A comprehensive evaluation of Hippo pathway silencing in sarcomas

TAZ and YAP are transcriptional coactivators negatively regulated by the Hippo pathway that have emerged as key oncoproteins in several cancers including sarcomas. We hypothesized that loss of expression of the Hippo kinases might be a mechanism of activating TAZ and YAP. By immunohistochemistry, TAZ/YAP activated clinical sarcoma samples demonstrated loss of MST1 (47%), MST2 (26%), LATS1 (19%), and LATS2 (27%). Western blot similarly demonstrated loss of MST1 (58%), MST2 (25%), and LATS2 (17%). Treatment with MG132 demonstrated an accumulation of MST2 in 25% of sarcoma cell lines, indicating that proteosomal degradation regulates MST2 expression. qRT-PCR in sarcoma cell lines demonstrated loss of expression of the Hippo kinases at the RNA level, most pronounced in MST1 (42%) and MST2 (25%). 5-azacytidine treatment in sarcoma cell lines modestly reversed expression of predominantly MST1 (8%) and MST2 (17%), indicating CpG island hypermethylation can silence expression of MST1 and MST2. Trichostatin A treatment reversed expression of MST1 (58%) and MST2 (67%), indicating histone deacetylation also plays a role in silencing expression of MST1 and MST2. Loss of expression of the Hippo kinases is frequent in sarcomas and is due to a variety of mechanisms including regulation at the post-translational level and epigenetic silencing.


INTRODUCTION
TAZ (WWTR1 is the gene) and YAP (YAP1 is the gene) are developmentally important transcriptional coactivators [1,2] that have emerged as central oncoproteins in a number of carcinomas including breast [3,4], colon [5], liver [6], lung [7,8], pancreas [9], and thyroid cancers [10]. TAZ (transcriptional coactivator with PDZ-binding motif) and YAP (yes associated protein) do not contain DNA binding domains of their own, and must complex with other transcription factors that contain DNA binding domains via their TEAD binding domain or WW domain [11][12][13][14][15]. The TEA domain (TEAD) family of transcription factors have been demonstrated to be the dominant transcription factors with regards to mediating the TAZ and YAP transcriptional programs [16,17]. TAZ and YAP are negatively regulated by the Hippo pathway, a series of serine/threonine kinases including the STE20-like protein 1 and 2 (MST1/2) [18][19][20] and the large tumor suppressor 1 and 2 (LATS1/2) [21,22]. The MOB1A/B [23] and Salvador proteins [20,24] have been shown to form a scaffold for the above kinases. Cell confluence [2] and detachment [25] activate Research Paper www.oncotarget.com the Hippo pathway, causing LATS1/2 to phosphorylate TAZ and YAP on several serines, including serine 89 on TAZ and serine 127 on YAP. Phosphorylation of these serines leads to binding of 14-3-3 proteins, ultimately resulting in translocation of TAZ/YAP from the nucleus into the cytoplasm, where they undergo ubiquitin-mediated degradation [1,2].
A number of different signal transduction pathways have been identified that modulate the activity of the Hippo kinases or TAZ and YAP directly. The Wnt signaling pathway has been shown to activate TAZ and YAP directly via their interaction with β-catenin [26,27]. The PI3 kinase pathway activates TAZ by inhibiting glycogen synthase kinase-3 [28] and activates YAP by promoting the dissolution of the Hippo kinase signaling cascade [29]. G protein coupled receptors have been shown to activate TAZ and YAP by dampening activity of the LATS1 and LATS2 kinases [30][31][32]. This has led to a paradigm that TAZ and YAP are activated predominantly via cross-talk with other signal transduction pathways.
In contrast, primary lesions affecting the Hippo kinases have been rarely identified. Although TAZ and YAP have been shown to be activated oncoproteins in a number of carcinomas [33,34] and sarcomas [35], genetic alterations are rare with the exception of the WWTR1-CAMTA1 [36][37][38] and YAP1-TFE3 gene fusions in epithelioid hemangioendothelioma (EHE) [39]. Mutations in the upstream Hippo kinases, MST1, MST2, LATS1, and LATS2 have also been rare. Occasional mutations have been identified in the scaffolding proteins MOB1A/B [33]. In silico studies have suggested that copy number changes (deletions) of genes upstream of the Hippo kinases (e.g. NF2) may result inactivation of TAZ and YAP [40], however the frequency of these genomic alterations is incompletely understood. More recently, the Itch ubiquitin ligase has been shown to reduce expression of the LATS1 kinase [41,42]. Scattered reports indicate the presence of promoter hypermethylation of several of the Hippo kinases, although evaluation of how promoter hypermethylation affects expression levels and functional activity of the kinases is incomplete [43][44][45]. In summary, although various lines of evidence indicate the presence of primary lesions of the Hippo pathway, their true frequency and significance is incompletely understood. To address this, we performed a comprehensive evaluation of expression of the Hippo kinases in sarcomas, a group of cancers that have recently been shown to harbor frequent activation of the TAZ and YAP oncoproteins [35].

Expression of the Hippo kinases is lost in TAZ/ YAP activated clinical sarcoma samples
We have recently demonstrated that TAZ and YAP are constitutively activated and located within the nucleus of the majority of sarcoma clinical samples. Evaluating expression of TAZ and YAP in multiple histological types of sarcoma revealed that approximately 50% of sarcoma clinical samples demonstrate activated YAP, while 66% demonstrate activated TAZ. ( Figure 1A) [35]. Mutations within components of the Hippo pathway have reported to be rare [33]. Evaluation of 259 sarcomas in The Cancer Genome Atlas demonstrated a mutation rate ranging from 0% for MST2 and LATS1 to 0.8% for LATS2 (Supplementary Figure 1A). Several lines of evidence indicate that silencing of the Hippo kinases is necessary for activation of TAZ and YAP. The TAZ-CAMTA1 fusion protein has been demonstrated to negate inhibition from the Hippo pathway [38]. Other lines of evidence indicate that the Hippo pathway is silenced secondarily via interactions with other pathways [26][27][28][29]. Some reports have addressed the possibility that the Hippo pathway is primarily silenced through promoter hypermethylation [43][44][45] or ubiquitin mediated degradation [41,42], however this has not been investigated in a comprehensive manner.
One mechanism by which the Hippo pathway could be primarily silenced is through loss of expression of the Hippo kinases. To assess this possibility, we evaluated expression of the Hippo kinases in an unbiased way through immunohistochemistry for MST1, MST2, LATS1, and LATS2 on the tissue microarray ( Figure 1B and 1C). Normal smooth muscle was utilized as a positive control (Supplementary Figure 1B) Figure 1C). The similar frequency of loss of the Hippo kinases in TAZ/ YAP activated sarcomas versus all sarcomas regardless of TAZ/YAP activation status indicates that loss of the Hippo kinases is associated with TAZ/YAP activation the majority of the time.
We next determined the frequency at which at least one of the Hippo kinases was lost among a panel of 110 sarcomas. Approximately 75% of the sarcomas demonstrate a loss of expression of at least one of the Hippo kinases ( Figure 1D). All of the synovial www.oncotarget.com sarcomas, malignant peripheral nerve sheath tumors, high grade osteosarcomas, soft tissue leiomyosarcomas, well-differentiated liposarcoma, and dedifferentiated liposarcomas demonstrated loss of expression of at least one of the Hippo kinases. The frequency of loss of expression in the remaining sarcomas ranged from 46% (12/26) in undifferentiated pleomorphic sarcoma to 86% (6/7) in uterine leiomyosarcoma ( Figure 1D). No angiosarcomas or chondrosarcomas were noted to exhibit loss of the Hippo kinases, but the sample sizes of these sarcomas were small.
We then looked at other combinations of loss of the Hippo kinases in YAP or TAZ activated sarcomas in an effort to ascertain what combinations of the Hippo kinases are required for activation of TAZ/YAP to occur. Loss of expression of MST1 and MST2 was identified in 15% (17/114) of sarcomas. Loss of expression of LATS1 and LATS2 was noted in 8% (9/113) of sarcomas. Loss of expression of all four Hippo kinases was identified in 1 of 110 (1%) of all sarcomas (Supplementary Figure 1D).

Expression of the Hippo kinases is lost in TAZ/ YAP activated sarcoma cell lines
As we have previously shown, TAZ and YAP are constitutively activated and located within the nucleus of sarcoma cell lines (Figure 2A, Supplementary Figures  2 and 3). Essentially all of the sarcoma cell lines assayed demonstrate nuclear localization of TAZ and YAP when grown to confluence ( Supplementary Figures 2 and 3), consistent with a lack of negative regulation by the Hippo pathway. With the exception of MOB1A/B [33], mutations in the upstream Hippo kinases (especially MST1/2) have been reported to be extremely rare. Mutations in TAZ and YAP have also been shown to be extremely rare, with the exception of the WWTR1-CAMTA1 and YAP1-TFE3 gene fusions which are disease defining genetic alterations found in essentially all epithelioid hemangioendotheliomas [37][38][39]. In the majority of cancers, WWTR1 and YAP1 have been reported to lack mutations [33]. To confirm this is the case in sarcomas, we performed targeted PCRbased Sanger sequencing of 12 sarcoma cell lines and GCT (giant cell tumor) for the presence of mutations in the serines in TAZ (S66, 89, 117, and 311) [46] and YAP (S61, 109, 127, 164, and 381) [2] phosphorylated by LATS1/2. GCT is an immortalized, non-sarcoma mesenchymal cell line, and expression of the Hippo kinases appeared to be similar as compared to MCF10A, an immortalized but non-transformed cell line commonly utilized as a negative control epithelial cell line in the field [3,47] (Supplementary Figure 4A). For this reason, this cell line was used as a control in the subsequent experiments. None of the 13 cell lines evaluated by Sanger sequencing demonstrated mutations in the above mentioned serine residues (Supplementary Figure 4B).
Expression levels of the Hippo kinases were then evaluated at the protein level by western blot. Qualitatively, expression of at least one of the Hippo kinases was reduced by approximately 2-fold in 10 of the 12 sarcoma cell lines (83%) compared to GCT ( Figures 2B and 7A), closely approximating the 75% of clinical samples that had lost at least one of the Hippo kinases ( Figure 1D). Quantitative loss of the MST1, MST2, and LATS2 kinases, the kinases whose expression was most commonly lost, was confirmed in a sampling of 3 cell lines, SKLMS1, RD, and A204 ( Figure 2C and Supplementary Figure 5A).
Protein expression levels can be regulated at multiple levels, including altered protein turnover. To test the hypothesis that protein degradation could play a role in Hippo kinase expression, the 12 sarcoma cell lines were treated with 10 µM MG132 for 12 hours. After 12 hours, MG132 was shown to quantitatively increase expression of MST2 in 3 of the 12 cell lines (25%) (Figure 3 and Supplementary Figure 5B), indicating that enhanced proteosomal degradation leads to decreased Hippo kinase expression in sarcoma cell lines. None of the other Hippo kinases showed evidence of increased proteosomal degradation. In some cell lines in which MST2 expression increased with MG132 treatment, LATS2 expression was found to demonstrate the opposite relationship and decreased.

Copy number changes (deletions) of the Hippo kinases are not a common event in sarcomas
To evaluate the possibility that loss of expression was occurring due to alterations at the genomic level, copy number changes/deletions were evaluated utilizing data from The Cancer Genome Atlas data set. Prior studies in sarcomas have highlighted in silico data emphasizing deletions in NF2 (merlin), a regulatory protein upstream of the Hippo pathway [40]. Potential genomic deletions in the Hippo kinases themselves have not been thoroughly investigated. In silico analysis demonstrated that 0.8% of sarcomas demonstrated genomic deletions of MST2 and LATS2 combined, indicating that deletions are not a common mechanism by which loss of expression of the Hippo kinases occurs ( Figure 4A). To confirm the in silico data, we performed fluorescence in situ hybridization (FISH) for MST1, MST2, LATS1, and LATS2 on cell lines ( Figure 4B-4G).
No genomic deletions of the Hippo kinases were identified by FISH in the sarcoma cell lines, with the exception of the RD cell line. The RD cell line demonstrated a disomic signal for LATS2 in a polyploid cell, indicating a deletion or loss of chromosome 13. Given that the LATS2 copy number is still two, the contribution to LATS2 expression is unclear.

Expression of MST1 and MST2 is reduced at the RNA level in sarcoma cell lines
Since genomic alterations/deletions of the Hippo kinases was not responsible for decreased expression of the Hippo kinases, and only MST2 appears to be regulated www.oncotarget.com at a protein level, we asked whether the Hippo kinases expression were regulated at a transcriptional level. To test this hypothesis, we evaluated whether decreased expression of the Hippo kinases at the protein level was due to decreased expression of the Hippo kinases at the RNA level. Quantitative RT-PCR was performed on all 12 sarcoma cell lines and expression normalized to GCT. Approximately 42% of cell lines demonstrated loss of expression of MST1 at the RNA level, while 25% of cell lines demonstrated loss of expression of MST2 at the RNA level ( Figure 5A). In contrast, loss of expression of the LATS1 and LATS2 kinases at the RNA level was less frequently observed. Approximately 8% of sarcoma cell lines demonstrated loss of expression of LATS1 at the RNA level, while no cell lines demonstrated loss of expression of LATS2 at the RNA level. The above findings suggest that loss of expression of MST1 and MST2 are regulated at a transcriptional level, while loss of expression of LATS1 and LATS2 are regulated by other mechanisms. Several authors have noted promoter hypermethylation of MST1, MST2, LATS1, and LATS2 in various cancers [43][44][45]. However evaluation of how promoter methylation affects expression has not been comprehensively evaluated. To determine whether promoter hypermethylation affected expression of the Hippo kinases, we evaluated TCGA methylation data acquired from Firebrowse.org [48]  Figure 6). A modest negative correlation was identified between methylation of CpG islands and RNA expression for MST2 (r = −0.4) ( Figure 5B).
To begin addressing whether hypermethylation of these promoters is functionally significant, we treated sarcoma cell lines with 10 µM 5-azacytidine (a covalent inhibitor of DNA methyltransferase I [49]) for 4 days. Quantitative RT-PCR demonstrated 6 cell lines that approximated at least a 1.5 fold increase in expression, however, only the RD cell line demonstrated a fold change increase greater than 2 in expression of MST1. Similarly, treatment with 5-azacytidine demonstrated 2 cell lines, RD and SW872, that had a greater than 2-fold increase in expression of MST2. Altogether, the data demonstrate that expression of MST1 and MST2 are modestly regulated by methylation of DNA, mirroring the TCGA in silico data. None of the cell lines demonstrated a two-fold increase in expression of LATS1 or LATS2 with treatment with 5-azacytidine ( Figure 5C).

Expression of MST1 and MST2 is silenced by histone deacetylation
Since treatment with 5-azacytidine (AZA) did not significantly rescue expression of MST1 and MST2 in many sarcoma cell lines, it suggested that promoter hypermethylation was not the predominant transcriptional mechanism regulating expression of MST1 and MST2. Promoter hypermethylation is linked structurally to histone acetylation. Histone deacetylases are known to bind to methylated DNA via the MBD2/3 and MeCP2 adaptor proteins [50,51].
This suggested that the promoter hypermethylation might coordinately regulate expression of the MST1 and MST2 kinases with histone deacetylation. To determine more directly whether promoter hypermethylation and histone deacetylation work in conjunction with one another to regulate expression of the Hippo kinases, we treated cells with 5 µM 5-azacytidine (AZA) for 4 days followed by treatment with 500 nM trichostatin A (inhibits class I, II, and IV histone deactylases (HDACs) by interacting with the Zn(II) in the catalytic site [52]) for 12 h ( Figure 6A-6D). A lower concentration of 5-azacytidine was used in the majority of cell lines, rather than 10 µM as in previous experiments, so that addition of trichostatin A could be tolerated in the cell lines. Some lines could only tolerate lower levels of trichostatin A, including SJCRH30 (0.125 µM trichostatin A) and SNF02.2 (0.25 µM trichostatin A). The HT1080, SKLMS1, and SW982 cell lines could only tolerate 1 µM 5-azacytidine in combination with trichostatin A. Trichostatin A (TSA) and AZA treatment alone controls were included. Seven of the 12 sarcoma cell lines (58%) demonstrated a two-fold or greater increase in expression of MST1 with treatment with TSA. TSA and AZA had a statistically significant additive effect in the HT1080 cell line only ( Figure 6A). TSA stimulated an increase in expression of MST2 in 8 of 12 sarcoma cell lines (67%) ( Figure 6B). No increase in expression of LATS1 was seen with TSA alone ( Figure  6C). TSA promoted an increase of expression of LATS2 in 1 of 12 sarcoma cell lines (8%). A statistically significant additive effect was seen with regards to LATS2 expression in the HT1080 and RD cell lines with simultaneous treatment with trichostatin A (TSA) and 5-azacytidine (AZA) ( Figure 6D).
The effectiveness of TSA in the above experiments in promoting histone acetylation was validated in three cell lines, RD, Saos2, and sNF96.2 that demonstrated an increase in expression of MST1 and MST2. These three lines were treated with 0.5 µM TSA for 24 hours. Lysates were then probed with a pan-acetyl histone H3 antibody recognizing acetylated K9, K14, K18, K23, and K27 (Supplementary Figure 7A) Figure 7B and 7C), confirming that reversal of expression with treatment with TSA was not an off-target effect but due to HDAC inhibition.

DISCUSSION
Although TAZ and YAP have been acknowledged as central oncoproteins in multiple cancers including breast, colon, liver, lung, pancreatic, and thyroid cancers [33], the dominant paradigm is that these two oncoproteins are activated relatively late in tumor progression. This is due to the observation that few mutations of TAZ and YAP or the upstream Hippo kinases are activated. Challenging this paradigm has been the discovery of the TAZ-CAMTA1 [36][37][38] and YAP-TFE3 [39] fusion proteins in epithelioid hemangioendothelioma (EHE), a vascular sarcoma. EHE contains a t(1;3)(p36;q25) [36][37][38] or t(X;11) [39]chromosomal translocations encoding the TAZ-CAMTA1 and YAP-TFE3 fusion proteins, respectively. EHE contains few other cytogenetic alterations other than the above mentioned chromosomal translocations, indicating that these chromosomal translocations and gene fusions are the initiating event. The TAZ-CAMTA1 fusion protein [38] and presumably the YAP-TFE3 fusion protein function by constitutively activating the N terminal portion of TAZ and YAP so that they are no longer negatively regulated by the Hippo pathway. Expression of the TAZ-CAMTA1 fusion protein has been demonstrated to transform cells and promote various hallmarks of cancer [38], demonstrating that activated TAZ and YAP can serve as the initiating oncoproteins in cancers.
The question of whether TAZ and YAP serve as the initiating oncoproteins in other cancers is still largely unresolved. Because the preponderance of evidence indicates that the Hippo kinases are inactivated and that TAZ and YAP are activated by a complex interaction with multiple signal transduction pathways, the dominant view is that they must be activated later in tumor progression due to the time required for mutations to occur in other signal transduction pathways [53]. The above paradigm focuses on post-translational mechanisms by which the Hippo kinases, TAZ, and YAP can be regulated. In this study we identified several mechanisms via which the www.oncotarget.com  Hippo kinases can directly be down-regulated, suggesting that TAZ and YAP might be activated earlier in tumor initiation/progression than previously stipulated.
Herein, utilizing an unbiased approach with clinical samples, we showed in sarcomas that the Hippo kinases can be primarily dysregulated by a variety of mechanisms resulting in their loss of expression ( Figure 7A, 7B). Loss of expression of the Hippo kinases was identified in clinical sarcoma samples demonstrating activated TAZ and YAP at a relatively high level, ranging from 19% (LATS1) to 47% (MST1). The majority of TAZ/YAP activated clinical sarcoma samples (75%) demonstrated loss of expression of one of the Hippo kinases, while other combinations of loss of expression of the Hippo kinases was much less frequent, suggesting that loss of expression of one of the Hippo kinases may be sufficient to activate TAZ or YAP. These findings were mirrored in sarcoma cell lines that demonstrated constitutive activation of TAZ/ YAP which demonstrated a similar frequency with regards to loss of expression of at least one of the Hippo kinases (83%) ( Figure 7A).
Identifying sarcoma cell lines that demonstrated loss of expression of the Hippo kinases allowed us to identify various mechanisms by which this loss of expression was occurring. These mechanisms include: regulation by ubiquitin-mediated degradation, as well as two distinct methods of epigenetic regulation, promoter hypermethylation and histone deacetylation ( Figure 7B). The Itch E3 ubiquitin protein ligase has already been implicated in regulation of LATS1 [41,42], we also identify that MST2's expression is negatively regulated by a ubiquitin ligase in several sarcoma cell lines. Additional studies to identify this ubiquitin ligase is needed. The finding that LATS2 expression decreased with MG132 treatment in some sarcoma cell lines suggests that treatment with proteosomal inhibitors for sarcomas with decreased MST2 expression may not be effective. Additional preclinical studies to study these potentially competing effects are warranted.
The presence of promoter hypermethylation in the Hippo kinases, including all four of the Hippo kinases has been noted for some time in sarcomas, astrocytomas, and breast cancer [43][44][45]. Although some studies showed a correlation between mRNA levels and the degree of promoter hypermethylation, it was unclear how tightly promoter hypermethylation regulated Hippo kinase expression. Our in silico analysis and in vitro experiments demonstrate that promoter hypermethylation modestly regulates Hippo kinase expression, but that it can be reversed in some contexts with DNA methyltransferase inhibitors such as 5-azacytidine. Because the effects of treatment with 5-azacytidine was modest, it suggested that other epigenetic mechanisms may be responsible for regulating expression of the Hippo kinases. Since promoter hypermethylation and histone deacetylation are tightly linked processes, due to the presence of adaptor proteins that are present, we hypothesized that histone deacetylation may play a role in negatively regulating the expression of the Hippo kinases. Indeed, inhibition of histone deactylase by trichostatin A resulted in a greater than 2 fold increase in MST1 and MST2 expression in the majority of cell lines, confirming that histone deacetylation plays a role in the regulation of MST1 and MST2. This data suggests a new mechanism of dampening MST1 and MST2 expression by histone deacetylation.
Additional preclinical studies are required to evaluate the feasibility of upregulating expression of the Hippo kinases in patients with TAZ/YAP activated sarcomas by modulating proteosomal degradation of MST2, and potentially inhibiting DNA methyltransferases and histone deacetylases.

Tissue microarray construction
A total of 159 untreated sarcomas were retrieved from the University of Iowa Department of Pathology with previous approval from the Institutional Review Board. The tissue microarray was constructed by arraying 1.0 mm cores taken from formalin fixed paraffin embedded tissue and assembled using a MTA-1 tissue arrayer from Beecher Instruments (Sun Prairie, WI) as previously described [35]. Sarcomas were classified according to World Health Organization criteria [54].

Western blot
Harvested cells were lysed in radioimmunoprecipitation assay (RIPA) complete lysis buffer with the addition of Complete Protease Inhibitor Cocktail (Roche) and PhosSTOP Phosphatase Inhibitor Cocktail (Roche) according to the manufacturer's instructions. Total protein concentration was measured using Pierce BCA™ Protein Assay Kit (ThermoFisher Scientific, Waltham, MA). Proteins were transferred to a polyvinylidene difluoride (PVDF) membrane and probed with antibodies described above. Immediately after transfer, membranes were stained with 10 mLs of Ponceau S (0.1% (w/v) in 5% acetic acid) (Sigma-Aldrich, catalog# P7170) [55].

Quantitation of Western blots
Samples utilized for quantiative Western blot were run in triplicate. Western blot images were exported as tiff files and converted to grey scale jpeg images with Photoshop. The images were then quantitated in Image J using "Grey Mean Value" measured using a previous described protocol [56]. A region of interest was defined and used across blots. Protein of interest and Ponceau S loading control values were acquired, densities were inverted and background values were subtracted. Net protein/net loading control ratios were calculated for each protein. The ratios were then normalized to an internal control (cell line or cell line/treatment condition) on each blot.

Immunofluorescence staining
Cell lines were fixed with 4% paraformaldehyde (in 1X PBS) for 15 min. After washing with PBS, cells were permeabilized and blocked with 0.3% Triton X-100 and 3% fetal bovine serum for 30 min. Cells were incubated with anti-TAZ and anti-YAP antibody diluted (1:100) in 3% fetal bovine serum at 4° C overnight in a humidity chamber. The primary antibody was removed, cells washed, then incubated with Alexa Fluor 488-conjugated secondary antibody (Invitrogen-Life Technologies) for 45 minutes to 1 hr at room temperature. Immunofluorescence was visualized using a Nikon Eclipse E600 microscope (Tokyo, Japan) with SPOT imaging software or an Olympus BX-61 microscope (Tokyo, Japan) with cellSens imaging software.

Cell culture
Sarcoma cell lines were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA) and were cultured in DMEM, RPMI, or McCoy's media (according to ATCC recommendations) containing 10% Fetal Bovine Serum (Invitrogen-Life Technologies), 1 mM sodium pyruvate, and 50µg/mL pen/strep. All cells were cultured at 37° C and 5% CO 2 . To detect proteosomal degradation, 10 µM MG132 was added to cells for 12 hours. To detect reconstitution of the Hippo kinases, either 5 or 10 µM 5-azacytidine (Sigma-Aldrich) was added to cell lines for 4 days. When 5-azacytidine and trichostatin A (Sigma-Aldrich) were combined, 0.125 µM, 0.25 µM, or 0.5 µM trichostatin A (maximal amount tolerated in combination with 5-azacytidine) was added for 12 hours on day 4 of the 5-azacytidine treatment. 1 µM MS-275 (Tocris, Minneapolis, MN), an additional HDAC inhibitor, was added to cell lines for 24 hours.

Quantitative reverse transcription polymerase chain reaction
Total RNA was isolated from sarcoma cell lines using TRIzol reagent (Invitrogen-Life Technologies). Total RNA was treated with DNase (Invitrogen-Life Technologies), and then column purified using the PureLink RNA mini kit (Ambion-Thermo Fisher Scientific). 1 µg of DNase treated RNA was converted to cDNA using Superscript III Reverse Transcriptase (Invitrogen-Life Technologies) and 50 ng of random primers (Promega, Madison, WI USA). PCR amplification was performed in technical triplicates on the Applied Biosystems QuantStudio 3 Real-Time PCR System (Applied Biosystems-Life Technologies). Relative quantitation was performed utilizing the delta-delta C T method and the geometric mean of ACTB (β-actin) and POLR2A (RNA Polymerase II) C T values as the reference control. The TaqMan Universal PCR Master Mix (Applied Biosystems-Life Technologies) was utilized as well as PrimeTime standard qPCR primer/probe sets from Integrated DNA Technologies (Iowa City, IA, USA). The Taqman based approach utilized the following primers and probes: POLR2A-R 5′-CGTATTCGCATCATGAACAGC-3′; POLR2A-Probe 5′-ACCACCTCTTCCTCCTCTTGCATCT-3′; The following qPCR cycling conditions were used: 95° C 10:00 (95° C 0:15 , 60° C 1:00 ) 40 .

Statistics
Standard deviation for the quantitative western blots was calculated from fold change expression derived from different western blot experiments. For quantitative RT-PCR, standard deviation was calculated from fold change values from each triplicate. Correlation between fraction methylation of CpG islands and expression of the Hippo kinases (RSEM [log 2 ]) was calculated by Spearman's correlation coefficient.

Author contributions
NM was involved in conceptualization, formal analysis, data curation, writing, review, and editing. CF, SH, QQ, CD, JT, and AM were involved in data curation. AD was involved in data curation, formal analysis, writing, review, and editing. BD was involved in data curation, formal analysis, writing, review, and editing. MT was involved in conceptualization, project administration, formal analysis, data curation, writing, review, and editing.