December 2020
Volume 61, Issue 14
Open Access
Retinal Cell Biology  |   December 2020
Involvement of Müller Glial Autoinduction of TGF-β in Diabetic Fibrovascular Proliferation Via Glial–Mesenchymal Transition
Author Affiliations & Notes
  • Di Wu
    Laboratory of Ocular Cell Biology and Visual Science, Department of Ophthalmology, Faculty of Medicine and Graduate School of Medicine, Hokkaido University, Sapporo, Japan
  • Atsuhiro Kanda
    Laboratory of Ocular Cell Biology and Visual Science, Department of Ophthalmology, Faculty of Medicine and Graduate School of Medicine, Hokkaido University, Sapporo, Japan
  • Ye Liu
    Laboratory of Ocular Cell Biology and Visual Science, Department of Ophthalmology, Faculty of Medicine and Graduate School of Medicine, Hokkaido University, Sapporo, Japan
  • Kousuke Noda
    Laboratory of Ocular Cell Biology and Visual Science, Department of Ophthalmology, Faculty of Medicine and Graduate School of Medicine, Hokkaido University, Sapporo, Japan
  • Miyuki Murata
    Laboratory of Ocular Cell Biology and Visual Science, Department of Ophthalmology, Faculty of Medicine and Graduate School of Medicine, Hokkaido University, Sapporo, Japan
  • Susumu Ishida
    Laboratory of Ocular Cell Biology and Visual Science, Department of Ophthalmology, Faculty of Medicine and Graduate School of Medicine, Hokkaido University, Sapporo, Japan
  • Correspondence: Atsuhiro Kanda, Department of Ophthalmology, Faculty of Medicine and Graduate School of Medicine, Hokkaido University, N-15, W-7, Kita-ku, Sapporo 060-8638, Japan; [email protected]
Investigative Ophthalmology & Visual Science December 2020, Vol.61, 29. doi:https://doi.org/10.1167/iovs.61.14.29
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Di Wu, Atsuhiro Kanda, Ye Liu, Kousuke Noda, Miyuki Murata, Susumu Ishida; Involvement of Müller Glial Autoinduction of TGF-β in Diabetic Fibrovascular Proliferation Via Glial–Mesenchymal Transition. Invest. Ophthalmol. Vis. Sci. 2020;61(14):29. https://doi.org/10.1167/iovs.61.14.29.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose: Müller glial–mesenchymal transition (GMT) is reported as the fibrogenic mechanism promoted by TGF-β–SNAIL axis in Müller cells transdifferentiated into myofibroblasts. Here we show the multifaceted involvement of TGF-β in diabetic fibrovascular proliferation via Müller GMT and VEGF-A production.

Methods: Surgically excised fibrovascular tissues from the eyes of patients with proliferative diabetic retinopathy were processed for immunofluorescence analyses of TGF-β downstream molecules. Human Müller glial cells were used to evaluate changes in gene and protein expression with real-time quantitative PCR and ELISA, respectively. Immunoblot analyses were performed to detect TGF-β signal activation.

Results: Müller glial cells in patient fibrovascular tissues were immunopositive for GMT-related molecular markers, including SNAIL and smooth muscle protein 22, together with colocalization of VEGF-A and TGF-β receptors. In vitro administration of TGF-β1/2 upregulated TGFB1 and TGFB2, both of which were suppressed by inhibitors for nuclear factor-κB, glycogen synthase kinase-3, and p38 mitogen-activated protein kinase. Of the various profibrotic cytokines, TGF-β1/2 application exclusively induced Müller glial VEGFA mRNA expression, which was decreased by pretreatment with small interfering RNA for SMAD2 and inhibitors for p38 mitogen-activated protein kinase and phosphatidylinositol 3-kinase. Supporting these findings, TGF-β1/2 stimulation to Müller cells increased the phosphorylation of these intracellular signaling molecules, all of which were also activated in Müller glial cells in patient fibrovascular tissues.

Conclusions: This study underscored the significance of Müller glial autoinduction of TGF-β as a pathogenic cue to facilitate diabetic fibrovascular proliferation via TGF-β–driven GMT and VEGF-A–driven angiogenesis.

Pathologic retinal angiogenesis seen in proliferative diabetic retinopathy (PDR) takes place in the form of epiretinal tissue growth called fibrovascular proliferation, a vision-threatening complication characterized by tractional retinal detachment and vitreous hemorrhage resulting from fibrocellular and neovascular components, respectively. Fibrocellular contraction of the fibrovascular tissue depends on transdifferentiated myofibroblasts harboring both fibrogenic and contractile features.1 To promote the pathogenesis of fibrovascular proliferation, profibrotic and proangiogenic molecules are theorized to function cooperatively with a biologically purposive link. Indeed, TGF-β and VEGF-A, the representative fibrogenic and angiogenic cytokines, were reported to correlate with high intravitreal levels in patients with PDR.2 Consistently, both of these cytokines were shown to be expressed in surgically excised fibrovascular tissues from patients with PDR.3,4 
In contrast with the epiretinal fibrovascular proliferation occurring in PDR, another vision-threatening disease, namely, neovascular AMD (nAMD), is complicated by subretinal fibrovascular proliferation, also known simply as choroidal neovascularization. Using patient surgical specimens, we recently demonstrated RPE cells as the cellular source of myofibroblasts through the epithelial–mesenchymal transition (EMT) in the pathogenesis of subretinal fibrovascular proliferation occurring in nAMD.5,6 TGF-β signaling via TGF-β receptor (TβR) was shown to be required for the upregulation of SNAIL, a key transcription factor triggering EMT in RPE cells and subsequent myofibroblastic differentiation.58 In parallel with TGF-β–initiated EMT, TGF-β induced the expression of TGF-β per se in RPE cells, suggesting the autoinduction of TGF-β as the amplifier of subretinal fibrosis in nAMD.6,7 
As concerns the cellular source of myofibroblasts in epiretinal fibrocellular proliferation seen in idiopathic epiretinal membrane, we recently revealed that Müller glial cells are equipped with fibrogenic EMT program for the expression of TβR, SNAIL, and smooth muscle protein (SM)22, which would thus be termed the Müller glial–mesenchymal transition (GMT) as an alternative to EMT.9 In fibrovascular proliferation associated with PDR, however, little is yet known about the involvement of Müller GMT promoted by the TGF-β–SNAIL axis, although Müller glial cells in patient tissues were suggested to bear a contractile phenotype expressing α-smooth muscle actin (SMA).10 It is also yet to be elucidated whether Müller glial cells have any capacity for the profibrotic TGF-β autoinduction as seen in RPE cells during subretinal fibrovascular proliferation.6 In the present study, we investigated the multifaceted involvement of TGF-β signaling in diabetic epiretinal fibrovascular proliferation through Müller glial bioactions including GMT, TGF-β autoinduction, and VEGF-A production, using patient-excised tissues and human cultured cells. 
Methods
Human Surgical Samples
Four fibrovascular tissues were excised from PDR eyes during vitrectomy and used for immunohistochemistry. This study was conducted in accordance with the tenets of the Declaration of Helsinki and after receiving approval from the Institutional Review Board of the Hokkaido University Hospital. Written informed consent was obtained from all patients after an explanation of the purpose and procedures of this study. 
Immunofluorescence Microscopy
Immunofluorescence analyses were performed as described previously.6,9 Serial paraffin sections at 4 µm thickness were cut and mounted onto microscope slides. Sections were incubated with primary antibodies followed by secondary antibodies. All antibodies are detailed in the Supplementary Information. Nuclei were counterstained with 4’,6-diamidino-2-phenylindole (DAPI), and sections were examined using a Biorevo BZ-9000 epifluorescence microscope (Keyence, Osaka, Japan). 
Cell Culture and Reagents
The human Müller glial cell line Moorfields/Institute of Ophthalmology-Müller 1 was provided from Dr. G. Astrid Limb (UCL Institute of Ophthalmology, London, UK).11 The Moorfields/Institute of Ophthalmology-Müller 1 cells were cultured in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum at 37°C. Human retinal microvascular endothelial cells (Cell Systems Corporation, Kirkland, WA) were cultured on type I collagen-coated plates with CS-C complete medium (Cell Systems Corporation). All reagents are detailed in the Supplementary Information
Real-Time Quantitative PCR Analyses
Total RNA was isolated by using TRI reagent (Molecular Research Center, Cincinnati, OH), and reverse transcription to cDNA was then performed using GoScript reverse transcriptase (Promega, Madison, WI), according to the manufacturers’ instructions. All primers are listed in Supplementary Table S1. Real-time quantitative PCR was performed with the GoTaq qPCR Master Mix (Promega) and StepOne Plus Systems (Thermo Fisher Scientific, Waltham, MA). Gene expression levels were calculated by the 2−ddCt method with glyceraldehyde-3-phosphate dehydrogenase levels as an internal control. 
Immunoblot Analyses
Collected cells were lysed in sodium dodecyl sulfate buffer. After quantifying protein concentrations using BCA reagent (Thermo Fisher Scientific), proteins were resolved by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and transferred to a polyvinylidene difluoride membrane by electroblotting. Membranes were blocked in 5% skim milk and incubated with primary antibodies followed by secondary antibodies. All antibodies are detailed in the Supplementary Information. Signals were visualized using a SuperSignal West Pico PLUS Chemiluminescent Substrate (Thermo Fisher Scientific). 
ELISA
Protein levels of VEGF-A in the supernatant were determined with Human VEGF Quantikine ELISA Kit (R&D Systems, Minneapolis, MN) according to the manufacturer's protocol. The optical density was measured using a microplate reader (Sunrise, TECAN, Männedorf, Switzerland). 
Cell Viability Assay
Cell Counting Kit-8 (Dojindo Molecular Technologies, Inc., Kumamoto, Japan) was used to quantify viable cells, according to the manufacturer's instructions. The optical density was measured using a microplate reader (Sunrise, TECAN). 
Statistical Analyses
All results are expressed as the mean ± SEM. The Student t-test was used for statistical comparison between groups, and one-way ANOVA followed by the Tukey-Kramer method as a post-hoc test was used for multiple comparison procedures. Differences between means were considered statistically significant when P values were less than 0.05. 
Results
Expression of GMT-Related Molecular Markers in Müller Glial Cells Migrated Into Diabetic Fibrovascular Tissues
Prerequisite factors for fibrogenic Müller GMT recently proved to be the essential combination of profibrotic signal transduction, mesenchymal transcriptional activation, and resultant myofibroblastic differentiation, corresponding respectively to the expression of TβR, SNAIL, and SM22.9 To investigate the involvement of GMT in diabetic fibrovascular proliferation, we carried out immunofluorescence analyses of these GMT-related molecular markers in fibrovascular tissues collected from PDR patients. Double-staining experiments for serial sections exhibited colocalization of an activated glial marker glial fibrillary acidic protein (GFAP) with a specific Müller glial marker glutamine synthetase (GS) (Figs. 1A–1C), TβRI (Figs. 1D–1F), TβRII (Supplementary Figs. S1A–S1C), TGF-β1 (Supplementary Figs. S1D–S1F), TGF-β2 (Supplementary Figs. S1G–S1I), SNAIL (Figs. 1G–1I), and a mesenchymal marker α-SMA (Figs. 1J–1L), suggesting the establishment of Müller GMT in diabetic fibrovascular proliferation. Furthermore, SM22, an actin-binding cytoskeletal protein used as an EMT or GMT marker, colocalized with glial markers GFAP (Figs. 1M–1O) and GS (Figs. 1P–1R), indicating the contractile phenotype of activated Müller glial cells in the fibrovascular tissue. 
Figure 1.
 
Expression of GMT-related molecular markers in Müller glial cells migrated into diabetic fibrovascular tissues. (AR) Double labeling of GS (green) and GFAP (red) (AC), TβRI (green) and GFAP (red) (DF), SNAIL (green) and GFAP (red) (GI), α-SMA (green) and GFAP (red) (JL), GFAP (green) and SM22 (red) (MO), and GS (green) and SM22 (red) (PR) in diabetic fibrovascular tissue specimens with DAPI (blue) counterstain to nuclei. Scale bar = 20 µm.
Figure 1.
 
Expression of GMT-related molecular markers in Müller glial cells migrated into diabetic fibrovascular tissues. (AR) Double labeling of GS (green) and GFAP (red) (AC), TβRI (green) and GFAP (red) (DF), SNAIL (green) and GFAP (red) (GI), α-SMA (green) and GFAP (red) (JL), GFAP (green) and SM22 (red) (MO), and GS (green) and SM22 (red) (PR) in diabetic fibrovascular tissue specimens with DAPI (blue) counterstain to nuclei. Scale bar = 20 µm.
Autoinduction of TGFB1/2 Expression via Its Noncanonical Signaling Pathways in Müller Glial Cells
TGF-β autoinduction in RPE cells was recently reported as the amplifier of subretinal fibrosis owing to TGF-β–initiated EMT in the pathogenesis of nAMD.6 In general, TGF-β/TβR–induced EMT requires signaling via SMAD-dependent (canonical) and -independent (noncanonical) pathways, the latter of which includes extracellular signal-regulated kinase (ERK)1/2 and phosphatidylinositol 3-kinase (PI3K)/AKT.1214 In particular, the activity of SMAD2, the key downstream molecule of TGF-β signaling, is critical for cancer EMT.15 SMAD signaling was also shown to be involved in proximal tubular epithelial TGF-β autoinduction together with ERK and p38 mitogen-activated protein kinase (MAPK) pathways.16 In the context of GMT, however, little is known about TGF-β signal transduction and TGF-β autoinduction in Müller glial cells. To explore the possible involvement of TGF-β autoinduction in Müller GMT, we next investigated the impact of TGF-β1/2 application on TGFB1/2 expression in human Müller glial cells. Stimulation with TGF-β1 and TGF-β2 to Müller cells significantly elevated mRNA levels of both TGFB1 (Fig. 2A) and TGFB2 (Supplementary Fig. S2A). To determine its downstream intracellular signaling pathways, we pretreated Müller cells with specific inhibitors for ERK1/2, nuclear factor (NF)-κB, PI3K, c-Jun N-terminal kinase (JNK), p38 MAPK, and glycogen synthase kinase (GSK)-3. TGF-β1/2–induced TGFB1 expression levels were significantly decrease by NF-κB, p38 MAPK, and GSK-3 inhibitors, but enhanced by PI3K inhibitor (Fig. 2B), which was also true of TGFB2 expression changes (Supplementary Fig. S2B). Transfection with small interfering RNA (siRNA)-1 and siRNA-2 against SMAD2 into Müller cells significantly suppressed SMAD2 mRNA expression (Fig. 2C). Silencing of SMAD2 did not cancel TGF-β1/2–induced expression of TGFB1 (Fig. 2D) or TGFB2 (Supplementary Fig. S2C), indicating the absence of canonical signaling in Müller glial autoinduction of TGF-β1/2. Consistently, the phosphorylation of NF-κB, p38, and GSK-3α/β increased substantially up to 24 hours after stimulation with TGF-β1 (Fig. 2E) and TGF-β2 (Supplementary Fig. S2D). In addition, the inhibition of PI3K alone did not increase TGFB1 or TGFB2 expression, but synergistically increased TGF-β1/2–induced TGFB1 and TGFB2 expression (Supplementary Figs. S2E, S2F). These results indicated that Müller glial autoinduction of TGF-β1/2 was positively regulated via noncanonical pathways by NF-κB, p38 MAPK, and GSK-3, and negatively by PI3K/AKT. 
Figure 2.
 
Autoinduction of TGFB1 expression via its noncanonical signaling pathways in Müller glial cells. (A) Müller glial cells were incubated with TGF-β1 or TGF-β2 (30 ng/mL) for 24 hours, and TGFB1 mRNA levels were analyzed. (B) Müller glial cells were pretreated with each inhibitor (U0126, ERK1/2; JSH-23, NF-κB; LY294002, PI3K; SP600125, JNK; SB203580, p38 MAPK; SB216763, GSK-3) at 10 µM for 30 minutes followed by stimulation with TGF-β1 or TGF-β2 (30 ng/mL) for 24 hours, and TGFB1 mRNA levels were analyzed. (C) SMAD2-siRNAs (siRNA-1 and siRNA-2) were transfected in Müller glial cells, and SMAD2 mRNA levels were analyzed. (D) Control siRNA-treated (Ctrl-siRNA) and SMAD2-knockdown (siRNA-1 and siRNA-2) Müller glial cells were stimulated with TGF-β1, and TGFB1 mRNA levels were analyzed. (E) Müller glial cells were incubated with TGF-β1, and protein levels of phosphorylated and total forms of NF-κB, p38, and GSK-3α/β were analyzed at various time points. (F, G) Müller glial cells were pretreated with normal IgG (300 ng/mL), anti-TβRI antibody (300 ng/mL), TβRI kinase inhibitor SB431542 (10 µg/mL), or anti-TβRII antibody (250 ng/mL) for 30 minutes followed by stimulation with TGF-β1 or TGF-β2 (30 ng/mL) for 24 hours, and TGFB1 mRNA levels were analyzed. n = 3, *P < 0.05, **P < 0.01. n.s., not significant.
Figure 2.
 
Autoinduction of TGFB1 expression via its noncanonical signaling pathways in Müller glial cells. (A) Müller glial cells were incubated with TGF-β1 or TGF-β2 (30 ng/mL) for 24 hours, and TGFB1 mRNA levels were analyzed. (B) Müller glial cells were pretreated with each inhibitor (U0126, ERK1/2; JSH-23, NF-κB; LY294002, PI3K; SP600125, JNK; SB203580, p38 MAPK; SB216763, GSK-3) at 10 µM for 30 minutes followed by stimulation with TGF-β1 or TGF-β2 (30 ng/mL) for 24 hours, and TGFB1 mRNA levels were analyzed. (C) SMAD2-siRNAs (siRNA-1 and siRNA-2) were transfected in Müller glial cells, and SMAD2 mRNA levels were analyzed. (D) Control siRNA-treated (Ctrl-siRNA) and SMAD2-knockdown (siRNA-1 and siRNA-2) Müller glial cells were stimulated with TGF-β1, and TGFB1 mRNA levels were analyzed. (E) Müller glial cells were incubated with TGF-β1, and protein levels of phosphorylated and total forms of NF-κB, p38, and GSK-3α/β were analyzed at various time points. (F, G) Müller glial cells were pretreated with normal IgG (300 ng/mL), anti-TβRI antibody (300 ng/mL), TβRI kinase inhibitor SB431542 (10 µg/mL), or anti-TβRII antibody (250 ng/mL) for 30 minutes followed by stimulation with TGF-β1 or TGF-β2 (30 ng/mL) for 24 hours, and TGFB1 mRNA levels were analyzed. n = 3, *P < 0.05, **P < 0.01. n.s., not significant.
We next investigated the role of TβRI/II in the mechanism of TGF-β1/2 autoinduction in Müller glial cells. TβRI blockade by pretreatment with anti-TβRI–neutralizing antibody and TβRI kinase inhibitor significantly attenuated TGF-β1/2–induced upregulation of TGFB1 (Fig. 2F) and TGFB2 (Supplementary Fig. S2G). Similarly, TβRII blockade by pretreatment with anti-TβRII neutralizing antibody significantly attenuated TGF-β1/2–induced upregulation of TGFB1 (Fig. 2G) and TGFB2 (Supplementary Fig. S2H). Moreover, TGFBR1 and TGFBR2 gene expression levels increased after TGF-β1/2 stimulation (Supplementary Figs. S2I, S2J). These data suggested the facilitatory role of TβRI/II in Müller glial autoinduction of TGF-β1/2. 
Screening of PDR-Associated Profibrotic Cytokines and Müller Glial VEGF-A Induction Exclusively by TGF-β1/2
During the pathogenesis of diabetic epiretinal fibrovascular proliferation, both profibrotic and proangiogenic cytokines such as TGF-β and VEGF-A were reported to increase often with mutual link in the vitreous of PDR patient eyes.2,1720 On top of its profibrotic nature, TGF-β is also a classically known proangiogenic factor that induces VEGF-A in human RPE cells21,22 and rat Müller glial cells,23 potentially promoting subretinal and epiretinal fibrovascular proliferations, respectively. Of various profibrotic stimuli with TGF-β1/2, connective tissue growth factor (CTGF), fibroblast growth factor (FGF) 2, nerve growth factor (NGF), and platelet-derived growth factor (PDGF)-BB, all of which were elevated in PDR eyes2,1720; TGF-β1/2 alone exclusively increased VEGFA mRNA levels in human Müller cells (Fig. 3A). The administration of TGF-β1/2 to Müller cells upregulated VEGFA expression in dose- and time-dependent manners (Supplementary Figs. S3A, S3B). Consistent with the mRNA expression results, VEGF-A protein levels in the culture medium were elevated by TGF-β1/2 stimulation in a dose-dependent manner (Fig. 3B). We next investigated the role of TβRI/II in the mechanism of VEGF-A induction in Müller glial cells. TβRI blockade by pretreatment with anti-TβRI–neutralizing antibody and TβRI kinase inhibitor significantly attenuated TGF-β1/2–induced upregulation of VEGFA (Fig. 3C). Similarly, TβRII blockade by pretreatment with anti-TβRII neutralizing antibody significantly attenuated TGF-β1/2–induced upregulation of VEGFA (Fig. 3D). These data suggested the crucial role of TGF-β1/2–TβRI/II signaling in Müller glial expression of VEGF-A. 
Figure 3.
 
Screening of PDR-associated profibrotic cytokines and Müller glial VEGF-A induction exclusively by TGF-β1/2. (A) Müller glial cells were applied with TGF-β1, TGF-β2, CTGF, FGF2, NGF, and PDGF-BB at the dose of 10 ng/mL for 24 hours, and VEGFA mRNA levels were analyzed. (B) Müller glial cells were incubated with TGF-β1 or TGF-β2 at different doses, and protein levels of VEGF-A were analyzed. (C, D) Müller glial cells were pretreated with normal IgG (300 ng/mL), anti-TβRI antibody (300 ng/mL), TβRI kinase inhibitor SB431542 (10 µg/mL), or anti-TβRII antibody (250 ng/mL) for 30 minutes followed by stimulation with TGF-β1 or TGF-β2 (30 ng/mL) for 24 hours, and VEGFA mRNA levels were analyzed. (E) Müller glial cells were stimulated with TGF-β1 or TGF-β2 (30 ng/mL) for 24 hours, and supernatants were collected. Human retinal microvascular endothelial cells were preincubated with anti–VEGF-A drugs bevacizumab (0.3125 mg/mL), aflibercept (0.5 mg/mL), and normal IgG (0.5 mg/mL) for 15 minutes, followed by treatment with TGF-β1/2-stimulated Müller cell culture medium for 48 hours, and processed for cell viability assay. n = 3, *P < 0.05, **P < 0.01.
Figure 3.
 
Screening of PDR-associated profibrotic cytokines and Müller glial VEGF-A induction exclusively by TGF-β1/2. (A) Müller glial cells were applied with TGF-β1, TGF-β2, CTGF, FGF2, NGF, and PDGF-BB at the dose of 10 ng/mL for 24 hours, and VEGFA mRNA levels were analyzed. (B) Müller glial cells were incubated with TGF-β1 or TGF-β2 at different doses, and protein levels of VEGF-A were analyzed. (C, D) Müller glial cells were pretreated with normal IgG (300 ng/mL), anti-TβRI antibody (300 ng/mL), TβRI kinase inhibitor SB431542 (10 µg/mL), or anti-TβRII antibody (250 ng/mL) for 30 minutes followed by stimulation with TGF-β1 or TGF-β2 (30 ng/mL) for 24 hours, and VEGFA mRNA levels were analyzed. (E) Müller glial cells were stimulated with TGF-β1 or TGF-β2 (30 ng/mL) for 24 hours, and supernatants were collected. Human retinal microvascular endothelial cells were preincubated with anti–VEGF-A drugs bevacizumab (0.3125 mg/mL), aflibercept (0.5 mg/mL), and normal IgG (0.5 mg/mL) for 15 minutes, followed by treatment with TGF-β1/2-stimulated Müller cell culture medium for 48 hours, and processed for cell viability assay. n = 3, *P < 0.05, **P < 0.01.
Next, we further confirmed whether TGF-β–stimulated Müller glial cells can indeed affect vascular endothelial cells through VEGF-A bioactivity. Human retinal microvascular endothelial cells were incubated with TGF-β1/2–stimulated Müller cell culture medium. TGF-β1/2 stimulation to Müller cells significantly increased endothelial cell viability, which was potently inhibited by pretreatment with anti–VEGF-A drugs bevacizumab and aflibercept (Fig. 3E). These results indicated the potentially high dependency of endothelial activation on Müller glial VEGF-A production by stimulation with TGF-β1/2. 
Colocalization of VEGF-A With TβRI/II in Müller Glial Cells Migrated Into Diabetic Fibrovascular Tissues
GFAP-positive cells present in diabetic fibrovascular tissues were previously shown to be the source of VEGF-A production with in situ hybridization.4 To examine the association of TβRI/II-bearing cells with the expression of VEGF-A in diabetic fibrovascular proliferation, we performed immunofluorescence analyses of VEGF-A and TβRI/II in Müller glial cells migrated into fibrovascular tissues excised from eyes of PDR patients. Double-staining experiments for serial sections demonstrated colocalization of GFAP with GS (Figs. 4A–4C), TβRI (Figs. 4D–4F), and VEGF-A (Figs. 4G–4I). Importantly, VEGF-A colocalized with TβRI (Figs. 4J–4L) and TβRII (Figs. 4M–4O). These data suggested that Müller glial production of VEGF-A in the downstream of TβRI/II during diabetic fibrovascular proliferation, in consistence with the in vitro results showing the requirement of TβRI/II in TGF-β1/2–induced VEGFA expression (Figs. 3C, 3D). In contrast, VEGF-A application to Müller glial cells did not induce the expression of TGFB1 or TGFB2 (Supplementary Figs. S4A, S4B). 
Figure 4.
 
Colocalization of VEGF-A with TβRI/II in Müller glial cells migrated into diabetic fibrovascular tissues. (AO) Double labeling of GS (green) and GFAP (red) (AC), TβRI (green) and GFAP (red) (DF), VEGF-A (green) and GFAP (red) (GI), VEGF-A (green) and TβRI (red) (JL), and VEGF-A (green) and TβRII (red) (MO) in diabetic fibrovascular tissue specimens with DAPI (blue) counterstain to nuclei. Scale bar = 20 µm.
Figure 4.
 
Colocalization of VEGF-A with TβRI/II in Müller glial cells migrated into diabetic fibrovascular tissues. (AO) Double labeling of GS (green) and GFAP (red) (AC), TβRI (green) and GFAP (red) (DF), VEGF-A (green) and GFAP (red) (GI), VEGF-A (green) and TβRI (red) (JL), and VEGF-A (green) and TβRII (red) (MO) in diabetic fibrovascular tissue specimens with DAPI (blue) counterstain to nuclei. Scale bar = 20 µm.
TGF-β1/2–Induced VEGF-A Expression via Its Canonical and Noncanonical Signaling Pathways in Müller Glial Cells
Although TGF-β was reported to induce VEGF-A expression in rat Müller glial cells,23 its intracellular signal transduction remains to be clarified. We, therefore, used the same set of inhibitors with TGF-β autoinduction (Fig. 2) to detect signaling pathways to induce VEGFA expression in TGF-β1/2–stimulated human Müller cells. TGF-β1–induced VEGFA expression levels were significantly reduced by PI3K and p38 MAPK inhibitors, but enhanced by JNK and GSK-3 inhibitors (Fig. 5A). Silencing of SMAD2 significantly suppressed TGF-β1–induced expression of VEGFA (Fig. 5B), indicating involvement of canonical signaling. Moreover, we verified the impact of pharmacologic inhibition of PI3K and p38 MAPK (Fig. 5C) and siRNA-based knockdown of SMAD2 (Fig. 5D) on TGF-β1–induced production of VEGF-A protein as well. Consistently, the phosphorylation of AKT and SMAD2 increased substantially up to 24 hours after stimulation with TGF-β1 (Fig. 5E). TGF-β1–induced phosphorylation of SMAD2 was mitigated by inhibition of p38 MAPK but not PI3K (Fig. 5F), suggesting the activation of SMAD2 in the downstream of p38 MAPK. Similar results were obtained in the above series of experiments using TGF-β2 (Supplementary Figs. S5A–S5F). 
Figure 5.
 
TGF-β1-induced VEGF-A expression via its canonical and noncanonical signaling pathways in Müller glial cells. (A) Müller glial cells were pretreated with each inhibitor (U0126, ERK1/2; JSH-23, NF-κB; LY294002, PI3K; SP600125, JNK; SB203580, p38 MAPK; SB216763, GSK-3) at 10 µM for 30 minutes followed by stimulation with TGF-β1 (30 ng/mL) for 24 hours, and VEGFA mRNA levels were analyzed. (B) Control siRNA-treated (Ctrl-siRNA) and SMAD2-knockdown (siRNA-1 and siRNA-2) Müller glial cells were stimulated with TGF-β1 (30 ng/mL) for 24 hours, and VEGFA mRNA levels were analyzed. (C, D) Müller glial cells were pretreated with each inhibitor (C) and siRNAs against SMAD2 (D) followed by stimulation with TGF-β1 (30 ng/mL) for 48 hours, and protein levels of VEGF-A were analyzed. n = 3, *P < 0.05, **P < 0.01. (E) Müller glial cells were incubated with TGF-β1 (30 ng/mL), and protein levels of phosphorylated and total forms of AKT and SMAD2 were analyzed at various time points. (F) Müller glial cells were pretreated with each inhibitor followed by stimulation with TGF-β1 (30 ng/mL) for 30 minutes, and protein levels of phosphorylated and total forms of SMAD2 were analyzed.
Figure 5.
 
TGF-β1-induced VEGF-A expression via its canonical and noncanonical signaling pathways in Müller glial cells. (A) Müller glial cells were pretreated with each inhibitor (U0126, ERK1/2; JSH-23, NF-κB; LY294002, PI3K; SP600125, JNK; SB203580, p38 MAPK; SB216763, GSK-3) at 10 µM for 30 minutes followed by stimulation with TGF-β1 (30 ng/mL) for 24 hours, and VEGFA mRNA levels were analyzed. (B) Control siRNA-treated (Ctrl-siRNA) and SMAD2-knockdown (siRNA-1 and siRNA-2) Müller glial cells were stimulated with TGF-β1 (30 ng/mL) for 24 hours, and VEGFA mRNA levels were analyzed. (C, D) Müller glial cells were pretreated with each inhibitor (C) and siRNAs against SMAD2 (D) followed by stimulation with TGF-β1 (30 ng/mL) for 48 hours, and protein levels of VEGF-A were analyzed. n = 3, *P < 0.05, **P < 0.01. (E) Müller glial cells were incubated with TGF-β1 (30 ng/mL), and protein levels of phosphorylated and total forms of AKT and SMAD2 were analyzed at various time points. (F) Müller glial cells were pretreated with each inhibitor followed by stimulation with TGF-β1 (30 ng/mL) for 30 minutes, and protein levels of phosphorylated and total forms of SMAD2 were analyzed.
In addition, the inhibition of JNK and GSK-3 alone per se increased VEGFA expression, and synergistically increased TGF-β1–induced VEGFA expression (Supplementary Figs. S5G, S5H). These results indicated that TGF-β–induced VEGF-A expression was positively regulated by PI3K/AKT and p38 MAPK with its downstream SMAD2 canonical signaling, and negatively by JNK and GSK-3 in Müller glial cells. 
Activation of TGF-β1/2–Induced Signaling Molecules in Müller Glial Cells Migrated Into Diabetic Fibrovascular Tissues
To confirm the activation of TGF-β1/2–induced signaling molecules (Figs. 2 and 5) in diabetic fibrovascular proliferation, we performed immunofluorescence analyses of their phosphorylated forms in Müller glial cells migrated into fibrovascular tissues excised from eyes of PDR patients. Double-staining experiments for serial sections demonstrated colocalization of GFAP with GS (Figs. 6A–6C), phosphorylated NF-κB (Figs. 6D–6F), phosphorylated p38 MAPK (Figs. 6G–6I), phosphorylated GSK-3 (Figs. 6J–6L), phosphorylated AKT (Fig. 6M–6O), and phosphorylated SMAD2 (Figs. 6P–6R). In concert with the in vitro signal blocking results (Figs. 2 and 5), these findings suggested that Müller glial activation of signaling molecules related to TGF-β autoinduction and TGF-β–induced VEGF-A production during diabetic fibrovascular proliferation. 
Figure 6.
 
Activation of TGF-β1/2–induced signaling molecules in Müller glial cells migrated into diabetic fibrovascular tissues. (AR) Double labeling of GS (green) and GFAP (red) (AC), phosphorylated NF-κB (green) and GFAP (red) (DF), phosphorylated p38 (green) and GFAP (red) (GI), phosphorylated GSK-3 (green) and GFAP (red) (J–L), phosphorylated AKT (green) and GFAP (red) (MO) and phosphorylated SMAD2 (green) and GFAP (red) (PR) in diabetic fibrovascular tissue specimens with DAPI (blue) counterstain to nuclei. Scale bar = 20 µm.
Figure 6.
 
Activation of TGF-β1/2–induced signaling molecules in Müller glial cells migrated into diabetic fibrovascular tissues. (AR) Double labeling of GS (green) and GFAP (red) (AC), phosphorylated NF-κB (green) and GFAP (red) (DF), phosphorylated p38 (green) and GFAP (red) (GI), phosphorylated GSK-3 (green) and GFAP (red) (J–L), phosphorylated AKT (green) and GFAP (red) (MO) and phosphorylated SMAD2 (green) and GFAP (red) (PR) in diabetic fibrovascular tissue specimens with DAPI (blue) counterstain to nuclei. Scale bar = 20 µm.
Discussion
In contrast with the subretinal space attached with the RPE cell monolayer, no cells of epithelial origin are present at the epiretinal (i.e., intravitreal) space where fibrovascular proliferation occurs secondary to PDR; therefore, the cellular source of myofibroblasts and the mechanism of transdifferentiation remain incompletely understood. The present study demonstrated the following novel findings on the multifaceted role of Müller glial TGF-β together with its regulatory signaling pathways in the pathogenesis of diabetic epiretinal fibrovascular proliferation. First, TGF-β1/2–initiated Müller GMT, recently recognized as the etiology of idiopathic epiretinal membrane,9 could also apply to epiretinal fibrovascular proliferation in patients with PDR (Fig. 1). Second, TGF-β1/2 autoinduction, as seen in RPE cells during EMT for subretinal fibrovascular proliferation owing to nAMD,6 would also be appropriated to the amplifying process of Müller GMT in diabetic fibrovascular proliferation (Figs. 2 and 6). Third, screening of PDR-associated profibrotic cytokines identified TGF-β1/2 as the exclusive inducer of VEGF-A production in Müller glial cells (Fig. 3). Finally, Müller glial autoinduction of TGF-β1/2 concurrently accelerated autocrine signal transduction for VEGF-A production (Figs. 46). Collectively, complicated cross-talk existed between multiple signaling pathways in Müller glial cells, implying the possible modification of balance between fibrogenic Müller GMT and angiogenic VEGF-A production (Fig. 7). 
Figure 7.
 
Dual involvement of Müller glial autoinduction of TGF-β1/2 in diabetic fibrovascular proliferation via GMT and VEGF-A production. Müller glial autoinduction of TGF-β1/2 dually causes the pathogenic cue to facilitate and amplify diabetic fibrovascular proliferation via TGF-β1/2-driven fibrogenic GMT and VEGF-A–driven angiogenesis.
Figure 7.
 
Dual involvement of Müller glial autoinduction of TGF-β1/2 in diabetic fibrovascular proliferation via GMT and VEGF-A production. Müller glial autoinduction of TGF-β1/2 dually causes the pathogenic cue to facilitate and amplify diabetic fibrovascular proliferation via TGF-β1/2-driven fibrogenic GMT and VEGF-A–driven angiogenesis.
The currently observed TGF-β–initiated Müller GMT in diabetic fibrovascular proliferation was originally identified as the molecular mechanism of idiopathic epiretinal membrane,9 a clinical entity characterized by avascular fibrocellular proliferation with much lower levels of VEGF-A in the vitreous compared with PDR.24 One possible explanation for the PDR-specific potent induction of VEGF-A in TGF-β–stimulated Müller glial cells would be much higher levels of angiotensin II in the vitreous of eyes with PDR.25 Because angiotensin II was shown to upregulate TGF-β1 expression in Müller glial cells,26 the increased concentration of angiotensin II in PDR eyes could thus reach and promote the vicious cycle of TGF-β1/2 autoinduction, leading to sufficiently augmented levels of TGF-β1/2 for Müller glial VEGF-A production. Indeed, TβRI/II-positive Müller glial cells were immunoreactive for VEGF-A in our patient tissues, in accordance with the positive correlation between VEGF-A and TGF-β1/22 as well as angiotensin II25 in PDR eyes. Consistently, our previous data with diabetic mice demonstrated that angiotensin II signaling contributed to retinal expression of VEGF-A,27 the concentration of which reasonably correlated with renin enzymatic activity28 and soluble (pro)renin receptor24 in the vitreous of patients with PDR. Independently of TGF-β, angiotensin II was also shown to directly induce VEGF-A expression in pericytes29 and macrophages,30 further enhancing the angiogenic activity of PDR. 
The cellular sources of myofibroblasts with their molecular markers in diabetic fibrovascular tissues were previously suggested to include Müller glial cells positive for GS, GFAP, and α-SMA10; vascular endothelial cells positive for CD31 and α-SMA31; and bone marrow-derived fibrocytes positive for CD45, CD11b, and α-SMA,31 as well as CD45 and C-C chemokine receptor 2 and type I collagen.32 Additionally, the mesenchymal transcription factor SNAIL was shown to be immunoreactive despite no demonstration of its cellular origin in diabetic fibrovascular tissues.33 As concerns avascular fibrocellular proliferation associated with diabetic macular edema, the cellular source of myofibroblasts with its molecular markers in epiretinal membranes was shown to be hyalocytes positive for CD45, CD64, and α-SMA,34 which would possibly be involved in more advanced fibrovascular tissues as resident macrophages at the epiretinal space. Resultant phenotypes in these kinds of cells stimulated with TGF-β were reported to include collagen gel hypercontraction,35 increased cell migration and proliferation, and upregulation of SNAIL, SM22, α-SMA, type I collagen, and fibronectin9 in Müller glial cells; upregulation of SNAIL, SM22, and fibroblast-specific protein-1 in vascular endothelial cells31; upregulation of α-SMA in fibrocytes32; and collagen gel hypercontraction and upregulation of α-SMA in hyalocytes.36 In contrast, TGF-β–induced phenotypic changes in contractility and α-SMA expression were only marginal in astrocytes,36 the major gliosis-related cellular component in non-neovascular epiretinal membranes associated with diabetic macular edema37 as well as PDR.38 In addition to Müller GMT, diabetic fibrovascular proliferation would therefore result from myofibroblastic differentiation from other cellular sources, including vascular endothelial cells (commonly known as EndoMT or endothelial-mesenchymal transition), fibrocytes, and possibly hyalocytes. Notably, high glucose stimulation to Müller cells was shown to induce SNAIL-mediated CTGF and fibronectin expression,33 which would imply the potential involvement of TGF-β–independent Müller GMT in diabetic fibrovascular proliferation. 
In the present study, Müller glial VEGF-A production was achieved exclusively by stimulation with TGF-β1/2 alone out of various profibrotic cytokines. Similarly, TGF-β–specific induction of VEGF-A was also found in RPE cells via p38 MAPK and PI3K/AKT, although ERK and JNK dependencies were different compared with Müller glial VEGF-A production.21,22 Interestingly, TGF-β solely was shown to induce, in an exclusive manner as well, Müller GMT9 and EMT from RPE cells,5 both of which were mediated via SNAIL. In the laser-induced model of choroidal neovascularization or subretinal fibrovascular proliferation, TGF-β inhibition in mice reduced SMAD2/3 phosphorylation, VEGF-A production, and the size of lesions, all of which were otherwise augmented.39 SMAD3-deficient mice in this model, also mitigating the size of lesions, suppressed the expression of VEGF-A and TGF-β compared with wild-type animals,7 suggesting the faciliatory role of TGF-β–SMAD2/3 canonical signaling in the expression of both VEGF-A and TGF-β (i.e., autoinduction of TGF-β) in RPE cells. In contrast, this study revealed the activation of SMAD2 canonical signaling in Müller glial VEGF-A production but not TGF-β autoinduction. Subretinal and epiretinal fibrovascular proliferations, although the signaling pathways were somewhat different between RPE and Müller cells, would consequently share the resultant common process whereby self-amplified TGF-β signaling dually governs angiogenesis and fibrosis. In this sense, nAMD and PDR, both of which are the main cause of blindness worldwide, could reasonably be regarded as TGF-β–induced fibrovascular proliferation disorders targeting RPE and Müller cells, respectively. 
Anti–VEGF-A treatments have hitherto brought a limited or negligible efficacy to ocular fibrosis in the clinical setting. Although preclinical investigation would be quite a hurdle at present owing to a lack of appropriate animal models of diabetic fibrovascular proliferation, TGF-β signaling is considered a promising candidate to be targeted for the management of PDR on the basis of the current and previous data.2,3,9 In contrast, TGF-β signaling was shown to be required for pericyte differentiation and vascular integrity in the retina of neonatal mice,40 thus leading to the notion that TGF-β would rather be protective in eyes with early stage non-PDR, which begins with pericyte loss.41 These accumulated piles of evidence warrant further research to seek for future therapeutic strategy, given that TGF-β may function in a stage-specific manner. Nevertheless, TGF-β would thus be an important molecular candidate targeted alone or in combination with anti–VEGF-A treatment in the long-term management of diabetic patients with PDR as well as severe non-PDR to inhibit fibrovascular proliferation. 
Acknowledgments
The authors thank Ikuyo Hirose and Shiho Yoshida (Hokkaido University) for their skilled technical assistance. 
Supported by Grant-in-Aid for Scientific Research (C) (19K09944 to A.K.) of the Japan Society for the Promotion of Science. D.W. and Y.L. are recipients of a scholarship from the Otsuka Toshimi Scholarship Foundation and China Scholarship Council, respectively. 
Disclosure: D. Wu, None; A. Kanda, None; Y. Liu, None; K. Noda, None; M. Murata, None; S. Ishida, None 
References
Walshe R, Esser P, Wiedemann P, Heimann K. Proliferative retinal diseases: myofibroblasts cause chronic vitreoretinal traction. Br J Ophthalmol. 1992; 76: 550–552. [CrossRef] [PubMed]
Dai Y, Wu Z, Wang F, Zhang Z, Yu M. Identification of chemokines and growth factors in proliferative diabetic retinopathy vitreous. Biomed Res Int. 2014; 2014: 486386. [PubMed]
Bochaton-Piallat ML, Kapetanios AD, Donati G, Redard M, Gabbiani G, Pournaras CJ. TGF-beta1, TGF-beta receptor II and ED-A fibronectin expression in myofibroblast of vitreoretinopathy. Invest Ophthalmol Vis Sci. 2000; 41: 2336–2342. [PubMed]
Ishida S, Shinoda K, Kawashima S, Oguchi Y, Okada Y, Ikeda E. Coexpression of VEGF receptors VEGF-R2 and neuropilin-1 in proliferative diabetic retinopathy. Invest Ophthalmol Vis Sci. 2000; 41: 1649–1656. [PubMed]
Hirasawa M, Noda K, Noda S, et al. Transcriptional factors associated with epithelial-mesenchymal transition in choroidal neovascularization. Mol Vis. 2011; 17: 1222–1230. [PubMed]
Wu D, Kanda A, Liu Y, Kase S, Noda K, Ishida S. Galectin-1 promotes choroidal neovascularization and subretinal fibrosis mediated via epithelial-mesenchymal transition. FASEB J. 2019; 33: 2498–2513. [CrossRef] [PubMed]
Iwanishi H, Fujita N, Tomoyose K, et al. Inhibition of development of laser-induced choroidal neovascularization with suppression of infiltration of macrophages in Smad3-null mice. Lab Invest. 2016; 96: 641–651. [CrossRef] [PubMed]
Ishikawa K, Kannan R, Hinton DR. Molecular mechanisms of subretinal fibrosis in age-related macular degeneration. Exp Eye Res. 2016; 142: 19–25. [CrossRef] [PubMed]
Kanda A, Noda K, Hirose I, Ishida S. TGF-beta-SNAIL axis induces Muller glial-mesenchymal transition in the pathogenesis of idiopathic epiretinal membrane. Sci Rep. 2019; 9: 673. [CrossRef] [PubMed]
Guidry C, King JL, Mason JO, 3rd. Fibrocontractive Muller cell phenotypes in proliferative diabetic retinopathy. Invest Ophthalmol Vis Sci. 2009; 50: 1929–1939. [CrossRef] [PubMed]
Limb GA, Salt TE, Munro PM, Moss SE, Khaw PT. In vitro characterization of a spontaneously immortalized human Muller cell line (MIO-M1). Invest Ophthalmol Vis Sci. 2002; 43: 864–869. [PubMed]
Xu J, Lamouille S, Derynck R. TGF-beta-induced epithelial to mesenchymal transition. Cell Res. 2009; 19: 156–172. [CrossRef] [PubMed]
Zhang YE. Non-Smad pathways in TGF-beta signaling. Cell Res. 2009; 19: 128–139. [CrossRef] [PubMed]
Saika S. TGFbeta pathobiology in the eye. Lab Invest. 2006; 86: 106–115. [CrossRef] [PubMed]
Hoot KE, Lighthall J, Han G, et al. Keratinocyte-specific Smad2 ablation results in increased epithelial-mesenchymal transition during skin cancer formation and progression. J Clin Invest. 2008; 118: 2722–2732. [PubMed]
Zhang M, Fraser D, Phillips A. ERK, p38, and Smad signaling pathways differentially regulate transforming growth factor-beta1 autoinduction in proximal tubular epithelial cells. Am J Pathol. 2006; 169: 1282–1293. [CrossRef] [PubMed]
Kuiper EJ, Van Nieuwenhoven FA, de Smet MD, et al. The angio-fibrotic switch of VEGF and CTGF in proliferative diabetic retinopathy. PLoS One. 2008; 3: e2675. [CrossRef] [PubMed]
Praidou A, Papakonstantinou E, Androudi S, Georgiadis N, Karakiulakis G, Dimitrakos S. Vitreous and serum levels of vascular endothelial growth factor and platelet-derived growth factor and their correlation in patients with non-proliferative diabetic retinopathy and clinically significant macula oedema. Acta Ophthalmol. 2011; 89: 248–254. [CrossRef] [PubMed]
Sivalingam A, Kenney J, Brown GC, Benson WE, Donoso L. Basic fibroblast growth factor levels in the vitreous of patients with proliferative diabetic retinopathy. Arch Ophthalmol. 1990; 108: 869–872. [CrossRef] [PubMed]
Boss JD, Singh PK, Pandya HK, et al. Assessment of neurotrophins and inflammatory mediators in vitreous of patients with diabetic retinopathy. Invest Ophthalmol Vis Sci. 2017; 58: 5594–5603. [CrossRef] [PubMed]
Nagineni CN, Samuel W, Nagineni S, et al. Transforming growth factor-beta induces expression of vascular endothelial growth factor in human retinal pigment epithelial cells: involvement of mitogen-activated protein kinases. J Cell Physiol. 2003; 197: 453–462. [CrossRef] [PubMed]
Bian ZM, Elner SG, Elner VM. Regulation of VEGF mRNA expression and protein secretion by TGF-beta2 in human retinal pigment epithelial cells. Exp Eye Res. 2007; 84: 812–822. [CrossRef] [PubMed]
Behzadian MA, Wang XL, Al-Shabrawey M, Caldwell RB. Effects of hypoxia on glial cell expression of angiogenesis-regulating factors VEGF and TGF-beta. Glia. 1998; 24: 216–225. [CrossRef] [PubMed]
Kanda A, Noda K, Saito W, Ishida S. (Pro)renin receptor is associated with angiogenic activity in proliferative diabetic retinopathy. Diabetologia. 2012; 55: 3104–3113. [CrossRef] [PubMed]
Funatsu H, Yamashita H, Nakanishi Y, Hori S. Angiotensin II and vascular endothelial growth factor in the vitreous fluid of patients with proliferative diabetic retinopathy. Br J Ophthalmol. 2002; 86: 311–315. [CrossRef] [PubMed]
Dong Y, Kanda A, Noda K, Saito W, Ishida S. Pathologic roles of receptor-associated prorenin system in idiopathic emiretinal Membrane. Sci Rep. 2017; 7: 44266. [CrossRef] [PubMed]
Nagai N, Izumi-Nagai K, Oike Y, et al. Suppression of diabetes-induced retinal inflammation by blocking the angiotensin II type 1 receptor or its downstream nuclear factor-kappaB pathway. Invest Ophthalmol Vis Sci. 2007; 48: 4342–4350. [CrossRef] [PubMed]
Kanda A, Noda K, Saito W, Ishida S. Vitreous renin activity correlates with vascular endothelial growth factor in proliferative diabetic retinopathy. Br J Ophthalmol. 2013; 97: 666–668. [CrossRef] [PubMed]
Otani A, Takagi H, Oh H, et al. Angiotensin II-stimulated vascular endothelial growth factor expression in bovine retinal pericytes. Invest Ophthalmol Vis Sci. 2000; 41: 1192–1199. [PubMed]
Nagai N, Oike Y, Izumi-Nagai K, et al. Angiotensin II type 1 receptor-mediated inflammation is required for choroidal neovascularization. Arterioscler Thromb Vasc Biol. 2006; 26: 2252–2259. [CrossRef] [PubMed]
Abu El-Asrar AM, De Hertogh G, van den Eynde K, et al. Myofibroblasts in proliferative diabetic retinopathy can originate from infiltrating fibrocytes and through endothelial-to-mesenchymal transition (EndoMT). Exp Eye Res. 2015; 132: 179–189. [CrossRef] [PubMed]
Tamaki K, Usui-Ouchi A, Murakami A, Ebihara N. Fibrocytes and fibrovascular membrane formation in proliferative diabetic retinopathy. Invest Ophthalmol Vis Sci. 2016; 57: 4999–5005. [CrossRef] [PubMed]
Zhou T, Che D, Lan Y, et al. Mesenchymal marker expression is elevated in Muller cells exposed to high glucose and in animal models of diabetic retinopathy. Oncotarget. 2017; 8: 4582–4594. [CrossRef] [PubMed]
Hagenau F, Vogt D, Ziada J, et al. Vitrectomy for diabetic macular edema: optical coherence tomography criteria and pathology of the vitreomacular interface. Am J Ophthalmol. 2019; 200: 34–46. [CrossRef] [PubMed]
Guidry C, Bradley KM, King JL. Tractional force generation by human muller cells: growth factor responsiveness and integrin receptor involvement. Invest Ophthalmol Vis Sci. 2003; 44: 1355–1363. [CrossRef] [PubMed]
Kohno RI, Hata Y, Kawahara S, et al. Possible contribution of hyalocytes to idiopathic epiretinal membrane formation and its contraction. Br J Ophthalmol. 2009; 93: 1020–1026. [CrossRef] [PubMed]
Gandorfer A, Rohleder M, Grosselfinger S, Haritoglou C, Ulbig M, Kampik A. Epiretinal pathology of diffuse diabetic macular edema associated with vitreomacular traction. Am J Ophthalmol. 2005; 139: 638–652. [CrossRef] [PubMed]
Hosoda Y, Okada M, Matsumura M, Ogino N, Honda Y, Nagai Y. Epiretinal membrane of proliferative diabetic retinopathy: an immunohistochemical study. Ophthalmic Res. 1993; 25: 289–294. [CrossRef] [PubMed]
Wang X, Ma W, Han S, et al. TGF-beta participates choroid neovascularization through Smad2/3-VEGF/TNF-alpha signaling in mice with Laser-induced wet age-related macular degeneration. Sci Rep. 2017; 7: 9672. [CrossRef] [PubMed]
Braunger BM, Leimbeck SV, Schlecht A, Volz C, Jagle H, Tamm ER. Deletion of ocular transforming growth factor beta signaling mimics essential characteristics of diabetic retinopathy. Am J Pathol. 2015; 185: 1749–1768. [CrossRef] [PubMed]
Wheeler SE, Lee NY. Emerging Roles of transforming growth factor beta signaling in diabetic retinopathy. J Cell Physiol. 2017; 232: 486–489. [CrossRef] [PubMed]
Figure 1.
 
Expression of GMT-related molecular markers in Müller glial cells migrated into diabetic fibrovascular tissues. (AR) Double labeling of GS (green) and GFAP (red) (AC), TβRI (green) and GFAP (red) (DF), SNAIL (green) and GFAP (red) (GI), α-SMA (green) and GFAP (red) (JL), GFAP (green) and SM22 (red) (MO), and GS (green) and SM22 (red) (PR) in diabetic fibrovascular tissue specimens with DAPI (blue) counterstain to nuclei. Scale bar = 20 µm.
Figure 1.
 
Expression of GMT-related molecular markers in Müller glial cells migrated into diabetic fibrovascular tissues. (AR) Double labeling of GS (green) and GFAP (red) (AC), TβRI (green) and GFAP (red) (DF), SNAIL (green) and GFAP (red) (GI), α-SMA (green) and GFAP (red) (JL), GFAP (green) and SM22 (red) (MO), and GS (green) and SM22 (red) (PR) in diabetic fibrovascular tissue specimens with DAPI (blue) counterstain to nuclei. Scale bar = 20 µm.
Figure 2.
 
Autoinduction of TGFB1 expression via its noncanonical signaling pathways in Müller glial cells. (A) Müller glial cells were incubated with TGF-β1 or TGF-β2 (30 ng/mL) for 24 hours, and TGFB1 mRNA levels were analyzed. (B) Müller glial cells were pretreated with each inhibitor (U0126, ERK1/2; JSH-23, NF-κB; LY294002, PI3K; SP600125, JNK; SB203580, p38 MAPK; SB216763, GSK-3) at 10 µM for 30 minutes followed by stimulation with TGF-β1 or TGF-β2 (30 ng/mL) for 24 hours, and TGFB1 mRNA levels were analyzed. (C) SMAD2-siRNAs (siRNA-1 and siRNA-2) were transfected in Müller glial cells, and SMAD2 mRNA levels were analyzed. (D) Control siRNA-treated (Ctrl-siRNA) and SMAD2-knockdown (siRNA-1 and siRNA-2) Müller glial cells were stimulated with TGF-β1, and TGFB1 mRNA levels were analyzed. (E) Müller glial cells were incubated with TGF-β1, and protein levels of phosphorylated and total forms of NF-κB, p38, and GSK-3α/β were analyzed at various time points. (F, G) Müller glial cells were pretreated with normal IgG (300 ng/mL), anti-TβRI antibody (300 ng/mL), TβRI kinase inhibitor SB431542 (10 µg/mL), or anti-TβRII antibody (250 ng/mL) for 30 minutes followed by stimulation with TGF-β1 or TGF-β2 (30 ng/mL) for 24 hours, and TGFB1 mRNA levels were analyzed. n = 3, *P < 0.05, **P < 0.01. n.s., not significant.
Figure 2.
 
Autoinduction of TGFB1 expression via its noncanonical signaling pathways in Müller glial cells. (A) Müller glial cells were incubated with TGF-β1 or TGF-β2 (30 ng/mL) for 24 hours, and TGFB1 mRNA levels were analyzed. (B) Müller glial cells were pretreated with each inhibitor (U0126, ERK1/2; JSH-23, NF-κB; LY294002, PI3K; SP600125, JNK; SB203580, p38 MAPK; SB216763, GSK-3) at 10 µM for 30 minutes followed by stimulation with TGF-β1 or TGF-β2 (30 ng/mL) for 24 hours, and TGFB1 mRNA levels were analyzed. (C) SMAD2-siRNAs (siRNA-1 and siRNA-2) were transfected in Müller glial cells, and SMAD2 mRNA levels were analyzed. (D) Control siRNA-treated (Ctrl-siRNA) and SMAD2-knockdown (siRNA-1 and siRNA-2) Müller glial cells were stimulated with TGF-β1, and TGFB1 mRNA levels were analyzed. (E) Müller glial cells were incubated with TGF-β1, and protein levels of phosphorylated and total forms of NF-κB, p38, and GSK-3α/β were analyzed at various time points. (F, G) Müller glial cells were pretreated with normal IgG (300 ng/mL), anti-TβRI antibody (300 ng/mL), TβRI kinase inhibitor SB431542 (10 µg/mL), or anti-TβRII antibody (250 ng/mL) for 30 minutes followed by stimulation with TGF-β1 or TGF-β2 (30 ng/mL) for 24 hours, and TGFB1 mRNA levels were analyzed. n = 3, *P < 0.05, **P < 0.01. n.s., not significant.
Figure 3.
 
Screening of PDR-associated profibrotic cytokines and Müller glial VEGF-A induction exclusively by TGF-β1/2. (A) Müller glial cells were applied with TGF-β1, TGF-β2, CTGF, FGF2, NGF, and PDGF-BB at the dose of 10 ng/mL for 24 hours, and VEGFA mRNA levels were analyzed. (B) Müller glial cells were incubated with TGF-β1 or TGF-β2 at different doses, and protein levels of VEGF-A were analyzed. (C, D) Müller glial cells were pretreated with normal IgG (300 ng/mL), anti-TβRI antibody (300 ng/mL), TβRI kinase inhibitor SB431542 (10 µg/mL), or anti-TβRII antibody (250 ng/mL) for 30 minutes followed by stimulation with TGF-β1 or TGF-β2 (30 ng/mL) for 24 hours, and VEGFA mRNA levels were analyzed. (E) Müller glial cells were stimulated with TGF-β1 or TGF-β2 (30 ng/mL) for 24 hours, and supernatants were collected. Human retinal microvascular endothelial cells were preincubated with anti–VEGF-A drugs bevacizumab (0.3125 mg/mL), aflibercept (0.5 mg/mL), and normal IgG (0.5 mg/mL) for 15 minutes, followed by treatment with TGF-β1/2-stimulated Müller cell culture medium for 48 hours, and processed for cell viability assay. n = 3, *P < 0.05, **P < 0.01.
Figure 3.
 
Screening of PDR-associated profibrotic cytokines and Müller glial VEGF-A induction exclusively by TGF-β1/2. (A) Müller glial cells were applied with TGF-β1, TGF-β2, CTGF, FGF2, NGF, and PDGF-BB at the dose of 10 ng/mL for 24 hours, and VEGFA mRNA levels were analyzed. (B) Müller glial cells were incubated with TGF-β1 or TGF-β2 at different doses, and protein levels of VEGF-A were analyzed. (C, D) Müller glial cells were pretreated with normal IgG (300 ng/mL), anti-TβRI antibody (300 ng/mL), TβRI kinase inhibitor SB431542 (10 µg/mL), or anti-TβRII antibody (250 ng/mL) for 30 minutes followed by stimulation with TGF-β1 or TGF-β2 (30 ng/mL) for 24 hours, and VEGFA mRNA levels were analyzed. (E) Müller glial cells were stimulated with TGF-β1 or TGF-β2 (30 ng/mL) for 24 hours, and supernatants were collected. Human retinal microvascular endothelial cells were preincubated with anti–VEGF-A drugs bevacizumab (0.3125 mg/mL), aflibercept (0.5 mg/mL), and normal IgG (0.5 mg/mL) for 15 minutes, followed by treatment with TGF-β1/2-stimulated Müller cell culture medium for 48 hours, and processed for cell viability assay. n = 3, *P < 0.05, **P < 0.01.
Figure 4.
 
Colocalization of VEGF-A with TβRI/II in Müller glial cells migrated into diabetic fibrovascular tissues. (AO) Double labeling of GS (green) and GFAP (red) (AC), TβRI (green) and GFAP (red) (DF), VEGF-A (green) and GFAP (red) (GI), VEGF-A (green) and TβRI (red) (JL), and VEGF-A (green) and TβRII (red) (MO) in diabetic fibrovascular tissue specimens with DAPI (blue) counterstain to nuclei. Scale bar = 20 µm.
Figure 4.
 
Colocalization of VEGF-A with TβRI/II in Müller glial cells migrated into diabetic fibrovascular tissues. (AO) Double labeling of GS (green) and GFAP (red) (AC), TβRI (green) and GFAP (red) (DF), VEGF-A (green) and GFAP (red) (GI), VEGF-A (green) and TβRI (red) (JL), and VEGF-A (green) and TβRII (red) (MO) in diabetic fibrovascular tissue specimens with DAPI (blue) counterstain to nuclei. Scale bar = 20 µm.
Figure 5.
 
TGF-β1-induced VEGF-A expression via its canonical and noncanonical signaling pathways in Müller glial cells. (A) Müller glial cells were pretreated with each inhibitor (U0126, ERK1/2; JSH-23, NF-κB; LY294002, PI3K; SP600125, JNK; SB203580, p38 MAPK; SB216763, GSK-3) at 10 µM for 30 minutes followed by stimulation with TGF-β1 (30 ng/mL) for 24 hours, and VEGFA mRNA levels were analyzed. (B) Control siRNA-treated (Ctrl-siRNA) and SMAD2-knockdown (siRNA-1 and siRNA-2) Müller glial cells were stimulated with TGF-β1 (30 ng/mL) for 24 hours, and VEGFA mRNA levels were analyzed. (C, D) Müller glial cells were pretreated with each inhibitor (C) and siRNAs against SMAD2 (D) followed by stimulation with TGF-β1 (30 ng/mL) for 48 hours, and protein levels of VEGF-A were analyzed. n = 3, *P < 0.05, **P < 0.01. (E) Müller glial cells were incubated with TGF-β1 (30 ng/mL), and protein levels of phosphorylated and total forms of AKT and SMAD2 were analyzed at various time points. (F) Müller glial cells were pretreated with each inhibitor followed by stimulation with TGF-β1 (30 ng/mL) for 30 minutes, and protein levels of phosphorylated and total forms of SMAD2 were analyzed.
Figure 5.
 
TGF-β1-induced VEGF-A expression via its canonical and noncanonical signaling pathways in Müller glial cells. (A) Müller glial cells were pretreated with each inhibitor (U0126, ERK1/2; JSH-23, NF-κB; LY294002, PI3K; SP600125, JNK; SB203580, p38 MAPK; SB216763, GSK-3) at 10 µM for 30 minutes followed by stimulation with TGF-β1 (30 ng/mL) for 24 hours, and VEGFA mRNA levels were analyzed. (B) Control siRNA-treated (Ctrl-siRNA) and SMAD2-knockdown (siRNA-1 and siRNA-2) Müller glial cells were stimulated with TGF-β1 (30 ng/mL) for 24 hours, and VEGFA mRNA levels were analyzed. (C, D) Müller glial cells were pretreated with each inhibitor (C) and siRNAs against SMAD2 (D) followed by stimulation with TGF-β1 (30 ng/mL) for 48 hours, and protein levels of VEGF-A were analyzed. n = 3, *P < 0.05, **P < 0.01. (E) Müller glial cells were incubated with TGF-β1 (30 ng/mL), and protein levels of phosphorylated and total forms of AKT and SMAD2 were analyzed at various time points. (F) Müller glial cells were pretreated with each inhibitor followed by stimulation with TGF-β1 (30 ng/mL) for 30 minutes, and protein levels of phosphorylated and total forms of SMAD2 were analyzed.
Figure 6.
 
Activation of TGF-β1/2–induced signaling molecules in Müller glial cells migrated into diabetic fibrovascular tissues. (AR) Double labeling of GS (green) and GFAP (red) (AC), phosphorylated NF-κB (green) and GFAP (red) (DF), phosphorylated p38 (green) and GFAP (red) (GI), phosphorylated GSK-3 (green) and GFAP (red) (J–L), phosphorylated AKT (green) and GFAP (red) (MO) and phosphorylated SMAD2 (green) and GFAP (red) (PR) in diabetic fibrovascular tissue specimens with DAPI (blue) counterstain to nuclei. Scale bar = 20 µm.
Figure 6.
 
Activation of TGF-β1/2–induced signaling molecules in Müller glial cells migrated into diabetic fibrovascular tissues. (AR) Double labeling of GS (green) and GFAP (red) (AC), phosphorylated NF-κB (green) and GFAP (red) (DF), phosphorylated p38 (green) and GFAP (red) (GI), phosphorylated GSK-3 (green) and GFAP (red) (J–L), phosphorylated AKT (green) and GFAP (red) (MO) and phosphorylated SMAD2 (green) and GFAP (red) (PR) in diabetic fibrovascular tissue specimens with DAPI (blue) counterstain to nuclei. Scale bar = 20 µm.
Figure 7.
 
Dual involvement of Müller glial autoinduction of TGF-β1/2 in diabetic fibrovascular proliferation via GMT and VEGF-A production. Müller glial autoinduction of TGF-β1/2 dually causes the pathogenic cue to facilitate and amplify diabetic fibrovascular proliferation via TGF-β1/2-driven fibrogenic GMT and VEGF-A–driven angiogenesis.
Figure 7.
 
Dual involvement of Müller glial autoinduction of TGF-β1/2 in diabetic fibrovascular proliferation via GMT and VEGF-A production. Müller glial autoinduction of TGF-β1/2 dually causes the pathogenic cue to facilitate and amplify diabetic fibrovascular proliferation via TGF-β1/2-driven fibrogenic GMT and VEGF-A–driven angiogenesis.
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×