NT5DC2 promotes leiomyosarcoma tumour cell growth via stabilizing unpalmitoylated TEAD4 and generating a positive feedback loop

Abstract 5′‐Nucleotidase Domain Containing 2 (NT5DC2) is a novel oncoprotein, the regulatory effects of which have not been well characterized. This study aimed to investigate the expression profile and functional regulation of NT5DC2 and its potential interplay with TEAD4 in leiomyosarcoma (LMS). Bioinformatic analysis was conducted using data from The Cancer Genome Atlas (TCGA) and Genotype‐Tissue Expression (GTEx) program. LMS cell lines SK‐LMS‐1 and SK‐UT‐1 were used for both in vitro and in vivo analysis. Results showed that NT5DC2 is aberrantly upregulated in LMS. Its overexpression was associated with unfavourable survival. Deletion of NT5DC2 significantly reduced the expression of cyclin B1, cyclin A2, cyclin E1 and CDK1 and increased G1 phase arrest in LMS cell lines, and suppressed their proliferation both in vitro and in vivo. NT5DC2 interacted with unpalmitoylated TEAD4, and this association reduced TEAD4 degradation via the ubiquitin‐proteasome pathway. TRIM27 is a novel E3 ubiquitin ligase that induces K27/48‐linked ubiquitination of unpalmitoylated TEAD4 at Lys278. TEAD4 inhibition significantly suppressed LMS cell growth both in vitro and in vivo. Dual‐luciferase assay demonstrated that TEAD4 could bind to the NT5DC2 promoter and activate its transcription. Based on these findings, we infer that the NT5DC2‐TEAD4 positive feedback loop plays an important role in LMS development and might serve as a potential therapeutic target.

5′-Nucleotidase Domain Containing 2 (NT5DC2) is a protein with potential metal ion binding and 5′-nucleotidase activity, according to Gene Ontology (GO) annotations. However, the exact molecular function is still not quite clear. Some recent studies revealed that this protein might have an important role in cancer biology. It enhances glioma stem-like cell tumorsphere formation and cell viability in vitro and promotes tumorigenesis in vivo. 3 Mechanistically, it binds to Fyn, an Src family proto-oncogene, and suppresses its degradation. 3 Its overexpression leads to increased proliferation and reduced apoptosis of non-small-cell lung cancer cells, via reducing p53 expression. 4 It promotes the proliferation and colony formation of hepatocellular carcinoma (HCC) in vitro and facilitates tumour growth in vivo. 5 Its physical interaction with the epidermal growth factor receptor (EGFR) directly reduces EGFR ubiquitination and degradation, 5 which is a high potential therapeutic target in multiple cancers. 6 These findings imply that NT5DC2 might exert critical oncogenic effects via acting as an oncoprotein stabilizer.
The TEAD family of transcription factors (TEAD1, TEAD2, TEAD3 and TEAD4) are the terminal nuclear effectors of the Hippo pathway, 7 which is usually overactivated in sarcoma. 8 TEAD activation plays critical roles in cancer biology, including epithelialto-mesenchymal transition (EMT), proliferation, metastasis, chemoresistance and cancer stem cell properties via their transcriptional target genes. 7 Recent studies revealed that post-translational S-palmitoylation of TEAD is important for its stability and activity. 9,10 After depalmitoylation by depalmitoylases (such as APT2 and ABHD17A), unpalmitoylated TEAD4 can be degraded via the ubiquitin-mediated proteasome pathway. 11 However, the regulatory mechanisms that dynamically control TEAD4 degradation and their functional role in LMS have not yet been identified. In this study, we aimed to investigate the expression profile and functional regulation of NT5DC2 and its potential interplay with unpalmitoylated TEAD4 in LMS.

| Data retrieved from The Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression (GTEx) program
Data extraction from TCGA and GTEx was performed using the UCSC Xena Browser, 12 as we described previously. 13  The integrated pathway level (IPL) for TEAD4, which was calculated by the PARADIGM (PAthway Recognition Algorithm using Data Integration on Genomic Models) 14 in 80 LMS cases was extracted.
In brief, this algorithm infers IPL scores for genes, complexes and processes by integrating diverse high-throughput genomics information with known signalling pathways from a single cell line or patient sample. 14

| Immunohistochemistry (IHC)
NT5DC2 IHC in normal smooth muscle and HCC tissues were reviewed using data provided in the Human Protein Atlas, 15 in which NT5DC2 was stained by HPA050683 (Sigma-Aldrich).
LMS tissue array was purchased from Taibsbio. IHC staining of NT5DC2 was performed using anti-NT5DC2 (HPA050683, Sigma-Aldrich), according to the method introduced previously. 16
The cell line was authenticated by matching the short-tandem repeat (STR) DNA profiles according to the provider. Lentiviral shRNA targeting NT5DC2 and TEAD4 were constructed by HanBio Technology, with the shuttle plasmid pHBLV-U6-Puro. shRNA sequences used are provided in Table S1.
The genes encoding ubiquitin and its mutants with all ly-

| Reverse transcription-quantitative PCR (RT-qPCR)
Total RNA was extracted using RNeasy Mini Kit (QIAGEN). Then, the first-strand cDNA was synthesized using a PrimeScript™ RT Kit (TaKaRa). qPCR reactions were conducted using an ABI 7900 HT thermocycler (Applied Biosystems), with SYBR Green Master Mix (Applied Biosystems). The threshold cycle (Ct value) of each sample was recorded and used to calculate relative gene expression using the 2 −ΔΔCt method. The primers used are provided in Table S1.
GAPDH mRNA expression was used as an internal control.

| Western blotting and co-immunoprecipitation (co-IP)
Total cellular protein was extracted using RIPA lysis buffer with protease inhibitor. 30 μg of the total protein was loaded to each lane, separated using 10% SDS-PAGE, and electrotrans-

| CCK-8 assay
CCK-8 kit was used to detect cell viability according to the manufacturer's instruction. 5000 cells with or without lentiviral mediated gene suppression were seeded into each well of a 96-well plate. Cell viability was measured at 24, 48, 72 and 96 hours after seeding.
Absorbance at 450 nm was measured and recorded.

| Colony formation assay
Colony formation was performed as previously described. Briefly, 1000 cells were seeded into a six-well plate and cultured for two weeks. The cells were then fixed and stained with 0.1% crystal violet for 15 minutes. Image J software was used to quantify the colonies.

| Flow cytometric assay
Cells infected with lentiviral shNT5DC2 were harvested 48 hours after infection. Then, cells were fixed in 70% ethanol for 1 hour and then stained with propidium iodide (PI) solution containing RNase A (Sigma-Aldrich) for 30 minutes at room temperature in the dark.
Cell-cycle distribution was then analysed by the Cellometer Vision CBA image cytometer (Nexcelom).

| Immunofluorescent staining
SK-LMS-1 cells grown on coverslips were rinsed, fixed with PBS containing 4% paraformaldehyde for 10 minutes at room temperature and then permeabilized with PBS containing 0.1% Triton X-100 for 10 minutes. After washing with PBS, cells were blocked for 30 minutes with 1% BSA in PBS at room temperature and then were incubated with primary antibody, including mouse Flag tag (66008-3-IG, Proteintech) and rabbit anti-Myc tag (ab9106 Abcam) or rabbit anti-TRIM27 (12205-1-AP, Proteintech) overnight at 4℃. After that, the cells were washed and incubated with DAPI (Beyotime) for 5 minutes. Then, the coverslips were sealed. Fluorescence images were captured using an AxioImager Z1 ApoTome microscope system (Carl Zeiss).

| High NT5DC2 expression is associated with poor survival of LMS
To characterize NT5DC2 expression, we compared RNA-seq data from three representative soft muscle tissues (colon, small intestine, and vagina) from GTEx and LMS tissue from TCGA-SARC. Results showed that LMS group had significantly higher NT5DC2 expression than the three normal groups (P < .001, Figure 1A). TCGA Research Network indicated that ULMS and STLMS had significantly different mRNA expression signatures. 1 However, no significant difference was observed in NT5DC2 expression between the two subgroups (P = .29, Figure 1B). Then, we performed K-M survival analysis in all LMS cases and subgroups, respectively. Results showed that LMS cases with high NT5DC2 expression had significantly worse PFS, DSS and OS than the counterparts with low NT5DC2 expression ( Figure 1C,D). Subgroup analysis confirmed the consistent trends in ULMS and STLMS ( Figure 1E-H). By retrieving IHC staining data in the HPA, we found normal smooth muscle tissue has low expression of NT5DC2 protein ( Figure 1I, up). NT5DC2 expression was confirmed in HCC ( Figure 1I, down), as reported previously. 5 Using the same antibody in the HPA, we confirmed that both ULMS and STLMS cases had NT5DC2 expression at the protein level, mainly in the cytoplasmic part of tumour cells ( Figure 1J).

| NT5DC2 promotes LMS cell-cycle progression and proliferation in vitro and tumour growth in vivo
Previous studies imply that NT5DC2 enhances cell proliferation of multiple cancers. 5,17 Therefore, we hypothesized that NT5DC2 might also play an important role in LMS cell proliferation. By checking the correlation between NT5DC2 and multiple cell-cycle related genes, we found strong correlations (R ≥ .6) with CCNB1 and CCNA2 and moderate correlations (.6 > R ≥ .4) with CCNE1 and CDK1 (Figure 2A).
To validate the potential cell-cycle regulation of NT5DC2, SK-LMS-1 and SK-UT-1 cells were transiently infected with lentiviral-shNT5DC2 ( Figure 2B,C). NT5DC2 inhibition significantly decreased the expression of cyclin B1, cyclin A2, cyclin E1 and CDK1 at the protein level in the two cell lines ( Figure 2D). Flow cytometric assay showed that after NT5DC2 knockdown, both SK-LMS-1 and SK-UT-1 cells had increased G0/G1 arrest and reduced cells in the S phase ( Figure 2E).

| NT5DC2 interacts with unpalmitoylated TEAD4 and reduces its ubiquitin-mediated degradation
Using RNA-seq data in TCGA-LMS (N = 80), we observed a moderate positive correlation (Pearson's R = .40) between NT5DC2 and TEAD4 expression ( Figure 3A). If we added the normal smooth muscle cases from colon, small intestine and vagina in GTEx, the correlation R-value drastically increased to .80 ( Figure 3B). Like NT5DC2, TEAD4 was also significantly upregulated in LMS cases, compared with the normal smooth muscle groups ( Figure 3C). Under the best cut-off model, high TEAD4 expression was associated with worse PFS ( Figure 3D).
After NT5DC2 inhibition, both SK-LMS-1 and SK-UT-1 cells had downregulated TEAD4 expression at the mRNA and protein levels ( Figure S1A,B and Figure 3E,F). One previous study reported that depalmitoylation of the TEAD4 protein triggers degradation via the ubiquitin-proteasome pathway. 11 Since NT5DC2 might act as an inhibitor of ubiquitin-mediated degradation, 3,5 we explored whether NT5DC2 inhibits the ubiquitination and degradation of unpalmitoylated TEAD4. Co-IP assay confirmed that endogenous NT5DC2 was co-immunoprecipitated by anti-TEAD4 in SK-LMS-1 and SK-UT-1 cells ( Figure 3G). Cycloheximide pulse-chase assay was performed in SK-LMS-1 cells ( Figure 3H). Silencing of NT5DC2 decreased the stability of unpalmitoylated TEAD4, while NT5DC2 overexpression increased the half-life of the protein ( Figure 3H).
To validate the interaction between NT5DC2 and unpalmitoylated TEAD4, we co-expressed myc-NT5DC2 and Flag-mtTEAD4 in SK-LMS-1 cells. IF staining indicated that mtTEAD4 had both nuclear and cytoplasm distribution. NT5DC2 showed co-localization with mtTEAD4 in the cytoplasmic part ( Figure 3I). Co-IP assay confirmed the interaction between NT5DC2 and mtTEAD4 ( Figure 3J

| TRIM27 induces K27/48-linked ubiquitination of unpalmitoylated TEAD4 at Lys278
Although ubiquitination mediated TEAD4 degradation has been characterized, E3 ubiquitin ligase is a large class with hundreds of members, the specific E3 ligases linked to TEAD4 have not been fully understood. One previous study reported that STUB1/CHIP is a key enzyme related to this process. 11 By checking TEAD4 interacting proteins in BioGRID, we found two E3 ubiquitin ligases, TRIM27 and TRIM54 might interact with TEAD4. TRIM27 and STUB1, but not TRIM54 overexpression, significantly reduced mtTEAD4 expression at the protein level but not the mRNA level in SK-LMS-1 cells (Figure S1C-I and Figure 4A), suggesting that TRIM27 might also participate in the degradation of unpalmitoylated TEAD4.
Cycloheximide pulse-chase assay showed that TRIM27 overexpression significantly decreased the half-life of mtTEAD4 ( Figure 4B,C).
IF assay indicated co-localization of TRIM27 and mtTEAD4 in the nuclear part of SK-LMS-1 cells ( Figure 4D). Co-IP experiment confirmed the interaction between TRIM27 and mtTEAD4 ( Figure 4E,F).
Seven potential ubiquitination sites at lysine residues were observed in the TEAD4 protein ( Figure S2). To identify the site of ubiquitination, we subsequently constructed TEAD4 mutants in which the lysine residues were replaced with arginine. Results indicated that the mutant with K278R had significantly reduced polyubiquitination compared to other mutants ( Figure 4H). In addition, TRIM27-mediated mtTEAD4 polyubiquitination could be detected in the presence of K27-Ub or K48-Ub, but not with other Ub mutants ( Figure 4I). shRNA or scrambled control. F, SK-LMS-1 and SK-UT-1 cells infected with NT5DC2 shRNA or scramble control were treated with MG132 (10 μmol/L, 6 h). The expression levels of TEAD4 and NT5DC2 were assessed by western blot assay. G, Co-IP assays were performed to check the interaction between endogenous NT5DC2 and TEAD4 in SK-LMS-1 (left) and SK-UT-1 (right) cells. IgG was used as a control. H, Cycloheximide pulse-chase assay was performed in SK-LMS-1 cells with mtTEAD4 overexpression alone or in combination with NT5DC2 knockdown or overexpression. 36 h after lentiviral infection, cells were treated with 10 μmol/L CHX for the indicated time, followed by western blot analysis. I, Co-localization of NT5DC2 (red) and mtTEAD4 (green) in SK-LMS-1 cells, by immunofluorescent staining. J and K, SK-LMS-1 cells were coinfected with Flag-mtTEAD4 and myc-NT5DC2 expression lentiviruses. The interaction between NT5DC2 and mtTEAD4 complexes were co-immunoprecipitated with anti-Flag (J) or anti-Myc (K) antibodies. L, SK-LMS-1 cells were coinfected with the indicated lentiviruses (Flag-mtTEAD4 and Myc-NT5DC2 or shNT5DC2) for 36 h, followed by treatment with MG132 (10 μmol/L, 6 h). Then, cell lysates were immunoprecipitated with an anti-Flag antibody. Ubiquitinated mtTEAD4 was detected by western blotting with an anti-HA antibody. *P <.05, **P <.01, ***P <.001 F I G U R E 4 TRIM27 induces K27/48-linked ubiquitination of unpalmitoylated TEAD4 at Lys278. A, SK-LMS-1 cells were coinfected with the indicated lentiviruses for 48 h, for Flag-mtTEAD4 expression alone or in combination with myc-TRIM27, myc-TRIM54 or myc-STUB1. Then, western blot analysis was conducted to analyse the expression of total TEAD4 and Flag-mtTEAD4. B and C, Cycloheximide pulsechase assay was performed in SK-LMS-1 cells with mtTEAD4 overexpression alone or in combination with TRIM27 overexpression. 36 h after lentiviral infection, cells were treated with 10 μmol/L CHX for the indicated time, followed by western blot analysis. The indicated proteins were analysed (B), and the relative mtTEAD4 protein level is illustrated graphically (C). D, Co-localization of TRIM27 (red) and Flag-mtTEAD4 (green) in SK-LMS-1 cells, by immunofluorescent staining. E and F, SK-LMS-1 cells were coinfected with Flag-mtTEAD4 and Myc-TRIM27 expression lentiviruses. The interaction between NT5DC2 and mtTEAD4 complexes were co-immunoprecipitated with anti-Flag (E) or anti-Myc (F) antibodies. G, SK-LMS-1 cells were coinfected with the indicated lentiviruses (Flag-mtTEAD4 and Myc-TRIM27 or shTRIM27) for 36 h, followed by treatment with MG132 (10 μmol/L, 6 h). Then, cell lysates were immunoprecipitated with an anti-Flag antibody.

| TEAD4 inhibition suppresses LMS cell proliferation in vitro and tumour growth in vivo
Then, we studied the functional role of TEAD4 in LMS cell proliferation. Both SK-LMS-1 and SK-UT-1 cells were transiently infected with lentiviral-shTEAD4 ( Figure 5A-D). TEAD4 inhibition remarkably decreased the proliferation and colony-forming ability of the two cell lines in vitro ( Figure 5E-G). Subcutaneous xenograft tumour models confirmed that knocking down of TEAD4 substantially reduced tumour growth in vivo ( Figure 5H-I). These findings demonstrated that TEAD4 might also act as a carcinogenetic driver in LMS.

| TEAD4 transcriptionally activates NT5DC2 expression in LMS
Interestingly, by suppressing TEAD4 expression, we also observed downregulated NT5DC2 expression at the protein level ( Figure 5C).
These findings suggested that the activation of the TEAD4 downstream pathway is common in LMS and might contribute to disease progression.
We performed bioinformatic screening of the possible down- By setting TCGA-SARC as a whole group, we observed that high NT5DC2 expression was generally associated with significantly shorter PFS and DSS ( Figure S4A,C). In comparison, high TEAD4 expression was generally linked to significantly shorter PFS ( Figure S4B), but not DSS ( Figure S4D). However, subgroup analysis in different sarcoma subtypes (with over 30 cases) only found an association between high NT5DC2 expression and unfavourable PFS in undifferentiated pleomorphic sarcoma/ myxofibrosarcoma (UPS/MFS) cases ( Figure S4I). No significant association was observed in other subgroup analyses ( Figure S4E-H, J-L). These findings imply that the tumour-promoting effects of NT5DC2 and TEAD4 might be quite specific in different sarcoma subtypes.

| D ISCUSS I ON
In this study, we found that NT5DC2 is aberrantly upregulated in LMS, and its overexpression was associated with unfavourable survival, suggesting that it might serve as a prognostic biomarker in LMS. Several recent studies revealed that NT5DC2 upregulation might enhance the malignant phenotypes of some tumours, such as facilitated cell-cycle progression, proliferation, EMT, angiogenesis, metastasis and cancer stem cell properties. 3,4,17 Bioinformatic analysis in this study implied that NT5DC2 expression was strongly correlated with multiple cell-cycle related genes in TCGA-LMS cases.
Findings in this study showed that deletion of NT5DC2 significantly reduced the expression of cyclin B1, cyclin A2, cyclin E1 and CDK1 and increased G1 phase arrest in LMS cell lines. Subsequent cellular studies confirmed that inhibiting endogenous NT5DC2 reduced LMS cell proliferation both in vitro and in vivo.
Functionally, NT5DC2 can act as a stabilizer of some oncoproteins by reducing ubiquitin-mediated degradation, including fyn in glioma 3 and EGFR in HCC. 5 In this study, we confirmed the physical interaction between NT5DC2 and unpalmitoylated TEAD4 and this association reduced TEAD4 degradation via the ubiquitinproteasome pathway. Elevated TEAD4 expression was observed in multiple tumour tissues. Its expression was highly correlated with clinicopathological parameters and worse prognosis. [20][21][22] In sarcoma, TAZ and YAP, two transcription co-activators, can bind to TEAD and enhance the oncogenic effect of the complex. 23 TEAD-YAP/TAZ complex has been demonstrated as a powerful enhancer of tumour cell proliferation, due to its transcriptional effects on the expression of multiple tumour-promoting genes. 19,24 Although palmitoylation is dispensable for the binding of TEAD4 with YAP/TAZ, it is important for TEAD4 stability. 11,25 After depalmitoylation, unpalmitoylated TEAD4 can be degraded via the ubiquitin-proteasome pathway. 11 STUB1 has been demonstrated as an E3 ubiquitin ligase involved in this process. 11 However, the type of polyubiquitination and the ubiquitination sites at lysine residues TEAD4 protein have not been identified. In this study, we further demonstrated that TRIM27 is a novel E3 ubiquitin ligase that induces K27/48-linked ubiquitination of unpalmitoylated TEAD4 at Lys278. TAZ and YAP are highly expressed and constitutively activated in SK-LMS-1 cells. 23 It was showed previously that the TEAD4-YAP/ TAZ complex can be disrupted pharmacologically by verteporfin, which is a heme analogue. 26  and NT5DC2 on FOXM1 transcription. This finding suggested that the transcriptional regulation of TEAD4 is tissue/tumour-specific.
By applying dual-luciferase assay, we demonstrated that TEAD4 could bind to the NT5DC2 promoter and activate its transcription.
Therefore, TEAD4 might act as one of the dominant oncoproteins in LMS.
This study also has some limitations. Besides ubiquitin-mediated degradation, we also observed that NT5DC2 inhibition reduced TEAD4 expression at the mRNA level. However, the underlying molecular mechanism was not identified in the current study. Although we confirmed a positive feedback loop between NT5DC2 and TEAD4, their downstream effectors in LMS were not well characterized. These issues are supposed to be answered in our future studies.
In conclusion, this study revealed a novel NT5DC2-TEAD4 positive feedback loop, in which TEAD4 binds to the NT5DC2 promoter and activates its transcription, while NT5DC2 has a physical interaction with unpalmitoylated TEAD4, reducing its ubiquitin-mediated degradation. This feedback loop helps to explain the highly activated TEAD4 downstream signalling pathways in LMS and might serve as a potential therapeutic target.

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data used in this study were included in the manuscript and Supporting Information.