibited little toxicity to normal cell MEF. Interestingly, proteomics data showed that low concentrations of ADT-OH are more related to tumor migration, while high concentrations of ADT-OH can more affect cell apoptosis and survival. Moreover, next results showed that ADT-OH at low concentration (such as 6.3 μM) can signi cantly inhibit the metastasis of melanoma cells, but cannot cause melanoma cell apoptosis, indicating that the inhibition of tumor cell migration caused by ADT-OH at lower concentrations is not achieved by promoting tumor cell apoptosis.
EMT is a critical cellular phenomenon regulating tumor metastasis, which is characterized by tumor cells losing typical epithelial traits including cell polarity and cell-cell adhesion, and the acquisition of the mesenchymal characteristics (28). Intriguingly, EMT is involved in cancer-related mortality by implicating critical processes, including cell migration and invasion. Currently, many studies have con rmed that EMT exerts a determinant role in the progression and metastasis of tumors (28)(29)(30)(31). In this study, we observed that ADT-OH could obviously increase levels of the epithelial marker (e.g., E-cadherin), and decrease levels of mesenchymal markers (e.g., N-cadherin and Vimentin) in both A375 and B16F10 cells. These ndings suggest that ADT-OH negatively regulates EMT, and signi cantly attenuates cell migration, invasion, and adhesion in vitro. Recently, lots of investigations have revealed that H 2 S suppressed the EMT process induced by TGF-β1 in A549 cells and HK-2 cells (26,32). The anti-metastasis effect of ADT-OH on melanoma cells in our study are thus consistent with these reported works.
FA kinase (FAK) is a nonreceptor tyrosine kinase that participates in FA complex formation. Its dysregulation is found in various types of cancer in relation to tumor metastasis (33)(34)(35)(36). Paxillin, which is a structural protein of the FA complex, also contributes to metastasis (37). FAK is a key regulator of integrin-mediated adhesion, which can be auto-phosphorylated at Tyr-397 upon recruitment of this kinase to FA sites following the binding of the transmembrane integrin receptor to ECM. An activated FAK provides both signal transduction and scaffolding functions. Paxillin is an important FA-associated cytoskeletal adaptor protein that provides a docking site for FAK. In turn, FAK can phosphorylate paxillin at Tyr-118 to regulate its function. Phosphorylation of both proteins is required for FA formation, cell motility, and metastasis. Previous researches have demonstrated that H 2 S and H 2 S donors covalently react with functional proteins to in uence several signaling pathways, including the EGF, NF-κB, MAPK, PI3K/Akt and integrin pathways (32,38,39). However, the role of ADT-OH in melanoma cell adhesion as well as the involvement of FAK or paxillin in this biological process remains largely unknown. Therefore, targeted inhibition for FAK signaling pathways has been considered as promising strategies for the treatment of melanoma metastasis. In this study, ADT-OH treatment decreased the phosphorylation of FAK in a dose-dependent manner in both A375 and B16F10 cells. In contrast, overexpression of FAK could reverse the effects and restore the phenotype exerted by ADT-OH in melanoma cells, suggesting a critical role of FAK signaling pathway in ADT-OH-induced inhibitory effect on cell migration.
Furthermore, ADT-OH also decreased the expression of CSE and CBS, which further elucidated the mechanism of ADT-OH inhibiting melanoma migration. We thus speculated that the inhibition of ADT-OH on melanoma metastasis may, in part, be attributed to its ability in inhibiting the CSE/CBS and FAK/Paxillin pathways. To the best of our knowledge, this is the rst time to demonstrate the antimetastasis mechanism of ADT-OH on melanoma cells.

Conclusion
Taken together, our results imply that ADT-OH exhibits a potent anti-metastasis effect in vitro and in vivo through inhibition of CSE/CBS and FAK/Paxillin signaling pathways, leading to the suppression of EMT (Supplementary Fig. S13). Besides, our ndings may provide a new development of therapies for the inhibition of melanoma metastasis. Moreover, ADT-OH shows reduced cytotoxicity in normal human cells (e.g., MEF and 2B cells). Therefore, novel ADT-OH derivatives should be developed to improve its e cacy and stability through a structure-activity relationship study. Further studies are needed to determine the full mechanism and potential of ADT-OH in anti-metastasis of melanoma. High throughput screening analyses, such as a microarray experiment, can also be conducted to assess the potential biomarkers that are affected by the treatment of ADT-OH. Availability of data and material All data generated or analyzed during this study are included in this published article.

Competing interests
The authors declare no con icts of interest.  Figure S12. ROS production. The ROS concentration was determined in B16F10 and A375 cells following treatment with ADT-OH (12.5 μM, 25 μM and 50 μM) for 24 hours. Data are represented as mean ± SD for different experiments performed in duplicate. *P < 0.05, **P < 0.01, ***P < 0.005 compared with the vehicle group. Supplementary Figure S13. A proposed signaling pathway by which ADT-OH inhibits melanoma metastasis.
Additional le 3: The raw data for differential proteomics analysis. Figure 1 ADT-OH inhibits melanoma invasion in vivo in the palm transfer model. a The level of the CBS and CSE genes are associated with the prognosis of patients with different cancers (bladder urothelial carcinoma, lung adenocarcinoma and skin cutaneous melanoma). Kaplan-Meier curves for recurrence-free survival were created using a Kaplan-Meier plotter (www.kmplot.com), in which bladder urothelial carcinoma, lung adenocarcinoma and skin cutaneous melanoma patients were classi ed according to high and low CBS/CSE gene expression. The hazard ratio (with 95% con dence interval) and the log rank p value were calculated. b Animal treatment methods. c Representative imaging of Footpad model mice and the tumor volume at the footpad of each group at the end of the experiment. Data are represented as mean ± SD. *p < 0.05, **p < 0.01. d Representative H&E tissue staining of lungs in footpad model mice.

Figure 2
Page 20/26 ADT-OH inhibits melanoma invasion in vivo in mice tail vein injection model. a Animal treatment methods. b Representative lung imaging of mice after tail vein injection with B16F10 cells for 3 weeks. c H&E stained lung sections. The representative images showed lung tumor distribution of the lung metastasis model mice treated with vehicle, 17.5 mg/kg ADT-OH, 37.5 mg/kg ADT-OH, 75 mg/kg ADT-OH, 1.4 mg/kg NaHS, 2.8 mg/kg NaHS and 5.6 mg/kg NaHS. Scale bars, 100 μm. d-e Representative lung imaging (d) and H&E tissue staining (e) of mice injected with A375 cells in the tail vein. The lung metastasis model mice were treated with vehicle, 37.5 mg/kg ADT-OH, 75 mg/kg ADT-OH and 2.8 mg/kg NaHS for 3 weeks. Scale bars, 100 μm. Figure 3 ADT-OH suppresses melanoma cells migration in vitro. a-c Melanoma cell line B16F10, B16F1 and A375 cell line were seeded into 6-well plates at 2 × 105 cells / well. After incubation with ADT-OH at 0 (Control), 6.3, 12.5, 100 μM for 24 h, the effect of ADT-OH on cell migration was measured by wound-healing assay; original magni cation 40 ×. d-f The migration rates of B16F10, B16F1 and A375 cells were calculated by the formula shown in Materials and Methods. Data are presented as mean ± SD of three independent experiments; *p < 0.05, **p < 0.01, ***p < 0.005 compared with vehicle group.

Figure 4
ADT-OH inhibits melanoma cells invasion in vitro. a-c Summary of results from the transwell invasion assay. After 12 h of incubation with or without ADT-OH, the cells that migrated to the lower chamber were xed, stained, and counted using a light microscope. Four random elds per lter were scanned for the presence of cells on the lower side of the membrane. The panels of the upper gures show the images obtained by cell invasion assay in B16F10, B16F1 and A375 cells. All experiments were performed thrice in triplicate. The data are presented as the mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.005 compared with vehicle group. d-f Representative images of F-actin immuno uorescence in B16F10, B16F1 and A375 cells. Scale bars correspond to 20 μm. Figure 5 ADT-OH regulates the protein level of FAK mainly through decreasing the stability of FAK. a The FAK mRNA levels of B16F10, B16F1 and A375 cells were detected by qPCR analysis after ADT-OH treatment for 6h respectively. b Detection of FAK promoter transcriptional activity after ADT-OH treatment with dual luciferase in B16F10, B16F1 and A375 cells. c B16F10, B16F1 and A375 cells were treated with or without Act D 0.05 μg/mL and ADT-OH (10 μM) for 12h, then the mRNA levels of FAK were detected. *P < 0.05, **P < 0.01. Data are expressed as mean ± SD of three independent experiments. Figure 8