Tetrahydrobiopterin (BH4) deficiency is associated with augmented inflammation and microvascular degeneration in the retina

hph-1 mice were bred in-house and genotyped to confirm homozygous, dominantly inherited GTPCH1 gene deficiency (data not shown). C57BL/6 mice (Charles River, ON, Canada) were utilized as wild type (WT) controls, since this is the background of the hph-1 mice utilized in this study. All experiments were conducted in accordance with ARVO statement regarding the use of animals in ophthalmic and vision research and approved by the Animal Care Committee of the Hôpital Maisonneuve-Rosemont and The Hospital for Sick Children in accordance with guidelines established by the Canadian Council on Animal Care.

Eyes were enucleated from WT mice and hph-1 mice at P1, P7, P14, and P22 and fixed in 4% paraformaldehyde for 1 hour at room temperature and then stored in PBS at 4 °C until used. The cornea and lens were removed, and the retina was gently separated from the underlying choroid and sclera under a dissecting microscope. Then, the retinas were incubated overnight at 4 °C in 1% Triton X-100-1 mM CaCl₂/phosphate-buffered saline (PBS) with the TRITC-conjugated lectin endothelial cell marker Bandeiraea simplicifolia (1:100; Sigma-Aldrich, St. Louis, MO), the specific microglia marker Iba-1 (1:500; Wako Chemicals USA, Inc], or the mouse monoclonal TSP-1 antibody (1:200; EMD Millipore). Iba-1 or TSP-1 antibodies were labeled for 2 h with Alexa-488-conjugated goat anti-rabbit (1:1000) or Alexa-594-conjugated goat anti-mouse (1:500), respectively, obtained from Molecular Probes (Eugene, OR). Retinas were washed in PBS and mounted on microscope slides (Bio Nuclear Diagnostics Inc., Toronto, ON) under cover slips with Fluoro-Gel® (Electron Microscopy Sciences, Hatfield, PA) as the mounting media. Retinas were photographed under a confocal microscope (Olympus, Richmond Hill, Canada) at 30× or an epifluorescence microscope at 10× using a Zeiss AxioObserver.Z1. Images were merged into a single file using the MosiaX option in the AxioVision 4.6.5 software (Zeiss). Vascular density was calculated for the full retina surface by using the software AngioTool as previously described [31]. AngioTool computes several morphological and spatial parameters including vascular density by assessing the variation in foreground and background pixel mass densities across an image [31].

Eyes were enucleated from WT and hph-1 mice at P22 and then fixed in 4% paraformaldehyde at room temperature for 2 h. Eyes were incubated  overnight at 4 °C in a 30% sucrose solution prior to embedding in OCT compound (TissueTek®). Coronal sections of the eyes were cut at a thickness of 10 μm by using a Cryostat (Leica). Sections were subsequently washed with PBS, blocked and permeabilized for 1 h at room temperature and subsequently incubated overnight with Lectin from Bandeiraea simplicifolia TRITC conjugate (Sigma; 1:100) for retinal vasculature and/or antibodies to rabbit Iba-1 (Wako Chemicals USA, Inc.; 1:400), goat IL-1β (R&D systems; 1:300), or rabbit anti-CD36 antibody (Abcam; 1:200). The primary antibodies were labeled for 2 h with Alexa-488 goat anti-rabbit or donkey anti-goat IgG obtained from Molecular Probes (Eugene, OR) and used at dilutions of 1:1000. Samples were visualized using 30× objective with an IX81 confocal microscope (Olympus), and images were obtained with Fluoview 3.1 software (Olympus).

Retinal tissues (pool of five retinas per sample, n = 3 per group) were collected from WT mice and hph-1 mice at postnatal day 7, 14, and 22. Determination of BH4 in retinal tissues was performed by liquid chromatography tandem mass spectrometry (LC-MS/MS) as previously described [32]. Briefly, analysis was performed on an AB Sciex 5500QTRAP mass spectrometer (Foster City, CA, USA) coupled with a Shimadzu Nexera ultrahigh pressure liquid chromatograph system (Kyoto, Japan). Pterins were separated by a binary gradient using reversed-phase HPLC on an EZfaast 250 × 2 mm 4 μm AAA-MS column, with a 4 × 2 mm SecurityGuard column. A calibration curve was prepared in water containing 0.2% dithiothreitol (DTT) (Sigma Aldrich, Oakville, ON) over the range of 25–1600 nmol/L for BH4. A deproteinization solution containing internal standards for ¹⁵N–BH4 was prepared in 0.1 M perchloric acid (Sigma Aldrich) containing 0.2% DTT at a final concentration of 1000 nmol/L each. Retinal tissue samples were deproteinized and then centrifuged at 14800 rpm at 4 °C for 10 min. Thirty microliter of supernatant was combined with 120 μl of water containing 0.2% DTT. Processed supernatants were transferred to a microtiter plate, and 10 μL was injected for analysis.

Microglia cell line (SIM-A9) was used and cultured as previously reported [33]. Briefly, microglial cells (800, 000 cells per well) were cultured in 6-well plates (Sarstedt, Inc., Newton, NC, USA) with DMEM/F12 (1:1) supplemented with 10% fetal bovine serum (FBS), 5% of horse serum (HS), and 1% penicillin/streptomycin. After 24 h, the cells were starved with DMEM/F12 (1:1) free of FBS and HS for 6 h. Then, microglial cells cultures in presence or absence of 100 μM of (6R)-5,6,7,8-Tetrahydrobiopterin dihydrochloride (BH4; Sigma, Cat. T4425) were exposed to hyperoxia (75% oxygen and 25% nitrogen; Hyp-MGCM) in a modular incubator chamber (Billups-Rothenberg, Inc) and maintained in a humidified CO₂ incubator at 37 °C for 24 h. Microglial cells in matching controls (Nor-MGCM) were incubated at 37 °C in an incubator with 95% air and 5% CO₂ and collected at the same time point. Cell lysates were quickly processed for RNA using RiboZol (Amresco, N580, OH, USA). The conditioned media was stored at −80 and later used in choroidal explant assay.

WT mice pups were sacrificed on P22, and eyes immediately enucleated and kept in ice-cold EBM-2 medium before dissection. Cornea and lens were removed from the anterior of the eye and choroid-scleral complex was separated from the retina and cut into approximately 1 mm × 1 mm. Choroid/sclera fragments were placed in 30 μL of growth factor-reduced BD Matrigel ^TM (BD Biosciences, Cat. 354,230) seeded in 24-well plates. After seeding the choroid, plates were incubated at 37 °C in a cell culture incubator without medium for 15 min for the BD Matrigel^TM to solidify. Five hundred microliters of EBM-2 medium (LONZA, Cat. CC-3156) supplemented with endothelium growth medium (EGM) kit (Lonza, Cat. CC-4147) and 50 units/mL of Penicillin/Streptomycin (GIBCO, Cat. 15,142) was then added to each well and incubated at 37 °C with 95% air and 5% CO₂ for 48 h before any treatment. After this time, the media was removed and choroidal explants were incubated with Nor-MGCM or Hyp-MGCM in presence or absence of a TSP-1 neutralizing antibody (1.2 μg/ml; EMD Millipore) for 48 h. Phase contrast photos of individual explants were taken at days 4 and 6 using a ZEISS AxioOberver microscope, and the microvascular sprouting area was quantified with the computer software ImageJ 1.46r (National Institute of Health).

Mice pups were exposed with their mothers in a 75% oxygen environment from postnatal day 7 to P9 using BioSpherix oxycycler (Model #A84XOV) to induce retinal vaso-obliteration (VO) as previously described [34, 35]. Animals were anesthetized in 3% isoflurane in oxygen and injected intravitreally at P7 with 100 μM of (6R)-5,6,7,8-Tetrahydrobiopterin dihydrochloride (BH4; Sigma, Cat. T4425) or vehicle (sterile PBS 1×) using a Hamilton syringe equipped with 50-gauge glass capillary. At P9, mice pups were sacrificed and retinas were dissected and stained overnight at 4 °C with fluorescein-labeled Griffonia Simplicifolia Lectin 1 (GSL 1), isolectin B4 (FL 1201, Vector Labs; 1:100) with 1 mM CaCl₂ in PBS as described above. Quantification of VO was assessed using the computer software ImageJ as previously described [36].

Results are expressed as mean ± SEM. Two-tailed independent Student t tests was used to analyze data after confirming that the data are normally distributed. Comparisons between groups were made using one-way ANOVA followed by the post hoc Bonferroni’s multiple comparison test. Statistical significance was set at p < 0.05.

To evaluate whether BH4 deficiency is associated with morphological changes in the eye, histological analysis of eyeballs from WT and hph-1 mice was analyzed by H&E staining at different post-natal ages (P1, P7, P14 and P22). Visual examination and histology revealed a decremented reduction in the size of the eyes from hph-1 eye mice (at all the ages analyzed) compared to WT animals (Fig. 1a and Additional file 1: Figure S1). Also, in approximately 50% of the eyes from hph-1 mice at P22, the iris appeared hypertrophic, causing the pupil to be smaller or absent (Fig. 1a, b). Persistence of fetal hyaloid vasculature was also detected in the eyes of hph-1 at P22, but absent in WT mice with the same age (Fig. 1a; arrows).

To assess the levels of BH4 in the retina, from three to five pools of retinas were collected from WT and hph-1 mice at postnatal age 7, 14, and 22 and evaluated by LC-MS/MS. LC-MS/MS analysis confirmed a significant decreased by approximately 90% in the concentration levels of BH4 in retinal tissue from hph-1 mice (0.0009 ± 0.0006; p < 0.0001, 0.01 ± 0.001; p < 0.0001 and 2.45 ± 0.40; p < 0.005) compared to the WT group (0.014 ± 0.001, 0.092 ± 0.01, and 23.13 ± 6.44) at P7, P14, and P22, respectively (Fig. 1c and Additional file 2: Figure S2).

We further investigated whether a deficiency in BH4 synthesis could be associated with microvascular changes in the retina. Retinal whole-mounts from WT and hph-1 mice at different postnatal days (P1, P7, P14, and P22) were analyzed (Fig. 2 and Additional file 3: Figure S3). In both groups of animals, the superficial retinal vascular sprouts were seen normally emerged from a ring-shaped vessel around the optic nerve head at P1 and reached the periphery at P7 (Additional file 3: Figure S3). At P14, the retinas from WT mice showed well-established superficial and deep vascular plexuses (Additional file 3: Figure S3), while the superficial and deep retinal vasculature from hph-1 mice at P22 showed markedly signs of vascular injury particularly localized in the central retina (Fig. 2a, b). A quantitative analysis of the superficial and deep retinal microvasculature density in hph-1 mice showed a significant reduction (p < 0.01 and p < 0.0001, respectively) in the number of vessels with respect to the control mice at P22 (Fig. 2c). However, no significant differences were detected in the density of both capillary plexuses in the retina from hph-1 mice at early ages (P1, P7, and P14; Additional file 4: Figure S4). These results suggest that a constant deficiency in BH4 synthesis is associated with a degeneration of the microvasculature in the retina with the time and are consistent with previous studies demonstrating that BH4 is an important regulator of inflammation and vascular remodeling [37].

Since a deficiency in BH4 can increase ROS production [6], which results in inflammatory disorders followed by endothelial dysfunction and tissue injury [8], we decided to evaluate the number of inflammatory cells in retinal flatmounts from hph-1 and WT stained with Iba-1, a specific marker for microglial cells. The number of retinal microglia in hph-1 mice at P1, P7, and P14 was very similar in comparison to the WT mice at the same ages (Fig. 3a–c). However, the number of microglial cells in retinas from hph-1 animals at P22 was significantly increased (p < 0.01) when compared with the same age control mice (Fig. 3d). Interestingly, these findings suggest a strong correlation between the number of inflammatory cells and retinal microvascular degeneration.

The next step was to determine whether activated inflammatory cells are associated with an overproduction of pro-inflammatory factors that may be involved in endothelial cell injury as we and others previously demonstrated [11, 38, 39]. Therefore, we evaluated the production of some pro-inflammatory cytokines as well as another’s possible factors involved in the vasculature damage in the retinas from hph-1 mice at P7, P14, and P22 (Fig. 4 and Additional file 5: Figure S5). qPCR analysis revealed a significant increase in the mRNA expression of the pro-inflammatory cytokine IL-1β (2.91-fold; p < 0.01; n = 9 and 8.6-fold; p < 0.005; n = 10), and the major inflammasome protein NOD-like receptor family pyrin domain containing-3 (NLRP3; 1.67-fold; p < 0.03; n = 9 and 4.4-fold; p < 0.03; n = 9) in retinas from hph-1 mice at P14 and P22, respectively, compared to the WT (Fig. 4 and Additional file 5: Figure S5). The expression of the pro-inflammatory cytokine IL-6 was particularly augmented at P22 (3.2-fold; p < 0.04; n = 10), but not at P14 (p = 0.097; n = 7). No significant changes in the expression levels of IL-1β, NLRP3 and IL-6 respect to the control were detected in retinal samples at P7 (Additional file 5: Figure S5). Interestingly, Norrin and its receptor Frizzled-4 (FZD4) strongly involved in the control of retinal vascular development and survival [40, 41] were decreased by 0.36-fold (p < 0.01; n = 10) and 0.2-fold (p < 0.0001; n = 12) and 0.55-fold (p < 0.01; n = 10) and 0.51-fold (p < 0.001; n = 6) respectively in retinas from hph-1 mice at P14 and P22 (Fig. 4 and Additional file 5: Figure S5). Norrin showed an increased expression by 2.28-fold (p < 0.013; n = 10) in retinal samples of hph-1 mice at P7, however, no significant changes in the expression of FZD4 were detected at this same age (p = 0.62; n = 10; Additional file 5: Figure S5). Furthermore, thrombospondin-1 (TSP-1), a matricellular glycoprotein secreted in response to injury and stress in the retina with anti-angiogenic properties and involved in vascular injury [15, 20, 21] was significantly augmented by 2.09-fold (p < 0.004; n = 10) and 17.5-fold (p < 0.005; n = 10) in the retinas from hph-1 mice at P14 and P22, respectively when compared to control animals (Fig. 4 and Additional file 5: Figure S5).

The increased number of microglial cells and inflammatory factors is suggestive of an acute inflammatory response in the retinas of hph-1 mice (Figs. 3 and 4). We focus our attention particularly on IL-1β and TSP-1, based in the fact that both molecules have been previously implicated in retinal microvascular degeneration [11, 15, 20]. Double immunolabeling in the retina by using different specific markers showed that both IL-1β and TSP-1 proteins were positively labeled in microglial cells (Iba-1 immunopositives) from hph-1 mice (Fig. 5a and Additional file 6: Figure S6A); while the immunoreactivity of both proteins in microglia cells in the retinas from WT mice was very weak (Fig. 5a and Additional file 6: Figure S6A). Because it has been proposed that IL-1β derived from activated microglia induces microvascular injury indirectly through the release of pro-apoptotic/repulsive factor semaphorin-3A (Sema3A) in the retina [11], we decide to evaluate the expression of Sema3A in retinas from hph-1 mice at P22. Surprisingly, mRNA expression of Sema3A was significantly decreased (0.5-fold; p < 0.0001; n = 10) suggesting that Sema3A is probably not involved in the vascular injury process in the retina from hph-1 mice (Additional file 6: Figure S6B).

Therefore, we focus our attention on TSP-1, a well-known anti-angiogenic factor that can directly act on endothelial cells by inhibiting their proliferation and migration or promoting their apoptosis [42‐44]. Previous studies in our laboratory have already demonstrated that nitrative stress can induce TSP-1 upregulation which cause retinal microvascular degeneration in a model of oxygen-ischemic retinopathy [15]. Here, TSP-1 was highly immuno localized in activated microglial cells in the retinas from hph-1 mice compared to WT (Fig. 5a). Besides this, the TSP-1 receptor (CD36) was strongly localized in the retinal microvasculature damaged in BH4-deficient mice (Fig. 5b), suggesting that TSP-1 secreted by activated microglial cells could act through its receptor highly expressed in endothelium, and be part of the mechanism implicated in the detrimental effect of the microvasculature. To evaluate this hypothesis, microglial cell cultures were exposed to hyperoxia-induced oxidative stress to induce its activation [11] and decrease BH4 levels as recently demonstrated [45]. Microglial cell cultures under hyperoxia were supplemented or not with an effective dose of BH4 [27] (100 μM; Fig. 6a). After 24 h of incubation, microglial cells lysates were used to evaluate TSP-1 expression by qPCR (Fig. 6b), while condition media was used on choroidal explants to evaluate the angiostatic activity in presence or absence of a neutralizing TSP-1 antibody (Fig. 7). As we expected, exposure of microglial cells to hyperoxia-induced oxidative stress for 24 h revealed a robust increase in TSP-1 mRNA expression (in lysates cells) and protein (by immunohistochemistry) (Fig. 6) compared to normoxia (21% O₂).

Interestingly, conditioned media (which contains TSP-1) from hyperoxia-exposed microglia was directly cytotoxic to the microvasculature from choroidal explants or microvascular endothelial cells in culture (Fig. 7). Conversely, BH4 supplementation markedly prevented hyperoxia-induced microglial activation as attested by diminished Iba-1 and TSP-1 expression (Fig. 6b) and prevented microvascular injury in choroidal explants (Fig. 7; corroborated by using a specific neutralizing TSP-1 antibody). Finally, by using the oxygen-induced retinopathy (OIR) model characterized by retinal vaso-obliteration associated with microglial activation during the first phase of the pathology [46]. We demonstrated the intravitreal injection of BH4 (100 μM final concentration) significantly decrease (p < 0.0002) retinal vascular degeneration after 48 h of hyperoxia (Additional file 7: Figure S7).

All these results show that in the absence of BH4, an augmented secretion of the anti-angiogenic TSP-1 derived from activated microglia (Fig. 5) contributes significantly to microvascular degeneration in the retina of hph-1 mice.