miR-193b regulates tumorigenesis in liposarcoma cells via PDGFR, TGFβ, and Wnt signaling

Liposarcoma is the most common soft tissue sarcoma. Molecularly targeted therapeutics have had limited efficacy in liposarcomas, in part because of inadequate knowledge of the complex molecular alterations in these tumors. Our recent study revealed the tumor suppressive function of miR-193b in liposarcoma. Considering the biological and clinical heterogeneity of liposarcoma, here, we confirmed the under-expression of miR-193b in additional patient liposarcoma samples and cell lines. Based on STRING analysis of protein-protein interactions among the reported putative miR-193b targets, we validated three: PDGFRβ, SMAD4, and YAP1, belonging to strongly interacting pathways (focal adhesion, TGFβ, and Hippo, respectively). We show that all three are directly targeted by miR-193b in liposarcoma. Inhibition of PDGFRβ reduces liposarcoma cell viability and increases adipogenesis. Knockdown of SMAD4 promotes adipogenic differentiation. miR-193b targeting of the Hippo signaling effector YAP1 indirectly inhibits Wnt/β-catenin signaling. Both a PDGFR inhibitor (CP-673451) and a Wnt/ β-catenin inhibitor (ICG-001) had potent inhibitory effects on liposarcoma cells, suggesting their potential application in liposarcoma treatment. In summary, we demonstrate that miR-193b controls cell growth and differentiation in liposarcoma by targeting multiple key components (PDGFRβ, SMAD4, and YAP1) in several oncogenic signaling pathways.

In our previous study, we detected transcriptomic and proteomic changes in miRNA-treated liposarcoma cells and identified 50 potential miR-193b targets using multiple miRNA target prediction tools 12 . To further understand miR-193b target networks, we applied the functional protein-protein interaction analysis method STRING, which revealed interactions among multiple signaling pathways (Fig. S1). Among them, focal adhesion and Hippo signaling are the top two, with the lowest false discovery rate values. Our prior study identified members of the focal adhesion pathway, FAK (PTK2) and CRKL, as direct targets of miR-193b 12 ; STRING analysis indicated that miR-193b targets both of those as well as PDGFRβ to repress the focal adhesion pathway (Fig. 2d). STRING also implicated SMAD4 and Yes-associated protein (YAP)-1, members of the Hippo pathway, as new targets of miR-193b, in addition to the previously identified YWHAZ 13 . These results suggest that these three novel miR-193b targets may interact with previously identified miR-193b targets to modulate multiple oncogenic signaling pathways in liposarcoma.
PDGFRβ is a direct target of miR-193b and inhibition of its activity attenuates differentiation and proliferation of liposarcoma cells. PDGFR signaling plays a crucial role in cancer development and progression 14 , so regulation of PDGFRβ by miR-193b could contribute to liposarcoma progression. To test this hypothesis, we overexpressed miR-193b in liposarcoma cells. Overexpression of miR-193b inhibited wild-type PDGFRβ-3′UTR reporter (WT) activity by 50% more than the reporter lacking the 3′UTR region (Fig. 3a). Mutation of seed sites in the 3′UTR reporter (Mut) completely blocked the repression induced by miR-193b (Fig. 3a), suggesting that miR-193b regulates PDGFRβ through its seed sites in the 3′UTR. Consistently, both mRNA and protein levels of PDGFRβ were repressed by miR-193b overexpression, while addition of anti-miRNA for miR-193b attenuated this regulation (Fig. 3b,c). These results demonstrate that miR-193b directly regulates PDGFRβ expression in WDLS/DDLS cells.
Besides suppressing tumor growth, miR-193b has been reported to promote brown fat differentiation in mouse cells 13 . Confirming our previous findings that miR-193b induces adipogenic differentiation in human adipose tissue-derived stem cells (ASCs) 12 , overexpression of miR-193b significantly increased both mRNA and protein levels of the adipogenic markers CEBPA, PPARG, and FABP4 (Fig. S2a). These markers were also induced Network nodes represent proteins. Same-colored target genes are enriched in the same pathway. Edges represent protein-protein associations, colored according to association types. Putative targets in italics are previously validated; those in bold were further investigated in this study.
www.nature.com/scientificreports www.nature.com/scientificreports/ by culture in differentiation medium (Fig. S2b). To assess PDGFRβ's role in miR-193b-mediated growth inhibition and adipogenic differentiation, we used siRNAs to knock down endogenous PDGFRβ in WDLS and DDLS cells. PDGFRβ knockdown significantly inhibited cell viability (by 30-40% at day 5) in both cell types (Fig. 3d). In addition, knockdown of PDGFRβ promoted adipogenic differentiation in both WDLS and ASCs, evident as increased CEBPA and PPARG protein expression and lipid droplet formation (Fig. 3e,f). Interestingly, PDGFRβ siRNA reduced levels of the adipogenic marker FABP4 in WDLS, but had no effects on FABP4 in ASCs (Fig. 3e), indicating that PDGFRβ may differentially regulate the differentiation of tumor cells versus normal cells. These data together suggest that miR-193b targets PDGFRβ to regulate cell proliferation and differentiation in WDLS/ DDLS and ASCs.
We further tested whether blocking PDGFR signaling using small molecules could affect WDLS/DDLS growth. The PDGFRβ inhibitor CP-673451 reduced viability of both DDLS and WDLS cells in a dose-dependent manner (Fig. 3g). To determine whether this decrease in viability involved apoptosis, we used propidium iodide and Annexin V staining. CP-673451 treatment (2.5 μM) caused 50% of DDLS and 30% of WDLS cells to undergo www.nature.com/scientificreports www.nature.com/scientificreports/ apoptosis, and 5 μM CP-673451 induced about 80% apoptosis in both cell types after 3 days of treatment (Fig. 3h). These results indicate that PDGFR signaling is essential for WDLS and DDLS cell survival.

miR-193b targets SMAD4 to regulate adipogenic differentiation. SMAD4 is a central regulator of
TGFβ signaling, and our STRING analysis indicated that it also interacts with YAP1 to regulate Hippo signaling. To determine whether miR-193b directly targets SMAD4, we used luciferase reporters. Addition of the wild-type SMAD4 sequence (WT) to the reporter resulted in a 25% reduction in luciferase activity, indicating that endogenous miR-193b recognizes the SMAD4 3′UTR (Fig. 4a). Compared to non-specific miRNA, transfection of miR-193b inhibited the luciferase activity of the reporter containing the WT SMAD4 3′UTR by 70% (Fig. 4a), confirming direct action of miR-193b on SMAD4 mRNA. Luciferase activity of the reporter containing a mutated version of the seed site for miR-193b was similar to that of the control, luciferase-only reporter, indicating specificity of miR-193b binding. The transfected miR-193b repressed mRNA and protein expression of SMAD4, while addition of anti-miR completely blocked the effects (Fig. 4b,c). Together, these results suggest that SMAD4 is a direct target of miR-193b in WDLS/DDLS cells.
Since miR-193b significantly reduces WDLS/DDLS cell viability, we tested whether inhibition of SMAD4 expression has similar effects. Knockdown of SMAD4 using SMARTpool SMAD4 siRNA had no effect on the viability of DDLS and WDLS cells compared with control siRNA (Fig. S3).
Knockdown of SMAD4 promoted adipogenic differentiation in ASCs as indicated by increased adipogenic marker expression and lipid droplet formation (Fig. 4d,e). This result indicates that, although inhibition of SMAD4 does not affect tumor cell viability (Fig. S3), miR-193b may regulate tumor differentiation by targeting SMAD4. To test this hypothesis, we knocked down SMAD4 before inducing differentiation in WDLS cells. Knockdown of SMAD4 indeed upregulated adipogenic markers in WDLS cells (Fig. 4f). These results indicate that miR-193b promotes adipogenesis of ASCs and regulates differentiation in WDLS/DDLS by directly targeting SMAD4.
To evaluate YAP1 function, we knocked down its expression using siRNA in DDLS and WDLS cells. YAP1 knockdown significantly reduced the viability of both DDLS and WDLS cells (Fig. 5d). YAP1 siRNA also caused apoptosis in about 30% of WDLS cells (Fig. 5e), but only induced modest cell cycle arrest at G0/G1 in DDLS cells (Fig. S4). YAP1 forms a complex with β-catenin to activate downstream gene expression 22 . To examine whether miR-193b regulates Wnt/β-catenin signaling in WDLS/DDLS cells, we employed the TOP-flash/FOP-flash system, in which the TOP-flash reporter indicates Wnt activity, while the FOP-flash reporter (lacking the β-catenin binding site) serves as a negative control. Recombinant human Wnt3a significantly increased TOP-flash activity but had no effects on FOP-flash (Fig. 5f). Both overexpression of miR-193b and knockdown of YAP1 reduced TOP-flash activity by 50% but did not affect that of FOP-flash (Fig. 5f).
Considering the heterogeneity of liposarcoma, we evaluated the function of miR-193b in additional liposarcoma cells (LPS141, RDD8107, WD7785-1). As shown in our previous study, endogenous miR-193b expression in these three cell lines is higher than that in DD8817 and WD4847-2, but much lower than that in ASCs 12 . Consistent with the results in DD8817 and WD4847-2 cells, miR-193b significantly inhibited cell growth in a dose-dependent manner (Fig. S6a) and induced approximately 50% reductions in target gene expression (PDGFRΒ, SMAD4, YAP1) (Fig. S6b). These results suggest that miR-193b functions similarly across multiple liposarcoma cell lines.

Discussion
In this study, we demonstrated that miR-193b regulates multiple signaling pathways, including PDGFR, TGFβ, and Wnt, by targeting the key regulators PDGFRβ, SMAD4, and YAP1 to reduce cell viability and promote adipogenic differentiation in WDLS and ASCs. Based on these results, we showed that inhibitors of PDGFR and Wnt/β-catenin signaling significantly reduce WDLS/DDLS cell viability and induce apoptosis.
Dysregulation of miR-193b is frequently observed in cancers. As a tumor suppressor, miR-193b has been reported to target cyclin D1 to regulate the cell cycle 23 , or directly inhibit oncogenes such as Mcl-1, c-KIT, MYB, and KRAS to attenuate tumorigenesis [24][25][26][27] . Although proteins whose expression is regulated directly by miR-193b have been identified using quantitative iTRAQ analysis in breast cancer cells, there are still many potential targets that have not been validated or identified in other cancers. We recently reported that miR-193b functions as a tumor suppressor in liposarcoma by targeting the FAK-SRC-CRKL axis 12 , and here, we identify three novel targets of miR-193b (PDGFRβ, SMAD4, and YAP1), which respectively play key roles in PDGFR, TGFβ, and Hippo/Wnt signaling.
Overactivity of PDGFR signaling is observed in sarcoma, and may drive tumor growth 28,29 . Consistent with that function, we found that inhibition of PDGFRβ expression reduces DDLS and WDLS cell viability. We also observed that, similar to SMAD4, knockdown of PDGFRβ promotes adipogenesis in both ASCs and WDLS cells, which is consistent with the function of PDGFRβ in adipogenic commitment of MSCs. Thus, unlike TGFβ/ SMAD4 signaling, PDGFR signaling may regulate tumorigenesis by controlling both cell viability and differentiation in WDLS/DDLS. The silencing of miR-193b that occurs in DDLS or a subset of WDLS could thus break multiple control mechanisms to keep liposarcoma cells from differentiating, and thereafter drive liposarcoma cell proliferation to promote tumor progression.
SMAD4 is a central mediator of TGFβ and bone morphogenetic protein (BMP) signaling pathways, which regulate differentiation, among other physiological processes. SMAD4 acts as a tumor suppressor in carcinomas of the pancreas and GI tract, where deletions or inactivating mutations in the SMAD4 gene have been reported [30][31][32] . However, SMAD4 was recently reported to exert a tumor-promoting role in hepatocellular carcinoma 33 . In WDLS/DDLS cells, knockdown of SMAD4 did not affect cell viability, but did promote adipogenic differentiation in ASCs and WDLS cells. This finding is consistent with recent studies showing that SMAD4 is essential for mouse adipogenesis 34 . Instead of affecting tumor growth, increased levels of SMAD4, as would result from the under-expression of miR-193b in DDLS, may contribute to liposarcoma progression and deregulation of differentiation in WDLS/DDLS.
Besides directly targeting key upstream regulators of the above signaling pathways, miR-193b also indirectly inhibits Wnt/β-catenin signaling by repressing the Hippo effector YAP1. YAP1 is reported to be essential for Wnt/β-catenin signaling activation in some cancer types 22,35,36 . Here, we demonstrated that miR-193b inhibits Wnt/β-catenin signaling activity by targeting YAP1 in WDLS/DDLS cells. (2019) 9:3197 | https://doi.org/10.1038/s41598-019-39560-0 www.nature.com/scientificreports www.nature.com/scientificreports/ STRING analysis revealed protein-protein interactions among the identified miR-193b targets (Fig. 1d), suggesting crosstalk of the related signaling pathways, as has been suggested by other studies. PDGFR is known to activate FAK to promote migration 37 , and signaling downstream of both kinases converges on MEK/ERK to www.nature.com/scientificreports www.nature.com/scientificreports/ support proliferation 38 . Similarly, cooperation of Hippo (YAP1) and TGFβ (SMADs) signaling has been found to promote carcinogenesis 39,40 . Not only does YAP1 regulate Wnt signaling, SMAD4 also controls proteasomal degradation of β-catenin 41 . Thus, miR-193b may dampen signaling via multiple oncogenic pathways, suggesting that combinations of inhibitors of FAK, PDGFR, TGFβ, and/or YAP1 may be more effective than single drugs. Given the challenge of identifying combinations that are safe and effective, administration of miR-193b itself (or a pharmacologically optimized version thereof) may represent a promising alternative, as it would function similarly to a combination of inhibitors.
We showed that the specific PDGFRβ inhibitor CP-673451 reduces viability and induces apoptosis in the tested WDLS/DDLS cells. However, the limited activity of similar agents in the clinic suggests that such drugs may only be effective for a subset of patients with these liposarcomas [42][43][44][45] . Responses of DDLS patients to pazopanib (which inhibits PDGFR, c-KIT, FGFR, and VEGFR) were better than those in myxoid liposarcoma, suggesting that the contribution of PDGFR signaling to DDLS progression is clinically relevant 44 . As we selected cell lines with low miR-193b expression, PDGFR or multi-kinase inhibitors may be especially effective in this subset of WDLS/DDLS. Studies in patient-derived orthotopic xenograft (PDX) models would provide further evidence in support of this idea; this approach has been used to show that PDGFRα-amplified pleomorphic liposarcoma is especially sensitive to pazopanib 46 .
In summary, miR-193b functions as a tumor suppressor in WDLS/DDLS. Loss of miR-193b expression appears to attenuate inhibitory control of multiple oncogenic signaling pathways (e.g. PDGFR, TGF, and Wnt) through the key regulators PDGFR, SMAD4, and YAP1. For the first time, oncogenic functions of these regulators are confirmed in liposarcoma cells. Restoring miR-193b activity or inhibition of PDGFRβ or Wnt/β-catenin signaling could represent a new therapeutic strategy in liposarcoma.

Methods
Patient samples. The study was approved by the MSKCC Institutional Review Board, and all participants gave written informed consent that their tissue samples could be used for this research, in accordance with HHS guidelines and the Declaration of Helsinki. Tumor and normal adipose tissue samples (all retroperitoneal) obtained during surgical resection were snap-frozen in liquid nitrogen and embedded in cryomolds. According to MSKCC's two-grade system, all WDLS tumors were low-grade and all DDLS tumors were high-grade. All tumors were primary except for DD1348, which was a local recurrence.
Plasmid construction. The pmirGLO plasmid (Promega, Madison, WI) was used for PDGFRβ, SMAD4, and YAP1 3′UTR reporters. Synthetic ~100-bp oligonucleotides representing regions of the SMAD4, PDGFRβ and YAP1 3′ UTRs containing miR-193b seed motifs were inserted into the Pmel/Xbal site of pmirGLO. Mutations of the same seed sequences in reporters were also generated. The oligonucleotide sequences are shown in Table S1. The TOPflash/FOPflash reporter plasmid system was applied for monitoring β-catenin-driven Wnt transcriptional activity. The TOPFlash reporter containing TCF binding sites is activated by β-catenin, and the FOPFlash with mutated TCF binding sites serves as a negative control. Both TOPflash and FOPflash were purchased from Upstate Biotechnology (NY).

Transient transfection and 3′UTR luciferase reporter assays. MicroRNAs and corresponding
inhibitors (Ambion) were transfected into 50% confluent cells with Oligofectamine at a concentration of 50 nM (Invitrogen). Protein and mRNA were collected at 72 h after transfection.
For 3′UTR luciferase reporter assays, miRNAs or siRNAs were co-transfected with 200 ng of reporters using Lipofectamine 2000 (Invitrogen). At 48 h post-transfection, cells were collected for luciferase assays. For TOP-/ FOP-Flash assays, reporters were transfected 24 h after miRNA or siRNA and cells were collected for measurement 24 h later. phRL-null Renilla luciferase plasmid was co-transfected for normalization. The luciferase activities were detected by using the dual luciferase reporter assay system according to the instructions from Promega.
Cell proliferation and cell cycle analysis. Cell proliferation was evaluated by using the CellTiter-Glo cell viability assay (Promega) following the manufacturer's instructions. Briefly, 1500 cells/well were plated in 96-well plates in the presence of miRNAs, siRNA, or inhibitors. The plates were then incubated for 3-7 days, then 100 μL www.nature.com/scientificreports www.nature.com/scientificreports/ of CellTiter-Glo reagent was added to lyse the cells. After a 10-min incubation at room temperature, luminescence was recorded in a luminometer with an integration time of 1 s per well.
For cell cycle analysis, transfected cells were harvested at 48 h and fixed at 4 °C. Propidium iodide (50 μg/mL) was used to stain the fixed cells. DNA content was analyzed using a FACSCalibur instrument (Becton Dickinson Bioscience). Cell cycle fractions were quantified with Multicycle Software (Phoenix Flow Systems) and analyzed by FlowJo software.
Annexin V assay. Apoptosis was evaluated using a Muse Annexin V and Dead Cell kit (EMD Millipore).
Cells were transfected with miRNAs or siRNAs or treated with inhibitors. After 72 h, cells were collected in 1% FBS medium, mixed with Muse Annexin V and Dead Cell Reagent, and analyzed using a Muse Cell Analyzer (EMD Millipore). Induction of adipogenic differentiation. As previously described 12 , confluent cells were cultured in differentiation-initiating medium (regular growth medium plus 100 nM insulin, 1 µM dexamethasone, 250 µM 3-isobutyl-1-methylxanthine (IBMX), 33 µM biotin, 17 µM pantothenic acid, and 5 µM of the PPARG agonist rosiglitazone). After 4 days, media were changed to maintenance medium (initiating medium lacking IBMX and rosiglitazone). Cells were fed with maintenance medium every 4 days thereafter.
Oil red O staining. After 10 days of adipogenic differentiation, lipid droplets were stained with Oil Red O as described 12 . Stained cells were quantified, and three biological replicates (approximately 400-500 cells) were analyzed for each treatment group.
RNA isolation and analysis. Total RNA was collected using TRIzol reagent (Invitrogen) and isolated by Direct-zol RNA mini prep kits (Zymo Research). cDNA was synthesized by the qScript DNA synthesis kit (Quanta Bioscience) or the QuantiMir RT kit (System Biosciences) according to the manufacturer's instructions. TaqMan gene expression assays (Life Technology) were used for relative gene expression. miRNA expression levels were detected by using SYBR Green miRNA-specific primers. Quantitative real-time PCR (qRT-PCR) was performed on the ABI Prism 7900HT Sequence Detection System (Applied Biosystems). All transcript levels were normalized to levels of GAPDH transcript (for regular gene expression) or U6 snRNA (for miRNA expression). All primers are listed in Table S1.
Immunoblotting. Cells were lysed in RIPA buffer (with 100x protease inhibitor cocktail and 25 μM MG132) and protein concentration was determined by the Lowry method. Equal amounts of total protein were isolated by SDS-PAGE and then transferred to PDVF membranes. After blocked, membranes were incubated with the appropriate primary antibodies, followed by HRP-conjugated secondary antibodies. Immunoreactive proteins were detected using Western Lightning chemiluminescence reagent. miRNA target prediction and network construction. We previously reported our methods of identifying 50 putative miR-193b targets 12 . Functional interactions of the 50 previously identified putative miR-193b targets were analyzed by STRING (https://string-db.org/). STRING interactions with a confidence score of 0.4 or higher are shown. Interactions were inferred from the STRING data mining and experimental databases. Statistical analysis. All data are presented as means ± S.E. Statistical significance was assessed using Student's t test, and p values < 0.05 were considered statistically significant.