Long Noncoding RNA MALAT1 Acts as a Competing Endogenous RNA to Regulate TGF-β2 Induced Epithelial-Mesenchymal Transition of Lens Epithelial Cells by a MicroRNA-26a-Dependent Mechanism

The aim of the present study was to characterize whether the long noncoding RNA metastasis-associated lung adenocarcinoma transcript 1 (MALAT1)/miR-26a/Smad4 axis is involved in epithelial–mesenchymal transition (EMT) of lens epithelial cells (LECs). Primary human LECs were separated and cultured. Microarray analysis showed that a total of 568 lncRNAs are differentially expressed in primary HLECs in the presence of TGF-β2 and MALAT1 is mostly significantly dysregulated lncRNAs, which is increased by nearly 17-fold. In addition, upregulation of MALAT1 and downregulation of miR-26a were detected in human posterior capsule opacification (PCO) attached LECs and the LECs obtained from patients with anterior polar cataracts by quantitative RT-PCR (qRT-PCR). Next, our results showed that TGF-β2 induces overexpression of EMT markers in primary HLECs via a MALAT1-dependent mechanism. The mechanism is that MALAT1 negatively regulates miR-26a and miR-26a directly targets Smad4 by luciferase reporter assays and RNA-binding protein immunoprecipitation assay. In summary, TGF-β2 induces MALAT1 overexpression, which in turn MALAT1 acts as a ceRNA targeting Smad4 by binding miR-26a and promotes the progression of EMT of LECs.

It is well known that only ∼2% of RNA is a blueprint for proteins, and the remaining ∼98% serves as noncoding RNAs (ncRNAs) [6]. Long noncoding RNAs (lncRNAs) represents a novel class of RNAs which are longer than 200 nucleotides in length without functional protein-coding ability, which have attracted much attention recently [7]. Recent studies have suggested that the lncRNA metastasis-associated lung adenocarcinoma transcript 1 (MALAT1), located on chromosome 11q13, which is involved in the process of EMT in the cancer [8,9]. Moreover, the emerging evidence has revealed a potential contribution of particular lncRNAs to the development of PCO [10,11]. However, to the best of our knowledge, whether MALAT1 can regulate EMT of LECs and contribute to the pathogenesis of PCO has not been fully elucidated.
Recently, a novel regulatory mechanism, competing endogenous RNAs (ceRNAs), in which RNAs can engage in crosstalk via competing for shared miRNAs, has been proposed [15,16]. Our analysis, using the online software program star-Base v2.0 (http://starbase.sysu.edu.cn/starbase2/index.php), predicted that microRNA-26a formed complementary base pairing with MALAT1 [17]. Therefore, the aim of the present study was to explore whether the lncRNA MALAT1/miR-26a axis is involved in EMT of LECs.

Patient Lens Epithelial Cell Collection and Culture. Patient
LECs collection and culture have been described in detail previously [3][4][5] Primary human lens epithelial cells (HLECs) were cultured according to our previous procedure [4]. Primary HLECs were used to determine the role of lncRNA MALAT1/miR-26a axis in EMT of LECs.

SRA01/04
Cell Culture. Human lens epithelial cell line SRA01/04 was obtained from the Cancer Institute and Hospital, Chinese Academy of Medical Sciences (Beijing, China). The method for SRA01/04 cell culture was cultured according to our previous procedure [3][4][5].
SRA01/04 cells were only used for RNA immunoprecipitation (RIP) and the luciferase study.

Microarray Analysis.
For Microarray analysis, the primary HLECs were treated with TGF-2 (5 ng/ml) in 6-well plates for 48 h. Microarray analysis was performed by human lncRNA Array v2.0 (Arraystar, Rockville, MD, USA), which target differential expression of lncRNAs on the primary HLECs treated with TGF-2 (experiment) and it treated without TGF-2 (control).

Luciferase Assay.
Luciferase assay was performed according to our previous procedure [3][4][5]. The 3'-UTR of Smad4 mRNA and lncRNA MALAT1 containing the predicted miR-26a binding sites or corresponding mutant sites were amplified by PCR. Reporter activities were assessed 24 h after transfection using the dual-luciferase reporter assay system (Promega, Madison, WI, USA).
2.9. Statistical Analysis. All data are presented as the mean ± SE. The SPSS for Windows Version 17.0 (SPSS, Inc., Chicago, IL, USA) software was used for statistical testing [18]. Student's t test was used to determine differences between two independent groups [18]. One-way analysis of variance (ANOVA) and post hoc test of Tukey's multiple comparisons were used to determine differences among multiple groups [18]. P < 0.05 was considered statistically significant.

TGF-2 Induces the Different Expression of LncRNAs in Primary
HLECs. TGF-is considered to be a crucial inducer of EMT-related changes in PCO and anterior subcapsular cataracts [19]. TGF-2, which is one of TGFsuperfamily, has been proposed as the major isoform within the aqueous humor [20,21]. To identify the involvement of lncRNAs in PCO development, we analyzed the different expression of lncRNAs in primary HLECs in the presence or absence of TGF-2 using human lncRNA Array v2.0. The heat map was used to show that lncRNAs were differentially expressed between primary HLECs treated with 5 ng/ml TGF-2 (experiment) and it treated without TGF-2 (control) (Figure 1(a)). A total of 568 lncRNAs exhibited significant differential expression (fold change ≥1.5, P ≤0.05) including 368 upregulated lncRNAs and 200 downregulated lncRNAs. The results showed that the expression of XIST, CCAT1, NEAT1, and MALAT1 is top upregulated lncRNAs, and MALAT1 is mostly significantly dysregulated lncRNAs, which is increased by nearly 17-fold in the presence of TGF-2 ( Figure 1(a)). To confirm the microarray data, the top four upregulated lncRNAs were detected in human PCO-attached LECs and normal-attached LECs using qRT-PCR. MALAT1 expression was the most increased by nearly 18-fold in human PCO-attached LECs compared with normal attached LECs (Figure 1 Consistent with these results, MALAT1 was most significantly upregulated by nearly 14-fold in LECs obtained from patients with anterior polar cataracts compared with patients with nuclear cataracts (Figure 1(c) and Supplementary Figure S1 E, F, G, and H). Next, to further confirm the microarray results and analyze whether TGF-2 upregulates MALAT1 in primary HLECs, we detected the expression of MALAT1 by qRT-PCR. TGF-2 significantly induced MALAT1 overexpression in primary HLECs in a dose-dependent manner and a timedependent manner (Figures 1(d) and 1(e)).

Smad4 Is a Target of miR-26a in Primary HLECs.
Our previous studies have identified that miR-26b inhibit EMT of LECs [4]. MiR-26a and miR-26b are members of the miR-26 family [14]. Based on these, we hypothesized miR-26a may be involved in EMT of LECs. Firstly, to confirm these, the levels of miR-26a were determined using qRT-PCR. The current data showed that miR-26a expression was decreased by nearly 5-fold in human PCO-attached LECs compared with normal attached LECs and downregulated by nearly 3-fold in LECs obtained from patients with anterior polar cataracts compared with patients with nuclear cataracts (Figure 3(a) PCO attached LECs Normal-attached LECs * * * * and Supplementary Figure S2 A, B). Next, the protein levels of Smad4 induced by TGF-2 were downregulated in miR-26a-overexpressing primary HLECs (Figure 3(b)). Furthermore, the protein expression of Smad4 induced by TGF-2 was upregulated in primary HLECs treated with anti-miR-26a (Figure 3(c)). In accord with Western blot analysis, qRT-PCR showed that Smad4 mRNA induced by TGF-2 was decreased in miR-26a-overexpressing primary HLECs (Figure 3(d)). Moreover, Smad4 mRNA was increased in primary HLECs treated with anti-miR-26a in the presence of TGF-2 (Figure 3(d)). Importantly, an inverse correlation was found between miR-26a and Smad4 mRNA expression levels in patients (Supplementary Figure S2 C, D). The mechanism that miR-26a downregulated the expression of Smad4 may be miR-26a binding directly to 3'-UTR of Smad4 using miRanda (Figure 3(e)). Finally, we used luciferase reporter assays to demonstrate that miR-26a directly targets Smad4 in HLECs (Figure 3(e)).

MALAT1 Negatively Regulated the Expression of miR-26a.
LncRNAs and miRNAs usually act as a competing endogenous RNA (ceRNA) [15,16]. The bioinformatics analysis showed that miR-26a was the potential miRNA which interacts with MALAT1. Next, a dual luciferase reporter assay was performed and confirmed that MALAT1 contains a binding site for miR-26a (Figure 4(a)). Furthermore, the levels of miR-26a were significantly upregulated by knockdown of MALAT1 (Figure 4(b)). To further identify the negative regulation of miR-26a by MALAT1, primary HLECs were treated with pcDNA3.1-MALAT1 vector or pcDNA3.1-MALAT1mut vector. The expression of MALAT1 was increased by transfecting with the pcDNA3.1-MALAT1 vector and mut vector (Figure 4(c)). The levels of miR-26a were decreased in primary HLECs treated with pcDNA3.1-MALAT1 vector (Figure 4(d)). However, there was no difference from control in the expression of miR-26a when the primary HLECs were treated with the pcDNA3.1-MALAT1-mut vector (mutations in the miRNA-26a response elements) (Figure 4(d)). Moreover, the levels of MALAT1 were unchanged after ectopic expression or knockdown of miR-26a in primary HLECs by treatment with miR-26a mimics and anti-miR-26a (Figure 4(e)). Importantly, an inverse correlation was found between miR-26a and MALAT1 expression levels in patients (Supplementary Figure S3 A, B).
Previous studies have already demonstrated that Ago2 is the only member in AGO family with catalytic activity during the silencing processes of RNA-induced silencing complex (RISC) [22]. Since miRNAs have been demonstrated involved in RNA silencing in an Ago2-dependent manner, we next investigated whether MALAT1 regulated miR-26a in such a manner. To this end, we performed an anti-Ago2 RIP assay on LECs. The endogenous MALAT1 pull-down was specifically enriched in LECs which were transiently overexpressed miR-26a (Figure 4(f)). Furthermore, the endogenous MALAT1 pull-down was decreased in LECs which were transiently anti-miR-26a. Based on these results, we identified that MALAT1 negatively regulated the expression of miR-26a through "sponging" miR-26a.

Knockdown of MALAT1 Inhibits Smad4: A Target of miR-26a.
Since we have confirmed that MALAT1 negatively regulates miR-26a and miR-26a directly targets Smad4, we next explored whether MALAT1 was involved in Smad4 expression induced by TGF-2 via directly controlling miR-26a in primary HLECs. The results showed that knockdown of MALAT1 significantly downregulated the protein levels of Smad4 induced by TGF-2 ( Figure 5(a)). However, anti-miR-26a ameliorated these effects ( Figure 5(a)). Next, qRT-PCR also showed similar effects ( Figure 5(b)). Taken together, these results indicated that MALAT1 partially controls Smad4 expression induced by TGF-2 via competing with miRNA-26a.

LncRNA MALAT1/Smad4 Axis Is Involved in TGF-2
Induced EMT of LECs. These findings, which indicated that MALAT1 acts as a ceRNA targeting Smad4 by binding miR-26a, prompted us to investigate whether MALAT1/Smad4 axis is involved in TGF-2 induced EMT of LECs. Firstly, TGF-2-induced fibronectin in primary HLECs was suppressed by MALAT1 knockdown and Smad4 knockdown (Figure 6(a)). Next, TGF-2 inhibited the levels of Ecadherin in primary HLECs, but the tendency was reversed by MALAT1 siRNA and Smad4 siRNA (Figure 6(a)). These effects were enhanced by MALAT1 knockdown together with Smad4 knockdown simultaneously. Furthermore, qRT-PCR also showed similar effects (Figures 6(b) and 6(c)). In addition, the protein expression of fibronectin was increased by transfecting with the pcDNA3.1-MALAT1 vector; these effects were inhibited using Smad4 siRNA (Figure 6(d)). Moreover, the protein expression of E-cadherin was decreased by treatment with the pcDNA3.1-MALAT1 vector, but the tendency was reversed by Smad4 siRNA (Figure 6(d)). Finally, qRT-PCR also showed similar effects (Figures 6(e) and 6(f)). These data indicated that lncRNA MALAT1/Smad4 axis is involved in TGF-2 induced EMT of LECs.

Discussion
In the present study, our data clearly identified that a total of 568 lncRNAs are differentially expressed in primary HLECs in the presence of TGF-2 and MALAT1 is mostly significantly dysregulated lncRNAs, which is increased by nearly 17-fold. Next, our results showed that TGF-2 induces the expression of EMT markers in primary HLECs via a MALAT1-dependent mechanism. The mechanism is that MALAT1 negatively regulates miR-26a and miR-26a directly targets Smad4. Finally, we demonstrated that MALAT1/miR-26a/Smad4 axis is involved in TGF-2 induced EMT of LECs. Although lncRNAs are noncoding transcripts and lack protein-coding capability, they play a pivotal role in regulation of gene expression [23,24]. Especially, accumulating evidence shows that lncRNAs are involved in the process of EMT in the cancer, including bladder cancer, nasopharyngeal carcinoma, breast cancer, gastric cancer, and esophageal cancer [25,26]. It is well known that EMT of LECs is a crucial cause of PCO formation [1][2][3][4][5]. Importantly, TGF-2 is considered to be an important inducer of EMTrelated changes in PCO [19]. In the present study, we firstly confirmed that TGF-2 induces the expression of MALAT1 in the primary HLECs in a dose-dependent manner and a time-dependent manner. Next, we provided further evidence of MALAT1 being significantly upregulated in human PCOattached LECs and in LECs obtained from patients with anterior polar cataracts. Based on these results, we identified that MALAT1 is involved in the pathogenesis of PCO. Then, to explore whether MALAT1 is involved in TGF-2 induced EMT of LECs, EMT markers, including E-cadherin and fibronectin, were detected. TGF-2-induced fibronectin protein and mRNA were suppressed by MALAT1 knockdown and MALAT1 knockdown ameliorate downregulation of Ecadherin due to TGF-2. Taken together, TGF-2 induced the expression of EMT markers in primary HLECs via a MALAT1-dependent mechanism.
Our previous studies have confirmed that miR-26b inhibit EMT of LECs [4]. Based on the fact that miR-26a and miR-26b are members of the miR-26 family, we hypothesized that miR-26a may be involved in EMT of LECs. Firstly, these results, where miR-26a expression was decreased in human PCO-attached LECs and in LECs obtained from patients with anterior polar cataracts, confirmed that miR-26a is involved in the pathogenesis of PCO. Next, consistent with the previous study, we identified that miR-26a play a crucial role in TGF 2-stimulated EMT [14]. During EMT, several transcription factors, such as Smad and Snail, are activated by TGF-2 [27]. The canonical TGF-2/SMAD4 signaling pathway is that TGF-2 activates its receptor and in turn phosphorylates its downstream target SMAD2/3, which hetero-oligomerizes with Smad4 and translocates to the nucleus [27][28][29]. Then, they recruit cotranscriptional factors to transactivate or repress target genes, which promote transdifferentiation of LECs to attain a mesenchymal phenotype [27][28][29]. The previous study has confirmed that miR-26a downregulates Smad1 and Smad4 in differentiating myoblasts by directly targeting their 3'UTRs [30]. Consistent with the previous study, we confirmed that miR-26a affect the levels of Smad4 induced by TGF-2 via overexpressing or knockdown of miR-26a in primary HLECs, respectively [30]. The mechanism, which miR-26a downregulated the expression of Smad4, is miR-26a binding directly to 3'-UTR of Smad4 in primary HLECs.
Recently, a study demonstrated that MALAT1 could sponge miR-211 as a competing endogenous RNA to suppress tumor growth and progression in ovarian carcinoma [31]. MALAT1 exerted its oncogenic function in osteosarcoma as a ceRNA to suppress miR-34a expression and upregulate CCND1 [32]. Furthermore, MALAT1 can act as a competing endogenous RNA to modulate miR-124/STAT3 in NSCLC [33]. Enlightened by the ceRNAs regulatory network and based on the fact that MALAT1 contains a target site of miR-26a (Figure 4(a)), we hypothesized that MALAT1 may be as a ceRNA to compete and sink miR-26a expression.
It is interesting that Wang and Sun have demonstrated that MALAT1 contains a target site of miR-26a-5p and MALAT1 directly regulates miR-26a-5p in osteosarcoma cells [34]. Consistent with the previous studies, the current data also confirmed that MALAT1 negatively regulated the expression of miR-26a through "sponging" miR-26a in primary HLECs. Finally, we identified that MALAT1/miR-26a/Smad4 axis is involved in TGF-2 induced EMT of LECs.

Conclusion
In summary, the current study showed that TGF-2 induces EMT in primary HLECs via a MALAT1-dependent mechanism. The mechanism is that MALAT1 negatively regulates the expression of miR-26a through "sponging" miR-26a and in turn induces Smad4. Therefore, these results also indicated that MALAT1/miR-26a/Smad4 axis is involved in TGF-2 induced EMT of LECs and MALAT1 is a potential therapeutic target for the treatment of PCO.

Data Availability
All relevant data used to support the findings of this study are included within the article.   Fibronectin mRNA expression in primary HLECs was detected using qRT-PCR. * P < 0.05 compared with TGF-2 group ((b) and (c)). # P < 0.05 compared with TGF-2 group and TGF-2+MALAT1 siRNAs group ((b) and (c)). ## P < 0.05 compared with TGF-2+MALAT1 siRNAs group and TGF-2+MALAT1 siRNAs+Smad4 siRNAs group ((b) and (c)). (d, e, f) The primary HLECs were treated with TGF-2 (5 ng/ml) for 48 h before incubation with pcDNA3.1-MALAT1 vector or Smad4 siRNA for 24 h. (d) E-cadherin and fibronectin protein levels in primary HLECs were detected by Western blot analysis. (e) E-cadherin mRNA levels in primary HLECs were detected by qRT-PCR. (f) Fibronectin mRNA expression in primary HLECs was detected using qRT-PCR. * P < 0.05 compared with normal group and pcDNA3.1-control group ((e) and (f)). # P < 0.05 compared with pcDNA3.1-MALAT1 group ((e) and (f)). All of data are presented as the mean± SE of six independent experiments.

Conflicts of Interest
The author declares that they have no conflicts of interest. PCO attached LECs. (E, F, G, H) The expression of XIST, CCAT1, NEAT1, and MALAT1 was significantly increased in LECs obtained from patients with anterior polar cataracts (n=33) compared with patients with nuclear cataracts (n=33). * P < 0.05 compared with anterior polar cataracts. All of data are presented as the mean± SE of six independent experiments. Supplementary Figure