Autophagy Modulates the Migration of Retinal Pericytes Induced by Advanced Glycation End Products

Retinal pericyte migration occurs in the early stage of diabetic retinopathy (DR), which is one of the important causes of pericyte loss. Autophagy has been found to play essential roles in the regulation of many types of cell migration. In this study, we explored the relationship between autophagy and retinal pericyte migration. In diabetic rats, the retinas became thinner, and the level of autophagy in each cell layer increased. In the primary culture of bovine retinal pericytes, we found that advanced glycation end products (AGEs) increased the migratory cell ability without influencing cell viability, which also increased the phosphorylation of focal adhesion kinase (FAK) and the expression of matrix metalloproteinase (MMP)-2 and decreased the expression of vinculin. AGEs-induced retinal pericyte autophagy and the inhibition of autophagy with chloroquine significantly inhibited cell migration, reversed AGEs-induced FAK phosphorylation, and changed vinculin and MMP-2 protein expression. These results provide a new insight into the migration mechanism of retinal pericytes. The early control of autophagy has a potential effect on regulating pericyte migration, which may contribute to keeping the integrity of retinal vessels in DR.


Introduction
Retinal pericytes are important constituents of the inner blood-retinal barrier (BRB); they are located at the basement membrane of retinal capillaries and tightly wrap around endothelial cells, keeping the integrity and stability of the BRB and regulating retinal blood fow and angiogenesis as well [1]. Te ratio of pericytes to endothelial cells in the retina is relatively high (1 : 1), compared with capillaries in other tissues [2], indicating the importance of pericytes in maintaining retinal functions. Selective pericytes loss is thought to be a hallmark of early diabetic retinopathy (DR) and also contributing to many clinical characteristics of DR.
Apoptosis is generally considered the cause of retinal pericytes loss in diabetic conditions [3,4]. In animal models of DR, pericyte apoptosis occurred 6 months after glucose elevation, whereas signifcant pericyte loss could be observed 3 months after diabetes formation, suggesting that there is a discrepancy between the time courses of pericyte apoptosis and pericyte loss in the early stage of DR [5,6].
Cell migration plays an important role in regulating multiple physiological and pathological processes in organisms. Previous studies showed that cerebrovascular pericytes migrated from the vascular wall and stayed in their position around the blood vessels after noxious stimuli [7,8]. Te early detachment of pericytes destroyed the integrity of the blood-brain barrier, thus resulting in vascular damage. Pfster et al. provided morphological evidence that showed the migration of retinal pericytes in diabetic models [9], which is considered another mechanism of pericyte loss in experimental DR and thus provides us a new insight into the prevention of early DR.
Autophagy occurs widely in the degradationregeneration systems of eukaryotes and plays an important role in the maintenance of cellular homeostasis. In the early stage of DR, optimal increases in levels of autophagy have been detected in the retinal outer plexiform layer, but there is no strong involvement of apoptosis [10]. Autophagy has also been observed in retinal pigment epithelial cells in high-glucose conditions, and the inhibition of autophagy increases high-glucose-induced production of intracellular reactive oxygen species and cell death [11]. We could consider autophagy a protective mechanism that allows cells to survive in unfavorable conditions and prevents apoptotic cell death. In the present studies, autophagy is demonstrated to be associated with cell migration, and autophagy has previously been demonstrated to facilitate the migration of dental pulp stem cells [12], human umbilical vein endothelial cells [13], bladder cancer cells [14], hepatocellular carcinoma cells [15], etc. However, the relationship between the migration of retinal pericytes and autophagy has not been discussed.
Te persistent high-glucose condition in vivo promotes the non-enzymatic glycosylation of proteins and forming advanced glycation end products (AGEs), the accumulation of AGEs is one of the most important pathogenesis of DR [16]. We have previously observed that AGE-modifed bovine serum albumin (AGE-BSA) induces retinal pericytes' migration before infuencing cell viability [17]. In this study, we further explored whether autophagy would participate in this process as a mechanism of protection.

Animal Studies.
Male Sprague-Dawley (SD) rats were obtained from the Experimental Animal Center of the Second Afliated Hospital of Harbin Medical University. After a week of routine feeding, the rats were randomly divided into a control group (n � 6) and a model group (n � 6). Te rats in the control group were fed a routine diet, and the rats in the model group were fed a high-fat diet. 8 weeks later, rats in the high-fat feeding group were intraperitoneally injected with streptozotocin (35 mg/kg), and the rats in the control group were injected with the same amount of citrate bufer. Rats with fasting blood glucose greater than 11.1 mmol/L for 3 consecutive days can be used as diabetic models. Te rats were anesthetized by ether, sacrifced by cervical dislocation, and eyeballs were immediately extracted. Te experiments conformed to the principles of the Declaration of Helsinki and were approved by the Ethics Committee of the First Afliated Hospital of Harbin Medical University.

Pericytes Culture and Identifcation.
Primary retinal pericytes were extracted from fetal bovine eyeballs and cultured in dulbecco's modifed eagle medium (DMEM) containing 20% foetal bovine serum (FBS), as described previously [18]. All cells were cultured at 37°C in 5% CO 2 , and the media were changed every 2-3 days. We use primary pericytes from passages 3-5 for this study. α-smooth muscle actin (α-SMA, Abcam, Cambridge, UK) and nerve/glial antigen 2 (NG2, Santa Cruz, Dallas, TX, USA) were used as positive stains to assess primary bovine retinal pericytes; factor VIII (Santa Cruz) and glial fbrillary acidic protein (Santa Cruz) were used as negative stains. Te cells were immobilized with paraformaldehyde (4%) for half an hour, then infltrated with Triton X-100 (0.5%) for 20 mins. Te primary antibodies were incubated at 4°C overnight; the secondary antibodies combined with FITC (ZSGB-BIO, Beijing, China) were incubated at 37°C for 1 h. Finally, the cell nuclei were stained by 4′6-diamino-2-phenylindole (Beyotime, Shanghai, China), and the cells were detected by fuorescence microscopy (Life Technologies, EVOS FL Auto). Subsequently, the pericytes were pretreated with or without chloroquine (CQ, 10 μM, Sigma, St. Louis, MO, USA) for 2 h and then exposed to AGE-BSA (Biovision, San Francisco, CA, USA) or Control-BSA (Biovision) for 24 h.

Measurement of Cell Viability.
Bovine retinal pericytes were seeded in a 96-well microplate (5 × 10 3 cell/well). After exposure to diferent concentrations of AGE-BSA and Control-BSA for 24 h, pericytes viability was detected with the CCK-8 detection kit (Dojindo, Shanghai, China) according to the manufacturer's protocol. Te viable cells were counted by absorbance measurement at 450 nm using a monochrome microplate reader.

Hematoxylin-Eosin Staining.
After rat eyeballs were extracted, they were placed into the eyeball's fxation fuid immediately. Parafn-embedded retinal tissue sections were counterstained with haematoxylin and eosin. After staining, the sections were dehydrated through ethanol and xylene. Finally, neutral glue is used for sealing. AperioCS2 Leica is used to scan the images.

Immunofuorescent
Staining. Te tissue sections undergo dewaxing, rehydration, and antigen repair extraction. Ten, LC3 primary antibody (1 : 200 dilution, Abcam) was used for tissue sections to be incubated at 4°C overnight, donkey anti-rabbit secondary antibody (1 : 200 dilution, Life Technologies) was incubated at 37°C for 45 min, the cell nucleus was stained with DAPI dye, and the tablet was sealed with fuorescent tablet. Photo by fuorescence microscope (Olympus).

Transmission Electron Microscopy.
Pericytes were immobilized in 0.1 m phosphate bufer in 2.5% glutaraldehyde for 2 h, then immobilized with 1% OsO 4 , washed, dehydrated in graded alcohol, and embedded in epoxy resin. Te processed ultramicrocuts were stained with uranyl acetate and lead citrate and observed under 80 kV with a transmission electron microscope (Hitachi, Japan).

GFP-mRFP-LC3 Staining.
Retinal pericytes were transfected with a tandem fuorescent GFP-mRFP-LC3 adenovirus (Vigenebio Bioscience, Jinan, China), which specifcally marks autophagosome formation. Te cells were viewed with a fuorescence microscope (Life Technologies). Te GFP dots primarily indicated autophagosomes, but in the acidic lysosomal environment they will lose their green fuorescence. Te red mRFP dots represented both autophagosomes and autolysosomes. In the combined image, red dots and green dots overlapped and appeared yellow, which is the symbol of autophagosomes. Te free red dots and green dots did not overlap and show red, which is the sign of autolysosomes. Te amounts of GFP and mRFP dots in fve high-power felds from three diferent cell preparations were determined by manual counting of the fuorescence dots.

Wound-Healing Assay.
Pericytes were cultured in 12well plates to full confuence and then treated with mitomycin C (10 μg/ml) for 2 h. Te monolayers were scratched with sterile pipette tips (10 μl), and after washing with PBS, pericytes were cultured for 24 hours with the intervention reagents using a microscope (Olympus, Japan) to collect the images of the wounds at diferent times after the wounds were generated. Using ImageJ software to estimate the percentage of the wound closures. Each assay was replicated 3 times. Te percentage of wound healing was calculated as follows: [1 − (empty area X h/empty area 0 h)] × 100 [19].

Transwell Assay.
Using transwell chambers (Corning, New York, NY, USA) to perform the transwell assays. Te lower chambers were added with complete medium, the upper chambers were seeded with suspended retinal pericytes (2 × 10 4 ). Te cells were incubated with 5% CO 2 at 37°C for 24 h, and then the flter inserts from the wells were removed. Cotton swabs are used to remove the cells from the upper surface of the flter, and then paraformaldehyde and crystal violet are used to fx and stain the remaining cells from the lower surface of the flter. Te cells were counted in nine randomly selected high-power felds using a microscope (Life Technologies, EVOS FL Auto) from three independent experiments to assess the average numbers of migrating cells.
2.11. Statistical Analysis. Te values are presented as the means ± standard deviation from at least three experiments. Te statistical analyses were processed via one-way ANOVA followed by Tukey's range signifcance diference test using GraphPad Prism software. Comparisons with p < 0.05 were considered statistically signifcant.

Autophagy in the Diabetic Retina.
Te results of HE staining showed that the retinas of normal rats were wellstructured, and the cells of each layer were regularly and neatly arranged (Figure 1(a)). Te retinas of diabetic rats had a disordered arrangement, fewer cells, and enlarged intercellular space (Figure 1(b)). In addition, the thickness of the retinas, defned as the distance from the retinal ganglion cell layer (GCL) to the pigment epithelium layer, became thinner in diabetic rats compared with normal rats (Figure 1(c)). Immunofuorescent staining analysis showed that in normal retinas, LC3 staining was faint and heterogeneously distributed (Figure 1(d)). However, diabetic retinas displayed intense LC3 signals (Figure 1(e)), and the immunofuorescence was localized mainly from the outer plexiform layer (OPL) to the GCL, indicating that LC3 immunoreactivity increased in the diabetic eyes.

AGE-BSA Induced Autophagy in Retinal Pericytes.
We used α-SMA and NG2 as positive staining to assess primary bovine retinal pericytes (Figures 2(a) and 2(b)) and used glial fbrillary acidic protein and factor VIII as negative staining (Figures 2(c) and 2(d)). Te measurement of cell activity showed that retinal pericytes viability was not altered by control-BSA but signifcantly decreased by AGE-BSA at a concentration of 150 μg/ml compared with the control-BSA group (Figure 2(e)). In order to investigate the migration and autophagy of retinal pericytes while excluding the infuence of cell viability, in the subsequent studies, the concentrations of AGE-BSA were no more than 100 μg/ml, and control-BSA was used for the control group.
To investigate whether autophagy could be induced by AGE-BSA, we detected the expressions of the autophagy indicators LC3A/B and SQSTM1. AGE-BSA dose-dependently decreased the expression of SQSTM1 and increased the LC3A/B II-to-I ratio; both of these efects were signifcant at the concentrations of 50 and 100 μg/ml compared with the control group (Control-BSA, 100 μg/ml; Figures 3(a) and 3(b)). To further verify the induction of autophagy, we used transmission electron microscopy to observe the autophagic changes. Te images displayed showed that in control-BSA-treated pericytes, the cytoplasm, mitochondria, and nuclei are normal. In contrast, a good deal of multiple-layered autophagic vacuoles (arrows) were detected in pericytes treated with 100 μg/ml AGE-BSA (Figure 3(c)). Next, we used GFP-mRFP-LC3 staining to further assess the autophagic fux. As illustrated in Figures 3(d)-3(f), 100 μg/ml AGE-BSA signifcantly enhanced the number of green and red dots and concomitantly increased the number of yellow dots in the retinal pericytes compared with the control group, indicating increased autophagosomes and autolysosomes. Tese observations strongly suggested that autophagy was activated by AGE-BSA in retinal pericytes.

Te Inhibition of Autophagy Suppressed Pericytes Migration Induced by AGE-BSA.
Because autophagy and migration were both increased in the AGE-BSA-induced pericytes, we further explored whether autophagy was related to cell migration. We decided to use the autophagy inhibitor CQ to pretreat the cells before the AGE-BSA intervention. Te transwell (Figures 5(a)

Discussion
Autophagy is a self-protection mechanism that exists widely in eukaryotic cells. It can promote the degradation and circulation of cellular components and play an important role in maintaining cell homeostasis. Under stressful

Evidence-Based Complementary and Alternative Medicine
conditions such as starvation, oxidative stress, hypoxia, and cytotoxicity, autophagy is basically an adaptive response [20][21][22]. Previous studies showed that autophagy is associated with the survival of DR-related pigment epithelial cells, endothelial cells, and Muller cells in a high-glucose condition [23][24][25]. In retinal pericytes, autophagy has been reported to be elicited by heavily-oxidised glycated LDL (HOG-LDL), and plays a dual role in this process [26]. In this study, we frst proved that autophagy was induced by AGE-BSA in retinal pericytes, the expressions of LC3A/B and SQSTM1 dose-dependently changed with the increase of the AGE-BSA concentrations, but cell viability was not afected until the AGE-BSA concentration reached to 150 μg/ml. We considered that autophagy might be a self-protective response in this AGEs-induced process to avoid further injuries to retinal pericytes. Cell migration is a complex signal transduction process, focal adhesions (FAs) are multiprotein assemblies' complex on plasma membrane, which connect cellular cytoskeleton to extracellular matrix [27] and play an important role in cellular migration. FAK is a nonreceptor protein tyrosine kinase, which is a converging signaling point and controls the assembly/disassembly of FAs [28]. Integrin stimulates FAK autophosphorylation at Tyr-397, leading to signaling cascades that eventually result in cytoskeleton recombination and cell migration [29]. Te results of this study showed that the level of FAK phosphorylation is elevated, indicating its relation to the promotion of migration.

Evidence-Based Complementary and Alternative Medicine
Vinculin is an important adapter protein for FAs, binding to talin and actin [30] and attaching microflaments to the membrane. Enhanced talin-vinculin-actin connections are a key factor in strengthening adhesion stability [31]. Matrix metalloproteinases (MMPs) are able to degrade the structural components of the extracellular matrix and degrade the stability of FAs. MMP-2 is one of the most widely distributed subtypes of the MMPs family. MMP-2 has been found in retinal pericytes, which plays an important role in promoting various cellular migrations as well [32][33][34]. Following the treatment with AGE-BSA, current data showed decreased expression of vinculin and increased expression of MMP-2, further confrming they both participate in promoting cell migration. In this study, the changes in FAs related functional protein FAK, structural proteins vinculin, and EMC protein MMP-2 were detected, which refected that AGE-BSA promoted pericytes migration from multiple perspectives. Autophagy has recently been found to play an essential role in the regulation of many types of cell migration [35][36][37], but the related mechanisms are not fully understood. Bressan and Saghatelyan showed that AMPKinduced autophagy is a key regulator of cell migration and plays an important role in the recycling of the focal adhesion molecule paxillin [38]. Kenifc and Debnath established that the autophagy cargo receptor NBR1 is an important mediator that supports cell migration; NBR1 interacts with focal adhesions through the ubiquitin binding domain and then targets autophagosomes to FAs, which leads to its disintegration by isolation of FA proteins [39]. Zhan et al. [35] demonstrated that toll-like receptor 4-and toll-like receptor 3-induced autophagy and increased the production of cytokines, including MMP-2, VEGFA, IL6, CCL20, and CCL2, which leads to the enhanced migration of lung cancer cells. As an inhibitor of lysosomes, chloroquine can inhibit the fusion of autophagosomes and lysosomes, thus inhibiting autophagy fux. Our results showed that AGE-BSAinduced cell migration was inhibited when autophagic fow was blocked, and the intracellular FAK phosphorylation, vinculin, and MMP-2 protein expressions were also changed. Terefore, we considered that AGE-BSA induced retinal pericytes autophagy is closely related to cell migration, which may directly or indirectly afect the stability of FAs ( Figure 6). Te loss of retinal pericytes will eventually lead to impairment of the blood-retinal barrier function and an increase in vascular leakage. Clarifying the mechanism of pericyte migration in the early stage of DR could provide intervention targets for the early prevention and treatment of DR in the future. Autophagy may play a role as a potential therapeutic target in regulating the stability of retinal vascular cells.

Conclusions
In conclusion, we suggested that autophagy and migration occurred in retinal pericytes in response to high-glucose metabolites, which may be associated with the increment of microvascular permeability and the development of diabetic retinopathy. In addition, we found that the inhibition of autophagy alleviated AGEs-induced retinal pericyte migration. Our results provide a new insight into the mechanisms underlying the migration of retinal pericytes. In addition, early control of autophagy may potentially contribute to the modulation of pericyte migration and sustain the integrity of the retinal vessels.

Data Availability
Te datasets used and/or analyzed in this study are available from the corresponding author upon reasonable request.

Conflicts of Interest
Te authors declare that they have no conficts of interest.

Authors' Contributions
WJL and HYK designed the research; WJL, XFM, and HRZ performed the study; WJL and XYY analyzed the data; CYX, YXH, and HXL participated in the animal section; WJL, MH, and QX prepared the manuscript; HYK modifed the manuscript. All authors approved the fnal version of the manuscript.