LIN28A blocks the EndMT process of high glucose‐induced HRMECs by stabilizing SIRT6 mRNA

ABSTRACT

Background

Diabetic retinopathy (DR) is a microvascular condition resulting from microangiopathy, causing gradual retinal damage and potential blindness. Endothelial–mesenchymal transition (EndMT) serves an important function in DR development. Exploring the molecular mechanism of EndMT in DR is needed.

Methods

Human retinal microvascular endothelial cells (HRMECs) were incubated with high glucose (HG) to induce an in vitro DR model. Lentivirus and small‐interfering RNAs were used to construct SIRT6 and LIN28A overexpression or knockdown in HRMECs, respectively. AMPK inhibitor, compound C, was used to block AMPK signaling in HRMECs. α‐SMA and PECAM1 levels were identified using immunofluorescence. CCK8 and transwell were used to detect cell viability and migration, respectively. The mRNA stability of SIRT6 was analyzed after HRMECs were exposed to actinomycin D. RNA immunoprecipitation was applied to verify the binding relationship between LIN28A and SIRT6 mRNA in HRMECs.

Results

In HG‐induced HRMECs, the cells undergo the EndMT process, and SIRT6 levels are downregulated. HG treatment reduced the level of p‐AMPK in HRMECs, and compound C reversed the inhibition of SIRT6 overexpression on EndMT in HG‐induced HRMECs. By binding to SIRT6 mRNA, LIN28A could enhance SIRT6 mRNA stability. Additionally, LIN28A overexpression inhibited EndMT in HG‐induced HRMECs, which was reversed by SIRT6 knockdown.

Conclusion

The stabilization of SIRT6 mRNA by LIN28A activated AMPK signaling, inhibiting the EndMT process in HRMECs caused by HG.

LIN28A promotes mRNA stability of SIRT6 to activate the AMPK signaling, thereby inhibiting the EndMT process induced by HG in HRMECs.

Abbreviations

AMPK

AMP‐activated protein kinase

ANOVA

analysis of variance

DR

diabetic retinopathy

EndMT

endothelial–mesenchymal transition

HG

high glucose

HRMECs

human retinal microvascular endothelial cells

LIN28A

Lin‐28 homolog A

RIP

RNA immunoprecipitation

RT‐qPCR

real‐time quantitative polymerase chain reaction

SD

standard deviation

SIRT6

Sirtuin 6

INTRODUCTION

As a microvascular disease stemming from microangiopathy, diabetic retinopathy (DR) is a critical and severe complication associated with diabetes mellitus (DM), potentially causing progressive retinal damage and ultimately blindness ¹ . It is estimated that 27.0% of diabetic individuals worldwide suffer from DR, resulting in around 0.4 million instances of blindness, especially among the working‐age adult population ² . The growing number of diabetes patients worldwide highlights the urgency of in‐depth research on the pathogenesis of DR and the development of effective prevention strategies.

Retinal vascular endothelial cells exhibit heightened vulnerability to damage from elevated blood glucose levels, particularly when subjected to persistent glucose exposure in the bloodstream ³ . This damage is the primary cause of DR and other diabetes‐related complications ⁴ , ⁵ . Hyperglycemia disrupts homeostatic regulation, inducing alterations in endothelial cells that result in dysfunction and subsequent tissue modifications ⁶ . Endothelial–mesenchymal transition (EndMT) refers to the process where endothelial cells shed their epithelial traits and transform into mesenchymal cells, which is a pathophysiological process considered a unique form and is present in numerous pathological states, particularly in vascular endothelial cells under various pathological conditions ⁷ , ⁸ . EndMT may be implicated in the pathological mechanisms underlying DR. For instance, melatonin mitigated DR by modulating the EndMT of retinal vascular endothelial cells ⁹ . In the context of DR, the methylation of lncRNA MEG3 promoted EndMT, thereby worsening disease progression ¹⁰ . Despite these insights, the cellular mechanisms underpinning EndMT in DR remain inadequately understood, necessitating further investigation.

Sirtuin 6 (SIRT6), an NAD+‐dependent histone deacetylase, has garnered considerable attention for its involvement in the regulation of transcription processes associated with glucose metabolism and the maintenance of inflammatory homeostasis ¹¹ . Noticeably, a study involving 52 patients with DR, 25 individuals without DR, and 27 healthy controls demonstrated that SIRT6 expression was reduced in DR, correlating with its pathogenicity ¹² . Moreover, the downregulation of SIRT6 induced by oxidative stress may contribute to the pathogenesis of DR ¹³ . Evidence has suggested that SIRT6 inhibited EndMT by reducing vascular endothelial inflammation ¹⁴ . In comparison to vascular endothelial cells derived from young mice, SIRT6 was found to inhibit EndMT in aged vascular endothelial cells ¹⁵ . Nevertheless, it is currently unknown whether SIRT6 can inhibit the EndMT process in DR. Extensive research has highlighted the important role of AMP‐activated protein kinase (AMPK) in DM, which reduces the concentration of blood glucose flowing into cells, improves the sensitivity of cells to insulin, and reduces insulin resistance, thus playing a therapeutic role in diabetes‐related diseases ¹⁶ . Importantly, AMPK ameliorated diabetes‐induced endothelial injury by inhibiting EndMT. Activating AMPK is an efficient tactic for DR treatment ¹⁷ . What is more, SIRT6 has been widely reported to activate the AMPK signaling ¹⁸ , and this activation can prevent the EndMT process ¹⁷ , implying that SIRT6 may inhibit EndMT by triggering AMPK signaling.

RNA‐binding proteins (RBPs) are the core molecules that regulate the life cycle of RNA, affecting the fate or function of the bound RNA, and play a key role from posttranscriptional processing to final degradation ¹⁹ . RBPs have been reported to be involved in eye development and the pathogenesis of diseases, as exemplified by HuR, which stabilizes VEGF mRNA to promote DR ²⁰ . Lin‐28 homolog A (LIN28A) was reported to be downregulated in high glucose (HG)‐induced ARPE‐19 cells, involved in DR progression, and specifically alleviated HG‐induced damage of retinal pigment epithelium through the activation of SIRT1‐dependent autophagy ²¹ . Through the POSTAR database prediction, we found that LIN28A was a potential RBP of SIRT6 mRNA. Therefore, it is suggested that the reduction of SIRT6 mRNA level in DR may be affected by the downregulation of LIN28A.

Hence, we proposed a hypothesis that by stabilizing SIRT6 mRNA, LIN28A promoted the activation of AMPK signaling, which consequently inhibited the EndMT process in HRMECs under HG conditions. The results may enhance comprehension of the DR EndMT mechanism and offer new insights for DR treatment.

METHODS

Cell culture and treatment

Primary HRMECs cells were obtained from the Procell System (Wuhan, China) and cultured in the endothelial cell medium (Gibco, Grand Island, USA), supplemented with 1% penicillin/streptomycin (Gibco), 1% endothelial cell growth supplement (Gibco), and 5% fetal bovine serum (Gibco), under 5% CO₂ at 37°C. HRMECs at a concentration of 5 × 10⁵/mL were placed in 12‐well plates. Following a 24 h starvation period, the cells were exposed to a particular concentration of glucose, which included high glucose (HG, 25 mmol/L D‐glucose) and normal glucose (Control, 5 mmol/L D‐glucose) ²² , the D‐glucose was purchased from Sigma‐Aldrich (Bellefonte, USA). In addition, treat HRMECs with 10 μM compound C (MedChem Express, USA) for 30 min to inhibit AMPK.

Lentivirus infections and small interfering RNA (siRNA) transfections

The lentivirus containing the SIRT6 overexpressing vector (oe‐SIRT6), LIN28A overexpressing vector (oe‐LIN28A), and control vector (oe‐vector) were obtained from Sangon Biotech Co. Ltd. (Shanghai, China). Lentiviruses were added to HRMECs with Polybrene (5 μg/mL; Genepharma, Shanghai, China). The transduced HRMECs were subsequently selected using puromycin (5 μg/mL; Sigma‐Aldrich) for further experiments. For knockdown of SIRT6 and LIN28A, siRNAs targeting SIRT6 (si‐SIRT6; guide: 5′‐ACAUUCUUCCACAAACAUGUU‐3′, passenger: 5′‐CAUGUUUGUGGAAGAAUGUGC‐3′), LIN28A (si‐LIN28A; guide: 5′‐ACUUCUUAAAGGUGAACUCCA‐3′, passenger: 5′‐GAGUUCACCUUUAAGAAGUCA‐3′), and AMPK (si‐AMPK; guide: 5′‐ACAGAUAUAAUCAAAUAGCUC‐3′, passenger: 5′‐GCUAUUUGAUUAUAUCUGUAA‐3′) molecules were chemically synthesized by Genechem Company (Shanghai, China) and transfected using Lipofectamine™ 3,000 (Thermo Fisher Scientific, USA). The nonspecific siRNA (si‐NC; guide: 5′‐UUCUCCGAACGUGUCACGUTT‐3′, passenger: 5′‐ACGUGACACGUUCGGAGAATT‐3′) consisted of a nontarget sequence used as a control.

Real‐time quantitative polymerase chain reaction (RT‐qPCR)

Total RNA was extracted using Trizol reagent (Invitrogen, USA), and then converted into cDNA with a PrimeScript™ RT Reagent Kit (Takara, Kyoto, Japan). mRNA levels were measured using the SYBR Green PCR Kit (Sigma‐Aldrich) on an Applied Biosystems 7,500 Real‐Time system (Applied Biosystems, CA, USA). β‐actin was used as the internal reference for mRNA. The primer sequences are listed as follows:

LIN28A‐F: 5′‐GTATTGGGAGTGAGAGGCGG‐3′, LIN28A‐R: 5′‐TAGGTTGGCTTTCCCTGTGC‐3′; SIRT6‐F: 5′‐CCAAGTTCGACACCACCTTT‐3′, SIRT6‐R: 5′‐CGGACGTACTGCGTCTTACA‐3′; S100A4‐161‐F: 5′‐GGTGTCCACCTTCCACAAGT‐3′, S100A4‐161‐R: 5′‐TGTTGCTGTCCAAGTTGCTC‐3′; α‐SMA‐100‐F: 5′‐TATCCCCGGGACTAAGACGGG‐3′, α‐SMA‐100‐R: 5′‐CAGAGCCCAGAGCCATTGTC‐3′; PECAM1‐170‐F: 5′‐GATGCCCAGTTTGAGGTCAT‐3′, PECAM1‐170‐R: 5′‐ACGTCTTCAGTGGGGTTGTC‐3′; CDH5‐189‐F:: 5′‐CCAAGTTCGACACCACCTTT‐3′, SIRT6‐R: 5′‐CGGACGTACTGCGTCTTACA‐3′; S100A4‐161‐F: 5′‐GGTGTCCACCTTCCACAAGT‐3′, S100A4‐161‐R: 5′‐TGTTGCTGTCCAAGTTGCTC‐3′; α‐SMA‐100‐F: 5′‐TATCCCCGGGACTAAGACGGG‐3′, α‐SMA‐100‐R: 5′‐CAGAGCCCAGAGCCATTGTC‐3′; PECAM1‐170‐F: 5′‐GATGCCCAGTTTGAGGTCAT‐3′, PECAM1‐170‐R: 5′‐ACGTCTTCAGTGGGGTTGTC‐3′; CDH5‐189‐F: 5′‐CCAAGTTCGACACCACCTTT‐3′, SIRT6‐R: 5′‐CGGACGTACTGCGTCTTACA‐3′; S100A4‐161‐F: 5′‐GGTGTCCACCTTCCACAAGT‐3′, S100A4‐161‐R: 5′‐TGTTGCTGTCCAAGTTGCTC‐3′; α‐SMA‐100‐F: 5′‐TATCCCCGGGACTAAGACGGG‐3′, α‐SMA‐100‐R: 5′‐CAGAGCCCAGAGCCATTGTC‐3′; PECAM1‐170‐F: 5′‐GATGCCCAGTTTGAGGTCAT‐3′, PECAM1‐170‐R: 5′‐ACGTCTTCAGTGGGGTTGTC‐3′; CDH5‐189‐F: 5′‐CACTTCCTACCCGTGGTCAT‐3′, CDH5‐189‐R: 5′‐CACTGTGATGGTGAGGATGC‐3′.

Western blot

Protein was extracted from HRMECs, and the concentration was determined with a BCA protein kit (Beyotime, Shanghai, China). Thirty micrograms of protein underwent electrophoresis on a 10% SDS‐PAGE (Millipore, Billerica, USA) and was transferred to PVDF membranes. After blocking the membrane with 5% BSA for 2 h, it was incubated with the primary antibody at 4°C overnight, including anti‐LIN28A (Abcam, ab279647, 1/1,000), anti‐SIRT6 (Abcam, ab191385, 1/2,000), anti‐S100A4 (Abcam, ab197896, 1/1,000), anti‐α‐SMA (Abcam, ab5831, 1/1,000), anti‐PECAM1 (Abcam, ab9498, 1/1,000), anti‐CDH5 (Abcam, ab318152, 1/1,000), anti‐p‐AMPK (Abcam, ab133448, 1/5,000), anti‐AMPK (Abcam, ab32047, 1/1,000), and anti‐β‐actin (Abcam, ab8227, 1/2,000). Then the membrane was washed with TBST and incubated with HRP‐labeled secondary antibody (Abcam, ab205718, 1/10,000; ab6728, 1/10,000) for 1.5 h. Finally, after washing with TBST, the strips were visualized using an enhanced chemiluminescence reagent (Beyotime). Gray values were analyzed using Image J software (National Institutes of Health, Bethesda, USA) to assess relative protein levels.

Immunofluorescence

Inoculating the HRMECs onto a culture dish cover glass, and when the cells approach a monolayer growth state, take out the cover glass. Subsequently, the cells were washed twice with PBS. Next, fix the cells using a 4% paraformaldehyde solution (Liankebio, Hangzhou, China) at room temperature for 15 min. After fixation, cover the cells with a 0.1–0.2% concentration of Triton X‐100 solution (Sorabio, Beijing, China) and incubate at room temperature for 15–20 min to achieve permeability. After completing the permeabilization, wash again with PBS three times. Then, HRMECs were incubated with anti‐α‐SMA (Abcam, ab220795), anti‐PECAM1 (Abcam, ab222783), and Goat Anti‐Rabbit IgG H&L (Abcam, ab150077). DAPI (Sorabio) was employed for nuclear staining. A fluorescent microscope (Olympus BX51) was used for image capture, and images were analyzed with ImageJ software (National Institutes of Health).

Transwell

To measure cell migration, a transwell assay was conducted. Cells were placed in the upper section of the cell transfer chamber (Corning Company, New York, USA), with 100 μL of serum‐free medium added, while 500 μL of medium with 5% fetal bovine serum was added to the lower section. After 24 h, the cells in the upper chamber were removed, and the migratory cells were fixed with formalin, staining them with 0.1% crystal violet (Sorabio), imaging using a microscope (Leica, China), and counting them using ImageJ (National Institutes of Health).

mRNA stability

A concentration of 5 μg/mL actinomycin D was applied to HRMECs, which sourced from the MedChem Express company (New Jersey, USA), for durations of 0, 2, and 4 h. After collecting the cells, total RNA was extracted, and the abundance of SIRT6 mRNA was detected by RT‐qPCR.

RNA immunoprecipitation (RIP)

RIP assay was conducted with the Magna RIP™ RNA‐Binding Protein Immunoprecipitation Kit (Millipore). Briefly, HRMECs were harvested and lysed in an RNA lysis buffer with protease and RNase inhibitors, with part of the lysate isolated as the Input. The other section was incubated overnight at 4°C with magnetic beads conjugated with an anti‐LIN28A (Abcam, ab279647) antibody. Anti‐IgG (Abcam, ab172730) served as a standard control. The complexes were washed and digested. Subsequently, SIRT6 mRNA levels were quantified through RT‐qPCR.

RNA pull‐down assay

To biochemically substantiate the direct interaction between the LIN28A protein and SIRT6 mRNA, an RNA pull‐down assay was employed. Biotinylated sense (target) and antisense (negative control) RNA probes were synthesized via in vitro transcription using T7 RNA polymerase (Thermo Fisher Scientific). These probes were subsequently immobilized on Dynabeads™ MyOne™ Streptavidin C1 beads (Thermo Fisher Scientific). Lysates from HRMECs, prepared with NP‐40 lysis buffer (Sigma‐Aldrich) supplemented with an RNase inhibitor (Thermo Fisher Scientific), were pre‐cleared and incubated with the RNA‐bound beads for 2 h at 4°C. After extensive washing with lysis buffer to remove nonspecific interactions, the bound proteins were eluted and analyzed by western blot using an anti‐LIN28A antibody (Abcam, ab279647).

Gene expression omnibus (GEO) analysis

GSE20886 dataset of streptozotocin (STZ)‐induced DR animal models was retrieved from the GEO database (http://www.ncbi.nlm.nih.gov/geo/), which contains retinal tissue data from Sprague–Dawley rats in the control group (n = 5) and the STZ group (n = 4). Differentially expressed genes (DEGs) between control and STZ groups were identified using the limma package (version 3.64.3) in R (version 4.5.0).

Statistical analysis

Results are displayed as mean ± standard deviation (SD). When comparing two groups, a Student's t‐test was employed, whereas a one‐way analysis of variance (anova) with Tukey's post hoc test was utilized for comparisons involving three or more groups. The data were analyzed with SPSS 19.0 (IBM Corp., USA), considering a P‐value below 0.05 as statistically significant.

RESULTS

SIRT6 inhibited the EndMT process of HG‐induced HRMECs

To investigate the role of SIRT6 in DR, an in vitro DR model was constructed based on a previous study ²³ . It was revealed that HG treatment could accelerate HRMECs migration (Figure 1a). In addition, in HG‐induced HRMECs, the levels of S100A4 and α‐SMA increased, whereas the levels of PECAM1 and CDH5 decreased (Figure 1b). Similarly, immunofluorescence assay showed that HG treatment increased α‐SMA expression and decreased PECAM1 expression in HRMECs (Figure 1c). The above results demonstrated that HG‐induced HRMECs to lose endothelial cell characteristics and transform into mesenchymal morphology. Moreover, we found a decrease in SIRT6 levels in HG‐induced HRMECs (Figure 1d,e). Next, the SIRT6 overexpression lentivirus was infected into HRMECs, followed by HG incubation. SIRT6 overexpression upregulated SIRT6 level in HG‐induced HRMECs (Figure 1f,g). In addition, the overexpression of SIRT6 inhibited HG‐induced HRMECs migration (Figure 1h). The upregulation of SIRT6 reversed the increased levels of S100A4, α‐SMA, and decreased levels of PECAM1 and CDH5 in HG‐incubated HRMECs (Figure 1i,j). HG treatment increased α‐SMA while decreasing PECAM1 expression in HRMECs, which was reversed by SIRT6 overexpression (Figure 1k). In summary, HG‐induced HRMECs EndMT was alleviated by SIRT6.

Figure 1.

SIRT6 inhibited the EndMT process of HG‐induced HRMECs. HRMECs were incubated with 25 mmol/L d‐glucose for 48 h. (a) Transwell assay was applied to detect cell migration ability. Scale bar = 100 μm. (b) RT‐qPCR assay was used to check the expressions of S100A4, α‐SMA, PECAM1, and CDH5. (c) Levels of α‐SMA and PECAM1 were assessed using immunofluorescence staining, with a scale bar of 100 μm. (d, e) SIRT6 mRNA and protein levels were quantified by RT‐qPCR and western blot assays, respectively. HRMECs were infected with SIRT6 overexpression lentivirus and incubated with 25 mmol/L d‐glucose for 48 h. (f, g) SIRT6 expression was determined by RT‐qPCR and western blot assays. (h) The cell migration capability was evaluated using transwell assay. Scale bar = 100 μm. (i, j) RT‐qPCR and western blot assays were conducted to analyze S100A4, α‐SMA, PECAM1, and CDH5 levels. (k) Staining for α‐SMA and PECAM1 levels was performed using immunofluorescence, with a scale bar of 100 μm. Data are representative of three independent experiments and presented as means ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

SIRT6 activated the AMPK signaling to inhibit HG‐treated HRMECs EndMT

Next, we further probed the mechanism by which SIRT6 inhibited HG‐treated HRMECs EndMT. We found that HG treatment reduced the proportion of p‐AMPK/AMPK in HRMECs, while overexpression of SIRT6 increased the proportion of p‐AMPK/AMPK in HG‐induced HRMECs (Figure 2a). We constructed SIRT6 overexpressing cells and incubated them with HG and/or compound C (AMPK inhibitor); the overexpression of SIRT6 increased p‐AMPK/AMPK proportion in HG‐induced HRMECs, while compound C treatment reversed the above results (Figure 2b). In HG‐induced HRMECs, SIRT6 overexpression reduced cell migration, while cell migration was promoted after treatment with compound C (Figure 2c). SIRT6 overexpression reduced the levels of S100A4 and α‐SMA, while PECAM1 and CDH5 showed the opposite trend, and compound C reversed the above phenomenon (Figure 2d). Consistent with the previous result, overexpression of SIRT6 resulted in decreased expression of α‐SMA, accompanied by increased expression of PECAM1 in HG‐induced HRMECs (Figure 2e). However, when treated with compound C, these changes were significantly reversed (Figure 2e). Furthermore, experiments involving the overexpression of SIRT6 and/or the knockdown of AMPK were conducted on HRMECs under HG conditions. The findings indicated that SIRT6 overexpression led to an increase in the p‐AMPK/AMPK ratio, whereas si‐AMPK treatment resulted in a reduction of both p‐AMPK and AMPK levels (Figure S1a). Additionally, SIRT6 overexpression was associated with a decrease in cell migration of HG‐induced HRMECs, in contrast to AMPK knockdown treatment, which enhanced cell migration (Figure S1b). Overexpression of SIRT6 resulted in decreased expression of S100A4 and α‐SMA, along with increased expression of PECAM1 and CDH5, while AMPK inhibition reversed these effects (Figure S1c). Immunofluorescence analysis corroborated these findings, showing that SIRT6 overexpression reduced α‐SMA levels and increased PECAM1 expression, whereas AMPK knockdown counteracted these changes (Figure S1d). Substantially, SIRT6 effectively inhibited the EndMT process of HRMECs in the HG environment by activating AMPK.

Figure 2.

SIRT6 activated the AMPK signaling to inhibit HG‐treated HRMECs EndMT. HRMECs were infected with SIRT6 overexpression lentivirus and then exposed to 25 mmol/L d‐glucose for 48 h. (a) Western blot assay was utilized to measure p‐AMPK and AMPK levels. After being infected with SIRT6 overexpression lentivirus, HRMECs were treated with 10 μM compound C for 30 min, followed by induction with 25 mmol/L d‐glucose for 48 h. (b) The levels of p‐AMPK and AMPK were measured using western blot assay. (c) Transwell assay was used to evaluate cell migration capability. Scale bar = 100 μm. (d) To evaluate S100A4, α‐SMA, PECAM1, and CDH5 expressions, the RT‐qPCR assay was utilized. (e) Immunofluorescence staining was employed to assess the levels of α‐SMA and PECAM1, with a scale bar of 100 μm. Data are representative of three independent experiments and presented as means ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

LIN28A upregulated SIRT6 by promoting SIRT6 mRNA stability

HRMECs and HG‐induced HRMECs were treated with actinomycin D, and it was found that HG could reduce the mRNA stability of SIRT6 (Figure 3a). We found a connection between LIN28A and SIRT6 in the POSTAR database, so we validated their binding relationship through RIP. As shown in Figure 3b, there was a binding relationship between LIN28A and SIRT6 mRNA in HRMECs, and the binding relationship was weakened after HG treatment. Besides, we synthesized a biotin‐labeled SIRT6 mRNA probe and incubated it with protein lysates extracted from cells under HG conditions. The results clearly demonstrated that biotin‐labeled SIRT6 mRNA probes efficiently enriched LIN28A protein, whereas no obvious LIN28A signal was detected in the control group (using a biotin‐labeled scrambled RNA probe) (Figure 3c). LIN28A was overexpressed or knocked down in HRMECs; the results showed that LIN28A overexpression increased SIRT6 levels, and its knockdown reduced SIRT6 levels in HRMECs (Figure 3d,e). Moreover, LIN28A overexpression enhanced the stability of SIRT6 mRNA, and LIN28A knockdown inhibited the stability of SIRT6 mRNA (Figure 3f). In addition, LIN28A overexpression increased the stability of SIRT6 mRNA under HG treatment (Figure 3g). Therefore, LIN28A increased SIRT6 levels by enhancing the stability of SIRT6 mRNA.

Figure 3.

LIN28A upregulated SIRT6 by promoting SIRT6 mRNA stability. HRMECs were treated with 25 mmol/L d‐glucose for 48 h and were subsequently induced with actinomycin D for intervals of 0, 2, and 4 h. (a) The SIRT6 mRNA level was checked through RT‐qPCR assay. (b, c) RIP and RNA pull down assays were employed to confirm the interaction between LIN28A and SIRT6 mRNA. Constructing LIN28A overexpression or knocking down in HRMECs. (d, e) The expressions of LIN28A and SIRT6 were detected by RT‐qPCR and western blot assays. (f) Following actinomycin D treatment (0, 2, and 4 h), the SIRT6 mRNA level was checked via RT‐qPCR assay. LIN28A overexpression was constructed in HRMECs and induced by HG conditions, followed by actinomycin D treatment (0, 2, and 4 h). (g) SIRT6 mRNA levels were measured through an RT‐qPCR assay. Data are representative of three independent experiments and presented as means ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

LIN28A upregulated SIRT6 to inhibit the HG‐induced HRMECs EndMT

Construction of LIN28A‐overexpressing and/or SIRT6‐knockdown then incubated with HG. HG inhibited the expressions of LIN28A and SIRT6 in HRMECs, and oe‐LIN28A upregulated the levels of LIN28A and SIRT6 in HG‐induced HRMECs, whereas si‐SIRT6 inhibited the upregulation of SIRT6 by oe‐LIN28A (Figure 4a,b). Overexpression of LIN28A blocked HG‐induced HRMECs migration (Figure 4c). However, cell migration increased after transfection with si‐SIRT6 (Figure 4c). Overexpression of LIN28A alleviated the promotion of S100A4 and α‐SMA by HG and the inhibition of PECAM1 and CDH5 by HG in HRMECs, while these effects were reversed by SIRT6 inhibition (Figure 4d). The increased α‐SMA and decreased PECAM1 level in HG‐treated HRMECs were reversed by LIN28A overexpression, whereas SIRT6 downregulation promoted α‐SMA expression and reduced PECAM1 expression (Figure 4e). Moreover, we performed bioinformatics analyses on the GSE20886 dataset. The results showed that the expression levels of LIN28A were downregulated in retina from STZ group (p = 0.0299); meanwhile, the expression of SIRT6 was downregulated, but this change was not significant (p = 0.0602) (Figure S2a,b). In short, by enhancing SIRT6 expression, LIN28A suppressed the EndMT process in HRMECs exposed to HG.

Figure 4.

LIN28A upregulated SIRT6 to inhibit the HG‐induced HRMECs EndMT. Lentivirus for overexpressing LIN28A and/or knocking down SIRT6 were introduced into HRMECs, which were then incubated with HG (25 mmol/L d‐glucose, 48 h). (a) LIN28A and SIRT6 expressions were detected by RT‐qPCR assay. (b) The detection of LIN28A and SIRT6 levels was performed with western blot assay. (c) The cell migration ability was confirmed by transwell assay. Scale bar = 100 μm. (d) The levels of S100A4, α‐SMA, PECAM1, and CDH5 were measured using RT‐qPCR assay. (e) Immunofluorescence staining was employed to assess the levels of α‐SMA and PECAM1, with a scale bar of 100 μm. Data are representative of three independent experiments and presented as means ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

CONCLUSION AND DISCUSSION

During the intricate biological process of EndMT, endothelial cells discard their specific markers, including PECAM1 and CDH5, and exhibit increased levels of mesenchymal markers such as α‐SMA ²⁴ . Notably, EndMT contributes to the progression of DR by regulating its core pathologic hallmarks—pericyte loss and neovascularization ²⁵ . In DR, EndMT is not only involved in the compromise of the blood–retinal barrier, causing vascular leakage, neovascularization, and fibrosis ²⁶ , but also closely associated with pericyte loss, a key early pathological event that disrupts retinal vascular integrity and precedes overt neovascularization ²⁵ . Therefore, gaining a deeper insight into the molecular mechanisms of the EndMT process could offer a new therapeutic strategy to alleviate the pathological progression of DR. The current research indicated that LIN28A increases the expression of SIRT6 by enhancing its mRNA stability, thereby activating the AMPK signaling and effectively suppressing the EndMT process experienced by HRMECs under HG conditions.

Over the last 10 years, the sirtuin family of class III histone deacetylases has garnered significant interest as posttranslational modification enzymes in regulating important cellular processes ²⁷ . Recent studies have shown that SIRT6 is an important sirtuin with unique and essential functions in maintaining endothelial homeostasis ²⁸ , ²⁹ . By promoting SIRT6‐mediated mitochondrial stability, FGF21 can prevent damage to blood–brain barrier endothelial cells ³⁰ . In human pulmonary lung microvascular endothelial cells, SIRT6 defended against inflammation triggered by lipopolysaccharides ³¹ . Moreover, in nonobese diabetic mice, SIRT6 expression was significantly decreased, and SIRT6 may be associated with early neurogenic events in DR ³² . Our findings demonstrated a downregulation of SIRT6 in HG‐induced HRMECs. The results are consistent with previous research, indicating that SIRT6 was downregulated in the fibrovascular membrane of proliferative DR patients and associated with the pathogenicity of proliferative DR ¹² . While this study mainly focused on the effect of SIRT6 on EndMT in DR and found that SIRT6 overexpression in HRMECs inhibited the HG‐induced EndMT process.

Research has extensively demonstrated that activating the AMPK signaling pathway has a protective effect in DR. By phosphorylating the AMPK/SIRT1 pathway, artemether triggers autophagy to mitigate DR ³³ . Besides, the AMPK/Nrf2/HO‐1 pathway, which was activated by diosgenin, was proven to protect retinal pigment epithelial cells from the inflammatory and oxidative stress effects of HG ³⁴ . According to reports, SIRT6 inhibits HG‐stimulated mitochondrial dysfunction and apoptosis in podocytes by activating AMPK phosphorylation ¹⁸ . In HFD/STZ‐induced mice and HG‐stimulated retinal cells, the activation of AMPK phosphorylation can drive autolysosomal biosynthesis to alleviate retinal damage ³⁵ . While in skeletal muscle, SIRT6 could maintain metabolic equilibrium by triggering AMPK ³⁶ . However, whether SIRT6 can activate the AMPK signaling pathway in DR is unknown. During the exploration process, we found that the overexpression of SIRT6 increased the proportion of p‐AMPK/AMPK in HG‐induced HRMECs, while treatment with the AMPK inhibitor (compound C) or AMPK knockdown could reduce the proportion of p‐AMPK/AMPK. SIRT6 has been shown to exhibit inhibitory effects on DR EndMT by activating the AMPK pathway.

Since being discovered, LIN28A has been identified as a regulator of various physiological processes, including stem cell renewal and differentiation, tissue repair, and glucose metabolism through both let‐7 dependent and independent pathways ³⁷ . Intriguingly, LIN28A has been demonstrated to inhibit DR by activating SIRT1‐dependent autophagy to alleviate HG‐induced retinal pigment epithelial damage ²¹ . Research has shown that LIN28A, as an RNA‐binding protein, has the sexual function of regulating mRNA stability ³⁸ . For example, in myocardial ischemia–reperfusion injury, LIN28A bound to BCAT1 mRNA and promoted its stability ³⁹ . Similarly, our results found that LIN28A was reduced by HG in HRMECs. Further investigations combining RIP and RNA pulldown results confirmed that LIN28A has the ability to bind to SIRT6 mRNA and enhance the stability of SIRT6 mRNA, thereby inhibiting the EndMT in HG‐induced HRMECs.

Moreover, we performed bioinformatics analyses on the GSE20886 dataset and found that the expression levels of LIN28A were downregulated in retina from the STZ group; meanwhile, the expression of SIRT6 was downregulated, but this change was not significant. We consider several potential explanations for these observations: (1) the dataset was derived from heterogeneous tissue samples (like whole retina), whereas the proposed LIN28A–SIRT6 axis is operative in specific cell types such as endothelial cells, leading to signal dilution; and (2) the number of tissue samples is limited.

However, this study primarily investigated the impact of the LIN28A/SIRT6/AMPK axis on EndMT within an in vitro DR cell model. Although in vitro cell models offer the advantage of controlled experimental conditions and the ability to simulate specific disease processes, they inherently lack the capacity to fully replicate the intricate physiological and pathological milieu of the human body. While in vivo validation using established DR models (e.g., STZ‐induced or db/db mice) represents a gold standard, these models typically require a prolonged induction period of 4–6 months to develop stable pathology ⁴⁰ , ⁴¹ , ⁴² , ⁴³ . Due to time limitations, we were unable to conduct these in vivo experiments within the current study. Consequently, to achieve a more comprehensive understanding of the therapeutic potential of the LIN28A/SIRT6/AMPK axis in DR EndMT, future studies will prioritize these critical in vivo validations based on the mechanistic insights gained here.

In summary, LIN28A promoted mRNA stability of SIRT6 to activate the AMPK signaling, thereby inhibiting the EndMT process induced by HG in HRMECs. In this study, we proposed a new insight into improving EndMT in DR, which may provide new approaches for future DR treatment.

FUNDING

This work was supported by Leading Talents of Zhongyuan Science and Technology (224200510013), Natural Science Foundation of Henan Province (252300421269), and The Basic Research Project of Henan Eye Institute (20JCZD001).

DISCLOSURE

The authors declare no conflict of interest.

Approval of the research protocol: N/A.

Informed consent: N/A.

Registry and the registration no. of the study/trial: N/A.

Animal studies: N/A.

Supporting information

Figure S1. SIRT6 activated the AMPK signaling to inhibit HG‐treated HRMECs EndMT. SIRT6 overexpression and/or AMPK knockdown were conducted on HRMECs and then treated with HG conditions. (a) Western blot assay was used to measure the levels of p‐AMPK and AMPK. (b) Cell migration capability was demonstrated using a transwell assay. Scale bar = 100 μm. (c) Levels of S100A4, α‐SMA, PECAM1, and CDH5 were assessed through RT‐qPCR assay. (d) The assessment of α‐SMA and PECAM1 levels was conducted through immunofluorescence staining, with a scale bar of 100 μm. Data are representative of three independent experiments and presented as means ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure S2. Bioinformatics analysis for the expression of LIN28A and SIRT6. (a) The box plot of LIN28A expression among different groups in GSE20886 dataset. (b) The box plot of SIRT6 expression among different groups in GSE20886 dataset.

DATA AVAILABILITY STATEMENT

All data needed to evaluate the conclusions are present in the paper and the Supplementary Files.