LncRNA MALAT1 Regulates miR-144-3p to Facilitate Epithelial-Mesenchymal Transition of Lens Epithelial Cells via the ROS/NRF2/Notch1/Snail Pathway

Diabetic cataract is a common complication of diabetes. The epithelial-mesenchymal transition (EMT) of lens epithelial cells (LECs) is a key event in the development of diabetic cataracts. Metastasis-associated lung adenocarcinoma transcript 1 (MALAT1) has been reported to be highly expressed in different tissues of diabetic patients. This study is aimed at investigating the function and mechanism of MALAT1 in the regulation of EMT in human LECs under high glucose conditions. MALAT1, α-smooth muscle actin (α-SMA), fibronectin (FN), and nuclear factor erythroid-derived 2-like 2 (NRF2) were highly expressed in the LECs of diabetic cataract patients and in the human LECs under high glucose conditions; meanwhile, the decreased expressions of E-cadherin and zonula occludens 1 (ZO-1) were detected. Knockdown of MALAT1 could significantly reduce ROS, prevent EMT, arrest S phase cell cycle, and suppress the expression of total NRF2 and its nucleus translocation in LECs. Furthermore, after NRF2 was knocked down, total NRF2, α-SMA, and FN in cells, and NRF2, Notch intracellular domain (NICD), and Snail were decreased in the nucleus. Using bioinformatics methods, we predicted that MALAT1 and NRF2 shared the same microRNA-144-3p (miR-144-3p) combining site. Luciferase reporter coupled with qRT-PCR assays revealed that miR-144-3p was a target of MALAT1, which was confirmed to downregulate miR-144-3p in the LECs. In addition, after transfection of miR-144-3p mimics or inhibitor, western blot assay demonstrated that miR-144-3p negatively regulated the expression of total NRF2, α-SMA, and FN in cells, and NRF2, NICD, and Snail in the nucleus without affecting Kelch-like ECH-associated protein 1 (KEAP1). Finally, we confirmed that transfection of shMALAT1 inhibited NRF2 expression, and its mediated EMT could be rescued by miR-144-3p inhibitor; transfection of pcDNA3.1-MALAT1 promoted NRF2 expression, and its mediated EMT could be reversed by miR-144-3p inhibitor. In summary, we demonstrate that MALAT1 regulates miR-144-3p to facilitate EMT of LECs via the ROS/NRF2/Notch1/Snail pathway.


Introduction
Diabetes is a metabolic disease characterized by elevated blood glucose. Complications of diabetes occur in all organ systems, such as diabetic nephropathy, diabetic neuropathy, diabetic retinopathy, and diabetic cataracts [1][2][3][4][5]. Diabetic patients suffer from cataracts earlier than nondiabetic patients [6], and cataract surgery on diabetic patients may lead to more complications, especially in hyperglycemia conditions [7]. Therefore, it is helpful to explore the pathogenesis of diabetic cataracts in order to find the most effective way to prevent them.
The proportions of cortical and subcapsular cataracts are higher in both type 1 and type 2 diabetic patients with cataracts than the proportions seen in age-related cataracts [8,9]. Posterior subcapsular cataracts occur because of abnormal cells and extracellular matrix under the lens posterior capsules in humans [10], and diabetic subcapsular cataracts in rats and mice are highly associated with the epithelial-mesenchymal transition (EMT) of lens epithelial cells (LECs) [11,12]. However, the mechanism of EMT in LECs in diabetic conditions is still unclear.
Long noncoding RNAs (lncRNAs) are defined as RNA molecules of more than 200 nucleotides without proteincoding capacity [13], which function in regulating the processes of apoptosis, autophagy, cell cycle, and EMT [14][15][16]. MALAT1 is a 6.5-knt lncRNA, which was first found in lung adenocarcinoma, playing significant roles in the pathophysiological process of diabetes and diabetes-related complications by modulating gene transcription [17]. It can promote oxidative stress in human LECs under high glucose conditions [16], and reactive oxygen species (ROS) have been reported to induce the EMT process in LECs of diabetic cataracts [18]. The role of MALAT1 in modulating ROS and EMT of LECs in diabetic cataracts remains unclear. KEAP1/NRF2 system is one of the main cellular defense mechanisms against oxidative stress in LECs. When ROS increases, KEAP1 is oxidized and covalently modified, and NRF2 translocates to the nucleus and induces the generation of antioxidant enzymes that protect cells from oxidative stress and apoptosis by reducing ROS accumulation in diabetic cataracts [19][20][21]. Meanwhile, nucleus translocation of NRF2 can regulate more than 100 genes, including Notch1, to promote cell proliferation, differentiation, and migration [22][23][24]. In hepatocellular carcinoma cells, ROS-induced nucleus translocation of NRF2 can activate the Notch1/Snail signaling pathway to accelerate EMT and metastasis of tumor cells [25]. This research implies that ROS may induce NRF2 nucleus translocation, subsequently trigger the Notch signaling pathway, and result in the EMT of LECs in the formation of diabetic cataracts.
The theory of competing endogenous RNA (ceRNA) proposes that lncRNA and mRNA transcripts can affect each other by competitively combining with a miRNA response element (MRE) to influence posttranscription [26]. MALAT1 can regulate target genes by acting as a "sponge" of some miRNAs [27,28], miR-144-3p in osteosarcoma cells that are regulated by MALAT1 function as competing endogenous RNAs (ceRNAs) [29]. It has been reported that miR-144-3p is widely expressed in the tissues of diabetic patients, and that it may negatively regulate NRF2 expression in retinal pigment epithelium (RPE) cells, plasma, retinal endothelial cells, and epicardial adipose tissues [30][31][32][33]. Taking all of these findings into consideration, we hypothesize that MALAT1 may promote ROS expression, induce nucleus translocation of NRF2, and regulate NRF2 expression via a competing endogenous RNA of miR-144-3p.
In the present study, we first confirmed that MALAT1 and NRF2 are highly expressed in LECs of diabetic patients with cataracts and in LECs in a hyperglycemic environment. Second, we found that knockdown of MALAT1 could reduce the level of ROS, expression of NRF2, nucleus translocation of NRF2, and EMT of LECs in a high-glucose environment. Third, we discovered that knockdown of NRF2 reduced EMT by blocking the Notch1/Snail pathway. Most importantly, we found that MALAT1 competitively sequestered miR-144-3p and relieved the inhibitory effect of miR-144-3p on NRF2, increasing NRF2 expression. Finally, our study clarified a novel role of MALAT1 in regulating miR-144-3p, ROS, and then activating the NRF2/Notch1 pathway to facilitate diabetic cataracts, and suggested that MALAT1 might be a target for the prevention of diabetic cataracts.

Materials and Methods
2.1. Human Anterior Lens Capsules. Anterior lens capsules with LECs were collected from diabetes patients with cataracts (diabetic cataract (DC) group, n = 18 eyes) and patients with age-related cataracts (control (CON) group, n = 18 eyes) during cataract surgery. The characteristics of all the patients are listed in Table 1. All donors provided written informed consent, and the protocol was performed in accordance with the principles of the Declaration of Helsinki and approved by the Institutional Research Ethics Committee of the Fourth Military Medical University.

Antibodies.
Primary and secondary antibodies used in this study are listed in Table 2. 2.3. Cell Culture. In human LECs, HLE-B3 (human lens epithelial, ATCC CRL-11421TM) cells were purchased from American Type Culture Collection (ATCC, Manassas, VA, USA). The cell line was authenticated by Qingke Biological Technology Company via short tandem repeat (STR) profiling to verify the human unique DNA profile and rule out intraspecies contamination. The STR analysis was performed on 20 core STR loci.
HLE-B3 cells were cultured in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (HyClone, Logan, Utah, USA), 100 mg/ml streptomycin, and 100 IU/ml penicillin (Gibco, Grand Island, New York, USA), and were incubated at 37°C in a humidified atmosphere containing 5% CO 2 . Culture medium was replaced every two days. When the cells were at 75-80% confluence, they were treated with 0.25% trypsin-0.02% EDTA solution and passaged. Cell number was counted by an automated cell counter (Ruiyu, Shanghai, China).
HLE-B3 cells were divided randomly into a normal control group (NC, 5.5 mM glucose) and high-glucose groups (25.5 mM glucose and 35.5 mM glucose). In the NC group, HLE-B3 cells were cultured in DMEM with 5.5 mM glucose, while in the high-glucose groups, the cells cultured in DMEM were treated with 20 mM glucose and 30 mM glucose, respectively, for 24 h and 48 h.
2.6. Western Blot Analysis. In brief, HLE-B3 cells were harvested and lysed by cell lysis buffer (Beyotime, Shanghai, China). Proteins in the cytoplasm and nucleus were extracted by cytoplasmic nuclear separation kit (Beyotime, Shanghai, China). Protein concentration was determined by BCA protein detection reagent (Beyotime, Shanghai, China). 30 μg/lane proteins were loaded on 10% or 15% SDS-PAGE gel for protein separation, after which the separated proteins were transferred to PVDF membranes. After being blocked at room temperature for 1 h, the membranes were probed with the primary antibodies (anti-ZO-1, anti-E-cadherin, anti-FN, anti-NRF2, anti-KEAP1, anti-α-SMA, anti-β-tubulin, anti-NICD, and anti-Snail) at 4°C overnight. After being incubated with HRP-conjugated secondary antibody for 1 h, the protein bands were visualized with enhanced chemiluminescence (ECL, ZETA, Beijing, China). Relative expressions of proteins were determined by densitometer and expressed as absorbance units. Each experiment was repeated three times.
2.8. Transwell Assay. Transwell chambers with 8 μm pore inserts (Millipore, Danvers, Massachusetts, USA) were used to investigate cell migration. For cell migration assay, 2 × 10 4 HLE-B3 cells or different transfected HLE-B3 cells were seeded in the upper chambers with 200 μl of serum-free DMEM (5.5 mM glucose), while in the lower chambers, the medium was 600 μl DEME in addition to 30 mM glucose (35.5 mM glucose) and 10% fetal bovine serum. After 24 h incubation, the cells in the upper chamber were carefully wiped off with a cotton swab. Next, the filters were fixed in 4% paraformaldehyde for 30 min and stained with 0.1% crystal violet for 20 min. Stained cells in five random fields per chamber were counted under an inverted microscope (Olympus, Nagoya, Japan). Every experiment was repeated three times independently.
2.9. Wound Healing Assay. Wound healing assays were performed to detect cell migration. First, 1 × 10 5 HLE-B3 cells, or different transfected HLE-B3 cells, were seeded in sixwell plates and reached 100% confluence overnight. A pipette tip of 1 ml was used to scratch the monolayer cells to make an artificial wound. After rinsing with PBS three times to remove the detached cells, the cells on the plates were cultured in DMEM with 5.5 mM glucose or 35.5 mM glucose for 24 h. Wound width was detected before treatment (0 h) and 24 h after treatment. Images were obtained by an inverted microscope (Olympus, Nagoya, Japan). Wound closure = wound width ð0 hÞ − wound width ð24 hÞ/wound width ð0 hÞ.
2.10. Immunofluorescence Staining Assay. Human anterior lens capsules with lens epithelial cells obtained from the cataract surgery were fixed with 4% polyformaldehyde immediately for 30 min. HLE-B3 cells were cultured and treated in a 12-well plate with a cover glass to prepare the cell slide. After different treatments, the cells on the cover glass were fixed with 4% polyformaldehyde for 30 min. The capsular specimens and cultured cells were washed with PBS three times, 5 min for each wash, and then treated with 1% Triton X-100 in PBS for 30 min to enhance the permeability of the cell membranes. In order to reduce the nonspecific immunoreactions, 1% bovine serum albumin in PBS was dropped on the cells for 30 min at room temperature, and then the capsules and cultured cells were incubated with the primary The gender and age between the CON and DC groups have no statistically significant difference. The blood glucose in the DC group is higher than that in the CON group (P < 0:0001).

Oxidative Medicine and Cellular Longevity
Synthesized by RiboBio (Guangzhou, China; Lot N0108) U6 Synthesized by Sangon Biotech (Shanghai, China; Lot B532451-0020) antibodies (anti-α-SMA, anti-E-cadherin, and anti-NRF2 for capsules; anti-FN, anti-NRF2, and anti-Snail for cultured cells) at 4°C overnight. After incubation with fluorophoreconjugated secondary antibody for 1 h, the cell nuclei were stained with 4 ′ ,6-diamidino-2-phenylindole (DAPI). Under the confocal scanning laser microscope (FV1000, Nagoya, Olympus, Japan), the stained cells were observed and the images were acquired. The information about primary and secondary antibodies used in this article is listed in Table 2.
2.11. ROS Detection Assay. Transfected HLE-B3 cells were incubated with 2 ′ ,7 ′ -dichlorofluorescein diacetate (DCFH-DA) (MCE, New Jersey, USA) for 30 min. After washing with cold PBS three times, the cells were observed, and the images were acquired under the confocal scanning laser microscope (FV1000, Nagoya, Olympus, Japan). The cells stained by DCFH-DA were harvested and washed with cold PBS, and then suspended in 1 ml PBS. The fluorescence intensity of the cells was measured by flow cytometry (BD, San Jose, CA, USA).

Cell Cycle Assay.
The treated cells were washed by cold PBS and fixed with 70% alcohol for 4 h. Then, the cells were stained with propidium iodide for 10 min at 37°C in darkness. Flow cytometry (BD, San Jose, CA, USA) was used to detect the percentage of cells in G0/G1, S, and G2/M phases.
2.13. Statistical Analysis. All data were expressed as the mean ± standard deviation and analyzed with SPSS statistical analysis software, version 18.0 (Chicago, IL, USA). For comparisons of multiple groups at different time points, statistical analysis was performed using repeated measures analysis of variance followed by Tukey's post hoc test. For groups at the same time point, Student's t-test was used. P < 0:05 was considered to indicate a statistically significant difference.

Results
3.1. The Expression of FN, α-SMA, E-cadherin, and ZO-1 in the Lens Epithelium of Diabetic Cataracts and High-Glucose-Treated HLE-B3 Cells. Compared with age-related cataracts (CON), the expressions of α-SMA and FN on RNA level in the lens epithelium of diabetic cataracts (DC) were significantly increased; meanwhile, the expressions of epithelial cell markers E-cadherin and ZO-1 were significantly decreased (Figure 1(a)). Immunofluorescence staining assay showed that the expression of α-SMA was higher in diabetic cataract lens epithelium tissues. However, the epithelial cell marker E-cadherin was lower in lens epithelium of diabetic cataracts compared to age-related cataracts ( Figure 1(b)).
In HLE-B3 cells, qRT-PCR and western blot analysis showed that the expressions of α-SMA and FN increased gradually and significantly with the concentration (25.5 mM and 35.5 mM) and duration of high glucose (24 h and 48 h) on both RNA and protein levels compared with the control group (5.5 mM); meanwhile, the expressions of epithelial cell markers E-cadherin and ZO-1 were significantly decreased (Figures 1(c)-1(g)). These results suggested the occurrence of EMT in human LECs of diabetic cataracts.

Increased Expression Levels of MALAT1 and NRF2 in the
Lens Epithelium of Diabetic Cataracts and High-Glucose-Treated HLE-B3 Cells. The expression of MALAT1 level in the LECs of diabetic cataracts was significantly increased compared with age-related cataracts (CON) (Figure 2(a)). In HLE-B3 cells stimulated by high glucose, the expression of MALAT1 increased significantly with the concentration of glucose (25.5 mM and 35.5 mM) compared with the control group (5.5 mM) at 24 h and was persistent at 48 h (Figure 2(b)).
Immunofluorescence staining assay showed that the expression of NRF2 was higher in lens epithelium of diabetic cataracts compared with age-related cataracts (Figure 2(c)). In HLE-B3 cells, the expression of NRF2 also elevated gradually with increasing glucose concentration on both protein and RNA levels compared with the control group (Figures 2(d)-2(f)). This was confirmed by the data showing the increase of NRF2 fluorescence intensity in the immunofluorescence staining assay (Figure 2(g)). Based on the above data, we selected the treatment of 35.5 mM glucose and 24 h for the following experiments.

Downregulation of MALAT1 in HLE-B3 Cells in a
Hyperglycemic Environment Suppressed ROS Production, NRF2 Expression, and Its Nucleus Translocation, Arrested S Phase Cell Cycle, and Attenuated EMT. It has been reported that ROS plays an important role in the pathogenesis of EMT [34,35]. To define whether upregulated MALAT1 is associated with production of ROS in high-glucose-treated HLE-B3 cells, we knocked down MALAT1 by transfection with sh-MALAT1 in the cells. As shown in Figure 3(a), among the four sh-MALAT1s prepared, the first shMALAT1 was the most effective at downregulating MALAT1, as revealed by qRT-PCR. So therefore, the first shMALAT1 was selected for the following experiments. After treatment with high glucose for 24 h, ROS levels in the cells increased significantly as shown by flow cytometry and immunofluorescence assays, and notably could be inhibited by shMA-LAT1 (Figures 3(b)-3(d)), suggesting that knockdown of MALAT1 was capable of preventing the production of ROS in HLE-B3 cells.
When shMALAT1 was transfected in high-glucoseinduced HLE-B3 cells, the elevated expressions of FN and α-SMA were suppressed significantly (Figures 3(e) and 3(f)). In addition, the cell migration of high-glucosestimulated cells increased significantly with Transwell and wound healing assays, while the transfection of shMA-LAT1 abolished the migration of HLE-B3 cells in high glucose conditions (Figures 3(g)-3(j)).
Cell cycle analysis demonstrated that the cells could be retarded in G0/G1 phase after treatment with high glucose. While after transfection with MALAT1, the percentage of cell in S phase increased and the percentage in G0/G1 phase reduced compared to NC (Figures 3(k) and 3(l)).
In this set of experiments, western blot assay showed that the expressions of both total NRF2 and NRF2 in the nucleus

Downregulation of NRF2 Prevented EMT and Cell
Migration in High-Glucose-Treated HLE-B3 Cells. Previous studies have shown that NRF2 contributed to EMT through activation of the Notch signaling pathway [25]. Notch1 is one of the transmembrane Notch family receptors, which can drive Notch signaling when combined with the Rbpjκ transcription factor. After canonically accepting ligands, the receptor undergoes cleavage to yield the NICD that translocates to the nucleus. It has been clearly shown that NRF2 can activate the Notch1 pathway [36]. The Notch1 signaling pathway can regulate the process of EMT by regulating the Notch1/Snail signaling pathway [37,38]. The efficacy of shNRF2 on knockdown of NRF2 in high-glucose-treated HLE-B3 cells was confirmed with western blot and immunofluorescence staining assays (Figures 4(a)-4(c)). We found that the transfection of shNRF2 could reduce the expression levels of NICD and Snail in the nucleus of HLE-B3 cells under high glucose conditions (Figures 4(a) and 4(b)). The reduction of Snail in the nucleus was confirmed by immunofluorescence staining assay (Figure 3(c)). Importantly, knockdown of NRF2 could suppress the expressions of α-SMA and FN (Figures 4(d) and 4(e)) and prevent cell migration in high-glucose-treated HLE-B3 cells (Figures 4(f)-4(i)).
Taken together, NRF2 may function as one of the downstream of MALAT1 in regulating EMT in high-glucosetreated HLE-B3 cells via the Notch/Snail pathway.
To investigate the role of miR-144-3p on NRF2 expression and EMT in high-glucose-treated HLE-B3 cells, we examined the expressions of FN, NRF2, α-SMA, and KEAP1 in the cells and furthermore the expressions of NRF2, NICD, and Snail in the nucleus. The expression of total NRF2 and expressions of NRF2, NICD, and Snail in the nucleus were all downregulated in high-glucose-stimulated HLE-B3 cells treated with the miR-144-3p mimics. In contrast, these protein expressions could be upregulated by miR-144-3p inhibitor. Interestingly, the expression of KEAP1 had no obvious changes between these groups (Figures 5(j) and 5(k)). In addition, the expressions of FN and α-SMA were downregulated by miR-144-3p mimics and were upregulated by miR-144-3p inhibitor in high-glucose-stimulated HLE-B3 cells (Figures 4(l) and 4(m)).
Thus, it is concluded that MALAT1 could negatively regulate miR-144-3p, and through the negative regulation of miR-144-3p on NRF2, it could be involved in EMT of HEL-B3 cells in a high-glucose environment.

MALAT1 Promoted EMT and Cell Migration through
Reducing miR-144-3p and Activating the NRF2/Notch1/ Snail Signaling Pathway. First, we found that transfection of shMALAT1 in high-glucose-treated HLE-B3 cells resulted in downregulation of total NRF2 in cells as well as downregulation of NRF2, NICD, and Snail in the nucleus compared with levels in the transfection with the shMALAT1 NC group and high-glucose group. Conversely, suppressions of these proteins could be partially rescued by cotransfecting miR-144-3p inhibitor, but not by miR-144-3p mimics (Figures 6(a) and 6(b)). Furthermore, upregulation of MALAT1 by transfecting pcDNA3.1-MALAT1 in highglucose-treated HLE-B3 cells resulted in increases of total NRF2 in cells, as well as NRF2, NICD, and Snail increases in the nucleus compared with those in the transfection with the pcDNA3.1 and high-glucose groups. Conversely, increases of these proteins could be partially reversed by cotransfecting miR-144-3p mimics, but not by miR-144-3p inhibitor (Figures 6(c) and 6(d)).
We used Transwell assay to determine whether MALAT1 and miR-144-3p could regulate cell migration in highglucose-treated HLE-B3 cells. Compared with the control groups (transfection of the pcDNA3.1 group and highglucose group), cell migration was promoted after transfection with pcDNA3.1-MALAT1 in high-glucose-treated HLE-B3 cells, which could be significantly reversed by transfection of a miR-144-3p mimics, but not by transfection of a miR-144-3p inhibitor. Compared with cells transfected with shMALAT1 NC, cell migration was inhibited in the cells transfected with shMALAT1, which could be promoted by cotransfecting miR-144-3p inhibitor, but not by cotransfecting miR-144-3p mimics (Figures 6(i) and 6(j)).

Discussion
In the current study, we found that MALAT1 promoted the EMT of LECs by negatively regulating miR-144-3p and subsequently activating the ROS/NRF2/Notch1/Snail signaling pathway under high glucose conditions. MALAT1 is widely upregulated in the aqueous humor, lens epithelium, and fibrovascular membranes of diabetic patients [41]. It can promote EMT by activating Wnt, Notch1, and Akt/mTOR signaling in cancer tissues and cells [42][43][44]. It also can facilitate EMT in HK-2 cells in diabetic nephropathy [45]. Consistent with previous studies, we also found that MALAT1 was aberrantly expressed in LECs of diabetic cataract patients and in high glucose conditions.
In this study, we demonstrated that MALAT1 could induce ROS production and NRF2 expression in LECs. However, previous studies have shown that MALAT1 can reduce the expression of NRF2 and promote the production of ROS [35]. On the contrary, some studies have shown that MALAT1 can reduce the production of ROS by activating NRF2 [46]. The role of MALAT1 in

20
Oxidative Medicine and Cellular Longevity regulating NRF2 and ROS seems controversial. We think that MALAT1 may control the ROS level in balance by promoting NRF2 expression. ROS are highly expressed in LECs during the onset of diabetic cataracts stimulated by high glucose [18,21], and could induce EMT by activating several transcription factors, such as Snail, STAT3, and ZEB1, which can be activated by the Notch, G-CSF, MAPK, and PI3K/Akt signaling pathways in cancer cells [47]. It is well known that the classic signaling pathway of KEAP1/NRF2 is the main approach against ROS [48]. KEAP1 binds with NRF2 by acting as an adaptor protein and is a negative regulator of NRF2. Under normal ROS conditions, this interaction is a short incident, with NRF2 exhibiting a short half-life of 13-21 min [49,50]. While under hyperactive ROS conditions, because of modification of critical cysteine residues in KEAP1 or NRF2 coupled with phosphorylation, NRF2 can detach from KEAP1 and then translocate into the nucleus to promote the transcription of target genes by binding to promoter regions [21]. NRF2 can also be regulated by posttranslational, transcriptional, translational, and epigenetic mechanisms, as well as by several microRNAs [51]. In recent years, many studies have shown that ROS can induce NRF2 into the nucleus to decrease the sensitivity of cancer to radiotherapy and increase the metastasis of cancer [25,[52][53][54]. NRF2 expression is low in both epithelial and mesenchymal cells, while NRF2 expression is high in mixed epithelial and mesenchymal phenotypes, suggesting that the highly expressed NRF2 in cells may activate EMT [52]. Excessive ROS can lead to apoptosis in LECs, while NRF2 may reduce apoptosis through antioxidant production [19,55]. In our study, we found that high-glucose-induced ROS may promote NRF2 nucleus translocation. Increased nucleus translocation of NRF2 can promote EMT in LECs, which may occur through the Notch1/Snail pathway. So this study indicates that the increase of NRF2 expression may not always be beneficial in the prevention of cataracts.
Previous studies have indicated that MALAT1 can induce EMT in cancer cells by acting as ceRNAs to sink miR-141, miR-3064-5p, miR-200b, and miR-144-3p [28,29,56,57]. NRF2 can be negatively regulated by miRNAs, including miR-153, miR-27a, miR-142-5p, and miR-144-3p, through direct repression of NRF2 messenger RNA in a KEAP1independent manner, or by enhancing KEAP1 expression. In the present study, we used bioinformatics software to predict that MALAT1 and NRF2 shared the same miR-144-3p binding sites. Additionally, MALAT1 can inhibit the expression of miR-144-3p, which can downregulate NRF2 expression independent of KEAP1. The miR-144-3p expression and function in diabetic cataracts have not been reported before. We detected that miR-144-3p was downregulated in LEC lining with anterior lens capsules and in LECs under high glucose conditions. Even though the highly expressed miR-144-3p can inhibit the EMT of tumor tissues and cells by targeting EZH2, PBX3, and MAP3K8 [58][59][60], we found that EMT could be negatively regulated by miR-144-3p through the ROS/NRF2/Notch1/Snail pathway in highglucose-stimulated LECs.

Conclusions
In summary, we have found that MALAT1 is upregulated in LECs under high glucose conditions. Upregulated MALAT1 can promote ROS production and competitively combine with miR-144-3p. Downregulation of miR-144-3p can promote NRF2 expression. NRF2 could be stimulated by ROS, translocate to the nucleus, and activate the Notch1/Snail pathway, resulting in the EMT of LECs (Figure 7). Therefore, MALAT1 may be a potential target for the prevention and treatment of diabetic cataracts.

Data Availability
All data related to this paper may also be requested from the corresponding author.

Conflicts of Interest
On behalf of all authors, the corresponding author states that there is no conflict of interest.

Authors' Contributions
Wei Ye, Jiyuan Ma, and Fang Wang contributed equally to this work. Jian Zhou reviewed the manuscript. Jian Zhou conceived the experiments and provided thoughtful comments and suggestions during this study. All authors read and approved the final manuscript.