Introduction

Neovascular glaucoma (NVG) secondary to proliferative diabetic retinopathy (PDR) is a potentially devastating condition, often refractory to medical treatment [1]. Retinal ischemia due to PDR can lead to neovascularization of the iris and anterior chamber angle and subsequent formation of fibrovascular membranes (FVMs). The condition may eventually obstruct the trabecular meshwork and cause angle closure, resulting in an intractable elevation of the intraocular pressure (IOP) [2].

The FVMs consist of stromal tissues and cellular components, such as proliferated myofibroblasts and vascular endothelial cells that can contribute to the development of fibrovascular proliferation [3, 4]. As the aqueous humor provides nutrients to tissues in the anterior segments and drains through the outflow pathway, which includes the trabecular meshwork (TM) and Schlemm’s canal (SC), several molecules in the aqueous humor are considered to play roles in the pathophysiology of NVG. Previous studies have indicated possible involvement of growth factors and cytokines, such as vascular endothelium growth factor (VEGF), transforming growth factor-beta 1 (TGF-β1) and transforming growth factor-beta 2 (TGF-β2), hepatocyte growth factor (HGF), basic fibroblast growth factor (bFGF), and pigment epithelium-derived factor (PEDF) [5,6,7,8,9], in the fibrovascular proliferation associated with NVG, by showing that their levels were elevated in the aqueous humor. However, the molecular mechanisms regulating FVM formation have not been fully determined.

We previously performed genome-wide gene expression profiling of human pre-retinal FVMs excised from patients with PDR who underwent vitrectomy surgery. This study found significant upregulation of matricellular proteins, such as periostin (PN) and tenascin-C (TNC) [10]. Matricellular proteins are characterized by their potential to bind to both extracellular matrix and cell surface receptors, which can enhance tissue remodeling in conditions such as wound healing. We subsequently demonstrated that PN and TNC were significantly upregulated in the vitreous humor of patients with PDR and proliferative vitreoretinopathy, and facilitated both retinal angiogenesis and fibrosis in the vitreoretinal diseases [11,12,13,14]. The involvement of PN and TNC in NVG pathology has not been studied.

Here, we investigate PN and TNC expression in human specimens obtained from patients with NVG secondary to PDR to explore their possible involvement in NVG pathogenesis.

Methods

Patients and sample collection

This cross-sectional study was approved by the Ethics Committee of the Kyushu University Hospital, and the surgical specimens were handled in accordance with the ethical standards of the 1989 Declaration of Helsinki. Participants with NVG, PDR, primary open-angle glaucoma (POAG), or cataract who visited Kyushu University Hospital between May 2017 and November 2018 were enrolled. We aimed to obtain ten subjects per study group. Inclusion criteria were tractional retinal detachment and vitreous hemorrhage with active retinal neovascularization associated with PDR. NVG was diagnosed by neovascularization of the iris and/or iridocorneal angle, with a history of IOP of more than 21 mmHg. Patients who had undergone intraocular surgery within 6 months prior to sample collection were excluded. Aqueous humor samples were collected from the anterior chamber at the start of surgery for cataract and glaucoma, or at intravitreal injection of anti-VEGF drug for PDR according to our previous report. Briefly, at the start of surgery, each patient was placed in a supine position on the bed. A 50–100 μl volume of aqueous humor was obtained through limbal paracentesis using a syringe attached with a 30-gauge needle (Nipro, Osaka, Japan). For control, aqueous humor samples were collected from patients with primary open-angle glaucoma (POAG) or cataract without diabetes mellitus. During trabeculectomy surgery, we first fashioned a superficial scleral flap. After which, we excised a deeper scleral block to create a sclerostomy channel. The excised block (sclerostomy tissues) included the sclera, trabecular meshwork, and cornea. We obtained such sclerostomy tissues from a POAG patient and two NVG patients who underwent previous retinal photocoagulation but not anti-VEGF therapy. The dissected specimens were immediately fixed in 4% paraformaldehyde at 4 °C for 24 h and subsequently underwent histopathological analysis.

Enzyme-linked immunosorbent assay (ELISA)

PN and TNC concentrations were measured with a sandwich ELISA kit as described previously [14, 15]. VEGF and TGF-β2 concentrations were measured with human VEGF and TGF-β2 immunoassay kits (R&D Systems, Minneapolis, MN, USA), according to the manufacturer’s instructions. Concentrations were determined by measuring absorbance at 450 nm wavelength using a microplate reader (PerkinElmer, Waltham, MA, USA).

Immunohistochemistry and immunofluorescence staining

Fixed sclerostomy tissues were embedded in paraffin and cut at a 3-μm thickness. After removing the paraffin, the sections were rehydrated, blocked, and incubated with primary antibodies (Abs) for 24 h. For immunohistochemistry, the bound Ab was visualized using an aminoethylcarbazole kit (Life Technologies, Carlsbad, CA, USA) according to the manufacturer’s protocol. Nuclei were counterstained using a hematoxylin and eosin (H&E) staining procedure. For immunofluorescence analyses, the sections were incubated with secondary Abs for 1 h at room temperature. Nuclei were counterstained with Hoechst 33342 (Molecular Probes, Eugene, OR, USA). The specimens were flat-mounted in Crystal/Mount (Biomeda Corp., Foster City, CA, USA) and examined by fluorescence microscopy (BZ-9000; Keyence, Osaka, Japan). The primary Abs were PN (MAB3548 1 g/ml; R&D Systems), TNC (NB110-68136 1:50 dilution; Novus Biologicals, Centennial, CO, USA), and FITC-labeled alpha-smooth muscle actin (α-SMA) (F3777 1:250 dilution; Sigma-Aldrich). The secondary Ab was Alexa Fluor 647 (A21247 1:800 dilution; Molecular Probes).

Statistical analysis

Statistical analyses were performed using JMP V13 software package (SAS Institute, Cary, NC, USA). Tukey–Kramer honestly significant difference tests were performed to compare the growth factor levels between the subgroups for multiple variables. Differences were considered significant at values of p < 0.05.

Pearson product-moment correlation coefficients were calculated to determine the correlations between the aqueous humor levels.

Results

Patient demographics

Aqueous humor samples were collected from 37 eyes of 37 patients who were grouped into (1) NVG secondary to PDR (NVG; n = 8); (2) PDR without NVG (PDR; n = 9); (3) primary open-angle glaucoma (POAG; n = 11); and (4) cataract surgery patients as a control group (CG; n = 9). None of the PDR patients underwent advanced anti-vascular endothelial growth factor (VEGF) therapy. Characteristics of the patients and the eyes are summarized in Table 1.

Table 1 Patient characteristics
Full size table

Concentrations of PN, TNC, VEGF, and TGF-β2 in the aqueous humor

We quantified the concentrations of PN and TNC in aqueous humor samples from patients with NVG, PDR, POAG, or CG. VEGF is known to promote neovascularization of the iris and trabecular meshwork resulting in high IOP [7, 8]. TGF-β2 contributes to intraocular fibrosis of PDR [16] and is elevated in the aqueous humor of NVG [6, 17]. Therefore, we measured the concentration of VEGF and total (activated and latent) TGF-β2 in the same aqueous humor samples that were used for PN and TNC measurements.

PN and TNC levels were below detection threshold in the POAG and CG groups. The NVG group had significantly higher levels of PN and TNC compared with the PDR group (84.7 ng/ml vs 2.2 ng/ml (p < 0.05) and 18.5 ng/ml vs 4.6 ng/ml (p < 0.05), respectively). The average levels of VEGF and TGF-β2 were higher in NVG compared with the PDR group with no significant difference (4064 ng/ml vs 209 pg/ml (p = 0.1) and 2548 ng/ml vs 1643 pg/ml (p = 0.1), respectively) (Fig. 1).

Fig. 1
figure 1

Levels of periostin, tenascin-C, VEGF, and total TGF-β2 in the aqueous humor. The concentrations were measured by ELISA. NVG secondary to PDR (NVG), n = 8; PDR without NVG (PDR), n = 9; primary open-angle glaucoma (POAG), n = 11; and cataract surgery patients as a control group (CG), n = 9. Bars show mean levels. *p < 0.05, Tukey–Kramer method

Full size image

There was a significant correlation between PN and TNC levels (r = 0.86, p < 0.05) and VEGF and TGF-β2 levels (r = 0.7314, p < 0.05) in the NVG group. No significant correlations were seen between the levels of PN and VEGF and TGF-β2 and between TNC and VEGF and TGF-β2. In the NVG group, there were significant relationships between peripheral anterior synechiae (PAS) indices and PN and TNC levels (r = 0.84 and 0.82, p < 0.05 and p < 0.05, respectively), but not between IOP and PN and TNC levels.

PN and TNC expression in the trabecular meshwork and Schlemm’s canal of sclerostomy tissues from NVG patients

Immunohistochemical analysis of sclerostomy tissues excised from patients with NVG showed that prominent PN expression could be extensively observed in the Schlemm’s canal (SC) and trabecular meshwork (TM) (Fig. 2a (cases 1 and 2)), whereas relatively weak TNC expression could be seen in the SC and partially in the TM (Fig. 3a (cases 1 and 2)). No apparent signal of PN and TNC was seen in the SC and the TM of the POAG eye (Figs. 2a and 3a (case 3)). We found no significant signal in sections stained with control IgG antibody (data not shown).

Fig. 2
figure 2

Localization of periostin in the trabecular meshwork and Schlemm’s canal in the eyes of patients with NVG secondary to PDR and POAG. Histological section of the trabecular meshwork (TM) and Schlemm’s canal (SC) in the sclerostomy tissues excised from patients with NVG (cases 1 and 2) or POAG (case 3). (a) Periostin immunohistochemical staining. Nuclei are stained blue. Scale bars: 50 μm. (b) Triple-immunofluorescence staining for α-SMA (green) and periostin (red). Nuclei are counterstained with DAPI (blue). Scale bars: 50 μm.

Full size image
Fig. 3
figure 3

Localization of tenascin-C in the trabecular meshwork and Schlemm’s canal in the eyes of patients with NVG secondary to PDR and POAG. Histological section of the trabecular meshwork (TM) and Schlemm’s canal (SC) in the sclerostomy tissues excised from patients with NVG (case 1 and 2) or POAG (case 3). (A) Tenascin-C immunohistochemical staining. Nuclei are stained blue. Scale bars, 50 μm. (b) Triple-immunofluorescence staining for α-SMA (green) and tenascin-C (red). Nuclei are counterstained with DAPI (blue). Scale bars, 50 μm.

Full size image

Next, to determine the cellular expression of PN and TNC in the SC and the TM, we stained the sections with Abs against PN, TNC, and α-SMA. α-SMA is a marker for myofibroblasts that have been reported as the main sources of PN and TNC in intraocular fibrovascular diseases [12,13,14,15, 18]. Immunofluorescence analyses showed that PN was co-expressed with α-SMA-positive cells in the SC and the TM (Fig. 2b). In contrast, TNC was not co-expressed with α-SMA-positive cells (Fig. 3b). These findings indicate that PN but not TNC can be produced by or bound to myofibroblasts in the SC and the TM of eyes with NVG associated with PDR.

Discussion

This study introduces for the first time the expression and localization of PN and TNC proteins in the aqueous humor and in the TM and SC of sclerostomy tissues from NVG patients.

Since PN and TNC are upregulated in the vitreous of PDR patients [12, 14], PN and TNC expression in the aqueous humor of PDR patients could result from diffusional reflux from the vitreous cavity. In contrast, the significant higher levels of PN and TNC in the aqueous humor in NVG compared with PDR without NVG could be caused by increased local production and/or hyperpermeability of the iris neovessels that may cause influx of PN and TNC into the anterior ocular segments in NVG patients. It has been shown that matricellular proteins including PN, TNC, and fibronectin cooperate with each other to form the extracellular matrix (ECM) meshwork architecture during the process of scaffold remodeling in angiogenesis and fibrosis, such as in bronchial asthma and intrahepatic cholangiocarcinoma [19,20,21]. Thus, the significant correlation between PN and TNC levels in NVG aqueous humors could imply their cooperation in ECM formation in eyes with NVG.

Our observation that PN but not TNC signal was localized in α-SMA-positive cells in the TM suggests possible local production of one or both and possible binding of PN at the site. TM cells normally function to maintain IOP by regulating aqueous outflow resistance, and under stress conditions, TM cells display different behaviors typical of endothelia, macrophages, fibroblasts, and smooth muscle that characterizes the clinical features of glaucoma [22]. The fibroblasts and smooth muscle–like phenotypes have the potential to secrete a number of ECM proteins [23]. We previously reported that PN is released from α-SMA-positive cells (myofibroblasts) and acts in an autocrine fashion to promote fibrovascular proliferation in proliferative vitreoretinal disease [13, 14]. Taken together, our findings suggest that α-SMA-positive cells in TM could produce PN that is released into the aqueous humor to promote formation of FVMs in the anterior chamber angle of NVG eyes.

Although TNC level was increased in aqueous humor, cellular expression of TNC was not evident in the TM and SC of NVG eyes. Our previous report demonstrated that TNC was produced and secreted from vascular smooth muscle cells in FVMs excised from PDR patients [12]. Similarly, in the eyes of NVG patients, the source of TNC may be vascular smooth muscle cells in the FVMs in the anterior chamber; however, further study is required to identify TNC-secreting cells.

VEGF and TGF-β are considered to play major roles in the neovascularization and fibrosis in the pathophysiology of NVG, respectively [6,7,8, 17]. PN and TNC act downstream of TGF-β2, and significant correlation has been seen between PN and TGF-β2 levels in the vitreous in PDR and PVR [13, 24, 25]. However, our results showed no correlation in aqueous humors of NVG patients. For VEGF, we previously found a weak correlation between TNC and VEGF in the vitreous in PDR [12]; however, the present study did not find correlations between PN and TNC and VEGF. These results suggest that TGF-β2 and VEGF might not regulate PN and TNC expression in the aqueous humor of NVG. Since a wide variety of cytokines and growth factors are elevated in NVG eyes [9, 26], simultaneous measurements of multiple factors would be needed to seek upstream regulators of PN and TNC in the aqueous humor of NVG eyes.

A limitation of our current study is its small sample size. Although we found statistically higher PN and TNC levels in the NVG group compared with the other groups, the levels were highly variable. The clinical significance of this variation should be considered with a larger sample size. In addition, while the mean VEGF concentration in the aqueous humor of NVG eyes was high and consistent with previous reports, it was not significantly different compared with the other groups [7, 8]. This could possibly be due to the small sample sizes in our study.

As NVG progresses, neovessels can appear over the angle structures and PAS can ultimately occur, and this is accompanied by FVM formation and subsequently leads to complete angle closure. It has been reported that intravitreal anti-VEGF drug injection in stage 1 and 2 NVG patients with open iridocorneal angle could have beneficial effects on IOP control and regression of the iris and angle neovascularization [27]. However, anti-VEGF treatment is not effective for lowering IOP after occurrence of angle closure by PAS, and surgical intervention is usually inevitable. Thus, in the management of PDR, it is important to control ischemic stimuli primarily by panretinal photocoagulation and/or anti-VEGF treatment for preventing subsequent development of NVG. In addition, development of novel therapeutic agents against fibrovascular proliferation is also required to prevent the consequent angle closure. Our group proposed PN and TNC as important players in retinal fibrosis and showed that their blockade could suppress generation of subretinal and pre-retinal fibrous membrane in animal models [13, 15]. The present study showing expressions of PN and TNC in NVG eyes and the relationships between PAS indices and PN and TNC levels may suggest their role in fibrovascular proliferation in NVG pathogenesis. Taken together, PN and TNC might be candidate therapeutic targets to inhibit FVM development in the treatment of NVG.