Exosome‐transferred long non‐coding RNA ASMTL‐AS1 contributes to malignant phenotypes in residual hepatocellular carcinoma after insufficient radiofrequency ablation

Abstract Objectives Long non‐coding RNAs (lncRNAs) are emerging RNA regulators in cancer progression, including in hepatocellular carcinoma (HCC). Recently, insufficient radiofrequency ablation (RFA) has been reported to lead to recurrence and metastasis of residual HCC tumours. Herein, we aimed to the role of ASMTL‐AS1 in residual HCC after insufficient RFA. Materials and methods In vitro insufficient RFA model was simulated in Huh7 cells and subsequently named Huh7‐H cells. In vitro and in vivo assays were conducted to investigate ASMTL‐AS1 function in HCC. Results LncRNA ASMTL‐AS1 low expressed in normal human liver was found to be highly expressed in HCC tissues and further increased in tumours after insufficient RFA. ASMTL‐AS1 expression was related to stage, metastasis and prognosis in HCC. Huh7‐H possessed higher ASMTL‐AS1 level and more aggressive than Huh7 cells. ASMTL‐AS1 contributed to the malignancy of HCC cells both in vitro and in vivo. Mechanistically, ASMTL‐AS1 was trans‐activated by MYC and promoted NLK expression to activate YAP signalling via sequestering miR‐342‐3p in HCC. Interestingly, ASMTL‐AS1 could be wrapped by exosomes and then convey malignancy through NLK/YAP axis between cells even in residual HCC after insufficient RFA. Conclusions Exosomal ASMTL‐AS1 aggravates the malignancy in residual HCC after insufficient RFA via miR‐342‐3p/NLK/YAP signalling, opening a new road for the treatment of HCC and the prevention of recurrence or metastasis of residual HCC after insufficient RFA.


| INTRODUC TI ON
Hepatocellular carcinoma (HCC), which always results in high morbidity and mortality, is one of the most aggressive cancers all over the world.1 In China, HCC ranks third in terms of the leading causes of cancer-related death.2 Although liver transplantation is the preferred treatment for HCC patients, its application usually be limited due to deficiency of good donor organs. 3 In this case, surgical resection and radiofrequency ablation (RFA) are extensively adopted to cure this disease.4 However, the option of RFA or surgery depends on the specific conditions of patients.5 Disappointingly, clinical studies indicated that sufficient ablation is hardly achieved in HCC patients,6 which might be attributed to insufficient ablative margin,7 heat loss in tumours via vessels8 or a low-temperature area in a large tumour when ablation.9 More seriously, residual tumours after insufficient RFA always lead to recurrence and metastasis.10 Hence, understanding the molecular mechanism through which residual tumours after insufficient RFA contributed to tumorigenesis and metastasis is the urgent work for researchers.
In past few decades, long non-coding RNAs (lncRNAs), defined as a subclass of non-coding RNAs (without protein-coding ability) with more than 200 nts in length, 11 have been increasingly recognized to be involved in cancer development12 including HCC. 13 As an example, lncRNA CASC9 facilitates HCC proliferation by interacting with HNRNPL to regulate AKT signalling.14 However, the significance of lncRNAs in the development of residual HCC after insufficient RFA has hardly been discussed yet, except a previous report which illustrated that lncRNA FUNDC2P4 suppresses epithelial-mesenchymal Interestingly, LncBook (https://bigd.big.ac.cn/lncbo ok/index) and NONCODE (http://www.nonco de.org/) indicated ASMTL-AS1 is low or even un-expressed in normal human liver tissues. On the contrary, starBase suggested a potential up-regulation tendency (although not significant, which may due to limited samples and individual differences) in liver hepatocellular carcinoma (LIHC) tissues compared with normal tissues. On this basis, we suspected that ASMTL-AS1 might be involved in HCC development. More intriguingly, lncLocator (http://www.csbio.sjtu.edu.cn/bioin f/lncLo cator /) predicted ASMTL-AS1 might mainly exist in exosomes of cells, which implied an involvement of exosomes during cancer progression. In this study, according to above data, we probed into the role of ASMTL-AS1 in residual HCC after insufficient RFA and whether exosomes were also implicated in to convey ASMTL-AS1 between HCC cells.

| Cell lines
Human HCC cell lines (Hep2G, Huh7, HCCLM3 and SMMC7721) and normal human hepatic cell line (THLE-2) were available from the Chinese Academy of Sciences (Shanghai, China), maintained in 5% CO 2 at 37°C. RPMI 1640 medium (Gibco) was applied for culturing cells with 10% foetal bovine serum (FBS; Gibco) and 1% pen/strep. treatment of HCC and the prevention of recurrence or metastasis of residual HCC after insufficient RFA.
To mimic in vitro cancerous cells after insufficient RFA, Huh7 cells in 7 cm diameter petri dishes were placed at 1 × 10 5 cells per well for 24 hours, blocked with parafilm and submerged into 47°C water bath.
Following incubation under normal condition, confluent cells were heated in 47°C water bath, and surviving cells were named as Huh7-H.

| RNA extraction and QRT-PCR
Total RNA was extracted using TRIzol reagent (Thermo Fisher Scientific) as per user guidebook, denatured and reversely transcribed into cDNA using Reverse Transcription Kit (TaKaRa). qRT-PCR was carried out by BioRad SYBR Green Super Mix (Bio-Rad), with GAPDH or U6 as internal control. Relative expression levels of genes were calculated by the 2 −∆∆Ct method.

| EdU incorporation assay
Cells in each group were planted in 96-well plates for incubating with EdU incorporation assay kit (Ribobio) as instructed by manufacturer.
Cells were observed after adding DAPI solution under fluorescent microscope (Leica).

| Colony formation assay
Clonogenic cells at logarithmic growth phase were planted in 6-well plates for 14-day culture process. After fixation in 4% paraformaldehyde, cells were processed with 0.1% crystal violet. Colonies were counted.

| Transwell assay
Transwell assays were conducted using 8 μm pore transwell insert (Millipore) with or without Matrigel for invasion or migration. 2 × 10 4 cells in serum-free medium were seeded into upper chamber. Lower chamber was filled with 700 μL of complete medium. Forty-eight hours later, the migrated or invaded cells were fixed for 20 minutes and stained with crystal violet for 10 minutes. Number of cells was determined as the mean of cell number in 5 random filed using microscopy.

| Western blot
Proteins lysates obtained in RIPA lysis buffer were subjected to 10% SDS-PAGE and transferred onto the PVDF membrane.
Membranes were sealed with 5% skimmed milk and incubated with specific primary antibodies (Abcam) overnight at 4°C. Followed by washing in TBST, secondary antibodies conjugated with HRP were used. Proteins were visualized using a detection system of enhanced chemiluminescence (ECL). GAPDH served as the internal reference.

| Plasmid transfection
The pcDNA3.1 vectors and NC vectors were available from Genepharma Company to overexpress ASMTL-AS1, MYC and NLK in cell samples using transfection kit Lipofectamine 2000 (Invitrogen).
The miR-342-3p mimics and miR-NC, as well as the shRNAs specific to ASMTL-AS1, MYC and NC shRNAs, were also acquired from Genepharma Company. Forty-eight hours later, cells were reaped from each group for analysis.

| In situ hybridization
The tissue sections were dehydrated in ethanol to hybridize with 40 nmol/L of specific RNA in situ hybridization (ISH) probes for 1 hour and then rinsed in PBS. After nuclear counterstaining with DAPI, RNA level was assayed by microscope.

| Immunohistochemistry
Tissue samples from animal study were fixed by 4% paraformaldehyde and embedded in paraffin for cutting. The successive sections (4 μm) were acquired and cultured with specific antibodies (Abcam) for immunohistochemistry (IHC) analysis.

| Chromatin immunoprecipitation assay
Cells were fixed for 20 minutes for forming the DNA and protein cross-linking, then fragmented by ultrasonic and immunoprecipitated with anti-MYC antibody or control IgG antibody (Abcam).
30 μL magnetic beads were then added for 2 hours, and precipitated DNA was purified for qRT-PCR.

| Fluorescence in situ hybridization analysis
The ASMTL-AS1-specific fluorescence in situ hybridization analysis (FISH) probe was available from Ribobio and employed as per the user guide. Cell nucleus was analysed by DAPI staining using microscope.

| RNA immunoprecipitation assay
RNA immunoprecipitation (RIP) assay was implemented as guided by the protocol of EZ-Magna RIP RNA Binding Protein Immunoprecipitation Kit (Millipore). Antibodies against Ago2 and control IgG were acquired from Abcam. Precipitated RNAs were collected by adding beads and measured via qRT-PCR.

| Co-immunoprecipitation assay
Cells were processed with IP lysis buffer to acquire cell lysates and then mixed with specific antibodies in constant speed at 4°C. Normal IgG served as control. Then, the mixture was cultured with beads for 2 hours, rinsed in IP lysis buffer and then eluted for Western blot analysis.

| Subcellular fractionation
Subcellular fractionation assay was conducted to acquire the cell nuclei and cell cytoplasm fractions using PARIS™ Kit (Ambion) in line with the standard method. The isolated RNAs were monitored by Western blot method.

| Immunofluorescence
After cells were adhered to the culture slides, cells in PBS were blocked in 5% BSA for probing with primary antibodies for 1.5 hours at room temperature. After washing, the secondary antibodies were added for 40 minutes. Cells were finally treated with DAPI and visualized by fluorescent microscope.

| Exosome morphology and size analysis
The morphology of exosomes was observed under transmission electron microscopy (TEM). Exosomes derived from each group were visualized and photographed with digital camera (Olympus).
The size and number of exosomes were measured and determined using nanoparticle tracking analysis (NTA) measurements and tunable resistive pulse sensing (TRPS, IZON qNano).

| Statistical analyses
Data were expressed as means ± SD (standard deviation) from three independent bio-repeats. Overall survival was estimated by Kaplan-Meier analysis. Correlation analysis was examined by Pearson's chisquared analysis. Statistical difference was calculated with Student's t test or one-way analysis of variance (ANOVA) using SPSS 19.0 (SPSS), with P < .05 as threshold.

| Up-regulation of ASMTL-AS1 enhances the malignancy of residual HCC cells after insufficient RFA
In this research, we studied a novel lncRNA ASMTL-AS1 which had never been explored yet. First of all, we were curious about the characteristics of ASMTL-AS1. Normally, ASMTL-AS1 was suggested by LncBook and NONCODE to be low expressed and even nearly deficient in human liver tissues ( Figure S1A,B). By contrast, ASMTL-AS1 expression tended to be up-regulated in liver hepatocellular carcinoma (LIHC) tissues relative to the normal ones ( Figure S1C). Also, the results from LncBook showed that ASMTL-AS1 had no proteincoding ability by three prediction tools ( Figure S1D). In this case, we were strongly interested in the potential role of ASMTL-AS1 in HCC.
Therefore, we then analysed the correlation of ASMTL-AS1 with HCC development. As indicated in Figure 1A, we found that the expression level of ASMTL-AS1 was markedly elevated in HCC tissues compared to adjacent non-cancerous tissues and its level seemed to be gradually increased along with cancer development and distant metastasis. In the meantime, we also found that the expression of ASMTL-AS1 in HCC tissues was strongly correlated with tumour size, distant metastasis and TNM stage (Table 1). Besides, patients with higher ASMTL-AS1 level always underwent worse outcomes ( Figure 1B). More interestingly, ASMTL-AS1 expression was further enhanced in residual HCC tissues collected from patients with insufficient RFA ( Figure 1C). Meanwhile, the prognosis of HCC patients who suffered insufficient RFA was further deteriorated when possessing high ASMTL-AS1 expression ( Figure 1D). All these results suggested ASMTL-AS1 might play an accelerating role in the development of residual HCC after insufficient RFA.
To make sure the exact function of ASMTL-AS1 in residual HCC after insufficient RFA, in vitro experiments were conducted subsequently. It was showed that ASMTL-AS1 was highly expressed in HCC cell lines compared to normal human liver immortalized cell line THLE-2 ( Figure 1E). Afterwards, Huh7 and HCCLM3 cells processing heat treatment were employed to simulate the residual tumour cells after insufficient RFA in vitro and such cells were then called Huh7-H and HCCLM3-H cells, respectively. As revealed by qRT-PCR, the expression of ASMTL-AS1 was robustly stimulated after heating ( Figure 1F and Figure S2A). Furthermore, such change caused by heating made Huh7-H cells grow faster and have more invasiveness and migration capability ( Figure 1G-J), so was in HCCLM3-H cells ( Figure S2B). Taken together, ASMTL-AS1 is significantly up-regulated in residual HCC tissues after insufficient RFA as well as in heated Huh7 cells.

| ASMTL-AS1 facilitates HCC cell growth and metastasis both in vitro and in vivo
For the next step, we probed the relationship between ASMTL-AS1 expression and the malignancy of residual HCC cells. In this situation, gain-and loss-of-function assays were carried out. It was confirmed that the level of ASMTL-AS1 was successfully overexpressed in To provide in vivo evidence, we also performed in vivo experiments here. Results indicated that in contrast to tumours derived from control cells, those from ASMTL-AS1-overexpressed Huh7 cells grew much faster and finally had bigger sizes and weights ( Figure 2G,H). In addition, tumours originated from Huh7 cells with ectopic ASMTL-AS1 expression were proved to possess higher ASMTL-AS1 level and stronger Ki67 staining than those from control group ( Figure 2I). In the meantime, the impact of ASMTL-AS1 on the metastasis of HCC cells was also estimated through conducting in vivo metastatic experiments. As a consequence, more metastatic nodules were generated in the liver of mice injected with ASMTL-AS1-overexpressed Huh7 cells relative to that of mice from control group ( Figure 2J). Altogether, above results certified that ASMTL-AS1 can aggravate malignant phenotypes in HCC both in vitro and in vivo.

| ASMTL-AS1 is trans-activated by MYC in residual HCC cells
Afterwards, we wanted to know why and how ASMTL-AS1 was upregulated in residual HCC cells after insufficient RFA. According to the prediction of UCSC and PROMO, there were three transcription factors, including MYC, YY1 and ELK1, that might regulate ASMTL-AS1 expression ( Figure 3A). Nevertheless, only the expression of MYC was evidently potentiated after heat treatment ( Figure 3B). Besides, MYC expression was generally enhanced in HCC cell lines contrast to THLE-2 cells ( Figure 3C). Also, the level of MYC was markedly augmented in HCC tissues compared to paratissues and it was further increased in HCC tissues after insufficient RFA ( Figure 3D). Importantly, we discovered a significant positive correlation between MYC and ASMTL-AS1 expressions in HCC tissues and a higher significance of their correlation in HCC tissues after insufficient RFA ( Figure 3E). Also, it was suggested that ASMTL-AS1 expression was increased or decreased in the wake of MYC up-regulation or down-regulation, respectively ( Figure 3F,G).
Moreover, results of ChIP assay revealed that ASMTL-AS1 promoter was prominently captured by anti-MYC but not anti-IgG ( Figure 3H and Figure S2F), while luciferase reporter assay testified that the luciferase activity of ASMTL-AS1 promoter was revived by MYC overexpression but suppressed under MYC knockdown ( Figure 3I).
Hence, we found ASMTL-AS1 transcription was positively modulated by MYC in residual HCC cells.
Thereafter, we explored the precise sequence through which MYC bound to ASMTL-AS1 promoter. As indicated in Figure 3Ja

| ASMTL-AS1 prompts NLK expression in HCC cells via miR-342-3p
In depth, we attempted to identify the detailed mechanism to explain the contribution of ASMTL-AS1 to the malignancy of residual HCC. The FISH analysis suggested that regardless of heating or not, ASMTL-AS1 was predominantly present in the cytoplasm of Huh7 and HCCLM3 cells although heat treatment markedly elevated ASMTL-AS1 signals ( Figure 4A and Figure S3A). This result highlighted the potential for ASMTL-AS1 to function as a competing endogenous RNA (ceRNA) by sponging certain miRNAs. Interestingly, starBase predicted three miRNAs (including miR-1343-3p, miR-342-3p and miR-6783-3p) that might interact with ASMTL-AS1, among which only miR-342-3p was overtly pulled down by Bio-ASMTL-AS1 ( Figure 4B). Besides, clinical data represented a noticeable reduction of miR-342-3p in HCC tissues and its level further declined in residual HCC tissues after insufficient RFA ( Figure S3B).
Meanwhile, miR-342-3p was revealed to be low expressed in HCC cell lines and further down-regulated after heat treatment ( Figure S3C,D). Moreover, we verified that both miR-342-3p and ASMTL-AS1 were harvested by anti-Ago2 ( Figure 4C) and that the luciferase activity of ASMTL-AS1-WT was silenced by ectopic expression of miR-342-3p ( Figure 4D). In this regard, miR-342-3p was screened out as the downstream of ASMTL-AS1 in HCC.  (Figure 4Fb). Moreover, NLK was negatively correlated with miR-342-3p but positively related to ASMTL-AS1 in all HCC tissues and also in residual HCC tissues after insufficient RFA ( Figure 4G). Further, the co-enrichment of ASMTL-AS1, miR-342-3p and NLK in anti-Ago2-induced immunoprecipitate was observed in both Huh7 and Huh7-H cells ( Figure 4H), as well as in HCCLM3 and HCCLM3-H cells ( Figure S3E). Importantly, enhanced expression of miR-342-3p impaired the luciferase activity of NLK-WT, whereas such impairment was reversed by ASMTL-AS1 overexpression ( Figure 4I). Collectively, it could be concluded that ASMTL-AS1 enhances NLK expression in HCC by secluding miR-342-3p.   Figure 5C and Figure S3H). Further, such changes made it easier for YAP to translocate to nucleus under ASMTL-AS1
Collectively, ASMTL-AS1 facilitates the nuclear translocation of YAP in HCC cells by NLK-mediated manner.

| ASMTL-AS1 aggravates the malignancy of HCC cells through targeting NLK
In subsequence, we probed whether the facilitating role of ASMTL-AS1 in the malignancy of HCC was mediated by NLK. As proved in Figure 5F, both the mRNA and protein expressions of NLK that were increased in ASMTL-AS1-up-regulated Huh7 cells were normalized in response to NLK knockdown. Moreover, such normalization on NLK expression led to recovered cell proliferation, motility and EMT in Huh7 cells with ASMTL-AS1 overexpression ( Figure 5G).
On the contrary, forced NLK expression counteracted the impact of ASMTL-AS1 suppression on the biological behaviours of Huh7-H cells ( Figure 5H,I). All in all, these data illustrated that ASMTL-AS1 accelerates the malignant phenotypes of HCC cells by NLK-relied pathway.

| Exosome-transmitted ASMTL-AS1 confers malignant behaviours between residual HCC cells through NLK/YAP signalling
Recently, the importance of tumour microenvironment has been highlighted in cancer development. Exosomes are a subset of extracellular vesicles that exist in tumour microenvironment to confer malignant information to recipient cells. 25 Interestingly, the online lncLocator indicated that ASMTL-AS1 was largely in the exosomes of cells ( Figure 6A). Besides, it was revealed that the expression of ASMTL-AS1 in the culture medium (CM) of Huh7 and Huh7-H cells was unaffected by sole treatment with RNase A but was largely abolished after the co-treatment of RNase A and Triton X-100 ( Figure 6B), suggesting that extracellular ASMTL-AS1 did not exist as a released form but was protected through being wrapped by membrane. Subsequently, exosomes was extracted and purified from CM and further identified via exosome markers CD63 and TSG101 ( Figure 6C). More importantly, we proved that there were no significant differences in the size and number of exosomes between Huh7 and Huh7-H cells ( Figure 6D). Furthermore, ASMTL-AS1 expression in exosomes was altered in line with its level in whole cells ( Figure 6E).
Next, we launched an investigation into the function of exosomes containing ASMTL-AS1 in HCC. As shown in Figure 6F Kaplan-Meier analysis. E, qRT-PCR revealed the level of NLK in para-tumours, primary HCC tumours without RFA treatment and residual HCC tumours after insufficient RFA. F, The levels of indicated proteins in above three kinds of tumours were analysed by Western blot. G, YAP localization in above tumours was also determined through Western blot. PT meant para-carcinoma tissues. T/N referred to HCC tissues from patients with no RFA treatment, and T/R indicated HCC tissues from patients with insufficient RFA. *P < .05, **P < .01 HCC after insufficient RFA ( Figure 7C). More importantly, although high exosomal ASMTL-AS1 expression resulted in disappointing survival rates in HCC patients without RFA treatment, it gave rise to worse outcomes in those with insufficient RFA ( Figure 7D). Additionally, we found NLK expression continuously augmented, just as the trend of ASMTL-AS1 in different tissues obtained from HCC patients ( Figure 7E). As a result, more YAP was phosphorylated at Ser 128 and therefore induced more YAP translocating to nucleus in tumours from patients after insufficient RFA than in those from patients without RFA due to higher NLK protein ( Figure 7F,G). On the whole, ASMTL-AS1 transferred by exosomes enhances the malignancy of residual HCC tumours after insufficient RFA through NLK/YAP signalling. to their ability to convey information to recipient cells. 43 In our research, we certified exosomes could affect the biological behaviours of Huh7 and Huh7-H cells depending on ASMTL-AS1 expression they contained to some extent. More importantly, cell received exosomes with high or low ASMTL-AS1 level could finally affect NLK expression and therefore influence YAP nuclear translocation, which was validated in vitro and further evidenced by clinical data.

| D ISCUSS I ON
In conclusion, this research illustrated the participation of an ASMTL-AS1/miR-342-3p/NLK/YAP signalling in reinforcing the malignancy of HCC cells, especially those in residual tumours after insufficient RFA. Also, we detected ASMTL-AS1 could be transmitted between HCC cells to activate such signalling in the recipient cells as well. These findings provide ASMTL-AS1 as a novel target that is potentially effective for the treatment of HCC patients, especially those after insufficient RFA. Moreover, it also suggests ASMTL-AS1 may be helpful to prevent the recurrence and metastasis of residual HCC after RFA.

ACK N OWLED G EM ENT
We thank each partner in our team and the kind people who gave us a hand during this work.

CO N FLI C T O F I NTE R E S T
All authors ensure no conflicts of interest in our work.