Unconventional myosin VIIA promotes melanoma progression

Melanoma is characterized by its increasing incidence, aggressive clinical behavior and propensity for lethal metastasis (Jemal et al., 2013; Marsh Durban et al., 2013). However, the mechanisms underlying these properties remain elusive. Unconventional myosins are a superfamily of actin-based motor proteins and have vital roles in numerous cellular processes, including intracellular transport, organization of F-actin, mitotic spindle regulation and gene transcription (Berg et al., 2001; Foth et al., 2006; Richards and Cavalier-Smith, 2005; Woolner and Bement, 2009). Myosins consist of three distinct regions, a head, neck and tail. The heads are motor domains that are homologous among different myosins, whereas the tails display diversity and have combinations of a variety of functional domains, which determine the individual cellular functions (Hirano et al., 2011).

Myosin VIIA (Myo7a) is an unconventional myosin closely associated with inherited hearing loss. Previous studies have suggested that Myo7a is mainly expressed in the stereocilia of inner ear hair cells, and is essential for maintenance of the staircase-like bundle and the resting tension (Frolenkov et al., 2004; Kros et al., 2002; Rzadzinska et al., 2009). Mutations in MYO7A cause syndromic (USH1B) (Weil et al., 1995) and non-syndromic (DFNB2 and DFNA11) (Liu et al., 1997a,b; Tamagawa et al., 2002; Weil et al., 1997) deafness in humans. Mutations in Myo7a are also responsible for recessive deafness in shaker-1 mice (Gibson et al., 1995; Mburu et al., 1997). In addition, Myo7a is expressed in the photoreceptor and retinal pigment epithelial cells of the retina, participating in opsin transport (Liu et al., 1999), melanosome transport along the apical processes (Futter, 2006; Futter et al., 2004; Lopes et al., 2007) and the phagocytosis of retinal pigment epithelial cells to the rod outer segments shed from photoreceptor cells (Soni et al., 2005). Most of studies have been focused on the roles of Myo7a in auditory and visual processes, whereas little is known about its functions in other systems, including cancer.

Structurally, the tail of Myo7a contains two protein 4.1, ezrin, radixin, moesin (FERM) domains. Many FERM-containing proteins play important roles in tumor development. Some of these inhibit tumor motility, such as FRMD5 (Wang et al., 2012), while others promote tumor progression such as ERM proteins (Gautreau et al., 2002). Our previous studies have demonstrated that the Myo10 FERM domain interacts with an NPXY motif that is located within the β-integrin cytoplasmic domain, and silencing of Myo10 through siRNA impaired integrin function in cell adhesion (Zhang et al., 2004). Another study has reported that mutant p53-associated Myo10 upregulation promotes breast cancer invasion and metastasis (Arjonen et al., 2014). As Myo7a contains FERM domains, we propose that Myo7a may play an important role in tumor progression. Notably, it has been reported that MYO7A variant S1666C (rs2276288) is associated with increased risk of malignant melanoma in a case-control study, suggesting that Myo7a might play important roles in melanoma development (Fernandez et al., 2009). Importantly, our previous study demonstrated that Myo7a interacts with integrin β5 and selectively promotes integrin αvβ5-mediated cell adhesion and migration (Liu et al., 2014). These findings imply that the biological functions of Myo7a are related to tumor progression. However, few reports have identified the biological functions of Myo7a, or the relationship between Myo7a and tumor progression. In the present study, we demonstrated that Myo7a promotes melanoma cell growth, cell migration and tumor metastasis. We further investigated the underlying molecular mechanisms of the role of Myo7a in melanoma progression.

To explore the biological functions of Myo7a in tumors, we had previously detected the expression of Myo7a in various tumor cell lines, including B16, MCF-7, MDA-MB-231, H1299, SW480 and HeLa cells, and our results showed that B16 is the only cell line that endogenously expresses high levels of Myo7a (see supplementary data Fig. 1B in Liu et al., 2014). Therefore, B16 cells can be used for the loss of function study of Myo7a. Next, we assessed the expression of Myo7a in seven human melanoma cell lines; the results revealed that Myo7a was weakly expressed in A375 and WM983a cells, whereas it was undetectable in other cells (Fig. 1A). Moreover, Myo7a mRNA levels were measured and the results also showed high expression of Myo7a in B16 cells and weak expression in A375 and WM983a cells (Fig. 1B). Two Myo7a-specific siRNAs were synthesized, and these significantly inhibited the protein expression of endogenous Myo7a in B16 cells (Fig. 1C). To explore the functional role of Myo7a in melanoma cells, a B16 cell line with stable knockdown of Myo7a (Fig. 1D) and an A375 cell line stably overexpressing Myo7a (Fig. 1E) were established.

First, we examined the effects of Myo7a on proliferation of melanoma cells. The results showed that knockdown of Myo7a significantly inhibited the proliferation of B16 cells in vitro (Fig. 2A). By the fifth day, the absorbance of Myo7a shRNA-transfected B16 cells was nearly half of that in control cells (Fig. 2B). Next, a soft agar colony formation assay, the results of which are considered to closely relate to the level of in vivo tumorigenicity, was performed to assess the anchorage-independent growth of B16 cells. Compared to control cells, fewer colonies were observed in Myo7a knockdown B16 cells (Fig. 2C,D), which suggests that knockdown of Myo7a strongly inhibited the colony formation of B16 cells. To further identify the roles of Myo7a in melanoma cell proliferation in vivo, xenograft models were established to determine the contribution of Myo7a to tumor growth. In this experiment, six C57BL/6 mice were inoculated with B16 Myo7a shRNA and B16 control shRNA cells at different sides on the back, and the tumors were visualized after 2 weeks. Consistent with in vitro studies, Myo7a knockdown B16 cells grew more slowly than control cells in vivo (Fig. 2E,F), and the tumor mass was also significantly lower in mice inoculated with Myo7a knockdown B16 cells (Fig. 2G). On the other hand, Myo7a-overexpressing A375 cells had a greater tumorigenic ability than control cells (Fig. 2H,I), and tumor masses exhibited the same tendency (Fig. 2J). Taken together, these results suggest that Myo7a promotes tumor growth both in vitro and in vivo.

Myo7a is a motor protein that has been shown to regulate cellular movement, and, in our previous study, we found that Myo7a promoted integrin αvβ5-mediated cell migration (Liu et al., 2014). Thus we sought to further confirm the effects of Myo7a on melanoma cell mobility. To verify this, confluent and quiescent monolayers of B16 cells transfected with Myo7a shRNA and control shRNA were scratched and cell mobility was observed at different time points (0, 6 and 16 h). The results showed that knockdown of Myo7a diminished the rate of wound closure in B16 cells (Fig. 3A), indicating that Myo7a promoted lateral cell migration. In addition, we observed random cell motility of B16 cells, as monitored by time-lapse confocal microscopy. Cell tracking of ten B16 cells selected randomly is presented in Fig. 3B. Cell tracking of another 20 B16 cells in each group is shown in Fig. S1. Similarly, the results showed that knockdown of Myo7a in B16 cells greatly reduced both cell migration velocity and total cell migration distance (Fig. 3C,D). Moreover, Transwell migration assays were performed to uncover the effects of Myo7a on vertical migration of tumor cells. The results revealed that knockdown of Myo7a suppressed the vertical migration of B16 tumor cells (Fig. 3E), while overexpression of Myo7a greatly promoted the migration of A375 cells (Fig. 3G). Moreover, in vitro invasion assays were performed to investigate the effects of Myo7a on tumor cell invasion. As shown in Fig. 3F, Myo7a knockdown cells displayed a significantly decreased ability to invade through a Matrigel-coated chamber, in comparison to control cells. Consistent with the above results, overexpression of Myo7a obviously promoted invasion by A375 cells (Fig. 3H). Taken together, these results indicate that Myo7a promotes melanoma cell migration and invasion in vitro.

To further validate the above results, we investigated the effects of Myo7a on melanoma metastasis in vivo through tail vein injection of B6 cells with stable knockdown of Myo7a. At 14 days after injection, mice were killed, and then several organs including lungs and kidneys were removed and fixed. Representative pictures of lungs from each group are shown in Fig. 4A. A drastic reduction in melanoma nodules was found in the lungs of the Myo7a knockdown group, as compared to what was found for the control group. Consistently, the weight of the lungs from treated mice was significantly lower than in controls (Fig. 4B). Furthermore, there were significantly fewer metastatic foci in lung tissue treated with Myo7a shRNA than in the control groups and these foci were also smaller (Fig. 4C). H&E staining also revealed a high density of tumor nodules that had invaded most of the lung parenchyma in the control group, but many fewer tumor nodules in the Myo7a knockdown group (Fig. 4D). Furthermore, kidney metastasis was only observed in control group animals, which was confirmed by H&E staining (Fig. 4E,G). Surprisingly, there was femur metastasis in control cells, and bone marrow smears revealed melanoma cells scattered among bone marrow blasts, which were not observed in mice injected with Myo7a shRNA transfected cells (Fig. 4F,G). Moreover, H&E staining of the lung tissue sections also showed broader areas of tumors for mice with Myo7a overexpressing cells (Fig. 4H). These data indicate that Myo7a regulates melanoma cell metastasis to the lung and bone.

To elucidate the molecular mechanisms through which Myo7a is involved in the proliferation and metastasis of melanoma cells, we performed high-throughput RNA sequencing (RNA-seq) to analyze the transcriptomes of B16 cells transfected with Myo7a shRNA versus those of the control group. The list of 1348 differentially expressed genes between the two groups can be found as Table S1, and a heat map showed 140 genes [those with reads per kilobase million (RPKM) values greater than 1 and values of log₂ (fold change) greater than 1.5 or less than −1.5] among them, including genes of interest (Fig. 5A). Some of these genes encode proteins associated with cell cycle and tumor metastasis, such as p21 (also called Cdkn1a), Cdk2 and MMP9. Next, real-time quantitative PCR (RT-qPCR) was used to validate the RNA-seq results. Indeed, knockdown of Myo7a increased the expression of p21 and inhibited the expression of Cdk2 at the transcriptional level, and the expression of MMP9 was also significantly diminished in Myo7a knockdown cells (Fig. 5B), an observation further supported by western blots (Fig. 5C). We also found that overexpression of Myo7a downregulated p21 expression in A375 cells (Fig. S2). It is known that p21, a cyclin-dependent kinase inhibitor, plays a critical role in the regulation of the G1 to S phase transition during the cell cycle. To determine whether knockdown of Myo7a inhibited cell proliferation of melanoma cells through blocking the G1-S transition, a flow cytometry cell cycle analysis was performed. The results showed that knockdown of Myo7a caused a significant increase in the G1 population and a decrease in the S population in B16 cells (Fig. 5D,E). In contrast, Myo7a overexpression led to a decrease of the G1 population and an increase in the S population in A375 cells (Fig. 5F; Fig. S3). To determine whether p21 is required for the Myo7a knockdown-mediated cell cycle inhibition, p21 was knocked down by means of siRNA in B16 Myo7a shRNA cells and in control B16 cells (Fig. 5G). As expected, p21 knockdown partially rescued the cell cycle arrest induced by Myo7a knockdown, whereas no obvious changes were observed in control cells (Fig. 5H). Representative FACS results are shown in Fig. S3. Collectively, these data indicate that knockdown of Myo7a induced G1 arrest by upregulating the expression of p21 and inhibiting the expression of its downstream protein Cdk2.

To explore the molecular mechanisms of Myo7a-regulated tumor cell motility, two-dimensional gel electrophoresis following mass spectrometry was used to isolate proteins differentially expressed between B16 Myo7a shRNA cells and control cells. Five protein spots were found to be differentially expressed between the two groups (Fig. 6A; Fig. S4). Among the five protein spots, we found that Rho GDP-dissociation inhibitor 2 (RhoGDI2; also known as ARHGDIB), which promotes the activation of Rho GTPases essential for cell motility, was downregulated in B16 Myo7a knockdown cells (Fig. 6A). We then validated the expression of RhoGDI2 protein by means of western blotting (Fig. 6B). Our results also showed that overexpression of Myo7a upregulated RhoGDI2 expression in A375 cells (Fig. S2). Moreover, knockdown of Myo7a led to reduction of the mRNA level of RhoGDI2, which is consistent with the data from RNA-seq (Fig. 6C). Knockdown of RhoGDI2 efficiently reduced cell migration ability in B16 cells (Fig. 6D,E). To determine whether RhoGDI2 is required for Myo7a-induced cell motility, a Flag-tagged RhoGDI2 expression vector or empty vector was transfected into B16 Myo7a shRNA cells and into control cells, respectively (Fig. 6F). Overexpressed RhoGDI2 promoted cell migration and rescued the inhibition of cell motility caused by knockdown of Myo7a (Fig. 6G). Next, we tested the activities of the endogenous Rho GTPases in B16 cells by performing a GST pulldown assay. The results showed that knockdown of Myo7a resulted in a significant decrease in Rac1 activity (Fig. 6H). However, there was no obvious difference in the activity of Cdc42 between the control cells and Myo7a shRNA-transfected cells. On the other hand, Rac1 activity was significantly elevated when Flag-tagged RhoGDI2 was transfected into B16 Myo7a shRNA cells and into control cells. In summary, the above results suggest that Myo7a promotes the invasiveness of melanoma cells partially by regulating RhoGDI2 expression and the activity of Rac1.

Given that Myo7a regulates melanoma progression in cells and mouse xenografts, as shown above, we speculated whether Myo7a expression is increased in human melanoma. To this end, we examined Myo7a levels by immunohistochemical staining in both normal and tumor tissues of melanoma patients (n=15). We found that Myo7a expression was greatly increased in the melanoma tissues of 11 patients compared to normal tissues. Results from three patients are shown in Fig. 7A. Other results are provided in Fig. S5. Furthermore, upregulated expression levels of Myo7a in melanoma were confirmed also seen in the Oncomine data bank (https://www.oncomine.org/) (Fig. 7B). These results further indicate that Myo7a plays an important role in the regulation of melanoma progression.

Myo7a was initially identified as the gene defective in shaker-1 mice and in human USH1B (Gibson et al., 1995; Weil et al., 1995). Although Myo7a has been extensively studied since then, our knowledge of mammalian Myo7a is still limited to its functions in auditory and visual processes. No studies have characterized the effects of Myo7a in tumor progression. In this study, we demonstrated for the first time that Myo7a plays important roles in regulating the cell proliferation, motility, invasion and metastasis of melanoma cells. Importantly, we found that Myo7a expression is significantly upregulated in human melanomas.

Previous studies have shown that Myo7a is expressed in cells with cilia or microvilli, such as hair cells, retinal pigment epithelium (RPE) cells, kidney tubules, lung bronchi and in Sertoli cells of the testis (Wolfrum et al., 1998). The expression of Myo7a in normal skin melanocytes has not been reported, and Myo7a was undetectable in rat skin (Hasson et al., 1995). In our study, we found that Myo7a was highly expressed in mouse B16 melanoma cells, which were derived from a skin melanoma in C57BL/6 mice.

In the present investigation, we identified that Myo7a modulated melanoma cell motility. This conclusion was supported by several lines of evidence. First, depletion of Myo7a in B16 cells inhibited cell migration and invasion in vitro. Second, overexpression of Myo7a in A375 cells promoted tumor cell migration and invasion in vitro. Third, depletion of Myo7a in B16 cells suppressed tumor metastasis in an experimental model of metastasis and overexpression of Myo7a in A375 cells had the opposite effect. Since invasion and metastasis are important characteristics of malignant tumors, these findings indicate that Myo7a plays essential roles in melanoma development.

Mechanistically, we found that Myo7a regulated tumor motility by modulating the expression of RhoGDI2. RhoGDI2 inhibits the dissociation of GDP and further regulates GTPase function, and Rho GTPases are well-established regulators of cell migration and are implicated in the process of tumor cell invasion and metastasis (Raftopoulou and Hall, 2004; Rathinam et al., 2011; Ridley, 2001). RhoGDI2 expression has been reported in a series of tumor types. Intriguingly, RhoGDI2 plays bipartite roles in tumor progression (Griner and Theodorescu, 2012). RhoGDI2 acts as a tumor suppressor in bladder cancer and suppresses experimental lung metastasis (Gildea et al., 2002). However, RhoGDI2 expression is positively correlated with tumor progression and metastasis in gastric tumors (Cho et al., 2009). RhoGDI2 also promotes tumor growth, invasion and metastasis in ovarian carcinoma (Stevens et al., 2011) and in breast cancer (Gu et al., 2008; Zhang et al., 2009; Zhang and Zhang, 2006). It has been reported that RhoGDI2 suppresses anchorage-independent growth in melanoma B16 cells (Wang et al., 2011). In our study, we demonstrated that depletion of Myo7a inhibited melanoma cell motility by downregulating expression of RhoGDI2. Previous studies have suggested that RhoGDI2 inhibits tumor metastasis in bladder cancer but does not affect tumor cell growth or colony formation in vitro, or tumorigenesis in vivo (Gildea et al., 2002). Although RhoGDI2 mediates the metastatic processes of melanoma, understanding the underlying mechanisms requires further investigation. Furthermore, we examined the activation status of the Rho family of small GTPases, Rac1 and Cdc42. We found that knockdown of Myo7a resulted in significant decrease in Rac1 activation while not affecting the activation of Cdc42 (data not shown).

In our study, knockdown of Myo7a strikingly diminished cell proliferation in vitro and tumor growth in vivo, and vice versa. Gene expression profiling via RNA-seq revealed changes in gene expression of certain cell cycle-related genes, including p21 and Cdk2. It is known that p21, a Cdk inhibitor, inhibits Cdk2 and plays critical roles in the regulation of the G1 to S phase transition during the cell cycle (Waldman et al., 1995). Taken together, our findings suggest that Myo7a inhibits p21, which raises the level of Cdk2 and thereby promotes tumor growth, and that Myo7a also upregulates RhoGDI2 to promote tumor cell migration and metastasis (Fig. 7C). However, further investigation is required to pinpoint the mechanisms of Myo7a-regulated p21 expression.

In conclusion, we identified that Myo7a is able to function in non-specialized cells by demonstrating that Myo7a regulates melanoma proliferation, migration, invasion and metastasis via control of p21 and RhoGDI2 expression, which expands our current knowledge concerning Myo7a. Enhanced expression of Myo7a in human melanoma suggests an important role of Myo7a in the regulation of melanoma progression. Thus, Myo7a may be a candidate therapeutic target in human melanoma.

Female BALB/c nude mice and C57BL/6 mice that were 4–6 weeks old were purchased from the Animal Department of Peking University Health Science Center with approval of the Animal Care and Use Committee of Peking University Health Science Center (Permit Number: LA2011-73). Cell lines used in the present investigation were purchased from Cell Lines Collection Center of China (Beijing, PRC). Human melanoma cells A375 and M21, human breast cancer cells MCF-7 and MDA-MB-231, human lung cancer cells H1299, human cervix cancer cells HeLa and human colon cancer cells SW480 were cultured in Dulbecco's modified Eagle's medium (DMEM; Invitrogen, Carlsbad, CA) and mouse melanoma cells B16, human melanoma cells Mel57, WM983a, WM266-4 and WM1205 were cultured in RPMI 1640 (Invitrogen, Carlsbad, CA). Cell culture media were supplemented with 10% fetal bovine serum (FBS) (Invitrogen, Carlsbad, CA), 100 IU/ml penicillin and 100 μg/ml streptomycin. The cells were cultured at 37°C with 5% CO₂ in a humidified incubator.

Cells were plated in six-well plates and transfected with Myo7a siRNA, p21 siRNA, RhoGDI2 siRNA or control siRNA (QIAGEN, Hilden, Germany) at a final concentration of 100 nM using Lipofectamine RNAiMAX (Invitrogen, Carlsbad, CA). Sequences of RNAi oligonucleotides were as follows: Myo7a siRNA1, 5′-CAGAGTCATTCTCCTCCAGAA-3′; Myo7a siRNA2, 5′-CCAGGTGTTCTTCATGAAGAA-3′; p21 siRNA1, 5′-AGACCAGCCUGACAGAUUU-3′; p21 siRNA2, 5′-AACGGUGGAACUUUGACUUCG-3′; RhoGDI2 siRNA, 5′-GGATGACGAGAGTCTAACCAA-3′; control siRNA, 5′-CGAGUGGUCUAGUUGAGAA-3′. Myo7a siRNA2 with higher knockdown efficiency was chosen to generate Myo7a shRNA. One pair of oligonucleotides was synthesized by Invitrogen. The forward oligonucleotide sequence was 5′-GATCCCCCCAGGTGTTCTTCATGAAGAATTCAAGAGATTCTTCATGAAGAACACCTGGTTTTTA-3′ and the reverse oligonucleotide sequence was 5′-AGCTTAAAAACCAGGTGTTCTTCATGAAGAATCTCTTGAATTCTTCATGAAGAACACCTGGGGG-3′. The oligonucleotides were annealed and inserted into pSUPER vector to produce Myo7a shRNA, and the recombinant vector was verified by DNA sequencing.

B16 cells were transfected with control shRNA or Myo7a shRNA, and then were selected by 800 μg/ml G418 (Gibco, CA) for 3–4 weeks to establish stable clones. A375 cells were transfected with Flag or Flag-tagged Myo7a, which were also selected by 800 μg/ml G418 for 3–4 weeks to establish stable clones.

Cells lysates were prepared using PBSTDS lysis buffer containing cocktail protease inhibitor (Boehringer Mannheim, Mannheim, Germany). Proteins were denatured at 95°C for 5 min and then separated by SDS-PAGE. Separated proteins were transferred onto PVDF membrane (Millipore, Billerica, MA) following blocking with 5% milk in Tris-buffered saline (TBS) containing 1% Tween-20 for 1 h and incubated with primary antibodies overnight at 4°C. Primary antibodies used were: rabbit anti-Myo7a ab3481, Abcam, 1:1000; mouse anti-RhoGDI2, sc-376473, Santa Cruz Biotechnology, 1:200; mouse anti-p21, sc-6246, Santa Cruz Biotechnology, 1:500; rabbit anti-Rac1, #2465, Cell Signaling Technology, 1:1000; mouse anti-Cdc42, sc-8401, Santa Cruz Biotechnology, 1:500 and mouse anti-β-actin, sc-47778, Santa Cruz Biotechnology, 1:1000. All horseradish peroxidase (HRP)-coupled secondary antibodies (JIR) were used at 1:5000 dilution. The signal was detected using a chemiluminescence detection kit (Pierce, Rockford, IL).

Stable clones were seeded in 96-well plates at a density of 2000 cells per well and cultured for 5 days. 100 μl of 10% WST-1 was added to each well after removal of the growth medium. Then the cells were incubated for 1 h at 37°C, and the absorbance was measured at a wavelength of 450 nm on a microplate reader (Bio-Rad). Wells without cells were used as blank control and samples were analyzed in triplicate.

The colony formation assay in soft agar was carried out in a six-well plate. First, the basal layer was made by mixing 1.2% agarose (Invitrogen) and equivalent volume of 2× medium with 20% FBS. Then the stable cells were harvested and suspended in medium containing 0.3% agarose and plated over the basal layer in triplicate at a density of 5000 cells per well. After 2–3 weeks, cells were stained with 0.1% Crystal Violet when colonies were visible by the naked eye. The ability of colony formation was calculated by determining the total number of colonies per well.

The stable B16 Myo7a shRNA cells and the controls were trypsinized and washed twice with PBS and suspended in serum-free RPMI 1640. For implantations, six C57BL/6 mice were injected subcutaneously with 10⁵ cells at both sides of the back, with one side being injected with control cells and the other with Myo7a knockdown cells. In gain-of-function experiments, by contrast, 10⁶ A375 Flag–Myo7a and control cells were injected subcutaneously into the two sides on the back of the nude mice. Once tumors were observed, the tumor volumes were measured every other day and calculated using the formula: (length)×(width)²×(π/6). About 2 weeks later (4 weeks for nude mice), mice were killed when the biggest tumor reached 1 cm in diameter and then tumors were removed.

At 48 h post transfection with siRNA, cells were serum-starved overnight. Then a confluent cell monolayer was scratched by a sterile 200 μl pipette tip and the wounded monolayer was washed twice to remove non-adherent cells. Images were photographed at 0 h after wounding and at regular intervals during cell migration to close the wound. For stable clones, cells were seeded into six-well plates and the wound healing assay was performed in the same way.

Stable cells were resuspended in medium containing 1% FBS at 10⁶ cells/ml, and 100 μl of cell suspension was placed in the upper chamber (pore size of 8 μm, Costar, Corning, NY) [for invasion assays, 2 mg/ml Matrigel (BD Bioscience) was pre-coated]. Medium supplemented with 20% FBS was added to the bottom wells of the chambers. Samples were prepared in triplicate. After incubation for 5 h (for invasion assay, 48 h incubation) at 37°C, the cells on the upper side of the filters were gently removed using cotton swabs. Filters were fixed in 4% paraformaldehyde and stained with 0.1% Crystal Violet. Twelve microscopic fields were randomly chosen for analyses.

The random motility of B16 cells was monitored by using time-lapse confocal microscopy (Carl Zeiss LSM780, Germany) as previously described (An et al., 2010). Briefly, B16 stably transfected with Myo7a shRNA and control shRNA were seeded in 60 mm dishes incubated at 37°C for 6 h and transferred to the microscope for imaging. Cells were maintained during imaging in an incubator with humidified air (95%) and 5% CO₂ at 37°C. A total of 15 observation fields were randomly selected with two cells in each field, and time-lapse imaging was performed for 17 h using a 10× dry objective with an interval of 30 min. The obtained images were converted to movie files by using ZEN software. Cell tracking was performed with ImageJ software, and the graph was created using GraphPad Prism 5.

Twelve C57BL/6 mice were randomly divided into two groups. Approximately 10⁵ B16 Myo7a shRNA cells or control cells suspended in 100 μl serum-free medium were injected via the tail vein, respectively. After 14 days, mice in each group were killed, the lungs, kidneys and femurs were removed and photographed, and histochemical analysis was performed after fixation in Bouin's solution overnight.

The BALB/c nude mice were randomly divided into two groups consisting of six mice each. The mice were injected intravenously with 2×10⁶ A375 cells overexpressing Myo7a or control vector, which were suspended in 100 μl of serum-free medium, through tail vein. After 30 days, the mice were killed, and the lungs were resected and histochemically analyzed.

Total RNA was extracted from the cells by using Trizol reagent (Invitrogen). Approximately 2 μg of RNA was used for the reverse transcription reaction with MMLV Reverse Transcriptase (Promega, CA). Real-time qPCR was performed by using SYBR Green mix (Applied Biosystems, CA) with the PCR conditions: 95°C for 3 min; 95°C for 20 s, 60°C for 1 min, for 40 cycles. The following primers were used for PCR: mouse Myo7a forward primer, 5′-ATCCTCCTGCCTCATGTTCAG-3′ and reverse primer, 5′-CGGGGAAGTAGACCTTGTGGA-3′; human Myo7a forward primer, 5′-GCAGCACTCGTGGATTGAG-3′ and reverse primer, 5′-TGTACTTTCCGAAACGGCTTG-3′; mouse MMP9 forward primer, 5′-CGAACTTCGACACTGACAAGAAGT-3′ and reverse primer, 5′-GCACGCTGGAATGATCTAAGC-3′; mouse RhoGDI2 forward primer, 5′-ATGACGGAGAAGGATGCACAG-3′ and reverse primer, 5′-CTCCCAGCAGTGTTTTCTTGTA-3′; mouse p21 forward primer, 5′-CCTGGTGATGTCCGACCTG-3′ and reverse primer, 5′-CCATGAGCGCATCGCAATC-3′; mouse Cdk2 forward primer, 5′-CCTGCTTATCAATGCAGAGGG-3′ and reverse primer, 5′-TGCGGGTCACCATTTCAGC-3′; mouse GAPDH forward primer, 5′-GGCAAAGTGGAGATTGTTGC-3′ and reverse primer, 5′-AATTTGCCGTGAGTGGAGTC-3′; human β-Actin forward primer, 5′-CCTCGCCTTTGCCGATCCG-3′ and reverse primer, 5′-GCCGGAGCCGTTGTCGACG-3′. Samples were analyzed in triplicate and experiments were repeated three times.

The cell cycle distribution of cells was analyzed by flow cytometry assay following propidium iodide (PI) staining. Briefly, when the cells were ∼80% confluent, they were harvested and fixed in 0.5 ml cold 70% ethanol at −20°C overnight. The cells were then treated with 100 mg/ml of DNase-free RNase (Qiagen) and incubated for 30 min at 37°C. Before detection, PI (50 mg/ml; Sigma-Aldrich) was added directly to the cell suspension. Stained cells were subjected to fluorescence-activated cell sorting (FACS) analysis with a flow cytometer (FACSCalibur, BD Biosciences).

RNA sequencing was performed as described by De Bellis et al. (2014). In brief, total RNA was isolated by using Trizol (Invitrogen) according to the manufacturer's recommendations. Complementary DNA first and DscDNA synthesis was then performed by following supplier's instructions (Invitrogen). Cluster generation and sequencing-by-synthesis (36 bp) were performed by using the Illumina Hiseq2000 sequencing system according to standard protocols (Illumina). The gene expression values were obtained using Genomatix (www.genomatix.de) software. GO and Pathway analysis were performed using Panther (http://www.pantherdb.org).

Stable cells were harvested at 80% confluence and then were washed three times with PBS and homogenized on ice in lysis buffer containing 7 M urea, 2 M thiourea, 4% CHAPS, 65 mM dithiothreitol (DTT), 2 mM EDTA, 0.2% Bio-Lyte and protease inhibitor cocktail. After centrifugation at 17,500 g for 1 h, the supernatant was collected and the protein concentration was determined by the Bradford method. Approximately 200 μg protein was dissolved in rehydration buffer and loaded onto IPG strips (17 cm, pH 3-10, Bio-Rad). Isoelectric focusing was performed at 250 V for 30 min, followed by an increase in voltage to 10,000 V within 5 h, and this was maintained at 10,000 V until 60,000 volt hours were attained. Strips were placed in equilibration buffer (6 M urea, 2% SDS, 20% glycerol, 375 mM Tris-HCl pH 8.8) containing 2% DTT and shaken slowly for 15 min, then transferred to equilibration buffer containing 2.5% iodoacetamide and shaken for an additional 15 min. The strips were then transferred to 12% SDS polyacrylamide gels for separation in the second dimension. Proteins were visualized by silver staining, and the images were scanned and analyzed with PDQuest software, which was obtained from Bio-Rad. The differentially expressed spots were excised and verified by mass spectrometry (MS).

Affinity precipitation with GST–PBD was performed using an assay as previously described by Zhang et al. (2010). In brief, glutathione–Sepharose beads (Amersham Pharmacia Biotech) were used to capture the GST fusion proteins and the interacting proteins. Cells lysates were prepared using PBSTDS lysis buffer (0.01 M PBS, 1% Triton X-100, 5 g/l Sodium deoxycholate, 1 g/l SDS and 1 mM EDTA) containing cocktail protease inhibitor (Boehringer Mannheim, Mannheim, Germany). Equal amounts of protein supernatants were incubated at 4°C with GST–PAK1-CD fusion protein-coated glutathione–Sepharose beads overnight. The beads were washed, and the bound proteins were released by heat denaturing in protein loading buffer. Samples were then separated by 15% SDS-PAGE and detected by immunoblotting with antibodies against Rac1 (#2465, Cell Signaling Technology, 1:1000) and Cdc42 (sc-8401, Santa Cruz Biotechnology, 1:500).

Surgically removed melanoma tissues were collected from 15 patients at Peking University Cancer Hospital and Institute (Beijing, China); adjacent normal tissues used as the control. The experiments were approved by the Ethics Committee of Peking University Cancer Hospital and informed consent was obtained from all subjects.

Immunohistochemical staining for protein expression was performed on tissue sections. Briefly, sections were deparaffinized with xylene, followed by rehydration in ethanol. Sections were incubated overnight at 4°C with rabbit anti-Myo7a antibody (ab3481, Abcam, 1:100), and then incubated for 30 min with secondary antibodies (Dako, Carpinteria, CA). The immunostaining was examined with an Olympus BX51 microscope (Olympus, Tokyo, Japan).

All experiments were repeated three times, and data are presented as means±s.d. Statistical analyses were conducted using Graphpad Prism 5 (https://www.graphpad.com/scientific-software/prism/). A paired Student's t-test was used for comparison of tumor weight between the two groups, while an unpaired t-test was performed for other intergroup comparisons. *P<0.05;**P<0.01; ***P<0.001.