TMP Prevents Retinal Neovascularization and Imparts Neuroprotection in an Oxygen-Induced Retinopathy Model

Abstract

Purpose: To evaluate the effects of tetramethylpyrazine (TMP) on retinal neovascularization (NV) and neuroprotection in an oxygen-induced retinopathy (OIR) model.

Methods: Neonatal C57BL/6J mice were subjected to 75% oxygen from postnatal day 7 (P7) to P12 and then returned to room air. TMP (200 mg/kg) or normal saline was given daily from P12 to P17. Immunostaining, HE staining, TUNEL assay, and RT-PCR were used to assess the effects of TMP on retinal neurovascular repair.

Results. : TMP effectively prevented pathologic NV and accelerated physiologic revascularization by enhancing the formation of endothelial tip cells at the edges of the repairing capillary networks and preserving the astrocytic template in the avascular retina. TMP also prevented morphologic changes and significantly decreased TUNEL-positive cells in the avascular retina by rescuing neurons such as amacrine, rod bipolar, horizontal, and Müller cells. In TMP-treated mice retinas, there was a less obvious loss of amacrine cell bodies and their distinct bands; the number of both rod bipolar and horizontal cell bodies, as well as the density of their dendrites in the outer plexiform layer, was greater than that in OIR control mice. TMP not only decreased the loss of alignment of Müller cell bodies and distortion of processes but reduced the reactive expression of GFAP in Müller cells. Furthermore, HIF-1α and VEGF mRNA expression were downregulated in TMP-treated mice retinas.

Conclusions: TMP improved neurovascular recovery by preventing NV and protecting retinal astroglia cells and neurons from ischemia-induced cell death partially due to its downregulation of HIF-1α and VEGF mRNA expression.

Introduction

Pathologic ocular neovascularization (NV) and associated vascular leakage in diabetic retinopathy, exudative age-related macular degeneration, retinopathy of prematurity, and vascular occlusions are leading causes of blindness worldwide. ^1,2 Considerable scientific and clinical work has focused on identifying the mechanisms of vascular injury leading to pathologic vitreoretinal NV, whereas recent studies show that local neurons and glial cells are also affected, associated with abnormal growth of blood vessels. ^3–9  

Experiments in various animal models of ischemia have suggested that retinopathy is associated with changes in a spectrum of cells, including vascular endothelial cells, astrocytes, retinal neurons, and Müller glia. ^10–15 Intensive studies have demonstrated that neuronal apoptosis and subsequent degeneration occur in the ischemic retina. ^16–18 Furthermore, glial dysfunction has also been reported in the hypoxia retina. ^7,12 Numerous studies have shown that astrocytes and Müller glia are essential for guiding the retinal vasculature. ¹⁹ Astrocytes and Müller cells usually provide support for retinal neurons, ^20,21 secrete VEGF for angiogenic sprouts, ¹⁹ and impart blood retinal barrier properties to endothelia. ^9,22 Moreover, astrocytes form a template that provides guidance for the developing vascular network. ²³ Thus, the dysfunction of neurons and glial cells may exacerbate the aberrant vessel growth following ischemic injury and contribute to progression of the disease. ^24–26 Treatments with angiogenic inhibitors or genetic manipulations directed toward reversing vascular permeability and eliminating NV need to address not only the vascular changes but also the alterations in neuronal and glial function. ²⁷ Thus, an ideal therapeutic treatment for ischemic retinopathy should prevent pathologic vitreoretinal NV, rescue the retinal neurons and glial cells, and promote physiologic retinal revascularization. 

Tetramethylpyrazine (TMP) is one of the most important active ingredients of the traditional Chinese herbal medicine, Ligusticum wallichii Franchat (Chung Xiong). It has been widely used for treatments of neurovascular disorders, such as ischemic stroke and pulmonary hypertension secondary to chronic obstructive pulmonary diseases in China. ^28–30 Previous studies have suggested strong neuroprotective effects and potential antiangiogenic properties of TMP both in vitro and in vivo. ^30–35 TMP has been demonstrated to scavenge reactive oxygen species, inhibit platelet aggregation, dilate blood vessels, depress blood viscosity, improve microcirculation, and increase coronary and cerebral blood flow. ^28,36–38 It has been shown that TMP efficiently protects retinal cells against hydrogen peroxide–induced oxidative stress in mixed rat retinal cell cultures. ³⁹ In addition, TMP protects photoreceptor cells of rats against retinal damage. ⁴⁰ It is also demonstrated that TMP can inhibit laser-induced experimental choroidal neovascularization in a rat model. ³⁴ These studies imply the potential protective effects of TMP in ischemic retinopathy. However, the therapeutic effects of TMP in ischemic retinopathy remain largely uncharacterized. 

The aim of the present study was to investigate the therapeutic benefit of TMP during the ischemic hypoxia phase of ischemic retinopathy in an oxygen-induced retinopathy (OIR) mouse model. By using this model, the interplay among angiogenesis, neuronal preservation, and the glial response after TMP treatment was explored. Our results revealed that TMP prevented glial degeneration, inhibited retinal neuronal apoptosis, and promoted the establishment of the intraretinal vasculature while suppressing the neovascular response partially by downregulating HIF-1α and VEGF mRNA expression. 

Materials and Methods

Mouse Model of OIR

C57BL/6J mice from the Animal Laboratory of Zhongshan Ophthalmic Center (Guangzhou, China) were used. All procedures with animals in this study were performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Institutional Animal Care and Use Committee of Zhongshan Ophthamic Center. The OIR model was induced as a previously described method of Smith et al. ⁴¹ Briefly, neonatal C57BL/6J mice were exposed to 75% oxygen for 5 days from postnatal day 7 (P7) to P12. At P12, the mice were abruptly returned to room air (21% oxygen). 

TMP Preparation and Administration

TMP (Sigma, St. Louis, MO) solution was freshly prepared as 25 mM in vehicle (dimethylsulfoxide [DMSO]: normal saline = 1:1000, v/v). OIR mice were divided randomly into two groups, treated with 200 mg/kg TMP solution (TMP group) or an equivalent volume of vehicle (OIR group), respectively, by intraperitoneal injection once a day from P12 to P17. Age-matched mice kept in room air served as room-air controls (RA group). Both the OIR group and the RA group were treated with vehicle daily from P12 (see Fig. 1 for the experimental procedure). 

Figure 1.

 

A scheme of the OIR model. Neonatal mice were exposed to 75% oxygen for 5 days from postnatal day 7 (P7) to P12. At P12, the mice were returned to room air (RA; 21% oxygen), and were injected with either TMP (200 mg/kg) or an equal volume of DMSO/normal saline (for OIR controls) daily from P12 to P17 (A). Mice maintained in RA from P1 to P17 with DMSO/normal saline injection were RA controls (B). The eyes from all three groups (OIR, TMP, and RA) were harvested at P14, P15, or P17 for histologic analysis.

Immunostaining on Whole Mount Retinas

Mice were euthanized at P15 or P17 and eyes were enucleated and fixed with freshly prepared 4% paraformaldehyde for 1 hour. The corneas were removed with scissors along the limbus, then the intact retinas were dissected. Retinas were blocked and permeabilized in phosphate-buffered saline (PBS) containing 5% BSA and 0.5% Triton-X-100 overnight at 4°C. Then retinas were incubated with red light–absorbing dye labeled (Alexa Fluor 568; Invitrogen, Carlsbad, CA)Griffonia simplicifolia isolectin B4 (1:50, a marker for vessels and monocytes/microglia/macrophages; Invitrogen) and green light–absorbing dye (Alexa Fluor 488; Invitrogen) conjugated glial fibrillary acidic protein (GFAP) mouse mAb (GA5; 1:50, Cell Signaling Technology, Beverly, MA) overnight at 4°C with gentle rocking. Retinas being stained with isolectin and GFAP were washed with PBS and mounted on microscope slides in mounting medium (Aqua-Polymount; Polysciences, Inc., Warrington, PA). Retinas were examined by fluorescence microscopy (AxioCam MRC; Carl Zeiss, Thornwood, NY) and by confocal microscopy (Zeiss 510; Carl Zeiss). Areas of vaso-obliteration and vitreoretinal neovascular tufts were quantified by using commercial software (Image-Pro Plus; Media Cybernetics, Silver Spring, MD). 

Paraffin Embedded Tissue Processing

Mice were euthanized at P14, P15, or P17 by an overdose of sodium pentobarbital. Eyeballs were immediately enucleated and fixed in an admixture fixative solution consisting of formalin, dehydrated alcohol, and glacial acetic acid overnight at room temperature. Eyeballs were processed in an automated machine (Leica TP1020) and then embedded in paraffin blocks for sectioning. Cross-sections were cut sagittally using a sliding microtome (Leica SM2010R, Nussloch, Germany) at 3 μm and mounted an 3-aminopropyl-triethoxysilane–coated slides. 

Histologic Evaluation

Random paraffin sections crossing the optic nerve from each group were collected on the same glass slides. The selected sections were deparaffinized and stained with hematoxylin and eosin (HE). Two pictures from the central retina of each section were captured. The morphology changes and layer thickness (about 450–750 μm apart from the optic nerve head) in the retina were examined and analyzed (Image-Pro Plus software; Media Cybernetics). 

Detection of Apoptotic Cells by TUNEL

Apoptosis was examined by TUNEL assay (Roche, Mannheim, Germany) as previously described. ⁴² Sections were thoroughly washed with PBST (0.05% Tween 20) and incubated with 4′,6-diamidino-2-phenylindole (DAPI, 1:1000; Sigma-Aldrich Corp., St. Louis, MO) at room temperature for 5 minutes. Slides were mounted with aqueous medium (Aqua-Polymount; Polysciences, Inc.) and visualized with a confocal fluorescence microscope. For each section, total fluorescence was calculated from two separate high-power fields on either side of the optic nerve and the average values from three separate sections/eye crossing the optic nerve were combined to produce a mean value. The interval between each section was 15 μm apart. The field size of each picture was approximately 0.05 mm². TUNEL-positive nuclei staining green were counted. The intensity of the fluorescence detected in the area of the central retina (approximately 450–750 μm apart from the optic nerve head) was analyzed using commercial software (Image-Pro Plus; Media Cybernetics). 

Immunostaining on Retinal Sections

Heat antigen retrieval was performed using citrate buffer (pH 6.0) or Tris-EDTA buffer (pH 9.0) for 40 minutes at 95°C. The sections were then cooled at room temperature for 2 hours and blocked in PBS with 5.0% normal goat serum and 5.0% BSA for 1 hour. After they were blocked, the sections were incubated overnight at 4°C with primary antibodies. All primary antibodies used in this work (summarized in Table 1) have been used in several previous studies and are well characterized by us and other authors regarding the specific cell type molecular marker. ^43–48 After washing in PBST, sections were incubated for 1 hour with secondary antibody–conjugated (Alexa Fluor 488, green) anti-rabbit IgG (Cell Signaling Technology) at 1:500 dilution and were counterstained with DAPI for 5 minutes. 

Table 1.

 

Antibodies Used for Immunostaining of Retinal Cells

 

Semiquantification of Astrocyte Rescue in the Retinal Vascular Obliterated Areas

To assess the astrocyte, mouse retina whole mounts stained by GFAP were examined by fluorescence microscopy at P15. Retinas were dissected, flat-mounted, and stained by GFAP to visualize the astrocytes. Masked observers scored the extent of astrocyte persistence in the vascular obliterated areas on a 1–6 scale. A score of 1 or 2 indicated retinas with a large number of astrocytes in the vascular obliterated areas that were observed to form a completely or fairly normal astrocytic template. A score of 3 to 4 indicated retinas with substantial numbers of astrocytes remaining in the vascular obliterated areas (although still fewer than normal); however, these astrocytes lacked the normal cellular processes and the standard network observed in a normal astrocytic template was not present. A score of 5 to 6 indicated very few astrocytes remaining in the vascular obliterated areas (see Fig. 4A in the following text for examples of the scoring system). Twelve retinas per group were graded from animals at P15 after OIR exposure. 

RNA Isolation and RT-PCR

Mice were euthanized at P12 + 6h (6 hours after injection), P13, and P14, and the retinas were dissected from 18 OIR control mice and 18 TMP-treated mice for RNA isolation (n = 6 for each time point). Retinas were collected and pooled together from at least two mice (four eyes) in the same group and were immediately frozen in liquid nitrogen. To reveal the mRNA expression of HIF-1α and VEGF, RT-PCR was performed to amplify (for 30 cycles) the HIF-1α and VEGF cDNA. PCR products were separated by 2% agarose (Biowest, Seville, Spain) and visualized by double-stranded DNA binding fluorescent stain (GoldView dye; SBSgene, Shanghai, China), according to a previously described method in our laboratory. ⁴⁹ Mouse β-actin served as an internal standard of mRNA expression. Primers for HIF-1α, VEGF, and β-actin were used: mouse- HIF-1α (460 bp), sense, 5′-ACTTGATGTTCATCGTCCTC-3′; antisense, 5′-CGGCGAAGCAAAGAGT-3′; mouse- VEGF (190 bp), sense, 5′-AGCTACTGCCGTCCGATTGA-3′; antisense, 5′-GGTGAGGTTTGATCCGCATGA-3′; mouse-β-actin (232 bp), sense, 5′-GATTGTGATGGACTCCGGAGACGG-3′; antisense, 5′-CATCTCCTGCTGAAGTCTAGAGC-3′. 

Statistical Analysis

All data are presented as mean ± SD from three to five individual OIR experiments unless otherwise indicated. Statistical analysis was performed by independent Student's t-test, or one-way ANOVA followed by Bonferroni post hoc test (P < 0.05; P < 0.01) with commercial analytical software (SPSS 13.0 program; SPSS, Chicago, IL), for multiple comparisons. 

Results

TMP Reduces Pathologic Vitreous NV and Accelerates Physiologic Retinal Revascularization

Systemic injection of TMP in OIR (Fig. 1A, red arrows) from P12 to P17 resulted in a dramatic inhibition of OIR pathologic changes. Our data (Fig. 2A) showed that TMP prevented vitreoretinal NV and accelerated revascularization of the central retina compared with OIR controls at P17, at which time retinal NV was reduced by 55% (Fig. 2B) and the avascular area was reduced by 19% (Fig. 2C). Thus, TMP contributed to rebuild a relatively normal vascular morphology in ischemic retina. 

Figure 2.

 

TMP reduces NV tufts formation and accelerates retina revascularization in OIR. (A) Mouse retinas from OIR and TMP groups were harvested at P17 and subjected to whole mount immunostaining with isolectin B4. Representative images from OIR and TMP groups are shown. Scale bar: 500 μm. (B, C) NV areas (green, B) and avascular areas (yellow, C) were quantified (n = 12 retinas from 12 mice). *P < 0.05, **P < 0.01 comparing TMP vs. OIR groups.

TMP Increases the Formation of Endothelial Tip Cells

To investigate potential cellular mechanisms underlying TMP's protective effects, we analyzed the formation of endothelial tip cells at P17. Endothelial tip cells are vital for the development of new capillaries. They are characterized by using the specialized apical filopodia to attach to the astrocytes and to migrate. ²⁷ In OIR, these cells were noticeably sparse at the junction between the vascularized and avascular areas of the retina at P17 (Fig. 3A). In areas lacking endothelial tip cells, the capillaries were poorly developed, and pathologic extension of vessels into the vitreous occurred. On the other hand, in mice treated with TMP since P12, the density and distribution of endothelial tips cells was increased 2.85-fold compared with the OIR control group (Figs. 3A and 3C) and filopodia were extended from the endothelial cells planar to the protected astrocytes within the vaso-obliterated retina (Fig. 3B), suggesting that cellular factors critical to a normal angiogenic response were rescued. 

Figure 3.

 

TMP increases tip cells in OIR mice retinas. (A) Mouse retinas from OIR and TMP groups were harvested at P17 and subjected to whole mount immunostaining with isolectin B4 and anti-GFAP. Representative images of retinal tip cells (yellow asterisks) and filopodia (yellow arrowheads) are shown. Original magnification, ×40. (B) A high power image from TMP retina is shown to indicate the interactions between tips cells and GFAP-positive astrocytes. Original magnification, ×100. (C) Quantification of numbers of tip cells (n = 4 retinas from 4 mice). **P < 0.01 compared with OIR.

Figure 4.

 

TMP preserves the astrocytic morphology and template in OIR mice retinas. Mouse retinas from OIR and TMP groups were harvested at P15 and subjected to whole mount immunostaining with anti-GFAP. (A) Confocal images of GFAP staining were graded between 1 and 6. Original magnification, ×40. (B) The relative frequency of each grade was used for statistical analyses of OIR and TMP retinas. n = 12 retinas from 12 mice. *P < 0.05, and **P < 0.01. (C) Images of the interface between the vascularized and vaso-obliterated zones showing differences in astrocyte persistence (arrowheads) and Müller glia activation (arrow) at P15 between OIR and TMP groups. Original magnification, ×10. (D) 3-Dimensional renderings demonstrate GFAP staining of the characteristic transretinal processes of activated Müller glia and show association with the punctuate staining observed within the superficial plexus of OIR control retina at P15. All four images are of the same region, but rotated around the y-axis by 0, 30, 60, and 90° from left to right. Size bar = 50 μm.

TMP Preserves the Astrocytic Template in Central Vaso-Obliterated Retina

OIR mice retinal astrocytes survived and were observed in the vaso-obliterated areas at P12 (see Supplementary Fig. S1 ), retaining their normal stellate/dendritic morphology and distribution. However, astrocytes in the vaso-obliterated zone quickly began to degenerate and were mostly absent at P15 following return to normoxia. OIR retinas analyzed at P15 showed a large increase in quantifiable damage to astrocytes, and most of the retinas demonstrated a score of 3 to 6 (Figs. 4A and 4B). The loss of astrocytes was accompanied by an increase in Müller cell GFAP reactivity, which was observed as spotted staining from the Müller cell endfeet within the superficial vascular plexus. These activated Müller glia were observed as GFAP-positive cells with processes spanning the entire retina (Fig. 4D). In the peripheral retina, where capillaries remained intact, GFAP-positive astrocytes were preserved (Fig. 4C). However, in the central retina of TMP-treated mice, astrocytes formed a better network compared with that of OIR; furthermore, astrocytes retained their normal stellate/dendritic morphology and variations in density seen in OIR (Figs. 4B and 4C). Thus, the vascular rescue was associated with protection of the endogenous astrocytes within the vascular obliterated zones at P15. 

TMP Maintains the Relatively Normal Morphology in Central Vaso-Obliterated Retina

In addition to the protective effects of TMP on retinal vasculature and astrocytes, we further investigated its potential role in hypoxia-mediated local neuronal damage of avascular areas using HE staining. We found that there was an abundance of chromatin condensation, pyknotic nuclei (Fig. 5B, yellow arrowheads), and vacuoles (Fig. 5B, blue arrows) in the central inner nuclear layer (INL) of the avascular area at P14 in OIR control mice compared with the RA control retina. These changes were significantly reduced in the equivalent central INL of TMP-treated mice. At P17 these pathologic changes became less obvious in both the OIR and TMP-treated mice (Fig. 5B). However, there was a severe reduction in retinal layer thickness (approximately 450 to 750 μm apart from the optic nerve) of the central INL, inner plexiform layer (IPL), and outer plexiform layer (OPL) of the avascular area at P17 in the OIR control group, which was much less pronounced in TMP-treated mice (Figs. 5A and 5C). The ganglion cell layer (GCL) and outer nuclear layer (ONL) seemed to be relatively unchanged. Thus, TMP treatment contributed to the recovery of a relatively normal inner retinal morphology. 

Figure 5.

 

TMP prevents oxygen-induced morphologic changes and decreases TUNEL-positive cells in the central vaso-obliterated retinas. (A, B) Mouse retinal vertical paraffin sections from RA, OIR, and TMP groups at P14 and P17 were stained with hematoxylin and eosin. Representative images of layers thickness (A), chromatin condensation, pycnic nuclei (yellow arrowheads, B), and vacuoles (blue arrows, B) are shown. Original magnifications, ×40 and ×100. (C) The thickness of IPL, INL, and OPL in P17 retina was used for statistical analysis. n = 9 retinas from 9 mice. *P < 0.01 compared with RA, #P < 0.01 compared with OIR. (D) Mouse retinal vertical paraffin sections from RA, OIR, and TMP groups at P14 and P17 were stained by TUNEL (green) and DAPI (blue). Representative images are shown. Original magnification, ×40. (E) Quantification of the TUNEL-positive cells between three groups at P14 and P17 is shown. n = 9 retinas from 9 mice. *P < 0.01 compared with RA, #P < 0.01 compared with OIR. ON, optic nerve; GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer.

TMP Decreases TUNEL-Positive Cells in Central Vaso-Obliterated Retina

Next we used TUNEL assay to detect the signs of chromatin condensation and pyknotic nuclei, which indicated the apoptotic nature of the DNA strand breaks. ¹⁶ TUNEL-positive cells were observed mostly at P14 (Fig. 5D) and mainly in the central INL of the avascular areas in the OIR control group compared with the RA control retinas (30.04 ± 6.39 vs.1.02 ± 0.69 cells/100 μm; P < 0.001; Fig. 5E). However, after TMP administration, TUNEL-positive cells were diminished significantly to 5.33 ± 2.04 cells/100 μm at P14 in equivalent avascular areas. No significant difference was observed in retinal apoptotic cells among these three groups at P17 (Fig. 5E). These results were correlated with that detected by HE staining. The decrease in TUNEL-positive nuclei observed after TMP treatment corroborated the potential neuroprotective effects. 

TMP Prevents Loss of Retinal Neurons in Central Vaso-Obliterated Retina from Ischemia-Induced Cell Death

The INL consists predominantly of tightly packed neural cells including amacrine, bipolar, horizontal, and Müller cells. ^50,51 We next sought to identify the cell types of apoptotic neurons in the INL. 

Calretinin immunoreactivity was found in cell bodies of amacrine cells in the inner part of the INL and ganglion cells in the GCL at P17 in RA retinas. The synaptic terminals of these cells formed three distinct bands in IPL. In OIR central retinas, when compared with RA retinas, the extent of calretinin immunoreactivity showed a substantial loss of cell bodies in the INL of the avascular area; in addition, the innermost stratum in IPL appeared to be disorganized. However, TMP treatment obviously reduced the damage (Fig. 6A). 

Figure 6.

 

TMP prevents neuronal cells loss in the central vaso-obliterated retina of OIR. Mouse retinal vertical paraffin sections at P17 from RA, OIR, and TMP groups were immunostained (green) with specific cell type molecular marker. Nuclei were counterstained with DAPI (blue). Representative fluorescence microscopy images of retinas are shown. (A) Calretinin-stained retinas showed a more profuse loss of amacrine cell bodies (red arrowheads), disorganized distinct bands (red arrows), and a slightly more loss of calretinin-positive ganglion cell bodies (yellow arrows) in OIR control mice than that observed in TMP-treated or RA mice. (B) Rod bipolar-specific staining for PKC-α showed fewer cells (arrowheads) and retraction dendrites (arrows) in OIR control mice than in TMP-treated or RA mice. (C) Horizontal-specific staining for calbindin showed less horizontal cell bodies (arrowheads) and terminals (arrows) in OIR control mice than in TMP-treated or RA mice. (D) GS-stained retinas showed a more obvious loss of alignment of Müller cell bodies (arrowheads) and distortion of processes (arrows) in OIR control mice than in TMP-treated or RA mice. (E) TMP reduced GFAP expression in Müller cells. Arrowheads: Müller cell processes expressing GFAP. n = 5 retinas from 5 mice. Original magnifications, ×40 and ×100.

PKC-α is a marker for rod bipolar cells. In RA retinas, the cell bodies of PKC-α–positive rod bipolar cells were located in the outer border of INL. Dendritic terminals of rod bipolar cell established connections with rod spherules arbor in the OPL, and their end-bulb axon terminals extended into the inner border of IPL. ^47,52,53 In OIR central retinas at P17, The number of immunopositive cell bodies appeared to decrease and their dendrites retracted, which could explain the thinning of IPL, INL, and OPL (Fig. 5). By contrast, TMP treatment preserved the rod bipolar cell bodies and their dendrites (Fig. 6B). 

Calbindin is distributed in horizontal, amacrine, and ganglion cells in the mouse retina, but it is strongly expressed in horizontal cells and thus has been used as a horizontal cell marker. ⁵⁴ In vertical sections of RA retinas, strong calbindin immunoreactivity was found in horizontal cell somata in the outermost INL and their dendrites in the OPL and relatively faint immunoreactivity was observed in the inner part of the INL, IPL, and GCL. The horizontal cell number and dendritic tips that connected with both rod and cone photoreceptors became faint and broken down in OIR control retinas. In contrast, in TMP-treated mice retinas, more horizontal cells and their terminals could be observed instead (Fig. 6C). 

Glutamine synthetase (GS) is a marker for retinal Müller cells. Normally, the processes and bodies of Müller glia were positive for GS. In OIR central retinas, loss of alignment of the Müller cell bodies and distortion of processes in INL could be observed. However, TMP inhibited the pathologic change of Müller cells (Fig. 6D). 

TMP Reduces GFAP Expression in Müller Cells

Müller cells are activated and involved in the development of pathologic angiogenesis. GFAP is expressed in Müller cells when they are damaged. In mice exposed exclusively to room air, GFAP expression was confined to astrocytes in GCL both centrally and peripherally, but not in Müller glia (Fig. 6E). In OIR control mice, the loss of astrocytes was accompanied by an increase of GFAP reactive Müller cells, which was observed as spotted staining (Figs. 4A, 4C, and 4D) from the Müller cell endfeet in the avascular area. These activated Müller glia cell processes demonstrated strong GFAP expression in both the central and peripheral retinas. However, TMP reduced the reactive expression of GFAP in the Müller glia (Fig. 6E). 

TMP Decreases HIF-1α and VEGF mRNA Expression in OIR Mice Retinas

To identify the molecular mechanism involved in the TMP's neuroprotection and antiangiogenic properties, we examined the mRNA expression of HIF-1α and VEGF in the retinas of OIR mice by RT-PCR. Our results showed that the levels of HIF-1α mRNA expression at P12 + 6h, P13, and P14 in the TMP group were reduced by 27.5%, 26.9%, and 43.4% (Fig. 7A), respectively, compared with the OIR group. Similarly, VEGF mRNA was reduced by 27.9%, 13.8%, and 27.2% at P12 + 6h, P13, and P14 of the TMP group, respectively. 

Figure 7.

 

TMP reduces HIF-1α and VEGF mRNA expression in OIR mice retinas. Mouse retinas from OIR and TMP (200 mg/kg) groups were harvested at P12 + 6h (6 hours after injection), P13, P14, and detected by standard RT-PCR. Representative pictures for mRNA expression of HIF-1α and VEGF are shown. n = 12 retinas from 6 mice. *P < 0.05, **P < 0.01 comparing TMP vs. OIR groups.

Discussion

Strong associations exist between retinal neurons, retinal blood vessels, and vascular associated glial cells such as astrocytes and Müller cells. ⁵⁵ Understanding the interplay among neuronal cells, angiogenesis, and the glial response in OIR is critical in determining the mechanisms underlying diseases such as retinopathy of prematurity. Pathologic vitreoretinal NV, glial cell activation, and neuronal death are major concerns in ischemic retinopathy. ^12,56 Neuroprotective agents, alone or in combination with antiangiogenic agents that can retard or prevent these damages, are beneficial in treating many ocular diseases. In our present study, we showed that TMP prevented astrocyte degeneration and reduced Müller cell reactivity, which correlated with its antiangiogenic effect of inhibiting pathologic vitreoretinal NV as well as accelerating establishment of the intraretinal vasculature. Moreover, TMP inhibited retinal INL neuronal apoptosis, specifically by rescuing amacrine, rod bipolar, and horizontal neurons, and maintained retinal thickness during the ischemic hypoxia phase of ischemic retinopathy in an OIR mouse model. 

We did a dose response (100, 200, and 400 mg/kg) for TMP in vivo and found TMP injections, in a dose-dependent manner, significantly suppressed NV by 24.8%, 55%, and 62.3%, respectively, compared with OIR controls at P17 (see Supplementary Fig. S2 ), whereas TMP injections, at a dose of 400 mg/kg or more, began to induce the deaths of some mice likely partially due to its slight toxicity. So we picked 200 mg/kg as the dose of TMP to inject into the mice. It has been reported that TMP can penetrate through the blood–brain and blood–eye barriers. ³² We proved that TMP could pass the blood retinal barrier and enter into retinas of OIR mice through intraperitoneal injection of 200 mg/kg TMP detected by reversed phase high performance liquid chromatography in our previous research (data not shown). Yang JN et al. ⁴⁰ have claimed that systemic injection of TMP can suppress photoreceptor cell apoptosis and decrease its loss in the peripheral retina induced by N-methyl-N-nitrosourea in rat retinal damage. TMP is also reported to significantly protect hippocampal neurons from degeneration in rats following kainate-induced prolonged seizures. ^28,57 Moreover, TMP has also been shown to reduce neurologic deficit scores and infarction volume on transient focal cerebral ischemia/reperfusion injury in rats. ⁵⁸ Thus far, research on the neovascular effects of TMP is still ongoing. However, to our knowledge, no investigation of TMP in an animal study of ischemic retinopathy has been reported. The major goal of this study was to explore the protective effects of TMP against ischemic retinopathy in the OIR model. TMP exerts antiangiogenic properties and neuroprotective effects by acting as an antioxidant and a calcium antagonist. ³² Zhang et al. ³⁶ have demonstrated that TMP decreases the hypoxia-induced rat pulmonary microvascular endothelial cells monolayer permeability and attenuated the elevation of reactive oxygen species, HIF-1α, and VEGF protein levels. Li et al. ⁵⁹ have found that TMP protects human umbilical vein endothelial cells from LPS-induced IL-8 production due to its potential antiinflammatory effect. Our results also suggested that TMP prevented the development of NV partially by downregulating proangiogenic factors, such as HIF-1α and VEGF mRNA levels in OIR mice. Moreover, TMP also improved normal vasculature repair by enhancing the survival of astrocytes during hypoxia. 

During retinal vascular development, retinal astrocytes provide the template over which endothelial cells migrate to form the retinal vascular network. ^24,27 Endothelial cells' migration along the underlying astrocytic template is thought to be mediated by VEGF and the cell adhesion molecule R-cadherin, which is expressed on retinal astrocytes. ^{23,25,60–62} Our immunoreactivity analysis of GFAP expression indicated that in OIR control mice retinal astrocytes disappeared from the avascular, hypoxic areas of the retina and this was associated with slow recovery and formation of large pathologic, preretinal vascular tufts. In contrast, astrocytic density remained high within the avascular areas of the TMP-treated mouse retinas and the survival of astrocytes strongly contributed to reduce pathologic intravitreal NV and accelerate physiologic revascularization. Müller cells do not express GFAP under normal physiologic situations. Activation of Müller glia and subsequent GFAP reactivity are commonly associated with retinal stress and vascular disease. ^9,12 We have demonstrated that TMP diminished the activation of Müller glia and improved the status of the astrocyte network in the avascular areas, facilitating the formation of a template over which filopodia-mediated endothelial cells migrate to promote vascular recovery. 

Filopodia is devoted to intercellular communication, cell migration, and cell adhesion in endothelial tip cells which lead to the outgrowth of blood vessels. ^61,63,64 Our data indicated that endothelial filopodia extended along the astrocytic template during revascularization, suggesting that the vessels are indeed responding to underlying guidance cues generated by the astrocytes. The emergence of increased numbers and normal morphology of endothelial tip cells to control vessel sprouting and branching may be a possible cellular mechanism that TMP improved vascular recovery. The presence of appropriately directed endothelial tip cells in TMP-treated mice retinas indicates active, directional tip cells migration, implying physiologic rather than pathologic regulation of angiogenesis. ^27,64  

Interactions among astrocytes, endothelial cells, and neurons regulate retinal vascular development. ^64,65 For example, neurons secrete PDGF-A to stimulate proliferation of astrocytes, ⁶⁶ which in turn promote vascular growth by secreting VEGF. ^23,64 Thus, it is interesting to focus on the role of TMP in retinal neurons besides astroglia and retinal blood vessels. In the OIR model, apoptosis and subsequent thinning in the inner retina have been well documented. ^16,67 Ischemic retinopathy induces neuronal death in the inner retina, especially INL at P14 during the ischemic phase. ^16,18,68 Consistent with this, we found that abundant apoptotic nuclei in the central inner retinal layers, especially in the INL of the avascular ischemic areas detected by histologic evaluation or TUNEL assay. Sennlaub et al. ¹⁶ have reported the thinning of the IPL and INL in the central avascular area at P17 in a mouse model of OIR without the founding of OPL thinning. However, apart from the thinning of IPL and INL, our data showed a significant thinning or even absence of OPL in some particular central avascular area where may suffer from severe hypoxia and ischemia damage. It is known that IPL is established by the synapses between bipolar, amacrine, ganglion cells, and Müller cells, whereas OPL is established by the synapses between bipolar, horizontal, photoreceptor, and Müller cells. ^51,69 The thickness of IPL and OPL became thinner, indicating that the synapses of neurons were destroyed accompanied by their apoptosis. Lachapelle et al. ⁷⁰ have also described a significant thinning of the OPL in a rat model of OIR, which oxygen exposures consist of high oxygen levels interrupted by multiple daily low oxygen level periods. This discrepancy is probably due to the different genotype of animals and different models when compared with that in our study. With regard to different mice, Sennlaub and colleagues used C57BL/6 129SvEv mice, whereas we used C57BL/6J, suggesting that a different phenotype of animals may cause different responses in OIR. 

Judging from the extent of cell death (sometimes as much as two to four rows of cells in the INL), it seems likely that all prevalent cell types are undergoing apoptosis observed. Therefore, we sought to identify the cell types of apoptotic neurons in retina. We discovered that retinal ischemic and hypoxia caused detrimental injury to amacrine cells, as indicated by the great reduction in calretinin expression and the disorganization of IPL stratification. It was in the inner INL where abundant loss of calretinin expression was observed and coincided with the localization of TUNEL-positive cells, implying that most of the apoptotic neurons occurred during ischemic phase were amacrine cells. Rod bipolar and horizontal cell immunoreactivity showed similar but less prominent changes. The decreased length of the bipolar cells that occurred in OIR resulted in the thinning of INL, IPL, and OPL. Müller cells had elevated glutamine levels and were most gliotic. Loss of alignment of the Müller cell bodies and distortion of their processes in INL were observed. However, significantly fewer apoptotic cells were seen in INL of TMP-treated ischemic retinas. The detrimental effects on amacrine, rod bipolar, horizontal, and Müller cells due to ischemic injury were also minimized. It has been proposed the cytoprotective effects of TMP on neurons are exerted through quenching of reactive oxygen species (ROS), ⁷¹ the inhibition of HIF-1α activation ⁷² and proapoptotic pathways, ^33,40,72,73 and the suppression of inflammatory reaction. ⁷³ Yang et al. ³⁹ found that TMP protects both microtubule-associated protein-2–labeled neuronal and nonneuronal cells from hydrogen peroxide attacks by striking inhibition of both lipid peroxidation and mitochondrial ROS production in mixed rat retinal cell cultures, of which the majority of them are identified as GFAP-positive astrocytes. Neuronal cells are especially sensitive to oxidative damage during the ischemic phase of the retinopathy. ^8,56 In our experiments, TMP prevented neurons against apoptosis likely due to its antioxidant potency and the inhibition of HIF-1α activation. Thus, our data elucidated that beneficial protection of TMP is comparable to that of other neuroprotective agents such as curcumin, lutein, and lycium barbarum. ^52,74,75  

In the final analysis, our results demonstrated that TMP improved neurovascular recovery by rescuing neurons and astroglia cells of the ischemic retina partially due to its downregulation of HIF-1α and VEGF mRNA expression, contributing to restorative angiogenesis. Further investigation is needed to define the molecular and cellular mechanisms of TMP's cytoprotective effects involved in tip cell formation, vascular repair, glial rescue, and neuronal protection in ischemic proliferative retinopathies. 

Supplementary Materials

Doc S1 

Doc S2 

Footnotes

 Supported in part by National Natural Science Foundation of China Grant 30973899 (XiL) and Fundamental Research Funds of State Key Laboratory of Ophthalmology Grant CX-10 (XiL).

Footnotes

⁵  These authors contributed equally to the work presented here and should therefore be regarded as equivalent authors.

Footnotes

 Disclosure: X. Liang, None; H. Zhou, None; Y. Ding, None; J. Li, None; C. Yang, None; Y. Luo, None; S. Li, None; G. Sun, None; X. Liao, None; W. Min, None