Selective Early Glial Reactivity in the Visual Pathway Precedes Axonal Loss, Following Short-Term Cerebrospinal Fluid Pressure Reduction

Abstract

Purpose: To examine the early glial reactivity and neuron damage in response to short-term cerebrospinal fluid pressure (CSFp) reduction, as compared with intraocular pressure (IOP) elevation.

Methods: The experiment included 54 male Sprague-Dawley rats with elevated translaminar cribrosa pressure difference (TLPD), defined as IOP minus CSFp. These rats underwent either continuous CSF drainage for 6 hours (n = 18), or unilateral IOP elevation to 40 mm Hg for 6 hours (n = 18). For control, 18 normal rats were anesthetized for 6 hours. Orthograde axonal transport was examined by intravitreal injection of 3% rhodamine-β-isothiocyanate. We also used transmission electron microscopy to display the ultrastructural features of retinal ganglion cell axons in the optic nerve head. Early glial reactivity in the retina, lateral geniculate nucleus (LGN), and superior colliculus (SC) was detected by immunostaining and Western blot for the glial fibrillary acidic protein (GFAP) and glutamine synthetase (GS). We also observed the glial reactivity in the inferior colliculus and hippocampus to rule out possible influences of CSF dynamics and composition.

Results: Anterograde staining with 3% rhodamine-β-isothiocyanate revealed decreased fluorescence intensity of the SC and LGN projected from both lower CSFp and higher IOP eyes. Transmission electron microscopy showed loss of axons from the optic nerve head in the high-IOP group, but not in the low-CSFp group. Compared with the anesthesia control group, GFAP expression was significantly increased in the retina, LGN, and SC, whereas GS expression was only increased in the retina following CSFp reduction. However, their expressions were not significantly elevated in the inferior colliculus and hippocampus. In the high-IOP group, expressions of GFAP and GS were significantly increased in the retina, LGN, and SC.

Conclusions: Visual system neurons may be much more sensitive than other nervous tissues. Following short-term CSFp reduction, early glial reactivity may precede axonal loss. Changes of translaminar cribrosa pressure difference in both experimental low-CSFp and high-IOP groups induce selective early glial reactivity. The neuron damage from abnormally low CSFp may be pathogenetically similar to high IOP.

Glaucoma is a multifactorial ocular disease that results in loss of retinal ganglion cells (RGCs) and optic nerve damage.^1,2 The pathologic mechanism of glaucoma is not fully understood yet. Intraocular pressure (IOP) is one of the most important risk factors involved in the development of glaucoma. In addition to IOP, there are many factors that may play an important role in glaucoma pathogenesis. In some cases, glaucoma progression was not fully prevented by either medical or surgical reduction of IOP.^3,4 Some patients with normal IOP also showed retinal nerve fiber layer defects (RNFLD), which may be accompanied by abnormally low cerebrospinal fluid pressure (CSFp).^5,6 Some patients with ocular hypertension without RGC damage may also be associated with high CSFp.^7–9 

Recently, the orbital CSFp against IOP across the lamina cribrosa has been found to play an increasingly important role in the pathogenesis of glaucoma. A series of clinical and experimental evidence, as well as biomathematical models, suggest that both low CSFp and elevated TLPD may be formidable risk factors for glaucoma or glaucoma-like optic neuropathy.^6,9–14 Abnormally low CSFp may suggest glaucomatous disease progression, such as optic disc hemorrhage and visual field (VF) loss; if CSFp is corrected, VF may be partly recovered.¹⁵ Also, postexperiment follow-ups show that CSFp modification has both structural and functional effects on the retina and optic nerve head (ONH) of animals.^16–20 

Furthermore, the dynamics of CSF, which are, in small quantities, not heterogeneous throughout all CSF compartments (especially in the optic nerve subarachnoid space [ONSAS], which is organized by a complex system of arachnoid trabeculae and septa), may also participate in the optic neuropathic procedure.^21–25 

The glia-neuron partnership is important for retinal and cerebral homeostasis, and any disturbance may result in neuron damage.^26,27 Glial cells act as a buffer system, which has both protective and cytotoxic effects on neurons.²⁸ The activation of glial cells is a hallmark of neural injury in the nervous system. Glial cells may be activated in a variety of retinal pathologic conditions, such as ischemia, inflammation, and trauma^29–31; activated glial cells have also been observed in afferent projections to the lateral geniculate nucleus (LGN) and superior colliculus (SC) in the early stage of high-IOP eyes.^32–34 

In this study, we attempted to examine the early influences of high TLPD, as induced by CSFp reduction and IOP elevation, on neuron damage and glial reactivity in visual system. To rule out the possible influences of CSF dynamics on the whole brain, we also detected the glial reactivity in the inferior colliculus (IC), which is adjacent to the SC, and the hippocampus, which is sensitive to stress. 

Materials and Methods

Animals

Fifty-four male Sprague-Dawley rats aged 6 to 8 weeks old (200–220 g) were used in this study. All of them were provided with standard food and water and kept in temperature-controlled rooms with a 12-hour light/dark cycle. The study was approved and monitored by the Institutional Animal Care and Use Committee of the Capital Medical University of Beijing (approval number AEEI–2016–093) and performed in accordance with the ARVO Statement for Use of Animals in Ophthalmic and Vision Research. 

Animal Model

The rats were divided into three groups: (1) experimental acute-CSFp-reduction group (n = 18), (2) experimental acute-IOP-elevation group (n = 18), and (3) control group with anesthetized rats (n = 18). The experimental procedures were described previously.^19,20 In brief, the rats were anesthetized with an intraperitoneal injection of chloral hydrate (400 mg/kg), followed by a midline scalp incision. The nuchal muscles were cut in the midline and stripped laterally to expose the atlantooccipital membrane with the dorsal surface of medulla oblongata. The membrane was pierced with a 22-gauge needle to expose the transparent dura mater. Then, using an operating microscope, we inserted a taper glass capillary tube into the cisterna magna. For rats in the CSFp-reduction group, CSF was aspirated every 15 minutes over a study period of 6 hours. The ventricular CSFp was measured using the pressure catheter (Scisense catheter, Catalog no. FTH-1611B-0018; Transonic Scisense, Inc., London, Ontario, Canada) and connected to the BIOPAC Systems MP150 workstation (BIOPAC Systems Co., Gleeta, CA, USA). 

Rats with elevated IOP were anesthetized with chloral hydrate and a topical conjunctival application of 0.5% proparacaine hydrochloride. The anterior chamber of the right eye was cannulated with a 30-gauge needle connected to a saline-filled bottle of adjustable height. The IOP was adjusted to a level of 40 mm Hg by calibrating the height of the bottle for the whole study period of 6 hours. 

Eighteen normal rats were anaesthetized with chloral hydrate for 6 hours as an anesthesia control to rule out possible neurotoxic effects. 

Imaging Study

To detect the connection and degeneration of the visual system from the eye to the LGN and SC, dyes were injected as previously described.^20,35 In brief, 2 μl of freshly prepared 3% rhodamine-β-isothiocyanate (RITC) was injected intravitreally with a 28-gauge needle at the pars plana of each right eye. The brains were removed 3 days after injection. The differences in LGN and SC staining were macroscopically observed in the low-CSFp, high-IOP, and anesthesia control groups by using the Kodak In-Vivo Imaging System FX Pro (n = 5 per group). Fluorescence intensity in the left LGN and SC of low-CSFp and high-IOP brains were compared to those in the anesthesia control group. The relative average value of RITC fluorescence brightness in the LGN and SC was analyzed with the Carestream MI software. 

Axonal Changes in the ONH

Transmission electron microscopy was used to observe the ultrastructure of the retina and ONH, as described previously. In brief, at 3-days postoperation, rats were deeply anaesthetized, transcardially perfused with PBS (0.01 M; pH, 7.4), and then fixed with 2% glutaraldehyde and 1% formaldehyde mixture in PBS. The removed globe anterior and posterior segments were postfixed in 1% buffered osmium tetroxide. The ultrathin sections were collected on mesh nickel grids and examined using a Hitachi H-7650 electron microscope (Tokyo, Japan; n = 3 per group).¹⁹ 

Immunocytochemical Analysis

Samples were cut into 5-mm sections by using a cryostat (Leica Microsystem, Wetzlar, Germany). The slices were rinsed with PBS, washed 3 times for 5 minutes at room temperature (RT), and then blocked in 5% bovine serum albumin for 30 minutes at RT. Following that, the slices were transferred to a moist chamber with a monoclonal mixture of mouse anti-glial fibrillary acidic protein (GFAP; 1:300) and rabbit anti-glutamate synthetase (GS; 1:200) antibodies overnight at 4°C. After that, they were rinsed three times with PBS. Then, secondary antibodies of Alexa Fluor 488 anti-mouse (1:1000) and Alexa Fluor 594 anti-rabbit (1:1000) were incubated in dark for 1 hour at RT. Cell nuclei were counter-stained with 4′6-diamidimo-2-phenylindoke (DAPI). Fluorescence signals were visualized by Leica fluorescence microscopy (n = 5 per group). 

Western Blot

Eyes were enucleated under deep chloral hydrate anesthesia. The retinas, LGN, SC, IC, and hippocampus were collected (n = 5 per group). All samples were immediately frozen to −80°C and were lysed later in ice-cold RIPA buffer (Sigma-Aldrich Corp., St. Louis, MO, USA) containing protease and phosphatase inhibitor. The lysates were centrifuged for 15 minutes at 12,000 g at 4°C, and the supernatants were assayed for protein concentration. With 50 μg of protein loaded in each lane, proteins were transferred to polyvinylidene difluoride membranes (Bio-Rad, Richmond, CA, USA). The membrane was blocked with 5% nonfat dry milk in Tris-buffered saline and Tween-20 (TBST) and incubated overnight at 4°C with mouse or rabbit monoclonal anti-GFAP, and rabbit anti-GS at a concentration of 1:1000. Peroxidase-conjugated goat anti-mouse/rabbit secondary antibody (CWBio, Beijing, China) was used at a concentration of 1:1000. Immunoblots were visualized by enhanced chemiluminescence (CWBio). Images were analyzed with Image Lab (Bio-Rad Laboratories, Hercules, CA, USA), and band densities were normalized using actin or glyceraldehyde 3-phosphate dehydrogenase (GAPDH) as an internal calibrator.¹⁹ 

Statistics

Statistical analysis was performed using SPSS 21.0. All results were expressed as mean ± standard deviation (SD). Data normal distribution was tested using one-sample Kolmogorov-Smirnov test. The homogeneity of variance was examined using Levene's test. Normally distributed parameters were analyzed using ANOVA with Bonferroni-corrected post hoc tests. Nonparametric parameters were analyzed using ANOVA on ranks with the Tukey test used for post hoc paired analysis. P ≤ 0.05 was considered as statistically significant. 

Results

In the experimental acute-CSFp-reduction group, the mean CSFp was reduced successfully, according to standard procedures, as previously demonstrated (Fig. 1).^19,20 

Macroscopic Analysis of the Visual Tract in Low-CSFp and High-IOP Rat Models

Upon injection of 3% RITC into the right eye (brains having been removed 3 days after injection), LGN and SC damage on the left side was clearly visible (Fig. 2). The mean fluorescence intensity was 794.2 ± 34.3 arbitrary (arb.) units in the left LGN and SC of CSFp rats (Fig. 2B) and 239.8 ± 39.7 arb. units in those of IOP rats (Fig. 2C). The average fluorescence intensity of anesthesia control rats was 998.2 ± 16.4 arb. units in the left LGN and SC (Fig. 2A). 

Figure 1

 

Changes of the CSFp in Sprague-Dawley rats pre- and postcontinuous cerebrospinal fluid drainage. n = 5 rats per group.

Figure 2

 

Macroscopic images of the LGN and SC in the anesthesia control (A), and experimental low-CSFp, high-IOP rats (B, C) n = 5 rats per group.

The average fluorescence intensity in the left LGN and SC was significantly decreased in high-IOP and low-CSFp rats as compared with the anesthesia controls (P < 0.01). 

Axon Loss in the ONH

To further identify axon loss in the ONH, we used transmission electron microscopy to monitor the cross-sections of ONH that had been subjected to acute CSFp reduction or IOP elevation. Three days after operation, the axons were normal and showed close membrane contact with astrocytes processes in the low-CSFp and anesthesia control groups (Figs. 3A, 3B). However, shrunken or orphaned axons were observed at 3-days postoperation (white arrow) only in the high-IOP group (Fig. 3C) 

Figure 3

 

Transmission electron micrographs of ONH sections. In the anesthesia control ONH, normal RGC axons, and astrocytes processes were identified (A). In the low-CSFp ONH, axons and astrocyte processes were observed free of axon damages at 3-days postoperation (B). In the high-IOP ONH, the shrunken or orphaned axons (white arrows) without astrocytes processes were visible, and the bold extracellular space (black arrowhead) was left by the degeneration of axons and glial processes (C). n = 3 rats per group.

Expressions of Retinal GFAP and GS in the Low-CSFp and High-IOP Rat Models

Compared with the anesthesia control group, foot processes of Müller cells at the inner limiting membrane and external limiting membrane respond with reactive gliosis, demonstrated with GFAP and GS labelling in both high-IOP and low-CSFp eyes (Figs. 4A–G). 

Figure 4

 

Changes of GFAP (red) and GS (green) expressions in Müller cells of the anesthesia control, CSFp reduction, and IOP elevation rat retinas. A1–G1, immunofluorescence labeling showed GFAP in rat retinal vertical slices taken from anesthesia controls, and those obtained at days 1, 3, and 7 after CSFp reduction and IOP elevation operations. A2–G2, immunofluorescence staining showed GS expressions in the same slices as in A1–G1, respectively. A3–G3, merged images of A1–G1, A2–G2. Scale bar: 100 μm for all images. GCL, ganglion cells layer; INL, inner nuclear layer; ONL, outer nuclear layer. n = 5 rats per group.

Compared with the anesthesia group, the expressions of GFAP were 1.28 ± 0.02-fold, 1.78 ± 0.01-fold, and 1.83 ± 0.18-fold at days 0, 3, and 7, respectively, after acute CSFp reduction, which was significantly increased from the anesthesia control group despite the fact that GFAP expression between days 3 and 7 was relatively stable. As for the IOP-elevation group, GFAP proteins were 1.54 ± 0.08-fold, 2.04 ± 0.11-fold, and 2.63 ± 0.14-fold at days 0, 3, and 7, respectively, with significant increases especially at day 7, with the elevation persisting throughout the study period (Figs. 5A, 5B). 

Figure 5

 

Western blotting results of GFAP and GS in retinas. (A–D) Quantitative results revealed increases of GFAP and GS expression in retina over time after acute IOP elevation or CSFp reduction. The relative intensity of the chemiluminescence for each protein band was normalized using GAPDH, and the data were presented as mean ± SD of the fold increase compared with anesthesia controls. n = 5 rats per group. *Versus control and P ≤ 0.05. n = 5 rats per group.

In the CSFp-reduction group, GS proteins were 1.43 ± 0.04-fold, 1.71 ± 0.05-fold, and 1.78 ± 0.01-fold at days 0, 3, and 7, respectively, with significant increases compared with those of the anesthesia control group. In the higher-IOP group, GS proteins were 1.57 ± 0.04-fold, 1.80 ± 0.02-fold, and 1.80 ± 0.02-fold at days 0, 3, and 7, respectively, with significant increases compared with the control group (Figs. 5C, 5D). In addition, there was no difference in GS expression between postsurgery days 3 and 7 in both groups. 

Expressions of GFAP and GS in LCN of Both Low-CSFp and High-IOP Rat Models

In both acute-CSFp-reduction and IOP-elevation groups, Western blot analysis revealed a significant increase in GFAP protein levels in the LGN (Figs. 6A, 6B), whereas the GS expression was significantly changed only in the high-IOP group (Figs. 6A, 6C). 

Figure 6

 

(A–C) The relative intensity of the chemiluminescence for each protein band was normalized using GAPDH, and the data were presented as mean ± SD of the fold increase compared with controls. *Versus control and P ≤ 0.05. n = 5 rats per group.

Expressions of GFAP and GS in SC of Both Low-CSFp and High-IOP Rat Models

Immunofluorescence labeling for GFAP was performed in the SC (Figs. 7A–C). Western blot analysis was also performed to assess the protein levels of GFAP and GS in the SC. GFAP expressions for acute CSFp-reduction and IOP-elevation groups were significantly increased in the LGN, and the amount of GFAP notably reached a peak at day 3 in the low-CSFp group. The GFAP was significantly increased especially at day 7, and the elevation persisted throughout the study period (Figs. 7D, 7E). 

Figure 7

 

Immunohistochemistry and Western blot analysis of GFAP and GS in SC of low-CSFp, high-IOP, and anesthesia groups. A1–C1, immunofluorescence labeling showed GFAP in rat SC vertical slices. A2–C2, immunofluorescence staining showed DAPI in the same slices as in A1–C1, respectively. A3–G3, merged images of A1–C1, A2–C2. Scale bar: 100 μm for all images. Western blot results of GFAP and GS in SC. Quantitative results revealed the expressions of GFAP after acute IOP elevation or CSFp reduction (D, E). Quantitative results revealed an increase of GS after acute IOP elevation or CSFp reduction (D, F). The relative intensity of the chemiluminescence for each protein band was normalized using actin, and the data were presented as mean ± SD of the fold increase compared with the controls. n = 5 rats per group. *Versus control and P ≤ 0.05.

In the CSFp-reduction group, GS expressions did not change significantly compared to the anesthesia control group, whereas the IOP-elevation group saw significant increases (Figs. 7D, 7F). 

Expressions of GFAP and GS in the IC of Low-CSFp Rat Model

In order to rule out the influence of CSF dynamics on the whole brain and determine whether early glial reactivity selectively occurred in the whole-visual pathway, we further explored the early glial reactivity in the IC, as adjacent to the SC, and hippocampus, which was sensitive to damage in the CSFp-reduction group as a control. 

We estimated that GFAP and GS expressions in the acute-CSFp-reduction group would increase slightly in the IC but without significant statistical difference (Fig. 8). Western blot analysis revealed 1.19 ± 0.03, 1.04 ± 0.01, and 0.90 ± 0.05-fold increase of GFAP expressions at days 0, 3, and 7 postsurgery, respectively (Figs. 8A, 8B), and 1.19 ± 0.01, 1.04 ± 0.01, and 1.01 ± 0.01-fold increase of GS expressions (Figs. 8A, 8C) in the IC at days 0, 3, and 7 postsurgery, respectively. 

Figure 8

 

Western blot results of GFAP and GS in the IC. (A–C) Quantitative results revealed no significant change of GFAP (A, B) and GS (A, C) expressions in the IC at days 0, 3, and 7 in the CSFp-reduction group, respectively. The relative intensity of the chemiluminescence for each protein band was normalized using actin, and the data was presented as mean ± SD of the fold increase compared with the controls. n = 5 rats for each group.

GFAP and GS Expressions in Hippocampus of Acute-Lower-CSFp Rat Models

We further estimated GFAP and GS expressions in the hippocampus of the acute-CSFp-reduction group. Western blot analysis showed slight GFAP and GS changes without any significant statistical difference (Fig. 9). GFAP expressions changed by 0.97 ± 0.14, 1.35 ± 0.61, and 0.76 ± 0.06-fold at days 3 and 7 and 1 month, respectively, compared with the control group (Figs. 9A, 8B). GS expressions were 1.06 ± 0.22, 0.88 ± 0.29, and 0.55 ± 0.10-fold at days 3 and 7 and 1 month, respectively (Figs. 9A, 9C). 

Figure 9

 

Western blot results of GFAP and GS in hippocampus 3 and 7 days and 1 month after CSFp reduction compared with the control group. (A–C) Quantitative results revealed no significant change of GFAP (A, B) and GS (A, C) expressions in the hippocampus over a period of 3 and 7 days and 1 month after CSFp reduction, respectively. The relative intensity of the chemiluminescence for each protein band was normalized using GAPDH, and the data were presented as mean ± SD of the fold increase compared with the controls. n = 5 rats per group.

Discussion

In this study, we showed that the anterograde axonal flow from the retina to the SC and LGN was destructed in both experimental low-CSFp and high-IOP eyes, whereas shrunken axons were observed only in the high-IOP group. The optic neuropathy in both experimental low-CSFp and high-IOP groups was accompanied with early glial reactivity in the visual system. In the experimental CSFp-reduction group, GFAP expressions were significantly increased in the retina, LGN, and SC, whereas the GS expression only increased in the retina compared with the control group. In the high-IOP group, GFAP and GS expressions were also elevated in the retina, LGN, and SC. However, GFAP and GS expressions were not markedly changed in the IC and hippocampus in the low-CSFp group. We propose that the early selective glial cells' reaction might have resulted from changes in the TLPD and might have a correlation with optic neuropathy following acute CSFp reduction and IOP elevation. 

The idea that orbital CSFp may play a contributory role in the pathogenesis of glaucoma has been discussed for more than 40 years.¹⁰ It has been demonstrated that CSFp is lower in primary open angle glaucoma and normal tension glaucoma (NTG) and elevated in ocular hypertensive patients.^6,9,12,13 However, some reports do not support these opinions. In a retrospective analysis, Pircher et al.³⁶ found the TLPD was not significantly correlated with the mean defect of VF in NTG patients. This result was later supported by a prospective case-control study, where the CSFp (lamina cribrosa) was not reduced in NTG patients as compared with healthy controls, either in supine or upright positions; however, the CSFp (lamina cribrosa) and TLPD were not measured directly but calculated.³⁷ Further prospective or experimental studies were needed to determine whether CSFp has a fundamental role in the pathogenesis of glaucoma. 

It is noteworthy that our lateral ventricle pressure of 11.3 ± 1.9 mm Hg at baseline is much higher than what was found by most other researchers, when taking pressure from the lateral ventricle.^38–42 The highest pressure recorded by any other research was about 8 ± 1.6 mm Hg,⁴⁰ and the lowest CSFp was 3.4 (1.4–5.4) mm Hg.⁴² A comparison between these studies and the present study reveals the main difference may be due to the different CSFp measurement methods. The other studies preferred the fluid filled system, whereas we chose the pressure sensor system. When our solid pressure sensor catheter was inserted into the lateral ventricle, the CSFp increased given the minor change in ventricular volume. The undetectable leakage occurring at the insertion site of the fluid filled catheter should also be considered.⁴⁰ The reported lateral ventricle CSF pressures based on the fluid filled system were quite different from each other. We suppose that the repeatability may be improved even when using the same measuring technique and same measuring site. In some cases, our lateral ventricle CSFp was much lower or similar to the previous reports, in which the intraparenchymal^40,43 or lateral ventricle^19,20 CSFp was about 10 to 11 mm Hg, with most being in the 5-mm Hg range. It may support the opinion that ICP measurements vary widely depending on the location of measurement and technique used.⁴⁴ 

Direct ventricular cannulation was developed by Lundberg in 1960 and considered the “gold standard” for CSFp measurement.⁴⁵ The cisterna magna measurement was less reliable and produced significantly lower readings,⁴⁰ and the intraparenchymal device produced greater cortical damage compared with ventricle CSFp. In this research, the ventricular CSFp monitoring was the preferred method, because it is accurate and induces the least brain damage.⁴⁰ We have used a standardized technique and specialized equipment for the ventricle CSFp recording.^19,20 The protocol includes setting the 0 level very precisely and careful avoidance of puncturing the dura mater and leakage when placing the pressure sensor catheter. Both experimental low-CSFp and anesthesia control rats were investigated with the same protocol. We are therefore confident that the CSFp measurements in the present study are reliable. 

In this study, we found that the anterograde axonal flow from the retina to the SC and LGN was obstructed in both experimental low-CSFp and high-IOP eyes, which was consistent with our previous study.²⁰ Furthermore, it may be that the axonal transport proteins were mainly blocked on the ONH, where the axons passed through, and which was sensitive to pressure as it en faces the anterior IOP and posterior retroocular CSFp.¹⁹ As IOP elevates, the high TLPD may be accompanied with ONH deformation that might be the result of posterior laminar deformation, neural canal expansion, lamina cribrosa thickening, and posterior bowing of peripapillary sclera.^46–50 Similar to IOP elevation, the high TLPD induced by low CSFp may also be followed by ONH movement and functional changes,^18,49 and may further influence the axonal transport. The peripheral nerves can tolerate an absolute pressure increase of up to 3800 mm Hg⁵¹; however, a minor pressure gradient of 4.5 mm Hg/100 μm will significantly reduce the orthograde axonal transport.⁵² Consistent with low CSFp, the change of TLPD caused by elevated CSFp may also lead to papilledema that has been thought to be a disturbance of axoplasmic flow.^53,54 

GFAP has been considered a sensitive and reliable biomarker of reactive cell gliosis or neuropathy.^55,56 GS is a glial enzyme and specific marker for retinal Müller cells, which aberrantly converts glutamate to glutamine to protect neurons from glutamate excitotoxicity.^33,57 Unlike the high-IOP group, retinal GFAP and GS expressions were stable at 3 and 7 days after acute CSFp reduction. We speculate that in the low-CSFp group, the initial glial response was relatively mild and remained stable or even partially reversible over time, as, in a brief and mild IOP-elevation model, the morphologic remodeling following glial cell reactivity might be reversible.³⁴ In the low-CSFp group, the gliosis was not accompanied with shrunken axons, which was consistent with our previous study.²⁰ We propose that this subtle injury, induced by a short-term reduction of CSFp, may play an initially protective role by reestablishing the extracellular medium and by supplementing neurons with factors that promote their survival.³² 

In our study, both experimental low-CSFp and high-IOP groups were accompanied with early glial reactivity in the visual system. Glial reactivity is involved in the initial phases of glaucoma, and it has been observed that it may influence the whole visual pathway.^32,33 It is different from the high-IOP group, the results showed that GS expressions in the LGN and SC were relatively stable, which was not the case with GFAP expressions in the low-CSFp group. The dissociation between GFAP and GS expressions was not surprising, given that it has been previously demonstrated.⁵⁸ The retinal GS expression remained stable after optic nerve crush, whereas GFAP remained elevated over 2 weeks.⁵⁶ 

It has been established that CSF plays an important role in mechanical cushioning of the brain. It also nurtures neurons, axons, and glial cells and acts as a transport system for the removal of waste material. The dynamics of CSF may also participate in the optic neuropathic procedure.^23–25 The CSF dynamics are complex, and the flow of CSF between intracranial and ONSAS is neither continuous nor bidirectional, as ONSAS is divided by a complex system of arachnoid trabeculae and septa.^21,22 Despite that, the ONSAS pressure is largely dependent on the lateral ventricular CSF pressure, when the latter is within a certain range.^16,59,60 Patients with higher CSFp showed unilateral optic disc swelling and asymmetric papilledema with distension of the perioptic nerve sheath, whereas in patients with intracranial hypotension, the ONSAS failed to be visualized with magnetic resonance imaging, even after the CSFp was again increased to normal levels.⁶¹ 

In this study, CSF dynamics were influenced by the drainage of CSF. We speculate that it was hard to distinguish whether the damage induced by CSFp reduction was glaucoma or a glaucoma-like disease. As an attempt to rule out possible influences of CSF dynamics, we studied the early glial cell response in the IC that was adjacent to the SC as control; we also studied long-term glial reactivity in the hippocampus that is a medial temporal lobe structure and is highly susceptible to stress as astrocytes activate.^62,63 Our study revealed that both GFAP and GS expressions in the IC and hippocampus were changed without any significant statistical difference in the low-CSFp group, which was consistent with our previous study that found almost no additional damages to the whole visual pathway.¹⁹ 

An obvious question was, why did selective early glial reactivity influence the retina, LGN, and SC but not the IC or hippocampus? The mechanism of selective visual pathway influence remained elusive. Firstly, it had been suggested that the retina was more sensitive to damage because of its particularly energy-hungry projection compared with other nervous tissues.⁶⁴ Secondly, the unique anatomic characteristics of retroocular optic nerves, as surrounded by CSF throughout its entire length, may suffer directly from the changeable mechanical stress of CSFp. Thirdly, the dynamics and components of orbital CSF may be different with ventricular CSF that turns over up to 5 times a day and replenishes rapidly.^25,65 Last, but not least, the cumulative effects might be well tolerated by brain tissues, as the period of CSF drainage was not long enough. 

This research also had its weaknesses. First of all, despite our best simulation efforts, our experimental low-CSFp rat models were fundamentally distinct from NTG or primary open angle glaucoma patients with lower CSFp. A chronic low-CSFp rodent model to simulate clinical patients was needed. Second, the TLPD between the low-CSFp group and the high-IOP group was quite different. Third, the follow-up in both groups was limited. Glial cell response may still continue, and whether the glial reactivity sustains or diminishes in a longer period was not clear. Fourth, the ventricle CSFp and the retroocular CSFp may be quite different. It was hard to define the exact retroocular orbital pressure and the TLPD. 

In conclusion, we showed that the anterograde axonal flow from the retina to the SC and LGN was obstructed in both experimental low-CSFp and high-IOP eyes. The selective early glial reactivity may result from changes in the TLPD, whereas shrunken axons were observed only in the high-IOP group. The neuron damage from abnormally low CSFp and elevated IOP might share a similar mechanism. 

Acknowledgments

Supported by the National Natural Science Foundation of China (81470635, 81600725); Beijing Natural Science Foundation (7162037); and Beijing Tongren Hospital Affiliated to Capital Medical University research foundation (2015-YJJ-ZZL-003). 

Disclosure: X.X. Li, None; Z. Zhang, None; H.Y. Zeng, None; S. Wu, None; L. Liu, None; J.X. Zhang, None; Q. Liu, None; D.Y. Yang, None; N.L. Wang, None