LncRNA SND1-IT1 facilitates TGF-β1-induced epithelial-to-mesenchymal transition via miR-124/COL4A1 axis in gastric cancer

The transformation of tumor cells from an epithelial to a mesenchymal-like phenotype, designated as epithelial-to-mesenchymal transition (EMT), represents a key hallmark of human cancer metastasis, including gastric cancer (GC). However, a large set of non-coding RNAs have been studied for their functions that initiate or inhibit this phenotypic switch in GC cells by regulating oncogenes or tumor suppressors. In this paper, we aimed to identify lncRNA SND1-IT1, miR-124, and COL4A1 gene in the context of GC with a specific focus on their effects on transforming growth factor β1 (TGF-β1)-induced EMT. The study included 52 paired samples of lesion tissues and adjacent lesion-free tissues surgically resected from patients diagnosed with GC. HGC-27 cells were stimulated with exogenous TGF-β1 (2 ng/mL). Expression of lncRNA SND1-IT1, miR-124, and COL4A1 was determined by RT-qPCR. CCK-8 assays, Transwell assays, immunoblotting analysis of EMT-specific markers, and tumor invasion markers were performed to evaluate cell viability, migration, and invasion of cultured HGC-27 cells. Luciferase activity assay was employed to examine miR-124 binding with lncRNA SND1-IT1 and COL4A1, respectively. LncRNA SND1-IT1 was upregulated in GC tissues and cells. TGF-β1-stimulated EMT and regulated lncRNA SND1-IT1, miR-124, and COL4A1 expressions in HGC-27 cells. LncRNA SND1-IT1 knockdown tempered HGC-27 cell viability, migration and invasion. LncRNA SND1-IT1 participated in TGF-β1-stimulated EMT in GC by sponging miR-124. MiR-124 attenuated TGF-β1-stimulated EMT in GC by targeting COL4A1. These results primarily demonstrated TGF-β1 can regulate cancer cell migration, invasion and stimulate EMT through the SND1-IT1/miR-124/COL4A1 axis in GC.


INTRODUCTION
Gastric cancer (GC) is one of the most lethal malignancies and contributes to a substantial health care burden to patients. Each year, more than 1 million new cases have been diagnosed worldwide [1]. Although the past decade has witnessed a better understanding in prevention and personalized treatment, GC still remains the sixth in incidence and second in mortality globally in 2018 [2]. GC is a complex, heterogeneous disease arising from unique epidemiologic backgrounds, such as Helicobacter pylori infection [3], diet, smoking, obesity [4], host genetic mutation and instability [5], such as E-cadherin gene (CDH1), PALB2, BRCA1, and RAD51C mutations [6,7]. Laparoscopy-assisted distal gastrectomy (LADG) and open distal gastrectomy (ODG) have been suggested as standard treatment options for GC at an early stage [8]. Due to the lack of marked symptoms in an early stage of GC, many patients present advanced disease at their initial diagnosis [9]. Epithelial-mesenchymal transition (EMT) is a prominent process in tumor development by converting epithelial phenotypic cells into mesenchymal phenotypic cells [10]. Transforming growth factor (TGF)-β1 is a well-characterized contributor to the process of EMT [11]. EMT usually limits total surgical resection and leads to therapeutic resistance, whereas there is limited knowledge about signaling pathways and effector molecules that initiate this phenotypic switch in GC. Thus, molecular processes and the downstream mechanisms involved in TGF-β1-stimulated EMT in GC remain enigmatic and merit further investigation.
Long non-coding RNAs (lncRNAs), a subgroup of RNAs with a length of more than 200 nucleotides, are abnormally expressed in various human cancers including GC [12]. Staphylococcal nuclease and Tudor domain-containing 1 intronic transcript 1 (SND1-IT1) is a newly identified lncRNA located on chromosome 7 (7q32.1) and found to modulate the promoter of oncogenes [13]. LncRNA SND1-IT1 has been reported to participate in the development of human diseases, including myocardial ischemia/reperfusion injury [14] and laryngeal squamous cell carcinoma [15]. However, the regulating signaling of lncRNA SND1-IT1 in GC development remains unknown. Plentiful lncRNAs govern a broad spectrum of biological processes, such as tumor initiation and metastasis by functioning as microRNA (miRNA) sponges and positive regulators of parental gene transcription [16]. The competitive endogenous RNA (ceRNA) hypothesis proposes that transcripts, including lncRNAs, share miRNA binding sites and compete for post-transcriptional control, which is becoming a new paradigm of ncRNA regulation [17]. For example, lncRNA SND1-IT1 has been suggested to promote osteosarcoma cancer cell growth by acting as a sponge to positively regulate POU2F1, the target of miR-655 (PMID: 31799644). An lncRNA-miRNA binding prediction indicates lncRNA SND1-IT1 and miR-124 share binding sites. MiR-124 was downregulated and characterized as a tumor suppressor in GC [18,19]. There is a feedback loop between miR-124 and TGF-β pathway that is crucial for NSCLC metastasis [20]. Another miRNA-mRNA prediction suggests putative miR-124 binding sites on the 3′UTR of the COL4A1. Accordingly, mutations in COL4A1 are pleiotropic and tightly linked with a broad spectrum of human diseases, including GC [21,22]. Herein, we proposed a hypothesis that lncRNA SND1-IT1 engages GC progression and metastasis by enhancing TGF-β1stimulated EMT through modulating miR-124 and COL4A1.
The present study demonstrated that lncRNA SND1-IT1 was notably upregulated in GC tissues. Moreover, lncRNA SND1-IT1 could function as a ceRNA to regulate COL4A1 by competing for miR-124, which underpins TGF-β1-stimulated EMT in GC. These findings extend our knowledge of the biological role and underlying mechanism of lncRNA SND1-IT1 in GC metastasis.

RESULTS
LncRNA SND1-IT1 was upregulated in GC tissues and cells Initially, lncRNA SND1-IT1 was quantified in human cancer tissue samples surgically resected from GC patients by RT-qPCR. As depicted in Fig. 1A, the results showed that the expression level of lncRNA SND1-IT1 was higher in lesion tissues than in lesion-free tissues. It revealed that relative expression of lncRNA SND1-IT1 was associated with invasion depth, lymph node metastasis, and TNM stage (P < 0.05, Table 1). We also found that lncRNA SND1-IT1 exhibited a higher expression level in four selected GC cells (HGC-27, AGS, NCI-N87, and MKN74) than in GES1 cells (Fig. 1B). Thus lncRNA SND1-IT1 was overexpressed in the context of GC.
LncRNA SND1-IT1 knockdown inhibited GC cell proliferation, migration, invasion, and EMT HGC-27 cells were transfected with siRNA targeting lncRNA SND1-IT1 and then cultured without the addition of TGF-β1. The siRNA targeting lncRNA SND1-IT1 significantly knocked down lncRNA SND1-IT1 in HGC-27 cells. As shown in Fig. 2A, si-SND1-IT1 HGC-27 cells were constructed. The results showed that knockdown of lncRNA SND1-IT1 inhibited HGC-27 cell viability, migration, and invasion ( Fig. 2B-D). To better understand whether lncRNA SND1-IT1 functionally influences migration and invasion properties of GC cells, we performed immunoblotting to determine alternations of (Fig. 2E), Not surprisingly, the results of immunoblotting showed a declined endogenous E-cadherin expression yet increased endogenous N-cadherin expression in si-SND1-IT1 HGC-27 cells. These results suggested that knockdown lncRNA SND1-IT1 led to inhibited GC cell migration and invasion along with enhanced EMT.
LncRNA SND1-IT1 sponged miR-124 in GC LncRNAs can act as competing endogenous RNAs (ceRNAs) to modulate gene expression by competing for the shared miRNAs in humans. Given that, we are curious about whether miR-124 is implicated in the regulation of lncRNA SND1-IT1 in the EMT in GC. We first quantified miR-124 expression in 52 paired samples of lesion tissues and lesion-free tissues surgically resected from GC patients by RT-qPCR. The results showed that miR-124 expression was lower in lesion tissues than in adjacent lesion-free tissues (Fig. 5A), and it was negatively correlated with lncRNA SND1-IT1 expression (Fig. 5B). Likewise, we examined miR-124 expression in four selected GC cells and GES1 and found that miR-124 exhibited a lower expression level in four selected GC cells than in GES1 cells (Fig. 5C). An lncRNA-miRNA prediction showed putative miR-124 binding sites in the lncRNA SND1-IT1 transcripts (Fig. 5D). LncRNA SND1-IT1 binding with miR-124 was further demonstrated by dualluciferase reporter gene assay, evidenced by decreased luciferase luminosity at the promoter of the reporter gene containing wt-lncRNA SND1-IT1 compared with that containing mut-lncRNA SND1-IT1 in the presence of miR-124 mimic (Fig. 5E). Results of qRT-PCR indicated lncRNA SND1-IT1 overexpression reduced the miR-124, while lncRNA SND1-IT1 knockdown increased the miR-124 (Fig. 5F). These results indicated lncRNA SND1-IT1 sponged miR-124 in GC.

MiR-124 targeted and inhibited COL4A1 in GC
To the best of our knowledge, the miRNA regulatory network of a specific mRNA plays a crucial character in human diseases. We believed miR-124 regulating gene expression undertakes the regulation of lncRNA SND1-IT1 in GC. In the beginning, we examined COL4A1 expression in 52 paired samples of lesion tissues and lesion-free tissues by RT-qPCR. The results showed that COL4A1 expression was higher in lesion tissues than in lesion-free tissues (Fig. 7A), and it was negatively correlated with miR-124 expression (Fig. 7B). It was found that COL4A1 exhibited a higher expression level in four selected GC cells than in GES1 cells (Fig. 7C). We performed a web-available prediction of miR-124 binding sites in the COL4A1 mRNA 3′UTR in the starbase (Fig. 7D). We also observed a mark reduction in luciferase activity of the promoter of the reporter gene containing COL4A1-wt rather than COL4A1-mut in the presence of miR-124 mimic (Fig. 7E). In addition to dual-luciferase reporter gene assay, immunoblotting demonstrated targeted inhibition of COL4A1 by miR-124, as evidenced by declined COL4A1 in response to miR-124 mimic and increased COL4A1 in response to miR-124 inhibitor in HGC-27 cells (Fig. 7F). These data revealed that miR-124 targeted and inhibited COL4A1 in GC.
MiR-124 attenuated TGF-β1-stimulated EMT in GC by targeting COL4A1 Finally, we attempted to dissect out whether targeted inhibition of COL4A1 by miR-124 was involved in TGF-β1-stimulated EMT. The expression of COL4A1 was increased in HGC-27 cells after TGF-β1 treatment but this increase was prevented after miR-124 mimic treatment. In addition, it was observed the expression of COL4A1 was not different between HGC-27 cells with treatment of TGF-β1, miR-124 mimic, and COL4A1 expression vector and HGC-27 cells with TGF-β1 treatment alone (Fig. 8A). As shown in Fig. 8B-D, The contribution of TGF-β1 treatment to HGC-27 cell viability, migration, and invasion was negated by miR-124 mimic treatment. No significant difference concerning HGC-27 cell viability, migration, and invasion was observed between HGC-27 cells with treatment of TGF-β1, miR-124 mimic, and COL4A1 expression vector and HGC-27 cells with TGF-β1 treatment alone. Results of immunoblotting (Fig. 8E) displayed that a decrease in E-cadherin expression and an increase in N-cadherin expression following TGF-β1 treatment were reversed by miR-124 mimic treatment. No significant difference with regard to E-cadherin and N-cadherin expressions was found between HGC-27 cells with treatment of TGF-β1, miR-124 mimic, and COL4A1 expression vector and HGC-27 cells with TGF-β1 treatment alone. These results suggested that miR-124 attenuated TGF-β1-stimulated EMT in GC by targeting COL4A1.

DISCUSSION
Cancer metastasis is associated with EMT, which is a process whereby epithelial cells lose their polarity and acquire new features of mesenchyme. TGF-β1 has been reported as a predominant contributor in the process of EMT, but its mechanism remains to be elucidated [24]. Recently, a study showed that TGF-β1-induced upregulation of lncRNAs promoted GC invasion and migration [25].
In this study, we found TGF-β1 stimulation upregulated lncRNA SND1-IT1 and lncRNA SND1-IT1 promoted TGF-β1-stimulated EMT in GC. More than that, we also found lncRNA SND1-IT1 sponging miR-124 was involved in TGF-β1-stimulated EMT. Once miR-124 was inhibited, its target gene COL4A1 was triggered and thus enhanced TGF-β1-stimulated EMT. These results and our functional evidence for lncRNA SND1-IT1-miR-124-COL4A1 network might exhibit great value to explore the molecular mechanism underlying TGF-β1-stimulated EMT in GC. LncRNA SND1-IT1 was found to be a promoter for metastasis of GC. As our results have shown, lncRNA SND1-IT1 was abundantly expressed in GC tissues and cells. Prior work has documented that lncRNA NEAT1 expression was enhanced in the context of osteosarcoma [26], suggesting its oncogenic role in human cancers. Additionally, cellular migration can be promoted by an accelerated EMT process in GC [27]. Not surprisingly, our results showed that overexpression of lncRNA SND1-IT1 led to a declined expression of E-cadherin but an increased N-cadherin expression in HGC-27 cells. The molecular pathogenesis of GC is closely associated with E-cadherin, and suppressed biological function of GC cells is correlated with upregulated E-cadherin level, considerably downregulated N-cadherin [28]. Moreover, we found TGF-β1 stimulation upregulated lncRNA SND1-IT1 and lncRNA SND1-IT1 promoted TGF-β1-stimulated EMT, as evidenced by an upregulated lncRNA SND1-IT1 in HGC-27 cells upon TGF-β1 stimulation. TGF-β1-induced upregulation of lncRNAs promoted GC invasion and migration [25]. TGF-β1 is known as a  Values are expressed as mean ± SD from three technical repeats. *P < 0.05, **P < 0.01, ***P < 0.001. promoter of tumor metastasis by inducing EMT process in GC [29]. Following stimulation of TGF-β1, levels of Vimentin and α-SMA are increased while reduced E-cadherin expression is observed [30], which is in line with our data derived from this study. In contrast, knockdown of lncRNA SND1-IT1 suppressed EMT among GC cells by downregulating N-cadherin, Vimentin, and snail. The abovementioned literature supports our finding that overexpression of lncRNA SND1-IT1 promoted cellular migration, invasion, and TGF-β1-stimulated EMT in GC cells.
The downstream molecular mechanism underlying lncRNA SND1-IT1 was subsequently explored. Our data exhibited that lncRNA SND1-IT1 could specifically bind to miR-124 which targeted and suppressed expression of COL4A1. Results derived from our study showed that miR-124 was downregulated in GC cells and tissues. Similarly, miR-124 has been found to be poorly expressed in GC [31]. Also, decreased expression of miR-124 has been associated with poor outcomes in GC patients [32], which demonstrated the tumor-suppressive role of miR-124 in GC. Moreover, upregulated miR-124 significantly suppresses cell proliferation, invasion, and downregulates levels of MMP2 and MMP9 among osteosarcoma cells [33]. We next found the regulatory mechanism between lncRNA SND1-IT1 and miR-124 in GC. Exogenous addition of TGF-β1 inhibited miR-124 expression, while following lncRNA SND1-IT1 knockdown further enhanced miR-124 expression in HGC-27 cells, suggesting the involvement of lncRNA SND1-IT1/miR-124/TGF-β1 in the process of EMT in GC. As previously reported, Zhang et al. demonstrated lncRNA UCA1/miR-124 axis modulates TGFβ1-induced EMT through JAG1/Notch signaling in tongue cancer [34]. COL4A1 directly targeted by miR-124 was found to be suppressed by miR-124. COL4A1 expression is reported to be associated with the TGFβ1 gene in subcutaneous adipose tissues [35]. Furthermore, another collagen gene family member, COL10A1 has been proved to induce EMT by increasing N-cadherin and Vimentin expression and reducing E-cadherin expression [36]. Our study further demonstrated that COL4A1 could negate the suppression of miR-124 elevation on TGFβ1-induced EMT. It could be indicated that overexpression of miR-124 restrained GC cell migration, invasion, and attenuated TGFβ1-induced EMT by silencing COL4A1 expression. Fig. 7 MiR-124 targeted and inhibited COL4A1. A The expression level of COL4A1 was determined in tumor tissues (n = 52) by RT-qPCR and compared to adjacent non-tumor tissues (n = 52). B Spearman correlation analysis between miR-124 and COL4A1 in GC tissues. C The expression level of COL4A1 was determined in four selected GC cells and GES1 cells by RT-qPCR and immunoblots. D Putative miR-124 binding sites in the COL4A1 mRNA 3′UTR in the starbase database. E Luciferase activity of the promoter of the reporter gene containing COL4A1-wt or COL4A1-mut in the presence of miR-124 mimic. F Representative immunoblots and quantitative analysis of COL4A1 protein in response to exogenous addition of miR-124 mimic and miR-124 inhibitor. Values are expressed as mean ± SD from three technical repeats. **P < 0.01, ***P < 0.001. The results were normalized to the β-actin expression. Values are expressed as mean ± SD from three technical repeats. *P < 0.05, **P < 0.01, ***P < 0.001.
Conclusively, the study provides evidence that lncRNA SND1-IT1 could function as a ceRNA to regulate COL4A1 by competing for miR-124, which underpins TGF-β1-stimulated EMT in GC. However, further in vivo and in vitro investigations are required to validate whether there is positive feedback between lncRNA SND1-IT1 and TGFβ1, negative feedback between miR-124 and TGFβ1, and a ceRNA regulatory network of lncRNA SND1-IT1, miR-124, and COL4A1 in mediating the process of EMT in GC and all these regulatory mechanisms to explain TGFβ1-induced EMT in GC. Nevertheless, our data still revealed an oncogenic role for lncRNA SND1-IT1 in GC tumorigenesis, which may offer a strategy for using the lncRNA-based strategy as a potential therapeutic target for patients with GC.

MATERIALS AND METHODS Human tissue specimens
The study included 52 paired samples of lesion tissues and lesion-free tissues from 52 GC patients who underwent gastrectomy at our hospital from February of 2018 to August of 2020. Among 52 patients, there were 42 males and 10 females, with age ranging from 32 to 68 years. No patient had received radiotherapy or chemotherapy before surgery. Sample collection was performed with informed consent for a protocol approved by the ethics committee of the hospital.
RNA extracts, reverse transcription, and real-time qPCR (RT-qPCR) Total RNA was extracted using TRIzol reagents (Takara, Dalian, China). Synthesis of cDNA was carried out using the PrimeScript RT reagent Kit (RR047A, Takara, Dalian, China) following the instructions provided by the manufacturer. RT-qPCR was run on with the SYBR®Premix ExTaqTM II (Perfect Real Time) kit (DRR081, Takara, Japan) and analyzed using the PCR instrument (ABI, Foster City, CA, USA), with each reaction run in triplicate.
Primers (Table 2) were generated by Sangon (Shanghai, China). MiR-124 expression level was normalized to U6 and SND1-IT1 expression level to GAPDH. Fold changes in expression were calculated by using 2 −ΔΔCt . ΔCt represents the difference between the threshold cycle (Ct) for each target and housekeeping mRNA or miRNA levels.

Immunoblotting
Total protein was obtained in the RIPA buffer (Beyotime Biotechnology), and 50 μg of protein samples for each run were loaded into the wells of 10% SDS-PAGE for separation and then wet-transferred onto the poly (vinylidene fluoride) (PVDF) membrane. The membrane immunoreacted with antibodies (Abcam, Cambridge, UK) to E-cadherin (ab15148), N-cadherin (ab18203), Collagen type IV (ab6586), and β-actin (ab227387). Immunoblots were visualized by horseradish peroxidaselabeled IgG with the enhanced chemiluminescence (ECL) reagents (BB-3501, Amersham Pharmacia, UK) and quantified using a gel documentation system (Bio-Rad Quantity One Software v4.6.2, Bio-Rad Laboratories, Richmond, California, USA). The gray value of the targeted proteins was normalized to that of β-actin.

Transwell assays
Cell migration and invasion assays were performed in a 24-well transwell chamber (8-mm pore size). In brief, HGC-27 cells were subject to serumfree starvation for 24 h, and they were resuspended into 3 × 10 5 cells/mL in serum-free DMEM. The suspension (100 μL) of HGC-27 cells were added into the upper chambers coated with (for cell invasion assays) or without (for cell migration assays) 50 μL Matrigel pre-incubated by a serum-free medium. Following 24-h incubation at 37°C, the cells that transferred to the lower chamber containing 500 μL 10% FBS-supplemented DMEM were subject to 4% paraformaldehyde fixation (10 min) and 0.1% crystal violet staining (10 min). Stained cells were counted in six random microscopic fields.

Statistics
Data were obtained from at least three replicates and expressed as mean ± standard deviation (SD). Student's t-test and one-way analysis of variance (ANOVA) with Tukey's adjustments were used for statistical comparison, with P < 0.05 as a level of statistical significance. Data entry, management, and analysis were conducted using SPSS 19.0 software (IBM, Armonk, NY, USA).