Eribulin Suppresses Clear Cell Sarcoma Growth by Inhibiting Cell Proliferation and Inducing Melanocytic Differentiation Both Directly and Via Vascular Remodeling

Clear cell sarcoma (CCS) is a rare but aggressive soft-tissue sarcoma (STS), appearing predominantly in adolescents and young adults. It usually arises in the lower extremities close to tendons, fasciae, and aponeuroses, although it occasionally occurs in the upper extremities or the trunk (1, 2). CCS is termed “malignant melanoma of soft parts” due to melanocytic differentiation driven by expression of microphthalmia-associated transcription factor (MITF), which upregulates tyrosinase (TYR), MART-1, and HMB45, and in some cases melanin production (1). CCS is characterized cytogenetically by a t(12;22)(q13;q12) chromosomal translocation and its resultant fusion oncogene EWSR1–ATF1 (3). EWSR1–ATF1 fusion oncoprotein constitutively activates the MITF promoter, leading to melanocytic differentiation in CCS (4).

The mainstay CCS treatment is complete surgical resection, although chemotherapy and/or radiotherapy are also used depending on disease status. Despite multidisciplinary treatment, earlier reports found 5-year overall survival (OS) rates of only 30% to 67% (1, 2, 5, 6). Notably, chemotherapy was reported to be largely ineffective in patients at advanced stages, resulting in poor prognosis (7). Therefore, novel therapeutic strategies including effective chemotherapy regimens are needed to improve patient prognosis.

MITF is a basic helix–loop–helix leucine zipper transcription factor that regulates key processes in several cell lineages, including melanocytes, retinal pigment epithelial cells, osteoclasts, mast cells, and melanoma cells (8). In melanoma, MITF controls not only differentiation but also proliferation (8). However, melanomas with extremely high MITF levels exhibit greater susceptibility to cell-cycle arrest and differentiation, reducing tumorigenicity (9, 10). On the other hand, lower MITF activity increases the invasive and metastatic properties of melanoma cells by conferring a stem cell–like phenotype (9–11). In melanoma, MITF activity is regulated at transcriptional and posttranscriptional levels, and these regulatory mechanisms include epigenetic and microenvironmental signals (11, 12). However, the mechanic details of CCS are unclear and so as yet have not been exploited successfully for treatment.

Eribulin mesilate (eribulin), an analog of the natural marine compound halichondrin B, is a nontaxane synthetic microtubule dynamics inhibitor currently used in many countries for advanced or metastatic breast cancer and STS (13, 14). Two phase III clinical trials of patients with metastatic breast cancer reported that eribulin improved OS without corresponding effects on progression-free survival (PFS) compared with treatment of physician's choice or capecitabine (14–16). Similar findings of prolonged OS in the absence of effects on PFS were also reported in a phase III clinical trial of patients with liposarcoma and leiomyosarcoma compared with dacarbazine (13). Thus, eribulin is considered a promising agent for other cancer types.

The anticancer mechanisms of eribulin appear to be unique, although still not fully understood. Eribulin was shown to destabilize microtubules by inhibiting the growth parameters at the plus ends without effect on microtubule shortening parameters, thereby giving rise to irreversible mitotic disruption and apoptosis (17, 18). In addition, vascular remodeling by eribulin treatment, resulting in increased microvessel density (MVD) and enhanced tumor perfusion, has been shown in human patients with breast cancer and mouse xenograft models (19–21). Moreover, accumulating evidence indicates that eribulin inhibits the epithelial–mesenchymal transition of breast cancer cell lines and induces differentiation of STS cell lines in addition to regular cytotoxic effects, which have not been reported for other microtubule-targeting agents such as taxanes and vinca alkaloids (22, 23).

In this study, we first report potent antitumor activity of eribulin against CCS cell lines and two mouse xenograft models. Further, eribulin induces melanocytic differentiation and vascular remodeling in CCS tumors. At the molecular level, eribulin upregulates MITF protein levels by inhibiting ERK1/2. In addition, we show that reoxygenation through vascular remodeling potentially leads to tumor differentiation and decreases cell proliferation. Finally, knockdown of MITF attenuates the cytotoxic effect of eribulin in CCS cells.

We utilized four human CCS cell lines, Hewga-CCS, MP-CCS-SY, KAS, and SU-CCS1, to examine the antiproliferative properties of eribulin. The types of EWSR1–ATF1 chimeric transcripts in these CCS cells were confirmed as described previously (24). The Hewga-CCS line was established in our laboratory from a primary tumor (5), whereas MP-CCS-SY and KAS lines were kindly provided by Dr. Moritake (Miyazaki University, Miyazaki, Japan) and Dr. Nakamura (Japanese Foundation for Cancer Research, Tokyo, Japan), respectively (25, 26), and SU-CCS1 was purchased from the ATCC. These cell lines were passaged soon after receipts, divided and stocked at −80°C. All cells were maintained in DMEM (Nacalai Tesque) supplemented with 100 μg/mL streptomycin sulfate, 100 U/mL penicillin G (Life Technologies), and 10% heat-inactivated FBS (Life Technologies) at 37°C in a humidified atmosphere of 5% CO₂. When the cells reached subconfluence, they were detached using 0.25% trypsin plus EDTA (Life Technologies) and reseeded for experimental treatments. Cell lines were authenticated by examination of morphology, genotyping by PCR and growth characteristics, and were mycoplasma free. Experiments were performed within 2 months of thawing.

Eribulin was manufactured at and provided by Eisai. Co. Ltd. The ERK1/2 inhibitor SCH772984 was purchased from Cayman Chemical. The drugs were prepared in DMSO before addition to cell cultures for in vitro examinations according to the manufacturer's instructions. Eribulin was diluted in physiologic saline to the desired concentration for in vivo experiments.

All CCS lines were seeded into 96-well plates at 2 × 10³ cells/well in triplicate, including three control wells to provide “time zero” for absorbance readings. They were incubated overnight and then treated with the experimental drugs or the vehicle (DMSO) for the indicated time. The cell proliferation rate was assessed using the Premix WST-1 Cell Proliferation Assay System (TAKARA) according to the manufacturer's instructions. Absorbance at 450 and 650 nm (reference wavelength) was measured using a spectrophotometer. The relative cell proliferation rate was determined by subtracting the average value of “time zero” measurements from sample measurements.

All CCS lines were seeded at 1 × 10⁶ per dish and cultured overnight, followed by eribulin or vehicle treatment. After 0 to 24 hours treatment at the indicated concentration, cells were harvested and stained with propidium iodide (PI) solution (25 μg/mL PI, 0.03% NP-40, 0.02 mg/mL RNase A, 0.1% sodium citrate) for 30 minutes at room temperature. For cell-cycle analysis, we used the BD FACSVerse flow cytometer (BD Bioscience) and the BD FACSSuite Software Application (BD Bioscience) according to the manufacturer's protocol.

For lysate preparation, cells were first washed with PBS and lysed in RIPA (Thermo Fisher Scientific) supplemented with 1% protease/phosphatase inhibitor cocktail. Protein concentrations were determined using the bicinchoninic acid (BCA) method (Thermo Fisher Scientific). Cell proteins were separated on 4% to 12% Bis-Tris gels (Life Technologies) and transferred to polyvinylidene difluoride membranes (Nippon Genetics). The membranes were blocked in tris-buffered saline (TBS) containing 5% skim milk and Tween 20 (TBS-T) at room temperature and then incubated with primary antibodies in Can Get Signal Solution 1 (TOYOBO) at 4°C overnight, followed by incubation with secondary antibodies in Can Get Signal Solution 2 (TOYOBO) at room temperature for 1 hour. The primary antibodies are available in Supplementary Table S1.

Five-week-old female BALB/c nu/nu mice (SLC) were housed at the Institute of Experimental Animal Sciences, Osaka University Medical School, in accordance with guidelines approved by the Animal Care and Use Committee of the Osaka University Graduate School of Medicine. For the xenograft tumor growth assay, 1 × 10⁷ CCS cells were injected subcutaneously into the left side of the back. Tumor dimensions were measured from the skin using a caliper and volume calculated as (A × B²)/2, where A is the longest diameter and B is the shortest diameter. When the average tumor diameter reached 5 mm, the mice were randomized into eribulin- and vehicle-treated groups. Eribulin was intravenously injected through the tail vein at 1 or 3 mg/kg on days 1 and 8. Controls received equal-volume saline injections at the same time. Xenograft tumor volume and mouse body weight were measured twice weekly. When all mice were euthanized, the tumor weights were measured. All protocols were approved by the Animal Care and Use Committee of the Osaka University Graduate School of Medicine.

Excised xenograft tumors were fixed in 10% neutral-buffered formalin, embedded in paraffin, cut into 3-μm-thick sections, and examined by IHC, hematoxylin–eosin (HE) staining, and Fontana–Masson staining. For immunostaining, paraffin-embedded sections were first deparaffinized and dehydrated. Antigens were retrieved at 95°C for 10 minutes in a 10 mmol/L citrate buffer. After quenching endogenous peroxidase activity for 10 minutes with methanol containing 3% H₂O₂, the sections were blocked for 1 hour with TBS containing 2% BSA at room temperature. The sections were then incubated with primary antibodies at 4°C overnight, followed by 1 hour incubation with HRP-conjugated secondary antibodies. Finally, label sections were stained with 3,30-diaminobenzidine tetrahydrochloride (Dako) and counterstained using hematoxylin. The primary antibodies are available in Supplementary Table S1. Fontana–Masson staining was performed to identify melanin synthesis as described previously (24).

CCS cell lines were seeded in six-well plates at 2 × 10⁵ cells/well and incubated for 24 hours. The plating medium was then exchanged with fresh medium containing 10 nmol/L eribulin or vehicle. Cells were incubated for an additional 96 hours, washed twice in PBS, detached using 0.25% trypsin plus EDTA, and transferred to 1.5 mL tubes. After centrifugation for 10 minutes at 14,000 rpm, the cell pellets were dissolved in 1 mol/L NaOH (100 μL) for 60 minutes at 80°C. The optical density of the supernatant was measured at 405 nm using a spectrophotometer, and the result was compared with a standard curve of synthetic melanin (Sigma-Aldrich).

CCS cells were cultured in six-well plates at 2 × 10⁵ cells/well. After 24 hours, the plating medium was exchanged for fresh medium including 10 nmol/L eribulin or vehicle. Cells were incubated for an additional 96 hours, washed with PBS, and lysed in RIPA buffer supplemented with 1% protease/phosphatase inhibitor cocktail. Protein concentrations were determined using the BCA method. The lysate was centrifuged, and supernatant samples containing 50 μg total protein were transferred to 96-well plates and mixed with 100 μL samples of 0.2% 3,4-dihydroxy-l-phenylalanine (Sigma-Aldrich) in phosphate buffer. After incubation at 37°C for 1 hour, the absorbance at 475 nm was measured using a microplate reader.

Total RNA was purified using the RNeasy Mini Kit (Qiagen) and reverse transcribed using the High-Capacity cDNA Reverse Transcription Kit (Life Technologies). Real-time PCR was performed using a StepOnePlus Real-Time PCR System (Life Technologies) and SYBR Green Realtime PCR Master Mix (TOYOBO). Expression values were normalized to that of β-actin. The PCR primers used (forward and reverse) were as follows: β-actin (5′-ATTGCCGACAGGATGCAGAA-3′ and 5′-GCTGATCCACATCTGCTGGAA-3′), MITF (5′-GAGGCAGTGGTTTGGGCTT-3′ and 5′-AATTCTGCACCCGGGAATC-3′), EWSR1–ATF1 (type1, type3) (5′-GAGGCATGAGCAGAGGTGG-3′ and 5′-GAAGTCCCTGTACTCCATCTGTG-3′), EWSR1–ATF1 (type2) (5′-CCTACAGCCAAGCTCCAAGTC-3′ and 5′-GCCTGGACTTGCCAACTGTA-3′).

To analyze the degradation kinetics of MITF, CCS cells were seeded in 10 cm dishes at 5 × 10⁵ cells/dish for cycloheximide chase assays. After 24 hours, the medium was exchanged with fresh medium containing 10 nmol/L eribulin or vehicle. Cells were then incubated for an additional 48 or 96 hours, treated with cycloheximide (60 μg/mL) (Wako Pure Chemical Industries) for the indicated times, and harvested. Equal amounts of cell protein were separated by 4% to 12% Bis-Tris gel electrophoresis and MITF expression analyzed by immunoblotting. The integrated optical densities of protein bands were quantified by TotalLab Quant software (Nonlinear Inc.).

Monolayers of CCS cells were scraped (wounded) in a straight line with a 10 μL pipette tip. The detached cells were removed by washing with medium and the remaining CCS cells incubated under 3% oxygen, 5% oxygen, normoxia. Images of each sample culture were obtained at 0 and 24 hours using a microscope and wound width measured as an index of cell motility.

CCS cells were seeded at a density of 3 × 10³ cells/well in 96-well plates and grown overnight. Cells were transfected with 1 nmol/L siRNAs for 24 hours using Lipofectamine RNAiMAX Transfection Reagent (Life Technologies). Two kinds of siRNAs targeting MITF (constructs I and II; #s8791 and #s8972) and a nontargeting negative control siRNA (#4390844) were purchased from Thermo Fisher Scientific.

All data are expressed as mean ± SD. Two-tailed Student t test or one-way ANOVA was used to determine statistical differences. Values of P < 0.05 were considered significant and the specific values are indicated in legends to figures as P < 0.05 (*), P < 0.01 (**).

To assess the potential antiproliferative efficacy of eribulin against CCS, we examined the growth rates of four CCS cell lines following exposure to therapeutic concentrations of eribulin (0–30 nmol/L) for 96 hours by WST-1 assay. Eribulin dose-dependently reduced viable cell numbers of all four lines (Fig. 1A), with the highest potency against the Hewga-CCS line (IC₅₀ = 0.69 nmol/L), followed by KAS (1.91 nmol/L), MP-CCS-SY (2.21 nmol/L), and SU-CCS1 (2.21 nmol/L). Thus, the proliferation of all four lines exhibited high sensitivity to eribulin, in accordance with the effects on other sarcoma cell lines including fibrosarcoma (HT-1080), liposarcoma (SK-UT-1), and leiomyosarcoma (SK-LMS-1; refs. 23, 27). Next, flow cytometry was performed to assess the effects of eribulin on the cell cycle. At 10 nmol/L, eribulin treatment induced G₂-M cell-cycle arrest in all four CCS lines beginning as early as 3 hours after exposure and increasing with time thereafter (Fig. 1B). Moreover, at 100 nmol/L, eribulin treatment induced more potent G₂–M cell-cycle arrest (Fig. 1C). In addition, the cellular level of cleaved caspase-3, the major effector of apoptosis, was enhanced after 72 hours exposure to more than 10 nmol/L eribulin as evidenced by Western blot analyses (Fig. 1D). These experiments suggest that eribulin exerts its cytotoxic effect by blocks of metaphase stage in CCS cells.

We next evaluated the antitumor effects of eribulin against Hewga-CCS and MP-CCS-SY xenograft tumors in nude mice. Mice were injected subcutaneously with each of cells and tumors allowed to grow until the average diameter reached 5 mm. Mice were then treated intravenously with eribulin at 1 or 3 mg/kg or with equal-volume vehicle once every 7 days for two cycles. Treatment with both 1 and 3 mg/kg eribulin markedly suppressed tumor growth compared with the vehicle control in both xenograft models (Fig. 2A and B; Supplementary Fig. S1A). In 3 mg/kg eribulin-treated mice, mean body weight loss was approximately 7.5% and 17.5% at 11 days in Hewga-CCS and MP-CCS-SY xenografts, respectively, and recovered within about 10 days after the second injection (Supplementary Fig. S1B), suggesting moderate nontarget toxicity. Hematoxylin and eosin staining revealed reduced cell density and increased fibrotic area in eribulin-treated tumors compared with vehicle-treated tumors (Supplementary Fig. S1C). IHC analyses also revealed fewer cells immunopositive for the proliferating cell marker PCNA and increased numbers expressing the apoptosis marker cleaved caspase-3 in eribulin-treated tumors compared with controls (Supplementary Fig. S1D and S1E). These results confirm that eribulin suppresses tumor cell proliferation and induces apoptosis in vivo as well as in culture.

Given that eribulin triggers phenotypic alterations in breast cancer, liposarcoma, and leiomyosarcoma cell lines (22, 23), we next examined whether eribulin altered CCS cell phenotype. Specifically, we assessed if eribulin-induced melanocytic features, including expression of proteins involved in melanin production and actual melanin accumulation. Indeed, Fontana–Masson staining revealed the increased number of melanin-positive cells in eribulin-treated xenografts compared with vehicle-treated controls (Fig. 2C), indicating that eribulin induces melanocytic differentiation in CCS cells.

Furthermore, it has been reported that eribulin can remodel tumor vasculature, so tumor sections were stained with anti-CD31 as a vascular endothelial cell marker and/or with anti-α-SMA as a pericyte marker (19, 21). Immunostaining for CD31 revealed that eribulin treatment significantly increased mean vessel density (MVD) in eribulin-treated xenograft tumors and living tumor cells were absent in perivascular areas of 3 mg/kg eribulin-treated tumors (Fig. 2D). Double staining for CD31 and α-SMA was also performed to quantify the extent of pericyte coverage, which is crucial for vessel maturation and stabilization. The vessel maturity index, calculated as the ratio of the α-SMA-positive area to the CD31-positive area, was significantly greater in both eribulin dose groups compared with control (Supplementary Fig. S1F), suggesting that eribulin promotes the normalization of vessel morphology and induces tumor reperfusion. To investigate whether these morphologic changes in tumor vessels cause functional changes, we examined the expression of HIF1α, which contributes to the cellular hypoxia response. Indeed, IHC and Western blot analyses revealed a dramatic decrease in HIF1α expression by eribulin-treated tumors compared with vehicle-treated controls, demonstrating that eribulin promotes reoxygenation (i.e., mitigates intratumoral hypoxia; Fig. 2E and F). These in vivo findings suggest that eribulin induces both melanocytic differentiation and vascular remodeling in CCS xenografts.

To determine whether eribulin directly induces melanocytic differentiation in CCS cells, we assessed the quantity of cellular melanin in cultured CCS cells treated with 10 nmol/L eribulin or vehicle for 96 hours. Melanin synthesis was significantly elevated in all eribulin-treated lines compared with vehicle-treated control cells (Fig. 3A). Because TYR is the rate-limiting enzyme in melanin production, we also measured TYR activity by the conversion of uncolored 3,4-dihydroxy-l-phenylalanine substrate to darkly colored dopaquinone product in CCS cell extracts. Coincident with increased melanin content, treatment with 10 nmol/L eribulin for 96 hours increased TYR activity significantly compared with vehicle-treated control cells (Fig. 3B). These observations suggest that the melanin synthesis pathway is activated by eribulin exposure in CCS cells.

MITF is the most important transcription factor for melanogenesis and is regulated in response to modulation of EWSR1–ATF1 activity in CCS (28). MITF also upregulates the expression levels of TYR, TRP1, and TRP2, which catalyze melanin biosynthesis (8). We thus examined the protein expression levels of EWSR1–ATF1, MITF, TYR, TRP1, and TRP2 by Western blot analysis. Consistent with TYR activity and melanin measurements, 10 nmol/L eribulin time-dependently enhanced the expression levels of MITF, TYR, TRP1, and TRP2, but not of EWSR1–ATF1, in CCS cells (Fig. 3C), indicating induction of melanocytic differentiation by eribulin treatment.

To dissect the mechanisms underlying MITF protein upregulation in eribulin-treated CCS cells, we first investigated the effects of eribulin on MITF gene transcription in CCS cells. Surprisingly, MITF mRNA expression was not induced by eribulin exposure (Fig. 4A), and EWSR1–ATF1 mRNA expression was not altered (Fig. 4B), suggesting that MITF protein upregulation is attributable to neither transcriptional activation nor an interaction between eribulin and EWSR1–ATF1.

On the basis of these results, we speculated that eribulin modulates the posttranscriptional regulation of MITF protein in CCS cells. To test this, we measured MITF protein degradation kinetics by cycloheximide (60 μg/mL) chase assay in the presence of 10 nmol/L eribulin or vehicle. The rate of MITF protein decline in the presence of cycloheximide was significantly reduced in Hewga-CCS cells treated with eribulin for 48 or 96 hours compared with vehicle-treated control cells, suggesting that eribulin prolongs MITF protein half-life (Fig. 4C).

In melanoma, it is well established that downregulation of MITF protein results from ERK1/2-stimulated phosphorylation at Ser 73 and recruitment of p300/CBP, which ultimately leads to ubiquitination and proteasome-mediated degradation (12). Thus, we examined the activation (phosphorylation) status of ERK1/2 in response to eribulin treatment of CCS cells. Intriguingly, ERK1/2 phosphorylation was time-dependently diminished in eribulin-treated CCS cell lines (Fig. 4D), suggesting that ERK1/2 inactivation may reduce MITF protein degradation, thereby enhancing MITF protein levels and activity in eribulin-treated CCS cells. To examine if ERK1/2 inactivation contributes to eribulin-induced melanocytic differentiation, we treated Hewga-CCS and MP-CCS-SY cells with SCH772984, a specific inhibitor of ERK1/2 signaling, and measured the effects on melanocytic differentiation markers and melanin synthesis (29). Consistent with eribulin-induced melanocytic differentiation through ERK1/2 inactivation, SCH772984 alone blocked ERK1/2 activation and time-dependently upregulated the melanocytic differentiation markers, MITF, TYR, TRP1, and TRP2, without any effect on EWSR1–ATF1 expression, in both CCS cell lines (Fig. 4E). Melanin synthesis was also elevated in SCH772984-treated cells compared with untreated cells (Supplementary Fig. S2A). Moreover, SCH772984 inhibited the proliferation of both CCS cell lines with an IC₅₀ of about 60 nmol/L (Supplementary Fig. S2B).

Tumor lysates extracted from Hewga-CCS and MP-CCS-SY xenografts were also examined for expression of ERK1/2 and melanocytic differentiation markers. Consistent with in vitro data, eribulin treatment markedly decreased ERK1/2 phosphorylation and increased MITF, TYR, and TRP2 expression levels with no change in EWSR1–ATF1 (Fig. 4F). These findings support the notion that inhibition of ERK1/2 phosphorylation causes increased MITF expression, at least in part, by direct effect of eribulin on CCS cells, which evokes melanocytic differentiation.

It has been reported that hypoxia in melanoma leads to transcriptional suppression of MITF (11), suggesting that tumor reoxygenation concomitant with eribulin-induced vascular remodeling may mediate MITF protein upregulation and ensuing melanocytic differentiation. To test this notion, Hewga-CCS and MP-CCS-SY cells were grown in 1% oxygen, 3% oxygen, atmospheric oxygen, or atmospheric oxygen with the hypoxia “mimetic” CoCl₂, which suppresses prolyl hydroxylase activity. As predicted, MITF protein expression was reduced in CCS cells under hypoxia (Fig. 5A and B). However, unexpectedly, MITF transcription was not reduced in either cell line by hypoxia (Fig. 5C). Similar results were obtained in both cell lines following CoCl₂ treatment (Supplementary Figs. S3A–S3C). These findings suggest that hypoxia-mediated reduction of MITF expression in CCS cells occurs via a posttranscriptional mechanism.

Activation of ERK1/2 has been observed in several cancer cell lines under hypoxia and is associated with cellular migration, angiogenesis, and enhanced resistance to apoptosis (30–32). Accordingly, we investigated whether ERK1/2 signaling was activated in CCS cells under hypoxic conditions. Indeed, both hypoxic and CoCl₂-treated CCS cells exhibited elevated ERK1/2 phospho-activation (p-ERK1/2 expression; Fig. 5A and B; Supplementary Figs. S3A and S3B). The reciprocal relationship between MITF and ERK1/2 expression levels under hypoxia was further confirmed using SCH772984. As expected, blockade of ERK1/2 signaling reversed the attenuation of MITF protein levels induced by hypoxia and CoCl₂, demonstrating that ERK1/2 signaling suppresses MITF expression in hypoxic CCS cells (Fig. 5A and B; Supplementary Figs. S3A and S3B). Our data suggest that reoxygenation via eribulin-induced vascular remodeling (i.e., reversal of tumor hypoxia) reduces ERK1/2 activity, leading to MITF upregulation and further melanocytic differentiation of CCS cells. Collectively, eribulin facilitates melanocytic differentiation through both direct effects on CCS cells and indirect effects through vascular remodeling, resulting in tumor reperfusion, ERK1/2 inactivation, reduced MITF phosphorylation, greater MITF stability, and upregulation of melanocytic MITF target genes.

To examine if this indirect process influences other aspects of tumor behavior, we examined the effects of hypoxia on CCS cell proliferation and motility, behaviors necessary for tumor growth and invasion/metastasis. Both 3% and 5% oxygen significantly increased Hewga-CCS and MP-CCS-SY cell proliferation (Fig. 5D). Moreover, both cell lines showed increased migration under 3% and 5% oxygen in the wound healing assay (Fig. 5E). These results suggest that reoxygenation affects CCS cell proliferation and motility, both of which may contribute to the therapeutic efficacy of eribulin.

To investigate whether MITF protein expression is involved in the cytotoxic effect of eribulin on CCS cells, MITF was silenced employing siRNA transduction in KAS and MP-CCS-SY cells. Two kinds of MITF siRNAs efficiently silenced MITF in both cells (Fig. 6A). This caused no significant difference in cell proliferation of both cells (Fig. 6B). Importantly, MITF knockdown partly rescued the antiproliferative effect of eribulin in both cells (Fig. 6C). In addition, cleaved caspase-3 expression was decreased compared with the negative controls in both cells by silencing of MITF (Fig. 6D). These findings suggest that MITF upregulation, at least in part, contributes to the antiproliferative effect of eribulin in CCS cells.

Phenotypic transition of cancer cells and alteration of the tumor microenvironment to a state less conducive to aggression are important factors affecting treatment outcome. In this study, we demonstrate that eribulin exerts both direct and indirect anticancer effects on CCS cells, resulting in suppression of cell proliferation, facilitation of tumor differentiation, reduced invasive capacity, and ultimately tumor shrinkage. Eribulin-induced CCS cell differentiation in vitro and in vivo was driven by augmentation of MITF expression mediated through ERK1/2 inactivation. Furthermore, reoxygenation caused by eribulin-induced vascular remodeling inhibited CCS cell proliferation and ERK1/2 signaling, leading to tumor differentiation (Fig. 6E).

Cancer can be regarded as a disease of cell differentiation. Cancer cells exhibit suspended differentiation properties compared with normal cells, and maintaining this undifferentiated state is critical for tumorigenesis (33, 34). Several agents with the ability to induce tumor differentiation, such as retinoic acid and 1α,25-dihydroxyvitamin D₃, have been identified and tested over the last few years, with some already in clinical use (35, 36). Therefore, targeting impaired terminal differentiation could be a promising general therapeutic strategy for malignancy. In CCS, we demonstrate that eribulin-induced upregulation of MITF expression leads to tumor differentiation, and that silencing of MITF modestly attenuates the antiproliferative effect of eribulin on CCS cells. Thus, differentiation status of tumor controlled by MITF expression levels may contribute somewhat to maintaining the sensitivity to eribulin in CCS. In addition, whether MITF as an oncogene is intimately involved in CCS proliferation is not clear in this study, whereas we found that MITF expression was implicated in CCS tumor differentiation and the cytotoxic effect of eribulin, and that eribulin remarkably inhibited CCS growth. These findings suggest that CCS growth is not dependent solely on MITF expression levels, and that other mechanisms, including mitotic and nonmitotic effects, as well as MITF upregulation may be potentially important for eribulin activity in CCS cells. Further investigation into other mechanisms of action for eribulin is required.

Intratumoral hypoxia is an indicator of tumor aggression and poor prognosis (37). Indeed, numerous studies have documented a correlation between intratumoral hypoxia and tumor progression. Eribulin was previously shown to increase tumor vessel density and tumor reoxygenation, resulting in enhanced antitumor activities compared with other cytotoxic agents (19, 20). Our in vivo data also demonstrated that eribulin treatment enhanced MVD and normalized vessel morphology, thus mitigating intratumoral hypoxia. These findings raise the possibility that eribulin can reduce tumor aggression by modifying the tumor microenvironment. Cheli and colleagues proposed that a hypoxic microenvironment promoted melanoma aggression by reducing MITF expression through the HIF1α-controlled transcriptional repressor chondrocytes protein 1 (DEC1), which subsequently binds and suppresses the MITF promoter (11, 38). Contrary to initial expectations, however, MITF gene transcription was not reduced in hypoxic CCS cells, although MITF protein level was downregulated. We then demonstrated that hypoxic CCS cells exhibit increased ERK1/2 phosphorylation (activation) and reduced MITF protein expression, suggesting that ERK1/2 is a post-transcriptional mediator of hypoxia-dependent MITF suppression in CCS cells. These findings indicate that MITF modulation in CCS cells is clearly different from modulation in melanoma cells, supporting the notion that CCS is a distinct disease from melanoma, despite designation as malignant melanoma of soft parts.

The tumor vascular remodeling in response to eribulin is intriguing. Prior preclinical studies suggested eribulin's ability to improve delivery of subsequently administered drugs (19, 20). Eribulin (1.0 mg/kg) synergistically enhanced antitumor effects of the subsequent Doxil (a liposomal anticancer agents) in a non–small cell lung cancer xenograft model compared with vinorelbine (20). Likewise, prior treatment with 1.0 mg/kg eribulin significantly augmented the antitumor activity of capecitabine or paclitaxel in breast cancer xenograft models (19, 20). This could be beneficial in the case of residual tumor remaining after treatment with eribulin. Furthermore, because conventional chemotherapy is largely ineffective in patients with CCS, it would be useful for CCS treatment to utilize eribulin to exert the effect of subsequently administered drugs more intensely.

In this study, the high dose of 3 mg/kg eribulin was employed to evaluate the efficacy of eribulin in CCS xenografts refractory to conventional chemotherapies in addition to 1 mg/kg eribulin. The dose of 1 mg/kg eribulin is well used in preclinical studies and more relevant to clinical practice. In our results, treatment with 1 mg/kg eribulin demonstrated remarkable inhibition of tumor growth, increased differentiation, and vascular remodeling compared with the vehicle control in both xenograft models, consistent with 3 mg/kg eribulin. Thus, Eribulin would be a promising therapeutic agent for patients with CCS.

Collectively, this study demonstrates that eribulin exerts potent anticancer activities on CCS cells both directly and via vascular remodeling. Eribulin gives rise to nonmitotic effects on residual tumor such as alteration of differentiation status and tumor microenvironment, following antimitotic, cytotoxic activity. The complexity of these antineoplastic mechanisms suggests beneficial therapeutic possibilities of combination anticancer regimens including eribulin.

No potential conflicts of interest were disclosed.

Conception and design: S. Nakai, T. Nakai, T. Wakamatsu, T. Tanaka, S. Takenaka, A. Myoui, H. Yoshikawa, N. Naka

Development of methodology: S. Nakai, H. Yoshikawa, N. Naka

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Nakai, N. Yasuda, H. Outani, S. Takenaka, N. Naka

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Nakai, Y. Imura, K. Hamada, N. Naka

Writing, review, and/or revision of the manuscript: S. Nakai, Y. Imura, H. Outani, S. Takenaka, H. Yoshikawa, N. Naka

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Nakai, H. Tamiya, N. Yasuda, T. Wakamatsu, S. Takenaka, N. Naka

Study supervision: S. Nakai, T. Wakamatsu, N. Araki, T. Ueda, H. Yoshikawa, N. Naka

We are grateful to Drs. Hiroshi Moritake and Tohru Sugimoto for kindly providing the human CCS cell line MP-CCS-SY, Dr. Takuro Nakamura for supplying the KAS cell line, and Eisai Co., Ltd. for providing eribulin. We also thank Ryota Chijimatsu, Yukiko Eguchi, and Mari Shinkawa for their technical support and Enago for providing high-quality editing service. This study was supported by grants from the Japan Society for the Promotion of Science, JSPS KAKENHI [Grant Nos.: JP16H05448 (to N. Naka), JP16K20050 (to S. Takenaka), JP18K16639 (to H. Tamiya) and JP18K16679 (to T. Tanaka)], the Osaka Medical Research Foundation for Intractable Diseases (to S. Nakai), the Japan Orthopaedics and Traumatology Research Foundation, Inc. 372 (to Y. Imura) and 378 (to H. Tamiya), and the Practical Research for Innovative Cancer Control from Japan Agency for Medical Research and Development, AMED (to N. Naka).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.