Activation of ERK1/2 Causes Pazopanib Resistance via Downregulation of DUSP6 in Synovial Sarcoma Cells

Synovial sarcoma (SS) is a rare high-grade malignant mesenchymal tumour with a relatively poor prognosis despite intensive multimodal therapy. Although pazopanib, a multi-kinase inhibitor, is often used for advanced SS, most cases eventually become resistant to pazopanib. In the present study, we investigated the mechanisms of acquired pazopanib resistance in SS. To examine acquired pazopanib resistance, two SS cell lines, SYO-1 and HS-SY-II, were isolated after multiple selection steps with increasing concentrations of pazopanib. SYO-1 was also used in vivo. Then, pazopanib-resistant clones were investigated to assess potential mechanisms of acquired pazopanib resistance. Stable pazopanib-resistant clones were established and exhibited enhanced cell cycle progression, cell growth with increased ERK1/2 phosphorylation, and higher sensitivity than parental cells to a MEK-inhibitor, trametinib, both in vitro and in vivo. Furthermore, addition of low-dose trametinib partially reversed the pazopanib resistance. In the pazopanib-resistant clones, dual specificity phosphatase 6 (DUSP6) was downregulated. Inhibition of DUSP6 expression in parental HS-SY-II cells partially recapitulated acquired pazopanib resistance. Acquired pazopanib resistance in SS was associated with activation of ERK1/2 through downregulation of DUSP6 expression. Simultaneous treatment with pazopanib and a MEK inhibitor could be a promising strategy to overcome pazopanib resistance in SS.

phase, regardless of pazopanib, in comparison with parental clones (Fig 1b and c). Then, we examined pazopanib resistance in a mouse model using SYO-1 clones, since of the two cell lines only SYO-1 is tumourigenic. The pazopanib-resistant xenograft mice treated with 30 mg/kg pazopanib showed no significant decrease in tumour burden, while the parental xenograft model showed a significant decrease (Fig. 1d). These results suggest that acquired pazopanib resistance in SS cells causes promotion of cell cycle progression and that resistance is preserved in a preclinical mice xenograft model. Pazopanib-resistant clones showed enhanced cell growth and activated phosphorylation of ERK1/2. We next performed a cell proliferation assay, since a previous study reported that accelerated cell cycle transitions were associated with deregulated cell proliferation 11 , and our cell cycle analysis showed that pazopanib-resistant clones were more frequently in S phase in the absence of pazopanib (Fig. 1b and c). Surprisingly, promotion of cell growth was observed in addition to cell cycle progression in pazopanib-resistant clones, compared with the parental clones in the absence of pazopanib (Fig. 2a). Consistent with in vitro experiments, mice inoculated with pazopanib-resistant SYO-1 cells showed a significant increase in tumour burden compared with the mice inoculated with parental cells (Fig. 2b). In order to address the mechanisms of cell cycle promotion and cell growth in pazopanib-resistant SS clones, as an initial approach we used a human phospho-antibody array (Human Phospho-Kinase Array, Proteome Profiler Array Kit) and semi-quantitate the levels of phosphorylation to study a subset of phosphorylation events in two SS cell lines to investigate signalling pathway profiles. In pazopanib-resistant clones, phosphorylation of ERK1/2 was increased in comparison with parental clones, while phosphorylation of Akt in the resistant clones was comparable or decreased ( Fig. 2c and Table 1). Consistent with the kinase array results, Western blot analysis revealed increased ERK1/2 phosphorylation in pazopanib-resistant clones ( Fig. 2d and Table 1).
A MEK1/2 inhibitor strongly inhibited cell growth and partially reversed acquired pazopanib resistance in pazopanib-resistant clones. We examined whether activated ERK1/2 might be a good therapeutic target for pazopanib-resistant SS cells. To determine the role of ERK1/2 in SS cells, MAPK signalling pathways were inhibited with a chemical inhibitor, and cell cycle and cell proliferation were examined. We used trametinib, an oral, reversible, selective allosteric inhibitor of MEK1/2 activation and kinase activity, to inhibit the ERK signalling pathway 12 . First, Western blotting showed that phosphorylation of ERK1/2 in pazopanib-resistant clones was inhibited to the same extent by trametinib as parental clones (Fig. 3a), indicating that trametinib inhibited the ERK1/2 signalling pathway in both parental and pazopanib-resistant SS clones. Second, we examined the effect of trametinib on the cell cycle. Pazopanib-resistant clones showed a higher proportion of cells in the G1 phase than parental clones after treatment with 10 nM trametinib (Fig. 3b). Third, we performed a chemosensitivity assay with trametinib. Interestingly, trametinib more effectively inhibited the cell growth of pazopanib-resistant clones than parental clones (Fig. 3c). Fourth, we examined the in vivo efficacy of trametinib on pazopanib-resistant SYO-1 clones in the mouse model. As we expected, pazopanib-resistant xenograft mice treated with 0.1 mg/kg trametinib showed a significant decrease in tumour burden while parental xenograft mice treated with the same dose of trametinib showed no significant decrease (Fig. 3d). Finally, to examine whether inhibition of activated ERK1/2 would overcome pazopanib resistance, the pazopanib-resistant SS cells were simultaneously treated with pazopanib and a low dose of trametinib. In the pazopanib-resistant SS cells, even low-dose trametinib partially reversed the resistance to pazopanib chemosensitivity (Fig. 3e). In the SYO-1 mouse model, pazopanib-resistant clones showed higher sensitivity to 10 mg/kg pazopanib along with low-dose trametinib (Fig. 3f). These results suggest that trametinib can be an effective drug for pazopanib-resistant SS, and increased phosphorylation of ERK1/2 is a key to acquired pazopanib resistance.

Mutational status of the RAS-RAF-MEK-ERK pathway and PDGFRα in pazopanib-resistant clones.
To address the mechanism of activation of ERK1/2 in pazopanib-resistant SS cells, we examined aspects of the signalling pathway upstream of ERK1/2. First, we investigated the phosphorylation status of ERK1/2 and PDGFRα , the latter of which is a main target of pazopanib in SS since the PDGF signalling pathway can initiate the MAPK pathway and activate ERK1/2 13 . Interestingly, Western blot analysis showed that phosphorylation of PDGFRα was inhibited in all parental and pazopanib-resistant SS clones in the presence of pazopanib (Fig. 4a). However, pazopanib alone could not sufficiently inhibit phosphorylation of ERK1/2 in only pazopanib-resistant clones (Fig. 4a). Next, we examined kinases upstream of ERK1/2. The activation profile of the RAS-RAF-MEK-ERK pathway was similar between pazopanib-resistant and parental clones in Western blot analysis, with the exception of phospho-ERK1/2 (Fig. 4b). Furthermore, we conducted whole-exome sequencing in the pazopanib-resistant and parental clones to search for any activating mutations of the RAS-RAF-MEK-ERK pathway or any gatekeeper mutations of PDGFRα . There were no such mutations in either the pazopanib-resistant clones (Supplementary Figure 1) or parental clones (data not shown), and no common pazopanib-resistant-specific mutations (data not shown). These findings suggest that upstream signalling elements of ERK1/2, including gatekeeper mutations of PDGFRα , are not involved in the activation of ERK1/2 in pazopanib-resistant SS cells, and pazopanib-resistant clones are resistant to pazopanib inhibition of ERK1/2 phosphorylation. Next, we were interested in specific regulators of ERK1/2. It is known that sustained activation of ERK1/2 does not always correlate with upstream kinases, and dual specificity phosphatases (DUSPs) regulate MAPK activity 14 . We performed gene expression microarray analysis of SYO-1 parental and pazopanib-resistant clones to comprehensively evaluate molecules downstream from ERK1/2 in each clone. We identified downregulation of the DUSP6 gene in the SYO-1 pazopanib-resistant sample using the same criteria described in the Methods section regarding all the DUSPs (Table 2). We next performed quantitative PCR and Western blot analysis to examine mRNA and protein expression of DUSP6 in all SS clones. The resistant clones showed lower expression levels of DUSP6 compared with the parental clones ( Fig. 4c and d). Taken together, these findings suggest that activation of ERK1/2 is sustained, at least in part, via downregulation of DUSP6 expression in pazopanib-resistant SS cells.

Downregulation of DUSP6 in parental SS cells recapitulates pazopanib resistance.
To examine the relevance of DUSP6 to ERK1/2 phosphorylation, cell growth promotion, and pazopanib resistance, we knocked down the expression of DUSP6 in parental clones using the CRISPR/Cas9 system. As shown in Fig. 4e, phosphorylation of ERK1/2 was increased as a result of decreased DUSP6 expression in HS-SY-II cells transfected with small guide RNA (sgRNA) targeting DUSP6.
We next examined the phenotypes of CRISPR-treated HS-SY-II cells. These cells showed enhanced cell growth (Fig. 4f), and in the chemosensitivity assay they exhibited a little greater resistance to pazopanib (Fig. 4g) than

Discussion
Pazopanib is an oral multi-targeted tyrosine kinase inhibitor, and the first molecular target drug approved for soft-tissue sarcoma in Japan. Although a subgroup analysis showed that patients with synovial sarcoma, leiomyosarcoma, vascular tumours, alveolar soft part sarcoma, solitary fibrous tumour, and desmoplastic small round cell tumour had better progression-free survival with pazopanib 15 , most sarcomas eventually acquire drug resistance to pazopanib. Pazopanib targets VEGFRs, PDGFRs, and c-Kit 16 . Since previous studies found that  immunohistochemistry of SS shows high protein expression of PDGFRs and VEGFRs 17 , pazopanib potentially offers advantages in the treatment of SS. Therefore, establishing cells with acquired pazopanib resistance and using SS cell lines to elucidate the mechanism underlying this drug resistance are critical to developing novel treatment strategies.
In the present study, the dose-sensitivity to pazopanib of parental SS clones was comparable to that in a previous study that showed an IC50 value between 2 and 10 μ M (Fig. 1a) 10 . We succeeded in developing SS cell lines with acquired pazopanib resistance with an IC50 of more than 20 μ M (Fig. 1a). It is reported that the recommended dose of pazopanib is 800 mg orally once daily, and a mean target trough concentration was 34 μ M in patients with solid tumours 18 . Additionally, pazopanib of the almost same dose is administered in patients with renal cell carcinoma which is another adaption disease of pazopanib 19 . Therefore, we consider that our experiments at the concentration of 20 μ M is clinically relevant.
Among the two SS cell lines utilised in this study, only SYO-1 is tumourigenic, and we therefore used it to perform preclinical animal experiments. The established pazopanib-resistant clones proved to have several characteristics. In addition to pazopanib resistance, promotion of cell growth was observed in pazopanib-resistant SS cell lines (Fig. 2a). Further, the resistant clones exhibited increases in phosphorylated ERK1/2 (Fig. 2c,d, and Table 1). In addition, we found that a couple of blots were also decreased commonly both SYO-1 and HS-SY-II resistant cells compared to their parental cells ( Table 1). The semi-quantitation revealed that CREB, PRAS40, HSP60 and WNK1 were decreased prominently in SYO-1 (Table 1). Because CREB, PRAS40, HSP60 and WNK1 were considered to contribute to aggressive phenotypes such as tumour progression and metastases 20-23 , we speculated that their downregulation would be less likely to cause the aggressive phenotype of resistant clones. Further investigations are warranted to determine their roles in pazopanib resistance.
It is known that the MAPK pathway regulates diverse cellular processes such as cell growth, apoptosis, malignant transformation, and drug resistance 24,25 . MAPK signalling is tightly regulated and is activated primarily by extracellular growth factor stimulation. Activation of ERK1/2 can be induced by activation of its upstream kinases, RAS, RAF, and MEK1/2 26 . While many reports have shown that the MAPK pathway plays important roles in cancer drug resistance 25 , only a few have reported the involvement of the MAPK pathway in acquired tyrosine kinase inhibitor resistance 27,28 . The inactivation of the MAPK pathway has been proposed as one of the most promising approaches to overcoming acquired resistance to various cytotoxic drugs 29,30 . Indeed, we found that trametinib exerted greater antitumour effects against pazopanib-resistant SS cells than the parental cells, not only in vitro but also in vivo (Fig. 3b, c and d), and that low dose of trametinib along with pazopanib effectively inhibited the growth of pazopanib-resistant SS clones in vitro (Fig. 3e) and in vivo (Fig. 3f). Thus our data as well as those of others suggest that the MAPK pathway is a good molecular target in cases of acquired pazopanib resistance in SS. A phase II clinical trial has just been initiated to evaluate trametinib in combination with pazopanib in patients with advanced gastrointestinal stromal tumour (GIST) (ClinicalTrials.gov Identifier: NCT02342600).  Similarly, simultaneous administration of trametinib and pazopanib was found to exhibit anti-tumour activity in a preclinical thyroid cancer model 31 . Further research regarding the clinical applications of trametinib in patients with advanced SS is required. The causative molecular mechanisms of increased phosphorylated ERK1/2 have been proposed to be the activation of kinases upstream of ERK1/2 26 and/or the inhibition of downstream phosphatases 32 . With regard to upstream molecules in the MAPK pathway, gatekeeper mutations of PDGFRα cause phosphorylation of ERK1/2 through the RAS-RAF-MEK pathway 13 . It has been reported that in cases of GIST, gatekeeper mutations of PDGFRα are frequently observed in resistance to other tyrosine kinase inhibitors such as imatinib 33 . However, a number of our findings suggest that it is unlikely that activation of ERK1/2 results from upstream signalling elements and activating mutations in acquired pazopanib-resistant SS cells: PDGFRα was not activated more in pazopanib-resistant SS cells than in parental SS cells (Fig. 4a), pazopanib was still able to inhibit the phosphorylation of PDGFRα in pazopanib-resistant SS cells (Fig. 4a), phosphorylation of ERK1/2 was partially inhibited by pazopanib (Fig. 4a), and there were no identified gatekeeper mutations of PDGFRα in both of the resistant clones (Supplementary Figure 1). Furthermore, Western blot analysis showed no differences between pazopanib-resistant and parental SS cells in expression levels and phosphorylation of KRAS, BRAF, and MEK1/2 (Fig. 4b).

Probe ID
With regard to downstream specific regulators in the MAPK pathway, DUSPs have been reported to modulate the duration, magnitude, and subcellular compartmentalization of MAPK activity 14 . It is known that DUSP dephosphorylates and inactivates MAP kinases 32 . Using a microarray system, we found that DUSP6 was down-regulated in pazopanib-resistant SYO-1 cells ( Table 2). Consistent with this microarray result, mRNA expression and protein expression of DUSP6 were down-regulated in all pazopanib-resistant clones ( Fig. 4c and d). Forced knockdown of DUSP6 in parental HS-SY-II cells reproduced the increases in ERK1/2 phosphorylation (Fig. 4e). A previous study showed that DUSP6 negatively and specifically modulated ERK1/2 kinase activity 34 . It has also been reported that down-regulation of DUSP6 expression is involved in drug resistance in ovarian cancer 35 , and up-regulation of DUSP6 mediated by p53 caused a cellular senescent phenotype 36 . Taken together with our results, these findings suggest that increased ERK1/2 phosphorylation in pazopanib-resistant SS cells is at least in part due to down-regulation of DUSP6 expression.
As mentioned above, it has been reported that the activated MAPK pathway is associated with acquired cancer drug resistance 25 . Harada et al. reported that acquired resistance to sorafenib, a multi-kinase inhibitor, resulted from the continuous activation of MAPK pathway 27 . In this study, we revealed that trametinib inhibited cell growth of pazopanib-resistant SS cells more than the parental cells (Fig. 3c). In addition, inhibition of ERK1/2 by low-dose trametinib partially reversed pazopanib resistance in vitro (Fig. 3e) and in vivo (Fig. 3f). Conversely, DUSP6 knockdown in parental HS-SY-II cells apparently induced pazopanib resistance, accompanied by an increase in phosphorylated ERK1/2 ( Fig. 4e and g), which recapitulated the acquired resistance to pazopanib in SS cells. Therefore, we conclude that prolonged exposure to pazopanib induces downregulation of DUSP6 by an unknown mechanism, and sustained phosphorylation of ERK1/2 promotes cell cycle progression and proliferation in SS cells.
Bridgeman et al. reported that the addition of trametinib could overcome acquired sunitinib resistance in renal cell carcinoma xenograft models 37 , which supports our conclusion that inhibition of ERK1/2 phosphorylation may be a key to overcome the acquired resistance to antiangiogenic tyrosine kinase inhibitors.
In this study, we found that pazopanib-resistant cells exhibited a marked increase in phosphorylated ERK1/2 (Fig. 2d), and the efficacy of trametinib on inhibiting the cell growth of pazopanib-resistant clones increased compared with the parental clones (Fig. 3c). Furthermore, down-regulation in DUSP6 expression levels were shown in pazopanib-resistant cell lines ( Fig. 4c and d). These results suggest that the increase in ERK1/2 phosphorylation and down-regulation in DUSP6 expression are certainly keys to acquired pazopanib resistance, but the mechanisms of DUSP6 down-regulation are unknown. Further investigations are needed to determine what factors are responsible for the emergence of DUSP6 down-regulation.
In conclusion, this is the first report to address acquired pazopanib resistance in soft-tissue sarcoma. This study reveals a critical role of canonical MAPK signalling in the acquisition of pazopanib resistance through the down-regulation of DUSP6 in SS cells. MEK1/2 inhibition with trametinib is a promising strategy to overcome pazopanib resistance in SS. Reagents. The primary antibodies are summarised in Supplementary Table S1. pazopanib could be solubilised in tissue culture media, and 20 μ M is clinically achievable and lower than the pazopanib concentration of 40 μ M found clinically in patients 40 . Chemosensitivity assay. For the chemosensitivity assay, 2 × 10 5 cells were seeded in 6-well plates. After 24 hours incubation, various concentrations of pazopanib or trametinib were added to the media. After 48 hours incubation, the number of cells was counted with a Z1 Coulter particle counter (Beckman Coulter, Brea, CA, USA) 41 .

Methods
Cell cycle analysis by flow cytometry. Cells were harvested with pazopanib or trametinib, incubated at 37 °C overnight, washed with PBS twice, and fixed with 70% ethanol. Then cells were centrifuged, washed with PBS twice, resuspended in PBS with 10 μ g/mL RNase A and 50 μ g/mL propidium iodide, and incubated for 30 minutes on ice. Alterations in cell distribution were analysed using a BD Accuri TM C6 Flow Cytometer (BD Biosciences, San Jose, CA, USA). For each sample, 20,000 events were scored. Mouse tumour xenograft model. week-old BALB/C nude mice were obtained from Charles River Japan (Fukuoka, Japan) and maintained in a "specific pathogen"-free environment throughout the experiment. Each SYO-1 clone (1 × 10 7 ) was suspended in 200 μ l BD Matrigel TM Basement Membrane Matrix obtained from BD Biosciences (San Jose, CA, USA) and injected subcutaneously. When the tumour volume reached 100-200 mm 3 , mice were randomly divided into three groups, each consisting of at least four mice. Pazopanib or trametinib suspended in dimethyl sulfoxide (DMSO) and PBS were administered orally three times weekly at 0, 10, 30, or 50 mg/kg, and 0, 0.1, 0.3, or 3 mg/kg, respectively. Tumour volume was measured twice a week by calipers using the following equation: tumour volume (mm 3  Target selection and sequencing. Exome sequencing was conducted for six DNA samples from each clone. Genomic DNA was extracted using QIAamp DNA Mini Kit (QIAGEN, Hilden, Germany), sheared into approximately 150-200 bp fragments, and used to make a library for multiplexed paired-end sequencing (Illumina, Tokyo, Japan). The constructed library was hybridised to biotinylated cRNA oligonucleotide baits from the SureSelect Human All Exon V6 Kit (Agilent Technologies, Santa Clara, CA, USA) for exome capture. Targeted sequences were purified by magnetic beads, amplified, and sequenced on an Illumina HiSeq2500 platform in a paired-end 100-bp configuration.
Gene expression microarrays. The total RNA was isolated from each cell using TRIzol Reagent (Invitrogen) and purified using SV Total RNA Isolation System (Promega). RNA samples were quantified by an ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA) and the quality was confirmed with an Experion System (Bio-Rad Laboratories, Hercules, CA, USA). The cRNA was amplified, labelled, and hybridised to a 60-K Agilent 60-mer oligomicroarray. All hybridised microarray slides were scanned by an Agilent scanner. Relative hybridization intensities and background hybridization values were calculated using Agilent Feature Extraction Software (9.5.1.1).
Data analysis and filter criteria. Raw signal intensities and flags for each probe were calculated from hybridization intensities (gProcessedSignal) and spot information (gIsSaturated, etc.), according to the procedures recommended by Agilent. (Flag criteria on GeneSpring Software. Absent (A): 'Feature is not positive and significant' and 'Feature is not above background' . Marginal (M): 'Feature is not Uniform' , 'Feature is Saturated' , and 'Feature is a population outlier' . Present (P): others.) The raw signal intensities of two samples were log 2 -transformed and normalised by quantile algorithm with the 'preprocessCore' library package 44 of Bioconductor software 45 . We selected probes with a 'P' flag call in both samples. To identify up-or down-regulated genes, we calculated Z-scores 46 and ratios (non-log scaled fold-change) from the normalised signal intensities of each probe for comparison between SYO-1 parental samples and pazopanib-resistant samples. Then we established criteria for regulated genes: up-regulated genes, Z-score ≥ 2.0 and ratio ≥ 1.5-fold; down-regulated genes, Z-score ≤ − 2.0 and ratio ≤ 0.66. A total of 1107 probes were obtained. RNA preparation and quantitative PCR. Total RNA from each clone was extracted using the RNeasy Lipid Tissue Mini kit (QIAGEN). Quantitative PCR was carried out using a LightCycler 1.5 as previously described (Perfect Real Time, Takara Bio, Otsu, Japan) 47 . The primers are summarised in Supplementary Table S2. Data were standardised using GAPDH as a housekeeping gene. A negative control was also prepared using distilled water instead of a DNA template. The assay was performed in triplicate and was repeated in at least three separate experiments. The expression of mRNA was calculated using LightCycler version 3.5 software (Roche Diagnostics). CRISPR/Cas9 system. The following were purchased from Thermo Fisher Scientific (Waltham, MA, USA): template DNA designed for gRNA, MEGAshortscript TM Kit, MEGAclear TM Kit, Lipofectamine TM CRISPRMAX TM Transfection Reagent, and GeneArt TM Platinum Cas9 Nuclease. Transfection to the parental clones was carried out using Cas9 protein and lipid-mediated transfection according to the manufacturer's instructions 48 . Statistical analysis. The Wilcoxon signed-rank test was used for two-group comparisons. P < 0.05 was considered to be statistically significant. Data in graphs are given as means ± standard deviation (S.D.). All statistical analyses were performed with the Statistical Analysis System (SAS) software package (JMP9, SAS Institute Inc., Cary, NC, USA).