Loss of SMARCB1 promotes autophagy and facilitates tumour progression in chordoma by transcriptionally activating ATG5

Abstract Objectives SWI/SNF‐related matrix‐associated actin‐dependent regulator of chromatin subfamily B member 1 (SMARCB1) loss is associated with a poor prognosis in chordoma, while the mechanism remains largely unclear. Here, we aim to explore the function and regulatory mechanisms of SMARCB1 in chordoma. Materials and Methods The effect of SMARCB1 on chordoma cells was investigated in vitro and in vivo. Chromatin immunoprecipitation (ChIP) sequencing was used to investigate the mechanisms of SMARCB1 in chordoma. The association between SMARCB1 and autophagy was validated by Western blot, immunofluorescence and transmission electron microscopy. In addition, the ATG5 expression in chordoma tissue was assessed using immunohistochemistry and correlated with patient survival. Results SMARCB1 inhibited the malignant phenotype of chordoma cells in vitro and in vivo, supporting a tumour suppressor role of SMARCB1 in chordoma. ATG5‐mediated autophagy was identified as a potential downstream pathway of SMARCB1. Mechanistically, SMARCB1 bound directly to the ATG5 promoter and epigenetically inhibited its transcription, which decreased ATG5 expression and impaired autophagy. Additionally, autophagy inhibitor chloroquine had a potential anti‐cancer effect on chordoma cells in vitro. Moreover, high ATG5 expression was observed in recurrent chordoma patients, which independently correlated with adverse outcomes. Conclusions Taken together, our results revealed that the SMARCB1/ATG5 axis is a promising therapeutic target for chordoma and autophagy inhibitors may be effective agents for chordoma treatment.

SMARCB1, a potential tumour suppressor gene located at chromosome 22q11.2, is a core component of the SWI/SNF complex that plays a prominent role in several cellular biological behaviours and the development of cancers. [9][10][11] Numerous studies have revealed decreases in SMARCB1 expression or inactivation of SMARCB1 in various cancers, such as malignant rhabdoid tumours, 12 familial schwannomatosis 13 and epithelioid sarcomas. 14 However, a recent study has indicated a potentially oncogenic role of SMARCB1 in liver cancer. 15 In chordoma, our previous study and other studies showed that the low expression or loss of SMARCB1 correlate with dismal survival. [16][17][18][19] Moreover, recent studies proposed chordoma with SMARCB1 loss as a potential novel subset of chordoma, poorly differentiated chordoma, which displays aggressive behaviour and a dismal prognosis. 17,18,20 However, the biological function and underlying mechanism of SMARCB1 in chordoma remain to be elucidated.
Autophagy is an evolutionarily conserved, cellular self-protection process that is essential to maintain intracellular homeostasis via the degradation of cellular materials. 21,22 Increasing evidence demonstrates that autophagy participates in tumour progression and drug resistance of tumours, which suggests the therapeutic value of targeting autophagy. [23][24][25] As an essential autophagy-related gene (ATG), ATG5 plays a vital role in autophagosome formation and correlates with poor survival of several cancers. [26][27][28] However, to date, the role of autophagy in chordoma remains largely unclear. Additionally, no previous study has investigated the expression, prognostic role or potential upstream regulators of ATG5 in chordoma.
Here, we examined the role and potential regulatory mechanism of SMARCB1 in chordoma. In knockdown and overexpression experiments, we found that SMARCB1 inhibited cell proliferation and invasion by regulating autophagy. Mechanistically, SMARCB1 bound directly to the promoter region of ATG5 and epigenetically inhibited transcription of ATG5, which impaired autophagy. Our results suggest the SMARCB1-ATG5-autophagy axis as a possible novel therapeutic target in chordoma.

| Cell culture
The human chordoma cell line UM-Chor1 was purchased from the American Type Culture Collection (ATCC), and MUG-Chor1 cells were kindly provided by the Chordoma Foundation. Cells were cultured in a mixture of IMDM (ATCC) and RPMI-1640 (ATCC) culture media (4:1) supplemented with 10% foetal bovine serum and 1% antibiotics in a 5% CO2 incubator at 37°C.

| RNA extraction and qRT-PCR
Total RNA was isolated from chordoma cells using TRIzol reagent (Invitrogen), and the SuperScript III First Strand Synthesis System (Invitrogen, USA) was used to synthesize cDNA. qRT-PCR was conducted on a QuantStudio 5 (Applied Biosystems) in triplicate. Target gene expression was normalized to GAPDH expression and shown as 2 −∆∆CT . Primer sequences are provided in Table S2.

| Cell counting kit-8 (CCK-8) and colony formation assays
Cell viability was evaluated using CCK-8 (Dojindo, Japan). Briefly, 2 × 10 3 UM-Chor1 cells per well and 6 × 10 3 MUG-Chor1 cells per well were seeded in a 96-well plate. CCK-8 was added at 0, 1, 2 and 3 days, and absorbance at 450 nm was then measured. For colony formation assays, 2 × 10 3 transfected chordoma cells were seeded in a six-well plate and incubated for 14 days. Colonies were then fixed with 4% paraformaldehyde, stained with crystal violet and and bedtools software were used for peak calling and identification of different binding peaks. Function annotation of peak-related genes was then conducted. ChIP-qPCR was applied to confirm the ChIP sequencing results. The primers of ChIP-qPCR are listed in Table S2.

| Electrophoretic mobility shift assay
Reaction mixtures that contained biotin-labelled ATG5 promoter probes (GeneCreate Biological Engineering Co., Ltd, Wuhan, China) and/or nuclear proteins of UM-Chor1 cells were incubated with or without the anti-SMARCB1 antibody (ab12167, Abcam) for 30 min according to the manufacturer's instructions. Then, the samples were subjected to 6% PAGE (80 V, 50 min), transferred to nylon membranes and finally visualized. The detailed probe information of the ATG5 promoter is shown in Table S3.

| Transmission electron microscopy
After collection using a cell scraper and centrifugation at 193 g for 5 min, chordoma cells were fixed with 2% glutaraldehyde and 1% osmium tetroxide, dehydrated in graded ethanol solutions and followed by embedment in epikote. Then, 50 nm ultrathin sections were constructed, stained with uranyl acetate and lead citrate and finally visualized under a transmission electron microscope (H-7650; Hitachi, Japan).

| Tumour specimens, immunohistochemistry (IHC) and analysis
A total of 84 histopathologically diagnosed skull base chordoma specimens from Beijing Tiantan Hospital between January 2008 and November 2010 were applied to IHC analysis. Additionally, 10 paired primary and recurrent specimens were also collected. The clinical information (age, classified by 55 years; gender; tumour volume, classified by 20 cm 3 ; blood supply, abundant (tumour resection surface with a tendency to bleed and is hard to aspirate clearly), poor (limited bleeding that is easy to aspirate clearly) and moderate (between poor and abundant); texture, classified as hard (tumours could hardly be excised without scissor or punch forceps), soft (tumours could be easily suctioned when using the aspirator) and moderate (between hard and soft); etc.) of the 84 patients was reviewed as our previous study reported and is detailed in Table S4. 4,29,30 IHC was performed using the Leica Bond III automated system as we have described previously. 2 Primary antibody against ATG5 (sc-133158, 1:50; Santa Cruz Biotechnology) was used for IHC of chordoma specimens. The IHC score of ATG5 was evaluated under a Leica Aperio AT2 scanner and defined as the product of the score of positive tumour cells as a percentage (0, no positive tumour cells; 1+, less than 25% cells; 2+, 25%-50% cells; 3+, >50% cells) × the score of the staining intensity (0, no staining; 1+, weak, light yellow; 2+, moderate, yellow-brown; 3+, strong, brown). 31,32 Low ATG5 expression was defined as an IHC score of <4, and IHC scores of ≥4 were defined as high ATG5 expression.  Technology). Tumour size was determined using the following formula: volume = (length × width 2 )/2. After 6 weeks, the tumour xenografts were extracted, weighed and applied to IHC staining for ATG5 (sc-133158, 1:200; Santa Cruz Biotechnology) and LC3B (NB100-2220, 1:400; Novus) and haematoxylin-eosin (HE) staining.

| Statistical methods
Data are presented as the mean ± SD. Statistical analysis was performed with SPSS 19.0 (IBM Corporation, USA) and GraphPad Prism 7.0 (GraphPad, USA). The Student's t test, paired t test and one-way ANOVA were used for statistical analyses between groups. The chi-squared test was used to compare clinicopathological characteristics between ATG5 subgroups. The Kaplan-Meier survival curve and Cox analysis were applied for survival analysis. Statistical significance was based on p < 0.05.

| SMARCB1 regulates the proliferation, migration, and invasion of chordoma cells
We first established stably transfected SMARCB1 knockdown and overexpressing chordoma cells, and the efficiencies were checked by qRT-PCR and Western blot ( Figure 1A, B). Next, we analysed the effect of SMARCB1 on cell viability using CCK-8 assay and observed increases in the viabilities of SMARCB1 knockdown UM-Chor1 and MUG-Chor1 cells ( Figure 1C). Correspondingly, overexpression of SMARCB1 led to impaired cell viability ( Figure 1D). Additionally, we found a similar in colony formation assay ( Figure 1E, F). We also in- Together, these findings revealed that changes in SMARCB1 expression affected the malignant phenotype of chordoma cells.

| SMARCB1 knockdown activates autophagy in chordoma cells
We then asked the potential downstream molecules mediated by SMARCB1 in chordoma cells by ChIP sequencing of UM-Chor1 cells incubated with or without an anti-SMARCB1 antibody. We identified 6714 peaks, which corresponded to 2957 genes, in the SMARCB1 group compared with the input control ( Figure S1A).
Significant enrichment of SMARCB1-associated peaks in the transcription start site (TSS) region was observed, which revealed the potential regulation of transcription by SMARCB1 (Figure 2A).
Interestingly, Gene Ontology (GO) analysis of related genes identified several autophagy-related terms that included regulation of autophagosome assembly, autophagosome and autophagosome assembly (Figures 2B, S1B-D). Thus, we speculated that SMARCB1 may function via autophagy in chordoma, and we then explored the ATGs in cells with changes in SMARCB1 expression. The mRNA expression of several ATGs was increased after SMARCB1 knockdown, which was decreased in cells that overexpressed SMARCB1 ( Figure 2C). We also found changes in expression of two autophagy markers, LC3B and P62, in these cells ( Figure 2D).

| SMARCB1 negatively regulates ATG5 expression by directly targeting the promoter of ATG5
We next explored the potential autophagy-related targets of SMARCB1. ChIP sequencing suggested potential enrichment of SMARCB1 in the promoter region of ATG5 ( Figure 3A). The subsequent ChIP-qPCR verified binding of SMARCB1 to the ATG5 promoter (+8 to +263 bp) ( Figure 3B). To further clarify the detailed binding site, we designed five probes of the binding region and performed an electrophoretic mobility shift assay (EMSA). The results showed probes 1, 2 and 4 formed a supershift band with the nuclear extract and SMARCB1 antibody ( Figure 3C). We also found augmentation of ATG5 expression in SMARCB1 knockdown cells and impaired ATG5 expression in SMARCB1-overexpressing chordoma cells ( Figure 3D and E).

| ATG5 knockdown and an autophagy inhibitor reverse the effect of SMARCB1 on chordoma cells
To determine whether ATG5 participates in the tumour suppressor role of SMARCB1 in chordoma, we examined the effect of ATG5 knockdown in UM-Chor1 and MUG-Chor1 cells with SMARCB1 knockdown. The mRNA and protein expression of ATG5 were successfully inhibited by ATG5 siRNA (Figure S2A and B). The CCK-8 assay revealed that the SMARCB1 knockdownmediated increase in proliferation was suppressed by ATG5 inhibition ( Figure 4A). Additionally, the colony number was consistently decreased after ATG5 inhibition ( Figure 4B). Transwell assays also identified suppressed migration and invasion of SMARCB1 knockdown cells treated with ATG5 siRNA ( Figure 4C). Of note, the enhanced autophagy in SMARCB1 knockdown cells was impaired by ATG5 siRNA, as suggested by the attenuated LC3-II level, decreased LC3B puncta and reduced autophagic vacuoles ( Figure 4D-F).
We also assessed the effect of autophagy inhibitor CQ in chordoma. CCK-8 assay revealed that CQ inhibited chordoma cell proliferation in a dose-dependent manner ( Figure 5A). Additionally, CQ suppressed the proliferation and colony formation of SMARCB1 knockdown cells ( Figure 5B, C). Moreover, the migration and invasion of SMARCB1 knockdown cells were significantly suppressed after CQ treatment ( Figure 5D). Together, these results indicated that ATG5-mediated autophagy was essential for the malignant phenotype of SMARCB1 knockdown chordoma cells.

| High expression of ATG5 correlates with a poor prognosis of chordoma patients
We next analysed the expression of ATG5 in chordoma tissues by IHC ( Figure S3). As a result, higher ATG5 expression was observed in recurrent chordoma compared with corresponding primary tumours ( Figure 6A, B), suggesting an oncogenic role of ATG5 in chordoma.
We then investigated potential associations between ATG5 expression and the clinical features and prognoses of the 84 chordoma patients (Table S4). Although no significant association between ATG5 and clinical features was identified, we found that high ATG5 expression was associated with shorter progression-free survival time (median, 20 months versus 80 months) and overall survival time (median, 97 months versus >168 months) comparing with the low ATG5 expression group ( Figure 6C, D). Moreover, multivariable Cox analysis identified that high ATG5 could independently predict an adverse prognosis of chordoma patients (Tables 1 and 2).

| SMARCB1 inhibits tumour growth and regulates autophagy in vivo
To confirm the tumour suppressive effect of SMARCB1 in chordoma in vivo, we established a xenograft model in BALB/c nude mice using MUG-Chor1 cells with stable transfection of SMARCB1. As shown in Figure 7A-C, chordoma cells with SMARCB1 overexpression showed a delay in tumour growth compared with the vector group.
Overexpression of SMARCB1 in the xenografted tissues was further confirmed ( Figure 7D). Additionally, increased P62 expression and attenuated ATG5 and LC3B expression in SMARCB1-overexpressing xenografts were observed, confirming the association between SMARCB1 and ATG5-mediated autophagy ( Figure 7D, E).

| DISCUSS ION
Our results confirmed the tumour suppressor role of SMARCB1 in chordoma in vitro and in vivo. Moreover, we identified ATG5 as a direct downstream target of SMARCB1 and revealed the association between autophagy and SMARCB1 in chordoma ( Figure 7F). These findings may provide novel insights into the molecular mechanism by which SMARCB1 regulates cancer progression. Expanding attention has been focussed on autophagy because of its Janus-faced effect on tumour progression. 38 One previous study on chordoma reported the presence of autophagosomes and autophagolysosomes in chordoma cell lines. 39 Additionally, a recent

ACK N OWLED G EM ENT
This study was supported by the National Natural Science Foundation of China (82071559).

CO N FLI C T O F I NTE R E S T
The authors report no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.