Lack of Blood–Brain Barrier Properties in Microvessels of the Prelaminar Optic Nerve Head

Abstract

purpose. To define the blood–brain barrier (BBB) characteristics of microvessels in the optic nerve head (ONH).

methods. Immunohistochemical staining of different regions of the ONH, retro-laminar optic nerve, and retina of human and monkey eyes was carried out, using antibodies against BBB markers (glucose transporter 1, transferrin receptor, and P-glycoprotein), the non-BBB marker PAL-E, and against plasma proteins fibrinogen and IgG, which serve as endogenous markers of nonspecific microvascular permeability. In the ONH of monkey eyes, the number of transport-related endothelial pinocytotic vesicles and their cellular distribution within the microvessels were determined by electron microscopy.

results. In both human and monkey eyes, only microvessels in the prelaminar region of the ONH were positive for the PAL-E antigen. The prelaminar region microvessels showed either no or weak expression of the transferrin receptor and P-glycoprotein but stained positive for glucose transporter 1. In human ONH, fibrinogen and IgG were present around microvessels in the prelaminar region but not in other parts of the optic nerve or retina. By electron microscopy, endothelial cells of prelaminar region microvessels contained a higher number of pinocytotic vesicles, located at the luminal and abluminal side of the endothelial cell membrane, in contrast to a mainly abluminal localization in microvessels of the retina and other parts of the optic nerve.

conclusions. Microvessels in the prelaminar region of the ONH lack classical BBB characteristics and display nonspecific permeability, possibly mediated by vesicular transport.

The prelaminar region (PLR) of the optic nerve head (ONH) is involved in several eye diseases. ^{1 2 3 4 5 6} In particular, a variety of optic neuropathies is located at the PLR ^{3 4 5 6} and circulatory disturbances in this area are thought to be one of the causes of glaucoma. ^{1 2} The reasons for the specific vulnerability of this region to pathologic conditions are as yet unclear. The vascular supply to the ONH originates from the short ciliary arteries and the central retinal artery in many different species (i.e., rat, rabbit, monkey, and human). ^{7 8 9 10 11 12} The vasculature of the optic nerve, and of the central nervous system in general, constitutes a blood–brain barrier (BBB). The BBB capillary endothelium is a specialized continuous endothelium, with properties distinct from permeable continuous endothelium found in skin and muscle and from the highly permeable fenestrated endothelium, as found in the choriocapillaris. In general, impermeable tight junctions and the presence of only few pinocytotic vesicles characterize the BBB endothelium. ¹³ The BBB endothelium can immunohistochemically be defined by its expression of specific membrane proteins, so-called BBB markers. Most of these are proteins involved in regulating the passage of substances through the capillary wall, e.g., the glucose transporter (Glut) 1, transferrin receptor (Transf-R), and P-glycoprotein (P-gP). ¹⁴  

It is generally assumed that microvessels in the entire optic nerve have BBB characteristics. ^{15 16 17 18} This notion was based on studies using exogenous tracers, such as horseradish peroxidase and sodium fluorescein, and from light and electron microscopic studies. ^{16 19 20 21 22 23 24} In these studies, it was observed, however, that extravasated tracer was present in the PLR and the lamina cribrosa (LC) of the ONH. Because no leakage of tracer could be observed via the tight junctions of the microvessels in the PLR and the LC, it was assumed that the extravasation of tracer was due to diffusion from the highly permeable peri-papillary choroid. ^{16 19 20 21 22 23 24}  

Only few studies addressed the specific properties of the microvessels in the PLR. ^{25 26} Knowledge of these properties is important, because differences in functional vascular characteristics in this area may play a role in the pathogenesis of glaucoma and other diseases. 

Recently, we reported on the distribution of the non-BBB marker PAL-E in the brain and in the eye. ^{27 28} In the brain, this PAL-E antigen is present only in the permeable capillaries, as found in the choroid plexus and the dura mater, but it is absent in brain capillaries possessing a BBB. ²⁸ In the eye, PAL-E is present in the permeable microvessels of the choroid and ciliary processes but is absent in capillaries possessing a blood–retinal barrier. ²⁷ In this study, we unexpectedly observed PAL-E–positive microvessels in the PLR, ²⁷ suggesting that the vasculature in this area lacks a BBB function. The purpose of the present study was to further define the presence or absence of BBB characteristics in microvessels of the different ONH regions. 

Materials and Methods

Tissues

Normal eyes of eight persons without known eye or vascular disease were obtained from the Corneabank Amsterdam, The Netherlands, after removal of corneal buttons for transplantation. Normal eyes of nine Rhesus monkeys (Macaca mulatta) were obtained from the Biomedical Primate Research Center, Toegepast-Natuurwetenschappelijk Onderzoek (TNO), Rijswijk, The Netherlands, and from the Central Animal Laboratory of the University of Nijmegen. These monkeys were killed after other unrelated studies with no known effects on the eyes. After death, the monkeys were perfused with phosphate-buffered saline (PBS, pH 7.4). All experimental procedures conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and the human eyes were treated in accordance with the Declaration of Helsinki on the use of human material for research. 

Immunohistochemistry

The distribution of BBB-type microvessels in the ONH was investigated using immunohistochemical staining with a panel of BBB markers. For this purpose, air-dried longitudinal cryosections (10μ m) of the posterior segment and retina of eight human and four monkey eyes were cut, fixed in cold acetone for 10 minutes, and air-dried for at least 2 hours at room temperature. Sections were preincubated with 10% normal goat serum (Jackson ImmunoResearch Laboratories, West Grove, PA) in PBS (pH 7.4) for 30 minutes to reduce nonspecific staining, followed by an overnight incubation at 4°C with one of the following monoclonal antibodies: PAL-E (1/500), anti-transferrin receptor (Transf-R, 1/20; DAKO, Glostrup, Denmark), anti–P-glycoprotein (P-gP, 1/250; kindly provided by P. Van de Valk, PhD, Department of Pathology, Free University, Amsterdam, The Netherlands), anti-glucose transporter 1 (Glut-1, 1/250; kindly provided by Lars Andersson, PhD, Uppsala, Sweden) and anti-fibrinogen (1/100; kindly provided by J. J. Emeis, PhD, Gaubius Laboratories, TNO, Leiden, The Netherlands), or with a anti-human IgG rabbit polyclonal antibody conjugated to horseradish peroxidase (1/250; DAKO). Sections were washed for 15 minutes with PBS and incubated with goat anti-mouse immunoglobulin conjugated to horseradish peroxidase (1/100; DAKO), except for the antibody against IgG. Sections were washed again for 15 minutes and stained with di-amino benzidine (10 mg/ml) with 0.01% H₂O₂ in PBS for 10 minutes. The reaction was stopped with H₂O. Sections were counterstained with hematoxylin, dehydrated through a series of graded ethanol, coverslipped with Entellan (Merck, Darmstadt, Germany), and examined light microscopically. For controls, the first antibodies were omitted or replaced by an irrelevant antibody (OX-43; Serotec, Oxford, England). 

To investigate the exact distribution of PAL-E staining in relation to expression of the Transf-R or Glut-1 in individual microvessels, double immunostaining was performed on cryosections of human eyes by incubating the sections with a mixture of PAL-E (IgG2a) and Glut-1 (IgG1) or a mixture of PAL-E and Transf-R (IgG1). First antibodies were used in the same dilution as described above, and incubations were performed overnight at 4°C. After incubation, the sections were washed in PBS for 30 minutes, followed by incubation with a mixture of the following secondary antibodies: goat anti-mouse IgG2a conjugated to FITC (Southern Biotechnology Associates Inc., Birmingham, AL) and goat anti-mouse IgG1 conjugated to TRITC (Southern Biotechnology Associates Inc.). Sections were rinsed for 30 minutes and coverslipped with Vectashield Mounting Medium (Vector Laboratories, Burlingame, CA). To ascertain the specificity of the double immunolabeling, control sections were incubated with only single primary antibody followed by the mixture of secondary antibodies. Sections were studied using a Leitz (Wetzlar, Germany) confocal laser-scanning microscope. 

Electron Microscopy

To investigate possible differences in the number and distribution of the transport-related pinocytotic vesicles of endothelial cells in the optic nerve, electron microscopic studies were carried out. These studies were performed on monkey material only, because we were unable to perfuse human material. Posterior segments of five perfused monkey eyes were fixed in 1.25% glutaraldehyde and 2% paraformaldehyde in PBS (pH 7.4) for 3 hours. After fixation, the tissues were cryoprotected with 15% glucose in PBS and embedded in 20% gelatin in PBS and frozen in liquid nitrogen. Sections of 80 μm were cut and postfixed in OsO₄ for 15 minutes, dehydrated with a series of graded ethanol, and flat-embedded in a thin film of epoxy resin. Different areas of ONH (i.e., PLR, LC, and retro-laminar region [RLR]) were dissected out of the flat-embedded sections and used for electron microscopy. Ultrathin sections (60–80 nm) were cut, and in each of the areas a random selection of microvessels (n ≥ 5) was photographed with a Philips 201 transmission electron microscope (Philips Industries, Eindhoven, The Netherlands) at 60 kV. 

Using the micrographs of individual microvessels, a quantitative evaluation was performed using a VIDAS image analysis system (Kontron, München, Germany). For each microvessel the length of the luminal and abluminal endothelial cell membrane was measured, and the total number of vesicles on the abluminal and luminal cell membrane was counted. Next, the number of pinocytotic vesicles per unit length along the luminal and abluminal cell membrane and the ratio of the luminal versus abluminal pinocytotic vesicles were calculated. For each microvessel the total number of cytoplasmic vesicles and the total area of the cytoplasm was measured using a square grid with lines 5 mm apart. The grid was superimposed over the electronmicrographs. Using calibrated magnifications, the number of vesicles perμ m² cytoplasm was determined. 

Statistical Analysis of Microvessels in the Monkey

For each individual monkey, the means and SEM of the ratio of luminal versus abluminal pinocytotic vesicles and the mean and SEM of the number of vesicles per μm² cytoplasm in the different regions of ONH and the retinas were determined. From these data the means and SEM for each ONH region and retina of all monkeys together could be calculated. An unpaired t-test was used to determine the level of significance of differences between the various regions studied. The level of significance was set at 5%. 

Results

Three different regions in the ONH were identified (Fig. 1) : (1) the PLR, localized at the level of the choroid and retina, which includes the superficial nerve fiber layer and is characterized by minimal connective tissue and the absence of clearly myelinated axons, (2) the LC, localized at the level of the sclera and characterized by large connective tissue septa, and (3) the RLR, localized outside the eye and characterized by myelinated axons and delicate connective tissue septa. These ONH regions were compared with the retina (Fig. 1) . 

Immunohistochemical Staining for BBB Markers in Human Eyes

Prelaminar Region.

The majority of PLR microvessels were positive for the non-BBB marker PAL-E (Fig. 2A ), whereas staining for BBB markers P-gP and Transf-R (Fig. 2G) was weak or absent. In contrast, the PLR microvessels were clearly positive for the BBB marker Glut-1 (Fig. 2J) . Endogenous IgG and fibrinogen, markers for microvascular permeability, were localized around PLR microvessels (Fig. 2D) . Double-staining experiments were performed to study whether these PAL-E–negative microvessels were positive for the Transf-R. These double-staining experiments for PAL-E and Transf-R demonstrated different populations of microvessels in the PLR: microvessels were positive for either the PAL-E antigen (red) or the Transf-R (Fig. 3A , green). Only a limited number of microvessels were positive for both markers (Fig. 3A , arrowhead). Doublestaining experiments using Glut-1 and PAL-E showed that most microvessels coexpressed these markers (results not shown). 

Lamina Cribrosa.

Microvessels in the LC were negative for the non-BBB marker PAL-E (Figs. 2B and 3B) , but positive for BBB-markers Transf-R (Figs. 2H and 3B) , P-gP, and Glut-1 (Fig. 2K) , suggesting a patent BBB in this area. Fibrinogen and IgG were present in the connective tissue of the LC and near the surrounding choroid (Fig. 2E) . However, marked perivascular staining of IgG and fibrinogen, as observed in the PLR, was not seen in the LC. 

Retro-laminar Region.

PAL-E staining was absent in the RLR (Fig. 2C) , and all microvessels were positive for the BBB-markers Glut-1 (Fig. 2L) , P-gP, and Transf-R (Fig. 2I) . Fibrinogen and IgG were present in the connective tissue of the PLR (Fig. 2F) , but no perivascular staining for these permeability markers was observed. 

Retina.

Retinal microvessels were negative for PAL-E (Fig. 4A ) and positive for the Transf-R (Fig. 4C) and Glut-1 (Fig. 4D) . However, capillary staining for Glut-1 was masked by staining of all other cells in the retina. In the adjacent choroid, marked staining of microvessels for PAL-E was found but not for Glut-1, Transf-R and P-gP. As reported previously, some arterioles in the choroid showed Glut-1–positive endothelium. ²⁷ In the retina, staining of extravasated fibrinogen and IgG was absent, in contrast to marked staining in the adjacent choroid (Fig. 4B) . 

Immunohistochemical Staining for BBB Markers in Monkey Eyes

Prelaminar Region.

As in the human eyes, in the PLR of monkey ONH, PAL-E–positive microvessels were present, whereas staining of microvessels for Glut-1 was moderate. Transf-R and P-gP were not studied because our antibodies do not recognize these proteins in monkey tissues. IgG and fibrinogen were present in the entire PLR, whereas in some parts the staining was concentrated around microvessels (results not shown). 

Lamina Cribrosa and Retro-laminar Region.

In LC and RLR, PAL-E–positive microvessels were not observed. All microvessels showed a strong staining for Glut-1. IgG and fibrinogen were present in the stroma of the entire LC, but no perivascular staining was observed. 

Retina.

In the retina, PAL-E–positive microvessels were not present. Capillary staining for Glut-1 was masked by positive staining of all retinal cells. IgG and fibrinogen staining was absent in the retina. In the choroid, PAL-E–positive microvessels were observed. Glut-1 positivity was observed in some arterioles. Marked staining for IgG and fibrinogen was observed in the choroid. 

All results are summarized in Table 1 . 

Electron Microscopy of Microvessels in Monkey Eyes

Prelaminar Region.

Ultrastructurally, microvessels of the PLR in monkey ONH displayed a continuous endothelium, with intercellular junctions suggesting the presence of tight junctions. In endothelial cells of these microvessels pinocytotic vesicles were located at the luminal and abluminal cell membrane (Fig. 5A) . When compared with microvessels in LC, RLR, and retina, a significantly higher ratio of luminal versus abluminal pinocytotic vesicles was found in individual microvessels in the PLR (P < 0.005; Fig. 6A ). Moreover, the endothelial cells of microvessels in the PLR had a significantly higher number of vesicles per μm² cytoplasm than microvessels in other ONH regions and the retina (P < 0.005; Fig. 6B ). 

Lamina Cribrosa and Retro-laminar Region.

Continuous endothelium with intercellular junctions, suggesting the presence of tight junctions, was observed in microvessels of the LC and RLR. This endothelium displayed few pinocytotic vesicles, mainly located at the abluminal cell membrane (Figs. 5B 5C and 6A , 6B ). 

Retina.

The retinal microvessels displayed a continuous endothelium with intercellular junctions, suggesting the presence of tight junctions. Pinocytotic vesicles of endothelial cells in the retina were mainly located at the abluminal cell membrane (Figs. 5D and 6A) . The number of vesicles per μm² cytoplasm in the retina was higher than in RLR and LC, but significantly lower than in PLR (P < 0.005; Fig. 6B ). 

Discussion

From our results, we conclude that microvessels in the PLR of the ONH of human and monkey eyes lack the typical BBB characteristics observed in microvessels of the other regions of the optic nerve and retina. 

We found that the PLR microvessels express the PAL-E antigen. In earlier work, PAL-E expression in the eye and brain was only observed in areas without a BBB ^{27 28 29 30 31} ; e.g., endothelial cells in the choroid, the ciliary process and choroid plexus stain for PAL-E, but microvessels in the retina and brain that possess a BBB are negative. ^{27 28 29 31}  

The true BBB-markers, which are endothelial membrane proteins associated with the BBB function, were consistently expressed on endothelium of microvessels of the retina and RLR but showed a variable staining pattern in PLR microvessels: No or very low levels of staining for Transf-R and P-gP was observed, suggesting absence of a BBB, but Glut-1 was expressed in this region. This may indicate that these microvessels have an intermediate functional phenotype related to transendothelial transport. However, we also found that all cells in the retina (vascular and nonvascular) and arterioles in the choroid, expressed Glut-1, suggesting an unknown function of this membrane receptor in retina and ONH, unrelated to the BBB. 

The actual barrier function of individual blood vessels in the ONH could be assessed by immunohistochemical staining for plasma proteins fibrinogen and IgG, ³⁰ and these results suggested that local extravasation of endogenous fibrinogen and IgG occurs from microvessels in the PLR but not from microvessels of other ONH regions or the retina. Fibrinogen and IgG were also found in the connective tissue of the LC and PLR but in a pattern suggesting diffusion from the choroid rather than from local leakage from LC microvessels. This is in line with results obtained in studies using exogenous tracers in animal experiments. ^{16 20 21 23 24}  

To find further differences related to BBB properties between microvessels in the PLR and other regions of the ONH, we studied the endothelial cells in these tissues at the ultrastructural level, with special emphasis on pinocytotic vesicles. These organelles are involved in transendothelial transport and a number of other functions in continuous endothelia. ³² BBB endothelium has few of these vesicles. ³³ It was recently shown by us that in unstimulated capillary endothelial cells of the retina these vesicles are mainly located at the abluminal endothelial cell membrane ³⁴ and that in a model of vascular endothelial growth factor–induced hyperpermeability in the monkey eye, more of these vesicles become located at the luminal endothelial cell membrane of retinal endothelial cells of leaky vessels. ³⁴ In the present study, PLR microvessels also showed a high number of pinocytotic vesicles and a localization at both the luminal and abluminal endothelial membrane, whereas microvessels in the other ONH regions had few vesicles, located abluminally, as in the normal retina. 

Hence, these results indicate that microvessels in the PLR differ in permeability characteristics from other ONH regions and retinal microvessels and do not possess a classical BBB. This is not in line with several other studies, in which it was concluded that the entire optic nerve has a BBB, ^{15 16 17 18} based on experiments using exogenous tracers. ^{16 21 23 24} In these studies, the extravasated tracer was frequently observed in the LC and PLR. This observation was solely explained by diffusion from the highly permeable fenestrated microvessels in the peri-papillary choroid ²⁰ and not by leakage from PLR microvessels. This was a logical conclusion, because at the ultrastructural level, intercellular junctions were observed in the PLR microvessels, suggesting the presence of tight junctions and therefore a patent BBB. ^{16 24} However, our results suggest that plasma proteins also extravasate from microvessels in the PLR, possibly by transendothelial transport via pinocytotic vesicles. 

Our findings are supported by clinical observations in routine fluorescein angiography (FA) in normal eyes. To our opinion, the very early diffuse hyperfluorescence of the ONH in FA cannot be explained by diffusion from the surrounding choroid ^{23 35 36} but, as suggested by our present results, may be caused by extravasation of fluorescein from ONH microvessels in the PLR. 

Why PLR microvessels in the normal eye have no classical BBB cannot be answered by the present study. However, the lack of a BBB may have a function in feeding the structures in the PLR. On the other hand, the absence of a BBB may lead to exposure of the nerve fibers in the PLR to plasma proteins and other possibly noxious, circulating substances. This may be of relevance to the vulnerability of this region for nerve fiber damage in glaucoma and other disease entities. Further studies are needed to ascertain whether these influences are involved in the pathogenesis of ONH disease. 

 Supported by Diabetes Fonds Nederland (Grant 95.103), Edmond en Marianne Blaauwfonds, Stichting Blindenhulp, Gelderse Blindenvereniging, and Stichting Dondersfonds.

 Submitted for publication August 11, 2000; revised November 9, 2000; accepted November 29, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

Longitudinal cryo-section through the human ONH showing the different regions studied. Immunostaining with the non-BBB marker PAL-E shows that PLR microvessels are positive for the PAL-E antigen; Re, retina.

Figure 2.

 

Immunoperoxidase staining of cryo-sections of human PLR (A, D, G, J), LC (B, E, H, K), and RLR (C, F, I, L) of the ONH with antibodies against non-BBB endothelium PAL-E, vascular leakage marker IgG, and BBB markers transferrin receptor (Transf-R) and glucose transporter-1 (Glut-1). Arrowheads, positive staining of microvessels; (∗), diffuse stromal staining for IgG.

Figure 4.

 

Immunoperoxidase staining of cryo-sections of human retina with antibodies against non-BBB marker PAL-E (A), vascular leakage marker IgG (B), and BBB markers transferrin receptor (Transf-R; C) and glucose transporter-1 (Glut-1; D). Arrowheads, positive staining. Capillary staining for Glut-1 was masked by staining of other cells in the retina. Note the diffuse staining for IgG in the choroid (∗).

Figure 3.

 

Immunofluorescence double staining of cryosections of human PLR of the ONH (A) and LC (B) with non-BBB marker PAL-E (red fluorescence) and transferrin receptor (green fluorescence). Arrow, a microvessel that coexpresses both markers.

Table 1.

 

Immunohistochemical Staining Results of Human and Monkey Eyes in PLR, LC, and RLR of the Optic Nerve and Retina

Species	Anatomic Region	PAL-E	IgG/Fibr	Transf-R	P-gP	Glut-1
Human	PLR	+	+	+/−	+/−	+
	LC	−	#	+	+	+
	RLR	−	#	+	+	+
	Re	−	#	+	+	+
Monkey	PLR	+	#	ns	ns	+/−
	LC	−	#	ns	ns	+
	RLR	−	#	ns	ns	+
	Re	−	−	ns	ns	+

 +, positive capillary staining; +/−, most microvessels negative;− , no positive capillary staining;

#  , positive staining of connective tissue; ns, not studied because these antibodies do not recognize these proteins in monkey tissues.

Figure 5.

 

Electron micrographs of microvessels in monkey PLR of the ONH (A), LC (B), RLR region of the ONH (C), and retina (D). Arrowheads, pinocytotic vesicles at the luminal endothelial membrane; arrows, pinocytotic vesicles at the abluminal endothelial membrane. L, lumen; E, endothelial cell. Note the luminal localization of pinocytotic vesicles in the PLR (A). Magnification,× 30,000.

Figure 6.

 

Bar graph showing the mean of the ratio of luminal versus abluminal endothelial pinocytotic vesicles of a series of microvessels (±SE of mean; A) and bar graph demonstrating the mean of the number of pinocytotic vesicles per μm² cytoplasm (±SE of mean; B) of a series of microvessels in the PLR, LC, and RLR of the ONH and in the retina (Re) of the monkey. n, number of monkeys studied.

The authors thank Liesbeth Pels and her coworkers of the Corneabank, Amsterdam, for making available the human tissues; the donors and their relatives for their generosity; BIS/Eurotransplant for its assistance in obtaining the tissues; Anneke de Wolf, Niko Bakker, Ton Put, and Marina Danzmann for preparing the photographs; and Jan van Marle from the department of electron microscopy, Academic Medical Center, Amsterdam, The Netherlands, for his help in the double immunofluorescence staining procedures.